From bd8b9fda8163258488123a78d775221760912700 Mon Sep 17 00:00:00 2001 From: Fred Atherden <43879983+fred-atherden@users.noreply.github.com> Date: Fri, 3 Nov 2023 11:36:29 +0000 Subject: [PATCH] Fixes (#107) * Fix for 88819 Relates to elifesciences/enhanced-preprints-import#1515 * Fix for 90293 Unblocks elifesciences/enhanced-preprints-import#1534 * Fix for 87238v2 Unblocks elifesciences/enhanced-preprints-import#1616 * Fix for 90333v2 * Minor update to 90333 ref fix --- README.md | 18 + src/2022.12.29.522272/fix-refs.xsl | 24 + src/2023.02.04.527146/fix-author-emails.xsl | 45 ++ src/2023.04.19.535725/equation-fix.xsl | 26 + src/2023.07.30.551185/fix-ref.xsl | 20 + .../fix-refs/2022.12.29.522272.xml | 334 +++++++++++++ .../fix-author-emails/2023.02.04.527146.xml | 338 +++++++++++++ .../equation-fix/2023.04.19.535725.xml | 456 ++++++++++++++++++ .../fix-ref/2023.07.30.551185.xml | 445 +++++++++++++++++ .../2022.12.29.522272/2022.12.29.522272.xml | 334 +++++++++++++ .../2023.02.04.527146/2023.02.04.527146.xml | 339 +++++++++++++ .../2023.04.19.535725/2023.04.19.535725.xml | 450 +++++++++++++++++ .../2023.07.30.551185/2023.07.30.551185.xml | 445 +++++++++++++++++ 13 files changed, 3274 insertions(+) create mode 100644 src/2022.12.29.522272/fix-refs.xsl create mode 100644 src/2023.02.04.527146/fix-author-emails.xsl create mode 100644 src/2023.04.19.535725/equation-fix.xsl create mode 100644 src/2023.07.30.551185/fix-ref.xsl create mode 100644 test/2022.12.29.522272/fix-refs/2022.12.29.522272.xml create mode 100644 test/2023.02.04.527146/fix-author-emails/2023.02.04.527146.xml create mode 100644 test/2023.04.19.535725/equation-fix/2023.04.19.535725.xml create mode 100644 test/2023.07.30.551185/fix-ref/2023.07.30.551185.xml create mode 100644 test/fixtures/2022.12.29.522272/2022.12.29.522272.xml create mode 100644 test/fixtures/2023.02.04.527146/2023.02.04.527146.xml create mode 100644 test/fixtures/2023.04.19.535725/2023.04.19.535725.xml create mode 100644 test/fixtures/2023.07.30.551185/2023.07.30.551185.xml diff --git a/README.md b/README.md index ba6166f8..6b3ce914 100644 --- a/README.md +++ b/README.md @@ -622,6 +622,24 @@ This xsl accounts for a bug in Encoda. In this preprint some text that precedes The xsl adds an empty `disp-quote` element between the paragraph and display equation in order for the two to be adequately separated in the JSON. +### [/src/2023.04.19.535725/equation-fix.xsl](/src/2023.04.19.535725/equation-fix.xsl) + +This xsl accounts for a bug in Encoda. In this preprint some text that precedes a display equation is pulled in as if it is a caption for that equation which it treats as a figure. + +The xsl adds an empty `disp-quote` element between the paragraph and display equation in order for the two to be adequately separated in the JSON. + +### [/src/2022.12.29.522272/fix-refs.xsl](/src/2022.12.29.522272/fix-refs.xsl) + +This xsl fixes a couple of references which are missing key information and as a result rendering poorly on EPP. + +### [/src/2023.02.04.527146/fix-author-emails.xsl](/src/2023.02.04.527146/fix-author-emails.xsl) + +This xsl is to ensure that the correct email is attributed to the correct author. bioRxiv capture author emails addresses in a `` inside the author notes. They do this becuase they intend to show the content as a string, instead of displaying the emails under each author it relates to. We have asked them to change this capture (capturing the email under the respecitve author contrib, as done in this xsl). + +### [/src/2023.07.30.551185/fix-ref.xsl](/src/2023.07.30.551185/fix-ref.xsl) + +This xsl fixes a reference which is missing key information and as a result rendering poorly on EPP. + # Modify bioRxiv XML in preparation for Encoda Prerequisites: diff --git a/src/2022.12.29.522272/fix-refs.xsl b/src/2022.12.29.522272/fix-refs.xsl new file mode 100644 index 00000000..4312d330 --- /dev/null +++ b/src/2022.12.29.522272/fix-refs.xsl @@ -0,0 +1,24 @@ + + + + + + + + + + + + + H. Pelicano, D. S. Martin, R. Xu, P. Huang, Glycolysis inhibition for anticancer treatment. Oncogene. 4633–4646 (2006). + + + + A. Walvekar, Z. Rashida, H. Maddali, S. Laxman, A versatile LC-MS / MS approach for comprehensive, quantitative analysis of central metabolic pathways [version 1; peer review : 2 approved]. Wellcome Open Res 1–15 (2018). + + + \ No newline at end of file diff --git a/src/2023.02.04.527146/fix-author-emails.xsl b/src/2023.02.04.527146/fix-author-emails.xsl new file mode 100644 index 00000000..20976e99 --- /dev/null +++ b/src/2023.02.04.527146/fix-author-emails.xsl @@ -0,0 +1,45 @@ + + + + + + + + + + + + + + + + + + paulhage@jhu.edu + + + + + + + yes + + shadmehr@jhu.edu + + + + + + + + + + + + + + \ No newline at end of file diff --git a/src/2023.04.19.535725/equation-fix.xsl b/src/2023.04.19.535725/equation-fix.xsl new file mode 100644 index 00000000..c0eeec24 --- /dev/null +++ b/src/2023.04.19.535725/equation-fix.xsl @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + \ No newline at end of file diff --git a/src/2023.07.30.551185/fix-ref.xsl b/src/2023.07.30.551185/fix-ref.xsl new file mode 100644 index 00000000..2f9db1e9 --- /dev/null +++ b/src/2023.07.30.551185/fix-ref.xsl @@ -0,0 +1,20 @@ + + + + + + + + + + + + + Abad C, Cook MM, Cao L, Jones JR, Rao NR, Dukes-Rimsky L, Pauly R, Skinner C, Wang Y, Luo F et al (2018) A Rare De Novo RAI1 Gene Mutation Affecting BDNF-Enhancer-Driven Transcription Activity Associated with Autism and Atypical Smith-Magenis Syndrome Presentation. Biology (Basel) 7 + + + \ No newline at end of file diff --git a/test/2022.12.29.522272/fix-refs/2022.12.29.522272.xml b/test/2022.12.29.522272/fix-refs/2022.12.29.522272.xml new file mode 100644 index 00000000..a72d27a4 --- /dev/null +++ b/test/2022.12.29.522272/fix-refs/2022.12.29.522272.xml @@ -0,0 +1,334 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2022.12.29.522272 +1.3 + + +Regular Article + + +New Results + + +Biochemistry + + + +The deubiquitinase Ubp3/Usp10 constrains glucose-mediated mitochondrial repression via phosphate budgeting + + + +http://orcid.org/0000-0002-1078-5476 +VengayilVineeth +1 +2 + + +NiphadkarShreyas +1 +2 + + +AdhikarySwagata +1 +2 + + +http://orcid.org/0000-0002-3609-4032 +VarahanSriram +1 + + +http://orcid.org/0000-0002-0861-5080 +LaxmanSunil +1 +* + +Institute for Stem Cell Science and Regenerative Medicine (DBT-inStem), GKVK Post Bellary Road, Bangalore 560065 +Manipal Academy of Higher Education, Manipal, Karnataka 576104, India + + +correspondence: sunil@instem.res.in + +2023 +2022.12.29.522272 + + +29 +12 +2022 + + +22 +8 +2023 + + +23 +8 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at http://creativecommons.org/licenses/by-nc-nd/4.0/ + + + +Abstract +

Many cells in high glucose repress mitochondrial respiration, as observed in the Crabtree and Warburg effects. Our understanding of biochemical constraints for mitochondrial activation is limited. Using a Saccharomyces cerevisiae screen, we identified the conserved deubiquitinase Ubp3 (Usp10), as necessary for mitochondrial repression. Ubp3 mutants have increased mitochondrial activity despite abundant glucose, along with decreased glycolytic enzymes, and a rewired glucose metabolic network with increased trehalose production. Utilizing Δubp3 cells, along with orthogonal approaches, we establish that the high glycolytic flux in glucose continuously consumes free Pi. This restricts mitochondrial access to inorganic phosphate (Pi), and prevents mitochondrial activation. Contrastingly, rewired glucose metabolism with enhanced trehalose production and reduced GAPDH (as in Δubp3 cells) restores Pi. This collectively results in increased mitochondrial Pi and derepression, while restricting mitochondrial Pi transport prevents activation. We therefore suggest that glycolytic-flux dependent intracellular Pi budgeting is a key constraint for mitochondrial repression.

+ + +Keywords +mitochondria +glycolysis +phosphate +deubiquitinase +Crabtree + + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Title and figures updated.

 + + + + + +Introduction +

Rapidly proliferating cells have substantial metabolic and energy demands in order to increase biomass (1, 2). This includes a high ATP demand, obtained from cytosolic glycolysis or mitochondrial oxidative phosphorylation (OXPHOS), to fuel multiple reactions (3). Interestingly, many rapidly proliferating cells preferentially rely on ATP from glycolysis/fermentation over mitochondrial respiration even in oxygen-replete conditions, and is the well-known Warburg effect (4, 5). Many such cells repress mitochondrial processes in high glucose, termed glucose-mediated mitochondrial repression or the Crabtree effect (6, 7). This is observed in some tumors (5), neutrophils (8), activated macrophages (9), stem cells (10–12), and famously Saccharomyces cerevisiae (7). Numerous studies have identified signaling programs or regulators of glucose-mediated mitochondrial repression. However, biochemical programs and regulatory processes in biology evolve around key biochemical constraints (13). The biochemical constraints for mitochondrial repression remain unresolved (14, 15).

+

There are two hypotheses on the biochemical principles driving mitochondrial repression. The first proposes direct roles for glycolytic intermediates in driving mitochondrial respiration, by inhibiting specific mitochondrial outputs (16, 17). The second hypothesizes that a competition between glycolytic and mitochondrial processes for mutually required metabolites/co-factors such as pyruvate, ADP or inorganic phosphate (Pi) could determine the extent of mitochondrial repression (14, 15, 18). These are not all mutually exclusive, and a combination of these factors might dictate mitochondrial repression in high glucose. However, any hierarchies of importance are unclear (19), and experimental data for the necessary constraints for mitochondrial repression remains incomplete.

+

One approach to resolve this question has been to identify regulators of metabolic state under high glucose. Post-translational modifications (PTMs) regulated by signaling systems can regulate mitochondrial repression (20–22). Ubiquitination is a PTM that regulates global proteostasis (23, 24), but the roles of ubiquitination-dependent processes in regulating mitochondrial repression are poorly explored. Ubiquitination itself is determined by the balance between ubiquitination, and deubiquitinase (DUB) dependent deubiquitination (25). Little is known about the roles of DUBs in regulating metabolic states, making the DUBs interesting candidate regulators of mitochondrial repression.

+

In this study, using an S. cerevisiae DUB knock-out library-based screen, we identified the evolutionarily conserved deubiquitinase Ubp3 (mammalian Usp10) as required for mitochondrial repression in high glucose. Loss of Ubp3 resulted in mitochondrial activation, along with a reduction in the glycolytic enzymes-phosphofructokinase 1 (Pfk1) and GAPDH (Tdh2 and Tdh3). This consequently reroutes glucose flux and increases trehalose biosynthesis. This metabolic rewiring increases Pi release from trehalose synthesis, and decreases Pi consumption in glycolysis, to cumulatively increase Pi pools. Using ubp3Δ cells along with independent analysis of wild-type cells, and isolated mitochondrial fractions, we establish that glycolytic-flux dependent Pi allocations to mitochondria determines mitochondrial activity. Through these data, we propose how intracellular Pi balance as controlled by glycolytic flux, is a biochemical constraint for mitochondrial repression.

+ + +Results + +A deubiquitinase deletion screen identifies Ubp3 as a regulator of glucose-mediated mitochondrial repression +

In cells such as S. cerevisiae, high glucose represses mitochondrial activity as well as OXPHOS-dependent ATP synthesis (7, 26) (illustrated in Figure 1A). Our initial objective was to identify proteostasis regulators of glucose-mediated mitochondrial repression. We generated and used a deubiquitinase (DUB) deletion strain library of S. cerevisiae (Figure S1A), to unbiasedly identify regulators of mitochondrial repression by measuring the fluorescence intensity of a potentiometric dye Mitotracker CMXRos (illustrated in Figure 1B). Using this screen, we identified DUB mutants with altered mitochondrial membrane potential (Figure 1C, Figure S1A). Note: WT cells in a respiratory medium (2% ethanol) were used as a control to estimate maximum mitotracker fluorescence intensity (Figure S1B).

+ + +A deubiquitinase deletion screen identifies Ubp3 as a regulator of glucose-mediated mitochondrial repression +

A) Schematic depicting glucose-mediated mitochondrial repression (Crabtree effect).

B) Schematic describing the screen with a yeast DUB KO library to identify regulators of Crabtree effect.

C) Identifying DUB knockouts with altered mitochondrial potential. Heat map shows relative mitochondrial membrane potential of 19 deubiquitinase deletions in high glucose, from 2 biological replicates. Also see figures S1A, S1B.

D) The deubiquitinase activity of Ubp3 and repression of mitochondrial membrane potential. WT, ubp3Δ, and Ubp3C469A were grown inhigh glucose and relative mitochondrial membrane potential was measured. Data represent mean ± SD from three biological replicates (n=3). Also see figure S1D.

E) Effect of loss of Ubp3 on ETC complex IV subunit Cox2. WT and ubp3Δ were grown in high glucose, and Cox2 was measured (western blot using an anti-Cox2 antibody). A representative blot (out of 3 biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

F) Basal oxygen consumption rate (OCR) in high glucose in ubp3Δ. WT and ubp3Δ were grown in high glucose, and OCR was measured. Basal OCR corresponding to ∼3 x 10^5 cells, from two independent experiments (n=2), normalized to the OD600 is shown. Bar graph representations are shown in the inset. Data represent mean ± SD.

G) Total ATP levels in ubp3Δ and WT. WT and ubp3Δ were grown in high glucose, and total ATP were measured. Data represent mean ± SD from three biological replicates (n=3).

H) Dependence of ubp3Δ on mitochondrial ATP. WT and ubp3Δ cells were grown in high glucose, and treated with 1 mM sodium azide for 45 minutes. Total ATP levels in sodium azide treated and untreated cells were measured. Data represent mean ± SD (n=3).

I) Requirement for mitochondrial respiration in high glucose in ubp3Δ. Serial dilution growth assay in high glucose in the presence/absence of sodium azide (1 mM) are shown. Also see figure S1F.

Data information: **p<0.01, ***p<0.001.

+ + +

A prominent ‘hit’ was the evolutionarily conserved deubiquitinase Ubp3 (Figure 1C), (homologous to mammalian Usp10) (Figure S1C). Due to its high degree of conservation across eukaryotes (Figure S1C) as well as putative roles in metabolism or mitochondrial function (27–29), we focused our further attention on this DUB. Cells lacking Ubp3 showed an ∼1.5-fold increase in mitotracker fluorescence (Figure 1C, Figure S1A). Cells with catalyticaly inactive Ubp3 (Ubp3C469A), showed increased mitochondrial potential comparable to ubp3Δ (Figure 1D). This data confirmed that Ubp3 catalytic activity of is required to fully repress mitochondrial activity under high glucose. The catalytic site mutation did not affect steady-state Ubp3 levels (Figure S1D).

+

Next, to assess the requirement of Ubp3 for mitochondrial function, we quantified the electron transport chain (ETC) complex IV subunit Cox2 (30)(Figure 1E). ubp3Δ had higher Cox2 than WT (Figure 1E). There was no increase in mitochondrial outer membrane protein Tom70 (Figure S1E), suggesting that the increased Cox2 is not merely because of higher total mitochondrial content. We next measured the basal oxygen consumption rate (OCR) of ubp3Δ, and basal OCR was higher in ubp3Δ, indicating higher respiration (Figure 1F).

+

Next, we asked if mitochondrial ATP synthesis was higher in ubp3Δ. The total ATP levels in WT and ubp3Δ were comparable (Figure 1G). However, upon treatment with the ETC complex IV inhibitor sodium azide, ATP levels in WT were higher than ubp3Δ, contributing to ∼60% of the total ATP (Figure 1H). In contrast, the ATP levels in ubp3Δ after sodium azide treatment were ∼ 40% of the total ATP (Figure 1H). These data suggest a higher contribution of mitochondrial ATP synthesis towards the total ATP pool in ubp3Δ.

+

We next asked if ubp3Δ required higher mitochondrial activity for growth. In high glucose, WT show minimal growth inhibition in the presence of sodium azide, indicating lower reliance on mitochondrial function (Figure 1I). Contrastingly, ubp3Δ or Ubp3C469A show a severe growth defect in the presence of sodium azide (Figure 1I). Deletion of ETC components Cox2, and the ATP synthase subunits Atp1 and Atp10 results in a severe growth defect in ubp3Δ compared to WT (Figure S1F). Together, these results indicate that the loss of Ubp3 makes cells dependent on mitochondrial ATP synthesis in high glucose.

+

Collectively these data show that in 2% glucose, ubp3Δ have high mitochondrial activity, respiration, and rely on this mitochondrial function for ATP production and growth. We therefore decided to use ubp3Δ cells to start delineating requirements for glucose-mediated mitochondrial repression.

+ + +Key glycolytic enzymes decrease and glucose flux is rerouted in <italic>ubp3Δ</italic> cells +

Glucose-6 phosphate (G6P) is the central node in glucose metabolism where carbon-allocations are made towards distinct metabolic arms, primarily: glycolysis, the pentose phosphate pathway (PPP), and trehalose biosynthesis (Figure 2A). We first compared amounts of two key (‘rate-controling’) glycolytic enzymes - Phosphofructokinase 1 (Pfk1), GAPDH isozymes (Tdh2, Tdh3) (3), along with the enolase isozymes (Eno1, Eno2) in WT and ubp3Δ cells. Pfk1, Tdh2, and Tdh3 substantially decreased in ubp3Δ (but not Eno1 and Eno2) (Figure 2B, S2A). Since deubiquitinases can control protein amounts by regulating proteasomal degradation, we asked if the decrease in Pfk1, Tdh2 and Tdh3 in ubp3Δ is due to proteasomal degradation. To test this, we measured the levels of these enzymes in ubp3Δ after treatement with proteasomal inhibitor MG132. We did not observe any rescue in the levels of these enzymes in MG132 treated samples, suggesting that the decreased levels of these enzymes were not due to increased proteasomal degradation (Figure S2B). To further understand if these enzyme transcripts are altered in ubp3Δ, we measured the mRNA levels of PFK1, TDH2 and TDH3 in WT, ubp3Δ and Ubp3C469A cells. The transcripts of all the three genes in ubp3Δ and Ubp3C469A cells decreased (Figure S2C), suggesting that Ubp3 regulates the transcripts of these glycolytic enzyme genes. The reduction in the Pfk and GAPDH enzyme amounts was intriguing, because the Pfk and GAPDH steps are critical in determining glycolytic flux (3, 31). A reduction in these enzymes could therefore decrease glycolytic flux, and would reroute glucose (G6P) allocations via mass action towards other branches of glucose metabolism, primarily the pentose phosphate pathway as well as trehalose biosynthesis (Figure 2A). To assess this, we first measured the steady-state levels of key glycolytic and pentose phosphate pathway (PPP) intermediates, and trehalose in WT or ubp3Δ using targeted LC-MS/MS. Glucose-6/fructose-6 phosphate (G6P/F6P) increased in ubp3Δ (Figure 2C). Concurrently, trehalose, and the PPP intermediates ribose 5-phosphate (R5P) and sedoheptulose 7-phosphate (S7P), increased in ubp3Δ (Figure 2C).

+ + +Key glycolytic enzymes decrease and glucose flux is rerouted in <italic>ubp3Δ</italic> cells +

A) A schematic illustrating directions of glucose-6-phosphate (G6) flux in cells. Glucose is converted to G6P, a precursor for trehalose, the pentose phosphate pathway (PPP), and glycolysis.

B) Effect of loss of Ubp3 on key glycolytic enzymes. WT and ubp3Δ were grown in high glucose and the Pfk1, Tdh2, and Tdh3 levels were measured by western blot using an anti-FLAG antibody. A representative blot (out of three biological replicates, n=3) and their quantification are shown. Data represent mean ± SD. Also see figure S2A.

C) Steady-state metabolite amounts in WT and ubp3Δ in high glucose. Relative steady-state levels of trehalose, major glycolytic, and PPP intermediates were estimated in WT and ubp3Δ. Data represent mean ± SD from three biological replicates (n=3). Also see Appendix Table S3.

D) Relative glycolytic and trehalose synthesis flux in WT and ubp3Δ. Relative 13C-label incorporation into trehalose and glycolytic intermediates, after a pulse of 1% 13C6 glucose is shown. Data represent mean ± SD from three biological replicates (n=3). Also see Appendix Table S3, figures S2B, S2C.

E) Ethanol production in ubp3Δ. WT and ubp3Δ were grown in high glucose and ethanol in the media was measured. Data represent mean ± SD from three biological replicates (n=3).

F) Relative rate of ethanol production in WT vs ubp3Δ. WT and ubp3Δ were grown in high glucose (to OD600 ∼ 0.6), equal numbers of cells were shifted to fresh medium (high glucose) and ethanol concentration in the medium was measured temporally. Data represent mean ± SD from three biological replicates (n=3)

Data information: *p<0.05, **p<0.01, ***p<0.001.

+ + +

Since steady-state metabolite amounts cannot separate production from utilization, in order to unambiguously assess if glycolytic flux is reduced in ubp3Δ cells, we utilized a pulse labeling of 13C6 glucose, following which the label incorporation into glycolytic and other intermediates was measured. Note that because glycolytic flux is very high in yeast, this experiment would require rapid pulsing and extraction of metabolites in order to stay in a linear range and avoid label saturation. We therefore established a very short time point of label-addition, quenching and metabolite extraction post 13C glucose pulse. Since flux saturates/reaches steady-state in seconds, we first ensured that the label incorporation into individual metabolites after the 13C glucose pulse was in the linear range, and for early glycolytic intermediates this was seconds after glucose addition. This new methodology is extensively described in materials and methods, with required controls shown in Figure S2E. WT and ubp3Δ cells were grown in high glucose, pulsed with 13C6 glucose, and the relative 13C label incorporation into glycolytic intermediates and trehalose were measured, as shown in the schematic (Figure S2D). In ubp3Δ, 13C label incorporation into G6P/F6P as well as trehalose substantially increased (Figure 2D). Contrastingly, 13C label incorporation into glycolytic intermediates F1,6BP, G3P, 3PG and PEP decreased, indicating decreased glycolytic flux (Figure 2D). We next measured ethanol concentrations and production rates, as an additional output of relative glycolytic rates. We observed decreased steady-state ethanol levels, as well as ethanol production rates in ubp3Δ (Figure 2F).

+

Collectively, these results reveal that that reduced Pfk1 and GAPDH (in ubp3Δ) decrease glucose flux via glycolysis, which results in rewired glucose flux towards trehalose biosynthesis and the PPP.

+ + +Rerouted glucose flux results in phosphate (Pi) accumulation +

We therefore asked if the proteomic state, as observed in ubp3Δ cells, could provide clues to explain the coupling between mitochondrial derepression and rerouted glucose flux. A recent study by Isasa et al., had systematically quantified the changes in protein levels in ubp3Δ cells (28). We therefore reanalyzed this extensive dataset, looking for changes in proteins that would correlate with these metabolic processes. Notably, we observed increased levels of proteins of the mitochondrial ETC and respiration, and decreased amounts of glucose metabolizing enzymes in ubp3Δ (Figure 3A). Additionally, multiple proteins involved in regulating phosphate (Pi) homeostasis were decreased in ubp3Δ (Figure 3A) (28). We had recently uncovered a reciprocal coupling of Pi homeostasis with the different arms of glucose metabolism, particularly trehalose biosynthesis (32, 33), and therefore wondered if a glycolytic-flux dependent change in Pi homeostatsis had any role in mitochondrial respiration. Our hypothesis was refined based on the reasoning given below.

+ + +Rerouted glucose flux results in phosphate (Pi) accumulation +

A) Changes in protein levels in ubp3Δ (dataset from Isasa et al., 2015). ubp3Δ cells have an increase in proteins involved in mitochondrial respiration and decrease in proteins involved in glucose and phosphate metabolism.

B) Schematic showing maintenance of Pi balance during glycolysis. Trehalose synthesis from G6P releases Pi, and the conversion of G3P to 1,3BPG by GAPDH consumes Pi. In ubp3Δ, trehalose biosynthesis (which releases Pi) increases. ubp3Δ have decreased GAPDH, which will decrease Pi consumption. This increase in Pi release along with decreased Pi consumption could increase cytosolic Pi.

C) Intracellular Pi levels in WT and ubp3Δ. WT and ubp3Δ were grown in high glucose and the total free phosphate (Pi) levels were estimated. WT in high Pi (2% glucose, 10mM Pi) was a positive control. Data represent mean ± SD from three biological replicates (n=3). Also see figure S3A.

D) Pho regulon responses in WT and ubp3Δ. Protein levels of Pho84-FLAG and Pho12-FLAG were compared between WT grown in high glucose and in high Pi, ubp3Δ in high glucose with or without a shift to a no-Pi medium for one hour, by western blot. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

E) Contribution of trehalose synthesis as a Pi source. WT, tps2Δ, ubp3Δ, and ubp3Δtps2Δ were grown in high glucose and the total Pi levels were estimated. Data represent mean ± SD from three biological replicates (n=3). Also see figure S3B.

F) Loss of GAPDH isozymes Tdh2 and Tdh3 and effect on Pi. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and total Pi was estimated. Data represent mean ± SD from three biological replicates (n=3).

G) Pi levels in ubp3Δ and tdh2Δtdh3Δ cells in ethanol medium. WT, ubp3Δ, and tdh2Δtdh3Δ cells were grown in ethanol medium and the total Pi levels were estimated. Data represent mean ± SD from three biological replicates (n=3).

Data information: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

+ + +

One explanation for mitochondrial repression can be an internal competition for shared metabolites/co-factors between (cytosolic) glycolytic and mitochondrial processes, that mitochondria might not be sufficiently able to access (14, 18, 19). In this context, a plausible role for inorganic phosphate (Pi) in regulating mitochondrial repression can be hypothesized. Cytosolic glycolysis requires rapid, high consumption of net Pi (19, 34, 35), and this could possibly limit the Pi that is continuously available for mitochondrial use (18, 36), thereby repressing mitochondria. Contextually, the balance between reactions releasing vs consuming Pi could explain changes in global Pi levels (33). A continuous hub of Pi consumption is glycolysis. Here, GAPDH catalyzes G3P to 1,3BPG, converting ADP to ATP, while concurrently consuming a molecule of Pi (35, 37). This Pi that goes into ATP will subsequently be used for nucleotide biosynthesis, polyphosphate biosynthesis and protein phosphorylation (33, 38). Therefore, we can surmise that in high glycolytic flux, the production of ATP, nucleotides and polyphosphates is concurrent with Pi consumption (37, 39, 40). Could this reaction therefore limit cytosolic Pi for the mitochondria (as illustrated in Figure 3B)? Notably, ubp3Δ have reduced GAPDH levels and decreased glycolytic flux. Second, trehalose synthesis is a Pi releasing reaction, and a major source of free Pi that is critical for Pi homeostasis (32, 35). Flux through this reaction is also substantially higher in ubp3Δ. Therefore, these cells might have increased Pi release (via trehalose), coupled with decreased Pi consumption (via decreased GAPDH). We therefore asked if total Pi increases in ubp3Δ (Figure 3B)?

+

To test this, we directly assessed total Pi levels in ubp3Δ and WT. ubp3Δ cells had higher Pi than WT, and this was comparable to Pi in WT grown in excess Pi (Figure 3C). Similarly, Pi amounts also increased in Ubp3C469A (Figure S3A). Therefore, the loss of Ubp3 increases intracellular Pi levels. Next, we asked if ubp3Δ cells exhibit signatures of a ‘high Pi’ state. S. cerevisiae maintains internal Pi balance by controlling the expression of multiple genes collectively known as the Pho regulon (41). The Pho regulon is induced under Pi limitation, and repressed during Pi sufficiency (32, 41). We assessed two major Pho proteins (Pho84: a high-affinity membrane Pi transporter, and Pho12: an acid phosphatase) in WT and ubp3Δ in high glucose. ubp3Δ cells have lower amounts of Pho84 and Pho12 (Figure 3D). Further, upon shifting to low Pi for one hour, Pho84 and Pho12 increased in ubp3Δ. These data suggest that reduced Pho84 and Pho12 amounts in ubp3Δ are because of increased Pi, and not due to altered Pho regulon function itself (Figure 3D). These data clarify earlier observations from ubp3Δ which noted reduced Pho proteins (28). Therefore, ubp3Δ constitutively have higher Pi, and likely a consequent decrease in Pho proteins.

+

We next asked if the increased Pi in ubp3Δ is because of altered G6P allocations towards different end-points, particularly trehalose synthesis, which can be a major node of Pi restoration (32, 35). We assessed the contribution of increased trehalose synthesis towards the high Pi in ubp3Δ, by estimating Pi levels in the absence of trehalose 6-phosphate phosphatase (Tps2), which catalyzes the Pi-releasing step in trehalose synthesis. Notably, loss of Tps2 in ubp3Δ decreased Pi (Figure 3E). There was no additive difference in Pi between tps2Δ and ubp3Δtps2Δ (Figure 3E). As an added control, we assessed trehalose in WT and ubp3Δ in the absence of Tps2, and found no difference (Figure S3B). Therefore, increasing G6P flux towards trehalose biosynthesis is a major source of the increased Pi in ubp3Δ.

+

Since the major GAPDH isozymes, Tdh2 and Tdh3, are reduced in ubp3Δ, we directly asked if reducing GAPDH can decrease Pi consumption and increase Pi. To assess this, we generated tdh2Δtdh3Δ cells, which exhibit a growth defect, but are viable, permitting further analysis. tdh2Δtdh3Δ had higher Pi in high glucose (Figure 3F). Expectedly, we observed a decrease in ethanol in tdh2Δtdh3Δ (Figure S3C), along with an accumulation of F1,6 BP and G3P, and decreased 3PG and PEP (Figure S3D). However, G6P and trehalose levels between WT and tdh2Δtdh3Δ were comparable (Figure S3D, S3E). These data sugest that unlike in ubp3Δ, the increased Pi in tdh2Δtdh3Δ comes mainly from decreased Pi consumption (GAPDH step). To further assess the role of reduced glycolytic flux in increasing Pi (ubp3Δ and tdh2Δtdh3Δ), we measured the Pi in these cells growing in a gluconeogenic medium – 2%, ethanol. In this scenario, the GAPDH catalyzed reaction will be reversed, converting 1,3BPG to G3P, which should release and not consume Pi. Compared to WT, the Pi levels decreased in ubp3Δ and tdh2Δtdh3Δ (Figure 3G), suggesting that the changes in Pi in these mutants is driven by the relative change in Pi release vs consumption.

+

These results collectively indicate that the combined effect of increased Pi release coming from trehalose synthesis, and decreased Pi consumption from reduced GAPDH increase Pi levels in ubp3Δ.

+ + +Mitochondrial Pi availability correlates with mitochondrial activity in <italic>ubp3Δ</italic> +

We therefore wondered if this observed Pi increase from the combined rewiring of glucose metabolism resulted in more Pi becoming accessible to the mitochondria. This would effectively result in Pi budgeting between cytosolic glycolysis and mitochondria based on the relative flux in different arms of glucose metabolism, by increasing Pi availability for the mitochondria (Figure 4A). To test if this were so, we first asked if the mitochondria in ubp3Δ have increased Pi. Mitochondria were isolated by immunoprecipitation from WT and ubp3Δ, the isolation efficiency was analyzed (Figure S4A), and relative Pi levels compared. Pi levels were normalised to Idh1 (isocitrate dehydrogenase) in isolated mitochondria, since Idh1 protein levels did not decrease in ubp3Δ (Figure S4B), and consistent with (28). Mitochondrial Pi was higher in ubp3Δ (Figure 4B). We next asked if the increased mitochondrial activity in ubp3Δ is a consequence of high Pi. If this were so, tdh2Δtdh3Δ should partially phenocopy ubp3Δ with respect to mitochondrial activity. Consistently, tdh2Δtdh3Δ have higher mitotracker intensity as well as Cox2 levels compared to WT (Figure S4C, Figure 4C). Also consistent with this, a high basal OCR in tdh2Δtdh3Δ was observed (Figure 4D), together indicating high mitochondrial activity in tdh2Δtdh3Δ, which is comparable to ubp3Δ.

+ + +Mitochondrial Pi availability correlates with mitochondrial activity in <italic>ubp3Δ</italic> +

A) A hypothetical mechanism of cytosolic free Pi controling mitochondrial activity by regulating mitochondrial Pi availability.

B) Mitochondrial Pi amounts in WT vs ubp3Δ. Mitochondria were isolated by immunoprecipitation from WT and ubp3Δ and mitochondrial Pi estimated. Mitochondrial Pi levels (normalised to Idh1) are shown. Data represent mean ± SD from three biological replicates (n=3). Also see figure S4A, B.

C) Cox2 protein in tdh2Δtdh3Δ. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and Cox2 protein was estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

D) Basal OCR levels in tdh2Δtdh3Δ. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD. Also see figure S4C.

E) Comparative Pi amounts and Cox2 levels in ubp3Δ, tdh2Δtdh3Δ, WT cells. WT cells were grown in high glucose, ubp3Δ and tdh2Δtdh3Δ were grown in high glucose and low Pi, and Cox2 protein was estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S4D, S4F.

F) Pi amounts and basal OCR in ubp3Δ and tdh2Δtdh3Δ vs WT cells. WT cells were grown in high glucose, ubp3Δ and tdh2Δtdh3Δ were grown in high glucose and low Pi, and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD.

G) Effect of loss of mitochondrial Pi transporter Mir1 on Cox2protein. WT, ubp3Δ, mir1Δ, and mir1Δubp3Δ were grown in high glucose and Cox2 amounts compared. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

H) Relationship of mitochondrial Pi transport and basal OCR in WT vs ubp3Δ. WT, ubp3Δ, mir1Δ, and mir1Δubp3Δ cells were grown in high glucose and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD.

Data information: *p<0.05, **p<0.01, ****p<0.0001.

+ + +

We next asked if higher Pi is necessary to increase mitochondrial activity in ubp3Δ and tdh2Δtdh3Δ. Since glycolysis is defective in ubp3Δ and tdh2Δtdh3Δ, it is necessary to distinguish the effect of high Pi vs. only the effect of low glycolysis in activating mitochondria. Logically, if decreased glycolysis (independent of Pi) is sufficient to activate mitochondria, bringing down the Pi levels in ubp3Δ to that of WT should not affect mitochondrial activity. Notably, ubp3Δ grown in low (1 mM) Pi have Pi levels similar to WT in standard (normal Pi) medium (Figure S4D). ubp3Δ grown in low Pi also had decreased ethanol, suggesting reduced glycolysis (Figure S4E). Therefore, we used this condition to further understand the role of Pi in inducing mitochondrial activity. Mitotracker fluorescence decreased in both ubp3Δ and tdh2Δtdh3Δ in low Pi (Figure S4F). Note: basal mitotracker fluorescence in WT also decreases in low Pi, which is consistent with a required role of Pi for mitochondrial activity (Figure S4F). Similarly, Cox2 levels were reduced in both ubp3Δ and tdh2Δtdh3Δ (Figure 4E). Consistent with both reduced mitotracker intensity and Cox2 levels, basal OCR also decreased in both ubp3Δ and tdh2Δtdh3Δ in low Pi (Figure 4F). These data collectively suggest that high intracellular Pi is necessary to increase mitochondrial activity in ubp3Δ and tdh2Δtdh3Δ.

+

Next, we asked how mitochondrial Pi transport regulates mitochondrial activity. Mir1 and Pic2 are mitochondrial Pi transporters, with Mir1 being the major Pi transporter (42, 43). We first limited mitochondrial Pi availability in ubp3Δ by knocking-out MIR1. In mir1Δ, the increased Cox2 observed in ubp3Δ was no longer observed (Figure 4G). Consistent with this, we observed no further increase in the basal OCR in mir1Δubp3Δ compared to mir1Δ (Figure 4H). Furthermore, mir1Δ showed decreased Cox2 as well as basal OCR even in WT cells (Figure 4G, Figure 4H). These data together suggest that mitochondrial Pi transport is critical for increasing mitochondrial activity in ubp3Δ, and in maintaining basal mitochondrial activity even in high glucose.

+

As a control, no significant increase in Mir1 and Pic2 was observed in ubp3Δ (Figure S4G), suggesting that ubp3Δ do not increase mitochondrial Pi by merely increasing the Pi transporters, but rather by increasing available Pi pools.

+

Taken together, these data suggest the possibility that increased cytosolic Pi increases the mitochondrial Pi pool, which increases mitochondrial activity. Decreasing mitochondrial Pi by either reducing total Pi, or by reducing mitochondrial Pi transport decreases mitochondrial activity.

+ + +Mitochondrial Pi availability constrains mitochondrial activity under high glucose +

So far, these data suggest that the cytosolic Pi available for the mitochondria can dermine the extent of mitochondrial activity. Therefore, we further investigated if mitochondrial Pi allocation was a necessary constraint for glucose-mediated mitochondrial repression.

+

To test this, we first asked how important mitochondrial Pi transport was to switch to increased respiration. In WT yeast, low glucose or glycolytic inhibition will result in increased respiration (22). What happens therefore if we restrict mitochondrial Pi in this context? For this, we measured the basal OCR in WT and mir1Δ after switching from high (2%) to low (0.1%) glucose. We observed a significant increase in the basal OCR in WT but not in mir1Δ (Figure 5A). The alternate scenario is after glycolytic inhibition. We assessed the role of mitochondrial Pi in this context, by inhibiting glycolysis using 2-deoxyglucose (2DG). WT, but not mir1Δ, increased their OCR (respiration) upon a 1-hour treatment with 2DG (Figure 5B). Consistent with this, mitotracker fluorescence increased with an increase in 2DG in WT, but not in mir1Δ (Figure S5A). We further asked if the mitochondrial Pi transporter itself glucose repressed, and therefore assessed Mir1 amounts in high and low glucose. Mir1 levels are higher upon a shift to low glucose, and in cells grown in 2% ethanol, suggesting that Mir1 is glucose repressed (Figure 5C, S5B). These data suggest that mitochondrial Pi transport is necessary for increasing mitochondrial activity after glucose derepression.

+ + +Mitochondrial Pi availability constrains mitochondrial activity under high glucose +

A) Relationship of mitochondrial Pi transport and respiration after glucose removal. WT and mir1Δ cells were cultured in high (2%) glucose and shifted to low (0.1%) glucose for 1 hour. The normalized basal OCR, from two independent experiments (n=2) are shown. Data represent mean ± SD.

B) Requirement of mitochondrial Pi transport for switch to respiration upon glycolytic inhibition by 2DG. WT and mir1Δ cells were cultured in high glucose and treated with or without 0.25% 2DG for 1 hour. Basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD. Also see figure S5A.

C) Glucose-dependent regulation of Mir1. Cells (with Mir1-HA) were grown in high glucose and shifted to low glucose (0.1% glucose) for 1 hour, and Mir1 levels compared. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S5B.

D) Increasing Pi concentrations and mitochondrial activity in isolated mitochondria. Mitochondria were isolated from WT cells grown in high glucose, incubated with 1 mM pyruvate, 1 mM malate, 0.5 mM ADP and 0-50 mM KH2PO4. The mitochondrial activity was estimated by mitotracker fluorescence intensity, and intensities relative to the sample with 0 mM KH2PO4 is shown. Data represent mean ± SD from three biological replicates (n=3).

E) Effect of overexpressing Mir1 on Cox2 protein. WT (containing empty vector) and Mir1 overexpressing (Mir1OE) cells were grown in high glucose and Cox2 levels were estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S5F.

F) Effect of overexpressing Mir1 on basal OCR. The basal OCR in WT (containing empty vector) and Mir1OE in high glucose was measured from three independent experiments (n=3). Data represent mean ± SD.

G) Mitochondrial Pi transport requirement for growth after 2DG treatment. Shown are serial dilution growth assays in high glucose in the presence and absence of 0.1% 2DG, using WT and mir1Δ cells. The results after 40hrs incubation/30°C are shown.

H) A model illustrating how mitochondrial Pi availability controls mitochondrial activity. In high glucose, the decreased Pi due to high Pi consumption in glycolysis, along with the glucose-mediated repression of mitochondrial Pi transporters, decreases mitochondrial Pi availability. This reduces mitochondrial activity. In low glucose, increased mitochondrial Pi transporters and lower glycolytic flux increases mitochondrial Pi, leading to enhanced mitochondrial activity. In ubp3Δ cells in high glucose, high trehalose synthesis and lower glycolytic flux results in an increase in Pi. This increases mitochondrial Pi availability and thereby the mitochondrial activity.

Data information: *p<0.05, **p<0.01, ***p<0.001.

+ + +

We next asked whether just adding external Pi was sufficient to increase mitochondrial activity, when cells are in high glucose. In medium supplemented with excess Pi, the internal Pi increases as seen earlier (Figure 3C). Therefore, a simplistic assumption would be that the addition of external Pi to cells in high glucose would also increase mitochondrial Pi. However, an alternate possibility presents itself whererin since glycolytic flux is already high in glucose, supplementing Pi will continue to fuel glycolysis. Indeed, this was originally obseved by Harden and Young in 1908, where adding Pi increased fermentation (44). In such a scenario, mitochondrial Pi will not increase. We estimated mitochondrial Pi in this condition where excess Pi was externally supplemented. Notably, cells grown in high Pi had decreased mitochondrial Pi (Figure S5C). Furthermore, directly adding Pi to cells growing in high glucose decreased basal OCR (Figure S5D), consistent with decreased mitochondrial Pi. These data indicate that in high glucose, simply supplementing Pi will not increase Pi access to the mitochondria. We therefore now asked, if we inhibit glycolytic flux and then supplement Pi, what would happen to mitochondrial activity. For this, we treated cells with 2DG, and subsequently added Pi and measured the OCR (Figure S5E). In this case, supplementing Pi increased the basal OCR (Figure S5E). Collectively, these data suggest that a combination of decreasing glycolysis and increasing Pi can together increase respiration. Next, in order to directly test mitochondrial activation based on external Pi availability, we isolated mitochondria, and estimated activity in vitro upon adding increasing Pi. Mitochondrial activity increased with increased Pi, with maximum activity observed with 25 mM Pi supplemented (Figure 5D). In a complementary experiment, we overexpressed the Mir1 transporter in wild-type cells, to increase Pi within mitochondria (Figure S5F). Mir1-OE cells have higher Cox2 levels and basal OCR (Figure 5E, 5F). Therefore, increasing Pi transport to mitochondria is sufficient to increase mitochondrial activity in high glucose.

+

Mitochondrial pyruvate transport is also required for mitochondrial respiration (45). We asked where Pi availability stands in a hierarchy of constraints for mitochondrial derepression, as compared to mitochondrial pyruvate transport. We measured the amounts of the Mpc3 subunit of the mitochondrial pyruvate carrier (MPC) complex (45, 46). Mpc3 protein increases in ubp3Δ (Figure S5G), correlating with the increased mitochondrial activity. Interestingly, in ubp3Δ grown in low Pi, Mpc3 further increased (Figure S5G), but as shown earlier this condition cannot increase OCR or mitochondrial activity (Figure 4F, S4F). Basal Mpc3 levels decrease in mir1Δ, but upon shifting to 0.1% glucose, Mpc3 increases in both WT and mir1Δ, with higher levels in mir1Δ (Figure S5H). Therefore, even where Mpc3 is high (ubp3Δ in low Pi, and mir1Δ in low glucose), mitochondrial activity remains low if Pi is restricted (Figure 4F, 4H). There was also no decrease in basal OCR in mpc3Δ in high glucose, and the basal OCR increased to the same level as of WT after shifting to 0.1% glucose (Figure S5I). Since Mpc3 changes with mitochondrial Pi availability (Figure S5G, S5H), we also measured Mpc3 in the Mir1OE. No further changes in Mpc3 were observed in Mir1OE (Figure S5J), indicating that increasing mitochondrial Pi alone need not increase Mpc3. Overall, although Mpc3 levels corelate with decreased glycolysis (ubp3Δ -Figure S5G, ubp3Δ in low Pi-Figure S5G, low glucose -Figure S5H), increased Mpc3 alone cannot increase mitochondrial activity and respiration in the absence of adequate mitochondrial Pi.

+

Finally, we asked how important mitochondrial Pi transport was for growth, under 2DG treatment. In the presence of 2DG, WT cells have only slightly decreased growth. In contrast, mir1Δ show a severe growth defect upon 2DG treatment (Figure 5G), revealing a synergetic effect of combining 2DG with inhibiting mitochondrial Pi transport. Therefore, the combined inhibition of glycolysis and mitochondrial Pi transport restricts the growth of glycolytic cells.

+

Collectively, mitochondrial Pi availability constrains glucose-mediated mitochondrial repression. Increasing available pools of Pi to enter the mitochondria is sufficient to induce mitochondrial activity.

+ + + +Discussion +

In this study, we highlight a role for Pi budgeting between cytosolic glycolysis and mitochondrial processes (which compete for Pi) in constraining mitochondrial repression (Figure 5H). Ubp3 controls this process (Figure 1), by maintaining the amounts of the glycolytic enzymes Pfk1 and GAPDH (Tdh2 and Tdh3) and thereby allows high glycolytic flux. At high fermentation rates, glycolytic enzymes levels at maximal activity maintain high glycolytic rates (effectively following zero-order kinetics), and therefore, changes in the enzyme levels will have a direct, proportionate effect on flux (47). The loss of Ubp3 decreases glycolytic flux, resulting in a systems-level, mass-action based rewiring of glucose metabolism where more G6P is routed towards trehalose synthesis and PPP (Figure 2, Figure 5H). Indeed, inhibiting just phosphofructokinase can reroute glucose flux from glycolysis to PPP (48–50), and our study now permits contextualized interpretations of these results. Such reallocations of glucose flux will collectively increase overall Pi, coming from the combined effect of increased Pi release from trehalose synthesis, and decreased Pi consumption via reduced GAPDH. This altered intracellular Pi economy increases Pi pools available to mitochondria, and increasing mitochondrial Pi is necessary and sufficient to increase respiration in high glucose (Figure 4, Figure 5). Mitochondrial Pi transport maintains mitochondrial activity in high glucose, and increases mitochondrial activity in low glucose (Figure 4, Figure 5). Finally, the mitochondrial Pi transporter Mir1 itself decreases in high glucose (Figure 5). Therefore, this glucose-dependent repression of Mir1 also restricts mitochondrial Pi availability (and thereby activity) in high glucose.

+

Traditionally, loss-of-function mutants of metabolic enzymes are used to understand metabolic state regulation. This approach negates nuanced investigations, since metabolic enzymes are often essential for viability. Furthermore, metabolic pathways have multiple, contextually regulated nodes, through which cells maintain their metabolic state. Therefore, alternate approaches to identify global regulators of metabolic states (as opposed to single enzymes), might uncover ways via which multiple nodes are simultaneously tuned, and can reveal unanticipated systems-level principles of metabolic state rewiring. In this study of glucose-mediated mitochondrial repression, the loss of the DUB Ubp3 decreases glycolytic flux by reducing the enzymes at two critical nodes in the pathway - Pfk1 and GAPDH. Unlike loss of function mutants, a reduction in amounts will only rewire metabolic flux. By ‘hitting’ multiple steps in glycolysis simultaneously, ubp3Δ have decreased Pi consumption, as well as increased Pi release. Such a cumulative phenomenon reveals more than inhibiting only GAPDH, where increased Pi comes only from reduced Pi consumption, and not from increased trehalose biosynthesis. Our serendipitous identification of a regulator which regulates multiple steps in glucose metabolism to change the metabolic environment, now suggests a general basis of mitochondrial regulation that would have otherwise remained hidden. Separately, finding the substrates of Ubp3 and whether Ubp3 directly regulates glycolytic enzymes are exciting future research questions requiring concurrent innovations in accessible chemical-biological approaches to study DUBs.

+

Because phosphates are ubiquitous, it is challenging to identify hierarchies of Pi-dependent processes in metabolic state regulation (33). Phosphate transfer reactions are the foundation of metabolism, driving multiple, thermodynamically unfavorable reactions (51, 52). Contextually, the laws of mass action predict that the relative rates of these reactions will regulate overall Pi balance, and contrarily the Pi allocation to Pi-dependent reactions will determine reaction rates (33, 35). Additionally, cells might control Pi allocations for different reactions via compartmentalizing Pi in organelles, to spatially restrict Pi availability (53–55). Our data collectively suggest a paradigm where the combination of factors regulates mitochondrial Pi and thereby activity. In glycolytic yeast cells growing in high glucose, mitochondrial Pi availability becomes restricted due to higher utilization of Pi in glycolysis compared to mitochondria. Consistent with this, a rapid decrease in Pi upon glucose addition has been observed (18, 19, 37). Interesting, in vitro studies with isolated mitochondria from tumor cells also find that decreasing Pi levels decreases respiration (19), which would be consistent with this scenario. Further, supplementing Pi correlates with decreased mitochondrial repression in tumors (18, 56). By increasing Pi through a systems-level rewiring of glucose metabolism (such as in ubp3Δ cells), cells can collectively increase mitochondrial access to Pi. This Pi budgeting determines mitochondrial activity. Supplementing Pi under conditions of low glycolysis (where mitochondrial Pi transport is enhanced), as well as directly supplementing Pi to isolated mitochondria, increases respiration (Figure 5, S5). Therefore, in order to deprepress mitochondria, a combination of increased Pi along with decreased glycolysis is required. An additional systems-level phenomenon that might regulate Pi transport to the mitochondria is the decrease in cytosolic pH upon decreased glycolysis (57, 58). Since mitochondrial Pi transport itself dependeds on the proton gradient, a low cytosolic pH would favour mitochondrial Pi transport (59). Therefore, decreasing glycolysis (2DG treatment, deletion of Ubp3, and decreased GAPDH activity), and increasing cytosolic Pi might synergestically increase mitochondria Pi transport, and increase respiration. Alternately, increasing mitochondrial Pi transporter amounts can achieve the same result, as seen by overexpressing Mir1 (Figure 5). A similar observation has been reported in Arabidopsis, reiterating an evolutionarily conserved role for mitochondrial Pi in controlling respiration (60). Relatedly, glycolytic inhibition can suppress cell proliferation in Warburg-positive tumors (61, 62), or inflammatory responses (63). However, these cells survive by switching to mitochondrial respiration (64, 65), requiring alternate approaches to prevent their proliferation (66). Inhibiting mitochondrial Pi transport in combination with glycolytic inhibition could restrict the proliferation of Warburg/Crabtree positive cells.

+

Since its discovery in the 1920s, the phenomenon of accelerated glycolysis with concurrent mitochondrial repression has been intensely researched. Yet, the biochemical constraints for glucose-mediated mitochondrial repression remains unresolved. One hypothesis suggests that the availability of glycolytic intermediates might determine the extent of mitochondrial repression. F1,6BP inhibits complex III and IV of the electron transport chain (ETC) in Crabtree positive yeast (14–17). Similarly, the ratio between G6P and F1,6BP regulates the extent of mitochondrial repression (16, 17). Although G6P/F6P accumulates in ubp3Δ (Figure 2C), this is not the case in tdh2Δtdh3Δ (GAPDH mutant) (Figure S3D), suggesting that G6P/F6P accumulation in itself is not the criterion to increase mitochondrial activity. Separately, the competition for common metabolites/co-factors between glycolysis and respiration (such as ADP, Pi or pyruvate) could drive this phenomenon (14, 18). Here we observe that Mpc3 mediated mitochondrial pyruvate transport alone cannot increase respiration. An additional consideration is the possible contribution of changes in ADP in regulating mitochondrial activity, where the use of ADP in glycolysis might limit mitochondrial ADP. Therefore, when Pi changes as a consequence of glycolysis, it coud be imagined that a change in ADP balance can coincidentally occur. However, prior studies show that even though cytosolic ADP decreases in the presence of glucose, this does not limit mitochondrial ADP uptake, or decrease respiration, due to the very high affinity of the mitochondrial ADP transporter (14, 19). These collectively reiterate the importance of Pi access and transport to mitochondria in constraining mitochondrial respiration. Indeed, this interpretation can also contextually explain observations from other model systems where mitochondrial Pi transport seems to regulate respiration (67, 68).

+

We parsimoniously suggest that Pi access to the mitochondria as a key constraint for mitochondrial repression under high glucose. In a hypothetical scenario, a single-step event in evolution, reducing mitochondrial Pi transporter amounts, and/or increasing glycolytic flux (to deplete cytosolic Pi), will result in whole-scale metabolic rewiring to repress mitochondria. More elaborate regulatory events can easily be imagined as subsequent adaptations to enforce mitochondrial repression. Given the central role played by Pi, something as fundamental as access to Pi will constrain mitochondrial repression. Concurrently, the rapid incorporation of Pi into faster glycolysis can give cells a competitive advantage, while also sequestering Pi in the form of usable ATP. Over the course of evolution, this could conceivably drive other regulatory mechanisms to enforce mitochondrial repression, leading to the currently observed complex regulatory networks and signaling programs observed in the Crabtree effect, and other examples of glucose-dependent mitochondrial repression.

+ + +Materials and methods + +Statistics and graphing +

Unless otherwise indicated, statistical significance for all indicated experiments were calculated using unpaired Student’s t-tests (GraphPad prism 9.0.1). Graphs were plotted using GraphPad prism 9.0.1

+ + +Yeast strains, media, and growth conditions +

A prototrophic CEN.PK strain of Saccharomyces cerevisiae (WT) (69) was used. Strains are listed in appendix table S1. Gene deletions, chromosomal C-terminal tagged strains were generated by PCR mediated gene deletion/tagging (70). Media compositions, growth conditions and CRISPR-Cas9 based mutagenesis are described in extended methods (SI appendix).

+ + +Mitotracker fluorescence +

Mitotracker fluorescence was measured using Thermo VarioscanTM LUX multimode plate reader (579/599 excitation/emission). Detailed protocol is described in extended methods. Mitotracker fluorescence were normalized using OD600 of each sample and relative fluorescence intensity calculated.

+ + +Protein extraction and Western blotting +

Total protein was precipitated, extracted using trichloroacetic acid (TCA) as described earlier (71). Blots were quantified using ImageJ software. Detailed protocol is described in extended methods.

+ + +Basal OCR measurement +

The basal OCR was measured using Agilent Seahorse XFe24 analyzer. Basal OCR readings were normalized for cell number (using OD600 of samples) in each well. The detailed methods are described in extended methods.

+ + +RNA extraction and RT-qPCR +

The RNA extraction was done using the hot phenol extraction method as described in (71). The isolated RNA was DNase treated, and used for cDNA synthesis. Superscript III reverse transcriptase enzyme (Invitrogen) was used for cDNA synthesis and RT-qPCR was performed using KAPA SYBR FAST qRT PCR kit (KK4602, KAPA Biosystems). Taf10 was used as a control for normalisation and the fold change in mRNA levels were calculated by 2^-ddct method.

+ + +ATP, ethanol and Pi measurements +

ATP levels were measured by ATP estimation kit (Thermo Fisher A22066). Ethanol concentration in the medium was estimated using potassium dichromate based assay described in (72) with modifications. Pi was estimated using a malachite green phosphate assay kit (Cayman chemicals, 10009325). Detailed sample collection and assay protocols are described in extended methods.

+ + +Metabolite extraction and analysis by LC-MS/MS +

The steady state levels and relative 13C label incorporation into metabolites were estimated by quantitative LC-MS/MS methods as described in (73). Detailed methodology is extensively described in extended methods. Peak area measurements are listed in supplement file 1.

+ + +Mitochondrial isolation +

Mitochondria was isolated by immunoprecipitation as described in (74, 75) with modifications. The detailed protocol is described in extended methods (SI appendix). The eluted mitochondria were used in malachite green assay for Pi estimation, boiled with SDS-glycerol buffer for western blots or incubated with mitotracker CMXROS with mitochondria-activation buffer for mitotracker assays.

+ + + + + +Data availability +

The LC-MS/MS peak area for all the metabolites can be found in supplement file 1. This study includes no data deposited in external repositories.

+ + +Acknowledgements +

We thank all SL lab members, Benjamin Tu, Anand Bachhawat and Vijay Jayaraman for comments and suggestions. We acknowledge the NCBS/inStem/CCAMP mass spectrometry facility for LC-MS/MS support. We thank Aayushee Khanna for help with experiments. V.V acknowledges funding support from DST-INSPIRE fellowship (IF170236) from the Department of Science and Technology (DST), Govt. of India, S.N and SA acknowledges intramural funding from the DBT-inStem PhD program. S.L acknowledges support from the DST SERB CRG grant CRG/2019/004772, a DBT-Wellcome India Alliance Senior Fellowship (IA/S/21/2/505922), and institutional support from inStem.

+ + +References +J. Zhu, C. B. Thompson, Metabolic regulation of cell growth and proliferation. Nat. Rev. Mol. Cell Biol. 20, 436–450 (2019). +L. Cai, B. P. Tu, Driving the Cell Cycle Through Metabolism. Annu. Rev. Cell Dev. Biol. 28, 59–87 (2011). +D. L. (David L. Nelson, D. L. (David L. Nelson, A. L. Lehninger, M. M. Cox, Lehninger principles of biochemistry (2008) (March 10, 2022). +O. Warburg, The metabolism of carcinoma cells 1. J. Cancer Res. 9, 148–163 (1925). +M. G. Vander Heiden, L. C. Cantley, C. B. Thompson, Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation. Science (80-.). 324, 1029–1033 (2009). +H. G. Crabtree, Observations on the carbohydrate metabolism of tumours. Biochem. J. 23, 536–545 (1929). +R. H. De Deken, The Crabtree effect: a regulatory system in yeast. J. Gen. Microbiol. 44, 149–156 (1966). +L. Xia, et al., The cancer metabolic reprogramming and immune response. Mol. Cancer 20, 1–21 (2021). +M. D. Kornberg, The immunologic Warburg effect: Evidence and therapeutic opportunities in autoimmunity. Wiley Interdiscip. Rev. Syst. Biol. Med. 12, 1–17 (2020). +N. Pacini, F. Borziani, Cancer stem cell theory and the warburg effect, two sides of the same Coin? Int. J. Mol. Sci. 15, 8893–8930 (2014). +A. M. Abdel-Haleem, et al., The emerging facets of non-cancerous Warburg effect. Front. Endocrinol. (Lausanne). 8, 1–7 (2017). +E. Tsogtbaatar, C. Landin, K. Minter-Dykhouse, C. D. L. Folmes, Energy Metabolism Regulates Stem Cell Pluripotency. Front. Cell Dev. Biol. 8, 1–16 (2020). +A. Cornish-Bowden, Biochemical evolution : the pursuit of perfection. Garl. Sci. (2016). +R. Diaz-Ruiz, M. Rigoulet, A. Devin, The Warburg and Crabtree effects: On the origin of cancer cell energy metabolism and of yeast glucose repression. Biochim. Biophys. Acta - Bioenerg. 1807, 568–576 (2011). +N. Hammad, M. Rosas-Lemus, S. Uribe-Carvajal, M. Rigoulet, A. Devin, The Crabtree and Warburg effects: Do metabolite-induced regulations participate in their induction? Biochim. Biophys. Acta - Bioenerg. 1857, 1139–1146 (2016). +R. Díaz-Ruiz, et al., Mitochondrial Oxidative Phosphorylation Is Regulated by Fructose 1,6-Bisphosphate: A POSSIBLE ROLE IN CRABTREE EFFECT INDUCTION? *. J. Biol. Chem. 283, 26948–26955 (2008). +M. R. Lemus, et al., The role of glycolysis-derived hexose phosphates in the induction of the Crabtree effect. J. Biol. Chem. 293, 12843–12854 (2018). +D. H. Koobs, Phosphate mediation of the Crabtree and Pasteur effects. Science (80-.). 178, 127–133 (1972). +S. Rodríguez-Enríquez, O. Juárez, J. S. Rodríguez-Zavala, R. Moreno-Sánchez, Multisite control of the Crabtree effect in ascites hepatoma cells. Eur. J. Biochem. 268, 2512–2519 (2001). +F. Tripodi, R. Nicastro, V. Reghellin, P. Coccetti, Post-translational modifications on yeast carbon metabolism: Regulatory mechanisms beyond transcriptional control. Biochim. Biophys. Acta - Gen. Subj. 1850, 620–627 (2015). +T. Hitosugi, J. Chen, Post-translational modifications and the Warburg effect. Oncogene 33, 4279–4285 (2014). +J. R. Broach, Nutritional control of growth and development in yeast. Genetics 192, 73–105 (2012). +A. Hershko, A. Ciechanover, THE UBIQUITIN SYSTEM. Annu. Rev. Biochem. 67, 425–79 (1998). +D. Komander, M. Rape, The Ubiquitin Code. Annu. Rev. Biochem. 81, 203–229 (2012). +C. M. Pickart, M. J. Eddins, Ubiquitin: Structures, functions, mechanisms. Biochim. Biophys. Acta - Mol. Cell Res. 1695, 55–72 (2004). +E. Postma, C. Verduyn, W. A. Scheffers, J. P. Van Dijken, Enzymic analysis of the crabtree effect in glucose-limited chemostat cultures of Saccharomyces cerevisiae. Appl. Environ. Microbiol. 55, 468–77 (1989). +B. Ossareh-Nazari, et al., Cdc48 and Ufd3, new partners of the ubiquitin protease Ubp3, are required for ribophagy. EMBO Rep. 11, 548–554 (2010). +M. Isasa, et al., Multiplexed, Proteome-Wide Protein Expression Profiling: Yeast Deubiquitylating Enzyme Knockout Strains. J. Proteome Res. 14, 5306–5317 (2015). +R. Nostramo, S. N. Varia, B. Zhang, M. M. Emerson, P. K. Herman, The Catalytic Activity of the Ubp3 Deubiquitinating Protease Is Required for Efficient Stress Granule Assembly in Saccharomyces cerevisiae. Mol. Cell. Biol. 36, 173 (2016). +F. Fontanesi, I. C. Soto, D. Horn, A. Barrientos, Assembly of mitochondrial cytochrome c-oxidase, a complicated and highly regulated cellular process. Am. J. Physiol. - Cell Physiol. 291, 1129–1147 (2006). +A. A. Shestov, et al., Quantitative determinants of aerobic glycolysis identify flux through the enzyme GAPDH as a limiting step. Elife 3, 1–18 (2014). +R. Gupta, et al., A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis. Elife 8, 1–33 (2019). +R. Gupta, S. Laxman, Cycles, sources, and sinks: Conceptualizing how phosphate balance modulates carbon flux using yeast metabolic networks. Elife 10, 1–14 (2021). +P. W. Mason, D. P. Carbone, R. A. Cushman, A. S. Waggoner, The importance of inorganic phosphate in regulation of energy metabolism of Streptococcus lactis. J. Biol. Chem. 256, 1861–1866 (1981). +J. H. Van Heerden, et al., Lost in transition: Start-up of glycolysis yields subpopulations of nongrowing cells. Science (80-.). 343 (2014). +P. C. Brazy, L. J. Mandel, V. W. Dennis, S. R. Gullans, Metabolic Requirement for Inorganic Phosphate by the Rabbit Proximal Tubule: EVIDENCE FOR A CRABTREE EFFECT Metabolic Requirement for Inorganic Phosphate by the Rabbit Proximal Tubule EVIDENCE FOR A CRABTREE EFFECT. J Clin Invest 70, 53 (1982). +S. Hohmann, W. Bell, M. J. Neves, D. Valckx, J. M. Thevelein, Evidence for trehalose-6-phosphate-dependent and-independent mechanisms in the control of sugar influx into yeast glycolysis. Mol. Microbiol. 20, 981–991 (1996). +T. Hunter, Why nature chose phosphate to modify proteins. Philos. Trans. R. Soc. B Biol. Sci. 367, 2513–2516 (2012). +P. O. Ljungdahl, B. Daignan-Fornier, Regulation of amino acid, nucleotide, and phosphate metabolism in Saccharomyces cerevisiae. Genetics 190, 885–929 (2012). +S. Austin, A. Mayer, Phosphate Homeostasis − A Vital Metabolic Equilibrium Maintained Through the INPHORS Signaling Pathway. Front. Microbiol. 11, 1367 (2020). +J. M. Mouillon, B. L. Persson, New aspects on phosphate sensing and signalling in Saccharomyces cerevisiae. FEMS Yeast Res. 6, 171–176 (2006). +H. Murakami, G. Blobel, D. Pain, Isolation and characterization of the gene for a yeast mitochondrial import receptor. Nat. 1990 3476292 347, 488–491 (1990). +V. Zara, et al., Yeast mitochondria lacking the phosphate carrier/p32 are blocked in phosphate transport but can import preproteins after regeneration of a membrane potential. Mol. Cell. Biol. 16, 6524–6531 (1996). +A. Harden, W. J. Young, The alcoholic Ferment of Yeast-juice.Part III.-The Function of Phosphates in the Fermentation o f Glucose hy Yeast-juice. Proc. R. Soc. London. Ser. B, 80, 299–311 (1908). +A. Timoń-Gómez, M. Proft, A. Pascual-Ahuir, Differential Regulation of Mitochondrial Pyruvate Carrier Genes Modulates Respiratory Capacity and Stress Tolerance in Yeast. PLoS One 8, e79405 (2013). +T. Bender, G. Pena, J.-C. Martinou, Regulation of mitochondrial pyruvate uptake by alternative pyruvate carrier complexes. EMBO J. 34, 911–924 (2015). +P. Grigaitis, B. Teusink, An excess of glycolytic enzymes under glucose-limited conditions may enable Saccharomyces cerevisiae to adapt to nutrient availability. FEBS Lett. (2022) 10.1002/1873-3468.14484 (November 28, 2022). +W. Yi, et al., PFK1 Glycosylation Is a Key Regulator of Cancer Cell Growth and Central Metabolic Pathways. Science 337, 975 (2012). +H. Miyazawa, et al., Rewiring of embryonic glucose metabolism via suppression of PFK-1 and aldolase during mouse chorioallantoic branching. Dev. 144, 63–73 (2017). +W. D. Hollinshead, et al., Examining Escherichia coli glycolytic pathways, catabolite repression, and metabolite channeling using Δpfk mutants. Biotechnol. Biofuels 9, 1–13 (2016). +F. H. Westheimer, Why Nature Chose Phosphates. Science (80-.). 235, 1173–1178 (1987). +S. C. L. Kamerlin, P. K. Sharma, R. B. Prasad, A. Warshel, Why nature really chose phosphate. Q. Rev. Biophys. 46, 1–132 (2013). +J. W. Booth, G. Guidotti, Phosphate Transport in Yeast Vacuoles *. J. Biol. Chem. 272, 20408–20413 (1997). +T. Vila, S. Frases, F. M. Gomesid, Lessons from protozoans: Phosphate sensing and polyphosphate storage in fungi. PLOS Pathog. 18, e1010298 (2022). +M. E. Solesio, et al., Depletion of mitochondrial inorganic polyphosphate (polyP) in mammalian cells causes metabolic shift from oxidative phosphorylation to glycolysis. Biochem. J. 478, 1631–1646 (2021). +M. Brin, R. W. McKee, The Inhibition of Respiration by Glucose, Fructose, and Mannose in the Ehrlich Mouse Ascites Tumor. Cancer Res. 16, 364–368 (1956). +R. Orij, S. Brul, G. J. Smits, Intracellular pH is a tightly controlled signal in yeast. Biochim. Biophys. Acta - Gen. Subj. 1810, 933–944 (2011). +R. Dechant, et al., Cytosolic pH is a second messenger for glucose and regulates the PKA pathway through V-ATPase. EMBO J. 29, 2515–2526 (2010). +P. Hamel, et al., Redundancy in the function of mitochondrial phosphate transport in Saccharomyces cerevisiae and Arabidopsis thaliana. Mol. Microbiol. 51, 307–317 (2004). +F. Jia, et al., Overexpression of Mitochondrial Phosphate Transporter 3 Severely Hampers Plant Development through Regulating Mitochondrial Function in Arabidopsis. PLoS One 10, e0129717 (2015). +H.Pelicano, D. S.Martin, R.Xu, P.Huang, Glycolysis inhibition for anticancer treatment. Oncogene. 4633–4646 (2006). +S. O’Neill, R. K. Porter, N. McNamee, V. G. Martinez, L. O’Driscoll, 2-Deoxy-D-Glucose inhibits aggressive triple-negative breast cancer cells by targeting glycolysis and the cancer stem cell phenotype. Sci. Rep. 9, 1–11 (2019). +G. Soto-Heredero, M. M. Gómez de las Heras, E. Gabandé-Rodríguez, J. Oller, M. Mittelbrunn, Glycolysis – a key player in the inflammatory response. FEBS J. 287, 3350–3369 (2020). +R. Shiratori, et al., Glycolytic suppression dramatically changes the intracellular metabolic profile of multiple cancer cell lines in a mitochondrial metabolism-dependent manner. Sci. Rep. 9, 1–15 (2019). +C. L. Lu, et al., Tumor Cells Switch to Mitochondrial Oxidative Phosphorylation under Radiation via mTOR-Mediated Hexokinase II Inhibition - A Warburg-Reversing Effect. PLoS One 10, e0121046 (2015). +G. Cheng, et al., Mitochondria-targeted drugs synergize with 2-deoxyglucose to trigger breast cancer cell death. Cancer Res. 72, 2634–2644 (2012). +M. Scheibye-Knudsen, B. Quistorff, Regulation of mitochondrial respiration by inorganic phosphate; comparing permeabilized muscle fibers and isolated mitochondria prepared from type-1 and type-2 rat skeletal muscle. Eur. J. Appl. Physiol. 105, 279–287 (2009). +E. L. Seifert, E. Ligeti, J. A. Mayr, N. Sondheimer, G. Hajnóczky, The mitochondrial phosphate carrier: Role in oxidative metabolism, calcium handling and mitochondrial disease. Biochem. Biophys. Res. Commun. 464, 369–375 (2015). +J. B. J.P. van Dijken, et al., An interlaboratory comparison of physiological and genetic properties of four Saccharomyces cerevisiae strains. Enzyme Microb. Technol. 26, 706–714 (2000). +M. S. Longtine, et al., Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast 14, 953–961 (1998). +V. Vengayil, Z. Rashida, S. Laxman, The E3 ubiquitin ligase Pib1 regulates effective gluconeogenic shutdown upon glucose availability. J. Biol. Chem. 294, 17209–17223 (2019). +M. Sriariyanun, P. Mutrakulcharoen, S. Tepaamorndech, K. Cheenkachorn, K. Rattanaporn, A Rapid Spectrophotometric Method for Quantitative Determination of Ethanol in Fermentation Products. Orient. J. Chem. 35, 744–750 (2019). +A.Walvekar, Z.Rashida, H.Maddali, S.Laxman, A versatile LC-MS / MS approach for comprehensive, quantitative analysis of central metabolic pathways [version 1; peer review : 2 approved]. Wellcome Open Res1–15 (2018). +P. C. Liao, I. R. Boldogh, S. E. Siegmund, Z. Freyberg, L. A. Pon, Isolation of mitochondria from Saccharomyces cerevisiae using magnetic bead affinity purification. PLoS One 13, 1–15 (2018). +W. W. Chen, E. Freinkman, D. M. Sabatini, Rapid immunopurification of mitochondria for metabolite profiling and absolute quantification of matrix metabolites. Nat. Protoc. 12, 2215 (2017). + + +
diff --git a/test/2023.02.04.527146/fix-author-emails/2023.02.04.527146.xml b/test/2023.02.04.527146/fix-author-emails/2023.02.04.527146.xml new file mode 100644 index 00000000..52002a80 --- /dev/null +++ b/test/2023.02.04.527146/fix-author-emails/2023.02.04.527146.xml @@ -0,0 +1,338 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.02.04.527146 +1.3 + + +Regular Article + + +New Results + + +Neuroscience + + + +Effort cost of harvest affects decisions and movement vigor of marmosets during foraging + + +http://orcid.org/0000-0002-7006-9877HagePaulpaulhage@jhu.edu1 +http://orcid.org/0000-0002-1832-3743JangIn Kyu1 +http://orcid.org/0000-0002-9392-4178LooiVivian1 +http://orcid.org/0000-0002-7741-8689FakharianMohammad Amin1 +http://orcid.org/0000-0001-6092-7228OrozcoSimon P.1 +http://orcid.org/0000-0003-4036-2877PiJay S.1 +http://orcid.org/0000-0001-6732-1296Sedaghat-NejadEhsan1 +http://orcid.org/0000-0002-7686-2569ShadmehrRezashadmehr@jhu.edu1 +Laboratory for Computational Motor Control, Dept. of Biomedical Engineering Johns Hopkins School of Medicine, Baltimore, Maryland USA + + +

Author contributions: P.H., S.P.O., E.S., J.S.P. and R.S. conceived and performed experiments. P.H., I.K.J., V.L., M.A.F., E.S., and J.S.P. analyzed data, P.H. made figures and performed statistical analysis. R.S. and P.H. wrote the manuscript.

 +

Competing interests: None.

 +

Data availability: https://osf.io/54JS6

 + +2023 +2023.02.04.527146 + + +04 +2 +2023 + + +19 +7 +2023 + + +19 +7 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. + + + +Abstract

Our decisions are guided by how we perceive the value of an option, but this evaluation also affects how we move to acquire that option. Why should economic variables such as reward and effort alter the vigor of our movements? In theory, both the option that we choose and the vigor with which we move contribute to a measure of fitness in which the objective is to maximize rewards minus efforts, divided by time. To explore this idea, we engaged marmosets in a foraging task in which on each trial they decided whether to work by making saccades to visual targets, thus accumulating food, or to harvest by licking what they had earned. We varied the effort cost of harvest by moving the food tube with respect to the mouth. Theory predicted that the subjects should respond to the increased effort costs by choosing to work longer, stockpiling food before commencing harvest, but reduce their movement vigor to conserve energy. Indeed, in response to an increased effort cost of harvest, marmosets extended their work duration, but slowed their movements. These changes in decisions and movements coincided with changes in pupil size. As the effort cost of harvest declined, work duration decreased, the pupils dilated, and the vigor of licks and saccades increased. Thus, when acquisition of reward became effortful, the pupils constricted, the decisions exhibited delayed gratification, and the movements displayed reduced vigor.

+ +Significance statement +

Our results suggest that as the brainstem neuromodulatory circuits that control pupil size respond to effort costs, they alter computations in the brain regions that control decisions, encouraging work and delaying gratification, and the brain regions that control movements, reducing vigor and suppressing energy expenditure. This coordinated response suggests that decisions and actions are part of a single control policy that aims to maximize a variable relevant to fitness: the capture rate.

+ + + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Revisions based on feedback from eLife reviews.

 + + +

+https://osf.io/54JS6 +

 + + + + + +Introduction +

During foraging, animals work to locate a food cache and then spend effort harvesting what they have found. As they forage, their decisions appear to maximize a measure that is relevant to fitness: the sum of rewards acquired, minus efforts expended, divided by time, termed the capture rate (1–3). For example, a crow will spend effort extracting a clam from a sandy beach, but if the clam is small, it will abandon it because the additional time and effort required to extract the small reward --dropping it repeatedly from a height onto rocks -- can be better spent finding a bigger prize (4). In other words, if going to the bank entails waiting in a long line, one should go infrequently, but make each transaction a large amount.

+

Intriguingly, reward expectation not only affects decisions, it also affects movements: we not only prefer the less effortful option, we also move vigorously to obtain it (5, 6). This modulation of movement vigor can be justified if we consider that movements require expenditure of time and energy, which discount the value of the promised reward (7, 8). Thus, from a theoretical framework, it seems rational that the brain should have a mechanism to coordinate control of decisions with control of movements so that both contribute to maximizing a measure of fitness (9).

+

To study this coordination, we designed a task in which marmosets decided how long to work before they harvested their food. On a given trial, they made a sequence of saccades to visual targets and received an increment of food as their reward. However, the increment was small, and its harvest was effortful, requiring inserting their tongue inside a small tube. Theory predicted that in order to maximize the capture rate, harvest should commence only when there was sufficient reward accumulated to justify the effort required for its extraction. Indeed, the subjects chose to work in order to stockpile food, and only then initiated their harvest.

+

On some days the effort cost of harvest was low: the tube was placed close to the mouth. On other days the same amount of work, i.e., saccade trials, produced food that had a higher effort cost: the tube was located farther away. The theory made two interesting predictions: as the effort cost increased, the subjects should choose to work more trials, thus delaying their harvest so to stow more food, but reduce their movement vigor, thus saving energy. Indeed, when marmosets encountered an increased effort cost, they extended their work period, stockpiling food, but reduced their vigor, slowing their saccades during the work period, and slowing their licks during the harvest period.

+

What might be a neural basis for this coordinated response of the decision-making and the motor-control circuits? During the work and the harvest periods, momentary changes in pupil size closely tracked the changes in vigor: pupil dilation accompanied increases in vigor, while pupil constriction accompanied decreases in vigor. Remarkably, this was true regardless of whether the movement that was being performed was a saccade, or a lick. Moreover, in response to the increased effort cost, the pupils exhibited a global change, constricting during both the work and the harvest periods.

+

If we view the changes in pupil size as a proxy for activity in the brainstem noradrenergic circuits (10), our results suggest that as these circuits respond to effort costs (11), they alter computations in the brain regions that control decisions, delaying gratification and encouraging work, and the brain regions that control movements, promoting sloth and conserving energy.

+ + +Results +

We tracked the eyes and the tongue of head-fixed marmosets as they performed visually guided saccades in exchange for food (Fig. 1A). Each successful trial consisted of 3 visually guided saccades, at the end of which we delivered an increment of food (a slurry mixture of apple sauce and monkey chow). Because the reward amount was small (0.015-0.02 mL), the subjects rarely harvested following a single successful trial. Rather, they worked for a few trials, allowing the food to accumulate, then initiated their harvest by licking (Fig. 1B). The key variables were how many trials they chose to work before starting harvest, and how vigorously they moved their eyes and tongue during the work and the harvest periods.

+ + +

Elements of a foraging task. A. During the work period, marmosets made a sequence of saccades to visual targets. A trial consisted of 3 consecutive saccades, at the end of which the subject was rewarded by a small increment of food. We tracked the eyes, the tongue, and the food. B. An example of two consecutive work-harvest periods, showing reward-relevant saccades (eye velocity) and tongue endpoint displacement with respect to the mouth. C. Data for two sessions, one where the tube was placed close to the mouth (orange trace), and one where it was placed farther away (red trace). Two types of licks are shown: inner-tube licks, and outer-tube licks. Depending on food location, both types of licks can contact the food. Data on the right two panels show endpoint displacement and velocity of the tongue during inner-tube licks. Error bars are SEM. D. During the work period, the subjects attempted about 8 trials on average, succeeding in 4-5 trials before starting harvest, and then licked about 18 times to extract the food.

+ + +

Over the course of 2.5 years, we recorded 56 sessions in subject M (29 months) and 56 sessions in subject R (23 months). A typical work period lasted about 10 seconds, during which the subjects attempted ∼8 trials, and succeeded in 4-5 trials (Fig. 1D) (a successful trial was when all three saccades were within 1.25 of the center of each target). The work period ended when the subject decided to stop tracking the targets and instead initiated harvest, which lasted about 6 seconds, resulting in 16-18 licks. Subject M completed an average of 909.5 ± 61 successful trials per session (mean ± SEM), producing an average of 241 ± 13.9 work-harvest pairs, and subject R completed an average of 1431 ± 65 successful trials, producing an average of 263 ± 8.9 work-harvest pairs.

+

We delivered food via either the left or the right tube for 50-300 consecutive trials, and then switched tubes. We tracked the motion of the tongue using DeepLabCut (12), as shown for a typical session in Fig. 1B. The licks required precision because the tube was just large enough (4.4 mm diameter) to allow the tongue to penetrate. As a result, about 30% of the of the reward-seeking licks were successful and contacted food (30±1.6% for subject M, 28±2.5% for subject R), as shown in Supporting Information Video 1. Example of licks that failed to contact food are shown in Supporting Information Video 2, Video 3, and Video 4.

+ +Theory and predictions +

During the decision-making part of the task, the brain explicitly determined how long to work before initiating harvest. During the work and the harvest periods, the brain implicitly controlled the vigor of movements. We imagined that these two forms of behavior were not independent, but rather coordinated via a control policy that maximized a single utility: the sum of rewards acquired, minus efforts expended, divided by time, termed the capture rate. We chose this formulation because it presents a normative approach that ecologists have used to understand the decisions that animals make regarding how far to travel for food, what mode of travel to use, and how long to stay before moving on to another patch (4, 13, 14).

+

During a work period, our subjects decided to complete a number of saccade trials ns, a fraction βs of which were successful, earning food increment α, but expended effort cs that consumed time Ts for each trial. They then stopped working and initiated harvest, producing a number of licks nl, a fraction βl of which succeeded, thus expending effort cl and consuming time Tl for each lick. These actions produced the following capture rate: + + + +In the numerator of Eq. (1), the first term represents the fact that the food cache increased linearly with successful trials and was then consumed gradually with successful licks. The second term represents the effort expenditure of licking, and the third term represents the effort expenditure of working. Notably, the effort expenditure of work, grows faster than linearly as a function of trials. This nonlinearity is essential to reflect the idea that following a long work period, the capture rate must be more negative than following a short work period (i.e., more work trials produce a greater reduction in utility).

+

A control policy describes how long to work and harvest, and an optimal policy produces periods of working and harvesting, that maximize Eq. (1). A closed-form solution for the optimal policy can be obtained (Mathematica notebook simulations.nb) and Fig. 2A provides an example. As the work period concludes and the harvest period beings (nl = 0), the capture rate is negative. This reflects the fact that the subject has performed a few trials and stockpiled food, thus expended effort but has not been rewarded yet. The capture rate rises when licking commences. Critically, the peak capture rate is not an increasing function of the work period. Rather, there is an optimal work period ( red trace, Fig. 2A) associated with a given effort cost of licking cl. If we now move the tube away from the mouth, that is, increase the effort cost of licking cl, the peak of the capture rate shifts and the optimal work period changes: the proper response to an increased effort cost of licking is to work longer, stowing more food before commencing harvest.

+ + +

Theoretical results of an optimal control policy. A. Capture rate (Eq. 1) is plotted during the harvest period as a function of lick number following various number of work trials. When the effort cost of licking is low (left plot,), the optimal work period is (red trace). When the effort cost is higher (right plot,), it is best to work longer before initiating harvest. B. The metabolic cost of licking (Eq. 2) is minimized when a lick has a specific duration. Tube distance varied from 0.1 to 0.3. Optimal duration that minimizes lick cost grows linearly with tube distance. C. Optimal number of work trials and licks as a function food tube distance. As the effort cost of harvest increases, one should respond by working longer, delaying harvest. D. Optimal lick duration as a function of food tube distance. The lick duration that maximizes the capture rate is smaller than the one that minimizes the lick metabolic cost (Fig. 2B). That is, it is worthwhile moving vigorously to acquire reward. However, grows faster than linearly as a function of tube distance. Thus, as the tube moves farther, it is best to reduce lick vigor. Hunger, modeled as increased value of reward, should promote work and increase vigor, while effort cost of harvest (tube distance) should promote work but reduce vigor. Parameter values for all simulations:,,,, (low food value, less hunger), (high food value, hungry).

+ + +

Notably, the higher cost of licking inevitably reduces the maximum capture rate (Fig. 2A). This should impact movement vigor: animals tend to respond to a reduced capture rate by slowing their movements (9), which can be viewed as an effective way to save energy (8). To incorporate vigor into the capture rate, we tried to define the effort cost of a single lick cl in terms of its energetic cost, a relationship that is currently unknown. Fortunately, other movements provide a clue: the energetic cost of reaching (8, 15), as well as the energetic cost of walking (16, 17), are both concave upward functions of the movement’s duration. That is, from an energetic standpoint, there is a reach speed, and a walking speed, that minimizes the cost of each type of movement. We generalized these empirical observations to licking and assumed that the energetic cost of a single lick was a concave upward function of its duration: + + + +In Eq. (2), the lick is aimed at a tube located at distance d, and has a duration Tl. The parameter k describes the rate with which the cost grows as a function of duration. For example, the lick duration that minimizes the energetic cost is . Thus, for an energetically optimal lick, duration grows linearly with tube distance (Fig. 2B). However, our objective is not to minimize the cost of licking, but to maximize the capture rate. To do so, we insert Eq. (2) into Eq. (1) and find the optimal policy which now depends on the distance of the food tube to the mouth (Mathematica notebook simulations.nb).

+

The theory predicts that to maximize the capture rate (Eq. 1), the response to an increased effort cost of harvest (i.e., tube distance) should be as follows: should increase (Fig. 2C), should decrease (Fig. 2C), and should increase (Fig. 2D). Notably, the rate of increase in as a function of tube distance is faster than linear, while from an energetic point of view (Eq. 2), increase in distance should produce a linear increase in lick duration. Thus, as the harvest becomes more effortful, the subject should work longer to stockpile food, but move slower to save energy.

+

To test our theory further, we thought it useful to have a way to alter decisions in one direction (say work longer) but change movement vigor in the opposite direction (move faster). In theory, this is possible: if the subject is hungry (darker lines in Fig. 2C and 2D), i.e., the reward is more valuable, then they should again work longer before initiating harvest. Paradoxically, they should also move faster.

+

In summary, if decisions and actions are coordinated via a policy that aims to maximize the capture rate, then in response to an increased cost of harvest, one should work longer, but move with reduced vigor. In response to an increased reward value, as in hunger, one should also work longer, but now move with increased vigor.

+ + +Increased effort cost of harvest promoted work but reduced saccade vigor +

To vary the effort cost of harvest, we altered the tube distance to the mouth (but kept it constant during each session). Varying tube distance affected the decisions of the subjects: when the tube was placed farther, they chose to work longer before starting harvest (Fig. 3A, left subplot): they attempted more trials during each work period (ANOVA, subject M: F(2,7908)=41.5 p=5.2×10-25, subject R: F(2,10948)=88.2 p=7×10-50), and produced more successful trials per work period (ANOVA, subject M: F(2,7908)=63 p=2.8×10-24, subject R: F(2,10948)=163 p<10-50). This policy of delayed gratification was present throughout the recording session (Fig. 3A, middle plot). That is, when the harvest required more effort, the subjects worked longer to stockpile more food before initiating their harvest (Fig. 3A, right plot, effect of tube distance on food cached: subject M: F(2,9566)=176 p<10-50, subject R: F(2,8907)=204 p<10-50).

+ + +

As the effort cost of harvest increased, subjects chose to work more trials, but slowed their movements. A. Left: the number of trials attempted and succeeded per work period as a function of tube distance. Middle: successful trials per work period as a function of time during the recording session. Tube distance is with respect to a marker on the nose. Right: food available in the tube at the start of the harvest. B. Peak saccade velocity as a function of amplitude for reward-relevant and other saccades. C. Vigor of reward relevant saccades as a function of trial number during the work period. Saccade vigor was greater when the tube was closer. Pupil size is quantified during the same work periods. Accuracy is quantified as the magnitude of the saccade’s endpoint error vector (with respect to the target) and the variance of that error vector (determinant of the variance-covariance matrix), plotted as a function of the vigor of the saccade (bin size=0.05 vigor units). Error bars are SEM.

+ + +

During the work period the subjects made saccades to visual targets and accumulated their food. They also made saccades that were not toward visual targets and thus were not eligible for reward. For each animal we computed the relationship between peak saccade velocity and saccade amplitude across all sessions and then calculated the vigor of each saccade: defined as the ratio of the actual peak velocity with respect to the expected peak velocity for that amplitude (18, 19). For example, a saccade that exhibited a vigor of 1.10 had a peak velocity that was 10% greater than the average peak velocity of the saccades of that amplitude for that subject. As expected, the reward-relevant saccades, i.e., saccades made to visual targets (primary, corrective, and center saccades) were more vigorous than other saccades (Fig. 3B, 2-way ANOVA, effect of saccade type, subject M: F (1,391459)=7248 p<10-50, subject R: F (1,355839)=13641 p<10-50).

+

As a work period began, the reward-relevant saccades exhibited high vigor, but then trial-by-trial, this vigor declined, reaching a low vigor value just before the work period ended (Fig. 3C vigor). Remarkably, on days in which the tube was placed farther, saccade vigor was lower (RMANOVA, effect of tube distance, subject M: F(2,59033)=224 p<10-50, subject R: F(2,50103)=75.51 p=1.8×10-33). Thus, increasing the effort cost of extracting food during the harvest period reduced saccade vigor during the work period.

+

By definition, a more vigorous saccade had a greater peak velocity. This might imply that high vigor saccades should suffer from inaccuracy due to signal dependent noise (20). However, we observed the opposite tendency: as saccade vigor increased, both the magnitude and the variance of the endpoint error decreased (Fig. 3C, 2-way ANOVA, effect of vigor on error magnitude, subject M: F(8,59046)=480 p<10-50, subject R: F(8,50184)=252 p<10-50, effect of vigor on error variance, subject M: F(8,2673)=18200 p<10-50, subject R: F(8,2673)=4170 p<10-50). That is, reducing the effort costs of harvest not only promoted vigor, it also facilitated accuracy (21).

+

Cognitive signals such as effort and reward are associated with changes in pupil size (10), as well as transient activation of brainstem neuromodulatory circuits in locus coeruleus (11). We wondered if the changes in tube position altered the output of these neuromodulatory circuits, as inferred via pupil size. For each reward-relevant saccade, we measured the pupil size during a ±250 ms window centered on saccade onset, and then normalized this measure based on the distribution of pupil sizes that we had measured during the entire recording for that session in that subject, resulting in a z-score.

+

At the onset of each work period the pupils were dilated, but as the subjects performed more trials, the pupils constricted, exhibiting a trial-by-trial reduction that paralleled the changes in saccade vigor (Fig. 3C). Notably, the effort cost of harvest affected pupil size: during the work period the pupils were more dilated if the tube was placed closer to the mouth (Fig. 3C, RMANOVA, effect of tube distance, subject M: F(2,60502)=20 p=2×10-9, subject R: F(2,50431)=23.8 p=4.9×10-11). That is, when the effort cost of harvest was lower, the pupils dilated, and the saccades were invigorated.

+

In summary, when we increased the effort cost of harvest, both the movements and the decisions changed: the pupils constricted and the movements slowed, but they chose to work more trials before initiating harvest.

+ + +Increased effort cost of harvest reduced lick vigor +

The work period ended when the subject chose to stop tracking the target and initiated harvest via a licking bout. As in saccades, we defined lick vigor via the ratio of the actual peak velocity of the lick with respect to the expected velocity for that lick amplitude. As amplitude increased, lick peak velocity increased during both protraction and retraction (Fig. 4A). Some of the licks were reward seeking and directed toward the tube, while others were grooming licks, cleaning the tongue and the area around the mouth (Supporting Information Video 5). Reward-seeking licks were more vigorous than grooming licks (2-way ANOVA, effect of lick type, protraction, subject M: F(1,272233)=66 p=4.5×10-16, subject R: F(1,229052)=698 p<10-50), and retraction was more vigorous than protraction (reward seeking licks, retraction vs. protraction, subject M: t(241145)=532 p<10-50, subject R: t(213674)=665 p<10-50).

+ + +

As the effort cost of harvest increased, lick vigor declined, and the pupils constricted. A. Peak speed of reward-seeking and grooming licks during protraction and retraction as a function of lick amplitude. B. Vigor of reward seeking licks (protraction) and pupil size as a function of lick number during harvest at various tube distances. C. Lick vigor and pupil size as a function of time during the entire recording session. Line colors depict tube distance as in part B. D. Average lick vigor and pupil size during a harvest as a function of number of trials successfully completed in the previous work period. Lick vigor and pupil size were greater when more food had been stored. E. Following a successful lick (contact with food), the next lick was more vigorous, and pupils dilated. Following a failed lick, the next lick was slowed, and pupils were less dilated. F. We observed no consistent effect of lick vigor on lick accuracy across subjects or across tube distances. Error bars are SEM.

+ + +

As the harvest began, the first lick was very low vigor, but lick after lick, the movements gathered velocity, reaching peak vigor by the 3rd or the 4th lick (Fig. 4B). As the harvest continued, lick vigor gradually declined. Like saccades, licks had a lower vigor in sessions in which the tube was placed farther from the mouth (RMANOVA, effect of tube distance, subject M: F(2,59033)=222.5 p<10-50, subject R: F(2,133502)=224 p<10-50), and this pattern was present during the entire recording session (Fig. 4C, left subplot). Thus, an increased effort cost of harvest promoted sloth: reduced vigor of saccades during the work period, and reduced vigor of licks during the harvest period.

+

For each reward seeking lick, we measured pupil size during a ±250ms window centered on the moment of peak tongue displacement. During licking, the pupil size changed with a pattern that closely paralleled lick vigor: as the harvest began, pupil size was small, but it rapidly increased during the early licks, then gradually declined as the harvest continued (Fig. 4B, right subplot). Importantly, the pupils were more dilated in sessions in which the tube was closer to the mouth (Fig. 4C, right subplot, effect of tube distance, subject M: F(2,166742)=583 p<10-50, subject R: F(2,130493)=118 p<10-50). As a result, when the effort cost of reward increased, the pupils constricted, and the vigor of both saccades and licks decreased.

+

While the theory predicted that moving the tube farther would result in a longer work period and reduced movement vigor, it also predicted that the subjects would reduce their harvest duration (reduced licks, Fig. 2C). That is, it predicted that the subjects would work longer, stowing more food, but leave more of it behind. This last prediction did not agree with our data (see Discussion). For Subject R, the number of licks were approximately the same across the various tube distances, and for Subject M the number of licks increased with tube distance (Supplementary Fig. 1).

+

In summary, within a harvest period, lick vigor rapidly increased and then gradually declined. Simultaneous with the changes in vigor, the pupils rapidly dilated and then gradually constricted. In sessions where the tube was placed farther from the mouth, the licks had lower vigor, and the pupils were more constricted.

+ + +Expectation of greater reward increased lick vigor +

As the subject worked, they accumulated food, thus increasing the magnitude of the available reward. To check whether reward magnitude affected movement vigor, for each tube distance we computed the average lick vigor during the harvest as a function of the number of trials completed in the preceding work period. We found that when the work period had included many completed trials, then the movements in the ensuing harvest period were more vigorous (Fig. 4D, 2-way ANOVA, effect of trials, subject M: F(4, 164242)=353 p<10-50, subject R: F(4,123411)=152 p<10-50). Thus, the licks were invigorated by the amount of food that awaited harvest.

+

Because the tube was small, many of the licks missed their goal and failed to contact the food. The success or failure of a lick affected both the vigor of the subsequent lick, and the change in the size of the pupil. Following a successful lick there was a large increase in lick vigor (Fig. 4E, subject M: t(85182)=40 p<10-50, subject R: t(81378)=104 p<10-50), and a large increase in pupil size (subject M: F(84969)=57 p<10-50, subject R: t(80318)=94 p<10-50). In contrast, following a failed lick the subjects either reduced or did not increase their lick vigor (Fig. 4E, subject M: t(114159)=0.88 p=0.37, subject R: t(97164)=-44 p<10-50). This failure also produced a smaller increase in pupil size (comparison to successful lick, two sample t-test, subject M: t(198722)=14.8 p=4.3×10-50, subject R: t(176044)=53 p<10-50). Thus, a single successful lick led to acquisition of reward, which then was followed by a relatively large increase in pupil size, and an invigorated subsequent lick.

+

For saccades we had found that increased vigor was associated with greater accuracy. To quantify the relationship between lick vigor and accuracy, for each tube distance we labeled each reward seeking lick as being high or low vigor. For subject M, high vigor licks tended to be more successful, but this was not the case for subject R (Fig. 4F). Moreover, tube distance did not produce a consistent effect on lick success.

+

In summary, the subject licked more vigorously following a long work period in which they had accumulated more reward. Moreover, when a lick was successful in acquiring reward, they increased the vigor of the subsequent lick.

+ + +Hunger promoted work and increased vigor +

Our theory predicted that it should be possible to change decisions in one direction (say work longer), while altering movement vigor in the opposite direction (move faster). An increase in the subjective value of reward, as might occur when the subject is hungry, should have two effects: increase the number of trials that the subject chooses to perform before commencing harvest, and increase movement vigor.

+

We did not explicitly manipulate the weight of the subjects. Indeed, to maintain their health, we strived to keep their weights constant during the roughly 2.5-year period of these experiments. However, there was natural variability, which allowed us to test the predictions of the theory.

+

We found that when their weight was lower than average, the subjects chose to work a greater number of trials before commencing harvest (Fig. 5A, two sample t-test, subject M: t(11052)=7.9 p=3.4×10-25, subject M: t(12549)=10.1 p=9.3×10-24). This result was similar to the effect that we had seen when the effort cost of harvest was increased. However, the theory had predicted that the effect on vigor should be in the opposite direction: if hunger increased reward valuation, then one should speed the movements and hasten food acquisition. Notably, weight did not have a consistent effect on saccade vigor across the two subjects (Fig. 5A), yet during the harvest, both subjects licked with greater vigor when their weight was lower (Fig. 5A, subject M: t(219752)=88 p<10-50, subject R: t(205163)=22 p<10-50).

+ + +

Relatively low body weight, potentially reflecting a greater valuation of reward, coincided with longer work periods and greater vigor. Pupil size correlated with both vigor and decisions. A. Trials successfully completed during a work period as a function of normalized body weight at the start of the session. B. Left: saccade vigor as a function of trial number for low and high body weights. Right: pupil size during the same saccades. C. Left: lick vigor as a function of lick number during the harvest period. Right: pupil size during the same licks. D. Saccade vigor during the work period, and lick vigor during the harvest period, as a function of pupil size. E. Work duration and harvest duration as a function of pupil size. Error bars are SEM.

+ + +

Thus, while both the effort cost of reward and hunger promoted greater work, effort promoted sloth while hunger promoted lick vigor.

+ + +Pupil size variations strongly correlated with changes in decisions and movements +

Finally, we considered the data across both the work and the harvest periods and asked how well movement vigor tracked pupil size. The results demonstrated that in both the work and the harvest periods, for both saccades and licks, an increase in pupil size was associated with an increase in vigor (Fig. 5B, reward-relevant saccades, subject M: r=0.989 p=7.7×10-9, subject R: r=0.97 p=7.1×10-7, reward-seeking protraction licks, subject M: r=0.969 p=9.8×10-7, subject R: r=0.989 p=6.3×10-9). Moreover, when the pupil was dilated, the work periods tended to be shorter (Fig. 5C, subject M: r=-0.90 p=0.00014, subject R: r=-0.97 p=6.3×10-7), while harvest durations tended to be longer (Fig. 5C, subject M: r=0.894 p=0.00021, subject R: r=0.935 p=2.4×10-5). Thus, pupil dilation was associated with choosing to work less, while moving faster.

+ + + +Discussion +

What we choose to do is the purview of the decision-making circuits of our brain, while the implicit vigor with which we perform that action is the concern of the motor-control circuits. From a theoretical perspective (9), our brain should coordinate these two forms of behavior because both the act that we select, and its vigor, dictate expenditure of time and energy, contributing to a capture rate that affects longevity and fecundity (22). Does the brain coordinate decisions and movements to maximize a capture rate? If so, how might the brain accomplish this coordination?

+

Here, we designed a foraging task in which marmosets worked by making saccades, accumulating food for each successful trial, then stopped working and harvested their cache by licking. On every trial they decided whether to work, or to harvest. Their decision was carried out by the motor system, producing either a visually guided saccade, or a lick, each exhibiting a particular vigor. The theory predicted that to maximize the capture rate, the appropriate response to an increased effort cost of harvest was to do two things: work longer to cache more food but reduce vigor to conserve energy.

+

We varied the effort costs by moving the food tube with respect to the mouth. This changed the effort cost of harvest but not the effort cost of work. The subjects responded by altering how they worked as well as how they harvested. When the harvest was more effortful, they performed more saccade trials to stockpile food. They also slowed their movements, reducing saccade velocity during the work period, and reducing lick velocity during the harvest period. Notably, the most vigorous saccades were also the most accurate: as saccade vigor increased, so did endpoint accuracy.

+

The theory made a second prediction: as the value of reward increased, the subjects should again choose to work a longer period before initiating harvest, but unlike the effort costs, now respond by moving more vigorously. We did not directly manipulate the subjective value of reward, but rather relied on the natural fluctuations in body weight and assumed that when their weight was low, the subjects were hungrier for reward. Indeed, when their weight was low, the subjects again chose to work longer, but now elevated their vigor during the harvest period.

+

Finally, we quantified the effect of reward magnitude on vigor. Within a session, lick vigor increased robustly as a function of the number of trials completed in the preceding work period. Thus, the licks were invigorated by the amount of food that awaited harvest.

+

Notably, some of the predictions of the theory did not agree with the experimental data. An increased effort cost did not accompany a reduction in the duration of harvest, and hunger did not increase saccade vigor robustly. Indeed, earlier experiments have shown that if the effort cost of harvest increases, animals who expend the effort will then linger longer to harvest more of the reward that they have earned (2). This mismatch between observed behavior and theory highlights some of the limitations of our formulation. For example, our capture rate reflected a single work-harvest period, rather than a long sequence. Moreover, the capture rate did not consider the fact that the food tube had finite capacity, beyond which the food would fall and be wasted. This constraint would discourage a policy of working more but harvesting less. Finally, if we assume that a reduced body weight is a proxy for increased subjective value of reward, it is notable that we observed a robust effect on vigor of licks, but not saccades. A more realistic capture rate formulation awaits simulations, possibly one that describes capture rate not as the ratio of two sums (sum of gains and losses with respect to sum of time), but rather the expected value of the ratio of each gain and loss with respect to time (23, 24).

+

A shortcoming of our model is that we did not include a link between lick vigor and its probability of success. As a result, when we moved the food tube away, the model did not consider the possibility that maintaining lick accuracy may involve reduced vigor. The reason for this is that we searched for but could not find a consistent relationship, across subjects or effort conditions, between protraction speed of the tongue and its success probability. Thus, we cannot exclude this alternate hypothesis. However, the most interesting aspect of our results was that when we increased tube distance, making harvest more effortful, there was not only a reduction in lick vigor, but also a reduction in saccade vigor. That is, the decisions and actions during the work period responded to the increased effort cost of reward during the harvest period.

+

What might be a neural basis for this coordination of decisions and movements? A clue was the fact that the pupils were more constricted in sessions in which the effort cost of harvest was greater. This global change in pupil size accompanied delayed harvest and reduced vigor across sessions, but surprisingly, even within a session, transient changes in pupil size accompanied changes in vigor. During the work period the trial-to-trial reduction in saccade vigor accompanied trial-to-trial constriction of the pupil, and within a harvest period, the rapid rise and then the gradual fall in lick vigor paralleled rapid dilatation followed by gradual constriction of the pupil.

+

Pupil dilation is a proxy for activity in the brainstem neuromodulatory system (25) and is a measure of arousal (26). Control of pupil size is dependent on spiking of norepinephrine neurons in locus coeruleus (LC-NE): an increase in the activity of these neurons produces pupil dilation (27, 28). Some of these neurons show a transient change in their activity when acquisition of reward requires expenditure of either physical (11) or mental effort (29), even when there is no concomitant movement to be made. It is possible that in the present task, as the effort cost of harvest increased, LC-NE neurons decreased their activity, producing pupil constriction. If so, the reduced NE release may have had two simultaneous effects: encourage work and promote delayed gratification in brain regions that control decisions, discourage energy expenditure and promote sloth in brain regions that control movements. Thus, the idea that emerges is that the response of NE to economic variables, as inferred via changes in pupil size, might act as a bridge to coordinate the computations in the decision-making circuits with the computations in the motor-control circuits, aiming to implement a consistent control policy that improves the capture rate.

+

In addition to NE, the basal ganglia, and in particular the neurotransmitter dopamine, are likely the key contributors to the coordination of decisions with actions (30, 31). When the effort price of a preferred food increases, animals choose to work longer, pressing a lever a greater number of times (32, 33). This desire to expend effort to acquire a valuable reward is reduced if dopamine is blocked in the ventral striatum (34–36). Hunger activates circuits in the hypothalamic nuclei, disinhibiting dopamine release in response to food cues (37). Dopamine concentrations in the striatum drop when the effort price of a food reward increases (38), and dopamine release before onset of a movement tends to invigorate that movement (39). Thus, the presence of dopamine may not only alter decisions by encouraging expenditure of effort, but also modify movements by promoting vigor.

+

Experiments of Hayden et al. (40) and Barack et al. (41) suggest that the decision of when to stop work and commence harvest may rely on computations that are carried out in the cingulate cortex. They found that as monkeys deliberated between the choice of staying and acquiring diminishing rewards, or leaving and incurring a travel cost, these neurons encoded a decision-variable that reflected the value of leaving the patch. The prediction that emerges from our work is that the rate of rise of these decision variables may be modulated by the presence of NE.

+

From a motor control perspective, a surprising aspect of our results was that an increase in saccade vigor accompanied an improvement in endpoint accuracy. In our earlier work we found that during reaching, reward increased vigor without reducing accuracy (42). Thus, the brain has the means to increase movement vigor and improve its accuracy. How is this achieved?

+

We found that the high vigor saccades were produced when the pupils were dilated, implying an increased release of NE. In songbirds, increased NE release acts on the basal ganglia to suppress activity of spiny neurons, and this reduced activity in the basal ganglia accompanies reduced variance in the songs that the animal sings (43). Thus, NE may play a critical role in control of movement variability. For saccades, control of endpoint accuracy depends on the coordinated activity of Purkinje cells in the oculomotor region of the cerebellar vermis (44, 45). LC projects to the cerebellum, and stimulation of LC neurons increases the sensitivity of Purkinje cells to their inputs (46).

+

Is movement vigor increased following increased NE inputs from LC to the basal ganglia, and accuracy improved following increased NE inputs from LC to the cerebellum? Does decision-making shift toward greater work and delayed gratification following reduced NE inputs from LC to the frontal lobe? These are some of the questions that await future experiments.

+ + +Methods +

Behavioral and neurophysiological data were collected from two marmosets (Callithrix Jacchus, male and female, 350-390 g, subjects R and M, 6 years old). The neurophysiological data focused on the cerebellum and are described elsewhere (44, 47, 48). Here, our focus is on the behavioral data.

+

The marmosets were born and raised in a colony that Prof. Xiaoqin Wang has maintained at the Johns Hopkins School of Medicine since 1996. The procedures on the marmosets were evaluated and approved by the Johns Hopkins University Animal Care and Use Committee in compliance with the guidelines of the United States National Institutes of Health.

+ +Data acquisition +

Following recovery from head-post implantation surgery, the animals were trained to make saccades to visual targets and rewarded with a mixture of apple sauce and lab diet (47). They were placed in a monkey chair and head-fixed while we presented visual targets on an LCD screen (Curved MSI 32” 144 Hz - model AG32CQ) and tracked both eyes at 1000 Hz using an EyeLink-1000 system (SR Research, USA). Timing of target presentation on the video screen was measured using a photo diode. Tongue movements were tracked with a 522 frame per second Sony IMX287 FLIR camera, with frames captured at 100 Hz.

+

Each trial began with a saccade to the center target followed by fixation for 200 ms, after which a primary target (0.5×0.5 deg square) appeared at one of 8 randomly selected directions at a distance of 5-6.5 deg. Onset of the primary target coincided with presentation of a tone. As the animal made a saccade to the primary target, that target was erased, and a secondary target was presented at a distance of 2-2.5 deg, also at one of 8 randomly selected directions. The subject was rewarded if following the primary saccade it made a corrective saccade to the secondary target, landed within 1.5 deg radius of the target center, and maintained fixation for at least 200 ms. Onset of reward coincided with presentation of another distinct tone. Following an additional 150-250 ms period (uniform random distribution), the secondary target was erased, and the center target was displayed, indicating the onset of the next trial. Thus, a successful trial comprised of a sequence of 3 saccades: center, primary, and corrective, after which the subject received a small increment of food (0.015 mL).

+

The food was provided in two small tubes (4.4 mm diameter), one to the left and the other to the right of the animal (Fig. 1A). A successful trial produced a food increment in one of the tubes and would continue to do so for 50-300 consecutive trials, then switch to the other tube. Because the food increment was small, the subjects naturally chose to work for a few consecutive trials, tracking the visual targets and allowing the food to accumulate, then stopped tracking and harvested the food via a licking bout. The licking bout typically included a sequence of 15-40 licks. The subjects did not work while harvesting. As a result, the behavior consisted of a work period (targeted saccades), followed by a harvest period (targeted licking), repeated hundreds of times per session.

+

The critical variables were the number of trials that the subject chose to perform before initiating harvest, the vigor of their saccades during the work period, and the vigor of their licks during the harvest period.

+ + +Data analysis +

All saccades, regardless of whether they were instructed by presentation of a visual target or not, were identified using a velocity threshold. Saccades to primary, secondary, and central targets were labeled as reward-relevant saccades, while all remaining saccades were labeled as task irrelevant.

+

We analyzed tongue movements using DeepLabCut (12). Our network was trained on 89 video recordings of the subjects with 15-25 frames extracted and labeled from each recording. The network was built on the ResNet-152 pre-trained model, and then trained over 1.03×10 iterations with a batch size of 8, using a GeForce GTX 1080Ti graphics processing unit (49). A Kalman filter was further applied to improve quality and smoothness of the tracking, and the output was analyzed in MATLAB to quantify varying lick events and kinematics.

+

We tracked the tongue tip and the edge of the food in the tube, along with control locations (nose position and tube edges). We tracked all licks, regardless of whether they were aimed toward the tube (reward seeking), or not (grooming). Reward seeking licks were further differentiated based on whether they aimed to enter the tube (inner-tube licks), hit the outer edge of the tube (outer-edge licks), or fell below the tube (under tube). If any of these licks successfully contacted the tube, we labeled that lick as a success (otherwise, a failed lick).

+

Pupil area was measured during a ±250 ms period centered at the onset of each reward-relevant saccade, and the onset of each lick. We then normalized the pupil measurements by representing it as a z-score with respect to the mean value for that session.

+ + +Saccade and tongue vigor +

We relied on previous work to define vigor of a movement (5, 9, 18, 19). Briefly, if the amplitude of a movement is x and the peak speed of that movement is v, then for each subject the relationship between the two variables can be described as: + + + +In the above expression, α,β ≥ 0 and are subject-specific parameters. For a movement with amplitude x, its vigor was defined as the ratio of the actual peak speed with respect to the expected value of its peak speed, i.e., . Expected value was computed by fitting Eq. (3) to all the data acquired across all sessions. When vigor is greater than 1, the movement had a peak velocity that was higher than the mean value associated with that amplitude.

+ + +Model formulation +

We chose a formulation of utility (Eq. 1) based on a normative approach that ecologists have used to understand the decisions that animals make regarding how far to travel for food, what mode of travel to use, and how long to stay before moving on to another reward opportunity (4, 13, 14). In a typical formulation of the theory, the numerator represents the reward gained (in units of energy), minus the effort expended (also in units of energy), while the denominator represents the amount of time spent during that behavior. We represented this idea in Eq. (1) with saccades that produced reward accumulation, and licks that produced reward consumption. Thus, the utility that we aim to maximize is the rate of energy gained.

+

The specific functions that we used to represent the energy gained through reward acquisition, and the energy expended through effort expenditure, came either from experiment design, or from the measurements we have made in other experiments. We modeled reward accumulation as a linear rise in energy stored because successful saccades produced a linear increase in the food cache. We modeled harvesting of the food as a hyperbolic function of the number of licks to represent the fact that as the licking bout began, each successful lick depleted the food, and thus the first few licks produced a greater amount of food consumption than the last few licks. We modeled the effort cost of licking as a linear function of the number of licks.

+

A critical assumption that we made is that energy expended performing the saccade trials (which grew faster than linearly as a function of the number of trials attempted), grew faster than the time spent attempting those same trials (which grew linearly with the number of trials). This assumption is based on the heuristic that the average rate of energy lost following a large number of attempted trials is greater than the average rate of energy lost following a small number of attempted trials.

+

The model’s simplicity provided closed-form solutions across all parameter values, allowing us to make predictions without having to fit the model to the measured data. For example, for all parameter values that produce a real solution (as opposed to imaginary), the optimal number of saccade trials increases with the square root of the cost of licking. Thus, the basic prediction of the model is that to maximize the capture rate, regardless of parameter values, an increase in the effort required for harvest should be met with a greater willingness to work. The closed-form solutions are presented in the supplementary document (simulations.nb).

+ + +Other models of utility +

In composing our utility (Eq. 1), we chose to combine reward and effort additively. This is in contrast to other approaches in which effort discounts reward multiplicatively (50–52). Our reasoning is that multiplicative interactions have the limitation that they are incompatible with the observation that reward invigorates movements.

+

To compare additive and multiplicative approaches, let us consider an arbitrary function u r that specifies how effort varies with movement duration T. Typically, this is a U-shaped function that describes energy expenditure as a function of movement duration, as in (8). In the case of multiplicative interaction between reward and effort, we can consider the following representation of utility: + + + +In the above formulation, reward a is discounted hyperbolically with time, and an increase in reward increases the utility of the action. The optimum movement vigor has the duration T* that maximizes this utility. Notably, because increasing reward merely scales this utility, it has no effect on vigor. Thus, a utility in which reward is multiplied by a function of effort generally fails to predict dependence of movement vigor on reward.

+ + +Simulations +

The optimal policy specifies the decisions and movements that for the effort cost defined in Eq. (2), maximizes the capture rate defined in Eq. (1). This policy selects the number of saccade trials ns to perform during the work period, the number of licks nl to perform during the harvest period, and the vigor of each lick, represented by the average duration of a lick Tl. To compute the optimal policy, we found the derivative of the capture rate with respect to each policy variable ns, nl, and Tl, then set each derivative equal to zero, producing three simultaneous nonlinear equations. In all three cases, we were able to solve for the relevant control variable analytically (see Mathematica notebook simultatons.nb for the derivations). We found that if the solution was a real number, then regardless of parameter values, an increase in d (distance of the tube to the mouth), the optimal policy produced an increase in decrease in and increase in . Thus, the results illustrated in Fig. 2 are robust to changes in parameter values.

+

To generate the plots in Fig. 2A, we used the following parameter values: α = 20, βs = 0.5, βL = 0.3, cs = 0.5, Ts = 1, TL = 0.2, cL = 0.5 (low effort), cL = 2.5 (high effort). For the plots in Fig. 2C and 2D, we used the same parameter values, but cL was defined via Eq. (2). Thus, tube distance d varied, and TL was unknown and was solved for. In Eq. (2), kL = 1. In the simulations, to describe state of hunger, we set α = 20 for a sated state and a= 25 for a hungry state.

+ + +Statistical analysis +

Hypothesis testing was performed using functions provided by the MATLAB Statistics and Machine Learning Toolbox, version R2021b. For t-tests, across the one-sample, paired-sample, and two-sample conditions, p values were computed using the ttest and ttest2 functions with data that was combined across sessions, separated by condition. For ANOVA tests, in the one-way condition, p values were computed using a nonparametric Kruskal-Wallis test, using the kruskalwallis function. In the 2-way condition, the anovan function was used to compute p values, accounting for an unbalanced design resulting from a varied number of samples across conditions. In both cases, like in the t-tests, data was combined across sessions, separated by condition. In the repeated measures condition, each session was treated as a subject with multiple repeated measures representing a given variable (i.e., lick vigor per lick in a harvest period). To fit a repeated measures model, the fitrm function was used, then analyzed using the ranova function. In all cases of repeated measures ANOVA, compound symmetry assumptions were tested using the Mauchly sphericity test with the maulchy function. In cases where the assumption was violated (Maulchy test p < 0.05), epsilon adjustments were used, with the epsilon function, to compute corrected p values (for ε > .75, use Huynh-Feldt p value and for ε < .75, use Greenhouse-Geisser p values). For correlation analyses, Pearson’s correlation coefficient, r, and corresponding p values were computed using the corrcoef function.

+ + + + + +Acknowledgements +

The work was supported by grants from the NIH (R01-EB028156, R01-NS078311, R37-NS128416), and the Office of Naval Research (N00014-15-1-2312).

+ + +Supporting Information + + +

Number of licks per harvest as a function of tube distance.

+ + +

Video 1. Example of a successful inner-tube lick.

+

Video 2. Example of an under-tube lick that failed to contact food. Note the corrective sub-movements, as has been observed in mice (53).

+

Video 3. Example of an outer-tube lick that failed to contact food.

+

Video 4. Example of a lick that hit the outer edge of the tube and failed to contact food.

+

Video 5. Example of a grooming lick.

+ + +References +E. L. Charnov, Optimal foraging, the marginal value theorem. Theor.Popul.Biol. 9, 129–136 (1976). +R. J. Cowie, Optimal foraging in great tits (Parus major). Nature 268, 137–139 (1977). +R. Shadmehr, A. A. Ahmed, Vigor: neuroeconomics of movement control (MIT Press, 2020). +H. Richardson, N. A. M. Verbeek, Diet selection and optimization by northwestern crows feeding on Japanese littleneck clams. Ecology 67, 1219–1226 (1986). +T. Yoon, A. Jaleel, A. A. Ahmed, R. Shadmehr, Saccade vigor and the subjective economic value of visual stimuli. J. Neurophysiol. 123, 2161–2172 (2020). +C. C. Korbisch, D. R. Apuan, R. Shadmehr, A. A. Ahmed, Saccade vigor reflects the rise of decision variables during deliberation. Curr. Biol. (2022) 10.1016/j.cub.2022.10.053 (December 4, 2022). +R. Shadmehr, J. J. Orban de Xivry, M. Xu-Wilson, T. Y. Shih, Temporal discounting of reward and the cost of time in motor control. J.Neurosci. 30, 10507–10516 (2010). +R. Shadmehr, H. J. Huang, A. A. Ahmed, A representation of effort in decision-making and motor control. Curr.Biol. 26, 1929–1934 (2016). +T. Yoon, R. B. Geary, A. A. Ahmed, R. Shadmehr, Control of movement vigor and decision making during foraging. Proc.Natl.Acad.Sci.U.S.A. 115, E10476–E10485 (2018). +S. Joshi, J. I. Gold, Pupil Size as a Window on Neural Substrates of Cognition. Trends Cogn. Sci. 24, 466–480 (2020). +P. Bornert, S. Bouret, Locus coeruleus neurons encode the subjective difficulty of triggering and executing actions. PLOS Biol. 19, (2021). +A. Mathis, et al., DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 21, 1281–1289 (2018). +D. W. Stephens, J. R. Krebs, Foraging Theory (Princeton Univ. Press, 1986). +L. M. Bautista, J. Tinbergen, A. Kacelnik, To walk or to fly? How birds choose among foraging modes. Proc.Natl.Acad.Sci.U.S.A. 98, 1089–1094 (2001). +H. J. Huang, A. A. Ahmed, Older adults learn less, but still reduce metabolic cost, during motor adaptation. J.Neurophysiol. 111, 135–144 (2014). +H. J. Ralston, Energy-speed relation and optimal speed during level walking. Int.Z.Angew.Physiol. 17, 277–283 (1958). +G. J. Bastien, P. A. Willems, B. Schepens, N. C. Heglund, Effect of load and speed on the energetic cost of human walking. Eur.J.Appl.Physiol 94, 76–83 (2005). +T. R. Reppert, K. M. Lempert, P. W. Glimcher, R. Shadmehr, Modulation of Saccade Vigor during Value-Based Decision Making. J.Neurosci. 35, 15369–15378 (2015). +T. R. Reppert, et al., Movement vigor as a trait-like attribute of individuality. J.Neurophysiol. 120, 741–757 (2018). +C. M. Harris, D. M. Wolpert, Signal-dependent noise determines motor planning. Nature 394, 780– 784 (1998). +C. Wang, Y. Xiao, E. Burdet, J. Gordon, N. Schweighofer, The duration of reaching movement is longer than predicted by minumum variance. J.Neurophysiol. 116, 2342–2345 (2016). +W. C. Lemon, Fitness consequences of foraging behaviour in the zebra finch. Nature 352, 153–155 (1991). +M. Bateson, A. Kacelnik, Preferences for fixed and variable food sources: variability in amount and delay. J.Exp.Analysis.Behav. 63, 313–329 (1995). +M. Bateson, A. Kacelnik, Rate currencies and the foraging starling: the fallacy of the averages revisited. Behav. Ecol. 7, 341–352 (1996). +E. M. Vazey, D. E. Moorman, G. Aston-Jones, Phasic locus coeruleus activity regulates cortical encoding of salience information. Proc. Natl. Acad. Sci. 115, E9439–E9448 (2018). +S. Mathot, Pupillometry: Psychology, Physiology, and Function. J. Cogn. 1, 1–23 (2018). +S. Joshi, Y. Li, R. M. Kalwani, J. I. Gold, Relationships between Pupil Diameter and Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex. Neuron 89, 221–234 (2016). +V. Breton-Provencher, M. Sur, Active control of arousal by a locus coeruleus GABAergic circuit. Nat. Neurosci. 22, 218–228 (2019). +C. Contadini-Wright, K. Magami, N. Mehta, M. Chait, Pupil Dilation and Microsaccades Provide Complementary Insights into the Dynamics of Arousal and Instantaneous Attention during Effortful Listening. J. Neurosci. 43, 4856–4866 (2023). +D. Thura, P. Cisek, The Basal Ganglia Do Not Select Reach Targets but Control the Urgency of Commitment. Neuron 95, 1160–1170 (2017). +D. M. Herz, et al., Dynamic control of decision and movement speed in the human basal ganglia. Nat. Commun. 13, 7530 (2022). +J. D. Salamone, et al., Haloperidol and nucleus accumbens dopamine depletion suppress lever pressing for food but increase free food consumption in a novel food choice procedure. Psychopharmacol. Berl 104, 515–521 (1991). +J. E. Aberman, J. D. Salamone, Nucleus accumbens dopamine depletions make rats more sensitive to high ratio requirements but do not impair primary food reinforcement. Neuroscience. 92, 545– 552 (1999). +M. Koch, A. Schmid, H. U. Schnitzler, Role of nucleus accumbens dopamine D1 and D2 receptors in instrumental and Pavlovian paradigms of conditioned reward. Psychopharmacol. Berl 152, 67–73 (2000). +A. M. Farrar, et al., Nucleus accumbens and effort-related functions: behavioral and neural markers of the interactions between adenosine A2A and dopamine D2 receptors. Neuroscience 166, 1056–1067 (2010). +S. E. Yohn, et al., The role of dopamine D1 receptor transmission in effort-related choice behavior: Effects of D1 agonists. Pharmacol.Biochem.Behav. 135, 217–226 (2015). +R. M. Cassidy, Q. Tong, Hunger and Satiety Gauge Reward Sensitivity. Front. Endocrinol. 8 (2017). +S. A. Schelp, et al., A transient dopamine signal encodes subjective value and causally influences demand in an economic context. Proc.Natl.Acad.Sci.U.S.A 114, E11303–E11312 (2017). +J. A. da Silva, F. Tecuapetla, V. Paixao, R. M. Costa, Dopamine neuron activity before action initiation gates and invigorates future movements. Nature 554, 244–248 (2018). +B. Y. Hayden, J. M. Pearson, M. L. Platt, Neuronal basis of sequential foraging decisions in a patchy environment. Nat.Neurosci. 14, 933–939 (2011). +D. L. Barack, S. W. C. Chang, M. L. Platt, Posterior Cingulate Neurons Dynamically Signal Decisions to Disengage during Foraging. Neuron 96, 339–347 (2017). +E. M. Summerside, R. Shadmehr, A. A. Ahmed, Vigor of reaching movements: reward discounts the cost of effort. J.Neurophys. 119, 2347–2357 (2018). +J. Singh Alvarado, et al., Neural dynamics underlying birdsong practice and performance. Nature 599, 635–639 (2021). +E. Sedaghat-Nejad, J. S. Pi, P. Hage, M. A. Fakharian, R. Shadmehr, Synchronous spiking of cerebellar Purkinje cells during control of movements. Proc. Natl. Acad. Sci. 119, e2118954119 (2022). +S. Barash, et al., Saccadic dysmetria and adaptation after lesions of the cerebellar cortex. J.Neurosci. 19, 10931–10939 (1999). +H. C. Moises, B. D. Waterhouse, D. J. Woodward, Locus coeruleus stimulation potentiates Purkinje cell responses to afferent input: The climbing fiber system. Brain Res. 222, 43–64 (1981). +E. Sedaghat-Nejad, et al., Behavioral training of marmosets and electrophysiological recording from the cerebellum. J.Neurophysiol. 122, 1502–1517 (2019). +S. Z. Muller, et al., Complex spikes perturb movements, revealing the sensorimotor map of Purkinje cells. bioRxiv, 2023–04 (2023). +K. He, X. Zhang, S. Ren, J. Sun, Deep Residual Learning for Image Recognition in (2016), pp. 770– 778. +H. Sugiwaka, H. Okouchi, Reformative self-control and discounting of reward value by delay or effort. Jpn. Psychol. Res. 46, 1–9 (2004). +C. Prevost, M. Pessiglione, E. Metereau, M. L. Clery-Melin, J. C. Dreher, Separate valuation subsystems for delay and effort decision costs. J.Neurosci. 30, 14080–14090 (2010). +M. C. Klein-Flugge, S. W. Kennerley, A. C. Saraiva, W. D. Penny, S. Bestmann, Behavioral modeling of human choices reveals dissociable effects of physical effort and temporal delay on reward devaluation. PLoS.Comput.Biol. 11, e1004116 (2015). +T. Bollu, et al., Cortex-dependent corrections as the tongue reaches for and misses targets. Nature, 1–6 (2021). + + +
diff --git a/test/2023.04.19.535725/equation-fix/2023.04.19.535725.xml b/test/2023.04.19.535725/equation-fix/2023.04.19.535725.xml new file mode 100644 index 00000000..8b6fae14 --- /dev/null +++ b/test/2023.04.19.535725/equation-fix/2023.04.19.535725.xml @@ -0,0 +1,456 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.04.19.535725 +1.2 + + +Regular Article + + +New Results + + +Neuroscience + + + +Subchronic alteration of vestibular hair cells in mice: implications for multisensory gaze stabilization + + + +http://orcid.org/0009-0004-1298-8228 +SchenbergLouise +1 + + +PalouAïda +2 +3 +4 + + +SimonFrançois +1 +5 + + +BonnardTess +1 + + +BartonCharles-Elliot +1 + + +http://orcid.org/0000-0001-7328-9480 +FrickerDesdemona +1 + + +http://orcid.org/0000-0002-4905-788X +TagliabueMichele +1 + + +http://orcid.org/0000-0002-3894-9401 +LlorensJordi +2 +3 +4 +* + + +http://orcid.org/0000-0003-2722-0532 +BeraneckMathieu +1 +* + +Université Paris Cité, CNRS UMR 8002, INCC - Integrative Neuroscience and Cognition Center, F-75006, Paris, France +Departament de Ciències Fisiològiques, Universitat de Barcelona, 08907 l’Hospitalet de Llobrega, Catalunya, Spain +Institut de Neurociènces, Universitat de Barcelona, Barcelona, Catalunya, Spain +Institut d’Investigació Biomèdica de Bellvitge (IDIBELL), 08907 l’Hospitalet de Llobregat, Catalunya, Spain +Department of paediatric otolaryngology, Hôpital Necker-Enfants Malades, APHP, F-75015, Paris, France + + +Corresponding authors: Pr Jordi Llorens, Universitat de Barcelona, 08907 l’Hospitalet de Llobrega, Catalunya, Spain. E-mail : jllorens@ub.edu; Dr. M. Beraneck. CNRS UMR 8002, Université Paris Cité, 45 rue des St-Pères, Paris 75270, France. Email: mathieu.beraneck@cnrs.fr + +2023 +2023.04.19.535725 + + +19 +4 +2023 + + +19 +9 +2023 + + +19 +9 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. + + + +Abstract

The functional complementarity of the vestibulo-ocular reflex (VOR) and optokinetic reflex (OKR) allows for optimal combined gaze stabilization responses (CGR) in light. While sensory substitution has been reported following complete vestibular loss, the capacity of the central vestibular system to compensate for partial peripheral vestibular loss remains to be determined. Here, we first demonstrate the efficacy of a 6-week subchronic ototoxic protocol in inducing transient and partial vestibular loss which equally affects the canal- and otolith-dependent VORs. Immunostaining of hair cells in the vestibular sensory epithelia revealed that organ-specific alteration of type I, but not type II, hair cells correlates with functional impairments. The decrease in VOR performance is paralleled with an increase in the gain of the OKR occurring in a specific range of frequencies where VOR normally dominates gaze stabilization, compatible with a sensory substitution process. Comparison of unimodal OKR or VOR versus bimodal CGR revealed that visuo-vestibular interactions remain reduced despite a significant recovery in the VOR. Modeling and sweep-based analysis revealed that the differential capacity to optimally combine OKR and VOR correlates with the reproducibility of the VOR responses. Overall, these results shed light on the multisensory reweighting occurring in pathologies with fluctuating peripheral vestibular malfunction.

+ + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Figure 1, 2 and 3 revised; discussion updated to clarify the link between hair cells and vestibular afferents; conclusion added to discuss vestibular pathologies.

 + + + + + +Introduction +

The vestibular system is well-preserved amongst vertebrates, participating in essential functions such as balance, postural control and, together with the optokinetic system, gaze stabilization (Straka et al., 2016; Wibble et al., 2022). Beyond these recognized roles, vestibular signals also contribute to cognitive processes e.g spatial orientation and navigation (Cullen, 2019), or body representation (Lopez et al., 2012; Facchini et al., 2021). Because of its involvement in many basic functions important in our daily life, vestibular pathologies affecting the inner ear are associated with a significant deterioration of the well-being of patients (Möhwald et al., 2020) and represent an important public health concern (Agrawal et al., 2009, 2013).

+

Research on post-lesion plasticity following permanent vestibular loss has shed light on the neural plastic mechanisms that follow a chronic unilateral or bilateral vestibular lesion, a process referred as “vestibular compensation” (Brandt et al., 1997; Cullen et al., 2010; Beraneck and Idoux, 2012). The compensation taking place after the lesion is known to involve dynamical multisensory reweighting of proprioceptive and visual inputs and of internal efferent copies (Cullen et al., 2010; Sadeghi et al., 2012; Sadeghi and Beraneck, 2020). While total and permanent lesions offer the experimental opportunity to characterize drastic cellular and molecular changes triggered by the total silencing of the vestibular endorgans, they imperfectly mimic clinical situations where peripheral vestibular function loss is only partial and/or transient (Bisdorff et al., 2009; Lopez-Escamez et al., 2015; Brandt and Dieterich, 2017). To better model fluctuating inner ear function, protocols based on subchronic exposure to an ototoxic substance, 3,3ʹ-iminodiproprionitrile (IDPN) were first introduced in the rat (Seoane et al., 2001; Sedó-Cabezón et al., 2014, 2015; Martins-Lopes et al., 2019) and more recently in the mouse (Greguske et al., 2019). Subchronic exposure to IDPN in drinking water at low doses allowed for progressive ototoxicity, leading to a partial and largely reversible loss of function. Although the subchronic IDPN protocol was shown to cause postural and locomotor deficits (Martins-Lopes et al., 2019), its effects on the gaze stabilizing reflexes, namely the vestibulo-ocular and optokinetic reflexes, have not yet been described.

+

The primary objective of the present study is first to assess how subchronic exposure to IDPN may affect the function of the different vestibular endorgans. To that end, we took advantage of our recently described methodology (Simon et al., 2020, 2021) using canal-specific and otolith-specific tests. Quantification of the vestibulo-ocular reflexes is a sensitive and specific method to assess the functionality of the sensory-motor vestibular pathway; it is the most used test in clinics, and highly correlates to quality-of-life reports in patients suffering from acute peripheral diseases (Möhwald et al., 2020).

+

The secondary objective is to determine whether visual substitution occurs following transient and partial vestibular loss. Optogenetic stimulation of the vestibular pathway demonstrated the recruitment of circuits involved in visual processing at the midbrain, thalamic and cortical regions (Leong et al., 2019). Recent imaging studies in rodents have shown that acute vestibular loss triggers brain-wide adaptive plasticity in circuits known to be involved in visual processing (Zwergal et al., 2016; Grosch et al., 2021). In addition, it was previously shown that OKR plasticity is triggered during vestibular compensation following a permanent vestibular lesion (Faulstich et al., 2006; Nelson et al., 2017).

+

We report that 6 weeks of IDPN subchronic treatment affects both the canal- and otolith-dependent vestibulo-ocular reflexes and that organ-specific loss of type I hair cells (HC) correlates with individual mice’s impairments. We show that optokinetic adaptive compensation is frequency-specific and delayed with respect to the VOR changes. OKR changes occur at the frequencies where physiologically the vestibular inputs dominate visuo-vestibular gaze stabilization. We demonstrate that despite the significant recovery of their vestibulo-ocular reflexes, the visuo-vestibular integration remains notably impaired in some IDPN-treated mice. We suggest that the “noisiness” of the recovered vestibular signal affects their capacity to optimally combine visual and vestibular responses. Overall, these results shed light on the dynamic of multisensory reweighting in patients suffering from fluctuating peripheral vestibular malfunction.

+ + +Results + +Effects of the subchronic treatment of IDPN on the canal- and otolith-dependent VOR +

To investigate the effects of the IDPN on vestibulo-ocular reflexes (VOR), animals were exposed to the ototoxic compound in the drinking water for six weeks (Treatment period), followed by 6 weeks of standard drinking water without IDPN (Washout period). The VOR were quantified every two weeks using canal-specific and otolith-specific tests. Horizontal sinusoidal rotations in the dark were performed and oculomotor responses were recorded using video-oculography (Figure 1A) to study the impact of IDPN treatment on the canal-dependent angular VOR (aVOR). Typical raw aVOR traces are shown in Figure 1B. At week 6 (W6) the amplitude of the eye movements was distinctly reduced compared to W0, while at W12 the amplitude of the response appeared partially restored. The dynamics of decrease and recovery of the mean aVOR gain over the course of the protocol are reported in Figure 1C for both IDPN (n=21) and SHAM (n=22) groups. The evolution over time of the aVOR gain is significantly different for the two groups of mice (ANOVA Weeks x Group interaction, F(6,246)=29,949, p<10-4). Before treatment, both groups responded similarly to the sinusoidal stimulations and their aVOR gain remained unchanged through 2 weeks of treatment (W2). However, starting W4 the aVOR gain of IDPN group significantly decreased with respect to W0 (Newman-Keuls post hoc test: IDPN W0 vs W4 p<10-4, see Table 1) and was significantly lower compared with the SHAM group (W4, IDPN vs SHAM p<10-4). The aVOR gain of the IDPN group remained lower compared to SHAM through the rest of the protocol (W6 IDPN vs SHAM p<10-4., W8 IDPN vs SHAM p<10-4, W10 IDPN vs SHAM p<10-4, W12 IDPN vs SHAM p<10-4) with a minimum reached at W6 corresponding to ∼2/3 of aVOR loss (IDPN W4 vs W6 p<10-4, IDPN W6 vs W8 p=0,043762). At the end of the six weeks of washout, the mean aVOR significantly improved (IDPN W6 vs W12 p<10-4) to levels observed at W4 (IDPN W4 vs W12 p=0.561969). Notably, the amplitude and dynamic of gain changes were similar for all frequencies >0.2Hz (see Figure 1 supplement 1A). However, at the lowest frequency tested (0.2Hz), aVOR gain decrease reached significance only at W6.

+ + +Effects of subchronic IDPN on canal- and otolithic- dependent VOR. +

A) Illustration of the angular horizontal vestibulo-ocular reflex (aVOR) set-up. All tests are performed in complete dark. B) Examples of velocity aVOR traces in response to table rotations (1Hz at 30°/s) recorded in the dark in an IDPN mouse before (W0, corresponding gain of 0.785; VAF of 0.98), after 6 weeks of treatment (W6, gain: 0.14 ; VAF: 0.61) and 6 weeks of washout (W12, gain:0.48; VAF: 0.92). Right movement is represented up. C) Mean aVOR gain of SHAM (n=22) and IDPN (n=21) mice during the protocol. D) Illustration of the ocular-counter roll (OCR) set-up. E)Examples of raw OCR traces at W0, W6 and W12 in a IDPN mouse recorded in the dark. Tilt to the right is represented up (positive values). (Left) eye elevation is represented up. F) Mean OCR gain of SHAM (n=14) and IDPN (n=13) mice. We note that there was a significant difference between SHAM and IDPN during the initial measurements at W0. However, at this time point mice were not yet separated into different groups. This incidental difference completely disappeared on the measurement performed at W2. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + + + +Statistics table of the aVOR gain for the IDPN treated group. + + +

aVOR responses were further modified by significant phase leads that affected all frequencies starting W6 (Figure 1 supplement 1B, ANOVA repeated measures Weeks x Group Interactions F(6, 246)=14.528, p<10-4). Overall, canalar responses remained unaffected until week 2, but the amplitude and/or timing of the aVOR was abnormal from week 4 until the end of the protocol, despite a significant recovery of angular VOR responses observed during the washout period.

+

To determine whether otolith-dependent VOR was also affected by IDPN treatment, ocular-counter roll responses (OCR) were tested during static lateral inclination in the range ±40° (OCR; Figure 1D). Examples of raw traces and quantification of the response are shown in Figure 1E, and mean gain of the OCR of each group is plotted in Figure 1F (n=13 IDPN, n=14 SHAM). The modulations of responses amplitude were significantly different between IDPN and SHAM groups (ANOVA repeated measures Weeks x Group Interaction F(6, 150)= 7.7411 p<10-4). We note that there was a significant difference between SHAM and IDPN during the initial measurements at W0, before any treatment. However, at this timepoint mice were not yet separated into different groups. This incidental difference completely disappeared on the measurement performed at W2 (W2 IDPN vs SHAM p=0.8135). While the gain of the SHAM group stays in a 0.5-0.6 range over the whole duration of the protocol, the responses of the IDPN group significantly decreased at W4 compared to the SHAM group (W4 IDPN vs SHAM p<10-4). This decrease was larger at W6 (IDPN W4 vs W6 p=0.023) and stayed significantly different from the SHAM group until W8. At W10 and W12, the OCR of IDPN group recovered to a level comparable to the SHAM group (W10 IDPN vs SHAM p=0.1262, W12 IPDN vs SHAM p=0.3385).

+

Sub-chronic treatment of IDPN was also investigated through the dynamic Off Vertical Axis Rotation (OVAR) test (Figure 1 supplement 1C), which primarily reflects maculo-ocular (dominantly otolithic) responses integrated by central vestibular pathways. The maculo-ocular reflex (MOR) bias decreased significantly compared to the SHAM group (ANOVA Weeks x Group, F(6, 132)= 10.076, p<10-4) starting W2 of treatment (W2 IDPN vs SHAM p=0.0018), demonstrating that ototoxicity already affected the vestibular system at this early time point. The maximal decrease was reached at W6 (W6 IDPN vs SHAM p=0.00012) and recovery led to normal responses at W12 (W12 IDPN vs SHAM p=0.4709).

+ + +Comparison of otolith- and canal-dependent plasticity in individuals +

To compare the dynamic of canal- and otolith-dependent VOR alterations, the paired VOR and OCR gains measured in individuals from the IDPN group (n=13) are plotted together (Figure 2). Alteration in canal- and otolith-dependent responses followed a very similar time course (OCR vs aVOR at 1HZ on figure 2A; similar patterns were obtained with the other tested frequencies (not shown)).

+ + +Comparison of the dynamics of canalar and otolithic loss of function. +

A) Evolution of the Mean Gain of the aVOR (1Hz) and OCR responses of IDPN mice (n=13) during the treatment and washout periods. B) Individual gains for 1Hz aVOR (left) and OCR (right) of the 13 IDPN mice, identified with similar symbols. The grey symbols in the left panel correspond to aVOR gain values associated with VAF<0.5. C, D) Individual ΔaVOR gains as a function of individual ΔOCR gains at W6 (C) and W12 (D) compared to W0, for SHAM (n=14) and IDPN (n=13) mice. The linear regression corresponds to IDPN values is represented, as well as the 50% confidence interval of each group (shaded areas). The symbols for each animal are the same in panels B, C and D. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To investigate further the organ-specific responses, the individual values of VOR gain at 1Hz (left panel) and OCR slope (right panel) are plotted in Figure 2B. These plots show the variability of the responses observed in different individuals. 2/3 of the mice had aVOR responses lowered by more than 50%, while 1/3 had milder aVOR impairments. In most cases, however, the aVOR gain decreased notably between W4 and W6 and started to recover from W8 (compare individual slopes in Figure 2B). Similarly, OCR gains were variable between individuals. However, the dynamic of the variations in otolith-dependent responses followed a pattern comparable to that of the canal-dependent changes.

+

To determine whether the changes in canal- and otolith-dependent responses were proportional, individual variations in gain since pre-treatment of the aVOR and OCR were compared at W6 (Figure 2C) and W12 (Figure 2D). At W6, there was a significant correlation between the amplitude of the changes observed in aVOR and OCR, such that canalar and otolithic loss of function was proportional (slope of regression line: 0.5586). Despite the general recovery, this significant correlation was preserved at W12: the recovery in canal-dependent responses was proportional to the recovery in otolith-dependent responses (slope of regression line: 0.5566). Overall, these results suggest that the subchronic IDPN treatment similarly affects the vestibulo-ocular reflexes that depend on semi-circular canals and on the otolithic organs, respectively.

+ + +Effects of the IDPN exposure on the number of hair cells in the vestibular epithelia +

To correlate the vestibular loss of function to the structural changes induced by the ototoxic compound, vestibular epithelia were dissected and labelled to assess the number of hair cells (HC) in the organs. One horizontal semi-circular canal and one utricule were harvested at W6 for n=7 IDPN and n=4 SHAM, and at W12 for n=8 IDPN and n=4 SHAM, and each organ epithelium was divided into the central and peripheral region to differentiate the possible participation of zone- and organ-specific HC to the vestibular function. Vestibular HC were labelled with type-specific immunomarkers in both endorgans (Figure 3). Confocal immunostaining data are presented in Figure 3A1 for the horizontal semi-circular canals and Figure 3C1 for the utricule. 6-week long treatment of IDPN exposure led to a significant reduction in the labelling of type I HC in the canal crista (quantified in Figure 3A2, Non parametric Kruskal-Wallis test, SHAM vs IDPN Spp1, p=0.0322; SHAM vs IDPN CASPR1, p=0.0268; Figure 3 supplement 3A1 and B1 for the peripheral zones) and in the otolith macula (Figure 3C2, Non parametric Kruskal Wallis test SHAM vs IDPN Spp1, p=0.0279; SHAM vs IDPN CASPR1, p=0.0343). However, the treatment did not alter type II HC-specific labelling as the number of Calre+ cells in the IDPN group was similar to the SHAM in all regions of either vestibular endorgans (Figure 3A2, canal crista, Calre: Non parametric Kruskal-Wallis test IDPN vs SHAM, p>0.9999; Figure 3C2, otolith macula, Calre: SHAM vs IDPN, p>0.9999). Importantly, the total number of HC marked with Myo7a (labelling both type I and type II HC) was not significantly different between the IDPN-treated mice and the SHAM, suggesting that the alteration of type I HC is not associated with cell loss. Moreover, at W12 the number of type I HC labelled in the IDPN-treated mice was no longer significantly different from the SHAM group in either endorgans (Figure 3A3 for crista; Figure 3C3 for macula). The count of type II-specific markers and of non-specific HC markers was not statistically different between SHAM and IDPN either, so that overall no difference persisted at W12, indicating a structural recovery at the end of the washout period.

+ + +Immuno-labelling of HC in the central regions of the horizontal SCC ampulla and striolar region of the utricule Macula. +

A, C) Immunolabelling of type I HC (Spp1+ and CASPR1+), type II HC (Calre +), or all HC (Myo7a) for the SHAM, IDPN W6 and IDPN W12 groups in the central ampulla of the horizontal canal (A1) and central utricular maculae (C1). Cell count at W6 and W12 in the central horizontal ampulla (A2 and A3) and central utricular maculae (C2 and C3) for individual mice. B1, D1) Individual number of central Spp1+ type I HC, CASPR1 + type I HC and Calre+ type II HC, or all HC(Myo7a) as a function of the aVOR gain (B1) or OCR gain (D1) at W6 (circle) and W12 (squares) groups. The linear regressions correspond to all individuals (n=23 mice). B2, D2) Comparison of the number of CASPR1 type I HC as a function of the aVOR (B2) gain or OCR gain (D2) at W6 (black and circle) and at W12 (pink and square) for each IDPN mice of the W12 group (n=8). Note that all points are shifted toward the regression line (redrawn from respectively B1 and D1), indicating that the number of cells at W12 better correlates with the recovered aVOR. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To determine how the loss of HC correlates with organ-specific functional tests, the numbers of type I and II HC in the central region of the horizontal semi-circular canals are plotted as a function of the mean aVOR gain for mice at W6 and W12 (Figure 3B1). Both markers of type I HC (CASPR1 and Spp1, left panels) were significantly correlated with the aVOR gain (linear regression, Spp1: r2=0.5128, p=0.0001; CASPR1: r2=0.4948, p=0.0002) whereas the rather constant number of type II HC in the ampulla did not correlate with the variation observed in aVOR function. Finally, to determine if the correlation was due to the recovery occurring between W6 and W12, the number of CASPR+ cells of IDPN mice counted at W12 are plotted as a function either of the individual aVOR gain at W12 (pink squares) or aVOR gain at W6 (black circles), paired by an arrow (Figure 3B2). The recovery of the vestibular function between W6 and W12 induces a shift toward the W12 linear regression, reinforcing the notion that the recovery of function was related to the increase in the number of hair cells with normal expression of biochemical markers.

+

The numbers of type I HC and II HC found in the striolar region of the utricule are similarly plotted in relation to the OCR gain (linear regression, Spp1: r2=0.6011, p<0.0001; CASPR1: r2=0.6342, p<0.0001, Figure 3D1). Again, the otolithic function correlated with the number of type I HC, and not type II HC in the central region. Furthermore, the correlation found between the otolithic function and the number of type I HC related to the increase in the number of hair cells after the recovery period (Figure 3C2). A similar correlation between organ-specific function and the number of type I HC was also found in the peripheral regions of these organs (see Figure 3 supplement 3C1, 3C2 for the SCC, and Figure 3 supplement 3D1, 3D2 for the utricule).

+

Taken together, comparison at W6 and W12 of non-specific HC and type I-specific HC markers suggest that the ototoxic effect induces a transient biochemical alteration of type I HC rather than a definitive hair cell degeneration. Overall, IDPN-induced alteration of canal and otolith functions are correlated with the ototoxic effect on type I, and not type II, HC in the central and peripheral zones of vestibular endorgans.

+ + +Effects of the subchronic treatment of IDPN on the optokinetic reflex +

To determine the effects of the subchronic treatment of IDPN on the optokinetic reflex (OKR), mice were tested with sinusoidal rotations of a virtual drum (Figure 4A). An example of raw traces of the evoked horizontal eye movements is shown in Figure 4B for different timepoints (0.5Hz at 10deg.s-1 optokinetic stimulation). The mean OKR gain for all tested frequencies is plotted in Figure 4C for SHAM (n=12) and IDPN-treated mice (n=12). The mean OKR gain of IDPN mice was significantly different from the mean OKR gain of SHAM mice at W8 (interaction between Weeks and Group F(6,132)=2.9845, p=0.0091; IDPN W0 vs W8, p=0.0216, see Table 2) where it reached its peak (IDPN W0 vs W8, p=0.00037). The gain stayed significantly larger compared to the SHAM through the end of the washout period (W12 IDPN vs SHAM, p=0.046).

+ + +Effects of subchronic IDPN treatment on the OKR. +

A) Illustration of the optokinetic reflex (OKR) set-up. B) Example of raw traces of an OKR recorded in response to stimulation at 0.5Hz at a peak velocity of 10°/s before (W0), after 4 weeks of IDPN treatment (W4), 2 weeks (W8) and 6 weeks of washout (W12). All traces are from the same individual. C) Mean OKR gain of IDPN (n=12) and SHAM (n=12) mice. D) OKR gain and (E) phase for IDPN (n=12) and SHAM (n=12) for each frequency at W0, W6, W8 and W12. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + + + +Statistics table of the OKR gain for the IDPN treated group. + + +

To determine whether this OKR modulation was frequency-specific, the gain of the 5 tested frequencies was compared for W0, W6, W8 and W12 (Figure 4D). There was at W0 no difference between SHAM and IDPN mice (Figure 4D, left panel, W0). At W6, there was a significant difference between the gain of the IDPN and SHAM groups limited to the frequencies 0.5 and 1Hz (W6 IDPN vs. SHAM, 0.5Hz, p=0.013117; W6 IDPN vs. SHAM, 1Hz, p<10-4). At W8, the gains measured at frequencies above 0.2Hz were significantly increased compared to the SHAM group (W8 IDPN vs. SHAM, 0.33Hz, p=0.027687; 0.5Hz, p=0.0025; 1Hz, p=0.00015). At W12, the gain remained high for both 0.5 and 1Hz whereas the gain at 0.33Hz was no longer significantly larger than the SHAM group (W12 IDPN vs. SHAM, 0.33Hz, ns; 0.5Hz, p=0.0069; 1Hz, p=0.00078). Globally, responses at higher frequencies (0.33, 0.5 and 1 Hz) were significantly increased (ANOVA Weeks x Group x Frequencies interaction, F(24, 528)=6.5870, p<10-4) whereas responses at the lowest frequencies (0.1 and 0.2Hz) were not significantly modulated. Additionally, these changes in gain for frequencies >0.2Hz were not accompanied by changes in the timing (phase) of the OKR (Figure 4E) (ANOVA, Weeks x Group x Frequencies interaction, p=0.3802).

+ + +Relation between VOR decrease and optokinetic increase +

To correlate the changes in vestibular pathway with the frequency-specific changes observed in the optokinetic pathway, we compared the responses in aVOR and OKR obtained at the frequencies that were common between the 2 tests (i.e 0.2 Hz, 0.5 Hz and 1Hz) for n=12 SHAM and n=12 IDPN treated mice by comparing the increase in the OKR and decrease in VOR relative to W0 (Δmean gain, see Figure 5). To determine whether the changes in VOR and OKR were proportional, we compared the paired decrease and increase in the mean ΔaVOR and ΔOKR gain after 8 weeks of treatment, when the difference between the 2 reflexes peaked (Figure 5A). The vast majority (n=11/12) of IDPN-treated mice showed a decrease in aVOR and an increase in OKR (top left quadrant corner, Figure 5B), which was not the case for SHAM mice (blue squares). However, there was no correlation between the amplitude of the VOR decrease and the amplitude of the OKR increase, i.e mice that had the greatest VOR loss did not show the largest OKR increase (slope in the regression line: 0.27, p=0.0834).

+ + +Comparison of the IDPN treatment on OKR and aVOR. +

A) Mean Δ aVOR and Δ Gain OKR for 0.2, 0.5 and 1Hz for IDPN (n=12) and SHAM (n=12) mice. B) Individual ΔOKR gains as a function of individual ΔaVOR gains. The 50% confidence interval of each group is represented in the shaded areas. C) aVOR and OKR gains of IDPN mice (n=12) at W0 and W8 for frequencies of 0.2, 0.5 and 1Hz. D) Percentage of the individual vestibular weight (inset), as a function of the percentage of the individual OKR gains change for IDPN (n=12). The linear regression corresponds to all values (n=36). (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To determine whether the parallel changes in OKR and VOR are frequency-specific, Figure 5C compares the gains measured at W0 and W8 for each frequency. At 0.2Hz, there was a significant decrease in aVOR (IDPN W0 vs. W8 0.2Hz p<10-4), however the OKR was not modified. On the other hand, at 0.5 and 1 Hz significant VOR decreases (IDPN W0 vs. W8, 0.5Hz, p<10-4; 1Hz, p<10-4) were accompanied by a significant OKR increase (IDPN W0 vs. W8, 0.5Hz, p<10-4; 1Hz, p<10-4). To examine this frequency-selective increase, the percentage of change in the OKR gain is plotted as a function of the vestibular weight, determined as the ratio between the aVOR and aVOR+OKR values at week 0. The vestibular weight therefore represents the frequency-dependent relative influence of the vestibular signal on gaze stabilization. Figure 5D shows that at week 8, the change in the OKR occurred at the frequencies for which the vestibular weight is dominant (>50%), and that the increase in OKR was positively correlated with the vestibular weight among frequencies (slope of the regression line: 1.235, r²=0.6299, p<0.0001). Overall, these results suggest that the increase in OKR gain observed at high frequencies could correspond to a “visual substitution” occurring primarily in the range where the vestibular inputs are normally dominating gaze stabilization.

+ + +Visuo-vestibular interactions following alteration of vestibular inputs +

To investigate whether OKR increase did constitute a “visual substitution” that maintained optimal gaze stabilization at light despite vestibular loss, we investigated how IDPN-treated mice integrated vestibular and visual inputs. To this end, a session of combined visual and vestibular stimulation (aVOR in light, here referred as Combined Gaze Response or CGR condition) was performed at W6 and at W12 (n=19 IDPN; n=12 SHAM). A model (see methods) was implemented to predict the theoretical CGR (visuo-vestibular) gain and phase from aVOR and OKR (unimodal) measured values. The model was first optimized to correctly predict the average SHAM CGR gains and phases (Figure 6A) from their individual aVOR and OKR responses, and then applied to the individual IDPN data. The predicted and observed CGR gain (Model factor) were compared for each treatment group (ANOVA Group x Model, F(1,29)= 12.236, p=0.0013): as expected, model predictions for SHAM well-matched experimental data (Post-hoc Newman-keuls, SHAM MODEL vs SHAM CGR, p=0.41). When applied to IDPN mice the model predicted that their CGR gain should be lower compared to SHAM (shaded blue and red areas on Figures 6A and 6B, Post-hoc Newman-keuls, SHAM MODEL vs IDPN MODEL, p=0.00176), suggesting that even an optimal combination of unimodal responses could not restore completely normal CGR in IDPN-treated mice. The CGR gains observed for IDPN-treated mice actually revealed how their visuo-vestibular responses were not only reduced compared to SHAM (compare solid blue and red lines in Figure 6A-B, Post-hoc Newman-keuls, SHAM CGR vs IDPN CGR, p=0.00055), but were even inferior to the prediction of the model (compare solid red line and shaded red area in Figure 6B, Post-hoc Newman-keuls, IDPN MODEL vs IDPN CGR, p=0.00041). This suggests that the IDPN-treated mice sub-optimally combined their residual/compensated vestibular and visual reflexes to stabilize gaze. Figure 6C illustrates the relationship between the measured CGR gain and the prediction of the model for all tested frequencies. For SHAM animals, CGR gain values were positioned along the unity line that represents a close match between predicted and measured CGR (mean±SEM: SHAM MODEL 0.8255±0.053; SHAM CGR 0.8375±0.042). For IDPN-treated mice, the responses formed 2 subgroups that did not correspond to different frequencies (all frequencies found in either subgroup). The first subgroup was intermingled with the SHAM responses, with measured CGR values slightly lower than predicted values (MODEL 0.7457±0.028; CGR 0.6644±0.052). The second subgroup represented mice for which the measured CGR significantly underperformed the predicted CGR (MODEL 0.4685±0.022; CGR 0.2005±0.01). To investigate the difference between the two subgroups, the VOR, OKR and CGR values were plotted for the n=19 IDPN mice (Figure 6D). A hierarchical cluster analysis (see methods) suggested the presence of two main clusters that differ from the clusters observed considering only the aVOR gain (see dendrogram in Figure 6D). The first cluster (Δ+, n=12; in green in Figure 6D, E, and F) comprised individuals with CGR values higher than OKR (positive values on Δ=CGR-OKR), i.e with gaze stabilization at light better in the presence of vestibular stimulation. The second cluster (Δ-, n=7; depicted in orange) is composed of individuals with low CGR (<0.4). IDPN mice from the Δ-cluster have degraded responses during bimodal visuo-vestibular stimulation compared to unimodal OKR stimulation (negative value of Δ indicates CGR responses lower than OKR responses). Notably, this degraded CGR is not solely a consequence of a low VOR gain, as some Δ+ mice identified by the second differentiation of the clustering analyses (depicted with grey diamond symbols in Figure 6D, and indicated as Δ+(-)) also showed low unimodal VOR gain responses.

+ + +Visuo-Vestibular interactions following IDPN treatment. +

A) aVOR, OKR, CGR data (CGR) and predicted CGR (MODEL) gains (left panel), and CGR phase (data and model, right panel), for the SHAM mice (n=12) at W6 and W12. B) aVOR, OKR, CGR data (CGR) and predicted CGR (MODEL) gains for the IDPN mice (n=19) at W6 and W12. C) Comparison of the predicted versus measured CGR for all frequencies tested. The dotted line at 45° represents a perfect match between prediction and data (optimal CGR). D) aVOR, OKR, and CGR values of the IDPN mice (n=19) (left panel). Clustering analysis (right panel) distinguish 2 groups based on Delta (CGR-OKR), and a subgroup based on VOR. The horizontal coordinate of each cluster (vertical lines) represents the distance between two connected clusters. E) Δ(CGR-OKR) as a function of the VOR Sweep Amplitude of the Δ+ (green triangle), Δ+(-) (green diamonds) and Δ- (orange) IDPN mice. The inset panel represents typical raw cycles representative of the 3 subgroups. F) Delta (CGR-OKR) as a function of the VOR reproducibility of the Δ+ (green triangle), Δ+(-) (green diamonds) and Δ- (orange) IDPN mice. Regression line correspond to all values (n=19)

+ + +

We reasoned that the worsening of the gaze responses in the bimodal condition compared to the OKR-only condition could be a consequence of a poorly reliable, noisy vestibular signal. Based on a cycle-to-cycle sweep analysis, two additional features of the VOR responses were quantified (see methods): the mean amplitude of the raw sweep during VOR stimulation and the VOR response reproducibility between sweeps. The mean VOR raw sweep amplitude could not statistically differentiate Δ- and Δ+(-) mice (Figure 6E, orange triangle vs green diamonds p=0.0857), suggesting that vestibular signal of comparable amplitude improved or deteriorated the CGR response in the Δ+(-) and Δ-mice, respectively. On the other hand, the VOR reproducibility clearly discriminated Δ- and Δ+(-) mice (Figure 6F, orange triangle vs green diamonds, p=0.0001). There was a strong correlation between the inter-sweep reproducibility and the capacity of the mice to stabilize gaze in presence of vestibular inputs at light. This result suggests that the incapacity of the Δ-group to compute optokinetic signals in the presence of vestibular stimulation could relate to the poor reliability of the vestibular signal. Overall, these data suggest that in some situations the presence of a degraded vestibular signal of significant amplitude (panel 6E) but poorly reliable (panel 6F) can be detrimental to properly stabilize gaze in presence of visual and vestibular inputs.

+ + + +Discussion + +IDPN as a model of partial and transitory loss of vestibular function +

IDPN has long been established as an ototoxic compound targeting vestibular HC in the different vestibular endorgans of rats (Llorens et al., 1993, 1994; Llorens and Demêmes, 1994) and later in guinea-pig, frogs and mice (Soler-Martin et al., 2007; Greguske et al., 2019). Its ototoxic vestibular effects were used as a tool to study extraocular muscle development (Brueckner et al., 1999) or more recently to induce permanent vestibular loss in mice (Yang et al., 2019; Zeng et al., 2020) and study hair cell regeneration (Sayyid et al., 2019). The sub-chronic, reversible protocol used in our study was validated previously in rats (Sedó-Cabezón et al., 2015; Martins-Lopes et al., 2019; Maroto et al., 2021) and mice (Boadas-Vaello et al., 2017; Greguske et al., 2019) with postural/locomotor quantification of vestibular loss. We for the first time demonstrate that the subchronic protocol leads to a progressive and partly reversible loss of vestibulo-ocular reflexes. These previous studies further demonstrated some of the cellular mechanisms associated with the progressive loss of postural control: ototoxic effects lead to the early dismantlement of calyceal junction, followed by synaptic uncoupling, both of which were shown to be reversible, while continuation of the IDPN treatment would lead to hair cell extrusion and long-term, permanent lesion. HC loss was also demonstrated to occur in a central to peripheral order within vestibular epithelia, and in crista to utricule to saccule order (see Sedó-Cabezon et al. 2015). Maroto et al (2021) convincingly demonstrated that type I HC show greater sensitivity than type II HC to IDPN subchronic exposure. Given these data, a primary goal of the present study was to try to correlate the loss of VOR functions to organ-specific, zone-specific, and cell-type-specific effects.

+

Loss and substantial recovery of the vestibular function quantified by VOR measures were found to be correlated with the number of type I HC in both canals and otoliths. A parallel was previously established between HC integrity and VOR following ototoxic protocols in mice (Cassel et al., 2019; Yang et al., 2019; Zeng et al., 2020), but these were established at the population level. Here, we for the first time specifically correlated the loss of HC and loss of VOR function both at the individual level and in an organ-specific way. aVOR and OCR tests demonstrated a parallel decrease in canal- and otolith-dependent functions, respectively. The 6 week-long treatment was followed by a 6 week-washout period, allowing a significant but partial recovery of the aVOR, and complete recovery of the OCR. The individual correlations between these functional tests and the number of HC counted in the ampullae and maculae of the mice were found to be particularly significant, both during the trough (W6) of VOR function and after the recovery period (W12) (Figure 3). We, however, did not find any difference between the HC loss in the central vs. peripheral zones of the organs, and can therefore not conclude on any putatively differential implication of these areas in the VORs. Effects of the IDPN treatment on the aVOR was evidenced by a gain decrease starting W4, later associated with a significant phase lead at W6 (Figure 1 supplement 1A,1B). The dynamic of these changes might be related to the amount of HC progressively affected by the treatment. The modifications affected all frequencies; however, responses at the lowest tested frequency (0.2Hz) tended to be less affected and to recover first. These results could suggest that type I HC are crucial element to encode both the amplitude and timing of the VOR, particularly for the more dynamic stimulations. How IDPN progressively impairs the encoding of information by the HC, and how population-coding influence both VOR features in the natural range of head-movements should be the focus of future dedicated studies.

+

Overall, these organ-specific structure-function correlations confirm that the vestibulo-ocular reflex can serve as a proxy indicating the status of the vestibular endorgans. It further validates the use of the aVOR and OCR as relevant tests reflecting HC integrity at the level of the ampullae and maculae, respectively.

+ + +Differential alteration of type I versus type II hair cells +

The subchronic treatment affected type I HC more than type II HC, in accordance with previous reports (Llorens et al. 1993; Maroto et al. 2021). Susceptibility to IDPN was also reported to vary between different mouse strains (Boadas-Vaello et al., 2017; Wilkerson et al., 2018). Interestingly, we used a different strain of mice than the one used in a previous subchronic experiment (Greguske et al. 2019) and have not observed any type II HC loss either, nor differences between males and females (data not shown).

+

Hence, IDPN treatment induces a loss of cell markers specific to type I HC, with no effects on type II markers. Two types of markers were used to identify type I HC. Spp1 targeted a protein located on the neck of the hair cell (McInturff et al., 2018), while CASPR1 is located at the calyceal junction of the afferent terminal, and has been proven to be necessary for the functionality of the synapse between the HC and its connected afferent (Sousa et al., 2009).

+

Though it is not possible to compare for one individual the number of cells before and after treatment, the number of cells labelled for either of these proteins decreased significantly compared to SHAM group, whose numbers are consistent between W6 and W12. The loss of CASPR1 marker induced by IDPN sub-chronic treatment has been linked to HC detachment from the calyx terminal and loss of vestibulo-spinal reflexes, and both CASPR1 expression and vestibular function recover during washout (Sedó-Cabezón et al., 2015; Greguske et al., 2019; Maroto et al., 2021). Only after a longer IDPN exposure does HC extrusion occur, associated with persistent functional deficits. In theory, the loss of vestibular function observed in both canal- and otolith-dependent VOR at W6 could be linked to loss of type I HC. In adult rodents, however, the regeneration of hair cells leads to the formation of cells with type II HC features (González-Garrido et al., 2021), so the recovery of type I cellular markers at W12 is likely not a result of cell regeneration. Also, the apparent dramatic loss of CASPR1 and Spp1 HC (90%) at W6 was in contrast with the much smaller and not significant loss in Myo7a cells (29%), supporting the conclusion that the sensory cells persisted in the epithelium despite CASPR1 loss, and that the recovery in the number of type I HC observed at W12 was more likely the result of molecular repair, not cell regeneration. In addition, we observed a similar loss of Spp1 marker at W6, normally located in the neck of the hair cell. As such the loss of vestibular function could rather be attributed to a global disorganization of the synapse and defective hair cell function, associated with a transitory large decrease of the Spp1 and CASPR1 proteins and a smaller decrease in Myo7a. In accordance with this hypothesis, the number of Spp1 and CASPR1 positive cells after the washout period at W12 was not significantly different from the SHAM group and correlated to the recovery of the vestibular function observed at W12 for the IDPN mice (Figure 3C2). The altered expression of proteins in type I HC induced by the sub-chronic treatment of IDPN seems to be reversible in most individuals. One possibility is that definitive extrusion of type I HC occurred in the most susceptible individuals that did not show recovery after the end of treatment. This in fact was the case in the 2 individuals tested at W12 with immuno-histochemistry: CASPR1 mean value (±S.E.M) of 18.5±5.5 compared to 34.5±4.359 for IDPN with recovery and 41.75±1.215 for SHAM). It was previously reported in the pigeon that type I HC and calyx afferents take longer (12 weeks) to regenerate following aminoglycoside toxicity, while type II and boutons endings regenerated in a week (Zakir and Dickman, 2006). Similarly, recovered innervation for type I HC was delayed compared to type II HC in mice (Kim et al., 2022). A longer recovery period could be investigated in the case of the severely-affected mice to confirm the absence of recovery in the longer term.

+ + +Evidence for the role of type I HC in the vestibulo-ocular pathways +

To our knowledge, the direct implication of type I and type II HC in the VOR was never directly tested. Type I and type II HC differ by many features, including their anatomical location within the epithelium, electrophysiological properties, morphology and innervation by afferents (Eatock and Songer, 2011). Seemingly, irregular afferents have been described as predominantly innervating the central area and striolar zones of the ampullae and maculae with calyx and dimorphic synaptic contacts, while regular afferents make dimorphic and buttons contacts predominantly within the peripheral zones of cristae and extrastriolar zones of maculae (Goldberg, 2000; Eatock and Songer, 2011; Contini et al., 2022). Afferents with bouton terminals that contact only type II HC have regular discharge, while afferents with calyx terminals that contact only type I HC have irregular discharge. However, most regular and irregular afferents are fed by both type I and type II HC (Goldberg et al., 1992). Functionally, regular afferents recorded in the monkey show a lower detection threshold than irregular afferents (Sadeghi et al., 2007), while irregular afferents showing higher gains and phase leads (mouse: Lasker et al., 2008; Cullen, 2019) would be better optimized for encoding high dynamic stimuli (Cullen, 2019). Central VOR neurons receive a mixture of regular and irregular afferent inputs (Highstein et al., 1987; Boyle et al., 1992; Goldberg, 2000), with irregular afferents constituting ∼1/3 of their excitatory drive (Highstein et al., 1987; Boyle et al., 1991).

+

While the role of type I and type II HC in the VOR remains to be determined, previous studies have emphasized the importance of the regular afferents and regular central vestibular neurons in the vestibulo-ocular pathway. Functional ablation targeting irregular afferents suggested that VOR might function normally with only intermediate and regular afferent inputs (Minor and Goldberg, 1991). Functionally identified VOR neurons (i.e. PVP) were recently demonstrated to have heterogeneous discharge variability (high or low), with the most regular units particularly well-suited to faithfully encode the compensatory eye movements generated during natural stimulation (Mackrous et al., 2020).

+

Based on this literature, it would be tempting to infer that only the most regular and tonic elements of the entire vestibular pathway are responsible for the VOR. However, the correspondence between hair cell properties and afferents/central neurons properties is only partial, such that the two phasic/irregular and tonic/regular channels for head motion signals are constituted of both types I and type II HC (Baird et al., 1988; Goldberg et al., 1990 p.90) and afferents (irregular and regular) (Goldberg et al. 2000; Eatock and Songer 2011; Beraneck and Straka, 2011). Carey and colleagues (1996) previously reported a better correlation of VOR recovery with type I than with type II HC following ototoxic lesions in the chick. Overall, the result of ototoxic studies, including the present one, demonstrate a fundamental role of type I HC in the encoding of vestibular signals that drive the vestibulo-ocular reflexes, even in the relatively low range (Carriot et al., 2017) of head movements tested. It has also been recently shown that both type I and type II HC actually contribute to otolithic vestibular evoked potential responses (i.e. vestibulo-spinal pathway), previously described as mostly type I-specific (Sayyid et al, 2019). Overall, our results are compatible with the hypothesis of a convergence of heterogeneous peripheral neural elements at the level of central vestibular nuclei, where intrinsic properties of central vestibular neurons (Straka et al., 2005; Beraneck and Idoux, 2012) supplemented by network properties (Beraneck et al., 2007; Pfanzelt et al., 2008; Beraneck and Straka, 2011) would differentially integrate vestibular signals further processed in the different functional pathways (Sadeghi and Beraneck, 2020; Mackrous et al. 2020).

+ + +Visuo-vestibular interactions after IDPN treatment +

OKR plasticity following vestibular loss concerned gain, and not phase, modifications for frequencies >0.33Hz. Previous studies have reported an increase in OKR gain following a permanent vestibular lesion for a non-specific range of frequencies (Shinder et al., 2005; Faulstich et al., 2006; Nelson et al., 2017). One key to understand the frequency-specific OKR plasticity could be the physiological dominance of vestibular inputs in the gaze stabilization process at high frequencies (Faulstich et al., 2004). The visual inputs could be reweighted and potentiated specifically in the range where the vestibular loss has the most impact on gaze stabilization (Figure 5D).

+

The VOR and OKR work synergistically to stabilize gaze by compensating for head and visual surround movements, respectively (França de Barros et al., 2020). If the reflexive eye movements are not perfectly compensatory, an error signal (e.g retinal slip) is produced that drives adaptation of the VOR (Boyden et al., 2004; Dean and Porrill, 2014; Shin et al., 2014) and OKR (Glasauer, 2007; Kodama and du Lac, 2016). In an effort to understand the integration of VOR and OKR, Holland and colleagues (2020) recently proposed that OKR would account for the retinal slip not compensated by VOR. Within that framework, the changes in the OKR could happen preferentially as a function of the vestibular weight, i.e in the range of movements where the loss of vestibular inputs generates the largest retinal slip. While present data are evidence for frequency-specific adaptation, how this relates to alteration in vestibular inputs and/or visual feedback signals remains to be determined.

+

The increase in OKR gain could be seen as a substitution for the decrease in the VOR. However, even after VOR recovery, the optimal integration of OKR and VOR (as predicted by our MODEL) led to a degraded combined gaze response (CGR) with respect to SHAM mice. Even worse, the measured CGR was inferior to the predicted CGR (Figure 6B). This observation suggests that the integration of visual and vestibular signals during CGR is more than a mere summation of the gain and phase of the 2 unimodal reflexes, but is also affected by other factors. In fact, a subset of the IDPN individuals showed severely degraded bimodal responses, so that their combination of VOR and OKR is not only sub-optimal but also less effective than the unimodal (OKR) reflex. A sweep-based analysis suggested that in these individuals, the VOR unimodal responses were not reliable, i.e the low reproducibility indicates that the same stimuli are unfaithfully encoded into eye movements. We interpret this low reproducibility as a sign of noisiness within the VOR pathway that would not only preclude the optimal integration of both sensory signals, but also disturb the use of retinal information to stabilize the gaze. According to statistical optimality theories, as the Maximum Likelihood principle, the decrease in VOR reliability should lead to a reduction in its relative sensory weight. This filtering-out of a degraded signal has been shown during visuo-vestibular integration for monkey heading perception (Fetsch et al., 2010), or for human object discrimination using visual and haptic senses (Ernst and Banks, 2002).Our results suggest that the gaze stabilizing system is not able to optimally adapt to the degradation by filtering-out vestibular signals during combined visuo-vestibular combination.

+

The integration of visual and vestibular inputs occurs in several structures, including in the cerebellar flocculi (Jang et al., 2020) and brainstem (Carcaud et al., 2017) (see review De Zeeuw et al., 2021). In the case of vestibular loss, most of the defects are expected to concern the central vestibular neurons involved in the VOR, which also integrate visual inputs (ES, or eye-sensitive neurons; Beraneck and Cullen, 2007). After unilateral neurectomy, vestibular neurons in the Deiters nuclei responded to higher frequencies during visual stimulation (Cat: Zennou-Azogui et al., 1994), a change compatible with the hypothesis that OKR gain increase partly takes place in the vestibular nuclei. It was shown that inhibitory floccular inputs (part of the OKR indirect pathway) and excitatory vestibular inputs often colocalized on the dendrites of central vestibular neurons. One possibility is therefore that the massive disorganization of vestibular periphery inputs led in the long term to synaptic and intrinsic changes at the level of central vestibular neurons (Beraneck and Idoux 2012; Carcaud et al. 2017), thus impairing their capacity to optimally integrate both signals.

+ + +Conclusions +

Balance dysfunction occurs frequently in aging people (85% prevalence above 80; Agrawal et al., 2013) but also in younger people (prevalence of 35% in the 40 year-old). While animal models often use permanent vestibular lesions (Simon et al., 2020), many diseases in fact consist in a gradual, transient and/or partial loss of vestibular function. For instance Menière’s disease, which represents ∼9% of all vestibular pathologies in the adult and occurs in the <60 years old, is characterized by recurrent episodes with brief (<24h) fluctuating symptoms and otherwise normal vestibular function before long-term deterioration arises (Lopez-Escamez et al., 2015). Transitory vestibular symptoms are also commonly reported in vestibular neuritis where, although symptoms tend to rapidly disappear due to vestibular compensation, vestibular function can recuperate up to a year after the initial loss (Welgampola et al., 2019). Gradual and partial vestibular loss is also encountered as a side-effect of some ototoxic anti-cancer treatments (prevalence in treated patients ∼40%; cisplatin; Paken et al., 2016; Prayuenyong et al., 2018). Here, we took advantage of a subchronic ototoxic protocol to determine how animals adapt to partial and transitory loss of vestibular hair cells. We show that the loss, and then the partial recovery of the VOR, is correlated with the integrity of the type I HC, demonstrating for the first time their essential role in the VOR, whether of canalar or otolithic origin. We show that sensory (visual) substitution would theoretically compensate for vestibular loss, but that injured mice have suboptimal responses when combining visual and vestibular information. Finally, we show that this impairment in multisensory integration would be linked to the loss of "reliability" of the vestibular signal, degraded by ototoxicity. Overall, these results suggest that transitory peripheral infraction have long term consequences, and that the capacity of central vestibular structures to cope (vestibular compensation; Cullen et al., 2009; Beraneck and Idoux, 2012;(Lacour et al., 2016) with the sensorineural loss might critically depend on the integrity of the neural elements involved. Future studies should aim at obtaining information about the degradation of signal transmission following IDPN treatment and characterize the amount of peripheral population-coding necessary to preserve optimal vestibular function.

+ + + +Material & Methods + +Ethics +

A total of 56 C57/BL6J mice in an equal partition of males and females were used for the protocol. Animals were used in accordance with the European Communities Council Directive 2010/63/EU. All efforts were made to minimize suffering and reduce the number of animals included in this study. All procedures were approved by the ethical committee for animal research of the Université Paris Cité.

+ + +Headpost implantation surgery and animal care +

Surgical procedures, postoperative care, device fixation and animal surveillance during the protocol were performed as described previously in França de Barros et al., 2019, 2020. Briefly, 6-weeks-old mice anaesthetized with isoflurane gas had their heads shaved with small clippers. Then, lidocaine hydrochloride (2%; 2mg/Kg) was injected locally before a longitudinal incision of 1.5 cm was made to expose the skull. Just anterior to the lambda landmark, a small custom-built headpost (3x3x5mm; poly lactic acid) was cemented (C&B Metabond; Parkell Inc., Edgewood, NY) and laterally covered with resin (Heraeus) for protection. Animals were fully recovered 30 min after the end of the surgery, yet buprenorphine (0.05mg/kg) was provided for postoperative analgesia and they were closely monitored for the following 48 hours.

+ + +Subchronic ototoxic exposure +

28 mice were treated with 3,3ʹ-iminodiproprionitrile (IDPN, Sigma Aldrich, 317306, 90%) for 6 weeks (IDPN group) at 30mM concentration of IDPN in their drinking water at least 72h after the surgery. After 6 weeks of treatment, a washout period of 6 weeks followed. Previous experiments (Greguske et al., 2021) had demonstrated that at these concentrations, ototoxic lesions produced by IDPN are largely reversible. The SHAM group was tested the exact same way but was not exposed to IDPN in their drinking water. Video-oculography tests were performed before the beginning of the treatment and once every two weeks until week 12 (W12), for a total of 7 sessions of tests for each mouse.

+ + +Experimental groups +

3 different batches of mice were used, each composed of both IDPN-treated mice and SHAM. The mice of the first one (n=19) were subjected only to vestibular stimulations, to test the effect of IDPN on the canal-related or otolith-related VOR responses during the 12 weeks of protocol. The second group (n=24), designed to compare the dynamic of VOR and OKR changes was tested with angular VOR stimulations, otolithic test, optokinetic (visual) stimulations, and combined vestibular and visual stimulation (VOR in light; referred to as Combined Gaze Response or CGR). The third group (n=11) designed to correlate ototoxic effects of IDPN on vestibular hair cells (HC) and VOR function was tested for canal-related or otolith-related VOR responses, before treatment and after 6 weeks of treatment and immediately used for immunohistochemical assessment of the vestibular sensory epithelia (see below). A subset of the second group (n=12 out of 24) was also used for immunohistochemical assessment at the end of the 12 weeks of protocol.

+ + +Video oculography recording sessions +

The recording procedure was similar to the one presented in Carcaud et al. 2017 & França de Barros et al. 2020. The animals were head-fixed in a custom build Plexiglas tube, and their head was oriented in a 30° nose down position to align the horizontal canals to the yaw plane. Their left eye movements were recorded using a non-invasive video oculography system (ETL -200 Iscan, acquisition rate 120Hz, Stahl et al 2000). Eye, head, and virtual drum (OKR stimulation) position signals were digitally recorded (CED mk3 power 1401, acquisition rate 1kHz) with Spike 2 software. Signals were analysed offline in Matlab (Matlab, the MathWorks, Natick, MA, USA; RRID, SCR:001622) programming environment. The restraining apparatus was fixed on a rotation platform on top of an extended rig with a servo-controlled motor. Single VOR or OKR recording sessions lasted no longer than 45 minutes in total.

+ + +Vestibular stimulations and analysis +

All vestibular-specific tests were performed in a temperature-controlled room with all sources of light turned off except for computer screens. The turntable was further surrounded with a closed box to isolate the animal from remaining light, with a final luminance inside the box <0.02 lux. Myosis was induced with topical 2% pilocarpine applied 10 minutes before experimentation.

+

Vestibulo ocular reflex in dark (VORd) tests were performed in the dark with the mouse surrounded by an opaque black dome (Fig1.A). Sinusoidal angular rotations around the vertical axis were used to record the horizontal angular vestibulo-ocular reflex (aVOR), at different frequencies: 0.2, 0.5, 0.8; 1 and 2Hz at a peak velocity of 30°/s.

+

Angular vestibulo-ocular reflex analysis was similar to the one described in Carcaud et al. 2017. Segments of data with saccades were excluded from VOR slow-phase analysis. For horizontal sinusoidal rotations, at least 10 cycles were analyzed for each frequency. VOR gain and phase were determined by the least-squares optimization of the equation: + + + + + +where EHv(t) is eye horizontal velocity, g (gain) is a constant value, HHv (t) is head horizontal velocity, td is the dynamic lag time (in msec) of the eye movement with respect to the head movement, and Cte is an offset. The td was used to calculate the corresponding phase (φ°) of eye velocity relative to head velocity. The Variance-Accounted-For (VAF) of each fit was computed as + + + + + +where var represents variance, est represents the modeled eye velocity, and EHv represents the actual eye horizontal velocity. VAF values for VOR measures were typically between 0.70 – 1 (>95% of recordings), where a VAF of 1 indicates a perfect fit to the data. For IDPN-treated mice, abnormally low (<0.10) values of gain associated with VAF<0.5 were nevertheless included in the gain statistical analysis but specifically reported in grey (Figure 2 & Figure 1 supplement). Corresponding phase values were not included in the statistical analysis of the aVOR phase.

+

Static Ocular Counter-roll reflex tests were performed as described in (Simon et al., 2021) by measuring the vertical eye movement generated in response to different roll angles. The table was moved from left to right in incremental steps of 10° (from 0 to 40°), with static periods of at least 10s between oscillations (Fig1.D) to record the stabilized eye elevation and declination. The OCR gain corresponds to the slope of the linear regression of the vertical eye angle vs. the head tilt angles (Oommen and Stahl, 2008).

+

Off vertical axis rotation (OVAR) tests were performed with the vestibular turntable tilted with a 17° off-axis angle following the methodology described in Beraneck et al., 2012 & Idoux et al., 2018. 50°/s continuous stimulations were performed in a counter-clockwise and then clockwise direction. OVAR generates a continuous change in the head tilt angle, compensated through a maculo-ocular reflex (MOR) by the generation of a horizontal nystagmus compensating for the table constant rotation. This oculomotor response was quantified following the methodology described in Idoux et al 2018. First, quick-phases were identified and removed. During rotations, the velocity of horizontal slow phases is modulated (modulation, μ) around a constant bias (β). Both parameters (μ and β) were calculated from the sinusoidal fit of eye horizontal slow-phase velocity using the least-squares optimization of the equation: + + + + + +where SP(t) is slow-phase velocity, β is the steady-state bias slow phase velocity, μ is the modulation of eye velocity, f0 is the frequency of table rotation, td is the dynamic lag time (in msec) of the eye movement with respect to the head movement. The bias (Maculo-ocular reflex Bias) is reported here as the main index of the MOR response (Hess and Dieringer, 1990; Beraneck et al., 2012).

+ + +Optokinetic reflex tests and analysis +

Horizontal optokinetic stimulations were performed as previously described in França de Barros et al. 2020. The mice were placed under a semi-opaque plastic dome and all sources of light were turned off. The projected stimulation consisted of a randomly distributed white dots pattern on a black background image (250000 white dots, max width 0.075°). The optokinetic sinusoidal stimulations were tested at 0.1, 0.2, 0.33, 0.5 and 1Hz at a peak velocity of 10°/s. The gain and phase were obtained by the same least-squares optimization method described above for the aVOR. To prevent putative cross effects between visual and vestibular stimulations, VOR and OKR test sessions were performed on separate days.

+ + +Combined visual and vestibular stimulations +

Combined visual and vestibular stimulations measuring the combined gaze reflex (CGR) consisted of horizontal vestibular stimulations while projecting the fixed dotted pattern used for OKR on the surrounding dome. Horizontal angular sinusoidal rotations were performed at frequencies of 0.2, 0.5 and 1Hz with a peak velocity of 30°/s. The CGR gain and phase quantifications were performed following the same methodology as for aVOR. To avoid interference, these were performed on n=8 SHAM and n=12 IDPN after the end of the VOR and OKR test sessions on W12. To test the effects at W6, a specific group of mice (n= 4 SHAM and n=7 IDPN) was tested at W0 and W6.

+

A simple model has been developed to predict for each individual and for three stimulation frequencies (0.2, 0.5 and 1Hz) the theoretical CGR response gain and phase were given the same two parameters observed for their unimodal VOR and OKR responses. The experimental VOR and OKR gain and phase are used to compute the sinusoid representing the two unimodal responses. The gain and phase of the CGR sinusoid resulting from adding the VOR and OKR sinusoids are quantified by measuring its peak-to-peak amplitude and by identifying the time-lag that maximizes its cross-correlation with the sinusoidal stimulus (note that the sum of two sinusoids with the same period, even if with different phases, is always a sinusoid). Independently of the mouse, the unimodal VOR and OKR sinusoids are shifted in time before their summation, in order to improve, on average, the fit between predicted and observed CGR responses for SHAM mice. The two optimal shift parameters, one for VOR and one for OKR, obtained for each frequency, are then used also for all IDPN-treated mice.

+

To identify subpopulations of IDPN-treated mice, we computed an agglomerative hierarchical cluster tree on a dataset composed of individual aVOR gain and Δ (calculated as CGR-OKR gain). The classification method used unweighted average Euclidean distance between clusters.

+

To describe the mice VOR responses without any assumption about their sinusoidal nature the two following characteristics have been quantified: the reproducibility of the ocular response to the vestibular stimulation and the amplitude of the raw eye movements generated by the vestibular stimulation. These analyses were performed on the slow phases horizontal eye trajectories, EHp(t), recorded during each sweep of stimulation.

+

The reproducibility index, Rep, was obtained by computing the matrix of correlation coefficient, Ri,j, between each couple (i,j), of the N selected sweeps, and then computing the mean of all above-diagonal elements of the matrix as reported in the following equation (this choice aims at considering only once each Ri,j, since Rj,i=Ri,j and to exclude diagonal elements, Ri,i =1). + + + +The amplitude, Amp, of the raw eye movements was quantified by computing for each sweep how much on average, during a stimulation cycle, the eye moved horizontally away from its average position (σEHp). Before averaging this parameter over all the sweep, it had to be squared. Finally, to have a parameter value in degrees, the root mean squared of the mean was computed. + + + +

+ + +Immunolabelling of the vestibular HC +

Two groups of mice (n=11 at W6 and n=12 at W12) were used to perform immunofluorescence analysis on hair cells in the vestibular endorgans. Mice were anaesthetized with an overdose of intraperitoneal injection of ketamine hydrochloride (10%)/xylazine (5%) and decapitated. The histology was done following the protocol established by Lysakowski et al. (2011), as described previously (Maroto et al., 2021). The vestibular epithelia (one horizontal canal and one utricule) were dissected and fixed for 1h in a 4% solution of paraformaldehyde (PFA). PFA was washed twice with phosphate-buffered saline (PBS) and the samples were then embedded in a cryoprotective solution (34.5% glycerol, 30% ethylene glycol, 20% PBS, 15.5% distilled water) to be stored at -20°. Before the immunochemistry, samples were put at room temperature and rinsed twice in PBS. While under slow agitation, the samples were incubated twice, first for 1h with a 4% Triton X-100 (Sigma Aldrich) in PBS to permeabilise and a second time for 1h in 0.5% Triton X-100 1% fish gelatin (CAS #9000-70-8, Sigma-Aldrich) in PBS to block. The incubation with the primary antibodies was then performed in 0.1M Triton X-100, 1% fish gelatin in PBS at 4° for 24h. After rinsing, the second antibodies were incubated in the same conditions. The 2nd antibodies were rinsed and the vestibular epithelia were mounted on slides with fluoromount (F4680, Sigma-Aldrich) and were visualised with a confocal microscope Zeiss LSM880 (with an objective of 63x NA:1.4). To properly analyse the whole vestibular epithelium, Z-stacks of 0.5 μm were obtained and observed with ImageJ (National Institute of Mental Health, Bethesda, Malyland, USA). The primary antibodies used were rabbit anti-Myosin VIIa (Myo7a) (Proteus Biosciences, #25-6790), mouse anti-contactin-associated protein (CASPR1) (Neuromab #75-001), guinea pig anti calretinin (Synaptic Systems #214-104) and goat anti-osteopontin (Spp1) (R&D Systems #AF08). Their respective secondary antibodies were Dylight 405 donkey anti-rabbit igG H+L (Jackson Immuno Research #711-475-152, Alexa Fluor 488 donkey anti-mouse IgG H+L (Life Technologies #A21202), Alexa Fluor donkey anti-guinea-pig IgG H+L (Jackson ImmunoResearch #706-605-148) and Alexa Fluor 555 donkey anti-goat IgG H+L (Invitrogen #A21432).

+

HC counts were obtained from square areas (67.5 x 67.5 µm) of the central and peripheral parts of two vestibular organs (horizontal semi-circular canal and utricle) in the W6 group and the W12 group. The areas for counting were obtained from the same location within the epithelium in all animals. The global number of HC was assessed with the cytoplasmic labelling of the anti-Myo7a antibody (Hasson et al., 1997). Type I HC were labelled using two different antibodies: anti-Spp1 in the neck of the type I HC (McInturff et al., 2018) and the presence of anti-CASPR1 in the calyceal junctions of the calyx (Sousa et al., 2009 p.20). Type II HC were distinguished by the colocalization of Myo7a and calretinin (Maroto et al, 2021).

+ + +Statistical analysis +

For both OKR and VOR stimulations, the effect of the protocol on the gain and phase were statistically tested by performing repeated measures ANOVA. The treatment (SHAM or IDPN) was considered as between individual independent factor with the Weeks (W0, W2, W4, W6, W8, W10, W12) and the Frequencies (0.1, 0.2, 0.33, 0.5 and 1Hz for the OKR; 0.2, 0.5, 0.8, 1 and 2Hz for the VOR) were considered as within individual independent factors. The main effects of those factors and their interactions were tested and reported. In the case of the OCR and OVAR, mix-model ANOVA was used with only Weeks considered as within factors. For the CGR, measured at W6 and W12, mix-model ANOVA was used with the Frequencies and Model (Measured or Model CGR) used as within factors. For the comparison between the OKR and VOR Δ gain, a repeated measures ANOVA was applied on the Δ gain (Wx-W0) with the stimulation modality (OKR or VOR) as between individual independent factor and the Weeks as within individual independent factor. For the correlation between OCR and VOR, as well as the measured and theoretical CGR gain a linear regression model was fitted.

+

The effect of IDPN exposure on the HC count was reported with a Kruskal Wallis test. For all analyses the significance threshold was set at p<0.05 and Newman Keuls post hoc test was performed if the main effect or an interaction was found significant.

+ + + +Funding +

This work is supported by the Centre National d’Etudes Spatiales, the Centre National de la Recherche Scientifique, and the Université Paris Cité. MB and LS received funding from the Agence Nationale de la Recherche (ANR-20-CE37-0016 INVEST). MB, FS & DF received funding from the ERANET NEURON Program VELOSO (ANR-20-NEUR-0005). AP & JL received funding from the ERANET NEURON Program VELOSO (grant PCI2020-120681-2 from MCIN/AEI/10.13039/501100011033 and NextGenerationEU/PRTR).

+ + + + +<bold>Acknowledgment</bold> +

this study contributes to the IdEx Université de Paris ANR-18-IDEX-0001. This work has benefited from the support and expertise of the animal facility of BioMedTech Facilities at Université Paris Cité (Institut National de la Santé et de la Recherche Médicale Unité S36/Unité Mixte de Service 2009). The confocal microscopy studies were performed at the Centres Científics i Tecnològics de la Universitat de Barcelona (CCiTUB).

+ + +Author contribution +

Conceptualization: JL ; MB. Methodology: LS ; AP ; FS ; MT ; JL ; MB. Formal analysis: LS ; AP ; MT ; JL ; MB. Investigation: LS ; AP ; TB ; CEB . Writing original manuscript: LS ; MT ; MB. Writing review and editing: LS ; AP ; FS ; DF ; MT ; JL ; MB. Visualization: LS ; AP ; TB; MT ; JL ; MB. Supervision: LS ; DF ; MT ; JL ; MB. Project administration: JL; MB. Funding acquisition: DF, JL; MB.

+ + +Declaration of interests +

The authors declare no competing interests.

+

Figure 1 supplement: Effects of subchronic IDPN on canalar- and otolith-dependent VOR.

+

A, B) aVOR (A) Gain and (B) Phase across all frequencies and weeks tested for IDPN (n=21) and SHAM (n=22) mice. The grey symbols in the left panel correspond to aVOR gain values associated with VAF<0.5.C) OVAR MOR bias for IDPN (n=13) and SHAM (n=14) mice during the protocol. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+

Figure 3 supplement: Effects of the IDPN on the number of HC in the peripheral regions of the horizontal SCC ampulla and extrastriolar utricule Macula.

+

A) Individual cell count of peripheral horizontal semi-circular canal type I and type II HC for W6 (1) andW12 (2). B) Cell count of peripheral utricular type I and type II HC for W6 (1) and W12 (2). C) Individual number of peripheral Spp1+ type I HC (1), CASPR1 + type I HC (2) and Calre+ type II HC (3) as a function of the aVOR gain for W6 (circle) and W12 (squares). The linear regressionof both groups (n=23 mice) and the goodness of fit is represented. D) Individual number of peripheral Spp1+ type I HC (1), CASPR1 + type I HC (2) and Calre+ type II HC (3) as a function of the OCR gain for W6 (circle) and W12 (squares). The linear regression of both groups (n=23 mice) and the goodness of fit is represented. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+

Figure 4 supplement: Effect of the IDPN on optokinetic reflex amplitude and timing.

+

A) OKR Gain across all frequencies and weeks tested B) OKR Phase across all frequencies and weeks tested. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +References +Agrawal Y, Carey JP, Della Santina CC, Schubert MC, Minor LB (2009) Disorders of Balance and Vestibular Function in US Adults: Data From the National Health and Nutrition Examination Survey, 2001-2004. Arch Intern Med 169:938. +Agrawal Y, Ward BK, Minor LB (2013) Vestibular dysfunction: Prevalence, impact and need for targeted treatment Chabbert C, ed. J Vestib Res 23:113–117. +Baird RA, Desmadryl G, Fernandez C, Goldberg JM (1988) The vestibular nerve of the chinchilla. II. Relation between afferent response properties and peripheral innervation patterns in the semicircular canals. J Neurophysiol 60:182–203. +Beraneck M, Bojados M, Le Séac’h A, Jamon M, Vidal P-P (2012) Ontogeny of Mouse Vestibulo-Ocular Reflex Following Genetic or Environmental Alteration of Gravity Sensing Gilestro GF, ed. PLoS ONE 7:e40414. +Beraneck M, Cullen KE (2007) Activity of Vestibular Nuclei Neurons During Vestibular and Optokinetic Stimulation in the Alert Mouse. J Neurophysiol 98:1549–1565. +Beraneck M, Idoux E (2012) Reconsidering the Role of Neuronal Intrinsic Properties and Neuromodulation in Vestibular Homeostasis. Front Neurol 3 Available at: http://journal.frontiersin.org/article/10.3389/fneur.2012.00025/abstract. +Beraneck M, Pfanzelt S, Vassias I, Rohregger M, Vibert N, Vidal P-P, Moore LE, Straka H (2007) Differential Intrinsic Response Dynamics Determine Synaptic Signal Processing in Frog Vestibular Neurons. J Neurosci 27:4283–4296. +Beraneck M, Straka H (2011) Vestibular signal processing by separate sets of neuronal filters Peusner KD, ed. J Vestib Res 21:5–19. +Bisdorff A, Von Brevern M, Lempert T, Newman-Toker DE (2009) Classification of vestibular symptoms: Towards an international classification of vestibular disorders. J Vestib Res 19:1–13. +Boadas-Vaello P, Sedó-Cabezón L, Verdú E, Llorens J (2017) Strain and Sex Differences in the Vestibular and Systemic Toxicity of 3,3′-Iminodipropionitrile in Mice. Toxicol Sci:kfw238. +Boyden ES, Katoh A, Raymond JL (2004) CEREBELLUM-DEPENDENT LEARNING: The Role of Multiple Plasticity Mechanisms. Annu Rev Neurosci 27:581–609. +Boyle R, Carey JP, Highstein SM (1991) Morphological correlates of response dynamics and efferent stimulation in horizontal semicircular canal afferents of the toadfish, Opsanus tau. J Neurophysiol 66:1504–1521. +Boyle R, Goldberg JM, Highstein SM (1992) Inputs from regularly and irregularly discharging vestibular nerve afferents to secondary neurons in squirrel monkey vestibular nuclei. III. Correlation with vestibulospinal and vestibuloocular output pathways. J Neurophysiol 68:471–484. +Brandt T, Dieterich M (2017) The dizzy patient: don’t forget disorders of the central vestibular system. Nat Rev Neurol 13:352–362. +Brandt T, Strupp M, Arbusow V, Dieringer N (1997) Plasticity of the vestibular system: central compensation and sensory substitution for vestibular deficits. Adv Neurol:297–309. +Brueckner JK, Ashby LP, Prichard JR, Porter JD (1999) Vestibulo-ocular pathways modulate extraocular muscle myosin expression patterns. Cell Tissue Res 295:477–484. +Carcaud J, França de Barros F, Idoux E, Eugène D, Reveret L, Moore LE, Vidal P-P, Beraneck M (2017) Long-Lasting Visuo-Vestibular Mismatch in Freely-Behaving Mice Reduces the Vestibulo-Ocular Reflex and Leads to Neural Changes in the Direct Vestibular Pathway. eneuro 4:ENEURO.0290-16.2017. +Carey JP, Fuchs AF, Rubel EW (1996) Hair cell regeneration and recovery of the vestibuloocular reflex in the avian vestibular system. J Neurophysiol 76:3301–3312. +Carriot J, Jamali M, Chacron MJ, Cullen KE (2017) The statistics of the vestibular input experienced during natural self-motion differ between rodents and primates: Natural vestibular input in rodents and monkeys. J Physiol 595:2751–2766. +Cassel R, Bordiga P, Carcaud J, Simon F, Beraneck M, Le Gall A, Benoit A, Bouet V, Philoxene B, Besnard S, Watabe I, Pericat D, Hautefort C, Assie A, Tonetto A, Dyhrfjeld-Johnsen J, Llorens J, Tighilet B, Chabbert C (2019) Morphological and functional correlates of vestibular synaptic deafferentation and repair in a mouse model of acute onset vertigo. Dis Model Mech:dmm.039115. +Contini D, Holstein GR, Art JJ (2022) Simultaneous Dual Recordings From Vestibular Hair Cells and Their Calyx Afferents Demonstrate Multiple Modes of Transmission at These Specialized Endings. Front Neurol 13:891536. +Cullen K (2019) Vestibular processing during natural self-motion: implications for perception and action. :41. +Cullen KE, Minor LB, Beraneck M, Sadeghi SG (2010) Neural substrates underlying vestibular compensation: Contribution of peripheral versus central processing. J Vestib Res 19:171–182. +De Zeeuw CI, Lisberger SG, Raymond JL (2021) Diversity and dynamism in the cerebellum. Nat Neurosci 24:160–167. +Dean P, Porrill J (2014) Decorrelation Learning in the Cerebellum. In: Progress in Brain Research, pp 157–192. Elsevier. Available at: https://linkinghub.elsevier.com/retrieve/pii/B9780444633569000078. +Eatock RA, Songer JE (2011) Vestibular Hair Cells and Afferents: Two Channels for Head Motion Signals. Annu Rev Neurosci 34:501–534. +Ernst MO, Banks MS (2002) Humans integrate visual and haptic information in a statistically optimal fashion. Nature 415:429–433. +Facchini J, Rastoldo G, Xerri C, Péricat D, El Ahmadi A, Tighilet B, Zennou-Azogui Y (2021) Unilateral vestibular neurectomy induces a remodeling of somatosensory cortical maps. Prog Neurobiol 205:102119. +Faulstich BM, Onori KA, du Lac S (2004) Comparison of plasticity and development of mouse optokinetic and vestibulo-ocular reflexes suggests differential gain control mechanisms. Vision Res 44:3419–3427. +Faulstich M, van Alphen AM, Luo C, du Lac S, De Zeeuw CI (2006) Oculomotor Plasticity During Vestibular Compensation Does Not Depend on Cerebellar LTD. J Neurophysiol 96:1187–1195. +Fetsch CR, DeAngelis GC, Angelaki DE (2010) Visual-vestibular cue integration for heading perception: applications of optimal cue integration theory: Mechanisms of visual-vestibular cue integration. Eur J Neurosci 31:1721–1729. +França de Barros F, Carcaud J, Beraneck M (2019) Long-term Sensory Conflict in Freely Behaving Mice. J Vis Exp:59135. +França de Barros F, Schenberg L, Tagliabue M, Beraneck M (2020) Long term visuo-vestibular mismatch in freely behaving mice differentially affects gaze stabilizing reflexes. Sci Rep 10:20018. +Glasauer S (2007) Current Models of the Ocular Motor System. In: Developments in Ophthalmology, 2007 (Straube A, Büttner U, eds), pp 158–174. Basel: KARGER. Available at: https://www.karger.com/Article/FullText/100485. +Goldberg JM (2000) Afferent diversity and the organization of central vestibular pathways. Exp Brain Res 130:277–297. +Goldberg JM, Lysakowski A, Fernández C (1990) Morphophysiological and ultrastructural studies in the mammalian cristae ampullares. Hear Res 49:89–102. +Goldberg JM, Lysakowski A, Fernández C (1992) Structure and Function of Vestibular Nerve Fibers in the Chinchilla and Squirrel Monkey. Ann N Y Acad Sci 656:92–107. +González-Garrido A, Pujol R, López-Ramírez O, Finkbeiner C, Eatock RA, Stone JS (2021) The Differentiation Status of Hair Cells That Regenerate Naturally in the Vestibular Inner Ear of the Adult Mouse. J Neurosci 41:7779–7796. +Greguske EA, Carreres-Pons M, Cutillas B, Boadas-Vaello P, Llorens J (2019) Calyx junction dismantlement and synaptic uncoupling precede hair cell extrusion in the vestibular sensory epithelium during sub-chronic 3,3′-iminodipropionitrile ototoxicity in the mouse. Arch Toxicol 93:417–434. +Greguske EA, Llorens J, Pyott SJ (2021) Assessment of cochlear toxicity in response to chronic 3,3′-iminodipropionitrile in mice reveals early and reversible functional loss that precedes overt histopathology. Arch Toxicol 95:1003–1021. +Grosch M, Lindner M, Bartenstein P, Brandt T, Dieterich M, Ziegler S, Zwergal A (2021) Dynamic whole-brain metabolic connectivity during vestibular compensation in the rat. NeuroImage 226:117588. +Hasson T, Gillespie PG, Garcia JA, MacDonald RB, Zhao Y, Yee AG, Mooseker MS, Corey DP (1997) Unconventional Myosins in Inner-Ear Sensory Epithelia. J Cell Biol 137:1287–1307. +Hess BJM, Dieringer N (1990) Spatial Organization of the Maculo-Ocular Reflex of the Rat: Responses During Off-Vertical Axis Rotation. Eur J Neurosci 2:909–919. +Highstein SM, Goldberg JM, Moschovakis AK, Fernandez C (1987) Inputs from regularly and irregularly discharging vestibular nerve afferents to secondary neurons in the vestibular nuclei of the squirrel monkey. II. Correlation with output pathways of secondary neurons. J Neurophysiol 58:719–738. +Holland PJ, Sibindi TM, Ginzburg M, Das S, Arkesteijn K, Frens MA, Donchin O (2020) A Neuroanatomically Grounded Optimal Control Model of the Compensatory Eye Movement System in Mice. Front Syst Neurosci 14:13. +Idoux E, Tagliabue M, Beraneck M (2018) No Gain No Pain: Relations Between Vestibulo-Ocular Reflexes and Motion Sickness in Mice. Front Neurol 9:918. +Jang DC, Shim HG, Kim SJ (2020) Intrinsic Plasticity of Cerebellar Purkinje Cells Contributes to Motor Memory Consolidation. J Neurosci 40:4145–4157. +Kim GS, Wang T, Sayyid ZN, Fuhriman J, Jones SM, Cheng AG (2022) Repair of surviving hair cells in the damaged mouse utricle. Proc Natl Acad Sci 119:e2116973119. +Kodama T, du Lac S (2016) Adaptive Acceleration of Visually Evoked Smooth Eye Movements in Mice. J Neurosci 36:6836–6849. +Lacour M, Helmchen C, Vidal P-P (2016) Vestibular compensation: the neuro-otologist’s best friend. J Neurol 263:54–64. +Lasker DM, Han GC, Park HJ, Minor LB (2008) Rotational Responses of Vestibular–Nerve Afferents Innervating the Semicircular Canals in the C57BL/6 Mouse. J Assoc Res Otolaryngol 9:334–348. +Leong ATL, Gu Y, Chan Y-S, Zheng H, Dong CM, Chan RW, Wang X, Liu Y, Tan LH, Wu EX (2019) Optogenetic fMRI interrogation of brain-wide central vestibular pathways. Proc Natl Acad Sci 116:10122–10129. +Llorens J, Demêmes D (1994) Hair cell degeneration resenting from 3,3’-imiodiprpiunitril toxicity in the rat vestibular epithelia. Hear Res:78–86. +Llorens J, Demêmes D, Sans A (1993) The behavioural syndrome caused by 3’3-Iminodipropionitrile and related nitriles in the rat is associated with degeneration of the vestibular sensory hair cells. Toxicol Appl Pharmacol:199–210. +Llorens J, Demêmes D, Sans A (1994) The toxicity of IDPN on the vestibular system of the rat: new insights on its effects on behavior and neurofilament transport. NeuroToxicology:643–647. +Lopez C, Schreyer H-M, Preuss N, Mast FW (2012) Vestibular stimulation modifies the body schema. Neuropsychologia 50:1830–1837. +Lopez-Escamez JA, Carey J, Chung W-H, Goebel JA, Magnusson M, Mandalà M, Newman-Toker DE, Strupp M, Suzuki M, Trabalzini F, Bisdorff A (2015) Diagnostic criteria for Menière’s disease. J Vestib Res 25:1–7. +Lysakowski A, Gaboyard-Niay S, Calin-Jageman I, Chatlani S, Price SD, Eatock RA (2011) Molecular Microdomains in a Sensory Terminal, the Vestibular Calyx Ending. J Neurosci 31:10101–10114. +Mackrous I, Carriot J, Cullen KE, Chacron MJ (2020) Neural variability determines coding strategies for natural self-motion in macaque monkeys. eLife 9:e57484. +Maroto AF, Barrallo-Gimeno A, Llorens J (2021) Relationship between vestibular hair cell loss and deficits in two anti-gravity reflexes in the rat. Hear Res 410:108336. +Martins-Lopes V, Bellmunt A, Greguske EA, Maroto AF, Boadas-Vaello P, Llorens J (2019a) Quantitative Assessment of Anti-Gravity Reflexes to EvaluateVestibular Dysfunction in Rats. J Assoc Res Otolaryngol 20:553–563. +Martins-Lopes V, Bellmunt A, Greguske EA, Maroto AF, Boadas-Vaello P, Llorens J (2019b) Quantitative Assessment of Anti-Gravity Reflexes to EvaluateVestibular Dysfunction in Rats. J Assoc Res Otolaryngol 20:553–563. +McInturff S, Burns JC, Kelley MW (2018) Characterization of spatial and temporal development of Type I and Type II hair cells in the mouse utricle using new cell-type-specific markers. Biol Open 7:bio038083. +Minor L, Goldberg J (1991) Vestibular-nerve inputs to the vestibulo-ocular reflex: a functional-ablation study in the squirrel monkey. J Neurosci 11:1636–1648. +Möhwald K, Hadzhikolev H, Bardins S, Becker-Bense S, Brandt T, Grill E, Jahn K, Dieterich M, Zwergal A (2020) Health-related quality of life and functional impairment in acute vestibular disorders. Eur J Neurol 27:2089–2098. +Nelson AB, Faulstich M, Moghadam S, Onori K, Meredith A, du Lac S (2017) BK Channels Are Required for Multisensory Plasticity in the Oculomotor System. Neuron 93:211– 220. +Oommen BS, Stahl JS (2008) Eye orientation during static tilts and its relationship to spontaneous head pitch in the laboratory mouse. Brain Res 1193:57–66. +Paken J, Govender CD, Pillay M, Sewram V (2016) Cisplatin-Associated Ototoxicity: A Review for the Health Professional. J Toxicol 2016:1–13. +Pfanzelt S, Rossert C, Rohregger M, Glasauer S, Moore LE, Straka H (2008) Differential Dynamic Processing of Afferent Signals in Frog Tonic and Phasic Second-Order Vestibular Neurons. J Neurosci 28:10349–10362. +Prayuenyong P, Taylor JA, Pearson SE, Gomez R, Patel PM, Hall DA, Kasbekar AV, Baguley DM (2018) Vestibulotoxicity Associated With Platinum-Based Chemotherapy in Survivors of Cancer: A Scoping Review. Front Oncol 8:363. +Sadeghi SG, Beraneck M (2020) Task-Specific Differentiation of Central Vestibular Neurons and Plasticity During Vestibular Compensation. In: The Senses: A Comprehensive Reference, pp 290–308. Elsevier. Available at: https://linkinghub.elsevier.com/retrieve/pii/B978012809324524145X. +Sadeghi SG, Minor LB, Cullen KE (2007) Response of Vestibular-Nerve Afferents to Active and Passive Rotations Under Normal Conditions and After Unilateral Labyrinthectomy. J Neurophysiol 97:1503–1514. +Sadeghi SG, Minor LB, Cullen KE (2012) Neural Correlates of Sensory Substitution in Vestibular Pathways following Complete Vestibular Loss. J Neurosci 32:14685– 14695. +Sayyid ZN, Wang T, Chen L, Jones SM, Cheng AG (2019) Atoh1 Directs Regeneration and Functional Recovery of the Mature Mouse Vestibular System. Cell Rep 28:312–324.e4. +Sedó-Cabezón L, Boadas-Vaello P, Soler-Martín C, Llorens J (2014) Vestibular damage in chronic ototoxicity: A mini-review. NeuroToxicology 67. +Sedó-Cabezón L, Jedynak P, Boadas-Vaello P, Llorens J (2015) Transient alteration of the vestibular calyceal junction and synapse in response to chronic ototoxic insult in rats. Dis Model Mech:dmm.021436. +Seoane A, Demêmes D, Llorens J (2001) Pathology of the rat vestibular sensory epithelia during subchronic 3,3′-iminodipropionitrile exposure: hair cells may not be the primary target of toxicity. Acta Neuropathol (Berl) 102:339–348. +Shin S-L, Zhao GQ, Raymond JL (2014) Signals and Learning Rules Guiding Oculomotor Plasticity. J Neurosci 34:10635–10644. +Shinder ME, Perachio AA, Kaufman GD (2005) VOR and Fos response during acute vestibular compensation in the Mongolian gerbil in darkness and in light. Brain Res 1038:183–197. +Simon F, Pericat D, Djian C, Fricker D, Denoyelle F, Beraneck M (2020) Surgical techniques and functional evaluation for vestibular lesions in the mouse: unilateral labyrinthectomy (UL) and unilateral vestibular neurectomy (UVN). J Neurol 267:51– 61. +Simon F, Tissir F, Michel V, Lahlou G, Deans M, Beraneck M (2021) Implication of Vestibular Hair Cell Loss of Planar Polarity for the Canal and Otolith-Dependent Vestibulo-Ocular Reflexes in Celsr1–/– Mice. Front Neurosci 15:750596. +Soler-Martin C, Diez-Padrisa N, Boadas-Vaello P, Llorens J (2006) Behavioral Disturbances and Hair Cell Loss in the Inner Ear Following Nitrile Exposure in Mice, Guinea Pigs, and Frogs. Toxicol Sci 96:123–132. +Sousa AD, Andrade LR, Salles FT, Pillai AM, Buttermore ED, Bhat MA, Kachar B (2009) The Septate Junction Protein Caspr Is Required for Structural Support and Retention of KCNQ4 at Calyceal Synapses of Vestibular Hair Cells. J Neurosci 29:3103–3108. +Straka H, Vibert N, Vidal PP, Moore LE, Dutia MB (2005) Intrinsic membrane properties of vertebrate vestibular neurons: Function, development and plasticity. Prog Neurobiol 76:349–392. +Straka H, Zwergal A, Cullen KE (2016) Vestibular animal models: contributions to understanding physiology and disease. J Neurol 263:10–23. +Welgampola MS, Young AS, Pogson JM, Bradshaw AP, Halmagyi GM (2019) Dizziness demystified. Pract Neurol 19:492–501. +Wibble T, Pansell T, Grillner S, Pérez-Fernández J (2022) Conserved subcortical processing in visuo-vestibular gaze control. Nat Commun 13:4699. +Wilkerson BA, Artoni F, Lea C, Ritchie K, Ray CA, Bermingham-McDonogh O (2018) Effects of 3,3′-Iminodipropionitrile on Hair Cell Numbers in Cristae of CBA/CaJ and C57BL/6J Mice. J Assoc Res Otolaryngol 19:483–491. +Yang X, Zhou S, Wu J, Liao Q, Wang C, Liu M, Qu L, Zhang Y, Cheng C, Chai R, Zhang K, Yu X, Huang P, Liu L, Xiong W, Chen S, Chen F (2019) Surgery-free video-oculography in mouse models: enabling quantitative and short-interval longitudinal assessment of vestibular function. Neurosci Lett 696:212–218. +Zakir M, Dickman JD (2006) Regeneration of Vestibular Otolith Afferents after Ototoxic Damage. J Neurosci 26:2881–2893. +Zeng S, Ni W, Jiang H, You D, Wang J, Lu X, Liu L, Yu H, Wu J, Chen F, Li H, Wang Y, Chen Y, Li W (2020) Toxic Effects of 3,3 ′ -Iminodipropionitrile on Vestibular System in Adult C57BL/6J Mice In Vivo. Neural Plast 2020:1–11. +Zennou-Azogui Y, Xerri C, Harlay F (1994) Visual sensory substitution in vestibular compensation: neuronal substrates in the alert cat. Exp Brain Res 98 Available at: http://link.springer.com/10.1007/BF00233983. +Zwergal A, Schlichtiger J, Xiong G, Beck R, Günther L, Schniepp R, Schöberl F, Jahn K, Brandt T, Strupp M, Bartenstein P, Dieterich M, Dutia MB, la Fougère C (2016) Sequential [18F]FDG µPET whole-brain imaging of central vestibular compensation: a model of deafferentation-induced brain plasticity. Brain Struct Funct 221:159–170. + + +
diff --git a/test/2023.07.30.551185/fix-ref/2023.07.30.551185.xml b/test/2023.07.30.551185/fix-ref/2023.07.30.551185.xml new file mode 100644 index 00000000..128f3891 --- /dev/null +++ b/test/2023.07.30.551185/fix-ref/2023.07.30.551185.xml @@ -0,0 +1,445 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.07.30.551185 +1.3 + + +Regular Article + + +New Results + + +Neuroscience + + + +Smith-Magenis syndrome protein RAI1 regulates body weight homeostasis through hypothalamic BDNF-producing neurons and neurotrophin downstream signalling + + + +JavedSehrish +1 +2 + + +ChangYa-Ting +1 +2 + + +ChoYoobin +1 +2 + + +LeeYu-Ju +1 +2 + + +ChangHao-Cheng +1 +2 + + +HaqueMinza +1 +2 + + +LinYu Cheng +1 +2 + + +http://orcid.org/0000-0002-8411-9433 +HuangWei-Hsiang +1 +2 +* + +Department of Neurology and Neurosurgery, Centre for Research in Neuroscience, McGill University, Montréal, QC H3G 1A3, Canada +Brain Repair and Integrative Neuroscience Program, The Research Institute of the McGill University Health Centre, Montréal, QC H3G 1A3, Canada + + +Corresponding author. Email: wei-hsiang.huang@mcgill.ca + +2023 +2023.07.30.551185 + + +30 +7 +2023 + + +23 +10 +2023 + + +23 +10 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +This pre-print is available under a Creative Commons License (Attribution 4.0 International), CC BY 4.0, as described at http://creativecommons.org/licenses/by/4.0/ + + + +Abstract

Retinoic acid-induced 1 (RAI1) haploinsufficiency causes Smith-Magenis syndrome (SMS), a genetic disorder with symptoms including hyperphagia, hyperlipidemia, severe obesity, and autism phenotypes. RAI1 is a transcriptional regulator with a pan-neural expression pattern and hundreds of downstream targets. The mechanisms linking neural Rai1 to body weight regulation remain unclear. Here we find that hypothalamic brain-derived neurotrophic factor (BDNF) and its downstream signalling are disrupted in SMS (Rai1+/-) mice. Selective Rai1 loss from all BDNF-producing cells or from BDNF-producing neurons in the paraventricular nucleus of the hypothalamus (PVH) induced obesity in mice. Electrophysiological recordings revealed that Rai1 ablation decreased the intrinsic excitability of PVHBDNF neurons. Chronic treatment of SMS mice with LM22A-4 engages neurotrophin downstream signalling and delayed obesity onset. This treatment also partially rescued disrupted lipid profiles, insulin intolerance, and stereotypical repetitive behaviour in SMS mice. These data argue that RAI1 regulates body weight and metabolic function through hypothalamic BDNF-producing neurons and that targeting neurotrophin downstream signalling might improve associated SMS phenotypes.

+ + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

provided source images for western blotting

 + + + + + +Introduction +

Obesity is an increasing global concern: 20% of the world’s adult population is expected to become obese by 2025 if current trends persist (Collaboration, 2017). However, the complex environmental, behavioural, and genetic contributions to obesity make this a complex condition to model and address. Monogenic obesity disorders provide a consistent system where the pathophysiology, genetic predisposition, and neurobiology underlying phenotypes can potentially provide insights that translate to polygenic obesity (Loos & Yeo, 2022).

+

One such monogenic disorder is Smith-Magenis syndrome (SMS; OMIM #182290) (Smith et al, 1986), a neurogenetic condition frequently associated with obesity. 80% of SMS cases are caused by interstitial deletions of the 17p11.2 chromosomal region containing a dosage-sensitive gene, retinoic acid-induced 1 (RAI1) (Slager et al, 2003). Most SMS patients are overweight (40% > 95th percentile weight, 80% > 85th percentile weight) and exhibit over-eating tendencies by adolescence (Alaimo et al, 2015; Burns et al, 2010; Gandhi et al, 2022). 20% of SMS patients do not have 17p11.2 deletions but instead carry RAI1 point mutations that are still frequently associated with obesity (Boot et al, 2021; Edelman et al, 2007). These clinical data indicate that RAI is critical for body weight regulation.

+

The human RAI1 protein shares >80% sequence identity with the mouse RAI1 protein, allowing us to study RAI1 function in vivo (Huang et al, 2016; Javed et al, 2020). We previously found that Rai1 encodes a transcriptional regulator that binds to the promoter or enhancer regions of its target genes, most of which are involved in neurodevelopment and synaptic transmission (Huang et al., 2016). SMS (Rai1+/-) mice exhibit obesity associated with hyperphagia and increased adiposity (Burns et al., 2010; Huang et al., 2016). Interestingly, RAI1 may have a sexually dimorphic function, as female SMS mice exhibit more pronounced obesity than males (Burns et al., 2010; Huang et al., 2016).

+

Using a homozygous Rai1 conditional knockout (Rai1-cKO) mice, we found that Rai1 deletion in either subcortical excitatory neurons (vGlut2+) or single-minded homolog 1 (Sim1+) neurons induced obesity in mice (Huang et al., 2016). Sim1+ neurons reside in several brain regions including the nucleus of the lateral olfactory tract, the supraoptic nucleus, the medial amygdala, and the paraventricular nucleus of the hypothalamus (PVH), a region that is critical for body weight regulation (Balthasar et al, 2005). Sim1+ neurons in the PVH are composed of several molecularly defined cell types, including those that express oxytocin, vasopressin, corticotropin-releasing hormone, somatostatin, melanocortin-4-receptor (MC4R), growth hormone-releasing hormone (GHRH), or brain-derived neurotrophic factor (BDNF) (An et al, 2015). Deleting Bdnf from the PVH results in obesity due to increased food intake, reduced energy expenditure, and reduced locomotion (An et al., 2015). While PVH neuronal activity is reduced during feeding (Xu et al, 2019), activation of PVHBDNF neurons results in weight loss (Wu & Xu, 2022). We recently found that RAI1 promotes Bdnf expression in the hypothalamus throughout life (Javed et al, 2021). We hypothesized that RAI11 may be connected to BDNF-producing neurons in PVH, explaining how a neural transcription factor regulates body weight homeostasis.

+

BDNF regulates energy metabolism by binding to its cognate receptor tropomyosin receptor kinase B (TRKB) (An et al., 2015) and may thus provide a potential therapeutic target for SMS. However, the link between RAI1, BDNF, and SMS symptoms is unclear. Patients carrying monogenic mutations in either BDNF (Gray et al, 2006) or NTRK2 (encoding TRKB) (Yeo et al, 2004) experience severe hyperphagic obesity. Moreover, obesity and associated metabolic dysfunctions such as impaired glucose and insulin sensitivity are correlated with impaired PI3K-AKT pathway, a TRKB downstream target (Su et al, 2021). Given that (1) Bdnf overexpression during early embryonic development or (2) targeting Bdnf overexpression to the PVH of adolescent SMS (Rai1+/-) mice protected them from weight gain and metabolic defects (Javed et al., 2021), and that (3) a point mutation in RAI1 disrupts its ability to transcriptionally enhance BDNF expression in vitro (Abad et al, 2018), pharmacological targeting of neurotrophin downstream signalling is an attractive and yet unexplored therapeutic avenue for SMS.

+

Building on our previous discoveries, here we examine the neurobiological functions of RAI1 in BDNF-producing neurons, specifically those that reside in the PVH. Importantly, we evaluated the involvement of the BDNF-TRKB pathways in SMS pathogenesis using a small molecule drug (LM22A-4) to enhance neurotrophin downstream signalling. Our work indicates that PVHBDNF neurons depend on RAI1 to maintain normal neuronal activity and regulate body weight homeostasis, and dysfunction of BDNF downstream signalling contributes to obesity associated with SMS mice.

+ + +Results + +<italic>Rai1</italic> deficiency dysregulates specific downstream targets of neurotrophin in the mouse hypothalamus +

Multiple lines of evidence indicated that SMS mice show decreased Bdnf mRNA expression in the hypothalamus (Burns et al., 2010; Huang et al., 2016; Javed et al., 2021). To determine if BDNF protein levels were consequently altered in the hypothalamus, we performed an enzyme-linked immunosorbent assay (ELISA) and found that BDNF protein levels were significantly reduced in 7-week-old pre-symptomatic SMS mice relative to controls (Fig 1A). To quantify if molecular pathways associated with BDNF downstream signalling were altered, we subjected hypothalamic tissue lysates from 7-week-old pre-obese SMS and control (Ctrl) mice to reverse phase protein array (RPPA) (Fig 1B). Control and SMS samples are clustered separately in the principal component analysis (Fig 1C).

+ + +<italic>Rai1</italic> haploinsufficiency disrupts hypothalamic proteomic profile in mice

(A) ELISA showing that BDNF protein levels in hypothalamic tissues were significantly reduced in SMS mice compared to controls (n=4/genotype). *p<0.05, unpaired t-test, two-tailed.

(B) A schematic diagram showing the experimental strategy. Hypothalamic tissue lysates from 7- week-old control (n = 5, technical triplicates/sample) and SMS (n=8, technical triplicates/sample) mice were subjected to RPPA that utilizes a cocktail of 240 validated antibodies to probe several key intracellular signalling pathways by quantifying the expression levels of total and phosphorylated proteins

(C) Principal component analysis segregates the SMS and Ctrl groups where PC1 explains 47% of the variance, and PC2 explains 25% of the variance.

(D) Heat map showing hierarchical clustering of differentially expressed phospho-proteins: most showed reduced levels in SMS samples.

(E) Protein-protein interaction (PPI) analysis showing differentially expressed proteins and phospho-proteins in a crosstalk network. PPI enrichment p-value: < 1.0e-16. The nodes represent the proteins/phospho-proteins and the lines indicate physical interactions. Associations with specific molecular pathways are shown as color-coded outer rings.

(F-G) The protein levels of two neurotrophin pathway-related nodes, PIK3CA (F) and phospho-NF-kB (G), were significantly reduced in SMS samples. ****p<0.0001, unpaired t-test, two-tailed

+ + +

Hierarchical clustering analysis indicated that most differentially expressed phosphorylated proteins (Fig 1D) and total proteins (Figure 1—figure supplement 1) were downregulated in SMS mice. We mapped total proteins, as well as phospho-proteins, that were differentially expressed in Ctrl and SMS samples onto the interconnected biological network and identified a subset of connected nodes downregulated in SMS (FDR < 10e-5) belong to the neurotrophin, PI3K-AKT, and mTOR signalling cascades (Fig 1E). These molecules are known to play a critical role in regulating body weight homeostasis and can be activated by BDNF either directly (neurotrophin pathway) or indirectly (PI3K-AKT and mTOR pathways) (Cota et al, 2006; Schultze et al, 2012). The expression levels of PIK3CA, a common node in these pathways, were significantly reduced in the SMS mice compared to controls (Fig 1F). Moreover, the expression levels of phosphorylated NF-κB (nuclear factor-kappa B), a downstream target of the neurotrophin pathway, were significantly reduced in the SMS group (Fig 1G). Together, these results suggest that Rai1 loss impairs specific pathways, including neurotrophin signalling, and indicate that dysfunction of BDNF-producing cells could contribute to SMS pathogenesis.

+ + +BDNF-producing cells depend on RAI1 to regulate body weight homeostasis +

To target BDNF-producing cells, we used a BdnfCre/+ mouse line that carries a viral 2A oligopeptide and Cre recombinase immediately upstream of Bdnf’s endogenous STOP codon (Tan et al, 2016). We crossed the BdnfCre/+ mice with tdTomato-Ai9 reporter mice and found that Ai9 signals faithfully reflected endogenous BDNF protein expression in the PVH (Figure 2— figure supplement 1A-E). We also found that RAI1 is expressed in PVHBDNF cells that are distinct from larger magnocellular cells, such as the PVH oxytocin neurons (Figure 2—figure supplement 2A-I). Specifically, 55% of PVHBDNF neurons express RAI1, whereas only 2% of PVHOxytocin neurons express RAI1 (Figure 2—figure supplement 2I). These data suggest that RAI1 is enriched in BDNF-producing neurons in the mouse PVH.

+

To determine if BDNF-producing cells depend on RAI1 to regulate body weight, we deleted one or both copies of Rai1 using the BdnfCre/+ and the conditional Rai1fl/fl mice (Huang et al., 2016) (Fig 2A, Figure 2—figure supplement 3A). Conditional knockout mice (cKO: BdnfCre/+; Rai1fl/fl) showed a significant reduction in Rai1 expression in PVHBDNF neurons compared to controls (Fig 2B-C, Figure 2—figure supplement 3B). The total number of BDNF-producing cells in the PVH remained similar in Ctrl and cKO mice (Figure 2—figure supplement 3C), indicating that RAI1 loss did not affect PVH neurogenesis or survival.

+ + +RAI1 is required in BDNF-producing cells to regulate energy metabolism +

(A) A schematic diagram showing selective deletion of Rai1 from BDNF-producing cells in mice.

(B) Representative images showing that in Ctrl mice (BdnfCre/+; Ai9), many PVHBDNF neurons (magenta) express RAI1 (cyan, double-positive cells were indicated with yellow arrows). By contrast, PVHBDNF neurons lack RAI1 expression in the cKO group (BdnfCre/+; Rai1fl/fl; Ai9). Scale bars = 20µm.

(C) Percentage of PVHBDNF neurons co-expressing RAI1 in Ctrl (n=3) and cKO (n=3) mice. unpaired t-test, two-tailed, ****p<0.0001.

(D) Female cKO mice (n = 12) showed a significant weight gain when compared to female Ctrl mice (n = 10). Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

(E) Body composition was measured with Echo-MRI, showing an increased fat mass in 26- week-old mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. ****p<0.0001.

(F) Fat mass of brown adipocytes (BAT), subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) in Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. **p<0.01, ****p<0.0001.

(G) Representative images showing eWAT adipocyte hypertrophy of the cKO mice (left). Scale bar = 500µm. Frequency distribution of adipocytes at each cellular size (right). Two-way ANOVA with Šidák’s multiple comparisons test. ***p<0.001, ****p<0.0001.

(H) Female cKO mice showed significantly increased blood leptin levels. unpaired t-test, two-tailed, **p<0.01.

(I) Female cKO mice showed reduced energy expenditure during the dark phase. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(J) Female cKO mice showed reduced locomotor activity during the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(K) Female cKO mice showed similar food intake as control mice. ns indicates not significant. unpaired t-test, two-tailed.

(L) Glucose tolerance test showing that female cKO mice became hyperglycemic 20 minutes after intraperitoneal glucose administration, suggesting glucose intolerance. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

Data are shown as mean±SEM.

+ + +

We monitored body weight weekly and found that male and female cKO mice showed a significant increase in body weight beginning at 15 weeks of age compared to controls (Ctrl: BdnfCre/+) (Fig 2D, Figure 2—figure supplement 4A). However, conditional Rai1 heterozygous mice (BdnfCre/+; Rai1fl/+) did not show a pathological increase in body weight (Figure 2—figure supplement 3D, Figure 2—figure supplement 4B), suggesting that non-BDNF-producing cells also contribute to SMS pathogenesis. EcoMRI found that 26-week-old cKO male and female mice showed increased fat mass but not lean mass (Fig 2E, Figure 2—figure supplement 4C). Further dissection of fat tissues from different organs showed an increased mass of subcutaneous inguinal (S.C. Ing.) fat in cKO mice of both sexes and an increased mass of epididymal adipose tissues (eWAT) in female cKOs (Fig 2F, Figure 2—figure supplement 4D).

+

In contrast, the total mass of brown adipose tissue (BAT) remained unchanged between groups (Fig 2F, Figure 2—figure supplement 4D). In both male and female cKO mice, we found hypertrophic eWAT adipocytes (Fig 2G, Figure 2—figure supplement 4E) and elevated blood leptin levels (Fig 2H, Figure 2—figure supplement 4F). However, there was no significant alteration in blood lipids such as triglycerides (TG), high-density lipoprotein (HDL), low-density lipoproteins (LDL), and very low-density lipoprotein (VLDL) in either male or female cKO mice (Figure 2—figure supplement 3E-G, Figure 2—figure supplement 4G-I). These data suggest that obesity caused by ablating Rai1 from BDNF-producing cells is associated with increased fat mass, fat deposition, and leptin levels.

+

Obesity could result from increased food intake, reduced energy expenditure, or both. Similar to SMS mice, male and female cKO mice did not have defects in respiratory exchange rates when measured by indirect calorimetry (Figure 2—figure supplement 3H, Figure 2—figure supplement 4J). Interestingly, cKO mice showed sex-specific differences in energy expenditure. During the dark phase, female cKO mice showed reduced energy expenditure (Fig 2I) and decreased total locomotor activity (Fig 2J). In contrast, total food intake remained unchanged (Fig 2K). Male cKO mice became hyperphagic (Figure 2—figure supplement 4K), showed increased energy expenditure during the dark phase (Figure 2—figure supplement 4L), and exhibited normal locomotor activity when compared to controls (Figure 2—figure supplement 4M). These results suggest that obesity in male cKO mice is attributable to overeating despite increased energy expenditure.

+

To determine if Rai1 loss from BDNF-producing cells impacts metabolic function, a glucose tolerance test (GTT) was performed. Interestingly, cKO mice of both sexes showed normal blood glucose levels right before the test but became hyperglycemic 20 minutes (female cKO) and 45 minutes (male cKO) after glucose administration (Fig 2L, Figure 2—figure supplement 4N). This change was accompanied by a significantly higher area under the curve (AUC) for both groups (Figure 2—figure supplement 3I, Figure 2—figure supplement 4O). While the male cKO mice remained hyperglycemic throughout the GTT (Figure 2—figure supplement 4N), the glucose levels in female cKO mice returned to normal towards the end (Fig 2L). We also found increased blood insulin levels during the GTT in both sexes (Figure 2—figure supplement 3J, Figure 2—figure supplement 4P). In the insulin tolerance test (ITT), female cKO mice showed moderately elevated glucose levels 20 minutes into the test (Figure 2—figure supplement 3K). By contrast, male cKO mice showed considerably higher glucose levels than controls in the ITT (∼6.5 ng/ml for cKO and ∼2.0ng/ml for Ctrl) during the 60- to 120-minute intervals (Figure 2— figure supplement 4Q). These experiments demonstrate that BDNF-producing cells depend on RAI1 to regulate body weight, adiposity, glucose and insulin tolerance in mice.

+ + +<italic>Rai1</italic> deletion reduces PVH<sup>BDNF</sup> neuronal excitability +

BDNF-producing neurons are widely distributed in the hypothalamus, being found in the dorsomedial hypothalamus (Unger et al, 2007), ventromedial hypothalamus (Xu et al, 2003), and PVH (An et al., 2015). To understand how Rai1 loss affects intrinsic neuronal excitability and synaptic transmission of PVHBDNF neurons that underlie energy homeostasis (An et al., 2015), we fluorescently labelled them with Cre-dependent tdTomato (Fig. 3A). We found that a hyperpolarized step followed by increasing the holding voltage at resting membrane potentials (Rm) was able to initiate rebound firing or rebound repetitive firing in most PVHBDNF neurons (Figure 3—figure supplement 1A). Hyperpolarization-induced IH currents and low depolarization events were observed in some PVHBDNF neurons (Figure 3—figure supplement 1A). cKO PVHBDNF neurons showed reduced action potential (AP) firing frequency at a holding voltage of -60mV compared to Ctrl PVHBDNF neurons (Fig 3B-C). The resting membrane potential (Rm) of cKO (-52mV) neurons showed a 10mV positive shift compared to Ctrl neurons (-62mV) (Fig 3D). We also found that fewer cKO PVHBDNF neurons showed spontaneous firing at Rm (Fig 3E, Ctrl:10/13; cKO:5/12) and at a holding voltage of -70mV (Fig 3F, Ctrl:6/12; cKO:1/12) compared to Ctrl PVHBDNF neurons. The threshold of the first AP initiated by a ramp current injection (1 nA/sec) from Rm was significantly increased in cKO neurons (Fig 3G-H). Because a wider shoulder and lower AP amplitude were observed when held at more depolarized potentials (Figure 3—figure supplement 1B), we compared the AP waveforms at a holding voltage of -60mV, close to the Rm of both genotypes. The AP of the cKO group exhibited a lower peak amplitude (measuring from the threshold to peak) (Fig 3I), a normal threshold (Fig 3J), and a normal half-width (Figure 3—figure supplement 1C). In summary, we found that both passive and AP properties of cKO PVHBDNF neurons were altered due to RAI1 loss. The Rm of cKO neurons shifted to a more positive voltage than the gating threshold of voltage-gated sodium channels (6/12 were more positive than -55mV), which was unfavourable for their transition from the inactivation state to the closed state. Combined with the higher AP threshold of cKO neurons, these changes contribute to a decreased excitability in cKO PVHBDNF neurons.

+ + +<italic>Rai1</italic> loss reduces intrinsic neuronal excitability and enhances inhibitory synaptic transmission of PVH<sup>BDNF</sup> neurons

(A) A representative image showing a patched PVHBDNF neuron labelled by BdnfCre-dependent tdTomato fluorescence signals.

(B-C) Representative traces of spontaneous firing of control (black) and cKO (red) neurons at a holding voltage of -60mV (B). The average spontaneous AP firing frequencies are shown in (C). *p<0.05, unpaired student’s t-tests.

(D) The resting membrane potentials of control (black) and cKO (red) PVHBDNF neurons. **p<0.01, unpaired student’s t-tests.

(E-F) The number of control (black) and cKO (red) PVHBDNF neurons that showed spontaneous firing (open) or are silent (solid) at resting membrane potentials (E) and a holding voltage of - 70mV (F).

(G) Representative traces of elicited APs in control (black) and cKO (red) PVHBDNF neurons responding to an inject current (bottom, blue), which ramps up from the resting membrane potentials at 1 nA per second for 500ms.

(H) The threshold of the first AP initiated by a current ramp in control (black) and cKO (red) PVHBDNF neurons. *p<0.05, unpaired student’s t-tests.

(I-J) The amplitude (I) and threshold (J) for the first AP firing at the holding voltage of -60mV in control (black) and cKO (red) PVHBDNF neurons. *p<0.05, unpaired student’s t-tests.

(K) The cumulative fractions of mIPSC frequency were measured in control (black) and cKO(red) PVHBDNF neurons. The average frequency of events from a 2-minute stable recording of each neuron is shown in the inset, unpaired student’s t-tests.

(L) The cumulative fractions of mIPSC amplitude of control (black) and cKO (red) PVHBDNF neurons. The average amplitude of events from a 2-min stable recording of each neuron is normalized to its capacitance and shown in the inset, unpaired student’s t-tests.

(M) The area under the curve of each mIPSC event from a 2-minute recording is averaged. Controls are shown in black, and cKOs are shown in red. unpaired t-tests, two-tailed. *p<0.05.

Data are shown as mean±SEM. Statistics: neuron numbers are in brackets (more than 3 mice/genotype)

+ + +

PVN neurons receive glutamatergic and GABAergic inputs and are tonically inhibited by GABAergic neurons (Colmers & Bains, 2018; Gao et al, 2017). We observed a slight rightward shift of the probability of miniature inhibitory postsynaptic current (mIPSC) frequency in cKO PVHBDNF neurons, although the average frequency (Fig 3K) was not significantly different between groups. The probability of mIPSC amplitude also showed a right shift without a significant change (Fig 3L, Figure 3—figure supplement 1D). However, we observed a significantly increased area under the curve (Fig 3M). Interestingly, we found that cKO PVHBDNF neurons had a smaller cellular diameter (Figure 3—figure supplement 1E). Overall, our data argue that Rai1 deletion decreases the excitability and AP firing of PVHBDNF neurons without significant changes in synaptic inhibitory transmission.

+ + +RAI1 is necessary for PVH<sup>BDNF</sup> neurons to regulate energy homeostasis +

Reduced PVHBDNF neuronal activity in cKO mice suggested that RAI1 loss in BDNF-producing PVH neurons might be sufficient to induce defective energy homeostasis in vivo. To test this hypothesis by deleting Rai1 only in PVHBDNF neurons but not other BDNF-producing cells, we performed PVHBDNF-specific Rai1 deletion using a CRISPR/Cas9 system packaged into AAV9. Two viruses were delivered: one encoding a reporter molecule (GFP) and three single guide RNAs targeting the Rai1 gene (sgRai1), and the other carrying Cre-dependent Staphylococcus aureus CRISPR-associated protein 9 (saCas9) (Fig 4A). Hereafter, our experiments focus on female SMS mice since they showed more consistent defects in fat distribution (Burns et al., 2010) and body weight regulation (Huang et al., 2016) than males. Bilateral administration of two AAVs into the PVH of female BdnfCre/+ mice (Exp group) at 3 weeks of age resulted in a significant loss of Rai1 protein expression in PVHBDNF neurons when compared to BdnfCre/+ mice injected with an AAV encoding the sgRai1 and GFP (Ctrl group) (Fig 4B-C, Figure 4—figure supplement 1A). Quantification of AAV-infected (GFP+) BDNF neurons showed that while the total number of BDNF+ (Figure 4—figure supplement 1B) and BDNF+ GFP+ (Figure 4— figure supplement 1C) cells were not different in the two groups, the percentage of AAV- infected BDNF+ cells that express RAI1 was significantly reduced in the Exp group (Fig 4C, Figure 4—figure supplement 1D-E). Moreover, the virus expression did not spread to the non-PVH neighboring nuclei (Figure 4—figure supplement 1F). Deleting Rai1 from the PVHBDNF neurons resulted in a significant increase in body weight beginning at 17 weeks of age (Fig 4D). Similar to the cKO (BdnfCre/+; Rai1fl/fl) mice, obesity in 26-week-old Exp mice was due to increased fat mass (Fig 4E) and more specifically, increased subcutaneous inguinal (S.C. ing) and epididymal white adipose tissues (eWAT) but not BAT (Fig 4F). We also found that eWAT adipocytes became hypertrophic (Fig 4G). Similar to the cKO mice, food intake did not differ between Exp and Ctrl groups (Fig 4H), likely due to the high level of satiety signals in the brain caused by increased blood leptin levels (Fig 4I). We also found that energy expenditure (Fig 4J), respiratory exchange rate (Figure 4—figure supplement 1G), and blood lipid parameters, including TG (Figure 4—figure supplement 1H), HDL (Figure 4—figure supplement 1I), LDL and VLDL levels (Figure 4—figure supplement 1J), were similar in Exp and Ctrl groups, consistent with the cKO data (Figure 2). Interestingly, Exp mice showed significantly less locomotor activity during the dark phase than the Ctrl mice (Fig 4K). A GTT indicated that Exp mice have a hyperglycemic profile (Fig 4L), accompanied by a slightly higher area under the curve for blood glucose levels (p=0.05) (Fig 4M). Insulin levels in Exp mice were also significantly elevated 15 minutes after glucose administration (Fig 4N).

+ + +Selective <italic>Rai1</italic> ablation in PVH<sup>BDNF</sup> neurons induces obesity +

(A) A schematic showing stereotaxic injection of adeno-associated viruses (AAVs) expressing Cas9 protein and single-guide RNAs targeting Rai1 (sgRai1) to delete Rai1 from the PVHBDNF neurons of BdnfCre/+ female mice. Scale bar = 50µm.

(B) Representative images showing that RAI1 immunoreactivity (cyan) was dramatically reduced in BdnfCre/+; Ai9 mice injected with both sgRai1 and DIO-saCas9 viruses (Exp group n=3, bottom row) into the PVH. In contrast, many PVHBDNF neurons in BdnfCre/+; Ai9 mice injected only with the sgRai1 virus showed RAI1 expression (Ctrl group n =3, top row).

(C) The percentage of PVHBDNF neurons co-expressing RAI1 and GFP (virus) was significantly reduced in the Exp group, indicating a successful Rai1 deletion. unpaired t-test, two-tailed, ****p<0.0001

(D) PVHBDNF neuron-specific Rai1 deletion induced body weight gain in Exp mice (Ctrl group: n=5, Exp group: n =5). Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, **p < 0.01, ****p < 0.0001.

(E) Body composition was measured with Echo-MRI, showing an increased fat mass deposition in 26-week-old Exp mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. ***p<0.001, ****p < 0.0001.

(F) Fat deposition analysis shows that the Exp group has significantly more subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) mass (grams). ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ****p < 0.0001.

(G) Representative images showing eWAT adipocyte hypertrophy of the Exp group (left). Scale bar = 500µm. Frequency distribution of adipocytes at each cellular size (right). ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

(H) Exp and control mice showed similar food intake. ns indicates not significantly different, unpaired t-test, two-tailed.

(I) Blood leptin levels were significantly increased in the Exp mice. unpaired t-test, two-tailed, *p<0.05.

(J) Exp and control mice showed similar energy expenditure. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(K) Exp mice showed reduced locomotor activity in the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(L) Exp mice became hyperglycemic during the glucose tolerance test. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(M) Measurement of the area under the curve (AUC) in Exp and Ctrl mice during the glucose tolerance test. unpaired t-test, two-tailed, p=0.05.

(N) Exp mice showed increased plasma insulin levels during the glucose tolerance test. Two-way ANOVA with Šidák’s multiple comparisons test, **p<0.01.

Data are shown as mean±SEM.

+ + +

While Exp mice showed a trend towards increased glucose levels in insulin tolerance tests, the change was not statistically different (Figure 4—figure supplement 1K). Altogether, these experiments indicate that conditional deletion of Rai1 from PVHBDNF neurons during adolescence impairs the regulation of body weight, adiposity, glucose tolerance, and locomotion. These changes recapitulate most phenotypes observed in cKO mice with Rai1 deleted in BDNF- producing cells from early development.

+ + +Pharmacologically enhancing neurotrophin downstream signalling partially alleviates obesity in SMS mice +

Our findings so far indicate that multiple pathways, including neurotrophin signalling, are dysregulated in SMS mice, and that Rai1 loss from BDNF-producing cells, including PVHBDNF neurons, impairs energy homeostasis in vivo. Next we sought to examine the contribution of BDNF dysfunction in SMS pathogenesis by pharmacologically activating downstream targets of the neurotrophin pathway.

+

We systematically treated mice with LM22A-4, a potential small molecule BDNF loop domain mimetic (Massa et al, 2010). To measure the bioavailability of LM22A-4 in the SMS brain, we simultaneously administered LM22A-4 intranasally (5 mg/kg body weight) and intraperitoneally (50 mg/kg body weight), using the highest doses used in previous studies (Canals et al, 2004; Chang et al, 2006; Gu et al, 2022; Kron et al, 2012; Ogier et al, 2007). We then harvested forebrain and hypothalamic tissues one or three hours after drug administration. The spectrometric analysis found a significantly higher LM22A-4 concentration in both brain regions one-hour post-treatment, while its concentration decreased three hours after treatment (Figure 5—figure supplement 1A). Consistent with the reduced neurotrophin signalling observed in our proteomic analysis, saline-treated SMS mice showed reduced phosphorylated-AKT (p-AKT; Fig 5A, Figure 5—figure supplement 1B, Figure 5—source data 1, Figure 5—figure supplement 1 source data 1) levels in the hypothalamus. LM22A-4 treatment significantly increased the levels of p-AKT in the SMS mice 1-hr post-treatment (Fig 5A, Figure 5—figure supplement 1B, Figure 5—source data 1, Figure 5—figure supplement 1 source data 1), indicating increased signalling downstream of the BDNF pathway.

+ + +LM22A-4 treatment partially alleviates obesity and stereotypic behaviour in adult SMS mice +

(A) Representative western blot showing p-AKT, total AKT, Histone 3 (H3, a loading control) levels in the hypothalamus of saline-treated Ctrl, saline-treated SMS and LM22A-4-treated SMS mice.

(B) Quantification showing deficits in p-AKT levels of the hypothalamus were reversed by LM22A-4 treatment in SMS mice. Unpaired t-test, two-tailed, *p<0.05.

(C-E) Body weight of saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. Saline or LM22A-4 treatment periods were highlighted in orange. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, **p < 0.01, ****p < 0.0001.

(F) Food intake (in grams) of saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05, **p < 0.01.

(G) Locomotor activity (number of beam breaks) for saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t- test, two-tailed, *p<0.05.

(H) Fat deposition analysis shows that the saline-injected SMS group has significantly more subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) mass (grams) than saline-injected Ctrl mice. S.C.ing and eWAT mass in LM22A-4 injected mice is comparable to the saline-injected mice. Two-way ANOVA with Šidák’s multiple comparisons test. **p < 0.01, ***p<0.001, ****p < 0.0001.

(I) High-density lipoprotein (HDL) levels were restored to normal in LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed, **p < 0.01, ***p<0.001.

(J) Insulin levels in saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice right before the start of the insulin tolerance test (time point 0). ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05.

(K) Insulin tolerance test performed on saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice, two weeks post daily administration of LM22A-4 or saline, shows reductions in the blood glucose levels of LM22A-4 treated SMS mice. * Shows significance between saline-injected Ctrl and saline-injected SMS group. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. **p < 0.01.

(L) Stereotypic rearing behaviour was fully rescued in LM22A-4 treated SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed. *p<0.05, **p < 0.01.

Data are shown as mean±SEM.

+ + +

We next examined if chronic LM22A-4 treatment was sufficient to reverse or modify the progression of the obesity phenotype in adult SMS mice. Specifically, we treated SMS mice daily from 8 to 14 weeks of age, during which SMS mice began to gain significantly more weight. Saline-treated SMS mice (Fig 5C) were significantly heavier than saline-treated control mice from 9 weeks of age onward. Interestingly, LM22A-4-treated SMS mice showed a similar body weight to saline-treated control littermates until 13 weeks of age but became obese by 14 weeks of age (Fig 5D). Consistent with a partial rescue, LM22A-4-treated SMS mice were significantly less obese than saline-treated SMS mice throughout the treatment period (Fig 5E). Control mice treated with LM22A-4 did not show a further reduction in body weight (Figure 5—figure supplement 1C), indicating that the therapeutic effects were specific to SMS pathogenesis.

+

We measured food intake and found that LM22A-4 did not significantly rescue increased food intake in SMS mice (Fig 5F). SMS mice showed normal energy expenditure (Figure 5—figure supplement 1D) and respiratory exchange rate (Figure 5—figure supplement 1E), neither of which was altered by LM22A-4 treatment. Interestingly, we did observe a partial rescue of locomotor activity in LM22A-4-treated SMS mice that was not significantly different from saline-treated Ctrl or SMS mice (Fig 5G). We also found that the LM22A-4 treatment partially rescued fat deposition (Fig 5H) and completely rescued HDL levels (Fig 5I) in SMS mice. During an insulin tolerance test, we found that saline-treated SMS mice showed increased baseline insulin levels (Fig 5J) and blood glucose levels during the test (Fig 5K), both of which were partially rescued by LM22A-4 treatment.

+

In addition to a deficit in energy homeostasis, SMS mice showed two highly reproducible behavioural deficits: increased stereotypic rearing in the open field and decreased social dominance in the tube test (Huang et al, 2018; Rao et al, 2017). Saline-treated SMS mice showed significantly more rearing behaviour when compared to the saline-treated controls (Fig 5L). Interestingly, the LM22A-4 treatment reduced the repetitive rearing behaviour in SMS mice (Fig 5L). LM22A-4 treatment did not change repetitive behaviours in control mice (Figure 5— figure supplement 1F), suggesting the effect is specific to Rai1 haploinsufficiency. However, we did not observe a rescue of social dominance deficits in SMS mice (Figure 5—figure supplement 2A-E). Together, these data indicate that LM22A-4 treatment delayed the onset of obesity and fully rescued repetitive behaviour in SMS mice, suggesting that neurotrophin downstream signalling dysfunction contributes to changes in the mouse SMS brain linked to obesity and behavioural phenotypes.

+ + + +Discussion +

The hypothalamus and neurotrophin signalling in this region control body weight by regulating the intricate balance between energy intake and expenditure (Xu & Xie, 2016). The role of BDNF-producing neurons in human obesity disorders and how the activity of these cells is regulated remained poorly studied. We found that Rai1 deficiency, linked to Smith-Magenis syndrome (SMS), regulates BDNF expression and maintains intrinsic excitability in PVHBDNF neurons.

+

Consistent with this mechanism, we found that removing RAI1 from BDNF-producing neurons or PVHBDNF neurons results in imbalanced energy homeostasis and obesity. A small molecule drug LM22A-4 led to increased AKT phosphorylation and partially rescued locomotor activity, insulin intolerance, and HDL levels, while delaying the onset of obesity in SMS mice. These findings highlight a potential role for hypothalamic BDNF-producing neurons, specifically PVHBDNF neurons, in SMS pathogenesis and highlight BDNF signalling as a potentially druggable pathway to improve energy homeostasis defects and repetitive behaviours commonly observed in SMS patients (Burns et al., 2010; Moss et al, 2009).

+

The involvement of different PVH subtypes in human obesity is only beginning to be elucidated. Dysfunction of oxytocin PVH neurons is linked to the Prader-Willi syndrome (Lee et al, 2012), another genetic disorder associated with obesity. Since Rai1 expression is enriched in PVHBDNF neurons but not PVHOxytocin neurons, our data suggest that distinct pathophysiological mechanisms underlie SMS and Prader-Willi syndrome.

+

The conditional knockout analysis allowed us to precisely map the cell-specific origin of neuropathological phenotypes, such as SMS-like symptoms in mice. Our previous work found that deleting Rai1 from cortical excitatory neurons using an Emx1Cre allele (Gorski et al, 2002), from the GABAergic neurons using a Gad2Creallele (Taniguchi et al, 2011), or from astrocytes using a GfapCre allele (Garcia et al, 2004) did not induce SMS-like obesity (Huang et al., 2016). In contrast, obesity could be induced by ablating Rai1 from Sim1-lineage neurons and, to a lesser extent, SF1-lineage neurons (Huang et al., 2016). This analysis suggests that Rai1 expression in PVH is more important than in the ventromedial nucleus of the hypothalamus (VMH) for body weight regulation. Interestingly, Although BDNF is required in the VMH and DMH to regulate body weight (Unger et al., 2007), embryonic deletion of Bdnf from the SF1-lineage populations including the VMH did not result in obesity (Kamitakahara et al, 2016; Yang et al, 2016). By contrast, deleting Bdnf from Sim1-lineage neurons including the PVH resulted in a more severe obesity phenotype (An et al., 2015). This is consistent with the idea that BDNF expression in PVH is critical for energy homeostasis. We found that Rai1 expression in BDNF-producing cells regulates energy homeostasis in a sexually dimorphic manner. Specifically, female cKO mice showed normal food intake, decreased energy expenditure, and reduced locomotor activity. In contrast, male cKO mice showed increased food intake with no changes in locomotor activity. Intriguingly, energy expenditure was increased in male cKO mice, likely due to the higher demand to maintain normal activity levels in obese mice. In both sexes, cKO mice showed an increased fat mass and adiposity with no change in lean mass, highlighting an essential role for Rai1 in regulating fat deposition. Our data also indicate that BDNF-producing cells are only partially responsible for obesity in SMS: male and female cKO mice become overweight by 15 weeks, in contrast to SMS mice which become obese by 9 weeks of age, suggesting that non-BDNF neurons contribute to SMS phenotypes.

+

Notably, mice lacking one copy of Rai1 in the BDNF-producing cells do not exhibit obesity, whereas SMS patients and SMS mice show pronounced obesity (Burns et al., 2010; Huang et al., 2016; Smith et al, 2005). This indicates that although reduced Bdnf expression and BDNF-producing neurons contribute to regulating body weight, additional molecular changes and other hypothalamic populations also play important roles in regulating body weight homeostasis in SMS. Our RPPA data suggest that mTOR signalling is also misregulated in addition to the reduced activation of the neurotrophin downstream cascades. Hypothalamic mTORC1 is crucial to regulate glucose release from the liver, peripheral lipid metabolism, and insulin sensitivity (Burke et al, 2017; Caron et al, 2016; Smith et al, 2015), while mTORC2 regulates glucose tolerance and fat mass (Kocalis et al, 2014). How the impaired mTOR signalling contributes to energy homeostasis defects in SMS and the therapeutic potential of targeting this pathway to treat SMS-related obesity remains unclear and warrants future investigation.

+

What additional RAI1-dependent hypothalamic cell types residing in brain regions other than PVH regulate obesity in SMS? Other important cell types such as TRKB neurons within the PVH (An et al, 2020) and several RAI1-expressing hypothalamic nuclei including the arcuate nucleus, ventromedial nucleus of the hypothalamus (VMH), and lateral hypothalamus all play important roles in regulating energy homeostasis. POMC- and AGRP-expressing neurons within the arcuate nucleus are known to regulate food intake and glucose and insulin homeostasis (Quarta et al, 2021; Vohra et al, 2022). Therefore, Rai1 function in these neurons could contribute to obesity in SMS, a topic that awaits future investigation.

+

We found the metabolic phenotypes induced by deleting Rai1 from BDNF-producing cells were specifically due to Rai1 deletion in PVHBDNF neurons. Therefore, our study focused on PVHBDNF neurons. Electrophysiological experiments showed that most wild-type PVHBDNF neurons display high-frequency spontaneous AP firing (1-10 Hz) at the resting membrane potential. Most PVHBDNF neurons fire sharp APs held at a hyperpolarized potential and show repetitive overshoot during the current ramp test. Loss of Rai1 significantly reduced the intrinsic neuronal excitability of PVHBDNF neurons. Because GABAergic Rai1 loss did not induce obesity (Huang et al., 2016) and most BDNF-producing neurons are glutamatergic (Canals et al, 2001). The ionic changes in PVHBDNF neurons that precisely explain these excitability changes await future investigations. Different groups of PVH neurons show phasic bursting or slowly inactivating delayed rectifier potassium conductance, which are properties mediated by different potassium channels (Lee et al., 2012). Given that our previous RNA-seq experiment found that hypothalamic Rai1 loss induces misexpression of potassium channel subunits Kcnc2 and Kcnj11 (Huang et al., 2016), RAI1-dependent potassium channels are likely involved in altered neuronal activity. Furthermore, we found fewer low threshold depolarization events in cKO PVHBDNF neurons during the hyperpolarized voltage step, suggesting a potential involvement of HCN channels.

+

Our previous work found that Rai1 loss in the dentate gyrus resulted in neuronal hyperexcitability (Chang et al, 2022b). Here we demonstrated that Rai1 loss in the hypothalamus resulted in neuronal hypoactivity, suggesting that RAI1 regulates neuronal excitability in a cell type-specific manner. Reduced PVHBDNF neuronal activity upon Rai1 loss could potentially explain obesity, consistent with a previous report that found chemogenetic activation of PVHBDNF neurons promotes negative energy balance by decreasing feeding and increasing energy expenditure (Wu & Xu, 2022).

+

It is plausible that RAI1 regulates the expression of genes encoding inward rectifier K+ channels, which regulate neuronal activity and potentially energy homeostasis. For instance, KIR6 (a family of ATP-sensitive potassium channels, KATP) is widely expressed in the hypothalamus. Deleting the hypothalamic KIR6.2 subunit impairs KATP channel function and glucose tolerance (Miki et al, 2001; Parton et al, 2007). Moreover, reduced expression of hypothalamic GIRK4 (encoding an inwardly rectifying potassium channel) causes obesity (Perry et al, 2008). GABAergic neurotransmission from arcuate AGRP-expressing neurons to the PVH neurons is important to increase appetite by favouring hyperphagia (Atasoy et al, 2012). Disrupting the composition of these ion channels could contribute to reduced PVHBDNF neuronal firing, which awaits further investigations.

+

Female mice with a conditional knockout of Rai1 from BDNF-producing neurons do not display a noteworthy difference in food intake. Conversely, their male counterparts exhibit a significant increase in food intake. Although SMS individuals of both genders tend to overeat, male patients who are obese show significantly higher food consumption than their female counterparts (Gandhi et al., 2022). This observation raises the possibility that Rai1 regulates eating behaviours through multiple cell types in the hypothalamus and that a male-specific involvement of BDNF-producing neurons in regulating food intake, potentially provides a neurobiological basis for the observed pattern in SMS patients (Gandhi et al., 2022). In future work, it would be interesting to dissect the sex-specific role of RAI1 in regulating the functional development of PVHBDNF neurons. In addition to a role for RAI1 in regulating neuronal activity, RAI1 may regulate other aspects of PVHBDNF neuronal function, including projections to other brain regions to regulate energy homeostasis. In this regard, it would be interesting to dissect how RAI1 controls the circuit connectivity of PVHBDNF neurons and how Rai1 loss in PVHBDNF neurons affects nearby PVHTRKB neurons that receive BDNF signalling and are linked to body weight regulation (An et al., 2020).

+

At the molecular level, our RPPA data show that Rai1 depletion disrupted multiple intracellular signalling pathways and resulted in the hypoactivation of specific phospho-signalling proteins. Multiple early studies support the notion that Rai1 directly promotes Bdnf expression. First, hypothalamic Bdnf mRNA levels were reduced in SMS mice (Burns et al., 2010; Javed et al., 2021) or mice with Rai1 conditional deletions (Huang et al., 2016). Second, the RAI1 protein directly binds to a promoter region of Bdnf (Huang et al., 2016). Finally, when Rai1 expression was induced using CRISPR activation, Bdnf mRNA levels were significantly elevated (Chang et al, 2022a).

+

To lend further support, here we show that SMS mice have reduced hypothalamic BDNF protein levels, as well as reduced activation of a TRKB downstream target AKT. Reduced AKT activity is frequently associated with insulin resistance and obesity in mice (Shao et al, 2000) and children (Su et al., 2021). We also found that obesity, increased HDL levels, and insulin insensitivity in SMS mice could be partially alleviated using a small molecule drug (LM22A-4) that promotes AKT phosphorylation. LM22A-4 penetrates the blood-brain barrier reasonably well and has been shown to ameliorate symptoms of other neurological disorders associated with BDNF dysfunction including Rett syndrome, Huntington’s disease, Dravet syndrome, and refractory neonatal seizures (Canals et al., 2004; Chang et al., 2006; Gu et al., 2022; Kron et al., 2012; Ogier et al., 2007). We recognize that while several in vivo studies have demonstrated the potential of LM22A-4 in targeting neurotrophin downstream signalling (Kron et al, 2014; Li et al, 2017), an in vitro analysis failed to demonstrate the ability of LM22A-4 to activate TRKB directly (Boltaev et al, 2017). Therefore, the precise mechanism by which LM22A-4 enhances AKT cascades in the mammalian brain remains unclear and awaits further investigations. In the hypothalamus of SMS mice, LM22A-4 could indirectly engage neurotrophin downstream PI3K- AKT pathway through the G protein-coupled receptor-dependent transactivation of the TRKB receptor (Domeniconi & Chao, 2010) or other unknown mechanisms. Moreover, while LM22A-4 may have potential side effects, we found that wild-type mice treated with LM22A-4 did not show a further decrease in body weight, suggesting limited side effects regarding body weight regulation.

+

The partial rescue of SMS phenotypes that we observed contrasts with our previous findings that found (1) AAV-mediated BDNF overexpression in PVH at 3 weeks of age and (2) genetically overexpressing BDNF from early development could fully prevent the development of obesity in SMS mice (Javed et al., 2021). Therefore, further increasing TRKB signalling activation and earlier intervention might improve therapeutic outcomes. Alternatively, our RPPA data showed that in addition to disruption of neurotrophin signalling, multiple pathways including mTOR were downregulated in SMS. Increasing hypothalamic mTOR activity has been shown to decrease food intake and body weight (Cota et al., 2006). Therefore, it is possible that enhancing both mTOR and TRKB signalling could achieve a more robust rescue of obesity in SMS mice. Nevertheless, these data support the concept that drugging TRKB signalling may improve stereotypical repetitive behaviour and partially alleviate metabolic functions in a mouse model of SMS. Delayed symptom onset could provide a window of opportunity for other interventions to further modify disease progression, an area requiring further investigation.

+

This study has several limitations. First, whether the PVHBDNF neurons depend on Rai1 to regulate adaptive thermogenesis awaits future analysis. It is known that a subset of posterior PVHBDNF neurons project to the spinal cord and regulate thermogenesis (An et al., 2015). However, we found the weights of brown adipocytes were similar in control and cKO mice. Moreover, SMS mice have normal respiratory exchange rates and there are currently no reports of defective adaptive thermogenesis in SMS patients. This suggests that adaptive thermogenesis is not impacted in SMS. Another limitation is that we did not test different doses, durations, and time points for LM22A-4 treatment. The precise mechanisms by which LM22A-4 improves body weight, metabolic function, and repetitive behaviours remain unknown. Nevertheless, our work identifies RAI1 as a novel regulator of the neuronal excitability of PVHBDNF neurons. This regulation is critical for regulating food intake and energy expenditure. Our evidence suggests that BDNF-producing neurons are involved in SMS pathogenesis and that enhancing BDNF- TRKB signalling should be explored as a future therapeutic avenue for obesity and metabolic conditions associated with reduced neurotrophin signalling (Xu & Xie, 2016).

+ + +Materials and Methods + +Animal Management +

The animal protocol (MUHC-8127) was approved by the animal care committee of the Montreal General Hospital and was performed according to the guidelines of the Canadian Council on Animal Care. Mice were group-housed in 12-hour light/12-hour dark cycles with an ad libitum supply of regular chow and water. The exception was in the fasting studies when the animals were food deprived for 5 hours and in CALMS and food intake studies when animals were single-housed in a 12-hour light/12-hour dark cycle with an ad libitum supply of regular chow and water. All mice were maintained in a C57Bl/6J background. BdnfCre/Cre (RRID: IMSR_JAX:030189), Rosa26Ai9/Ai9(RRID: IMSR_JAX:007909), and ActbCre/Cre (RRID: IMSR_JAX:019099) mice were purchased from the Jackson laboratory. The Rai1fl/fl (RRID: IMSR_JAX:029103) strain was previously generated and characterized by us (Huang et al., 2016). SMS mice (Rai1+/-: whole-body germline Rai1 heterozygous knockout that genetically mimics SMS) were generated by crossing germline ActbCre/Cre mice with Rai1fl/+. To generate conditional knockout mice, BdnfCre/+; Rai1fl/+ mice were crossed with Rai1fl/+ to create control and experimental groups from the same littermate. 7 mice with malocclusion or lesions were excluded from the study: 4 with malocclusion (2 females and 2 males) and 3 with lesions (3 females).

+ + +Metabolic profiling +

Metabolic profiling was conducted in three experimental conditions: male and female conditional knockout mice, female virus-injected BdnfCre/+ mice, and female drug- or saline-injected SMS and control mice. Metabolic profiling included body weight and body composition analysis, food intake measurement, indirect calorimetry for locomotion activity, energy expenditure (EE), respiratory exchange rate (RER), measurement of serum lipids and leptin levels, adipose tissue histology analysis, and glucose and insulin tolerance tests. All experiments were performed as previously described (Javed et al., 2021). A brief description is detailed below.

+ +Body weight and body composition analysis +

The body weight of animals was measured weekly from week 3 (weaning age) onward. After the 25th week, mice were subjected to body composition analysis. Body composition, including total fat and lean mass, was assessed using a nuclear EchoMRI whole-body analyzer. Toward the end of the experiments, animals were euthanized to collect brown adipose, visceral, subcutaneous (inguinal) and ependymal white adipose tissues and weighed using an analytical scale (Sartorius).

+ + +Food Intake Measurement +

Mice were single-housed for food intake measurement. Food intake was measured over a one-week period during which food was weighed prior to placement in the cage with minimal bedding. Each morning, the remaining food inside the cage was weighed. The amount of total food consumed was averaged over 7 days.

+ + +Indirect Calorimetry +

A Comprehensive Lab Animal Monitoring System (CLAM) (Columbus instruments) was used to measure the RER and EE (normalized over lean mass). Animals were singled-housed in the CLAMS apparatus at 21◦C (70◦F) in a light–dark cycle matching their housing conditions for 24 hours (acclimation), followed by 48 hours of measurement.

+ + +Serum lipid level determination +

A fluorescence-based assay kit (Abcam ab65390) was used to measure serum lipid levels. The manufacturer’s instructions were followed except that we used half-area black 96-well microplates. Fluorescence measurements at 535/587 nm (Ex/Em) were recorded with the Ensight instrument from PerkinElmer. A new standard curve ranging from 0 to 10 μg/ml was produced for each microplate to accurately determine lipid levels in serum samples. Standards and samples were loaded and analyzed twice.

+ + +Histology of adipose tissue +

Ependymal white adipose tissues (WAT) were excised immediately after the animals were euthanized and fixed with 10% formalin in PBS. The fixed tissues were dehydrated, embedded in paraffin, and sectioned. Two 5-micrometre thick sections were taken from each animal and stained with hematoxylin and eosin. Images of sections were then captured using an Aperio ScanScope CS slide scanner at 20X optical magnification, and the resulting images were saved in jpeg format using the ImageScope software. The images were manually cleaned to remove non-adipose tissues such as lymph nodes. An automated method using the Adiposoft plugin version 1.16 was utilized to determine adipocyte size and number from the jpeg images in Fiji version 2.1.0/1.53c. The Adiposoft settings used excluded cells on edges and set a minimum and maximum cell diameter of 20 and 300 μm, respectively. An experimental value of 2.49 μm per pixel was determined using the Aperio scale bar and used for Adiposoft. To ensure accuracy, two images from each animal were analyzed blindly, examining more than 1600 cells per mouse and more than 9100 cells per condition. Adipocyte size frequencies were calculated using the frequency function in Excel.

+ + +Glucose and insulin tolerance tests +

For the glucose tolerance test, experimental mice were food deprived for 5 hours with ad libitum water supply. A bolus of glucose (1.5g/kg of lean body weight) was then administered intraperitoneally and glycemia was measured from tail vein blood samples using the Accu-chek Performa glucometer at T0 (before the injection) as well as at 15, 30, 45, 60, 90 and 120 min (after the injection). Tail vein blood samples were collected via a capillary for insulin assays at T0, 15 and 30 min. For insulin tolerance tests, experimental mice were food-deprived for 5 hours with ad libitum access to water. A bolus of insulin (1 U/kg of lean body weight) was administered via an IP injection and glycemia was measured from blood sampled at the tail vein using an Accu-chek Performa glucometer at T0 (before the injection), as well as at 15, 30, 45, 60, 90 and 120 min (after the injection). Tail vein blood samples were collected via a capillary for insulin assays at T0.

+ + + +Stereotaxic Injections +

BdnfCre/+ mice were anesthetized using isoflurane, and the ears were barred onto a stereotaxic device (David Kopf Instruments). Adeno-associated viruses (AAVs, volume = 250 nl) were bilaterally injected into the PVH using the following coordinates relative to the Bregma: -0.6 mm anteroposterior (AP), ±0.35 mm mediolateral (ML) and -5.52 mm dorsoventral (DV). A description of the AAV constructs and titre used is detailed below:

+

AAV9-CMV7-DIO-saCas9 (Viral title: >1 × 10^13), Vector Biolab

+

AAV9-CMV7-sgRai1-GFP (Viral title: >1 ×10^9), Applied Biological Materials Inc. Three sgRNAs were pooled to target Rai1 (sgRNA1: TTCCTCGCCAGAGTAGCGCC, sgRNA2: CCCAGCCTCATGATAGGCCG, sgRNA3: CCCAGCCTCATGATAGGCCG).

+ + +Immunostaining of mouse brain tissues +

Immunostaining experiments were performed as previously described (Huang et al, 2012). Briefly, mice were anesthetized and perfused with 1X PBS and 4% formaldehyde, and the brains were dissected and fixed in 4% formaldehyde overnight. Fixed brains were transferred to 30% sucrose solution and subsequently embedded in the OCT solution. The brains were then sectioned (40µm), washed in PBS, permeabilized with 0.5% Triton X-100, and incubated with 10% normal donkey serum in PBS for 2 hours. The sections were then incubated overnight with primary antibodies (Rai1: 1:500 dilution, RRID:AB_2921229 (Huang et al., 2016); BDNF: 1:500, RRID:AB_10862052, Abcam ab108319; p-TRKB: 1:500 dilution, RRID:AB_2721199, Millipore ABN1381; oxytocin: 1:1000 dilution, RRID:AB_11212999, Millipore MAB5296) at 4°C on a shaker. The following day, sections were washed with 1X PBS twice for 10 minutes and incubated with a secondary antibody (1:1000 dilution) for 3 hours at room temperature. Sections were washed with 1X PBS once before mounting on glass slides with the DAPI- mounting (DAPI-Fluoromount-G, SouthernBiotech) and then imaged using a confocal microscope (Olympus FV-1000 confocal laser scanning microscope) with 20x or 40x objectives (NA 0.85 and 1.3, respectively). Z-stacks were taken with a step size of 1.0 µm with a 1024 x 1024 resolution. Images were acquired as a stack with at least 13 optical sections. Image analysis was performed using the Fiji software.

+ + +Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) +

qRT-PCR was performed using mRNAs isolated from hypothalamic tissue samples. After the isolation of total RNA, mRNAs were reverse-transcribed with the SuperScript III First-Strand Synthesis System (Thermo-Fisher). Quantitative PCR reactions were conducted using SsoFast EvaGreen Supermix on a Bio-Rad qPCR system.

+ + +Enzyme-linked immunosorbent assay (ELISA) +

ELISA was performed using hypothalamic tissues from 7-week-old SMS mice to assess the mature BDNF protein levels using a commercially available BDNF ELISA kit (Mature BDNF Rapid ELISA Kit: Human, Mouse, Rat).

+ + +Western blotting analysis +

Mice were quickly decapitated, brains were rapidly removed, and hypothalamic tissues were collected, weighed, and snap-frozen with liquid nitrogen. Hypothalamic tissues were lysed using a tissue lysis buffer containing RIPA buffer (5M NaCl, 1M PH 8.0 Tris HCl, 1% NP.40, 10% C24H39NaO4, and 20% SDS) and the Halt™ Protease and Phosphatase Inhibitor Cocktail (Thermo-Fisher). Lysates were first sonicated on ice for 15 seconds and centrifuged at 10000 RPM at 4°C for 10 minutes. The supernatant was collected in a separate Eppendorf tube, and a BCA assay was performed to quantify protein levels. Proteins were then separated using SDS- PAGE gel and transferred onto nitrocellulose membranes. To determine the phosphorylation levels of AKT, membranes were cut and blocked in 5% milk in tris-buffered saline (TBS) for 1 hour and then incubated overnight at 4°C with anti-p-AKT (Ser473) antibody (1:1000 dilution, RRID: AB_2315049, Cell Signaling Technology 4060T) and anti-H3 antibody (internal control: 1:1000 dilution, RRID: AB_302613, Abcam ab1791). The next day, membranes were washed with TBST three times for five minutes before incubating with an HRP-conjugated secondary antibody (1:10000 dilution, Thermo Fisher 200-403-095S). Membranes were later stripped by incubating in 0.5M NaOH for 10 min and then re-probed for total Akt levels by incubating with anti-Akt (C67E7) antibody (1:1000 dilution, Cell Signaling Technology 4691T). The density of each immunoreactive phospho-protein band was expressed as a fraction of the band for the total AKT protein in the same lane.

+ + +Reverse-phase protein array (RPPA) +

A reverse-phase protein array (RPPA) was used for precise high-throughput quantification of total and phospho-protein levels in different signalling pathways. Protein lysates from hypothalamic tissues were extracted and transferred to 384-well microarray plates to print individual slides. The slides were then labelled with validated antibodies recognizing total protein and specific phosphorylation sites. SYPRO Ruby staining was used to measure total protein levels. The data obtained from RPPA were normalized for individual samples. Each antibody’s raw signal intensity (SI) was determined by subtracting the background from the antibody-specific signals during image analysis. The normalized antibody SI was expressed using the formula N = A - C/T x M, where N is the normalized SI, A is the raw antibody SI, C is the negative control SI, T is the SI of total protein, and M is the median SI of total protein from a spot within the same sample group. Antibodies with SI less than 200 were filtered out, and quality control (QC) scores were established to discard poor-quality reactions. The coefficient of variation (CV) score was determined for each antibody and defined as the percentage of samples on a slide with a CV < 20%. Antibodies with a median technical triplicate CV >20% were also excluded from further analysis. T-tests were performed for each antibody to identify differentially expressed proteins in the SMS compared to control groups. PCA plots were generated for all samples, annotated with their experimental group, and hierarchical clustering was performed using a Python analysis tool. The vertical axis of the dendrogram represents the dissimilarity (measured as distance) between protein expressions, and the horizontal axis represents the individual test samples. The colour code (ranging from red to yellow to green) specifies the expression levels of different proteins, where red indicates low expression, yellow indicates intermediate expression, and green indicates high expression. Enrichment analysis on the differentially expressed proteins was performed using an online data resource string: version 11.5.

+ + +Electrophysiology of acute brain slices +

Mouse brains were removed and placed in ice-cold carbonated slicing artificial cerebrospinal fluid (aCSF), which contained (in mM) 124 NaCl, 3 KCl, 1.4 NaH2PO4, 26 NaHCO3,1.3 Mg SO4, 10 D-glucose, 0.5 CaCl2, and 3.5 MgCl2. Coronal sections (300 µm) were cut on a vibratome. Slices were allowed to recover at 32 °C for 40 minutes and then at 30 °C for 30 minutes to 6 hours in the recording medium, carbonated aCSF (124 NaCl, 3 KCl, 1.4 NaH2PO4, 26 NaHCO3,1.3 Mg SO4, 10 D-glucose, 2.4 CaCl2, in mM, ∼300 mOsm). Signals were recorded with a 5× gain, low-pass filtered at 2 kHz, digitized at 10 kHz (Molecular Devices Multiclamp 700B) and analyzed with pClamp 11 (Molecular Devices). Whole-cell recordings were made using 3–5 MΩ pipettes filled with an internal solution that contained (in mM) 131 potassium gluconate, 8 NaCl, 20 KCl, 2 EGTA, 10 HEPES, 2 MgATP and 0.3 Na3GTP (pH 7.2, with KOH, 300–310 mOsm with sucrose). For miniature current recording, the internal solution contained (in mM) 130 gluconate acid, 8 CsCl, 1 NaCl, 2 EGTA, 0.8 CsOH, 10 HEPES, 2 MgATP and 0.3 Na3GTP (pH 7.2, with CsOH, 300–310 mOsm adjusted with sucrose). mEPSCs were recorded in the presence of 1 µM TTX at a holding potential of - 70 mV. mIPSCs were recorded at a holding potential of +10 mV. The mIPSCs were analyzed from events during a 2-minute continuous recording that contained > 15 events/neuron. Additional calculations, statistical analysis, and plotting were performed in Microsoft Excel or Prism (GraphPad).

+ + +Spectrometric analysis of LM22A-4 levels in the brain +

We assessed the efficacy of LM22A-4, a small molecule pharmacological agent, in 8-week-old SMS and control mice. LM22A-4, chemical name N1, N3, N5tris (2hydroxyethyl) 1, 3, 5 benzenetricarboxamide (Molecular formula: C15H21N3O6, Formula weight: 339.3), was manufactured by Cayman Chemical with purity specification >98%. A stock solution was made by first dissolving LM22A-4 in DMSO (solubility =30mg/ml) and then diluting it with 0.9% isotonic sterilized saline solution (Sigma). To maximize brain concentrations, LM22A-4 was administered both intranasally (5mg/kg) and intraperitoneally (50mg/kg), as previously reported (Simmons et al, 2013). The bioavailability of this drug in the hypothalamus using these doses and routes of administration was assessed with the reverse phase LC triple-quadrupole tandem mass spectroscopic detection (LC-MS/MS) by Absorption Systems. Mice were euthanized, and hypothalamic tissue was collected at either 1 h or 3 h post-injection (n= 4/timepoint). Known amounts of LM22A-4 were added to the blank sample to generate a standard curve to verify test accuracy.

+ + +LM22A-4 treatment +

Mice were divided into four groups: Saline-injected control, Saline-injected SMS, LM22A-4- injected SMS, and LM22A-4-injected control mice. Mice were injected daily, intra-nasally (5mg/kg) and intraperitoneally (50mg/kg), for 7 weeks. Body weight was measured between 8 and 14 weeks of age. Food intake was measured from 14 to 15 weeks of age. After 15 weeks of age, mice were euthanized to collect brown adipose, visceral, subcutaneous (inguinal) and ependymal white adipose tissues and weighed using an analytical scale (Sartorius). Hypothalamic brain tissues were collected for western blot analysis. Metabolic profiling was performed on a separate cohort of mice injected for two weeks prior to the start of experiments as well as at specific weeks during the experiment.

+ + +Vertical rearing +

To measure vertical repetitive rearing, mice were taken into the testing room and allowed to habituate for 1 h prior to testing. During the test, mice were placed in the center of a 45 cm (W) × 45 cm (D) × 40 cm (H) square arena, and their movement was recorded for 10 minutes.

+ + +Statistical analysis +

The sample sizes used were based on our previous studies (Huang et al., 2016; Javed et al., 2021). The data were subjected to statistical analysis for significance by using GraphPad Prism 0.9. Significance was defined as having a p-value < 0.05. The significance levels were as follows: * p< 0.05, **p<0.01, ***p<0.0001, **** p<0.0001. The statistical tests used for each analysis (e.g., Student’s t-test, analysis of variance [ANOVA]) are indicated in the text and legends.

+ + + +Data Availability +

This study includes no data deposited in external repositories.

+ + + + +Acknowledgements +

Sehrish Javed was supported by a Fonds de Recherche du Québec - Santé (FRQS) doctoral award. This work was supported by the Rare Diseases: Models & Mechanisms Network and the Smith-Magenis syndrome Research Foundation.

+ + +References +AbadC, CookMM, CaoL, JonesJR, RaoNR, Dukes-RimskyL, PaulyR, SkinnerC, WangY, LuoFet al (2018) A Rare De Novo RAI1 Gene Mutation Affecting BDNF-Enhancer-Driven Transcription Activity Associated with Autism and Atypical Smith-Magenis Syndrome Presentation. Biology (Basel)7 +Alaimo JT, Barton LV, Mullegama SV, Wills RD, Foster RH, Elsea SH (2015) Individuals with Smith-Magenis syndrome display profound neurodevelopmental behavioral deficiencies and exhibit food-related behaviors equivalent to Prader-Willi syndrome. Res Dev Disabil 47: 27–38 +An JJ, Kinney CE, Tan JW, Liao GY, Kremer EJ, Xu B (2020) TrkB-expressing paraventricular hypothalamic neurons suppress appetite through multiple neurocircuits. Nat Commun 11: 1729 +An JJ, Liao GY, Kinney CE, Sahibzada N, Xu B (2015) Discrete BDNF Neurons in the Paraventricular Hypothalamus Control Feeding and Energy Expenditure. Cell Metab 22: 175–188 +Atasoy D, Betley JN, Su HH, Sternson SM (2012) Deconstruction of a neural circuit for hunger. Nature 488: 172–177 +Balthasar N, Dalgaard LT, Lee CE, Yu J, Funahashi H, Williams T, Ferreira M, Tang V, McGovern RA, Kenny CD et al (2005) Divergence of melanocortin pathways in the control of food intake and energy expenditure. Cell 123: 493–505 +Boltaev U, Meyer Y, Tolibzoda F, Jacques T, Gassaway M, Xu Q, Wagner F, Zhang YL, Palmer M, Holson E et al (2017) Multiplex quantitative assays indicate a need for reevaluating reported small-molecule TrkB agonists. Sci Signal 10 +Boot E, Linders CC, Tromp SH, van den Boogaard MJ, van Eeghen AM (2021) Possible underreporting of pathogenic variants in RAI1 causing Smith-Magenis syndrome. Am J Med Genet A 185: 3167–3169 +Burke LK, Darwish T, Cavanaugh AR, Virtue S, Roth E, Morro J, Liu SM, Xia J, Dalley JW, Burling K et al (2017) mTORC1 in AGRP neurons integrates exteroceptive and interoceptive food-related cues in the modulation of adaptive energy expenditure in mice. Elife 6 +Burns B, Schmidt K, Williams SR, Kim S, Girirajan S, Elsea SH (2010) Rai1 haploinsufficiency causes reduced Bdnf expression resulting in hyperphagia, obesity and altered fat distribution in mice and humans with no evidence of metabolic syndrome. Hum Mol Genet 19: 4026–4042 +Canals JM, Checa N, Marco S, Akerud P, Michels A, Perez-Navarro E, Tolosa E, Arenas E, Alberch J (2001) Expression of brain-derived neurotrophic factor in cortical neurons is regulated by striatal target area. J Neurosci 21: 117–124 +Canals JM, Pineda JR, Torres-Peraza JF, Bosch M, Martin-Ibanez R, Munoz MT, Mengod G, Ernfors P, Alberch J (2004) Brain-derived neurotrophic factor regulates the onset and severity of motor dysfunction associated with enkephalinergic neuronal degeneration in Huntington’s disease. J Neurosci 24: 7727–7739 +Caron A, Labbe SM, Lanfray D, Blanchard PG, Villot R, Roy C, Sabatini DM, Richard D, Laplante M (2016) Mediobasal hypothalamic overexpression of DEPTOR protects against high-fat diet-induced obesity. Mol Metab 5: 102–112 +Chang HC, Lee YJ, Javed S, Haque M, Chang YT, Lin YC, Oram C, Huang WH (2022a) rAAV-CRISPRa therapy corrects Rai1 haploinsufficiency and rescues selective disease features in Smith-Magenis syndrome mice. J Biol Chem: 102728 +Chang Q, Khare G, Dani V, Nelson S, Jaenisch R (2006) The disease progression of Mecp2 mutant mice is affected by the level of BDNF expression. Neuron 49: 341–348 +Chang YT, Kowalczyk M, Fogerson PM, Lee YJ, Haque M, Adams EL, Wang DC, DeNardo LA, Tessier-Lavigne M, Huguenard JR et al (2022b) Loss of Rai1 enhances hippocampal excitability and epileptogenesis in mouse models of Smith-Magenis syndrome. Proc Natl Acad Sci U S A 119: e2210122119 +Collaboration NCDRF (2017) Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet 390: 2627–2642 +Colmers PLW, Bains JS (2018) Balancing tonic and phasic inhibition in hypothalamic corticotropin-releasing hormone neurons. J Physiol 596: 1919–1929 +Cota D, Proulx K, Smith KA, Kozma SC, Thomas G, Woods SC, Seeley RJ (2006) Hypothalamic mTOR signaling regulates food intake. Science 312: 927–930 +Domeniconi M, Chao MV (2010) Transactivation of Trk receptors in spinal motor neurons. Histol Histopathol 25: 1207–1213 +Edelman EA, Girirajan S, Finucane B, Patel PI, Lupski JR, Smith AC, Elsea SH (2007) Gender, genotype, and phenotype differences in Smith-Magenis syndrome: a meta-analysis of 105 cases. Clin Genet 71: 540–550 +Gandhi AA, Wilson TA, Sisley S, Elsea SH, Foster RH (2022) Relationships between food-related behaviors, obesity, and medication use in individuals with Smith-Magenis syndrome. Res Dev Disabil 127: 104257 +Gao H, Molinas AJR, Miyata K, Qiao X, Zsombok A (2017) Overactivity of Liver-Related Neurons in the Paraventricular Nucleus of the Hypothalamus: Electrophysiological Findings in db/db Mice. J Neurosci 37: 11140–11150 +Garcia AD, Doan NB, Imura T, Bush TG, Sofroniew MV (2004) GFAP-expressing progenitors are the principal source of constitutive neurogenesis in adult mouse forebrain. Nat Neurosci 7: 1233–1241 +Gorski JA, Talley T, Qiu M, Puelles L, Rubenstein JL, Jones KR (2002) Cortical excitatory neurons and glia, but not GABAergic neurons, are produced in the Emx1-expressing lineage. J Neurosci 22: 6309–6314 +Gray J, Yeo GS, Cox JJ, Morton J, Adlam AL, Keogh JM, Yanovski JA, El Gharbawy A, Han JC, Tung YC et al (2006) Hyperphagia, severe obesity, impaired cognitive function, and hyperactivity associated with functional loss of one copy of the brain-derived neurotrophic factor (BDNF) gene. Diabetes 55: 3366–3371 +Gu F, Parada I, Yang T, Longo FM, Prince DA (2022) Chronic partial TrkB activation reduces seizures and mortality in a mouse model of Dravet syndrome. Proc Natl Acad Sci U S A 119 +Huang WH, Guenthner CJ, Xu J, Nguyen T, Schwarz LA, Wilkinson AW, Gozani O, Chang HY, Shamloo M, Luo L (2016) Molecular and Neural Functions of Rai1, the Causal Gene for Smith-Magenis Syndrome. Neuron 92: 392–406 +Huang WH, Tupal S, Huang TW, Ward CS, Neul JL, Klisch TJ, Gray PA, Zoghbi HY (2012) Atoh1 governs the migration of postmitotic neurons that shape respiratory effectiveness at birth and chemoresponsiveness in adulthood. Neuron 75: 799–809 +Huang WH, Wang DC, Allen WE, Klope M, Hu H, Shamloo M, Luo L (2018) Early adolescent Rai1 reactivation reverses transcriptional and social interaction deficits in a mouse model of Smith-Magenis syndrome. Proc Natl Acad Sci U S A +Javed S, Lee YJ, Xu J, Huang WH (2021) Temporal dissection of Rai1 function reveals brain-derived neurotrophic factor as a potential therapeutic target for Smith-Magenis syndrome. Hum Mol Genet 31: 275–288 +Javed S, Selliah T, Lee YJ, Huang WH (2020) Dosage-sensitive genes in autism spectrum disorders: From neurobiology to therapy. Neurosci Biobehav Rev 118: 538–567 +Kamitakahara A, Xu B, Simerly R (2016) Ventromedial hypothalamic expression of Bdnf is required to establish normal patterns of afferent GABAergic connectivity and responses to hypoglycemia. Mol Metab 5: 91–101 +Kocalis HE, Hagan SL, George L, Turney MK, Siuta MA, Laryea GN, Morris LC, Muglia LJ, Printz RL, Stanwood GD et al (2014) Rictor/mTORC2 facilitates central regulation of energy and glucose homeostasis. Mol Metab 3: 394–407 +Kron M, Howell CJ, Adams IT, Ransbottom M, Christian D, Ogier M, Katz DM (2012) Brain activity mapping in Mecp2 mutant mice reveals functional deficits in forebrain circuits, including key nodes in the default mode network, that are reversed with ketamine treatment. J Neurosci 32: 13860–13872 +Kron M, Lang M, Adams IT, Sceniak M, Longo F, Katz DM (2014) A BDNF loop-domain mimetic acutely reverses spontaneous apneas and respiratory abnormalities during behavioral arousal in a mouse model of Rett syndrome. Dis Model Mech 7: 1047–1055 +Lee SK, Lee S, Shin SY, Ryu PD, Lee SY (2012) Single cell analysis of voltage-gated potassium channels that determines neuronal types of rat hypothalamic paraventricular nucleus neurons. Neuroscience 205: 49–62 +Li W, Bellot-Saez A, Phillips ML, Yang T, Longo FM, Pozzo-Miller L (2017) A small-molecule TrkB ligand restores hippocampal synaptic plasticity and object location memory in Rett syndrome mice. Dis Model Mech 10: 837–845 +Loos RJF, Yeo GSH (2022) The genetics of obesity: from discovery to biology. Nat Rev Genet 23: 120–133 +Massa SM, Yang T, Xie Y, Shi J, Bilgen M, Joyce JN, Nehama D, Rajadas J, Longo FM (2010) Small molecule BDNF mimetics activate TrkB signaling and prevent neuronal degeneration in rodents. J Clin Invest 120: 1774–1785 +Miki T, Liss B, Minami K, Shiuchi T, Saraya A, Kashima Y, Horiuchi M, Ashcroft F, Minokoshi Y, Roeper J et al (2001) ATP-sensitive K+ channels in the hypothalamus are essential for the maintenance of glucose homeostasis. Nat Neurosci 4: 507–512 +Moss J, Oliver C, Arron K, Burbidge C, Berg K (2009) The prevalence and phenomenology of repetitive behavior in genetic syndromes. J Autism Dev Disord 39: 572–588 +Ogier M, Wang H, Hong E, Wang Q, Greenberg ME, Katz DM (2007) Brain-derived neurotrophic factor expression and respiratory function improve after ampakine treatment in a mouse model of Rett syndrome. J Neurosci 27: 10912–10917 +Parton LE, Ye CP, Coppari R, Enriori PJ, Choi B, Zhang CY, Xu C, Vianna CR, Balthasar N, Lee CE et al (2007) Glucose sensing by POMC neurons regulates glucose homeostasis and is impaired in obesity. Nature 449: 228–232 +Perry CA, Pravetoni M, Teske JA, Aguado C, Erickson DJ, Medrano JF, Lujan R, Kotz CM, Wickman K (2008) Predisposition to late-onset obesity in GIRK4 knockout mice. Proc Natl Acad Sci U S A 105: 8148–8153 +Quarta C, Claret M, Zeltser LM, Williams KW, Yeo GSH, Tschop MH, Diano S, Bruning JC, Cota D (2021) POMC neuronal heterogeneity in energy balance and beyond: an integrated view. Nat Metab 3: 299–308 +Rao NR, Abad C, Perez IC, Srivastava AK, Young JI, Walz K (2017) Rai1 Haploinsufficiency Is Associated with Social Abnormalities in Mice. Biology (Basel) 6 +Schultze SM, Hemmings BA, Niessen M, Tschopp O (2012) PI3K/AKT, MAPK and AMPK signalling: protein kinases in glucose homeostasis. Expert Rev Mol Med 14: e1 +Shao J, Yamashita H, Qiao L, Friedman JE (2000) Decreased Akt kinase activity and insulin resistance in C57BL/KsJ-Leprdb/db mice. J Endocrinol 167: 107–115 +Simmons DA, Belichenko NP, Yang T, Condon C, Monbureau M, Shamloo M, Jing D, Massa SM, Longo FM (2013) A small molecule TrkB ligand reduces motor impairment and neuropathology in R6/2 and BACHD mouse models of Huntington’s disease. J Neurosci 33: 18712–18727 +Slager RE, Newton TL, Vlangos CN, Finucane B, Elsea SH (2003) Mutations in RAI1 associated with Smith-Magenis syndrome. Nat Genet 33: 466–468 +Smith AC, Magenis RE, Elsea SH (2005) Overview of Smith-Magenis syndrome. J Assoc Genet Technol 31: 163–167 +Smith AC, McGavran L, Robinson J, Waldstein G, Macfarlane J, Zonona J, Reiss J, Lahr M, Allen L, Magenis E (1986) Interstitial deletion of (17)(p11.2p11.2) in nine patients. Am J Med Genet 24: 393–414 +Smith MA, Katsouri L, Irvine EE, Hankir MK, Pedroni SM, Voshol PJ, Gordon MW, Choudhury AI, Woods A, Vidal-Puig A et al (2015) Ribosomal S6K1 in POMC and AgRP Neurons Regulates Glucose Homeostasis but Not Feeding Behavior in Mice. Cell Rep 11: 335–343 +Su X, Gu D, Xu L, Liang Z, Luo X, Yang P, Yang J (2021) PI3K/Akt pathway expression in children with different obesity degrees and its relationship with glucolipid metabolism and insulin resistance. Am J Transl Res 13: 6592–6598 +Tan CL, Cooke EK, Leib DE, Lin YC, Daly GE, Zimmerman CA, Knight ZA (2016) Warm-Sensitive Neurons that Control Body Temperature. Cell 167: 47–59 e15 +Taniguchi H, He M, Wu P, Kim S, Paik R, Sugino K, Kvitsiani D, Kvitsani D, Fu Y, Lu J et al (2011) A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71: 995–1013 +Unger TJ, Calderon GA, Bradley LC, Sena-Esteves M, Rios M (2007) Selective deletion of Bdnf in the ventromedial and dorsomedial hypothalamus of adult mice results in hyperphagic behavior and obesity. J Neurosci 27: 14265–14274 +Vohra MS, Benchoula K, Serpell CJ, Hwa WE (2022) AgRP/NPY and POMC neurons in the arcuate nucleus and their potential role in treatment of obesity. Eur J Pharmacol 915: 174611 +Wu SW, Xu B (2022) Rapid and Lasting Effects of Activating BDNF-Expressing PVH Neurons on Energy Balance. eNeuro 9 +Xu B, Goulding EH, Zang K, Cepoi D, Cone RD, Jones KR, Tecott LH, Reichardt LF (2003) Brain-derived neurotrophic factor regulates energy balance downstream of melanocortin-4 receptor. Nat Neurosci 6: 736–742 +Xu B, Xie X (2016) Neurotrophic factor control of satiety and body weight. Nat Rev Neurosci 17: 282–292 +Xu Y, Lu Y, Cassidy RM, Mangieri LR, Zhu C, Huang X, Jiang Z, Justice NJ, Xu Y, Arenkiel BR et al (2019) Identification of a neurocircuit underlying regulation of feeding by stress-related emotional responses. Nat Commun 10: 3446 +Yang H, An JJ, Sun C, Xu B (2016) Regulation of Energy Balance via BDNF Expressed in Nonparaventricular Hypothalamic Neurons. Mol Endocrinol 30: 494–503 +Yeo GS, Connie Hung CC, Rochford J, Keogh J, Gray J, Sivaramakrishnan S, O’Rahilly S, Farooqi IS (2004) A de novo mutation affecting human TrkB associated with severe obesity and developmental delay. Nat Neurosci 7: 1187–1189 + + +Figures and Figure legends + + +Differentially expressed proteins in the hypothalamus of SMS mice

A heat map showing hierarchical clustering of differentially expressed proteins in the hypothalamus of SMS mice.

+ + + + +The <italic>Bdnf<sup>Cre</sup></italic> allele labels BDNF-producing neurons in the paraventricular nucleus of the hypothalamus +

(A) Schematic showing a transgenic mouse expressing BdnfCre-dependent td-Tomato (Ai9) signals in BDNF-expressing cells.

(B-E) Colocalization of BdnfCre-dependent td-Tomato signals with endogenous BDNF protein. Black: DAPI, Magenta: BDNF-producing cells; Green: endogenous BDNF. Scale bar: 50 µm, yellow arrowheads indicate Ai9 marked BDNF cells coexpressing endogenous BDNF.

+ + + + +RAI1 expression in BDNF-expressing but not oxytocin-expressing magnocellular PVH neurons +

(A-H) Representative images of DAPI expression (blue, A), oxytocin (green, B), BDNF- producing cells (Ai9 reporter, magenta, C), and RAI1 protein (cyan, D) in the PVH. Note that BDNF-producing neurons are distinct from the oxytocin-expressing neurons (E) and that RAI1 is more selectively expressed in the BDNF-expressing neurons (F-H).

(I) Quantification showing the percentage of RAI1+ cells that co-express either BDNF (left bar, magenta) or oxytocin (right bar, green) in the PVH. n=3 mice/group. Data are shown as mean±SEM.

+ + + + +Ablating <italic>Rai1</italic> from the BDNF-producing cells induces body weight gain and defective energy homeostasis +

(A) Schematic representation of the conditional knockout mice carrying either normal Rai1 alleles and the Ai9 reporter in BDNF-producing cells (top) or homozygous Rai1 deletion and the Ai9 reporter in the BDNF-producing cells.

(B) Representative images showing the expression of DAPI (white), RAI1-expressing cells (green), and BDNF-expressing cells (Ai9 reporter, magenta) in the PVH. The top panel shows the Ctrl group with RAI1-Ai9 co-expression (yellow arrowheads) and the bottom panel shows reduced RAI1 and Ai9 colocalization (yellow arrowheads). Note that RAI1 expression was still detected in non-BDNF neurons in the PVH.

(C) Quantification showing the percentage of BDNF-producing cells was not altered by Rai1 deletion, indicating that RAI1 loss did not impair the generation or survival of PVHBDNF neurons. ns indicates not significantly different, unpaired t-test, two-tailed.

(D) The BdnfCre/+ and BdnfCre/+; Rai1fl/+ female mice had similar body weights. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test.

(E-G) Blood parameter analysis of triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein and very low-density lipoprotein (LDL+VLDL) did not differ between Ctrl and cKO mice. ns indicates not significantly different, unpaired t-test, two-tailed.

(H) The respiratory exchange rate did not differ between Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(I) In the glucose tolerance test, the cKO mice showed an increased area under the curve (AUC). ***p<0.001, unpaired t-test, two-tailed.

(J) In the glucose tolerance test, cKO mice showed increased insulin levels at the beginning of the test and 30 minutes after glucose administration. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons tests, *p < 0.05, **p<0.01.

(K) In the insulin tolerance test, blood glucose level is measured after insulin injection. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05.

Data are shown as mean±SEM.

+ + + + +Metabolic profiles of male cKO mice lacking RAI1 expression in the BDNF-producing cells +

(A) Male cKO mice gained significantly more weight than Ctrl mice beginning at 15 weeks of age. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05, **p<0.01, ****p < 0.0001.

(B) Male BdnfCre/+ and BdnfCre/+; Rai1fl/+ mice had similar body weights. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(C) Echo-MRI analysis showed that 26-week-old male cKO mice had a significantly increased body weight due to increased fat but not lean mass. ns indicates the difference is not significant. unpaired t-test, two-tailed, **p<0.01, ****p<0.0001.

(D) Fat disposition analysis showing the weight of brown adipocytes (BAT), subcutaneous inguinal (S.C.ing), and epididymal white adipose tissues (eWAT). ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05, **p<0.01.

(E) Male cKO mice showed eWAT cell hypertrophy. Top: Representative images of the eWAT tissues in Ctrl and cKO mice. Scale bar = 500µm. Bottom: Frequency distribution of cellular sizes. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05, **p<0.01, ***p<0.001, ****p < 0.0001.

(F-I) Blood parameter analysis showing that male cKO mice showed significantly increased leptin levels (F) without alterations in TG (G), HDL (H), and LDL+VLDL (I). ns indicates the difference is not significant. unpaired t-test, two-tailed. *p< 0.05.

(J) No differences were found in the respiratory exchange rate in male Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(K) Male cKO mice showed a significantly increased food intake compared to Ctrl littermates., unpaired t-test, two-tailed, *p< 0.05.

(L) Male cKO mice showed increased energy expenditure at the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test, *p<0.05.

(M) Male Ctrl and cKO mice showed similar locomotor activities. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(N-P) Glucose tolerance test shows that cKO male mice are glycemic 45 minutes post intraperitoneal glucose administration (N) and had increased AUC (unpaired t-test, two-tailed) (O), despite significantly higher insulin levels before and 30 minutes after glucose administration (P), suggesting potential insulin insensitivity. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons tests, *p<0.05, **p < 0.01, ***p<0.001.

(Q) Insulin tolerance test showing that male cKO mice showed significantly higher blood glucose levels 60 minutes after intraperitoneal insulin administration. Two-way ANOVA with Šidák’s multiple comparisons test, *p<0.05, **p < 0.01.

Data are shown as mean±SEM.

+ + + + +The cellular and synaptic properties of control and <italic>Rai1</italic>- deficient PVH<sup>BDNF</sup> neurons +

(A) Representative traces of rebound and rebound repetitive firing of PVHBDNF neurons in control (black) and cKO (red).

(B) Representative AP waveforms at holding voltages of -80mV, -60mV, and -50mV in control (left) and -50mV and -45mV in cKO (right).

(C) The half-width of AP initiated at a holding voltage of -60mV (neuronal numbers in brackets, ns indicates the difference is not significant, unpaired t-test, two-tailed).

(D) Representative traces of miniature recordings in control (black) and cKO (red) PVHBDNF neurons. The traces were recorded with a Cs-based internal at a holding voltage of -70mV in aCSF (top), spontaneous mEPSCs were recorded in aCSF containing 1uM TTX at a holding voltage of -70mV (middle), and mIPSCs were recorded at holding voltage of +10mV (bottom).

(E) Quantification showing that the cellular size of cKO PVHBDNF neurons was significantly smaller than Ctrl neurons. unpaired t-test, two-tailed. ****p<0.0001.

+ + + + +Metabolic profile of mice lacking <italic>Rai1</italic> in PVH<sup>BDNF</sup> neurons +

(A) Representative images showing co-expression of virus-infected cells (GFP/green), BdnfCre- dependent Ai9 signals (BDNF cells, magenta), and endogenous RAI1 (cyan). Yellow arrowheads indicate BDNF neurons infected with AAV and in these cells, RAI1 signals were diminished in Exp but not Ctrl mice. Scale bar = 50µm.

(B-C) The total number of PVHBDNF neurons (B) and total number of PVHBDNF neurons infected with AAVs (C) per slice show that Rai1 deletion or virus infection did not alter the total number of PVHBDNF neurons. ns indicates the difference is not significant. unpaired t-test, two-tailed.

(D) Rai1 expression was significantly reduced in PVHBDNF neurons infected with both AAVs compared to PVHBDNF neurons infected with sgRai1-GFP AAV alone. unpaired t-test, two-tailed, **p<0.01.

(E) The number of PVHBDNF neurons that co-express RAI1 was significantly reduced in the Exp group. unpaired t-test, unpaired, **p<0.01.

(F) Representative image showing lack of viral expression in the PVH neighbouring nuclei. ARC: arcuate nucleus. VMH: ventromedial nucleus of the hypothalamus. 3V: third ventricle. Scale bar = 50µm.

(G) The respiratory exchange rate remained similar in Ctrl and Exp mice. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(H-J) Blood parameter analysis showing TG, HDL, and LDL+VLDL levels were similar in Ctrl and Exp mice. ns indicates the difference is not significant. unpaired t-test, two-tailed.

(K) In the insulin tolerance test, the Exp group showed a trend towards increased blood glucose levels after insulin injection. Two-way ANOVA, Šídák’s multiple comparisons test.

Data are shown as mean±SEM.

+ + + + +LM22A-4 treatment does not disrupt energy homeostasis and repetitive rearing in the Ctrl mice +

(A) LM22A-4 concentration in the hypothalamus and forebrain of SMS mice 1 hour (black dots) and 3 hours (red dots) after simultaneous intranasal and IP injections.

(B) Four independent sets of western blotting analyses showing reduced p-AKT levels in the hypothalamus of saline-treated SMS mice, which were increased by LM22A-4 treatment. Total AKT protein was used as a control.

(C) Similar body weight of Ctrl mice treated with either saline or LM22A-4. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(D) Energy expenditure was not different between groups in the light vs. dark phase. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(E) Respiratory exchange rates were not different between groups in the light vs. dark phase. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(F) Repetitive rearing behaviour in Ctrl mice was not altered by LM22A-4 treatment. ns indicates the difference is not significant. unpaired t-test, two-tailed.

Data are shown as mean±SEM.

+ + + + +LM22A-4 treatment in adult SMS mice is insufficient to improve social interaction deficit +

(A) Schematic showing the tube test that assesses social interaction between stranger mice of different genotypes.

(B) 7-week-old Ctrl mice won significantly more than SMS mice before the LM22A-4 treatment. unpaired t-test, two-tailed. ****p<0.0001.

(C) 10-week-old saline-injected Ctrl mice won significantly more than the saline-injected SMS mice (two weeks post-LM22A-4 treatment). unpaired t-test, two-tailed, ***p<0.001.

(D) 10-week-old saline-injected Ctrl mice won significantly more than LM22A-4 injected SMS mice. unpaired t-test, two-tailed, *p<0.05.

(E) 10-week-old saline-injected SMS mice showed a similar winning rate to LM22A-4-injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed.

Data are shown as mean±SEM.

+ + + + + + + + + + + + + + + + + + + + + +
diff --git a/test/fixtures/2022.12.29.522272/2022.12.29.522272.xml b/test/fixtures/2022.12.29.522272/2022.12.29.522272.xml new file mode 100644 index 00000000..5de94848 --- /dev/null +++ b/test/fixtures/2022.12.29.522272/2022.12.29.522272.xml @@ -0,0 +1,334 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2022.12.29.522272 +1.3 + + +Regular Article + + +New Results + + +Biochemistry + + + +The deubiquitinase Ubp3/Usp10 constrains glucose-mediated mitochondrial repression via phosphate budgeting + + + +http://orcid.org/0000-0002-1078-5476 +VengayilVineeth +1 +2 + + +NiphadkarShreyas +1 +2 + + +AdhikarySwagata +1 +2 + + +http://orcid.org/0000-0002-3609-4032 +VarahanSriram +1 + + +http://orcid.org/0000-0002-0861-5080 +LaxmanSunil +1 +* + +Institute for Stem Cell Science and Regenerative Medicine (DBT-inStem), GKVK Post Bellary Road, Bangalore 560065 +Manipal Academy of Higher Education, Manipal, Karnataka 576104, India + + +correspondence: sunil@instem.res.in + +2023 +2022.12.29.522272 + + +29 +12 +2022 + + +22 +8 +2023 + + +23 +8 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at http://creativecommons.org/licenses/by-nc-nd/4.0/ + + + +Abstract +

Many cells in high glucose repress mitochondrial respiration, as observed in the Crabtree and Warburg effects. Our understanding of biochemical constraints for mitochondrial activation is limited. Using a Saccharomyces cerevisiae screen, we identified the conserved deubiquitinase Ubp3 (Usp10), as necessary for mitochondrial repression. Ubp3 mutants have increased mitochondrial activity despite abundant glucose, along with decreased glycolytic enzymes, and a rewired glucose metabolic network with increased trehalose production. Utilizing Δubp3 cells, along with orthogonal approaches, we establish that the high glycolytic flux in glucose continuously consumes free Pi. This restricts mitochondrial access to inorganic phosphate (Pi), and prevents mitochondrial activation. Contrastingly, rewired glucose metabolism with enhanced trehalose production and reduced GAPDH (as in Δubp3 cells) restores Pi. This collectively results in increased mitochondrial Pi and derepression, while restricting mitochondrial Pi transport prevents activation. We therefore suggest that glycolytic-flux dependent intracellular Pi budgeting is a key constraint for mitochondrial repression.

+ + +Keywords +mitochondria +glycolysis +phosphate +deubiquitinase +Crabtree + + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Title and figures updated.

 + + + + + +Introduction +

Rapidly proliferating cells have substantial metabolic and energy demands in order to increase biomass (1, 2). This includes a high ATP demand, obtained from cytosolic glycolysis or mitochondrial oxidative phosphorylation (OXPHOS), to fuel multiple reactions (3). Interestingly, many rapidly proliferating cells preferentially rely on ATP from glycolysis/fermentation over mitochondrial respiration even in oxygen-replete conditions, and is the well-known Warburg effect (4, 5). Many such cells repress mitochondrial processes in high glucose, termed glucose-mediated mitochondrial repression or the Crabtree effect (6, 7). This is observed in some tumors (5), neutrophils (8), activated macrophages (9), stem cells (10–12), and famously Saccharomyces cerevisiae (7). Numerous studies have identified signaling programs or regulators of glucose-mediated mitochondrial repression. However, biochemical programs and regulatory processes in biology evolve around key biochemical constraints (13). The biochemical constraints for mitochondrial repression remain unresolved (14, 15).

+

There are two hypotheses on the biochemical principles driving mitochondrial repression. The first proposes direct roles for glycolytic intermediates in driving mitochondrial respiration, by inhibiting specific mitochondrial outputs (16, 17). The second hypothesizes that a competition between glycolytic and mitochondrial processes for mutually required metabolites/co-factors such as pyruvate, ADP or inorganic phosphate (Pi) could determine the extent of mitochondrial repression (14, 15, 18). These are not all mutually exclusive, and a combination of these factors might dictate mitochondrial repression in high glucose. However, any hierarchies of importance are unclear (19), and experimental data for the necessary constraints for mitochondrial repression remains incomplete.

+

One approach to resolve this question has been to identify regulators of metabolic state under high glucose. Post-translational modifications (PTMs) regulated by signaling systems can regulate mitochondrial repression (20–22). Ubiquitination is a PTM that regulates global proteostasis (23, 24), but the roles of ubiquitination-dependent processes in regulating mitochondrial repression are poorly explored. Ubiquitination itself is determined by the balance between ubiquitination, and deubiquitinase (DUB) dependent deubiquitination (25). Little is known about the roles of DUBs in regulating metabolic states, making the DUBs interesting candidate regulators of mitochondrial repression.

+

In this study, using an S. cerevisiae DUB knock-out library-based screen, we identified the evolutionarily conserved deubiquitinase Ubp3 (mammalian Usp10) as required for mitochondrial repression in high glucose. Loss of Ubp3 resulted in mitochondrial activation, along with a reduction in the glycolytic enzymes-phosphofructokinase 1 (Pfk1) and GAPDH (Tdh2 and Tdh3). This consequently reroutes glucose flux and increases trehalose biosynthesis. This metabolic rewiring increases Pi release from trehalose synthesis, and decreases Pi consumption in glycolysis, to cumulatively increase Pi pools. Using ubp3Δ cells along with independent analysis of wild-type cells, and isolated mitochondrial fractions, we establish that glycolytic-flux dependent Pi allocations to mitochondria determines mitochondrial activity. Through these data, we propose how intracellular Pi balance as controlled by glycolytic flux, is a biochemical constraint for mitochondrial repression.

+ + +Results + +A deubiquitinase deletion screen identifies Ubp3 as a regulator of glucose-mediated mitochondrial repression +

In cells such as S. cerevisiae, high glucose represses mitochondrial activity as well as OXPHOS-dependent ATP synthesis (7, 26) (illustrated in Figure 1A). Our initial objective was to identify proteostasis regulators of glucose-mediated mitochondrial repression. We generated and used a deubiquitinase (DUB) deletion strain library of S. cerevisiae (Figure S1A), to unbiasedly identify regulators of mitochondrial repression by measuring the fluorescence intensity of a potentiometric dye Mitotracker CMXRos (illustrated in Figure 1B). Using this screen, we identified DUB mutants with altered mitochondrial membrane potential (Figure 1C, Figure S1A). Note: WT cells in a respiratory medium (2% ethanol) were used as a control to estimate maximum mitotracker fluorescence intensity (Figure S1B).

+ + +A deubiquitinase deletion screen identifies Ubp3 as a regulator of glucose-mediated mitochondrial repression +

A) Schematic depicting glucose-mediated mitochondrial repression (Crabtree effect).

B) Schematic describing the screen with a yeast DUB KO library to identify regulators of Crabtree effect.

C) Identifying DUB knockouts with altered mitochondrial potential. Heat map shows relative mitochondrial membrane potential of 19 deubiquitinase deletions in high glucose, from 2 biological replicates. Also see figures S1A, S1B.

D) The deubiquitinase activity of Ubp3 and repression of mitochondrial membrane potential. WT, ubp3Δ, and Ubp3C469A were grown inhigh glucose and relative mitochondrial membrane potential was measured. Data represent mean ± SD from three biological replicates (n=3). Also see figure S1D.

E) Effect of loss of Ubp3 on ETC complex IV subunit Cox2. WT and ubp3Δ were grown in high glucose, and Cox2 was measured (western blot using an anti-Cox2 antibody). A representative blot (out of 3 biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

F) Basal oxygen consumption rate (OCR) in high glucose in ubp3Δ. WT and ubp3Δ were grown in high glucose, and OCR was measured. Basal OCR corresponding to ∼3 x 10^5 cells, from two independent experiments (n=2), normalized to the OD600 is shown. Bar graph representations are shown in the inset. Data represent mean ± SD.

G) Total ATP levels in ubp3Δ and WT. WT and ubp3Δ were grown in high glucose, and total ATP were measured. Data represent mean ± SD from three biological replicates (n=3).

H) Dependence of ubp3Δ on mitochondrial ATP. WT and ubp3Δ cells were grown in high glucose, and treated with 1 mM sodium azide for 45 minutes. Total ATP levels in sodium azide treated and untreated cells were measured. Data represent mean ± SD (n=3).

I) Requirement for mitochondrial respiration in high glucose in ubp3Δ. Serial dilution growth assay in high glucose in the presence/absence of sodium azide (1 mM) are shown. Also see figure S1F.

Data information: **p<0.01, ***p<0.001.

+ + +

A prominent ‘hit’ was the evolutionarily conserved deubiquitinase Ubp3 (Figure 1C), (homologous to mammalian Usp10) (Figure S1C). Due to its high degree of conservation across eukaryotes (Figure S1C) as well as putative roles in metabolism or mitochondrial function (27–29), we focused our further attention on this DUB. Cells lacking Ubp3 showed an ∼1.5-fold increase in mitotracker fluorescence (Figure 1C, Figure S1A). Cells with catalyticaly inactive Ubp3 (Ubp3C469A), showed increased mitochondrial potential comparable to ubp3Δ (Figure 1D). This data confirmed that Ubp3 catalytic activity of is required to fully repress mitochondrial activity under high glucose. The catalytic site mutation did not affect steady-state Ubp3 levels (Figure S1D).

+

Next, to assess the requirement of Ubp3 for mitochondrial function, we quantified the electron transport chain (ETC) complex IV subunit Cox2 (30)(Figure 1E). ubp3Δ had higher Cox2 than WT (Figure 1E). There was no increase in mitochondrial outer membrane protein Tom70 (Figure S1E), suggesting that the increased Cox2 is not merely because of higher total mitochondrial content. We next measured the basal oxygen consumption rate (OCR) of ubp3Δ, and basal OCR was higher in ubp3Δ, indicating higher respiration (Figure 1F).

+

Next, we asked if mitochondrial ATP synthesis was higher in ubp3Δ. The total ATP levels in WT and ubp3Δ were comparable (Figure 1G). However, upon treatment with the ETC complex IV inhibitor sodium azide, ATP levels in WT were higher than ubp3Δ, contributing to ∼60% of the total ATP (Figure 1H). In contrast, the ATP levels in ubp3Δ after sodium azide treatment were ∼ 40% of the total ATP (Figure 1H). These data suggest a higher contribution of mitochondrial ATP synthesis towards the total ATP pool in ubp3Δ.

+

We next asked if ubp3Δ required higher mitochondrial activity for growth. In high glucose, WT show minimal growth inhibition in the presence of sodium azide, indicating lower reliance on mitochondrial function (Figure 1I). Contrastingly, ubp3Δ or Ubp3C469A show a severe growth defect in the presence of sodium azide (Figure 1I). Deletion of ETC components Cox2, and the ATP synthase subunits Atp1 and Atp10 results in a severe growth defect in ubp3Δ compared to WT (Figure S1F). Together, these results indicate that the loss of Ubp3 makes cells dependent on mitochondrial ATP synthesis in high glucose.

+

Collectively these data show that in 2% glucose, ubp3Δ have high mitochondrial activity, respiration, and rely on this mitochondrial function for ATP production and growth. We therefore decided to use ubp3Δ cells to start delineating requirements for glucose-mediated mitochondrial repression.

+ + +Key glycolytic enzymes decrease and glucose flux is rerouted in <italic>ubp3Δ</italic> cells +

Glucose-6 phosphate (G6P) is the central node in glucose metabolism where carbon-allocations are made towards distinct metabolic arms, primarily: glycolysis, the pentose phosphate pathway (PPP), and trehalose biosynthesis (Figure 2A). We first compared amounts of two key (‘rate-controling’) glycolytic enzymes - Phosphofructokinase 1 (Pfk1), GAPDH isozymes (Tdh2, Tdh3) (3), along with the enolase isozymes (Eno1, Eno2) in WT and ubp3Δ cells. Pfk1, Tdh2, and Tdh3 substantially decreased in ubp3Δ (but not Eno1 and Eno2) (Figure 2B, S2A). Since deubiquitinases can control protein amounts by regulating proteasomal degradation, we asked if the decrease in Pfk1, Tdh2 and Tdh3 in ubp3Δ is due to proteasomal degradation. To test this, we measured the levels of these enzymes in ubp3Δ after treatement with proteasomal inhibitor MG132. We did not observe any rescue in the levels of these enzymes in MG132 treated samples, suggesting that the decreased levels of these enzymes were not due to increased proteasomal degradation (Figure S2B). To further understand if these enzyme transcripts are altered in ubp3Δ, we measured the mRNA levels of PFK1, TDH2 and TDH3 in WT, ubp3Δ and Ubp3C469A cells. The transcripts of all the three genes in ubp3Δ and Ubp3C469A cells decreased (Figure S2C), suggesting that Ubp3 regulates the transcripts of these glycolytic enzyme genes. The reduction in the Pfk and GAPDH enzyme amounts was intriguing, because the Pfk and GAPDH steps are critical in determining glycolytic flux (3, 31). A reduction in these enzymes could therefore decrease glycolytic flux, and would reroute glucose (G6P) allocations via mass action towards other branches of glucose metabolism, primarily the pentose phosphate pathway as well as trehalose biosynthesis (Figure 2A). To assess this, we first measured the steady-state levels of key glycolytic and pentose phosphate pathway (PPP) intermediates, and trehalose in WT or ubp3Δ using targeted LC-MS/MS. Glucose-6/fructose-6 phosphate (G6P/F6P) increased in ubp3Δ (Figure 2C). Concurrently, trehalose, and the PPP intermediates ribose 5-phosphate (R5P) and sedoheptulose 7-phosphate (S7P), increased in ubp3Δ (Figure 2C).

+ + +Key glycolytic enzymes decrease and glucose flux is rerouted in <italic>ubp3Δ</italic> cells +

A) A schematic illustrating directions of glucose-6-phosphate (G6) flux in cells. Glucose is converted to G6P, a precursor for trehalose, the pentose phosphate pathway (PPP), and glycolysis.

B) Effect of loss of Ubp3 on key glycolytic enzymes. WT and ubp3Δ were grown in high glucose and the Pfk1, Tdh2, and Tdh3 levels were measured by western blot using an anti-FLAG antibody. A representative blot (out of three biological replicates, n=3) and their quantification are shown. Data represent mean ± SD. Also see figure S2A.

C) Steady-state metabolite amounts in WT and ubp3Δ in high glucose. Relative steady-state levels of trehalose, major glycolytic, and PPP intermediates were estimated in WT and ubp3Δ. Data represent mean ± SD from three biological replicates (n=3). Also see Appendix Table S3.

D) Relative glycolytic and trehalose synthesis flux in WT and ubp3Δ. Relative 13C-label incorporation into trehalose and glycolytic intermediates, after a pulse of 1% 13C6 glucose is shown. Data represent mean ± SD from three biological replicates (n=3). Also see Appendix Table S3, figures S2B, S2C.

E) Ethanol production in ubp3Δ. WT and ubp3Δ were grown in high glucose and ethanol in the media was measured. Data represent mean ± SD from three biological replicates (n=3).

F) Relative rate of ethanol production in WT vs ubp3Δ. WT and ubp3Δ were grown in high glucose (to OD600 ∼ 0.6), equal numbers of cells were shifted to fresh medium (high glucose) and ethanol concentration in the medium was measured temporally. Data represent mean ± SD from three biological replicates (n=3)

Data information: *p<0.05, **p<0.01, ***p<0.001.

+ + +

Since steady-state metabolite amounts cannot separate production from utilization, in order to unambiguously assess if glycolytic flux is reduced in ubp3Δ cells, we utilized a pulse labeling of 13C6 glucose, following which the label incorporation into glycolytic and other intermediates was measured. Note that because glycolytic flux is very high in yeast, this experiment would require rapid pulsing and extraction of metabolites in order to stay in a linear range and avoid label saturation. We therefore established a very short time point of label-addition, quenching and metabolite extraction post 13C glucose pulse. Since flux saturates/reaches steady-state in seconds, we first ensured that the label incorporation into individual metabolites after the 13C glucose pulse was in the linear range, and for early glycolytic intermediates this was seconds after glucose addition. This new methodology is extensively described in materials and methods, with required controls shown in Figure S2E. WT and ubp3Δ cells were grown in high glucose, pulsed with 13C6 glucose, and the relative 13C label incorporation into glycolytic intermediates and trehalose were measured, as shown in the schematic (Figure S2D). In ubp3Δ, 13C label incorporation into G6P/F6P as well as trehalose substantially increased (Figure 2D). Contrastingly, 13C label incorporation into glycolytic intermediates F1,6BP, G3P, 3PG and PEP decreased, indicating decreased glycolytic flux (Figure 2D). We next measured ethanol concentrations and production rates, as an additional output of relative glycolytic rates. We observed decreased steady-state ethanol levels, as well as ethanol production rates in ubp3Δ (Figure 2F).

+

Collectively, these results reveal that that reduced Pfk1 and GAPDH (in ubp3Δ) decrease glucose flux via glycolysis, which results in rewired glucose flux towards trehalose biosynthesis and the PPP.

+ + +Rerouted glucose flux results in phosphate (Pi) accumulation +

We therefore asked if the proteomic state, as observed in ubp3Δ cells, could provide clues to explain the coupling between mitochondrial derepression and rerouted glucose flux. A recent study by Isasa et al., had systematically quantified the changes in protein levels in ubp3Δ cells (28). We therefore reanalyzed this extensive dataset, looking for changes in proteins that would correlate with these metabolic processes. Notably, we observed increased levels of proteins of the mitochondrial ETC and respiration, and decreased amounts of glucose metabolizing enzymes in ubp3Δ (Figure 3A). Additionally, multiple proteins involved in regulating phosphate (Pi) homeostasis were decreased in ubp3Δ (Figure 3A) (28). We had recently uncovered a reciprocal coupling of Pi homeostasis with the different arms of glucose metabolism, particularly trehalose biosynthesis (32, 33), and therefore wondered if a glycolytic-flux dependent change in Pi homeostatsis had any role in mitochondrial respiration. Our hypothesis was refined based on the reasoning given below.

+ + +Rerouted glucose flux results in phosphate (Pi) accumulation +

A) Changes in protein levels in ubp3Δ (dataset from Isasa et al., 2015). ubp3Δ cells have an increase in proteins involved in mitochondrial respiration and decrease in proteins involved in glucose and phosphate metabolism.

B) Schematic showing maintenance of Pi balance during glycolysis. Trehalose synthesis from G6P releases Pi, and the conversion of G3P to 1,3BPG by GAPDH consumes Pi. In ubp3Δ, trehalose biosynthesis (which releases Pi) increases. ubp3Δ have decreased GAPDH, which will decrease Pi consumption. This increase in Pi release along with decreased Pi consumption could increase cytosolic Pi.

C) Intracellular Pi levels in WT and ubp3Δ. WT and ubp3Δ were grown in high glucose and the total free phosphate (Pi) levels were estimated. WT in high Pi (2% glucose, 10mM Pi) was a positive control. Data represent mean ± SD from three biological replicates (n=3). Also see figure S3A.

D) Pho regulon responses in WT and ubp3Δ. Protein levels of Pho84-FLAG and Pho12-FLAG were compared between WT grown in high glucose and in high Pi, ubp3Δ in high glucose with or without a shift to a no-Pi medium for one hour, by western blot. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

E) Contribution of trehalose synthesis as a Pi source. WT, tps2Δ, ubp3Δ, and ubp3Δtps2Δ were grown in high glucose and the total Pi levels were estimated. Data represent mean ± SD from three biological replicates (n=3). Also see figure S3B.

F) Loss of GAPDH isozymes Tdh2 and Tdh3 and effect on Pi. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and total Pi was estimated. Data represent mean ± SD from three biological replicates (n=3).

G) Pi levels in ubp3Δ and tdh2Δtdh3Δ cells in ethanol medium. WT, ubp3Δ, and tdh2Δtdh3Δ cells were grown in ethanol medium and the total Pi levels were estimated. Data represent mean ± SD from three biological replicates (n=3).

Data information: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

+ + +

One explanation for mitochondrial repression can be an internal competition for shared metabolites/co-factors between (cytosolic) glycolytic and mitochondrial processes, that mitochondria might not be sufficiently able to access (14, 18, 19). In this context, a plausible role for inorganic phosphate (Pi) in regulating mitochondrial repression can be hypothesized. Cytosolic glycolysis requires rapid, high consumption of net Pi (19, 34, 35), and this could possibly limit the Pi that is continuously available for mitochondrial use (18, 36), thereby repressing mitochondria. Contextually, the balance between reactions releasing vs consuming Pi could explain changes in global Pi levels (33). A continuous hub of Pi consumption is glycolysis. Here, GAPDH catalyzes G3P to 1,3BPG, converting ADP to ATP, while concurrently consuming a molecule of Pi (35, 37). This Pi that goes into ATP will subsequently be used for nucleotide biosynthesis, polyphosphate biosynthesis and protein phosphorylation (33, 38). Therefore, we can surmise that in high glycolytic flux, the production of ATP, nucleotides and polyphosphates is concurrent with Pi consumption (37, 39, 40). Could this reaction therefore limit cytosolic Pi for the mitochondria (as illustrated in Figure 3B)? Notably, ubp3Δ have reduced GAPDH levels and decreased glycolytic flux. Second, trehalose synthesis is a Pi releasing reaction, and a major source of free Pi that is critical for Pi homeostasis (32, 35). Flux through this reaction is also substantially higher in ubp3Δ. Therefore, these cells might have increased Pi release (via trehalose), coupled with decreased Pi consumption (via decreased GAPDH). We therefore asked if total Pi increases in ubp3Δ (Figure 3B)?

+

To test this, we directly assessed total Pi levels in ubp3Δ and WT. ubp3Δ cells had higher Pi than WT, and this was comparable to Pi in WT grown in excess Pi (Figure 3C). Similarly, Pi amounts also increased in Ubp3C469A (Figure S3A). Therefore, the loss of Ubp3 increases intracellular Pi levels. Next, we asked if ubp3Δ cells exhibit signatures of a ‘high Pi’ state. S. cerevisiae maintains internal Pi balance by controlling the expression of multiple genes collectively known as the Pho regulon (41). The Pho regulon is induced under Pi limitation, and repressed during Pi sufficiency (32, 41). We assessed two major Pho proteins (Pho84: a high-affinity membrane Pi transporter, and Pho12: an acid phosphatase) in WT and ubp3Δ in high glucose. ubp3Δ cells have lower amounts of Pho84 and Pho12 (Figure 3D). Further, upon shifting to low Pi for one hour, Pho84 and Pho12 increased in ubp3Δ. These data suggest that reduced Pho84 and Pho12 amounts in ubp3Δ are because of increased Pi, and not due to altered Pho regulon function itself (Figure 3D). These data clarify earlier observations from ubp3Δ which noted reduced Pho proteins (28). Therefore, ubp3Δ constitutively have higher Pi, and likely a consequent decrease in Pho proteins.

+

We next asked if the increased Pi in ubp3Δ is because of altered G6P allocations towards different end-points, particularly trehalose synthesis, which can be a major node of Pi restoration (32, 35). We assessed the contribution of increased trehalose synthesis towards the high Pi in ubp3Δ, by estimating Pi levels in the absence of trehalose 6-phosphate phosphatase (Tps2), which catalyzes the Pi-releasing step in trehalose synthesis. Notably, loss of Tps2 in ubp3Δ decreased Pi (Figure 3E). There was no additive difference in Pi between tps2Δ and ubp3Δtps2Δ (Figure 3E). As an added control, we assessed trehalose in WT and ubp3Δ in the absence of Tps2, and found no difference (Figure S3B). Therefore, increasing G6P flux towards trehalose biosynthesis is a major source of the increased Pi in ubp3Δ.

+

Since the major GAPDH isozymes, Tdh2 and Tdh3, are reduced in ubp3Δ, we directly asked if reducing GAPDH can decrease Pi consumption and increase Pi. To assess this, we generated tdh2Δtdh3Δ cells, which exhibit a growth defect, but are viable, permitting further analysis. tdh2Δtdh3Δ had higher Pi in high glucose (Figure 3F). Expectedly, we observed a decrease in ethanol in tdh2Δtdh3Δ (Figure S3C), along with an accumulation of F1,6 BP and G3P, and decreased 3PG and PEP (Figure S3D). However, G6P and trehalose levels between WT and tdh2Δtdh3Δ were comparable (Figure S3D, S3E). These data sugest that unlike in ubp3Δ, the increased Pi in tdh2Δtdh3Δ comes mainly from decreased Pi consumption (GAPDH step). To further assess the role of reduced glycolytic flux in increasing Pi (ubp3Δ and tdh2Δtdh3Δ), we measured the Pi in these cells growing in a gluconeogenic medium – 2%, ethanol. In this scenario, the GAPDH catalyzed reaction will be reversed, converting 1,3BPG to G3P, which should release and not consume Pi. Compared to WT, the Pi levels decreased in ubp3Δ and tdh2Δtdh3Δ (Figure 3G), suggesting that the changes in Pi in these mutants is driven by the relative change in Pi release vs consumption.

+

These results collectively indicate that the combined effect of increased Pi release coming from trehalose synthesis, and decreased Pi consumption from reduced GAPDH increase Pi levels in ubp3Δ.

+ + +Mitochondrial Pi availability correlates with mitochondrial activity in <italic>ubp3Δ</italic> +

We therefore wondered if this observed Pi increase from the combined rewiring of glucose metabolism resulted in more Pi becoming accessible to the mitochondria. This would effectively result in Pi budgeting between cytosolic glycolysis and mitochondria based on the relative flux in different arms of glucose metabolism, by increasing Pi availability for the mitochondria (Figure 4A). To test if this were so, we first asked if the mitochondria in ubp3Δ have increased Pi. Mitochondria were isolated by immunoprecipitation from WT and ubp3Δ, the isolation efficiency was analyzed (Figure S4A), and relative Pi levels compared. Pi levels were normalised to Idh1 (isocitrate dehydrogenase) in isolated mitochondria, since Idh1 protein levels did not decrease in ubp3Δ (Figure S4B), and consistent with (28). Mitochondrial Pi was higher in ubp3Δ (Figure 4B). We next asked if the increased mitochondrial activity in ubp3Δ is a consequence of high Pi. If this were so, tdh2Δtdh3Δ should partially phenocopy ubp3Δ with respect to mitochondrial activity. Consistently, tdh2Δtdh3Δ have higher mitotracker intensity as well as Cox2 levels compared to WT (Figure S4C, Figure 4C). Also consistent with this, a high basal OCR in tdh2Δtdh3Δ was observed (Figure 4D), together indicating high mitochondrial activity in tdh2Δtdh3Δ, which is comparable to ubp3Δ.

+ + +Mitochondrial Pi availability correlates with mitochondrial activity in <italic>ubp3Δ</italic> +

A) A hypothetical mechanism of cytosolic free Pi controling mitochondrial activity by regulating mitochondrial Pi availability.

B) Mitochondrial Pi amounts in WT vs ubp3Δ. Mitochondria were isolated by immunoprecipitation from WT and ubp3Δ and mitochondrial Pi estimated. Mitochondrial Pi levels (normalised to Idh1) are shown. Data represent mean ± SD from three biological replicates (n=3). Also see figure S4A, B.

C) Cox2 protein in tdh2Δtdh3Δ. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and Cox2 protein was estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

D) Basal OCR levels in tdh2Δtdh3Δ. WT, ubp3Δ, and tdh2Δtdh3Δ were grown in high glucose and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD. Also see figure S4C.

E) Comparative Pi amounts and Cox2 levels in ubp3Δ, tdh2Δtdh3Δ, WT cells. WT cells were grown in high glucose, ubp3Δ and tdh2Δtdh3Δ were grown in high glucose and low Pi, and Cox2 protein was estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S4D, S4F.

F) Pi amounts and basal OCR in ubp3Δ and tdh2Δtdh3Δ vs WT cells. WT cells were grown in high glucose, ubp3Δ and tdh2Δtdh3Δ were grown in high glucose and low Pi, and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD.

G) Effect of loss of mitochondrial Pi transporter Mir1 on Cox2protein. WT, ubp3Δ, mir1Δ, and mir1Δubp3Δ were grown in high glucose and Cox2 amounts compared. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD.

H) Relationship of mitochondrial Pi transport and basal OCR in WT vs ubp3Δ. WT, ubp3Δ, mir1Δ, and mir1Δubp3Δ cells were grown in high glucose and basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD.

Data information: *p<0.05, **p<0.01, ****p<0.0001.

+ + +

We next asked if higher Pi is necessary to increase mitochondrial activity in ubp3Δ and tdh2Δtdh3Δ. Since glycolysis is defective in ubp3Δ and tdh2Δtdh3Δ, it is necessary to distinguish the effect of high Pi vs. only the effect of low glycolysis in activating mitochondria. Logically, if decreased glycolysis (independent of Pi) is sufficient to activate mitochondria, bringing down the Pi levels in ubp3Δ to that of WT should not affect mitochondrial activity. Notably, ubp3Δ grown in low (1 mM) Pi have Pi levels similar to WT in standard (normal Pi) medium (Figure S4D). ubp3Δ grown in low Pi also had decreased ethanol, suggesting reduced glycolysis (Figure S4E). Therefore, we used this condition to further understand the role of Pi in inducing mitochondrial activity. Mitotracker fluorescence decreased in both ubp3Δ and tdh2Δtdh3Δ in low Pi (Figure S4F). Note: basal mitotracker fluorescence in WT also decreases in low Pi, which is consistent with a required role of Pi for mitochondrial activity (Figure S4F). Similarly, Cox2 levels were reduced in both ubp3Δ and tdh2Δtdh3Δ (Figure 4E). Consistent with both reduced mitotracker intensity and Cox2 levels, basal OCR also decreased in both ubp3Δ and tdh2Δtdh3Δ in low Pi (Figure 4F). These data collectively suggest that high intracellular Pi is necessary to increase mitochondrial activity in ubp3Δ and tdh2Δtdh3Δ.

+

Next, we asked how mitochondrial Pi transport regulates mitochondrial activity. Mir1 and Pic2 are mitochondrial Pi transporters, with Mir1 being the major Pi transporter (42, 43). We first limited mitochondrial Pi availability in ubp3Δ by knocking-out MIR1. In mir1Δ, the increased Cox2 observed in ubp3Δ was no longer observed (Figure 4G). Consistent with this, we observed no further increase in the basal OCR in mir1Δubp3Δ compared to mir1Δ (Figure 4H). Furthermore, mir1Δ showed decreased Cox2 as well as basal OCR even in WT cells (Figure 4G, Figure 4H). These data together suggest that mitochondrial Pi transport is critical for increasing mitochondrial activity in ubp3Δ, and in maintaining basal mitochondrial activity even in high glucose.

+

As a control, no significant increase in Mir1 and Pic2 was observed in ubp3Δ (Figure S4G), suggesting that ubp3Δ do not increase mitochondrial Pi by merely increasing the Pi transporters, but rather by increasing available Pi pools.

+

Taken together, these data suggest the possibility that increased cytosolic Pi increases the mitochondrial Pi pool, which increases mitochondrial activity. Decreasing mitochondrial Pi by either reducing total Pi, or by reducing mitochondrial Pi transport decreases mitochondrial activity.

+ + +Mitochondrial Pi availability constrains mitochondrial activity under high glucose +

So far, these data suggest that the cytosolic Pi available for the mitochondria can dermine the extent of mitochondrial activity. Therefore, we further investigated if mitochondrial Pi allocation was a necessary constraint for glucose-mediated mitochondrial repression.

+

To test this, we first asked how important mitochondrial Pi transport was to switch to increased respiration. In WT yeast, low glucose or glycolytic inhibition will result in increased respiration (22). What happens therefore if we restrict mitochondrial Pi in this context? For this, we measured the basal OCR in WT and mir1Δ after switching from high (2%) to low (0.1%) glucose. We observed a significant increase in the basal OCR in WT but not in mir1Δ (Figure 5A). The alternate scenario is after glycolytic inhibition. We assessed the role of mitochondrial Pi in this context, by inhibiting glycolysis using 2-deoxyglucose (2DG). WT, but not mir1Δ, increased their OCR (respiration) upon a 1-hour treatment with 2DG (Figure 5B). Consistent with this, mitotracker fluorescence increased with an increase in 2DG in WT, but not in mir1Δ (Figure S5A). We further asked if the mitochondrial Pi transporter itself glucose repressed, and therefore assessed Mir1 amounts in high and low glucose. Mir1 levels are higher upon a shift to low glucose, and in cells grown in 2% ethanol, suggesting that Mir1 is glucose repressed (Figure 5C, S5B). These data suggest that mitochondrial Pi transport is necessary for increasing mitochondrial activity after glucose derepression.

+ + +Mitochondrial Pi availability constrains mitochondrial activity under high glucose +

A) Relationship of mitochondrial Pi transport and respiration after glucose removal. WT and mir1Δ cells were cultured in high (2%) glucose and shifted to low (0.1%) glucose for 1 hour. The normalized basal OCR, from two independent experiments (n=2) are shown. Data represent mean ± SD.

B) Requirement of mitochondrial Pi transport for switch to respiration upon glycolytic inhibition by 2DG. WT and mir1Δ cells were cultured in high glucose and treated with or without 0.25% 2DG for 1 hour. Basal OCR was measured from two independent experiments (n=2). Data represent mean ± SD. Also see figure S5A.

C) Glucose-dependent regulation of Mir1. Cells (with Mir1-HA) were grown in high glucose and shifted to low glucose (0.1% glucose) for 1 hour, and Mir1 levels compared. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S5B.

D) Increasing Pi concentrations and mitochondrial activity in isolated mitochondria. Mitochondria were isolated from WT cells grown in high glucose, incubated with 1 mM pyruvate, 1 mM malate, 0.5 mM ADP and 0-50 mM KH2PO4. The mitochondrial activity was estimated by mitotracker fluorescence intensity, and intensities relative to the sample with 0 mM KH2PO4 is shown. Data represent mean ± SD from three biological replicates (n=3).

E) Effect of overexpressing Mir1 on Cox2 protein. WT (containing empty vector) and Mir1 overexpressing (Mir1OE) cells were grown in high glucose and Cox2 levels were estimated. A representative blot (out of three biological replicates, n=3) and their quantifications are shown. Data represent mean ± SD. Also see figure S5F.

F) Effect of overexpressing Mir1 on basal OCR. The basal OCR in WT (containing empty vector) and Mir1OE in high glucose was measured from three independent experiments (n=3). Data represent mean ± SD.

G) Mitochondrial Pi transport requirement for growth after 2DG treatment. Shown are serial dilution growth assays in high glucose in the presence and absence of 0.1% 2DG, using WT and mir1Δ cells. The results after 40hrs incubation/30°C are shown.

H) A model illustrating how mitochondrial Pi availability controls mitochondrial activity. In high glucose, the decreased Pi due to high Pi consumption in glycolysis, along with the glucose-mediated repression of mitochondrial Pi transporters, decreases mitochondrial Pi availability. This reduces mitochondrial activity. In low glucose, increased mitochondrial Pi transporters and lower glycolytic flux increases mitochondrial Pi, leading to enhanced mitochondrial activity. In ubp3Δ cells in high glucose, high trehalose synthesis and lower glycolytic flux results in an increase in Pi. This increases mitochondrial Pi availability and thereby the mitochondrial activity.

Data information: *p<0.05, **p<0.01, ***p<0.001.

+ + +

We next asked whether just adding external Pi was sufficient to increase mitochondrial activity, when cells are in high glucose. In medium supplemented with excess Pi, the internal Pi increases as seen earlier (Figure 3C). Therefore, a simplistic assumption would be that the addition of external Pi to cells in high glucose would also increase mitochondrial Pi. However, an alternate possibility presents itself whererin since glycolytic flux is already high in glucose, supplementing Pi will continue to fuel glycolysis. Indeed, this was originally obseved by Harden and Young in 1908, where adding Pi increased fermentation (44). In such a scenario, mitochondrial Pi will not increase. We estimated mitochondrial Pi in this condition where excess Pi was externally supplemented. Notably, cells grown in high Pi had decreased mitochondrial Pi (Figure S5C). Furthermore, directly adding Pi to cells growing in high glucose decreased basal OCR (Figure S5D), consistent with decreased mitochondrial Pi. These data indicate that in high glucose, simply supplementing Pi will not increase Pi access to the mitochondria. We therefore now asked, if we inhibit glycolytic flux and then supplement Pi, what would happen to mitochondrial activity. For this, we treated cells with 2DG, and subsequently added Pi and measured the OCR (Figure S5E). In this case, supplementing Pi increased the basal OCR (Figure S5E). Collectively, these data suggest that a combination of decreasing glycolysis and increasing Pi can together increase respiration. Next, in order to directly test mitochondrial activation based on external Pi availability, we isolated mitochondria, and estimated activity in vitro upon adding increasing Pi. Mitochondrial activity increased with increased Pi, with maximum activity observed with 25 mM Pi supplemented (Figure 5D). In a complementary experiment, we overexpressed the Mir1 transporter in wild-type cells, to increase Pi within mitochondria (Figure S5F). Mir1-OE cells have higher Cox2 levels and basal OCR (Figure 5E, 5F). Therefore, increasing Pi transport to mitochondria is sufficient to increase mitochondrial activity in high glucose.

+

Mitochondrial pyruvate transport is also required for mitochondrial respiration (45). We asked where Pi availability stands in a hierarchy of constraints for mitochondrial derepression, as compared to mitochondrial pyruvate transport. We measured the amounts of the Mpc3 subunit of the mitochondrial pyruvate carrier (MPC) complex (45, 46). Mpc3 protein increases in ubp3Δ (Figure S5G), correlating with the increased mitochondrial activity. Interestingly, in ubp3Δ grown in low Pi, Mpc3 further increased (Figure S5G), but as shown earlier this condition cannot increase OCR or mitochondrial activity (Figure 4F, S4F). Basal Mpc3 levels decrease in mir1Δ, but upon shifting to 0.1% glucose, Mpc3 increases in both WT and mir1Δ, with higher levels in mir1Δ (Figure S5H). Therefore, even where Mpc3 is high (ubp3Δ in low Pi, and mir1Δ in low glucose), mitochondrial activity remains low if Pi is restricted (Figure 4F, 4H). There was also no decrease in basal OCR in mpc3Δ in high glucose, and the basal OCR increased to the same level as of WT after shifting to 0.1% glucose (Figure S5I). Since Mpc3 changes with mitochondrial Pi availability (Figure S5G, S5H), we also measured Mpc3 in the Mir1OE. No further changes in Mpc3 were observed in Mir1OE (Figure S5J), indicating that increasing mitochondrial Pi alone need not increase Mpc3. Overall, although Mpc3 levels corelate with decreased glycolysis (ubp3Δ -Figure S5G, ubp3Δ in low Pi-Figure S5G, low glucose -Figure S5H), increased Mpc3 alone cannot increase mitochondrial activity and respiration in the absence of adequate mitochondrial Pi.

+

Finally, we asked how important mitochondrial Pi transport was for growth, under 2DG treatment. In the presence of 2DG, WT cells have only slightly decreased growth. In contrast, mir1Δ show a severe growth defect upon 2DG treatment (Figure 5G), revealing a synergetic effect of combining 2DG with inhibiting mitochondrial Pi transport. Therefore, the combined inhibition of glycolysis and mitochondrial Pi transport restricts the growth of glycolytic cells.

+

Collectively, mitochondrial Pi availability constrains glucose-mediated mitochondrial repression. Increasing available pools of Pi to enter the mitochondria is sufficient to induce mitochondrial activity.

+ + + +Discussion +

In this study, we highlight a role for Pi budgeting between cytosolic glycolysis and mitochondrial processes (which compete for Pi) in constraining mitochondrial repression (Figure 5H). Ubp3 controls this process (Figure 1), by maintaining the amounts of the glycolytic enzymes Pfk1 and GAPDH (Tdh2 and Tdh3) and thereby allows high glycolytic flux. At high fermentation rates, glycolytic enzymes levels at maximal activity maintain high glycolytic rates (effectively following zero-order kinetics), and therefore, changes in the enzyme levels will have a direct, proportionate effect on flux (47). The loss of Ubp3 decreases glycolytic flux, resulting in a systems-level, mass-action based rewiring of glucose metabolism where more G6P is routed towards trehalose synthesis and PPP (Figure 2, Figure 5H). Indeed, inhibiting just phosphofructokinase can reroute glucose flux from glycolysis to PPP (48–50), and our study now permits contextualized interpretations of these results. Such reallocations of glucose flux will collectively increase overall Pi, coming from the combined effect of increased Pi release from trehalose synthesis, and decreased Pi consumption via reduced GAPDH. This altered intracellular Pi economy increases Pi pools available to mitochondria, and increasing mitochondrial Pi is necessary and sufficient to increase respiration in high glucose (Figure 4, Figure 5). Mitochondrial Pi transport maintains mitochondrial activity in high glucose, and increases mitochondrial activity in low glucose (Figure 4, Figure 5). Finally, the mitochondrial Pi transporter Mir1 itself decreases in high glucose (Figure 5). Therefore, this glucose-dependent repression of Mir1 also restricts mitochondrial Pi availability (and thereby activity) in high glucose.

+

Traditionally, loss-of-function mutants of metabolic enzymes are used to understand metabolic state regulation. This approach negates nuanced investigations, since metabolic enzymes are often essential for viability. Furthermore, metabolic pathways have multiple, contextually regulated nodes, through which cells maintain their metabolic state. Therefore, alternate approaches to identify global regulators of metabolic states (as opposed to single enzymes), might uncover ways via which multiple nodes are simultaneously tuned, and can reveal unanticipated systems-level principles of metabolic state rewiring. In this study of glucose-mediated mitochondrial repression, the loss of the DUB Ubp3 decreases glycolytic flux by reducing the enzymes at two critical nodes in the pathway - Pfk1 and GAPDH. Unlike loss of function mutants, a reduction in amounts will only rewire metabolic flux. By ‘hitting’ multiple steps in glycolysis simultaneously, ubp3Δ have decreased Pi consumption, as well as increased Pi release. Such a cumulative phenomenon reveals more than inhibiting only GAPDH, where increased Pi comes only from reduced Pi consumption, and not from increased trehalose biosynthesis. Our serendipitous identification of a regulator which regulates multiple steps in glucose metabolism to change the metabolic environment, now suggests a general basis of mitochondrial regulation that would have otherwise remained hidden. Separately, finding the substrates of Ubp3 and whether Ubp3 directly regulates glycolytic enzymes are exciting future research questions requiring concurrent innovations in accessible chemical-biological approaches to study DUBs.

+

Because phosphates are ubiquitous, it is challenging to identify hierarchies of Pi-dependent processes in metabolic state regulation (33). Phosphate transfer reactions are the foundation of metabolism, driving multiple, thermodynamically unfavorable reactions (51, 52). Contextually, the laws of mass action predict that the relative rates of these reactions will regulate overall Pi balance, and contrarily the Pi allocation to Pi-dependent reactions will determine reaction rates (33, 35). Additionally, cells might control Pi allocations for different reactions via compartmentalizing Pi in organelles, to spatially restrict Pi availability (53–55). Our data collectively suggest a paradigm where the combination of factors regulates mitochondrial Pi and thereby activity. In glycolytic yeast cells growing in high glucose, mitochondrial Pi availability becomes restricted due to higher utilization of Pi in glycolysis compared to mitochondria. Consistent with this, a rapid decrease in Pi upon glucose addition has been observed (18, 19, 37). Interesting, in vitro studies with isolated mitochondria from tumor cells also find that decreasing Pi levels decreases respiration (19), which would be consistent with this scenario. Further, supplementing Pi correlates with decreased mitochondrial repression in tumors (18, 56). By increasing Pi through a systems-level rewiring of glucose metabolism (such as in ubp3Δ cells), cells can collectively increase mitochondrial access to Pi. This Pi budgeting determines mitochondrial activity. Supplementing Pi under conditions of low glycolysis (where mitochondrial Pi transport is enhanced), as well as directly supplementing Pi to isolated mitochondria, increases respiration (Figure 5, S5). Therefore, in order to deprepress mitochondria, a combination of increased Pi along with decreased glycolysis is required. An additional systems-level phenomenon that might regulate Pi transport to the mitochondria is the decrease in cytosolic pH upon decreased glycolysis (57, 58). Since mitochondrial Pi transport itself dependeds on the proton gradient, a low cytosolic pH would favour mitochondrial Pi transport (59). Therefore, decreasing glycolysis (2DG treatment, deletion of Ubp3, and decreased GAPDH activity), and increasing cytosolic Pi might synergestically increase mitochondria Pi transport, and increase respiration. Alternately, increasing mitochondrial Pi transporter amounts can achieve the same result, as seen by overexpressing Mir1 (Figure 5). A similar observation has been reported in Arabidopsis, reiterating an evolutionarily conserved role for mitochondrial Pi in controlling respiration (60). Relatedly, glycolytic inhibition can suppress cell proliferation in Warburg-positive tumors (61, 62), or inflammatory responses (63). However, these cells survive by switching to mitochondrial respiration (64, 65), requiring alternate approaches to prevent their proliferation (66). Inhibiting mitochondrial Pi transport in combination with glycolytic inhibition could restrict the proliferation of Warburg/Crabtree positive cells.

+

Since its discovery in the 1920s, the phenomenon of accelerated glycolysis with concurrent mitochondrial repression has been intensely researched. Yet, the biochemical constraints for glucose-mediated mitochondrial repression remains unresolved. One hypothesis suggests that the availability of glycolytic intermediates might determine the extent of mitochondrial repression. F1,6BP inhibits complex III and IV of the electron transport chain (ETC) in Crabtree positive yeast (14–17). Similarly, the ratio between G6P and F1,6BP regulates the extent of mitochondrial repression (16, 17). Although G6P/F6P accumulates in ubp3Δ (Figure 2C), this is not the case in tdh2Δtdh3Δ (GAPDH mutant) (Figure S3D), suggesting that G6P/F6P accumulation in itself is not the criterion to increase mitochondrial activity. Separately, the competition for common metabolites/co-factors between glycolysis and respiration (such as ADP, Pi or pyruvate) could drive this phenomenon (14, 18). Here we observe that Mpc3 mediated mitochondrial pyruvate transport alone cannot increase respiration. An additional consideration is the possible contribution of changes in ADP in regulating mitochondrial activity, where the use of ADP in glycolysis might limit mitochondrial ADP. Therefore, when Pi changes as a consequence of glycolysis, it coud be imagined that a change in ADP balance can coincidentally occur. However, prior studies show that even though cytosolic ADP decreases in the presence of glucose, this does not limit mitochondrial ADP uptake, or decrease respiration, due to the very high affinity of the mitochondrial ADP transporter (14, 19). These collectively reiterate the importance of Pi access and transport to mitochondria in constraining mitochondrial respiration. Indeed, this interpretation can also contextually explain observations from other model systems where mitochondrial Pi transport seems to regulate respiration (67, 68).

+

We parsimoniously suggest that Pi access to the mitochondria as a key constraint for mitochondrial repression under high glucose. In a hypothetical scenario, a single-step event in evolution, reducing mitochondrial Pi transporter amounts, and/or increasing glycolytic flux (to deplete cytosolic Pi), will result in whole-scale metabolic rewiring to repress mitochondria. More elaborate regulatory events can easily be imagined as subsequent adaptations to enforce mitochondrial repression. Given the central role played by Pi, something as fundamental as access to Pi will constrain mitochondrial repression. Concurrently, the rapid incorporation of Pi into faster glycolysis can give cells a competitive advantage, while also sequestering Pi in the form of usable ATP. Over the course of evolution, this could conceivably drive other regulatory mechanisms to enforce mitochondrial repression, leading to the currently observed complex regulatory networks and signaling programs observed in the Crabtree effect, and other examples of glucose-dependent mitochondrial repression.

+ + +Materials and methods + +Statistics and graphing +

Unless otherwise indicated, statistical significance for all indicated experiments were calculated using unpaired Student’s t-tests (GraphPad prism 9.0.1). Graphs were plotted using GraphPad prism 9.0.1

+ + +Yeast strains, media, and growth conditions +

A prototrophic CEN.PK strain of Saccharomyces cerevisiae (WT) (69) was used. Strains are listed in appendix table S1. Gene deletions, chromosomal C-terminal tagged strains were generated by PCR mediated gene deletion/tagging (70). Media compositions, growth conditions and CRISPR-Cas9 based mutagenesis are described in extended methods (SI appendix).

+ + +Mitotracker fluorescence +

Mitotracker fluorescence was measured using Thermo VarioscanTM LUX multimode plate reader (579/599 excitation/emission). Detailed protocol is described in extended methods. Mitotracker fluorescence were normalized using OD600 of each sample and relative fluorescence intensity calculated.

+ + +Protein extraction and Western blotting +

Total protein was precipitated, extracted using trichloroacetic acid (TCA) as described earlier (71). Blots were quantified using ImageJ software. Detailed protocol is described in extended methods.

+ + +Basal OCR measurement +

The basal OCR was measured using Agilent Seahorse XFe24 analyzer. Basal OCR readings were normalized for cell number (using OD600 of samples) in each well. The detailed methods are described in extended methods.

+ + +RNA extraction and RT-qPCR +

The RNA extraction was done using the hot phenol extraction method as described in (71). The isolated RNA was DNase treated, and used for cDNA synthesis. Superscript III reverse transcriptase enzyme (Invitrogen) was used for cDNA synthesis and RT-qPCR was performed using KAPA SYBR FAST qRT PCR kit (KK4602, KAPA Biosystems). Taf10 was used as a control for normalisation and the fold change in mRNA levels were calculated by 2^-ddct method.

+ + +ATP, ethanol and Pi measurements +

ATP levels were measured by ATP estimation kit (Thermo Fisher A22066). Ethanol concentration in the medium was estimated using potassium dichromate based assay described in (72) with modifications. Pi was estimated using a malachite green phosphate assay kit (Cayman chemicals, 10009325). Detailed sample collection and assay protocols are described in extended methods.

+ + +Metabolite extraction and analysis by LC-MS/MS +

The steady state levels and relative 13C label incorporation into metabolites were estimated by quantitative LC-MS/MS methods as described in (73). Detailed methodology is extensively described in extended methods. Peak area measurements are listed in supplement file 1.

+ + +Mitochondrial isolation +

Mitochondria was isolated by immunoprecipitation as described in (74, 75) with modifications. The detailed protocol is described in extended methods (SI appendix). The eluted mitochondria were used in malachite green assay for Pi estimation, boiled with SDS-glycerol buffer for western blots or incubated with mitotracker CMXROS with mitochondria-activation buffer for mitotracker assays.

+ + + + + +Data availability +

The LC-MS/MS peak area for all the metabolites can be found in supplement file 1. This study includes no data deposited in external repositories.

+ + +Acknowledgements +

We thank all SL lab members, Benjamin Tu, Anand Bachhawat and Vijay Jayaraman for comments and suggestions. We acknowledge the NCBS/inStem/CCAMP mass spectrometry facility for LC-MS/MS support. We thank Aayushee Khanna for help with experiments. V.V acknowledges funding support from DST-INSPIRE fellowship (IF170236) from the Department of Science and Technology (DST), Govt. of India, S.N and SA acknowledges intramural funding from the DBT-inStem PhD program. S.L acknowledges support from the DST SERB CRG grant CRG/2019/004772, a DBT-Wellcome India Alliance Senior Fellowship (IA/S/21/2/505922), and institutional support from inStem.

+ + +References +J. Zhu, C. B. Thompson, Metabolic regulation of cell growth and proliferation. Nat. Rev. Mol. Cell Biol. 20, 436–450 (2019). +L. Cai, B. P. Tu, Driving the Cell Cycle Through Metabolism. Annu. Rev. Cell Dev. Biol. 28, 59–87 (2011). +D. L. (David L. Nelson, D. L. (David L. Nelson, A. L. Lehninger, M. M. Cox, Lehninger principles of biochemistry (2008) (March 10, 2022). +O. Warburg, The metabolism of carcinoma cells 1. J. Cancer Res. 9, 148–163 (1925). +M. G. Vander Heiden, L. C. Cantley, C. B. Thompson, Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation. Science (80-.). 324, 1029–1033 (2009). +H. G. Crabtree, Observations on the carbohydrate metabolism of tumours. Biochem. J. 23, 536–545 (1929). +R. H. De Deken, The Crabtree effect: a regulatory system in yeast. J. Gen. Microbiol. 44, 149–156 (1966). +L. Xia, et al., The cancer metabolic reprogramming and immune response. Mol. Cancer 20, 1–21 (2021). +M. D. Kornberg, The immunologic Warburg effect: Evidence and therapeutic opportunities in autoimmunity. Wiley Interdiscip. Rev. Syst. Biol. Med. 12, 1–17 (2020). +N. Pacini, F. Borziani, Cancer stem cell theory and the warburg effect, two sides of the same Coin? Int. J. Mol. Sci. 15, 8893–8930 (2014). +A. M. Abdel-Haleem, et al., The emerging facets of non-cancerous Warburg effect. Front. Endocrinol. (Lausanne). 8, 1–7 (2017). +E. Tsogtbaatar, C. Landin, K. Minter-Dykhouse, C. D. L. Folmes, Energy Metabolism Regulates Stem Cell Pluripotency. Front. Cell Dev. Biol. 8, 1–16 (2020). +A. Cornish-Bowden, Biochemical evolution : the pursuit of perfection. Garl. Sci. (2016). +R. Diaz-Ruiz, M. Rigoulet, A. Devin, The Warburg and Crabtree effects: On the origin of cancer cell energy metabolism and of yeast glucose repression. Biochim. Biophys. Acta - Bioenerg. 1807, 568–576 (2011). +N. Hammad, M. Rosas-Lemus, S. Uribe-Carvajal, M. Rigoulet, A. Devin, The Crabtree and Warburg effects: Do metabolite-induced regulations participate in their induction? Biochim. Biophys. Acta - Bioenerg. 1857, 1139–1146 (2016). +R. Díaz-Ruiz, et al., Mitochondrial Oxidative Phosphorylation Is Regulated by Fructose 1,6-Bisphosphate: A POSSIBLE ROLE IN CRABTREE EFFECT INDUCTION? *. J. Biol. Chem. 283, 26948–26955 (2008). +M. R. Lemus, et al., The role of glycolysis-derived hexose phosphates in the induction of the Crabtree effect. J. Biol. Chem. 293, 12843–12854 (2018). +D. H. Koobs, Phosphate mediation of the Crabtree and Pasteur effects. Science (80-.). 178, 127–133 (1972). +S. Rodríguez-Enríquez, O. Juárez, J. S. Rodríguez-Zavala, R. Moreno-Sánchez, Multisite control of the Crabtree effect in ascites hepatoma cells. Eur. J. Biochem. 268, 2512–2519 (2001). +F. Tripodi, R. Nicastro, V. Reghellin, P. Coccetti, Post-translational modifications on yeast carbon metabolism: Regulatory mechanisms beyond transcriptional control. Biochim. Biophys. Acta - Gen. Subj. 1850, 620–627 (2015). +T. Hitosugi, J. Chen, Post-translational modifications and the Warburg effect. Oncogene 33, 4279–4285 (2014). +J. R. Broach, Nutritional control of growth and development in yeast. Genetics 192, 73–105 (2012). +A. Hershko, A. Ciechanover, THE UBIQUITIN SYSTEM. Annu. Rev. Biochem. 67, 425–79 (1998). +D. Komander, M. Rape, The Ubiquitin Code. Annu. Rev. Biochem. 81, 203–229 (2012). +C. M. Pickart, M. J. Eddins, Ubiquitin: Structures, functions, mechanisms. Biochim. Biophys. Acta - Mol. Cell Res. 1695, 55–72 (2004). +E. Postma, C. Verduyn, W. A. Scheffers, J. P. Van Dijken, Enzymic analysis of the crabtree effect in glucose-limited chemostat cultures of Saccharomyces cerevisiae. Appl. Environ. Microbiol. 55, 468–77 (1989). +B. Ossareh-Nazari, et al., Cdc48 and Ufd3, new partners of the ubiquitin protease Ubp3, are required for ribophagy. EMBO Rep. 11, 548–554 (2010). +M. Isasa, et al., Multiplexed, Proteome-Wide Protein Expression Profiling: Yeast Deubiquitylating Enzyme Knockout Strains. J. Proteome Res. 14, 5306–5317 (2015). +R. Nostramo, S. N. Varia, B. Zhang, M. M. Emerson, P. K. Herman, The Catalytic Activity of the Ubp3 Deubiquitinating Protease Is Required for Efficient Stress Granule Assembly in Saccharomyces cerevisiae. Mol. Cell. Biol. 36, 173 (2016). +F. Fontanesi, I. C. Soto, D. Horn, A. Barrientos, Assembly of mitochondrial cytochrome c-oxidase, a complicated and highly regulated cellular process. Am. J. Physiol. - Cell Physiol. 291, 1129–1147 (2006). +A. A. Shestov, et al., Quantitative determinants of aerobic glycolysis identify flux through the enzyme GAPDH as a limiting step. Elife 3, 1–18 (2014). +R. Gupta, et al., A tRNA modification balances carbon and nitrogen metabolism by regulating phosphate homeostasis. Elife 8, 1–33 (2019). +R. Gupta, S. Laxman, Cycles, sources, and sinks: Conceptualizing how phosphate balance modulates carbon flux using yeast metabolic networks. Elife 10, 1–14 (2021). +P. W. Mason, D. P. Carbone, R. A. Cushman, A. S. Waggoner, The importance of inorganic phosphate in regulation of energy metabolism of Streptococcus lactis. J. Biol. Chem. 256, 1861–1866 (1981). +J. H. Van Heerden, et al., Lost in transition: Start-up of glycolysis yields subpopulations of nongrowing cells. Science (80-.). 343 (2014). +P. C. Brazy, L. J. Mandel, V. W. Dennis, S. R. Gullans, Metabolic Requirement for Inorganic Phosphate by the Rabbit Proximal Tubule: EVIDENCE FOR A CRABTREE EFFECT Metabolic Requirement for Inorganic Phosphate by the Rabbit Proximal Tubule EVIDENCE FOR A CRABTREE EFFECT. J Clin Invest 70, 53 (1982). +S. Hohmann, W. Bell, M. J. Neves, D. Valckx, J. M. Thevelein, Evidence for trehalose-6-phosphate-dependent and-independent mechanisms in the control of sugar influx into yeast glycolysis. Mol. Microbiol. 20, 981–991 (1996). +T. Hunter, Why nature chose phosphate to modify proteins. Philos. Trans. R. Soc. B Biol. Sci. 367, 2513–2516 (2012). +P. O. Ljungdahl, B. Daignan-Fornier, Regulation of amino acid, nucleotide, and phosphate metabolism in Saccharomyces cerevisiae. Genetics 190, 885–929 (2012). +S. Austin, A. Mayer, Phosphate Homeostasis − A Vital Metabolic Equilibrium Maintained Through the INPHORS Signaling Pathway. Front. Microbiol. 11, 1367 (2020). +J. M. Mouillon, B. L. Persson, New aspects on phosphate sensing and signalling in Saccharomyces cerevisiae. FEMS Yeast Res. 6, 171–176 (2006). +H. Murakami, G. Blobel, D. Pain, Isolation and characterization of the gene for a yeast mitochondrial import receptor. Nat. 1990 3476292 347, 488–491 (1990). +V. Zara, et al., Yeast mitochondria lacking the phosphate carrier/p32 are blocked in phosphate transport but can import preproteins after regeneration of a membrane potential. Mol. Cell. Biol. 16, 6524–6531 (1996). +A. Harden, W. J. Young, The alcoholic Ferment of Yeast-juice.Part III.-The Function of Phosphates in the Fermentation o f Glucose hy Yeast-juice. Proc. R. Soc. London. Ser. B, 80, 299–311 (1908). +A. Timoń-Gómez, M. Proft, A. Pascual-Ahuir, Differential Regulation of Mitochondrial Pyruvate Carrier Genes Modulates Respiratory Capacity and Stress Tolerance in Yeast. PLoS One 8, e79405 (2013). +T. Bender, G. Pena, J.-C. Martinou, Regulation of mitochondrial pyruvate uptake by alternative pyruvate carrier complexes. EMBO J. 34, 911–924 (2015). +P. Grigaitis, B. Teusink, An excess of glycolytic enzymes under glucose-limited conditions may enable Saccharomyces cerevisiae to adapt to nutrient availability. FEBS Lett. (2022) 10.1002/1873-3468.14484 (November 28, 2022). +W. Yi, et al., PFK1 Glycosylation Is a Key Regulator of Cancer Cell Growth and Central Metabolic Pathways. Science 337, 975 (2012). +H. Miyazawa, et al., Rewiring of embryonic glucose metabolism via suppression of PFK-1 and aldolase during mouse chorioallantoic branching. Dev. 144, 63–73 (2017). +W. D. Hollinshead, et al., Examining Escherichia coli glycolytic pathways, catabolite repression, and metabolite channeling using Δpfk mutants. Biotechnol. Biofuels 9, 1–13 (2016). +F. H. Westheimer, Why Nature Chose Phosphates. Science (80-.). 235, 1173–1178 (1987). +S. C. L. Kamerlin, P. K. Sharma, R. B. Prasad, A. Warshel, Why nature really chose phosphate. Q. Rev. Biophys. 46, 1–132 (2013). +J. W. Booth, G. Guidotti, Phosphate Transport in Yeast Vacuoles *. J. Biol. Chem. 272, 20408–20413 (1997). +T. Vila, S. Frases, F. M. Gomesid, Lessons from protozoans: Phosphate sensing and polyphosphate storage in fungi. PLOS Pathog. 18, e1010298 (2022). +M. E. Solesio, et al., Depletion of mitochondrial inorganic polyphosphate (polyP) in mammalian cells causes metabolic shift from oxidative phosphorylation to glycolysis. Biochem. J. 478, 1631–1646 (2021). +M. Brin, R. W. McKee, The Inhibition of Respiration by Glucose, Fructose, and Mannose in the Ehrlich Mouse Ascites Tumor. Cancer Res. 16, 364–368 (1956). +R. Orij, S. Brul, G. J. Smits, Intracellular pH is a tightly controlled signal in yeast. Biochim. Biophys. Acta - Gen. Subj. 1810, 933–944 (2011). +R. Dechant, et al., Cytosolic pH is a second messenger for glucose and regulates the PKA pathway through V-ATPase. EMBO J. 29, 2515–2526 (2010). +P. Hamel, et al., Redundancy in the function of mitochondrial phosphate transport in Saccharomyces cerevisiae and Arabidopsis thaliana. Mol. Microbiol. 51, 307–317 (2004). +F. Jia, et al., Overexpression of Mitochondrial Phosphate Transporter 3 Severely Hampers Plant Development through Regulating Mitochondrial Function in Arabidopsis. PLoS One 10, e0129717 (2015). +H. Pelicano, D. S. Martin, R. Xu, P. Huang, Glycolysis inhibition for anticancer treatment. 4633–4646 (2006). +S. O’Neill, R. K. Porter, N. McNamee, V. G. Martinez, L. O’Driscoll, 2-Deoxy-D-Glucose inhibits aggressive triple-negative breast cancer cells by targeting glycolysis and the cancer stem cell phenotype. Sci. Rep. 9, 1–11 (2019). +G. Soto-Heredero, M. M. Gómez de las Heras, E. Gabandé-Rodríguez, J. Oller, M. Mittelbrunn, Glycolysis – a key player in the inflammatory response. FEBS J. 287, 3350–3369 (2020). +R. Shiratori, et al., Glycolytic suppression dramatically changes the intracellular metabolic profile of multiple cancer cell lines in a mitochondrial metabolism-dependent manner. Sci. Rep. 9, 1–15 (2019). +C. L. Lu, et al., Tumor Cells Switch to Mitochondrial Oxidative Phosphorylation under Radiation via mTOR-Mediated Hexokinase II Inhibition - A Warburg-Reversing Effect. PLoS One 10, e0121046 (2015). +G. Cheng, et al., Mitochondria-targeted drugs synergize with 2-deoxyglucose to trigger breast cancer cell death. Cancer Res. 72, 2634–2644 (2012). +M. Scheibye-Knudsen, B. Quistorff, Regulation of mitochondrial respiration by inorganic phosphate; comparing permeabilized muscle fibers and isolated mitochondria prepared from type-1 and type-2 rat skeletal muscle. Eur. J. Appl. Physiol. 105, 279–287 (2009). +E. L. Seifert, E. Ligeti, J. A. Mayr, N. Sondheimer, G. Hajnóczky, The mitochondrial phosphate carrier: Role in oxidative metabolism, calcium handling and mitochondrial disease. Biochem. Biophys. Res. Commun. 464, 369–375 (2015). +J. B. J.P. van Dijken, et al., An interlaboratory comparison of physiological and genetic properties of four Saccharomyces cerevisiae strains. Enzyme Microb. Technol. 26, 706–714 (2000). +M. S. Longtine, et al., Additional modules for versatile and economical PCR-based gene deletion and modification in Saccharomyces cerevisiae. Yeast 14, 953–961 (1998). +V. Vengayil, Z. Rashida, S. Laxman, The E3 ubiquitin ligase Pib1 regulates effective gluconeogenic shutdown upon glucose availability. J. Biol. Chem. 294, 17209–17223 (2019). +M. Sriariyanun, P. Mutrakulcharoen, S. Tepaamorndech, K. Cheenkachorn, K. Rattanaporn, A Rapid Spectrophotometric Method for Quantitative Determination of Ethanol in Fermentation Products. Orient. J. Chem. 35, 744–750 (2019). +A. Walvekar, Z. Rashida, H. Maddali, S. Laxman, A versatile LC-MS / MS approach for comprehensive, quantitative analysis of central metabolic pathways [version 1; peer review : 2 approved]. 1–15 (2018). +P. C. Liao, I. R. Boldogh, S. E. Siegmund, Z. Freyberg, L. A. Pon, Isolation of mitochondria from Saccharomyces cerevisiae using magnetic bead affinity purification. PLoS One 13, 1–15 (2018). +W. W. Chen, E. Freinkman, D. M. Sabatini, Rapid immunopurification of mitochondria for metabolite profiling and absolute quantification of matrix metabolites. Nat. Protoc. 12, 2215 (2017). + + +
diff --git a/test/fixtures/2023.02.04.527146/2023.02.04.527146.xml b/test/fixtures/2023.02.04.527146/2023.02.04.527146.xml new file mode 100644 index 00000000..fca30e28 --- /dev/null +++ b/test/fixtures/2023.02.04.527146/2023.02.04.527146.xml @@ -0,0 +1,339 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.02.04.527146 +1.3 + + +Regular Article + + +New Results + + +Neuroscience + + + +Effort cost of harvest affects decisions and movement vigor of marmosets during foraging + + +http://orcid.org/0000-0002-7006-9877HagePaul1 +http://orcid.org/0000-0002-1832-3743JangIn Kyu1 +http://orcid.org/0000-0002-9392-4178LooiVivian1 +http://orcid.org/0000-0002-7741-8689FakharianMohammad Amin1 +http://orcid.org/0000-0001-6092-7228OrozcoSimon P.1 +http://orcid.org/0000-0003-4036-2877PiJay S.1 +http://orcid.org/0000-0001-6732-1296Sedaghat-NejadEhsan1 +http://orcid.org/0000-0002-7686-2569ShadmehrReza1 +Laboratory for Computational Motor Control, Dept. of Biomedical Engineering Johns Hopkins School of Medicine, Baltimore, Maryland USA + + +Correspondence: Paul Hage or Reza Shadmehr, Johns Hopkins School of Medicine, 410 Traylor Building, 720 Rutland Ave., Baltimore, MD 21205. Email: paulhage@jhu.edu or shadmehr@jhu.edu +

Author contributions: P.H., S.P.O., E.S., J.S.P. and R.S. conceived and performed experiments. P.H., I.K.J., V.L., M.A.F., E.S., and J.S.P. analyzed data, P.H. made figures and performed statistical analysis. R.S. and P.H. wrote the manuscript.

 +

Competing interests: None.

 +

Data availability: https://osf.io/54JS6

 + +2023 +2023.02.04.527146 + + +04 +2 +2023 + + +19 +7 +2023 + + +19 +7 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. + + + +Abstract

Our decisions are guided by how we perceive the value of an option, but this evaluation also affects how we move to acquire that option. Why should economic variables such as reward and effort alter the vigor of our movements? In theory, both the option that we choose and the vigor with which we move contribute to a measure of fitness in which the objective is to maximize rewards minus efforts, divided by time. To explore this idea, we engaged marmosets in a foraging task in which on each trial they decided whether to work by making saccades to visual targets, thus accumulating food, or to harvest by licking what they had earned. We varied the effort cost of harvest by moving the food tube with respect to the mouth. Theory predicted that the subjects should respond to the increased effort costs by choosing to work longer, stockpiling food before commencing harvest, but reduce their movement vigor to conserve energy. Indeed, in response to an increased effort cost of harvest, marmosets extended their work duration, but slowed their movements. These changes in decisions and movements coincided with changes in pupil size. As the effort cost of harvest declined, work duration decreased, the pupils dilated, and the vigor of licks and saccades increased. Thus, when acquisition of reward became effortful, the pupils constricted, the decisions exhibited delayed gratification, and the movements displayed reduced vigor.

+ +Significance statement +

Our results suggest that as the brainstem neuromodulatory circuits that control pupil size respond to effort costs, they alter computations in the brain regions that control decisions, encouraging work and delaying gratification, and the brain regions that control movements, reducing vigor and suppressing energy expenditure. This coordinated response suggests that decisions and actions are part of a single control policy that aims to maximize a variable relevant to fitness: the capture rate.

+ + + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Revisions based on feedback from eLife reviews.

 + + +

+https://osf.io/54JS6 +

 + + + + + +Introduction +

During foraging, animals work to locate a food cache and then spend effort harvesting what they have found. As they forage, their decisions appear to maximize a measure that is relevant to fitness: the sum of rewards acquired, minus efforts expended, divided by time, termed the capture rate (1–3). For example, a crow will spend effort extracting a clam from a sandy beach, but if the clam is small, it will abandon it because the additional time and effort required to extract the small reward --dropping it repeatedly from a height onto rocks -- can be better spent finding a bigger prize (4). In other words, if going to the bank entails waiting in a long line, one should go infrequently, but make each transaction a large amount.

+

Intriguingly, reward expectation not only affects decisions, it also affects movements: we not only prefer the less effortful option, we also move vigorously to obtain it (5, 6). This modulation of movement vigor can be justified if we consider that movements require expenditure of time and energy, which discount the value of the promised reward (7, 8). Thus, from a theoretical framework, it seems rational that the brain should have a mechanism to coordinate control of decisions with control of movements so that both contribute to maximizing a measure of fitness (9).

+

To study this coordination, we designed a task in which marmosets decided how long to work before they harvested their food. On a given trial, they made a sequence of saccades to visual targets and received an increment of food as their reward. However, the increment was small, and its harvest was effortful, requiring inserting their tongue inside a small tube. Theory predicted that in order to maximize the capture rate, harvest should commence only when there was sufficient reward accumulated to justify the effort required for its extraction. Indeed, the subjects chose to work in order to stockpile food, and only then initiated their harvest.

+

On some days the effort cost of harvest was low: the tube was placed close to the mouth. On other days the same amount of work, i.e., saccade trials, produced food that had a higher effort cost: the tube was located farther away. The theory made two interesting predictions: as the effort cost increased, the subjects should choose to work more trials, thus delaying their harvest so to stow more food, but reduce their movement vigor, thus saving energy. Indeed, when marmosets encountered an increased effort cost, they extended their work period, stockpiling food, but reduced their vigor, slowing their saccades during the work period, and slowing their licks during the harvest period.

+

What might be a neural basis for this coordinated response of the decision-making and the motor-control circuits? During the work and the harvest periods, momentary changes in pupil size closely tracked the changes in vigor: pupil dilation accompanied increases in vigor, while pupil constriction accompanied decreases in vigor. Remarkably, this was true regardless of whether the movement that was being performed was a saccade, or a lick. Moreover, in response to the increased effort cost, the pupils exhibited a global change, constricting during both the work and the harvest periods.

+

If we view the changes in pupil size as a proxy for activity in the brainstem noradrenergic circuits (10), our results suggest that as these circuits respond to effort costs (11), they alter computations in the brain regions that control decisions, delaying gratification and encouraging work, and the brain regions that control movements, promoting sloth and conserving energy.

+ + +Results +

We tracked the eyes and the tongue of head-fixed marmosets as they performed visually guided saccades in exchange for food (Fig. 1A). Each successful trial consisted of 3 visually guided saccades, at the end of which we delivered an increment of food (a slurry mixture of apple sauce and monkey chow). Because the reward amount was small (0.015-0.02 mL), the subjects rarely harvested following a single successful trial. Rather, they worked for a few trials, allowing the food to accumulate, then initiated their harvest by licking (Fig. 1B). The key variables were how many trials they chose to work before starting harvest, and how vigorously they moved their eyes and tongue during the work and the harvest periods.

+ + +

Elements of a foraging task. A. During the work period, marmosets made a sequence of saccades to visual targets. A trial consisted of 3 consecutive saccades, at the end of which the subject was rewarded by a small increment of food. We tracked the eyes, the tongue, and the food. B. An example of two consecutive work-harvest periods, showing reward-relevant saccades (eye velocity) and tongue endpoint displacement with respect to the mouth. C. Data for two sessions, one where the tube was placed close to the mouth (orange trace), and one where it was placed farther away (red trace). Two types of licks are shown: inner-tube licks, and outer-tube licks. Depending on food location, both types of licks can contact the food. Data on the right two panels show endpoint displacement and velocity of the tongue during inner-tube licks. Error bars are SEM. D. During the work period, the subjects attempted about 8 trials on average, succeeding in 4-5 trials before starting harvest, and then licked about 18 times to extract the food.

+ + +

Over the course of 2.5 years, we recorded 56 sessions in subject M (29 months) and 56 sessions in subject R (23 months). A typical work period lasted about 10 seconds, during which the subjects attempted ∼8 trials, and succeeded in 4-5 trials (Fig. 1D) (a successful trial was when all three saccades were within 1.25 of the center of each target). The work period ended when the subject decided to stop tracking the targets and instead initiated harvest, which lasted about 6 seconds, resulting in 16-18 licks. Subject M completed an average of 909.5 ± 61 successful trials per session (mean ± SEM), producing an average of 241 ± 13.9 work-harvest pairs, and subject R completed an average of 1431 ± 65 successful trials, producing an average of 263 ± 8.9 work-harvest pairs.

+

We delivered food via either the left or the right tube for 50-300 consecutive trials, and then switched tubes. We tracked the motion of the tongue using DeepLabCut (12), as shown for a typical session in Fig. 1B. The licks required precision because the tube was just large enough (4.4 mm diameter) to allow the tongue to penetrate. As a result, about 30% of the of the reward-seeking licks were successful and contacted food (30±1.6% for subject M, 28±2.5% for subject R), as shown in Supporting Information Video 1. Example of licks that failed to contact food are shown in Supporting Information Video 2, Video 3, and Video 4.

+ +Theory and predictions +

During the decision-making part of the task, the brain explicitly determined how long to work before initiating harvest. During the work and the harvest periods, the brain implicitly controlled the vigor of movements. We imagined that these two forms of behavior were not independent, but rather coordinated via a control policy that maximized a single utility: the sum of rewards acquired, minus efforts expended, divided by time, termed the capture rate. We chose this formulation because it presents a normative approach that ecologists have used to understand the decisions that animals make regarding how far to travel for food, what mode of travel to use, and how long to stay before moving on to another patch (4, 13, 14).

+

During a work period, our subjects decided to complete a number of saccade trials ns, a fraction βs of which were successful, earning food increment α, but expended effort cs that consumed time Ts for each trial. They then stopped working and initiated harvest, producing a number of licks nl, a fraction βl of which succeeded, thus expending effort cl and consuming time Tl for each lick. These actions produced the following capture rate: + + + +In the numerator of Eq. (1), the first term represents the fact that the food cache increased linearly with successful trials and was then consumed gradually with successful licks. The second term represents the effort expenditure of licking, and the third term represents the effort expenditure of working. Notably, the effort expenditure of work, grows faster than linearly as a function of trials. This nonlinearity is essential to reflect the idea that following a long work period, the capture rate must be more negative than following a short work period (i.e., more work trials produce a greater reduction in utility).

+

A control policy describes how long to work and harvest, and an optimal policy produces periods of working and harvesting, that maximize Eq. (1). A closed-form solution for the optimal policy can be obtained (Mathematica notebook simulations.nb) and Fig. 2A provides an example. As the work period concludes and the harvest period beings (nl = 0), the capture rate is negative. This reflects the fact that the subject has performed a few trials and stockpiled food, thus expended effort but has not been rewarded yet. The capture rate rises when licking commences. Critically, the peak capture rate is not an increasing function of the work period. Rather, there is an optimal work period ( red trace, Fig. 2A) associated with a given effort cost of licking cl. If we now move the tube away from the mouth, that is, increase the effort cost of licking cl, the peak of the capture rate shifts and the optimal work period changes: the proper response to an increased effort cost of licking is to work longer, stowing more food before commencing harvest.

+ + +

Theoretical results of an optimal control policy. A. Capture rate (Eq. 1) is plotted during the harvest period as a function of lick number following various number of work trials. When the effort cost of licking is low (left plot,), the optimal work period is (red trace). When the effort cost is higher (right plot,), it is best to work longer before initiating harvest. B. The metabolic cost of licking (Eq. 2) is minimized when a lick has a specific duration. Tube distance varied from 0.1 to 0.3. Optimal duration that minimizes lick cost grows linearly with tube distance. C. Optimal number of work trials and licks as a function food tube distance. As the effort cost of harvest increases, one should respond by working longer, delaying harvest. D. Optimal lick duration as a function of food tube distance. The lick duration that maximizes the capture rate is smaller than the one that minimizes the lick metabolic cost (Fig. 2B). That is, it is worthwhile moving vigorously to acquire reward. However, grows faster than linearly as a function of tube distance. Thus, as the tube moves farther, it is best to reduce lick vigor. Hunger, modeled as increased value of reward, should promote work and increase vigor, while effort cost of harvest (tube distance) should promote work but reduce vigor. Parameter values for all simulations:,,,, (low food value, less hunger), (high food value, hungry).

+ + +

Notably, the higher cost of licking inevitably reduces the maximum capture rate (Fig. 2A). This should impact movement vigor: animals tend to respond to a reduced capture rate by slowing their movements (9), which can be viewed as an effective way to save energy (8). To incorporate vigor into the capture rate, we tried to define the effort cost of a single lick cl in terms of its energetic cost, a relationship that is currently unknown. Fortunately, other movements provide a clue: the energetic cost of reaching (8, 15), as well as the energetic cost of walking (16, 17), are both concave upward functions of the movement’s duration. That is, from an energetic standpoint, there is a reach speed, and a walking speed, that minimizes the cost of each type of movement. We generalized these empirical observations to licking and assumed that the energetic cost of a single lick was a concave upward function of its duration: + + + +In Eq. (2), the lick is aimed at a tube located at distance d, and has a duration Tl. The parameter k describes the rate with which the cost grows as a function of duration. For example, the lick duration that minimizes the energetic cost is . Thus, for an energetically optimal lick, duration grows linearly with tube distance (Fig. 2B). However, our objective is not to minimize the cost of licking, but to maximize the capture rate. To do so, we insert Eq. (2) into Eq. (1) and find the optimal policy which now depends on the distance of the food tube to the mouth (Mathematica notebook simulations.nb).

+

The theory predicts that to maximize the capture rate (Eq. 1), the response to an increased effort cost of harvest (i.e., tube distance) should be as follows: should increase (Fig. 2C), should decrease (Fig. 2C), and should increase (Fig. 2D). Notably, the rate of increase in as a function of tube distance is faster than linear, while from an energetic point of view (Eq. 2), increase in distance should produce a linear increase in lick duration. Thus, as the harvest becomes more effortful, the subject should work longer to stockpile food, but move slower to save energy.

+

To test our theory further, we thought it useful to have a way to alter decisions in one direction (say work longer) but change movement vigor in the opposite direction (move faster). In theory, this is possible: if the subject is hungry (darker lines in Fig. 2C and 2D), i.e., the reward is more valuable, then they should again work longer before initiating harvest. Paradoxically, they should also move faster.

+

In summary, if decisions and actions are coordinated via a policy that aims to maximize the capture rate, then in response to an increased cost of harvest, one should work longer, but move with reduced vigor. In response to an increased reward value, as in hunger, one should also work longer, but now move with increased vigor.

+ + +Increased effort cost of harvest promoted work but reduced saccade vigor +

To vary the effort cost of harvest, we altered the tube distance to the mouth (but kept it constant during each session). Varying tube distance affected the decisions of the subjects: when the tube was placed farther, they chose to work longer before starting harvest (Fig. 3A, left subplot): they attempted more trials during each work period (ANOVA, subject M: F(2,7908)=41.5 p=5.2×10-25, subject R: F(2,10948)=88.2 p=7×10-50), and produced more successful trials per work period (ANOVA, subject M: F(2,7908)=63 p=2.8×10-24, subject R: F(2,10948)=163 p<10-50). This policy of delayed gratification was present throughout the recording session (Fig. 3A, middle plot). That is, when the harvest required more effort, the subjects worked longer to stockpile more food before initiating their harvest (Fig. 3A, right plot, effect of tube distance on food cached: subject M: F(2,9566)=176 p<10-50, subject R: F(2,8907)=204 p<10-50).

+ + +

As the effort cost of harvest increased, subjects chose to work more trials, but slowed their movements. A. Left: the number of trials attempted and succeeded per work period as a function of tube distance. Middle: successful trials per work period as a function of time during the recording session. Tube distance is with respect to a marker on the nose. Right: food available in the tube at the start of the harvest. B. Peak saccade velocity as a function of amplitude for reward-relevant and other saccades. C. Vigor of reward relevant saccades as a function of trial number during the work period. Saccade vigor was greater when the tube was closer. Pupil size is quantified during the same work periods. Accuracy is quantified as the magnitude of the saccade’s endpoint error vector (with respect to the target) and the variance of that error vector (determinant of the variance-covariance matrix), plotted as a function of the vigor of the saccade (bin size=0.05 vigor units). Error bars are SEM.

+ + +

During the work period the subjects made saccades to visual targets and accumulated their food. They also made saccades that were not toward visual targets and thus were not eligible for reward. For each animal we computed the relationship between peak saccade velocity and saccade amplitude across all sessions and then calculated the vigor of each saccade: defined as the ratio of the actual peak velocity with respect to the expected peak velocity for that amplitude (18, 19). For example, a saccade that exhibited a vigor of 1.10 had a peak velocity that was 10% greater than the average peak velocity of the saccades of that amplitude for that subject. As expected, the reward-relevant saccades, i.e., saccades made to visual targets (primary, corrective, and center saccades) were more vigorous than other saccades (Fig. 3B, 2-way ANOVA, effect of saccade type, subject M: F (1,391459)=7248 p<10-50, subject R: F (1,355839)=13641 p<10-50).

+

As a work period began, the reward-relevant saccades exhibited high vigor, but then trial-by-trial, this vigor declined, reaching a low vigor value just before the work period ended (Fig. 3C vigor). Remarkably, on days in which the tube was placed farther, saccade vigor was lower (RMANOVA, effect of tube distance, subject M: F(2,59033)=224 p<10-50, subject R: F(2,50103)=75.51 p=1.8×10-33). Thus, increasing the effort cost of extracting food during the harvest period reduced saccade vigor during the work period.

+

By definition, a more vigorous saccade had a greater peak velocity. This might imply that high vigor saccades should suffer from inaccuracy due to signal dependent noise (20). However, we observed the opposite tendency: as saccade vigor increased, both the magnitude and the variance of the endpoint error decreased (Fig. 3C, 2-way ANOVA, effect of vigor on error magnitude, subject M: F(8,59046)=480 p<10-50, subject R: F(8,50184)=252 p<10-50, effect of vigor on error variance, subject M: F(8,2673)=18200 p<10-50, subject R: F(8,2673)=4170 p<10-50). That is, reducing the effort costs of harvest not only promoted vigor, it also facilitated accuracy (21).

+

Cognitive signals such as effort and reward are associated with changes in pupil size (10), as well as transient activation of brainstem neuromodulatory circuits in locus coeruleus (11). We wondered if the changes in tube position altered the output of these neuromodulatory circuits, as inferred via pupil size. For each reward-relevant saccade, we measured the pupil size during a ±250 ms window centered on saccade onset, and then normalized this measure based on the distribution of pupil sizes that we had measured during the entire recording for that session in that subject, resulting in a z-score.

+

At the onset of each work period the pupils were dilated, but as the subjects performed more trials, the pupils constricted, exhibiting a trial-by-trial reduction that paralleled the changes in saccade vigor (Fig. 3C). Notably, the effort cost of harvest affected pupil size: during the work period the pupils were more dilated if the tube was placed closer to the mouth (Fig. 3C, RMANOVA, effect of tube distance, subject M: F(2,60502)=20 p=2×10-9, subject R: F(2,50431)=23.8 p=4.9×10-11). That is, when the effort cost of harvest was lower, the pupils dilated, and the saccades were invigorated.

+

In summary, when we increased the effort cost of harvest, both the movements and the decisions changed: the pupils constricted and the movements slowed, but they chose to work more trials before initiating harvest.

+ + +Increased effort cost of harvest reduced lick vigor +

The work period ended when the subject chose to stop tracking the target and initiated harvest via a licking bout. As in saccades, we defined lick vigor via the ratio of the actual peak velocity of the lick with respect to the expected velocity for that lick amplitude. As amplitude increased, lick peak velocity increased during both protraction and retraction (Fig. 4A). Some of the licks were reward seeking and directed toward the tube, while others were grooming licks, cleaning the tongue and the area around the mouth (Supporting Information Video 5). Reward-seeking licks were more vigorous than grooming licks (2-way ANOVA, effect of lick type, protraction, subject M: F(1,272233)=66 p=4.5×10-16, subject R: F(1,229052)=698 p<10-50), and retraction was more vigorous than protraction (reward seeking licks, retraction vs. protraction, subject M: t(241145)=532 p<10-50, subject R: t(213674)=665 p<10-50).

+ + +

As the effort cost of harvest increased, lick vigor declined, and the pupils constricted. A. Peak speed of reward-seeking and grooming licks during protraction and retraction as a function of lick amplitude. B. Vigor of reward seeking licks (protraction) and pupil size as a function of lick number during harvest at various tube distances. C. Lick vigor and pupil size as a function of time during the entire recording session. Line colors depict tube distance as in part B. D. Average lick vigor and pupil size during a harvest as a function of number of trials successfully completed in the previous work period. Lick vigor and pupil size were greater when more food had been stored. E. Following a successful lick (contact with food), the next lick was more vigorous, and pupils dilated. Following a failed lick, the next lick was slowed, and pupils were less dilated. F. We observed no consistent effect of lick vigor on lick accuracy across subjects or across tube distances. Error bars are SEM.

+ + +

As the harvest began, the first lick was very low vigor, but lick after lick, the movements gathered velocity, reaching peak vigor by the 3rd or the 4th lick (Fig. 4B). As the harvest continued, lick vigor gradually declined. Like saccades, licks had a lower vigor in sessions in which the tube was placed farther from the mouth (RMANOVA, effect of tube distance, subject M: F(2,59033)=222.5 p<10-50, subject R: F(2,133502)=224 p<10-50), and this pattern was present during the entire recording session (Fig. 4C, left subplot). Thus, an increased effort cost of harvest promoted sloth: reduced vigor of saccades during the work period, and reduced vigor of licks during the harvest period.

+

For each reward seeking lick, we measured pupil size during a ±250ms window centered on the moment of peak tongue displacement. During licking, the pupil size changed with a pattern that closely paralleled lick vigor: as the harvest began, pupil size was small, but it rapidly increased during the early licks, then gradually declined as the harvest continued (Fig. 4B, right subplot). Importantly, the pupils were more dilated in sessions in which the tube was closer to the mouth (Fig. 4C, right subplot, effect of tube distance, subject M: F(2,166742)=583 p<10-50, subject R: F(2,130493)=118 p<10-50). As a result, when the effort cost of reward increased, the pupils constricted, and the vigor of both saccades and licks decreased.

+

While the theory predicted that moving the tube farther would result in a longer work period and reduced movement vigor, it also predicted that the subjects would reduce their harvest duration (reduced licks, Fig. 2C). That is, it predicted that the subjects would work longer, stowing more food, but leave more of it behind. This last prediction did not agree with our data (see Discussion). For Subject R, the number of licks were approximately the same across the various tube distances, and for Subject M the number of licks increased with tube distance (Supplementary Fig. 1).

+

In summary, within a harvest period, lick vigor rapidly increased and then gradually declined. Simultaneous with the changes in vigor, the pupils rapidly dilated and then gradually constricted. In sessions where the tube was placed farther from the mouth, the licks had lower vigor, and the pupils were more constricted.

+ + +Expectation of greater reward increased lick vigor +

As the subject worked, they accumulated food, thus increasing the magnitude of the available reward. To check whether reward magnitude affected movement vigor, for each tube distance we computed the average lick vigor during the harvest as a function of the number of trials completed in the preceding work period. We found that when the work period had included many completed trials, then the movements in the ensuing harvest period were more vigorous (Fig. 4D, 2-way ANOVA, effect of trials, subject M: F(4, 164242)=353 p<10-50, subject R: F(4,123411)=152 p<10-50). Thus, the licks were invigorated by the amount of food that awaited harvest.

+

Because the tube was small, many of the licks missed their goal and failed to contact the food. The success or failure of a lick affected both the vigor of the subsequent lick, and the change in the size of the pupil. Following a successful lick there was a large increase in lick vigor (Fig. 4E, subject M: t(85182)=40 p<10-50, subject R: t(81378)=104 p<10-50), and a large increase in pupil size (subject M: F(84969)=57 p<10-50, subject R: t(80318)=94 p<10-50). In contrast, following a failed lick the subjects either reduced or did not increase their lick vigor (Fig. 4E, subject M: t(114159)=0.88 p=0.37, subject R: t(97164)=-44 p<10-50). This failure also produced a smaller increase in pupil size (comparison to successful lick, two sample t-test, subject M: t(198722)=14.8 p=4.3×10-50, subject R: t(176044)=53 p<10-50). Thus, a single successful lick led to acquisition of reward, which then was followed by a relatively large increase in pupil size, and an invigorated subsequent lick.

+

For saccades we had found that increased vigor was associated with greater accuracy. To quantify the relationship between lick vigor and accuracy, for each tube distance we labeled each reward seeking lick as being high or low vigor. For subject M, high vigor licks tended to be more successful, but this was not the case for subject R (Fig. 4F). Moreover, tube distance did not produce a consistent effect on lick success.

+

In summary, the subject licked more vigorously following a long work period in which they had accumulated more reward. Moreover, when a lick was successful in acquiring reward, they increased the vigor of the subsequent lick.

+ + +Hunger promoted work and increased vigor +

Our theory predicted that it should be possible to change decisions in one direction (say work longer), while altering movement vigor in the opposite direction (move faster). An increase in the subjective value of reward, as might occur when the subject is hungry, should have two effects: increase the number of trials that the subject chooses to perform before commencing harvest, and increase movement vigor.

+

We did not explicitly manipulate the weight of the subjects. Indeed, to maintain their health, we strived to keep their weights constant during the roughly 2.5-year period of these experiments. However, there was natural variability, which allowed us to test the predictions of the theory.

+

We found that when their weight was lower than average, the subjects chose to work a greater number of trials before commencing harvest (Fig. 5A, two sample t-test, subject M: t(11052)=7.9 p=3.4×10-25, subject M: t(12549)=10.1 p=9.3×10-24). This result was similar to the effect that we had seen when the effort cost of harvest was increased. However, the theory had predicted that the effect on vigor should be in the opposite direction: if hunger increased reward valuation, then one should speed the movements and hasten food acquisition. Notably, weight did not have a consistent effect on saccade vigor across the two subjects (Fig. 5A), yet during the harvest, both subjects licked with greater vigor when their weight was lower (Fig. 5A, subject M: t(219752)=88 p<10-50, subject R: t(205163)=22 p<10-50).

+ + +

Relatively low body weight, potentially reflecting a greater valuation of reward, coincided with longer work periods and greater vigor. Pupil size correlated with both vigor and decisions. A. Trials successfully completed during a work period as a function of normalized body weight at the start of the session. B. Left: saccade vigor as a function of trial number for low and high body weights. Right: pupil size during the same saccades. C. Left: lick vigor as a function of lick number during the harvest period. Right: pupil size during the same licks. D. Saccade vigor during the work period, and lick vigor during the harvest period, as a function of pupil size. E. Work duration and harvest duration as a function of pupil size. Error bars are SEM.

+ + +

Thus, while both the effort cost of reward and hunger promoted greater work, effort promoted sloth while hunger promoted lick vigor.

+ + +Pupil size variations strongly correlated with changes in decisions and movements +

Finally, we considered the data across both the work and the harvest periods and asked how well movement vigor tracked pupil size. The results demonstrated that in both the work and the harvest periods, for both saccades and licks, an increase in pupil size was associated with an increase in vigor (Fig. 5B, reward-relevant saccades, subject M: r=0.989 p=7.7×10-9, subject R: r=0.97 p=7.1×10-7, reward-seeking protraction licks, subject M: r=0.969 p=9.8×10-7, subject R: r=0.989 p=6.3×10-9). Moreover, when the pupil was dilated, the work periods tended to be shorter (Fig. 5C, subject M: r=-0.90 p=0.00014, subject R: r=-0.97 p=6.3×10-7), while harvest durations tended to be longer (Fig. 5C, subject M: r=0.894 p=0.00021, subject R: r=0.935 p=2.4×10-5). Thus, pupil dilation was associated with choosing to work less, while moving faster.

+ + + +Discussion +

What we choose to do is the purview of the decision-making circuits of our brain, while the implicit vigor with which we perform that action is the concern of the motor-control circuits. From a theoretical perspective (9), our brain should coordinate these two forms of behavior because both the act that we select, and its vigor, dictate expenditure of time and energy, contributing to a capture rate that affects longevity and fecundity (22). Does the brain coordinate decisions and movements to maximize a capture rate? If so, how might the brain accomplish this coordination?

+

Here, we designed a foraging task in which marmosets worked by making saccades, accumulating food for each successful trial, then stopped working and harvested their cache by licking. On every trial they decided whether to work, or to harvest. Their decision was carried out by the motor system, producing either a visually guided saccade, or a lick, each exhibiting a particular vigor. The theory predicted that to maximize the capture rate, the appropriate response to an increased effort cost of harvest was to do two things: work longer to cache more food but reduce vigor to conserve energy.

+

We varied the effort costs by moving the food tube with respect to the mouth. This changed the effort cost of harvest but not the effort cost of work. The subjects responded by altering how they worked as well as how they harvested. When the harvest was more effortful, they performed more saccade trials to stockpile food. They also slowed their movements, reducing saccade velocity during the work period, and reducing lick velocity during the harvest period. Notably, the most vigorous saccades were also the most accurate: as saccade vigor increased, so did endpoint accuracy.

+

The theory made a second prediction: as the value of reward increased, the subjects should again choose to work a longer period before initiating harvest, but unlike the effort costs, now respond by moving more vigorously. We did not directly manipulate the subjective value of reward, but rather relied on the natural fluctuations in body weight and assumed that when their weight was low, the subjects were hungrier for reward. Indeed, when their weight was low, the subjects again chose to work longer, but now elevated their vigor during the harvest period.

+

Finally, we quantified the effect of reward magnitude on vigor. Within a session, lick vigor increased robustly as a function of the number of trials completed in the preceding work period. Thus, the licks were invigorated by the amount of food that awaited harvest.

+

Notably, some of the predictions of the theory did not agree with the experimental data. An increased effort cost did not accompany a reduction in the duration of harvest, and hunger did not increase saccade vigor robustly. Indeed, earlier experiments have shown that if the effort cost of harvest increases, animals who expend the effort will then linger longer to harvest more of the reward that they have earned (2). This mismatch between observed behavior and theory highlights some of the limitations of our formulation. For example, our capture rate reflected a single work-harvest period, rather than a long sequence. Moreover, the capture rate did not consider the fact that the food tube had finite capacity, beyond which the food would fall and be wasted. This constraint would discourage a policy of working more but harvesting less. Finally, if we assume that a reduced body weight is a proxy for increased subjective value of reward, it is notable that we observed a robust effect on vigor of licks, but not saccades. A more realistic capture rate formulation awaits simulations, possibly one that describes capture rate not as the ratio of two sums (sum of gains and losses with respect to sum of time), but rather the expected value of the ratio of each gain and loss with respect to time (23, 24).

+

A shortcoming of our model is that we did not include a link between lick vigor and its probability of success. As a result, when we moved the food tube away, the model did not consider the possibility that maintaining lick accuracy may involve reduced vigor. The reason for this is that we searched for but could not find a consistent relationship, across subjects or effort conditions, between protraction speed of the tongue and its success probability. Thus, we cannot exclude this alternate hypothesis. However, the most interesting aspect of our results was that when we increased tube distance, making harvest more effortful, there was not only a reduction in lick vigor, but also a reduction in saccade vigor. That is, the decisions and actions during the work period responded to the increased effort cost of reward during the harvest period.

+

What might be a neural basis for this coordination of decisions and movements? A clue was the fact that the pupils were more constricted in sessions in which the effort cost of harvest was greater. This global change in pupil size accompanied delayed harvest and reduced vigor across sessions, but surprisingly, even within a session, transient changes in pupil size accompanied changes in vigor. During the work period the trial-to-trial reduction in saccade vigor accompanied trial-to-trial constriction of the pupil, and within a harvest period, the rapid rise and then the gradual fall in lick vigor paralleled rapid dilatation followed by gradual constriction of the pupil.

+

Pupil dilation is a proxy for activity in the brainstem neuromodulatory system (25) and is a measure of arousal (26). Control of pupil size is dependent on spiking of norepinephrine neurons in locus coeruleus (LC-NE): an increase in the activity of these neurons produces pupil dilation (27, 28). Some of these neurons show a transient change in their activity when acquisition of reward requires expenditure of either physical (11) or mental effort (29), even when there is no concomitant movement to be made. It is possible that in the present task, as the effort cost of harvest increased, LC-NE neurons decreased their activity, producing pupil constriction. If so, the reduced NE release may have had two simultaneous effects: encourage work and promote delayed gratification in brain regions that control decisions, discourage energy expenditure and promote sloth in brain regions that control movements. Thus, the idea that emerges is that the response of NE to economic variables, as inferred via changes in pupil size, might act as a bridge to coordinate the computations in the decision-making circuits with the computations in the motor-control circuits, aiming to implement a consistent control policy that improves the capture rate.

+

In addition to NE, the basal ganglia, and in particular the neurotransmitter dopamine, are likely the key contributors to the coordination of decisions with actions (30, 31). When the effort price of a preferred food increases, animals choose to work longer, pressing a lever a greater number of times (32, 33). This desire to expend effort to acquire a valuable reward is reduced if dopamine is blocked in the ventral striatum (34–36). Hunger activates circuits in the hypothalamic nuclei, disinhibiting dopamine release in response to food cues (37). Dopamine concentrations in the striatum drop when the effort price of a food reward increases (38), and dopamine release before onset of a movement tends to invigorate that movement (39). Thus, the presence of dopamine may not only alter decisions by encouraging expenditure of effort, but also modify movements by promoting vigor.

+

Experiments of Hayden et al. (40) and Barack et al. (41) suggest that the decision of when to stop work and commence harvest may rely on computations that are carried out in the cingulate cortex. They found that as monkeys deliberated between the choice of staying and acquiring diminishing rewards, or leaving and incurring a travel cost, these neurons encoded a decision-variable that reflected the value of leaving the patch. The prediction that emerges from our work is that the rate of rise of these decision variables may be modulated by the presence of NE.

+

From a motor control perspective, a surprising aspect of our results was that an increase in saccade vigor accompanied an improvement in endpoint accuracy. In our earlier work we found that during reaching, reward increased vigor without reducing accuracy (42). Thus, the brain has the means to increase movement vigor and improve its accuracy. How is this achieved?

+

We found that the high vigor saccades were produced when the pupils were dilated, implying an increased release of NE. In songbirds, increased NE release acts on the basal ganglia to suppress activity of spiny neurons, and this reduced activity in the basal ganglia accompanies reduced variance in the songs that the animal sings (43). Thus, NE may play a critical role in control of movement variability. For saccades, control of endpoint accuracy depends on the coordinated activity of Purkinje cells in the oculomotor region of the cerebellar vermis (44, 45). LC projects to the cerebellum, and stimulation of LC neurons increases the sensitivity of Purkinje cells to their inputs (46).

+

Is movement vigor increased following increased NE inputs from LC to the basal ganglia, and accuracy improved following increased NE inputs from LC to the cerebellum? Does decision-making shift toward greater work and delayed gratification following reduced NE inputs from LC to the frontal lobe? These are some of the questions that await future experiments.

+ + +Methods +

Behavioral and neurophysiological data were collected from two marmosets (Callithrix Jacchus, male and female, 350-390 g, subjects R and M, 6 years old). The neurophysiological data focused on the cerebellum and are described elsewhere (44, 47, 48). Here, our focus is on the behavioral data.

+

The marmosets were born and raised in a colony that Prof. Xiaoqin Wang has maintained at the Johns Hopkins School of Medicine since 1996. The procedures on the marmosets were evaluated and approved by the Johns Hopkins University Animal Care and Use Committee in compliance with the guidelines of the United States National Institutes of Health.

+ +Data acquisition +

Following recovery from head-post implantation surgery, the animals were trained to make saccades to visual targets and rewarded with a mixture of apple sauce and lab diet (47). They were placed in a monkey chair and head-fixed while we presented visual targets on an LCD screen (Curved MSI 32” 144 Hz - model AG32CQ) and tracked both eyes at 1000 Hz using an EyeLink-1000 system (SR Research, USA). Timing of target presentation on the video screen was measured using a photo diode. Tongue movements were tracked with a 522 frame per second Sony IMX287 FLIR camera, with frames captured at 100 Hz.

+

Each trial began with a saccade to the center target followed by fixation for 200 ms, after which a primary target (0.5×0.5 deg square) appeared at one of 8 randomly selected directions at a distance of 5-6.5 deg. Onset of the primary target coincided with presentation of a tone. As the animal made a saccade to the primary target, that target was erased, and a secondary target was presented at a distance of 2-2.5 deg, also at one of 8 randomly selected directions. The subject was rewarded if following the primary saccade it made a corrective saccade to the secondary target, landed within 1.5 deg radius of the target center, and maintained fixation for at least 200 ms. Onset of reward coincided with presentation of another distinct tone. Following an additional 150-250 ms period (uniform random distribution), the secondary target was erased, and the center target was displayed, indicating the onset of the next trial. Thus, a successful trial comprised of a sequence of 3 saccades: center, primary, and corrective, after which the subject received a small increment of food (0.015 mL).

+

The food was provided in two small tubes (4.4 mm diameter), one to the left and the other to the right of the animal (Fig. 1A). A successful trial produced a food increment in one of the tubes and would continue to do so for 50-300 consecutive trials, then switch to the other tube. Because the food increment was small, the subjects naturally chose to work for a few consecutive trials, tracking the visual targets and allowing the food to accumulate, then stopped tracking and harvested the food via a licking bout. The licking bout typically included a sequence of 15-40 licks. The subjects did not work while harvesting. As a result, the behavior consisted of a work period (targeted saccades), followed by a harvest period (targeted licking), repeated hundreds of times per session.

+

The critical variables were the number of trials that the subject chose to perform before initiating harvest, the vigor of their saccades during the work period, and the vigor of their licks during the harvest period.

+ + +Data analysis +

All saccades, regardless of whether they were instructed by presentation of a visual target or not, were identified using a velocity threshold. Saccades to primary, secondary, and central targets were labeled as reward-relevant saccades, while all remaining saccades were labeled as task irrelevant.

+

We analyzed tongue movements using DeepLabCut (12). Our network was trained on 89 video recordings of the subjects with 15-25 frames extracted and labeled from each recording. The network was built on the ResNet-152 pre-trained model, and then trained over 1.03×10 iterations with a batch size of 8, using a GeForce GTX 1080Ti graphics processing unit (49). A Kalman filter was further applied to improve quality and smoothness of the tracking, and the output was analyzed in MATLAB to quantify varying lick events and kinematics.

+

We tracked the tongue tip and the edge of the food in the tube, along with control locations (nose position and tube edges). We tracked all licks, regardless of whether they were aimed toward the tube (reward seeking), or not (grooming). Reward seeking licks were further differentiated based on whether they aimed to enter the tube (inner-tube licks), hit the outer edge of the tube (outer-edge licks), or fell below the tube (under tube). If any of these licks successfully contacted the tube, we labeled that lick as a success (otherwise, a failed lick).

+

Pupil area was measured during a ±250 ms period centered at the onset of each reward-relevant saccade, and the onset of each lick. We then normalized the pupil measurements by representing it as a z-score with respect to the mean value for that session.

+ + +Saccade and tongue vigor +

We relied on previous work to define vigor of a movement (5, 9, 18, 19). Briefly, if the amplitude of a movement is x and the peak speed of that movement is v, then for each subject the relationship between the two variables can be described as: + + + +In the above expression, α,β ≥ 0 and are subject-specific parameters. For a movement with amplitude x, its vigor was defined as the ratio of the actual peak speed with respect to the expected value of its peak speed, i.e., . Expected value was computed by fitting Eq. (3) to all the data acquired across all sessions. When vigor is greater than 1, the movement had a peak velocity that was higher than the mean value associated with that amplitude.

+ + +Model formulation +

We chose a formulation of utility (Eq. 1) based on a normative approach that ecologists have used to understand the decisions that animals make regarding how far to travel for food, what mode of travel to use, and how long to stay before moving on to another reward opportunity (4, 13, 14). In a typical formulation of the theory, the numerator represents the reward gained (in units of energy), minus the effort expended (also in units of energy), while the denominator represents the amount of time spent during that behavior. We represented this idea in Eq. (1) with saccades that produced reward accumulation, and licks that produced reward consumption. Thus, the utility that we aim to maximize is the rate of energy gained.

+

The specific functions that we used to represent the energy gained through reward acquisition, and the energy expended through effort expenditure, came either from experiment design, or from the measurements we have made in other experiments. We modeled reward accumulation as a linear rise in energy stored because successful saccades produced a linear increase in the food cache. We modeled harvesting of the food as a hyperbolic function of the number of licks to represent the fact that as the licking bout began, each successful lick depleted the food, and thus the first few licks produced a greater amount of food consumption than the last few licks. We modeled the effort cost of licking as a linear function of the number of licks.

+

A critical assumption that we made is that energy expended performing the saccade trials (which grew faster than linearly as a function of the number of trials attempted), grew faster than the time spent attempting those same trials (which grew linearly with the number of trials). This assumption is based on the heuristic that the average rate of energy lost following a large number of attempted trials is greater than the average rate of energy lost following a small number of attempted trials.

+

The model’s simplicity provided closed-form solutions across all parameter values, allowing us to make predictions without having to fit the model to the measured data. For example, for all parameter values that produce a real solution (as opposed to imaginary), the optimal number of saccade trials increases with the square root of the cost of licking. Thus, the basic prediction of the model is that to maximize the capture rate, regardless of parameter values, an increase in the effort required for harvest should be met with a greater willingness to work. The closed-form solutions are presented in the supplementary document (simulations.nb).

+ + +Other models of utility +

In composing our utility (Eq. 1), we chose to combine reward and effort additively. This is in contrast to other approaches in which effort discounts reward multiplicatively (50–52). Our reasoning is that multiplicative interactions have the limitation that they are incompatible with the observation that reward invigorates movements.

+

To compare additive and multiplicative approaches, let us consider an arbitrary function u r that specifies how effort varies with movement duration T. Typically, this is a U-shaped function that describes energy expenditure as a function of movement duration, as in (8). In the case of multiplicative interaction between reward and effort, we can consider the following representation of utility: + + + +In the above formulation, reward a is discounted hyperbolically with time, and an increase in reward increases the utility of the action. The optimum movement vigor has the duration T* that maximizes this utility. Notably, because increasing reward merely scales this utility, it has no effect on vigor. Thus, a utility in which reward is multiplied by a function of effort generally fails to predict dependence of movement vigor on reward.

+ + +Simulations +

The optimal policy specifies the decisions and movements that for the effort cost defined in Eq. (2), maximizes the capture rate defined in Eq. (1). This policy selects the number of saccade trials ns to perform during the work period, the number of licks nl to perform during the harvest period, and the vigor of each lick, represented by the average duration of a lick Tl. To compute the optimal policy, we found the derivative of the capture rate with respect to each policy variable ns, nl, and Tl, then set each derivative equal to zero, producing three simultaneous nonlinear equations. In all three cases, we were able to solve for the relevant control variable analytically (see Mathematica notebook simultatons.nb for the derivations). We found that if the solution was a real number, then regardless of parameter values, an increase in d (distance of the tube to the mouth), the optimal policy produced an increase in decrease in and increase in . Thus, the results illustrated in Fig. 2 are robust to changes in parameter values.

+

To generate the plots in Fig. 2A, we used the following parameter values: α = 20, βs = 0.5, βL = 0.3, cs = 0.5, Ts = 1, TL = 0.2, cL = 0.5 (low effort), cL = 2.5 (high effort). For the plots in Fig. 2C and 2D, we used the same parameter values, but cL was defined via Eq. (2). Thus, tube distance d varied, and TL was unknown and was solved for. In Eq. (2), kL = 1. In the simulations, to describe state of hunger, we set α = 20 for a sated state and a= 25 for a hungry state.

+ + +Statistical analysis +

Hypothesis testing was performed using functions provided by the MATLAB Statistics and Machine Learning Toolbox, version R2021b. For t-tests, across the one-sample, paired-sample, and two-sample conditions, p values were computed using the ttest and ttest2 functions with data that was combined across sessions, separated by condition. For ANOVA tests, in the one-way condition, p values were computed using a nonparametric Kruskal-Wallis test, using the kruskalwallis function. In the 2-way condition, the anovan function was used to compute p values, accounting for an unbalanced design resulting from a varied number of samples across conditions. In both cases, like in the t-tests, data was combined across sessions, separated by condition. In the repeated measures condition, each session was treated as a subject with multiple repeated measures representing a given variable (i.e., lick vigor per lick in a harvest period). To fit a repeated measures model, the fitrm function was used, then analyzed using the ranova function. In all cases of repeated measures ANOVA, compound symmetry assumptions were tested using the Mauchly sphericity test with the maulchy function. In cases where the assumption was violated (Maulchy test p < 0.05), epsilon adjustments were used, with the epsilon function, to compute corrected p values (for ε > .75, use Huynh-Feldt p value and for ε < .75, use Greenhouse-Geisser p values). For correlation analyses, Pearson’s correlation coefficient, r, and corresponding p values were computed using the corrcoef function.

+ + + + + +Acknowledgements +

The work was supported by grants from the NIH (R01-EB028156, R01-NS078311, R37-NS128416), and the Office of Naval Research (N00014-15-1-2312).

+ + +Supporting Information + + +

Number of licks per harvest as a function of tube distance.

+ + +

Video 1. Example of a successful inner-tube lick.

+

Video 2. Example of an under-tube lick that failed to contact food. Note the corrective sub-movements, as has been observed in mice (53).

+

Video 3. Example of an outer-tube lick that failed to contact food.

+

Video 4. Example of a lick that hit the outer edge of the tube and failed to contact food.

+

Video 5. Example of a grooming lick.

+ + +References +E. L. Charnov, Optimal foraging, the marginal value theorem. Theor.Popul.Biol. 9, 129–136 (1976). +R. J. Cowie, Optimal foraging in great tits (Parus major). Nature 268, 137–139 (1977). +R. Shadmehr, A. A. Ahmed, Vigor: neuroeconomics of movement control (MIT Press, 2020). +H. Richardson, N. A. M. Verbeek, Diet selection and optimization by northwestern crows feeding on Japanese littleneck clams. Ecology 67, 1219–1226 (1986). +T. Yoon, A. Jaleel, A. A. Ahmed, R. Shadmehr, Saccade vigor and the subjective economic value of visual stimuli. J. Neurophysiol. 123, 2161–2172 (2020). +C. C. Korbisch, D. R. Apuan, R. Shadmehr, A. A. Ahmed, Saccade vigor reflects the rise of decision variables during deliberation. Curr. Biol. (2022) 10.1016/j.cub.2022.10.053 (December 4, 2022). +R. Shadmehr, J. J. Orban de Xivry, M. Xu-Wilson, T. Y. Shih, Temporal discounting of reward and the cost of time in motor control. J.Neurosci. 30, 10507–10516 (2010). +R. Shadmehr, H. J. Huang, A. A. Ahmed, A representation of effort in decision-making and motor control. Curr.Biol. 26, 1929–1934 (2016). +T. Yoon, R. B. Geary, A. A. Ahmed, R. Shadmehr, Control of movement vigor and decision making during foraging. Proc.Natl.Acad.Sci.U.S.A. 115, E10476–E10485 (2018). +S. Joshi, J. I. Gold, Pupil Size as a Window on Neural Substrates of Cognition. Trends Cogn. Sci. 24, 466–480 (2020). +P. Bornert, S. Bouret, Locus coeruleus neurons encode the subjective difficulty of triggering and executing actions. PLOS Biol. 19, (2021). +A. Mathis, et al., DeepLabCut: markerless pose estimation of user-defined body parts with deep learning. Nat. Neurosci. 21, 1281–1289 (2018). +D. W. Stephens, J. R. Krebs, Foraging Theory (Princeton Univ. Press, 1986). +L. M. Bautista, J. Tinbergen, A. Kacelnik, To walk or to fly? How birds choose among foraging modes. Proc.Natl.Acad.Sci.U.S.A. 98, 1089–1094 (2001). +H. J. Huang, A. A. Ahmed, Older adults learn less, but still reduce metabolic cost, during motor adaptation. J.Neurophysiol. 111, 135–144 (2014). +H. J. Ralston, Energy-speed relation and optimal speed during level walking. Int.Z.Angew.Physiol. 17, 277–283 (1958). +G. J. Bastien, P. A. Willems, B. Schepens, N. C. Heglund, Effect of load and speed on the energetic cost of human walking. Eur.J.Appl.Physiol 94, 76–83 (2005). +T. R. Reppert, K. M. Lempert, P. W. Glimcher, R. Shadmehr, Modulation of Saccade Vigor during Value-Based Decision Making. J.Neurosci. 35, 15369–15378 (2015). +T. R. Reppert, et al., Movement vigor as a trait-like attribute of individuality. J.Neurophysiol. 120, 741–757 (2018). +C. M. Harris, D. M. Wolpert, Signal-dependent noise determines motor planning. Nature 394, 780– 784 (1998). +C. Wang, Y. Xiao, E. Burdet, J. Gordon, N. Schweighofer, The duration of reaching movement is longer than predicted by minumum variance. J.Neurophysiol. 116, 2342–2345 (2016). +W. C. Lemon, Fitness consequences of foraging behaviour in the zebra finch. Nature 352, 153–155 (1991). +M. Bateson, A. Kacelnik, Preferences for fixed and variable food sources: variability in amount and delay. J.Exp.Analysis.Behav. 63, 313–329 (1995). +M. Bateson, A. Kacelnik, Rate currencies and the foraging starling: the fallacy of the averages revisited. Behav. Ecol. 7, 341–352 (1996). +E. M. Vazey, D. E. Moorman, G. Aston-Jones, Phasic locus coeruleus activity regulates cortical encoding of salience information. Proc. Natl. Acad. Sci. 115, E9439–E9448 (2018). +S. Mathot, Pupillometry: Psychology, Physiology, and Function. J. Cogn. 1, 1–23 (2018). +S. Joshi, Y. Li, R. M. Kalwani, J. I. Gold, Relationships between Pupil Diameter and Neuronal Activity in the Locus Coeruleus, Colliculi, and Cingulate Cortex. Neuron 89, 221–234 (2016). +V. Breton-Provencher, M. Sur, Active control of arousal by a locus coeruleus GABAergic circuit. Nat. Neurosci. 22, 218–228 (2019). +C. Contadini-Wright, K. Magami, N. Mehta, M. Chait, Pupil Dilation and Microsaccades Provide Complementary Insights into the Dynamics of Arousal and Instantaneous Attention during Effortful Listening. J. Neurosci. 43, 4856–4866 (2023). +D. Thura, P. Cisek, The Basal Ganglia Do Not Select Reach Targets but Control the Urgency of Commitment. Neuron 95, 1160–1170 (2017). +D. M. Herz, et al., Dynamic control of decision and movement speed in the human basal ganglia. Nat. Commun. 13, 7530 (2022). +J. D. Salamone, et al., Haloperidol and nucleus accumbens dopamine depletion suppress lever pressing for food but increase free food consumption in a novel food choice procedure. Psychopharmacol. Berl 104, 515–521 (1991). +J. E. Aberman, J. D. Salamone, Nucleus accumbens dopamine depletions make rats more sensitive to high ratio requirements but do not impair primary food reinforcement. Neuroscience. 92, 545– 552 (1999). +M. Koch, A. Schmid, H. U. Schnitzler, Role of nucleus accumbens dopamine D1 and D2 receptors in instrumental and Pavlovian paradigms of conditioned reward. Psychopharmacol. Berl 152, 67–73 (2000). +A. M. Farrar, et al., Nucleus accumbens and effort-related functions: behavioral and neural markers of the interactions between adenosine A2A and dopamine D2 receptors. Neuroscience 166, 1056–1067 (2010). +S. E. Yohn, et al., The role of dopamine D1 receptor transmission in effort-related choice behavior: Effects of D1 agonists. Pharmacol.Biochem.Behav. 135, 217–226 (2015). +R. M. Cassidy, Q. Tong, Hunger and Satiety Gauge Reward Sensitivity. Front. Endocrinol. 8 (2017). +S. A. Schelp, et al., A transient dopamine signal encodes subjective value and causally influences demand in an economic context. Proc.Natl.Acad.Sci.U.S.A 114, E11303–E11312 (2017). +J. A. da Silva, F. Tecuapetla, V. Paixao, R. M. Costa, Dopamine neuron activity before action initiation gates and invigorates future movements. Nature 554, 244–248 (2018). +B. Y. Hayden, J. M. Pearson, M. L. Platt, Neuronal basis of sequential foraging decisions in a patchy environment. Nat.Neurosci. 14, 933–939 (2011). +D. L. Barack, S. W. C. Chang, M. L. Platt, Posterior Cingulate Neurons Dynamically Signal Decisions to Disengage during Foraging. Neuron 96, 339–347 (2017). +E. M. Summerside, R. Shadmehr, A. A. Ahmed, Vigor of reaching movements: reward discounts the cost of effort. J.Neurophys. 119, 2347–2357 (2018). +J. Singh Alvarado, et al., Neural dynamics underlying birdsong practice and performance. Nature 599, 635–639 (2021). +E. Sedaghat-Nejad, J. S. Pi, P. Hage, M. A. Fakharian, R. Shadmehr, Synchronous spiking of cerebellar Purkinje cells during control of movements. Proc. Natl. Acad. Sci. 119, e2118954119 (2022). +S. Barash, et al., Saccadic dysmetria and adaptation after lesions of the cerebellar cortex. J.Neurosci. 19, 10931–10939 (1999). +H. C. Moises, B. D. Waterhouse, D. J. Woodward, Locus coeruleus stimulation potentiates Purkinje cell responses to afferent input: The climbing fiber system. Brain Res. 222, 43–64 (1981). +E. Sedaghat-Nejad, et al., Behavioral training of marmosets and electrophysiological recording from the cerebellum. J.Neurophysiol. 122, 1502–1517 (2019). +S. Z. Muller, et al., Complex spikes perturb movements, revealing the sensorimotor map of Purkinje cells. bioRxiv, 2023–04 (2023). +K. He, X. Zhang, S. Ren, J. Sun, Deep Residual Learning for Image Recognition in (2016), pp. 770– 778. +H. Sugiwaka, H. Okouchi, Reformative self-control and discounting of reward value by delay or effort. Jpn. Psychol. Res. 46, 1–9 (2004). +C. Prevost, M. Pessiglione, E. Metereau, M. L. Clery-Melin, J. C. Dreher, Separate valuation subsystems for delay and effort decision costs. J.Neurosci. 30, 14080–14090 (2010). +M. C. Klein-Flugge, S. W. Kennerley, A. C. Saraiva, W. D. Penny, S. Bestmann, Behavioral modeling of human choices reveals dissociable effects of physical effort and temporal delay on reward devaluation. PLoS.Comput.Biol. 11, e1004116 (2015). +T. Bollu, et al., Cortex-dependent corrections as the tongue reaches for and misses targets. Nature, 1–6 (2021). + + +
diff --git a/test/fixtures/2023.04.19.535725/2023.04.19.535725.xml b/test/fixtures/2023.04.19.535725/2023.04.19.535725.xml new file mode 100644 index 00000000..88eb0a35 --- /dev/null +++ b/test/fixtures/2023.04.19.535725/2023.04.19.535725.xml @@ -0,0 +1,450 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.04.19.535725 +1.2 + + +Regular Article + + +New Results + + +Neuroscience + + + +Subchronic alteration of vestibular hair cells in mice: implications for multisensory gaze stabilization + + + +http://orcid.org/0009-0004-1298-8228 +SchenbergLouise +1 + + +PalouAïda +2 +3 +4 + + +SimonFrançois +1 +5 + + +BonnardTess +1 + + +BartonCharles-Elliot +1 + + +http://orcid.org/0000-0001-7328-9480 +FrickerDesdemona +1 + + +http://orcid.org/0000-0002-4905-788X +TagliabueMichele +1 + + +http://orcid.org/0000-0002-3894-9401 +LlorensJordi +2 +3 +4 +* + + +http://orcid.org/0000-0003-2722-0532 +BeraneckMathieu +1 +* + +Université Paris Cité, CNRS UMR 8002, INCC - Integrative Neuroscience and Cognition Center, F-75006, Paris, France +Departament de Ciències Fisiològiques, Universitat de Barcelona, 08907 l’Hospitalet de Llobrega, Catalunya, Spain +Institut de Neurociènces, Universitat de Barcelona, Barcelona, Catalunya, Spain +Institut d’Investigació Biomèdica de Bellvitge (IDIBELL), 08907 l’Hospitalet de Llobregat, Catalunya, Spain +Department of paediatric otolaryngology, Hôpital Necker-Enfants Malades, APHP, F-75015, Paris, France + + +Corresponding authors: Pr Jordi Llorens, Universitat de Barcelona, 08907 l’Hospitalet de Llobrega, Catalunya, Spain. E-mail : jllorens@ub.edu; Dr. M. Beraneck. CNRS UMR 8002, Université Paris Cité, 45 rue des St-Pères, Paris 75270, France. Email: mathieu.beraneck@cnrs.fr + +2023 +2023.04.19.535725 + + +19 +4 +2023 + + +19 +9 +2023 + + +19 +9 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +The copyright holder for this pre-print is the author. All rights reserved. The material may not be redistributed, re-used or adapted without the author's permission. + + + +Abstract

The functional complementarity of the vestibulo-ocular reflex (VOR) and optokinetic reflex (OKR) allows for optimal combined gaze stabilization responses (CGR) in light. While sensory substitution has been reported following complete vestibular loss, the capacity of the central vestibular system to compensate for partial peripheral vestibular loss remains to be determined. Here, we first demonstrate the efficacy of a 6-week subchronic ototoxic protocol in inducing transient and partial vestibular loss which equally affects the canal- and otolith-dependent VORs. Immunostaining of hair cells in the vestibular sensory epithelia revealed that organ-specific alteration of type I, but not type II, hair cells correlates with functional impairments. The decrease in VOR performance is paralleled with an increase in the gain of the OKR occurring in a specific range of frequencies where VOR normally dominates gaze stabilization, compatible with a sensory substitution process. Comparison of unimodal OKR or VOR versus bimodal CGR revealed that visuo-vestibular interactions remain reduced despite a significant recovery in the VOR. Modeling and sweep-based analysis revealed that the differential capacity to optimally combine OKR and VOR correlates with the reproducibility of the VOR responses. Overall, these results shed light on the multisensory reweighting occurring in pathologies with fluctuating peripheral vestibular malfunction.

+ + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

Figure 1, 2 and 3 revised; discussion updated to clarify the link between hair cells and vestibular afferents; conclusion added to discuss vestibular pathologies.

 + + + + + +Introduction +

The vestibular system is well-preserved amongst vertebrates, participating in essential functions such as balance, postural control and, together with the optokinetic system, gaze stabilization (Straka et al., 2016; Wibble et al., 2022). Beyond these recognized roles, vestibular signals also contribute to cognitive processes e.g spatial orientation and navigation (Cullen, 2019), or body representation (Lopez et al., 2012; Facchini et al., 2021). Because of its involvement in many basic functions important in our daily life, vestibular pathologies affecting the inner ear are associated with a significant deterioration of the well-being of patients (Möhwald et al., 2020) and represent an important public health concern (Agrawal et al., 2009, 2013).

+

Research on post-lesion plasticity following permanent vestibular loss has shed light on the neural plastic mechanisms that follow a chronic unilateral or bilateral vestibular lesion, a process referred as “vestibular compensation” (Brandt et al., 1997; Cullen et al., 2010; Beraneck and Idoux, 2012). The compensation taking place after the lesion is known to involve dynamical multisensory reweighting of proprioceptive and visual inputs and of internal efferent copies (Cullen et al., 2010; Sadeghi et al., 2012; Sadeghi and Beraneck, 2020). While total and permanent lesions offer the experimental opportunity to characterize drastic cellular and molecular changes triggered by the total silencing of the vestibular endorgans, they imperfectly mimic clinical situations where peripheral vestibular function loss is only partial and/or transient (Bisdorff et al., 2009; Lopez-Escamez et al., 2015; Brandt and Dieterich, 2017). To better model fluctuating inner ear function, protocols based on subchronic exposure to an ototoxic substance, 3,3ʹ-iminodiproprionitrile (IDPN) were first introduced in the rat (Seoane et al., 2001; Sedó-Cabezón et al., 2014, 2015; Martins-Lopes et al., 2019) and more recently in the mouse (Greguske et al., 2019). Subchronic exposure to IDPN in drinking water at low doses allowed for progressive ototoxicity, leading to a partial and largely reversible loss of function. Although the subchronic IDPN protocol was shown to cause postural and locomotor deficits (Martins-Lopes et al., 2019), its effects on the gaze stabilizing reflexes, namely the vestibulo-ocular and optokinetic reflexes, have not yet been described.

+

The primary objective of the present study is first to assess how subchronic exposure to IDPN may affect the function of the different vestibular endorgans. To that end, we took advantage of our recently described methodology (Simon et al., 2020, 2021) using canal-specific and otolith-specific tests. Quantification of the vestibulo-ocular reflexes is a sensitive and specific method to assess the functionality of the sensory-motor vestibular pathway; it is the most used test in clinics, and highly correlates to quality-of-life reports in patients suffering from acute peripheral diseases (Möhwald et al., 2020).

+

The secondary objective is to determine whether visual substitution occurs following transient and partial vestibular loss. Optogenetic stimulation of the vestibular pathway demonstrated the recruitment of circuits involved in visual processing at the midbrain, thalamic and cortical regions (Leong et al., 2019). Recent imaging studies in rodents have shown that acute vestibular loss triggers brain-wide adaptive plasticity in circuits known to be involved in visual processing (Zwergal et al., 2016; Grosch et al., 2021). In addition, it was previously shown that OKR plasticity is triggered during vestibular compensation following a permanent vestibular lesion (Faulstich et al., 2006; Nelson et al., 2017).

+

We report that 6 weeks of IDPN subchronic treatment affects both the canal- and otolith-dependent vestibulo-ocular reflexes and that organ-specific loss of type I hair cells (HC) correlates with individual mice’s impairments. We show that optokinetic adaptive compensation is frequency-specific and delayed with respect to the VOR changes. OKR changes occur at the frequencies where physiologically the vestibular inputs dominate visuo-vestibular gaze stabilization. We demonstrate that despite the significant recovery of their vestibulo-ocular reflexes, the visuo-vestibular integration remains notably impaired in some IDPN-treated mice. We suggest that the “noisiness” of the recovered vestibular signal affects their capacity to optimally combine visual and vestibular responses. Overall, these results shed light on the dynamic of multisensory reweighting in patients suffering from fluctuating peripheral vestibular malfunction.

+ + +Results + +Effects of the subchronic treatment of IDPN on the canal- and otolith-dependent VOR +

To investigate the effects of the IDPN on vestibulo-ocular reflexes (VOR), animals were exposed to the ototoxic compound in the drinking water for six weeks (Treatment period), followed by 6 weeks of standard drinking water without IDPN (Washout period). The VOR were quantified every two weeks using canal-specific and otolith-specific tests. Horizontal sinusoidal rotations in the dark were performed and oculomotor responses were recorded using video-oculography (Figure 1A) to study the impact of IDPN treatment on the canal-dependent angular VOR (aVOR). Typical raw aVOR traces are shown in Figure 1B. At week 6 (W6) the amplitude of the eye movements was distinctly reduced compared to W0, while at W12 the amplitude of the response appeared partially restored. The dynamics of decrease and recovery of the mean aVOR gain over the course of the protocol are reported in Figure 1C for both IDPN (n=21) and SHAM (n=22) groups. The evolution over time of the aVOR gain is significantly different for the two groups of mice (ANOVA Weeks x Group interaction, F(6,246)=29,949, p<10-4). Before treatment, both groups responded similarly to the sinusoidal stimulations and their aVOR gain remained unchanged through 2 weeks of treatment (W2). However, starting W4 the aVOR gain of IDPN group significantly decreased with respect to W0 (Newman-Keuls post hoc test: IDPN W0 vs W4 p<10-4, see Table 1) and was significantly lower compared with the SHAM group (W4, IDPN vs SHAM p<10-4). The aVOR gain of the IDPN group remained lower compared to SHAM through the rest of the protocol (W6 IDPN vs SHAM p<10-4., W8 IDPN vs SHAM p<10-4, W10 IDPN vs SHAM p<10-4, W12 IDPN vs SHAM p<10-4) with a minimum reached at W6 corresponding to ∼2/3 of aVOR loss (IDPN W4 vs W6 p<10-4, IDPN W6 vs W8 p=0,043762). At the end of the six weeks of washout, the mean aVOR significantly improved (IDPN W6 vs W12 p<10-4) to levels observed at W4 (IDPN W4 vs W12 p=0.561969). Notably, the amplitude and dynamic of gain changes were similar for all frequencies >0.2Hz (see Figure 1 supplement 1A). However, at the lowest frequency tested (0.2Hz), aVOR gain decrease reached significance only at W6.

+ + +Effects of subchronic IDPN on canal- and otolithic- dependent VOR. +

A) Illustration of the angular horizontal vestibulo-ocular reflex (aVOR) set-up. All tests are performed in complete dark. B) Examples of velocity aVOR traces in response to table rotations (1Hz at 30°/s) recorded in the dark in an IDPN mouse before (W0, corresponding gain of 0.785; VAF of 0.98), after 6 weeks of treatment (W6, gain: 0.14 ; VAF: 0.61) and 6 weeks of washout (W12, gain:0.48; VAF: 0.92). Right movement is represented up. C) Mean aVOR gain of SHAM (n=22) and IDPN (n=21) mice during the protocol. D) Illustration of the ocular-counter roll (OCR) set-up. E)Examples of raw OCR traces at W0, W6 and W12 in a IDPN mouse recorded in the dark. Tilt to the right is represented up (positive values). (Left) eye elevation is represented up. F) Mean OCR gain of SHAM (n=14) and IDPN (n=13) mice. We note that there was a significant difference between SHAM and IDPN during the initial measurements at W0. However, at this time point mice were not yet separated into different groups. This incidental difference completely disappeared on the measurement performed at W2. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + + + +Statistics table of the aVOR gain for the IDPN treated group. + + +

aVOR responses were further modified by significant phase leads that affected all frequencies starting W6 (Figure 1 supplement 1B, ANOVA repeated measures Weeks x Group Interactions F(6, 246)=14.528, p<10-4). Overall, canalar responses remained unaffected until week 2, but the amplitude and/or timing of the aVOR was abnormal from week 4 until the end of the protocol, despite a significant recovery of angular VOR responses observed during the washout period.

+

To determine whether otolith-dependent VOR was also affected by IDPN treatment, ocular-counter roll responses (OCR) were tested during static lateral inclination in the range ±40° (OCR; Figure 1D). Examples of raw traces and quantification of the response are shown in Figure 1E, and mean gain of the OCR of each group is plotted in Figure 1F (n=13 IDPN, n=14 SHAM). The modulations of responses amplitude were significantly different between IDPN and SHAM groups (ANOVA repeated measures Weeks x Group Interaction F(6, 150)= 7.7411 p<10-4). We note that there was a significant difference between SHAM and IDPN during the initial measurements at W0, before any treatment. However, at this timepoint mice were not yet separated into different groups. This incidental difference completely disappeared on the measurement performed at W2 (W2 IDPN vs SHAM p=0.8135). While the gain of the SHAM group stays in a 0.5-0.6 range over the whole duration of the protocol, the responses of the IDPN group significantly decreased at W4 compared to the SHAM group (W4 IDPN vs SHAM p<10-4). This decrease was larger at W6 (IDPN W4 vs W6 p=0.023) and stayed significantly different from the SHAM group until W8. At W10 and W12, the OCR of IDPN group recovered to a level comparable to the SHAM group (W10 IDPN vs SHAM p=0.1262, W12 IPDN vs SHAM p=0.3385).

+

Sub-chronic treatment of IDPN was also investigated through the dynamic Off Vertical Axis Rotation (OVAR) test (Figure 1 supplement 1C), which primarily reflects maculo-ocular (dominantly otolithic) responses integrated by central vestibular pathways. The maculo-ocular reflex (MOR) bias decreased significantly compared to the SHAM group (ANOVA Weeks x Group, F(6, 132)= 10.076, p<10-4) starting W2 of treatment (W2 IDPN vs SHAM p=0.0018), demonstrating that ototoxicity already affected the vestibular system at this early time point. The maximal decrease was reached at W6 (W6 IDPN vs SHAM p=0.00012) and recovery led to normal responses at W12 (W12 IDPN vs SHAM p=0.4709).

+ + +Comparison of otolith- and canal-dependent plasticity in individuals +

To compare the dynamic of canal- and otolith-dependent VOR alterations, the paired VOR and OCR gains measured in individuals from the IDPN group (n=13) are plotted together (Figure 2). Alteration in canal- and otolith-dependent responses followed a very similar time course (OCR vs aVOR at 1HZ on figure 2A; similar patterns were obtained with the other tested frequencies (not shown)).

+ + +Comparison of the dynamics of canalar and otolithic loss of function. +

A) Evolution of the Mean Gain of the aVOR (1Hz) and OCR responses of IDPN mice (n=13) during the treatment and washout periods. B) Individual gains for 1Hz aVOR (left) and OCR (right) of the 13 IDPN mice, identified with similar symbols. The grey symbols in the left panel correspond to aVOR gain values associated with VAF<0.5. C, D) Individual ΔaVOR gains as a function of individual ΔOCR gains at W6 (C) and W12 (D) compared to W0, for SHAM (n=14) and IDPN (n=13) mice. The linear regression corresponds to IDPN values is represented, as well as the 50% confidence interval of each group (shaded areas). The symbols for each animal are the same in panels B, C and D. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To investigate further the organ-specific responses, the individual values of VOR gain at 1Hz (left panel) and OCR slope (right panel) are plotted in Figure 2B. These plots show the variability of the responses observed in different individuals. 2/3 of the mice had aVOR responses lowered by more than 50%, while 1/3 had milder aVOR impairments. In most cases, however, the aVOR gain decreased notably between W4 and W6 and started to recover from W8 (compare individual slopes in Figure 2B). Similarly, OCR gains were variable between individuals. However, the dynamic of the variations in otolith-dependent responses followed a pattern comparable to that of the canal-dependent changes.

+

To determine whether the changes in canal- and otolith-dependent responses were proportional, individual variations in gain since pre-treatment of the aVOR and OCR were compared at W6 (Figure 2C) and W12 (Figure 2D). At W6, there was a significant correlation between the amplitude of the changes observed in aVOR and OCR, such that canalar and otolithic loss of function was proportional (slope of regression line: 0.5586). Despite the general recovery, this significant correlation was preserved at W12: the recovery in canal-dependent responses was proportional to the recovery in otolith-dependent responses (slope of regression line: 0.5566). Overall, these results suggest that the subchronic IDPN treatment similarly affects the vestibulo-ocular reflexes that depend on semi-circular canals and on the otolithic organs, respectively.

+ + +Effects of the IDPN exposure on the number of hair cells in the vestibular epithelia +

To correlate the vestibular loss of function to the structural changes induced by the ototoxic compound, vestibular epithelia were dissected and labelled to assess the number of hair cells (HC) in the organs. One horizontal semi-circular canal and one utricule were harvested at W6 for n=7 IDPN and n=4 SHAM, and at W12 for n=8 IDPN and n=4 SHAM, and each organ epithelium was divided into the central and peripheral region to differentiate the possible participation of zone- and organ-specific HC to the vestibular function. Vestibular HC were labelled with type-specific immunomarkers in both endorgans (Figure 3). Confocal immunostaining data are presented in Figure 3A1 for the horizontal semi-circular canals and Figure 3C1 for the utricule. 6-week long treatment of IDPN exposure led to a significant reduction in the labelling of type I HC in the canal crista (quantified in Figure 3A2, Non parametric Kruskal-Wallis test, SHAM vs IDPN Spp1, p=0.0322; SHAM vs IDPN CASPR1, p=0.0268; Figure 3 supplement 3A1 and B1 for the peripheral zones) and in the otolith macula (Figure 3C2, Non parametric Kruskal Wallis test SHAM vs IDPN Spp1, p=0.0279; SHAM vs IDPN CASPR1, p=0.0343). However, the treatment did not alter type II HC-specific labelling as the number of Calre+ cells in the IDPN group was similar to the SHAM in all regions of either vestibular endorgans (Figure 3A2, canal crista, Calre: Non parametric Kruskal-Wallis test IDPN vs SHAM, p>0.9999; Figure 3C2, otolith macula, Calre: SHAM vs IDPN, p>0.9999). Importantly, the total number of HC marked with Myo7a (labelling both type I and type II HC) was not significantly different between the IDPN-treated mice and the SHAM, suggesting that the alteration of type I HC is not associated with cell loss. Moreover, at W12 the number of type I HC labelled in the IDPN-treated mice was no longer significantly different from the SHAM group in either endorgans (Figure 3A3 for crista; Figure 3C3 for macula). The count of type II-specific markers and of non-specific HC markers was not statistically different between SHAM and IDPN either, so that overall no difference persisted at W12, indicating a structural recovery at the end of the washout period.

+ + +Immuno-labelling of HC in the central regions of the horizontal SCC ampulla and striolar region of the utricule Macula. +

A, C) Immunolabelling of type I HC (Spp1+ and CASPR1+), type II HC (Calre +), or all HC (Myo7a) for the SHAM, IDPN W6 and IDPN W12 groups in the central ampulla of the horizontal canal (A1) and central utricular maculae (C1). Cell count at W6 and W12 in the central horizontal ampulla (A2 and A3) and central utricular maculae (C2 and C3) for individual mice. B1, D1) Individual number of central Spp1+ type I HC, CASPR1 + type I HC and Calre+ type II HC, or all HC(Myo7a) as a function of the aVOR gain (B1) or OCR gain (D1) at W6 (circle) and W12 (squares) groups. The linear regressions correspond to all individuals (n=23 mice). B2, D2) Comparison of the number of CASPR1 type I HC as a function of the aVOR (B2) gain or OCR gain (D2) at W6 (black and circle) and at W12 (pink and square) for each IDPN mice of the W12 group (n=8). Note that all points are shifted toward the regression line (redrawn from respectively B1 and D1), indicating that the number of cells at W12 better correlates with the recovered aVOR. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To determine how the loss of HC correlates with organ-specific functional tests, the numbers of type I and II HC in the central region of the horizontal semi-circular canals are plotted as a function of the mean aVOR gain for mice at W6 and W12 (Figure 3B1). Both markers of type I HC (CASPR1 and Spp1, left panels) were significantly correlated with the aVOR gain (linear regression, Spp1: r2=0.5128, p=0.0001; CASPR1: r2=0.4948, p=0.0002) whereas the rather constant number of type II HC in the ampulla did not correlate with the variation observed in aVOR function. Finally, to determine if the correlation was due to the recovery occurring between W6 and W12, the number of CASPR+ cells of IDPN mice counted at W12 are plotted as a function either of the individual aVOR gain at W12 (pink squares) or aVOR gain at W6 (black circles), paired by an arrow (Figure 3B2). The recovery of the vestibular function between W6 and W12 induces a shift toward the W12 linear regression, reinforcing the notion that the recovery of function was related to the increase in the number of hair cells with normal expression of biochemical markers.

+

The numbers of type I HC and II HC found in the striolar region of the utricule are similarly plotted in relation to the OCR gain (linear regression, Spp1: r2=0.6011, p<0.0001; CASPR1: r2=0.6342, p<0.0001, Figure 3D1). Again, the otolithic function correlated with the number of type I HC, and not type II HC in the central region. Furthermore, the correlation found between the otolithic function and the number of type I HC related to the increase in the number of hair cells after the recovery period (Figure 3C2). A similar correlation between organ-specific function and the number of type I HC was also found in the peripheral regions of these organs (see Figure 3 supplement 3C1, 3C2 for the SCC, and Figure 3 supplement 3D1, 3D2 for the utricule).

+

Taken together, comparison at W6 and W12 of non-specific HC and type I-specific HC markers suggest that the ototoxic effect induces a transient biochemical alteration of type I HC rather than a definitive hair cell degeneration. Overall, IDPN-induced alteration of canal and otolith functions are correlated with the ototoxic effect on type I, and not type II, HC in the central and peripheral zones of vestibular endorgans.

+ + +Effects of the subchronic treatment of IDPN on the optokinetic reflex +

To determine the effects of the subchronic treatment of IDPN on the optokinetic reflex (OKR), mice were tested with sinusoidal rotations of a virtual drum (Figure 4A). An example of raw traces of the evoked horizontal eye movements is shown in Figure 4B for different timepoints (0.5Hz at 10deg.s-1 optokinetic stimulation). The mean OKR gain for all tested frequencies is plotted in Figure 4C for SHAM (n=12) and IDPN-treated mice (n=12). The mean OKR gain of IDPN mice was significantly different from the mean OKR gain of SHAM mice at W8 (interaction between Weeks and Group F(6,132)=2.9845, p=0.0091; IDPN W0 vs W8, p=0.0216, see Table 2) where it reached its peak (IDPN W0 vs W8, p=0.00037). The gain stayed significantly larger compared to the SHAM through the end of the washout period (W12 IDPN vs SHAM, p=0.046).

+ + +Effects of subchronic IDPN treatment on the OKR. +

A) Illustration of the optokinetic reflex (OKR) set-up. B) Example of raw traces of an OKR recorded in response to stimulation at 0.5Hz at a peak velocity of 10°/s before (W0), after 4 weeks of IDPN treatment (W4), 2 weeks (W8) and 6 weeks of washout (W12). All traces are from the same individual. C) Mean OKR gain of IDPN (n=12) and SHAM (n=12) mice. D) OKR gain and (E) phase for IDPN (n=12) and SHAM (n=12) for each frequency at W0, W6, W8 and W12. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + + + +Statistics table of the OKR gain for the IDPN treated group. + + +

To determine whether this OKR modulation was frequency-specific, the gain of the 5 tested frequencies was compared for W0, W6, W8 and W12 (Figure 4D). There was at W0 no difference between SHAM and IDPN mice (Figure 4D, left panel, W0). At W6, there was a significant difference between the gain of the IDPN and SHAM groups limited to the frequencies 0.5 and 1Hz (W6 IDPN vs. SHAM, 0.5Hz, p=0.013117; W6 IDPN vs. SHAM, 1Hz, p<10-4). At W8, the gains measured at frequencies above 0.2Hz were significantly increased compared to the SHAM group (W8 IDPN vs. SHAM, 0.33Hz, p=0.027687; 0.5Hz, p=0.0025; 1Hz, p=0.00015). At W12, the gain remained high for both 0.5 and 1Hz whereas the gain at 0.33Hz was no longer significantly larger than the SHAM group (W12 IDPN vs. SHAM, 0.33Hz, ns; 0.5Hz, p=0.0069; 1Hz, p=0.00078). Globally, responses at higher frequencies (0.33, 0.5 and 1 Hz) were significantly increased (ANOVA Weeks x Group x Frequencies interaction, F(24, 528)=6.5870, p<10-4) whereas responses at the lowest frequencies (0.1 and 0.2Hz) were not significantly modulated. Additionally, these changes in gain for frequencies >0.2Hz were not accompanied by changes in the timing (phase) of the OKR (Figure 4E) (ANOVA, Weeks x Group x Frequencies interaction, p=0.3802).

+ + +Relation between VOR decrease and optokinetic increase +

To correlate the changes in vestibular pathway with the frequency-specific changes observed in the optokinetic pathway, we compared the responses in aVOR and OKR obtained at the frequencies that were common between the 2 tests (i.e 0.2 Hz, 0.5 Hz and 1Hz) for n=12 SHAM and n=12 IDPN treated mice by comparing the increase in the OKR and decrease in VOR relative to W0 (Δmean gain, see Figure 5). To determine whether the changes in VOR and OKR were proportional, we compared the paired decrease and increase in the mean ΔaVOR and ΔOKR gain after 8 weeks of treatment, when the difference between the 2 reflexes peaked (Figure 5A). The vast majority (n=11/12) of IDPN-treated mice showed a decrease in aVOR and an increase in OKR (top left quadrant corner, Figure 5B), which was not the case for SHAM mice (blue squares). However, there was no correlation between the amplitude of the VOR decrease and the amplitude of the OKR increase, i.e mice that had the greatest VOR loss did not show the largest OKR increase (slope in the regression line: 0.27, p=0.0834).

+ + +Comparison of the IDPN treatment on OKR and aVOR. +

A) Mean Δ aVOR and Δ Gain OKR for 0.2, 0.5 and 1Hz for IDPN (n=12) and SHAM (n=12) mice. B) Individual ΔOKR gains as a function of individual ΔaVOR gains. The 50% confidence interval of each group is represented in the shaded areas. C) aVOR and OKR gains of IDPN mice (n=12) at W0 and W8 for frequencies of 0.2, 0.5 and 1Hz. D) Percentage of the individual vestibular weight (inset), as a function of the percentage of the individual OKR gains change for IDPN (n=12). The linear regression corresponds to all values (n=36). (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +

To determine whether the parallel changes in OKR and VOR are frequency-specific, Figure 5C compares the gains measured at W0 and W8 for each frequency. At 0.2Hz, there was a significant decrease in aVOR (IDPN W0 vs. W8 0.2Hz p<10-4), however the OKR was not modified. On the other hand, at 0.5 and 1 Hz significant VOR decreases (IDPN W0 vs. W8, 0.5Hz, p<10-4; 1Hz, p<10-4) were accompanied by a significant OKR increase (IDPN W0 vs. W8, 0.5Hz, p<10-4; 1Hz, p<10-4). To examine this frequency-selective increase, the percentage of change in the OKR gain is plotted as a function of the vestibular weight, determined as the ratio between the aVOR and aVOR+OKR values at week 0. The vestibular weight therefore represents the frequency-dependent relative influence of the vestibular signal on gaze stabilization. Figure 5D shows that at week 8, the change in the OKR occurred at the frequencies for which the vestibular weight is dominant (>50%), and that the increase in OKR was positively correlated with the vestibular weight among frequencies (slope of the regression line: 1.235, r²=0.6299, p<0.0001). Overall, these results suggest that the increase in OKR gain observed at high frequencies could correspond to a “visual substitution” occurring primarily in the range where the vestibular inputs are normally dominating gaze stabilization.

+ + +Visuo-vestibular interactions following alteration of vestibular inputs +

To investigate whether OKR increase did constitute a “visual substitution” that maintained optimal gaze stabilization at light despite vestibular loss, we investigated how IDPN-treated mice integrated vestibular and visual inputs. To this end, a session of combined visual and vestibular stimulation (aVOR in light, here referred as Combined Gaze Response or CGR condition) was performed at W6 and at W12 (n=19 IDPN; n=12 SHAM). A model (see methods) was implemented to predict the theoretical CGR (visuo-vestibular) gain and phase from aVOR and OKR (unimodal) measured values. The model was first optimized to correctly predict the average SHAM CGR gains and phases (Figure 6A) from their individual aVOR and OKR responses, and then applied to the individual IDPN data. The predicted and observed CGR gain (Model factor) were compared for each treatment group (ANOVA Group x Model, F(1,29)= 12.236, p=0.0013): as expected, model predictions for SHAM well-matched experimental data (Post-hoc Newman-keuls, SHAM MODEL vs SHAM CGR, p=0.41). When applied to IDPN mice the model predicted that their CGR gain should be lower compared to SHAM (shaded blue and red areas on Figures 6A and 6B, Post-hoc Newman-keuls, SHAM MODEL vs IDPN MODEL, p=0.00176), suggesting that even an optimal combination of unimodal responses could not restore completely normal CGR in IDPN-treated mice. The CGR gains observed for IDPN-treated mice actually revealed how their visuo-vestibular responses were not only reduced compared to SHAM (compare solid blue and red lines in Figure 6A-B, Post-hoc Newman-keuls, SHAM CGR vs IDPN CGR, p=0.00055), but were even inferior to the prediction of the model (compare solid red line and shaded red area in Figure 6B, Post-hoc Newman-keuls, IDPN MODEL vs IDPN CGR, p=0.00041). This suggests that the IDPN-treated mice sub-optimally combined their residual/compensated vestibular and visual reflexes to stabilize gaze. Figure 6C illustrates the relationship between the measured CGR gain and the prediction of the model for all tested frequencies. For SHAM animals, CGR gain values were positioned along the unity line that represents a close match between predicted and measured CGR (mean±SEM: SHAM MODEL 0.8255±0.053; SHAM CGR 0.8375±0.042). For IDPN-treated mice, the responses formed 2 subgroups that did not correspond to different frequencies (all frequencies found in either subgroup). The first subgroup was intermingled with the SHAM responses, with measured CGR values slightly lower than predicted values (MODEL 0.7457±0.028; CGR 0.6644±0.052). The second subgroup represented mice for which the measured CGR significantly underperformed the predicted CGR (MODEL 0.4685±0.022; CGR 0.2005±0.01). To investigate the difference between the two subgroups, the VOR, OKR and CGR values were plotted for the n=19 IDPN mice (Figure 6D). A hierarchical cluster analysis (see methods) suggested the presence of two main clusters that differ from the clusters observed considering only the aVOR gain (see dendrogram in Figure 6D). The first cluster (Δ+, n=12; in green in Figure 6D, E, and F) comprised individuals with CGR values higher than OKR (positive values on Δ=CGR-OKR), i.e with gaze stabilization at light better in the presence of vestibular stimulation. The second cluster (Δ-, n=7; depicted in orange) is composed of individuals with low CGR (<0.4). IDPN mice from the Δ-cluster have degraded responses during bimodal visuo-vestibular stimulation compared to unimodal OKR stimulation (negative value of Δ indicates CGR responses lower than OKR responses). Notably, this degraded CGR is not solely a consequence of a low VOR gain, as some Δ+ mice identified by the second differentiation of the clustering analyses (depicted with grey diamond symbols in Figure 6D, and indicated as Δ+(-)) also showed low unimodal VOR gain responses.

+ + +Visuo-Vestibular interactions following IDPN treatment. +

A) aVOR, OKR, CGR data (CGR) and predicted CGR (MODEL) gains (left panel), and CGR phase (data and model, right panel), for the SHAM mice (n=12) at W6 and W12. B) aVOR, OKR, CGR data (CGR) and predicted CGR (MODEL) gains for the IDPN mice (n=19) at W6 and W12. C) Comparison of the predicted versus measured CGR for all frequencies tested. The dotted line at 45° represents a perfect match between prediction and data (optimal CGR). D) aVOR, OKR, and CGR values of the IDPN mice (n=19) (left panel). Clustering analysis (right panel) distinguish 2 groups based on Delta (CGR-OKR), and a subgroup based on VOR. The horizontal coordinate of each cluster (vertical lines) represents the distance between two connected clusters. E) Δ(CGR-OKR) as a function of the VOR Sweep Amplitude of the Δ+ (green triangle), Δ+(-) (green diamonds) and Δ- (orange) IDPN mice. The inset panel represents typical raw cycles representative of the 3 subgroups. F) Delta (CGR-OKR) as a function of the VOR reproducibility of the Δ+ (green triangle), Δ+(-) (green diamonds) and Δ- (orange) IDPN mice. Regression line correspond to all values (n=19)

+ + +

We reasoned that the worsening of the gaze responses in the bimodal condition compared to the OKR-only condition could be a consequence of a poorly reliable, noisy vestibular signal. Based on a cycle-to-cycle sweep analysis, two additional features of the VOR responses were quantified (see methods): the mean amplitude of the raw sweep during VOR stimulation and the VOR response reproducibility between sweeps. The mean VOR raw sweep amplitude could not statistically differentiate Δ- and Δ+(-) mice (Figure 6E, orange triangle vs green diamonds p=0.0857), suggesting that vestibular signal of comparable amplitude improved or deteriorated the CGR response in the Δ+(-) and Δ-mice, respectively. On the other hand, the VOR reproducibility clearly discriminated Δ- and Δ+(-) mice (Figure 6F, orange triangle vs green diamonds, p=0.0001). There was a strong correlation between the inter-sweep reproducibility and the capacity of the mice to stabilize gaze in presence of vestibular inputs at light. This result suggests that the incapacity of the Δ-group to compute optokinetic signals in the presence of vestibular stimulation could relate to the poor reliability of the vestibular signal. Overall, these data suggest that in some situations the presence of a degraded vestibular signal of significant amplitude (panel 6E) but poorly reliable (panel 6F) can be detrimental to properly stabilize gaze in presence of visual and vestibular inputs.

+ + + +Discussion + +IDPN as a model of partial and transitory loss of vestibular function +

IDPN has long been established as an ototoxic compound targeting vestibular HC in the different vestibular endorgans of rats (Llorens et al., 1993, 1994; Llorens and Demêmes, 1994) and later in guinea-pig, frogs and mice (Soler-Martin et al., 2007; Greguske et al., 2019). Its ototoxic vestibular effects were used as a tool to study extraocular muscle development (Brueckner et al., 1999) or more recently to induce permanent vestibular loss in mice (Yang et al., 2019; Zeng et al., 2020) and study hair cell regeneration (Sayyid et al., 2019). The sub-chronic, reversible protocol used in our study was validated previously in rats (Sedó-Cabezón et al., 2015; Martins-Lopes et al., 2019; Maroto et al., 2021) and mice (Boadas-Vaello et al., 2017; Greguske et al., 2019) with postural/locomotor quantification of vestibular loss. We for the first time demonstrate that the subchronic protocol leads to a progressive and partly reversible loss of vestibulo-ocular reflexes. These previous studies further demonstrated some of the cellular mechanisms associated with the progressive loss of postural control: ototoxic effects lead to the early dismantlement of calyceal junction, followed by synaptic uncoupling, both of which were shown to be reversible, while continuation of the IDPN treatment would lead to hair cell extrusion and long-term, permanent lesion. HC loss was also demonstrated to occur in a central to peripheral order within vestibular epithelia, and in crista to utricule to saccule order (see Sedó-Cabezon et al. 2015). Maroto et al (2021) convincingly demonstrated that type I HC show greater sensitivity than type II HC to IDPN subchronic exposure. Given these data, a primary goal of the present study was to try to correlate the loss of VOR functions to organ-specific, zone-specific, and cell-type-specific effects.

+

Loss and substantial recovery of the vestibular function quantified by VOR measures were found to be correlated with the number of type I HC in both canals and otoliths. A parallel was previously established between HC integrity and VOR following ototoxic protocols in mice (Cassel et al., 2019; Yang et al., 2019; Zeng et al., 2020), but these were established at the population level. Here, we for the first time specifically correlated the loss of HC and loss of VOR function both at the individual level and in an organ-specific way. aVOR and OCR tests demonstrated a parallel decrease in canal- and otolith-dependent functions, respectively. The 6 week-long treatment was followed by a 6 week-washout period, allowing a significant but partial recovery of the aVOR, and complete recovery of the OCR. The individual correlations between these functional tests and the number of HC counted in the ampullae and maculae of the mice were found to be particularly significant, both during the trough (W6) of VOR function and after the recovery period (W12) (Figure 3). We, however, did not find any difference between the HC loss in the central vs. peripheral zones of the organs, and can therefore not conclude on any putatively differential implication of these areas in the VORs. Effects of the IDPN treatment on the aVOR was evidenced by a gain decrease starting W4, later associated with a significant phase lead at W6 (Figure 1 supplement 1A,1B). The dynamic of these changes might be related to the amount of HC progressively affected by the treatment. The modifications affected all frequencies; however, responses at the lowest tested frequency (0.2Hz) tended to be less affected and to recover first. These results could suggest that type I HC are crucial element to encode both the amplitude and timing of the VOR, particularly for the more dynamic stimulations. How IDPN progressively impairs the encoding of information by the HC, and how population-coding influence both VOR features in the natural range of head-movements should be the focus of future dedicated studies.

+

Overall, these organ-specific structure-function correlations confirm that the vestibulo-ocular reflex can serve as a proxy indicating the status of the vestibular endorgans. It further validates the use of the aVOR and OCR as relevant tests reflecting HC integrity at the level of the ampullae and maculae, respectively.

+ + +Differential alteration of type I versus type II hair cells +

The subchronic treatment affected type I HC more than type II HC, in accordance with previous reports (Llorens et al. 1993; Maroto et al. 2021). Susceptibility to IDPN was also reported to vary between different mouse strains (Boadas-Vaello et al., 2017; Wilkerson et al., 2018). Interestingly, we used a different strain of mice than the one used in a previous subchronic experiment (Greguske et al. 2019) and have not observed any type II HC loss either, nor differences between males and females (data not shown).

+

Hence, IDPN treatment induces a loss of cell markers specific to type I HC, with no effects on type II markers. Two types of markers were used to identify type I HC. Spp1 targeted a protein located on the neck of the hair cell (McInturff et al., 2018), while CASPR1 is located at the calyceal junction of the afferent terminal, and has been proven to be necessary for the functionality of the synapse between the HC and its connected afferent (Sousa et al., 2009).

+

Though it is not possible to compare for one individual the number of cells before and after treatment, the number of cells labelled for either of these proteins decreased significantly compared to SHAM group, whose numbers are consistent between W6 and W12. The loss of CASPR1 marker induced by IDPN sub-chronic treatment has been linked to HC detachment from the calyx terminal and loss of vestibulo-spinal reflexes, and both CASPR1 expression and vestibular function recover during washout (Sedó-Cabezón et al., 2015; Greguske et al., 2019; Maroto et al., 2021). Only after a longer IDPN exposure does HC extrusion occur, associated with persistent functional deficits. In theory, the loss of vestibular function observed in both canal- and otolith-dependent VOR at W6 could be linked to loss of type I HC. In adult rodents, however, the regeneration of hair cells leads to the formation of cells with type II HC features (González-Garrido et al., 2021), so the recovery of type I cellular markers at W12 is likely not a result of cell regeneration. Also, the apparent dramatic loss of CASPR1 and Spp1 HC (90%) at W6 was in contrast with the much smaller and not significant loss in Myo7a cells (29%), supporting the conclusion that the sensory cells persisted in the epithelium despite CASPR1 loss, and that the recovery in the number of type I HC observed at W12 was more likely the result of molecular repair, not cell regeneration. In addition, we observed a similar loss of Spp1 marker at W6, normally located in the neck of the hair cell. As such the loss of vestibular function could rather be attributed to a global disorganization of the synapse and defective hair cell function, associated with a transitory large decrease of the Spp1 and CASPR1 proteins and a smaller decrease in Myo7a. In accordance with this hypothesis, the number of Spp1 and CASPR1 positive cells after the washout period at W12 was not significantly different from the SHAM group and correlated to the recovery of the vestibular function observed at W12 for the IDPN mice (Figure 3C2). The altered expression of proteins in type I HC induced by the sub-chronic treatment of IDPN seems to be reversible in most individuals. One possibility is that definitive extrusion of type I HC occurred in the most susceptible individuals that did not show recovery after the end of treatment. This in fact was the case in the 2 individuals tested at W12 with immuno-histochemistry: CASPR1 mean value (±S.E.M) of 18.5±5.5 compared to 34.5±4.359 for IDPN with recovery and 41.75±1.215 for SHAM). It was previously reported in the pigeon that type I HC and calyx afferents take longer (12 weeks) to regenerate following aminoglycoside toxicity, while type II and boutons endings regenerated in a week (Zakir and Dickman, 2006). Similarly, recovered innervation for type I HC was delayed compared to type II HC in mice (Kim et al., 2022). A longer recovery period could be investigated in the case of the severely-affected mice to confirm the absence of recovery in the longer term.

+ + +Evidence for the role of type I HC in the vestibulo-ocular pathways +

To our knowledge, the direct implication of type I and type II HC in the VOR was never directly tested. Type I and type II HC differ by many features, including their anatomical location within the epithelium, electrophysiological properties, morphology and innervation by afferents (Eatock and Songer, 2011). Seemingly, irregular afferents have been described as predominantly innervating the central area and striolar zones of the ampullae and maculae with calyx and dimorphic synaptic contacts, while regular afferents make dimorphic and buttons contacts predominantly within the peripheral zones of cristae and extrastriolar zones of maculae (Goldberg, 2000; Eatock and Songer, 2011; Contini et al., 2022). Afferents with bouton terminals that contact only type II HC have regular discharge, while afferents with calyx terminals that contact only type I HC have irregular discharge. However, most regular and irregular afferents are fed by both type I and type II HC (Goldberg et al., 1992). Functionally, regular afferents recorded in the monkey show a lower detection threshold than irregular afferents (Sadeghi et al., 2007), while irregular afferents showing higher gains and phase leads (mouse: Lasker et al., 2008; Cullen, 2019) would be better optimized for encoding high dynamic stimuli (Cullen, 2019). Central VOR neurons receive a mixture of regular and irregular afferent inputs (Highstein et al., 1987; Boyle et al., 1992; Goldberg, 2000), with irregular afferents constituting ∼1/3 of their excitatory drive (Highstein et al., 1987; Boyle et al., 1991).

+

While the role of type I and type II HC in the VOR remains to be determined, previous studies have emphasized the importance of the regular afferents and regular central vestibular neurons in the vestibulo-ocular pathway. Functional ablation targeting irregular afferents suggested that VOR might function normally with only intermediate and regular afferent inputs (Minor and Goldberg, 1991). Functionally identified VOR neurons (i.e. PVP) were recently demonstrated to have heterogeneous discharge variability (high or low), with the most regular units particularly well-suited to faithfully encode the compensatory eye movements generated during natural stimulation (Mackrous et al., 2020).

+

Based on this literature, it would be tempting to infer that only the most regular and tonic elements of the entire vestibular pathway are responsible for the VOR. However, the correspondence between hair cell properties and afferents/central neurons properties is only partial, such that the two phasic/irregular and tonic/regular channels for head motion signals are constituted of both types I and type II HC (Baird et al., 1988; Goldberg et al., 1990 p.90) and afferents (irregular and regular) (Goldberg et al. 2000; Eatock and Songer 2011; Beraneck and Straka, 2011). Carey and colleagues (1996) previously reported a better correlation of VOR recovery with type I than with type II HC following ototoxic lesions in the chick. Overall, the result of ototoxic studies, including the present one, demonstrate a fundamental role of type I HC in the encoding of vestibular signals that drive the vestibulo-ocular reflexes, even in the relatively low range (Carriot et al., 2017) of head movements tested. It has also been recently shown that both type I and type II HC actually contribute to otolithic vestibular evoked potential responses (i.e. vestibulo-spinal pathway), previously described as mostly type I-specific (Sayyid et al, 2019). Overall, our results are compatible with the hypothesis of a convergence of heterogeneous peripheral neural elements at the level of central vestibular nuclei, where intrinsic properties of central vestibular neurons (Straka et al., 2005; Beraneck and Idoux, 2012) supplemented by network properties (Beraneck et al., 2007; Pfanzelt et al., 2008; Beraneck and Straka, 2011) would differentially integrate vestibular signals further processed in the different functional pathways (Sadeghi and Beraneck, 2020; Mackrous et al. 2020).

+ + +Visuo-vestibular interactions after IDPN treatment +

OKR plasticity following vestibular loss concerned gain, and not phase, modifications for frequencies >0.33Hz. Previous studies have reported an increase in OKR gain following a permanent vestibular lesion for a non-specific range of frequencies (Shinder et al., 2005; Faulstich et al., 2006; Nelson et al., 2017). One key to understand the frequency-specific OKR plasticity could be the physiological dominance of vestibular inputs in the gaze stabilization process at high frequencies (Faulstich et al., 2004). The visual inputs could be reweighted and potentiated specifically in the range where the vestibular loss has the most impact on gaze stabilization (Figure 5D).

+

The VOR and OKR work synergistically to stabilize gaze by compensating for head and visual surround movements, respectively (França de Barros et al., 2020). If the reflexive eye movements are not perfectly compensatory, an error signal (e.g retinal slip) is produced that drives adaptation of the VOR (Boyden et al., 2004; Dean and Porrill, 2014; Shin et al., 2014) and OKR (Glasauer, 2007; Kodama and du Lac, 2016). In an effort to understand the integration of VOR and OKR, Holland and colleagues (2020) recently proposed that OKR would account for the retinal slip not compensated by VOR. Within that framework, the changes in the OKR could happen preferentially as a function of the vestibular weight, i.e in the range of movements where the loss of vestibular inputs generates the largest retinal slip. While present data are evidence for frequency-specific adaptation, how this relates to alteration in vestibular inputs and/or visual feedback signals remains to be determined.

+

The increase in OKR gain could be seen as a substitution for the decrease in the VOR. However, even after VOR recovery, the optimal integration of OKR and VOR (as predicted by our MODEL) led to a degraded combined gaze response (CGR) with respect to SHAM mice. Even worse, the measured CGR was inferior to the predicted CGR (Figure 6B). This observation suggests that the integration of visual and vestibular signals during CGR is more than a mere summation of the gain and phase of the 2 unimodal reflexes, but is also affected by other factors. In fact, a subset of the IDPN individuals showed severely degraded bimodal responses, so that their combination of VOR and OKR is not only sub-optimal but also less effective than the unimodal (OKR) reflex. A sweep-based analysis suggested that in these individuals, the VOR unimodal responses were not reliable, i.e the low reproducibility indicates that the same stimuli are unfaithfully encoded into eye movements. We interpret this low reproducibility as a sign of noisiness within the VOR pathway that would not only preclude the optimal integration of both sensory signals, but also disturb the use of retinal information to stabilize the gaze. According to statistical optimality theories, as the Maximum Likelihood principle, the decrease in VOR reliability should lead to a reduction in its relative sensory weight. This filtering-out of a degraded signal has been shown during visuo-vestibular integration for monkey heading perception (Fetsch et al., 2010), or for human object discrimination using visual and haptic senses (Ernst and Banks, 2002).Our results suggest that the gaze stabilizing system is not able to optimally adapt to the degradation by filtering-out vestibular signals during combined visuo-vestibular combination.

+

The integration of visual and vestibular inputs occurs in several structures, including in the cerebellar flocculi (Jang et al., 2020) and brainstem (Carcaud et al., 2017) (see review De Zeeuw et al., 2021). In the case of vestibular loss, most of the defects are expected to concern the central vestibular neurons involved in the VOR, which also integrate visual inputs (ES, or eye-sensitive neurons; Beraneck and Cullen, 2007). After unilateral neurectomy, vestibular neurons in the Deiters nuclei responded to higher frequencies during visual stimulation (Cat: Zennou-Azogui et al., 1994), a change compatible with the hypothesis that OKR gain increase partly takes place in the vestibular nuclei. It was shown that inhibitory floccular inputs (part of the OKR indirect pathway) and excitatory vestibular inputs often colocalized on the dendrites of central vestibular neurons. One possibility is therefore that the massive disorganization of vestibular periphery inputs led in the long term to synaptic and intrinsic changes at the level of central vestibular neurons (Beraneck and Idoux 2012; Carcaud et al. 2017), thus impairing their capacity to optimally integrate both signals.

+ + +Conclusions +

Balance dysfunction occurs frequently in aging people (85% prevalence above 80; Agrawal et al., 2013) but also in younger people (prevalence of 35% in the 40 year-old). While animal models often use permanent vestibular lesions (Simon et al., 2020), many diseases in fact consist in a gradual, transient and/or partial loss of vestibular function. For instance Menière’s disease, which represents ∼9% of all vestibular pathologies in the adult and occurs in the <60 years old, is characterized by recurrent episodes with brief (<24h) fluctuating symptoms and otherwise normal vestibular function before long-term deterioration arises (Lopez-Escamez et al., 2015). Transitory vestibular symptoms are also commonly reported in vestibular neuritis where, although symptoms tend to rapidly disappear due to vestibular compensation, vestibular function can recuperate up to a year after the initial loss (Welgampola et al., 2019). Gradual and partial vestibular loss is also encountered as a side-effect of some ototoxic anti-cancer treatments (prevalence in treated patients ∼40%; cisplatin; Paken et al., 2016; Prayuenyong et al., 2018). Here, we took advantage of a subchronic ototoxic protocol to determine how animals adapt to partial and transitory loss of vestibular hair cells. We show that the loss, and then the partial recovery of the VOR, is correlated with the integrity of the type I HC, demonstrating for the first time their essential role in the VOR, whether of canalar or otolithic origin. We show that sensory (visual) substitution would theoretically compensate for vestibular loss, but that injured mice have suboptimal responses when combining visual and vestibular information. Finally, we show that this impairment in multisensory integration would be linked to the loss of "reliability" of the vestibular signal, degraded by ototoxicity. Overall, these results suggest that transitory peripheral infraction have long term consequences, and that the capacity of central vestibular structures to cope (vestibular compensation; Cullen et al., 2009; Beraneck and Idoux, 2012;(Lacour et al., 2016) with the sensorineural loss might critically depend on the integrity of the neural elements involved. Future studies should aim at obtaining information about the degradation of signal transmission following IDPN treatment and characterize the amount of peripheral population-coding necessary to preserve optimal vestibular function.

+ + + +Material & Methods + +Ethics +

A total of 56 C57/BL6J mice in an equal partition of males and females were used for the protocol. Animals were used in accordance with the European Communities Council Directive 2010/63/EU. All efforts were made to minimize suffering and reduce the number of animals included in this study. All procedures were approved by the ethical committee for animal research of the Université Paris Cité.

+ + +Headpost implantation surgery and animal care +

Surgical procedures, postoperative care, device fixation and animal surveillance during the protocol were performed as described previously in França de Barros et al., 2019, 2020. Briefly, 6-weeks-old mice anaesthetized with isoflurane gas had their heads shaved with small clippers. Then, lidocaine hydrochloride (2%; 2mg/Kg) was injected locally before a longitudinal incision of 1.5 cm was made to expose the skull. Just anterior to the lambda landmark, a small custom-built headpost (3x3x5mm; poly lactic acid) was cemented (C&B Metabond; Parkell Inc., Edgewood, NY) and laterally covered with resin (Heraeus) for protection. Animals were fully recovered 30 min after the end of the surgery, yet buprenorphine (0.05mg/kg) was provided for postoperative analgesia and they were closely monitored for the following 48 hours.

+ + +Subchronic ototoxic exposure +

28 mice were treated with 3,3ʹ-iminodiproprionitrile (IDPN, Sigma Aldrich, 317306, 90%) for 6 weeks (IDPN group) at 30mM concentration of IDPN in their drinking water at least 72h after the surgery. After 6 weeks of treatment, a washout period of 6 weeks followed. Previous experiments (Greguske et al., 2021) had demonstrated that at these concentrations, ototoxic lesions produced by IDPN are largely reversible. The SHAM group was tested the exact same way but was not exposed to IDPN in their drinking water. Video-oculography tests were performed before the beginning of the treatment and once every two weeks until week 12 (W12), for a total of 7 sessions of tests for each mouse.

+ + +Experimental groups +

3 different batches of mice were used, each composed of both IDPN-treated mice and SHAM. The mice of the first one (n=19) were subjected only to vestibular stimulations, to test the effect of IDPN on the canal-related or otolith-related VOR responses during the 12 weeks of protocol. The second group (n=24), designed to compare the dynamic of VOR and OKR changes was tested with angular VOR stimulations, otolithic test, optokinetic (visual) stimulations, and combined vestibular and visual stimulation (VOR in light; referred to as Combined Gaze Response or CGR). The third group (n=11) designed to correlate ototoxic effects of IDPN on vestibular hair cells (HC) and VOR function was tested for canal-related or otolith-related VOR responses, before treatment and after 6 weeks of treatment and immediately used for immunohistochemical assessment of the vestibular sensory epithelia (see below). A subset of the second group (n=12 out of 24) was also used for immunohistochemical assessment at the end of the 12 weeks of protocol.

+ + +Video oculography recording sessions +

The recording procedure was similar to the one presented in Carcaud et al. 2017 & França de Barros et al. 2020. The animals were head-fixed in a custom build Plexiglas tube, and their head was oriented in a 30° nose down position to align the horizontal canals to the yaw plane. Their left eye movements were recorded using a non-invasive video oculography system (ETL -200 Iscan, acquisition rate 120Hz, Stahl et al 2000). Eye, head, and virtual drum (OKR stimulation) position signals were digitally recorded (CED mk3 power 1401, acquisition rate 1kHz) with Spike 2 software. Signals were analysed offline in Matlab (Matlab, the MathWorks, Natick, MA, USA; RRID, SCR:001622) programming environment. The restraining apparatus was fixed on a rotation platform on top of an extended rig with a servo-controlled motor. Single VOR or OKR recording sessions lasted no longer than 45 minutes in total.

+ + +Vestibular stimulations and analysis +

All vestibular-specific tests were performed in a temperature-controlled room with all sources of light turned off except for computer screens. The turntable was further surrounded with a closed box to isolate the animal from remaining light, with a final luminance inside the box <0.02 lux. Myosis was induced with topical 2% pilocarpine applied 10 minutes before experimentation.

+

Vestibulo ocular reflex in dark (VORd) tests were performed in the dark with the mouse surrounded by an opaque black dome (Fig1.A). Sinusoidal angular rotations around the vertical axis were used to record the horizontal angular vestibulo-ocular reflex (aVOR), at different frequencies: 0.2, 0.5, 0.8; 1 and 2Hz at a peak velocity of 30°/s.

+

Angular vestibulo-ocular reflex analysis was similar to the one described in Carcaud et al. 2017. Segments of data with saccades were excluded from VOR slow-phase analysis. For horizontal sinusoidal rotations, at least 10 cycles were analyzed for each frequency. VOR gain and phase were determined by the least-squares optimization of the equation: + + + +where EHv(t) is eye horizontal velocity, g (gain) is a constant value, HHv (t) is head horizontal velocity, td is the dynamic lag time (in msec) of the eye movement with respect to the head movement, and Cte is an offset. The td was used to calculate the corresponding phase (φ°) of eye velocity relative to head velocity. The Variance-Accounted-For (VAF) of each fit was computed as + + + +where var represents variance, est represents the modeled eye velocity, and EHv represents the actual eye horizontal velocity. VAF values for VOR measures were typically between 0.70 – 1 (>95% of recordings), where a VAF of 1 indicates a perfect fit to the data. For IDPN-treated mice, abnormally low (<0.10) values of gain associated with VAF<0.5 were nevertheless included in the gain statistical analysis but specifically reported in grey (Figure 2 & Figure 1 supplement). Corresponding phase values were not included in the statistical analysis of the aVOR phase.

+

Static Ocular Counter-roll reflex tests were performed as described in (Simon et al., 2021) by measuring the vertical eye movement generated in response to different roll angles. The table was moved from left to right in incremental steps of 10° (from 0 to 40°), with static periods of at least 10s between oscillations (Fig1.D) to record the stabilized eye elevation and declination. The OCR gain corresponds to the slope of the linear regression of the vertical eye angle vs. the head tilt angles (Oommen and Stahl, 2008).

+

Off vertical axis rotation (OVAR) tests were performed with the vestibular turntable tilted with a 17° off-axis angle following the methodology described in Beraneck et al., 2012 & Idoux et al., 2018. 50°/s continuous stimulations were performed in a counter-clockwise and then clockwise direction. OVAR generates a continuous change in the head tilt angle, compensated through a maculo-ocular reflex (MOR) by the generation of a horizontal nystagmus compensating for the table constant rotation. This oculomotor response was quantified following the methodology described in Idoux et al 2018. First, quick-phases were identified and removed. During rotations, the velocity of horizontal slow phases is modulated (modulation, μ) around a constant bias (β). Both parameters (μ and β) were calculated from the sinusoidal fit of eye horizontal slow-phase velocity using the least-squares optimization of the equation: + + + +where SP(t) is slow-phase velocity, β is the steady-state bias slow phase velocity, μ is the modulation of eye velocity, f0 is the frequency of table rotation, td is the dynamic lag time (in msec) of the eye movement with respect to the head movement. The bias (Maculo-ocular reflex Bias) is reported here as the main index of the MOR response (Hess and Dieringer, 1990; Beraneck et al., 2012).

+ + +Optokinetic reflex tests and analysis +

Horizontal optokinetic stimulations were performed as previously described in França de Barros et al. 2020. The mice were placed under a semi-opaque plastic dome and all sources of light were turned off. The projected stimulation consisted of a randomly distributed white dots pattern on a black background image (250000 white dots, max width 0.075°). The optokinetic sinusoidal stimulations were tested at 0.1, 0.2, 0.33, 0.5 and 1Hz at a peak velocity of 10°/s. The gain and phase were obtained by the same least-squares optimization method described above for the aVOR. To prevent putative cross effects between visual and vestibular stimulations, VOR and OKR test sessions were performed on separate days.

+ + +Combined visual and vestibular stimulations +

Combined visual and vestibular stimulations measuring the combined gaze reflex (CGR) consisted of horizontal vestibular stimulations while projecting the fixed dotted pattern used for OKR on the surrounding dome. Horizontal angular sinusoidal rotations were performed at frequencies of 0.2, 0.5 and 1Hz with a peak velocity of 30°/s. The CGR gain and phase quantifications were performed following the same methodology as for aVOR. To avoid interference, these were performed on n=8 SHAM and n=12 IDPN after the end of the VOR and OKR test sessions on W12. To test the effects at W6, a specific group of mice (n= 4 SHAM and n=7 IDPN) was tested at W0 and W6.

+

A simple model has been developed to predict for each individual and for three stimulation frequencies (0.2, 0.5 and 1Hz) the theoretical CGR response gain and phase were given the same two parameters observed for their unimodal VOR and OKR responses. The experimental VOR and OKR gain and phase are used to compute the sinusoid representing the two unimodal responses. The gain and phase of the CGR sinusoid resulting from adding the VOR and OKR sinusoids are quantified by measuring its peak-to-peak amplitude and by identifying the time-lag that maximizes its cross-correlation with the sinusoidal stimulus (note that the sum of two sinusoids with the same period, even if with different phases, is always a sinusoid). Independently of the mouse, the unimodal VOR and OKR sinusoids are shifted in time before their summation, in order to improve, on average, the fit between predicted and observed CGR responses for SHAM mice. The two optimal shift parameters, one for VOR and one for OKR, obtained for each frequency, are then used also for all IDPN-treated mice.

+

To identify subpopulations of IDPN-treated mice, we computed an agglomerative hierarchical cluster tree on a dataset composed of individual aVOR gain and Δ (calculated as CGR-OKR gain). The classification method used unweighted average Euclidean distance between clusters.

+

To describe the mice VOR responses without any assumption about their sinusoidal nature the two following characteristics have been quantified: the reproducibility of the ocular response to the vestibular stimulation and the amplitude of the raw eye movements generated by the vestibular stimulation. These analyses were performed on the slow phases horizontal eye trajectories, EHp(t), recorded during each sweep of stimulation.

+

The reproducibility index, Rep, was obtained by computing the matrix of correlation coefficient, Ri,j, between each couple (i,j), of the N selected sweeps, and then computing the mean of all above-diagonal elements of the matrix as reported in the following equation (this choice aims at considering only once each Ri,j, since Rj,i=Ri,j and to exclude diagonal elements, Ri,i =1). + + + +The amplitude, Amp, of the raw eye movements was quantified by computing for each sweep how much on average, during a stimulation cycle, the eye moved horizontally away from its average position (σEHp). Before averaging this parameter over all the sweep, it had to be squared. Finally, to have a parameter value in degrees, the root mean squared of the mean was computed. + + + +

+ + +Immunolabelling of the vestibular HC +

Two groups of mice (n=11 at W6 and n=12 at W12) were used to perform immunofluorescence analysis on hair cells in the vestibular endorgans. Mice were anaesthetized with an overdose of intraperitoneal injection of ketamine hydrochloride (10%)/xylazine (5%) and decapitated. The histology was done following the protocol established by Lysakowski et al. (2011), as described previously (Maroto et al., 2021). The vestibular epithelia (one horizontal canal and one utricule) were dissected and fixed for 1h in a 4% solution of paraformaldehyde (PFA). PFA was washed twice with phosphate-buffered saline (PBS) and the samples were then embedded in a cryoprotective solution (34.5% glycerol, 30% ethylene glycol, 20% PBS, 15.5% distilled water) to be stored at -20°. Before the immunochemistry, samples were put at room temperature and rinsed twice in PBS. While under slow agitation, the samples were incubated twice, first for 1h with a 4% Triton X-100 (Sigma Aldrich) in PBS to permeabilise and a second time for 1h in 0.5% Triton X-100 1% fish gelatin (CAS #9000-70-8, Sigma-Aldrich) in PBS to block. The incubation with the primary antibodies was then performed in 0.1M Triton X-100, 1% fish gelatin in PBS at 4° for 24h. After rinsing, the second antibodies were incubated in the same conditions. The 2nd antibodies were rinsed and the vestibular epithelia were mounted on slides with fluoromount (F4680, Sigma-Aldrich) and were visualised with a confocal microscope Zeiss LSM880 (with an objective of 63x NA:1.4). To properly analyse the whole vestibular epithelium, Z-stacks of 0.5 μm were obtained and observed with ImageJ (National Institute of Mental Health, Bethesda, Malyland, USA). The primary antibodies used were rabbit anti-Myosin VIIa (Myo7a) (Proteus Biosciences, #25-6790), mouse anti-contactin-associated protein (CASPR1) (Neuromab #75-001), guinea pig anti calretinin (Synaptic Systems #214-104) and goat anti-osteopontin (Spp1) (R&D Systems #AF08). Their respective secondary antibodies were Dylight 405 donkey anti-rabbit igG H+L (Jackson Immuno Research #711-475-152, Alexa Fluor 488 donkey anti-mouse IgG H+L (Life Technologies #A21202), Alexa Fluor donkey anti-guinea-pig IgG H+L (Jackson ImmunoResearch #706-605-148) and Alexa Fluor 555 donkey anti-goat IgG H+L (Invitrogen #A21432).

+

HC counts were obtained from square areas (67.5 x 67.5 µm) of the central and peripheral parts of two vestibular organs (horizontal semi-circular canal and utricle) in the W6 group and the W12 group. The areas for counting were obtained from the same location within the epithelium in all animals. The global number of HC was assessed with the cytoplasmic labelling of the anti-Myo7a antibody (Hasson et al., 1997). Type I HC were labelled using two different antibodies: anti-Spp1 in the neck of the type I HC (McInturff et al., 2018) and the presence of anti-CASPR1 in the calyceal junctions of the calyx (Sousa et al., 2009 p.20). Type II HC were distinguished by the colocalization of Myo7a and calretinin (Maroto et al, 2021).

+ + +Statistical analysis +

For both OKR and VOR stimulations, the effect of the protocol on the gain and phase were statistically tested by performing repeated measures ANOVA. The treatment (SHAM or IDPN) was considered as between individual independent factor with the Weeks (W0, W2, W4, W6, W8, W10, W12) and the Frequencies (0.1, 0.2, 0.33, 0.5 and 1Hz for the OKR; 0.2, 0.5, 0.8, 1 and 2Hz for the VOR) were considered as within individual independent factors. The main effects of those factors and their interactions were tested and reported. In the case of the OCR and OVAR, mix-model ANOVA was used with only Weeks considered as within factors. For the CGR, measured at W6 and W12, mix-model ANOVA was used with the Frequencies and Model (Measured or Model CGR) used as within factors. For the comparison between the OKR and VOR Δ gain, a repeated measures ANOVA was applied on the Δ gain (Wx-W0) with the stimulation modality (OKR or VOR) as between individual independent factor and the Weeks as within individual independent factor. For the correlation between OCR and VOR, as well as the measured and theoretical CGR gain a linear regression model was fitted.

+

The effect of IDPN exposure on the HC count was reported with a Kruskal Wallis test. For all analyses the significance threshold was set at p<0.05 and Newman Keuls post hoc test was performed if the main effect or an interaction was found significant.

+ + + +Funding +

This work is supported by the Centre National d’Etudes Spatiales, the Centre National de la Recherche Scientifique, and the Université Paris Cité. MB and LS received funding from the Agence Nationale de la Recherche (ANR-20-CE37-0016 INVEST). MB, FS & DF received funding from the ERANET NEURON Program VELOSO (ANR-20-NEUR-0005). AP & JL received funding from the ERANET NEURON Program VELOSO (grant PCI2020-120681-2 from MCIN/AEI/10.13039/501100011033 and NextGenerationEU/PRTR).

+ + + + +<bold>Acknowledgment</bold> +

this study contributes to the IdEx Université de Paris ANR-18-IDEX-0001. This work has benefited from the support and expertise of the animal facility of BioMedTech Facilities at Université Paris Cité (Institut National de la Santé et de la Recherche Médicale Unité S36/Unité Mixte de Service 2009). The confocal microscopy studies were performed at the Centres Científics i Tecnològics de la Universitat de Barcelona (CCiTUB).

+ + +Author contribution +

Conceptualization: JL ; MB. Methodology: LS ; AP ; FS ; MT ; JL ; MB. Formal analysis: LS ; AP ; MT ; JL ; MB. Investigation: LS ; AP ; TB ; CEB . Writing original manuscript: LS ; MT ; MB. Writing review and editing: LS ; AP ; FS ; DF ; MT ; JL ; MB. Visualization: LS ; AP ; TB; MT ; JL ; MB. Supervision: LS ; DF ; MT ; JL ; MB. Project administration: JL; MB. Funding acquisition: DF, JL; MB.

+ + +Declaration of interests +

The authors declare no competing interests.

+

Figure 1 supplement: Effects of subchronic IDPN on canalar- and otolith-dependent VOR.

+

A, B) aVOR (A) Gain and (B) Phase across all frequencies and weeks tested for IDPN (n=21) and SHAM (n=22) mice. The grey symbols in the left panel correspond to aVOR gain values associated with VAF<0.5.C) OVAR MOR bias for IDPN (n=13) and SHAM (n=14) mice during the protocol. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+

Figure 3 supplement: Effects of the IDPN on the number of HC in the peripheral regions of the horizontal SCC ampulla and extrastriolar utricule Macula.

+

A) Individual cell count of peripheral horizontal semi-circular canal type I and type II HC for W6 (1) andW12 (2). B) Cell count of peripheral utricular type I and type II HC for W6 (1) and W12 (2). C) Individual number of peripheral Spp1+ type I HC (1), CASPR1 + type I HC (2) and Calre+ type II HC (3) as a function of the aVOR gain for W6 (circle) and W12 (squares). The linear regressionof both groups (n=23 mice) and the goodness of fit is represented. D) Individual number of peripheral Spp1+ type I HC (1), CASPR1 + type I HC (2) and Calre+ type II HC (3) as a function of the OCR gain for W6 (circle) and W12 (squares). The linear regression of both groups (n=23 mice) and the goodness of fit is represented. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+

Figure 4 supplement: Effect of the IDPN on optokinetic reflex amplitude and timing.

+

A) OKR Gain across all frequencies and weeks tested B) OKR Phase across all frequencies and weeks tested. (*p< 0.05; **p < 0.01; ***p < 0.001). Error bars represent ±SEM.

+ + +References +Agrawal Y, Carey JP, Della Santina CC, Schubert MC, Minor LB (2009) Disorders of Balance and Vestibular Function in US Adults: Data From the National Health and Nutrition Examination Survey, 2001-2004. Arch Intern Med 169:938. +Agrawal Y, Ward BK, Minor LB (2013) Vestibular dysfunction: Prevalence, impact and need for targeted treatment Chabbert C, ed. J Vestib Res 23:113–117. +Baird RA, Desmadryl G, Fernandez C, Goldberg JM (1988) The vestibular nerve of the chinchilla. II. Relation between afferent response properties and peripheral innervation patterns in the semicircular canals. J Neurophysiol 60:182–203. +Beraneck M, Bojados M, Le Séac’h A, Jamon M, Vidal P-P (2012) Ontogeny of Mouse Vestibulo-Ocular Reflex Following Genetic or Environmental Alteration of Gravity Sensing Gilestro GF, ed. PLoS ONE 7:e40414. +Beraneck M, Cullen KE (2007) Activity of Vestibular Nuclei Neurons During Vestibular and Optokinetic Stimulation in the Alert Mouse. J Neurophysiol 98:1549–1565. +Beraneck M, Idoux E (2012) Reconsidering the Role of Neuronal Intrinsic Properties and Neuromodulation in Vestibular Homeostasis. Front Neurol 3 Available at: http://journal.frontiersin.org/article/10.3389/fneur.2012.00025/abstract. +Beraneck M, Pfanzelt S, Vassias I, Rohregger M, Vibert N, Vidal P-P, Moore LE, Straka H (2007) Differential Intrinsic Response Dynamics Determine Synaptic Signal Processing in Frog Vestibular Neurons. J Neurosci 27:4283–4296. +Beraneck M, Straka H (2011) Vestibular signal processing by separate sets of neuronal filters Peusner KD, ed. J Vestib Res 21:5–19. +Bisdorff A, Von Brevern M, Lempert T, Newman-Toker DE (2009) Classification of vestibular symptoms: Towards an international classification of vestibular disorders. J Vestib Res 19:1–13. +Boadas-Vaello P, Sedó-Cabezón L, Verdú E, Llorens J (2017) Strain and Sex Differences in the Vestibular and Systemic Toxicity of 3,3′-Iminodipropionitrile in Mice. Toxicol Sci:kfw238. +Boyden ES, Katoh A, Raymond JL (2004) CEREBELLUM-DEPENDENT LEARNING: The Role of Multiple Plasticity Mechanisms. Annu Rev Neurosci 27:581–609. +Boyle R, Carey JP, Highstein SM (1991) Morphological correlates of response dynamics and efferent stimulation in horizontal semicircular canal afferents of the toadfish, Opsanus tau. J Neurophysiol 66:1504–1521. +Boyle R, Goldberg JM, Highstein SM (1992) Inputs from regularly and irregularly discharging vestibular nerve afferents to secondary neurons in squirrel monkey vestibular nuclei. III. Correlation with vestibulospinal and vestibuloocular output pathways. J Neurophysiol 68:471–484. +Brandt T, Dieterich M (2017) The dizzy patient: don’t forget disorders of the central vestibular system. Nat Rev Neurol 13:352–362. +Brandt T, Strupp M, Arbusow V, Dieringer N (1997) Plasticity of the vestibular system: central compensation and sensory substitution for vestibular deficits. Adv Neurol:297–309. +Brueckner JK, Ashby LP, Prichard JR, Porter JD (1999) Vestibulo-ocular pathways modulate extraocular muscle myosin expression patterns. Cell Tissue Res 295:477–484. +Carcaud J, França de Barros F, Idoux E, Eugène D, Reveret L, Moore LE, Vidal P-P, Beraneck M (2017) Long-Lasting Visuo-Vestibular Mismatch in Freely-Behaving Mice Reduces the Vestibulo-Ocular Reflex and Leads to Neural Changes in the Direct Vestibular Pathway. eneuro 4:ENEURO.0290-16.2017. +Carey JP, Fuchs AF, Rubel EW (1996) Hair cell regeneration and recovery of the vestibuloocular reflex in the avian vestibular system. J Neurophysiol 76:3301–3312. +Carriot J, Jamali M, Chacron MJ, Cullen KE (2017) The statistics of the vestibular input experienced during natural self-motion differ between rodents and primates: Natural vestibular input in rodents and monkeys. J Physiol 595:2751–2766. +Cassel R, Bordiga P, Carcaud J, Simon F, Beraneck M, Le Gall A, Benoit A, Bouet V, Philoxene B, Besnard S, Watabe I, Pericat D, Hautefort C, Assie A, Tonetto A, Dyhrfjeld-Johnsen J, Llorens J, Tighilet B, Chabbert C (2019) Morphological and functional correlates of vestibular synaptic deafferentation and repair in a mouse model of acute onset vertigo. Dis Model Mech:dmm.039115. +Contini D, Holstein GR, Art JJ (2022) Simultaneous Dual Recordings From Vestibular Hair Cells and Their Calyx Afferents Demonstrate Multiple Modes of Transmission at These Specialized Endings. Front Neurol 13:891536. +Cullen K (2019) Vestibular processing during natural self-motion: implications for perception and action. :41. +Cullen KE, Minor LB, Beraneck M, Sadeghi SG (2010) Neural substrates underlying vestibular compensation: Contribution of peripheral versus central processing. J Vestib Res 19:171–182. +De Zeeuw CI, Lisberger SG, Raymond JL (2021) Diversity and dynamism in the cerebellum. Nat Neurosci 24:160–167. +Dean P, Porrill J (2014) Decorrelation Learning in the Cerebellum. In: Progress in Brain Research, pp 157–192. Elsevier. Available at: https://linkinghub.elsevier.com/retrieve/pii/B9780444633569000078. +Eatock RA, Songer JE (2011) Vestibular Hair Cells and Afferents: Two Channels for Head Motion Signals. Annu Rev Neurosci 34:501–534. +Ernst MO, Banks MS (2002) Humans integrate visual and haptic information in a statistically optimal fashion. Nature 415:429–433. +Facchini J, Rastoldo G, Xerri C, Péricat D, El Ahmadi A, Tighilet B, Zennou-Azogui Y (2021) Unilateral vestibular neurectomy induces a remodeling of somatosensory cortical maps. Prog Neurobiol 205:102119. +Faulstich BM, Onori KA, du Lac S (2004) Comparison of plasticity and development of mouse optokinetic and vestibulo-ocular reflexes suggests differential gain control mechanisms. Vision Res 44:3419–3427. +Faulstich M, van Alphen AM, Luo C, du Lac S, De Zeeuw CI (2006) Oculomotor Plasticity During Vestibular Compensation Does Not Depend on Cerebellar LTD. J Neurophysiol 96:1187–1195. +Fetsch CR, DeAngelis GC, Angelaki DE (2010) Visual-vestibular cue integration for heading perception: applications of optimal cue integration theory: Mechanisms of visual-vestibular cue integration. Eur J Neurosci 31:1721–1729. +França de Barros F, Carcaud J, Beraneck M (2019) Long-term Sensory Conflict in Freely Behaving Mice. J Vis Exp:59135. +França de Barros F, Schenberg L, Tagliabue M, Beraneck M (2020) Long term visuo-vestibular mismatch in freely behaving mice differentially affects gaze stabilizing reflexes. Sci Rep 10:20018. +Glasauer S (2007) Current Models of the Ocular Motor System. In: Developments in Ophthalmology, 2007 (Straube A, Büttner U, eds), pp 158–174. Basel: KARGER. Available at: https://www.karger.com/Article/FullText/100485. +Goldberg JM (2000) Afferent diversity and the organization of central vestibular pathways. Exp Brain Res 130:277–297. +Goldberg JM, Lysakowski A, Fernández C (1990) Morphophysiological and ultrastructural studies in the mammalian cristae ampullares. Hear Res 49:89–102. +Goldberg JM, Lysakowski A, Fernández C (1992) Structure and Function of Vestibular Nerve Fibers in the Chinchilla and Squirrel Monkey. Ann N Y Acad Sci 656:92–107. +González-Garrido A, Pujol R, López-Ramírez O, Finkbeiner C, Eatock RA, Stone JS (2021) The Differentiation Status of Hair Cells That Regenerate Naturally in the Vestibular Inner Ear of the Adult Mouse. J Neurosci 41:7779–7796. +Greguske EA, Carreres-Pons M, Cutillas B, Boadas-Vaello P, Llorens J (2019) Calyx junction dismantlement and synaptic uncoupling precede hair cell extrusion in the vestibular sensory epithelium during sub-chronic 3,3′-iminodipropionitrile ototoxicity in the mouse. Arch Toxicol 93:417–434. +Greguske EA, Llorens J, Pyott SJ (2021) Assessment of cochlear toxicity in response to chronic 3,3′-iminodipropionitrile in mice reveals early and reversible functional loss that precedes overt histopathology. Arch Toxicol 95:1003–1021. +Grosch M, Lindner M, Bartenstein P, Brandt T, Dieterich M, Ziegler S, Zwergal A (2021) Dynamic whole-brain metabolic connectivity during vestibular compensation in the rat. NeuroImage 226:117588. +Hasson T, Gillespie PG, Garcia JA, MacDonald RB, Zhao Y, Yee AG, Mooseker MS, Corey DP (1997) Unconventional Myosins in Inner-Ear Sensory Epithelia. J Cell Biol 137:1287–1307. +Hess BJM, Dieringer N (1990) Spatial Organization of the Maculo-Ocular Reflex of the Rat: Responses During Off-Vertical Axis Rotation. Eur J Neurosci 2:909–919. +Highstein SM, Goldberg JM, Moschovakis AK, Fernandez C (1987) Inputs from regularly and irregularly discharging vestibular nerve afferents to secondary neurons in the vestibular nuclei of the squirrel monkey. II. Correlation with output pathways of secondary neurons. J Neurophysiol 58:719–738. +Holland PJ, Sibindi TM, Ginzburg M, Das S, Arkesteijn K, Frens MA, Donchin O (2020) A Neuroanatomically Grounded Optimal Control Model of the Compensatory Eye Movement System in Mice. Front Syst Neurosci 14:13. +Idoux E, Tagliabue M, Beraneck M (2018) No Gain No Pain: Relations Between Vestibulo-Ocular Reflexes and Motion Sickness in Mice. Front Neurol 9:918. +Jang DC, Shim HG, Kim SJ (2020) Intrinsic Plasticity of Cerebellar Purkinje Cells Contributes to Motor Memory Consolidation. J Neurosci 40:4145–4157. +Kim GS, Wang T, Sayyid ZN, Fuhriman J, Jones SM, Cheng AG (2022) Repair of surviving hair cells in the damaged mouse utricle. Proc Natl Acad Sci 119:e2116973119. +Kodama T, du Lac S (2016) Adaptive Acceleration of Visually Evoked Smooth Eye Movements in Mice. J Neurosci 36:6836–6849. +Lacour M, Helmchen C, Vidal P-P (2016) Vestibular compensation: the neuro-otologist’s best friend. J Neurol 263:54–64. +Lasker DM, Han GC, Park HJ, Minor LB (2008) Rotational Responses of Vestibular–Nerve Afferents Innervating the Semicircular Canals in the C57BL/6 Mouse. J Assoc Res Otolaryngol 9:334–348. +Leong ATL, Gu Y, Chan Y-S, Zheng H, Dong CM, Chan RW, Wang X, Liu Y, Tan LH, Wu EX (2019) Optogenetic fMRI interrogation of brain-wide central vestibular pathways. Proc Natl Acad Sci 116:10122–10129. +Llorens J, Demêmes D (1994) Hair cell degeneration resenting from 3,3’-imiodiprpiunitril toxicity in the rat vestibular epithelia. Hear Res:78–86. +Llorens J, Demêmes D, Sans A (1993) The behavioural syndrome caused by 3’3-Iminodipropionitrile and related nitriles in the rat is associated with degeneration of the vestibular sensory hair cells. Toxicol Appl Pharmacol:199–210. +Llorens J, Demêmes D, Sans A (1994) The toxicity of IDPN on the vestibular system of the rat: new insights on its effects on behavior and neurofilament transport. NeuroToxicology:643–647. +Lopez C, Schreyer H-M, Preuss N, Mast FW (2012) Vestibular stimulation modifies the body schema. Neuropsychologia 50:1830–1837. +Lopez-Escamez JA, Carey J, Chung W-H, Goebel JA, Magnusson M, Mandalà M, Newman-Toker DE, Strupp M, Suzuki M, Trabalzini F, Bisdorff A (2015) Diagnostic criteria for Menière’s disease. J Vestib Res 25:1–7. +Lysakowski A, Gaboyard-Niay S, Calin-Jageman I, Chatlani S, Price SD, Eatock RA (2011) Molecular Microdomains in a Sensory Terminal, the Vestibular Calyx Ending. J Neurosci 31:10101–10114. +Mackrous I, Carriot J, Cullen KE, Chacron MJ (2020) Neural variability determines coding strategies for natural self-motion in macaque monkeys. eLife 9:e57484. +Maroto AF, Barrallo-Gimeno A, Llorens J (2021) Relationship between vestibular hair cell loss and deficits in two anti-gravity reflexes in the rat. Hear Res 410:108336. +Martins-Lopes V, Bellmunt A, Greguske EA, Maroto AF, Boadas-Vaello P, Llorens J (2019a) Quantitative Assessment of Anti-Gravity Reflexes to EvaluateVestibular Dysfunction in Rats. J Assoc Res Otolaryngol 20:553–563. +Martins-Lopes V, Bellmunt A, Greguske EA, Maroto AF, Boadas-Vaello P, Llorens J (2019b) Quantitative Assessment of Anti-Gravity Reflexes to EvaluateVestibular Dysfunction in Rats. J Assoc Res Otolaryngol 20:553–563. +McInturff S, Burns JC, Kelley MW (2018) Characterization of spatial and temporal development of Type I and Type II hair cells in the mouse utricle using new cell-type-specific markers. Biol Open 7:bio038083. +Minor L, Goldberg J (1991) Vestibular-nerve inputs to the vestibulo-ocular reflex: a functional-ablation study in the squirrel monkey. J Neurosci 11:1636–1648. +Möhwald K, Hadzhikolev H, Bardins S, Becker-Bense S, Brandt T, Grill E, Jahn K, Dieterich M, Zwergal A (2020) Health-related quality of life and functional impairment in acute vestibular disorders. Eur J Neurol 27:2089–2098. +Nelson AB, Faulstich M, Moghadam S, Onori K, Meredith A, du Lac S (2017) BK Channels Are Required for Multisensory Plasticity in the Oculomotor System. Neuron 93:211– 220. +Oommen BS, Stahl JS (2008) Eye orientation during static tilts and its relationship to spontaneous head pitch in the laboratory mouse. Brain Res 1193:57–66. +Paken J, Govender CD, Pillay M, Sewram V (2016) Cisplatin-Associated Ototoxicity: A Review for the Health Professional. J Toxicol 2016:1–13. +Pfanzelt S, Rossert C, Rohregger M, Glasauer S, Moore LE, Straka H (2008) Differential Dynamic Processing of Afferent Signals in Frog Tonic and Phasic Second-Order Vestibular Neurons. J Neurosci 28:10349–10362. +Prayuenyong P, Taylor JA, Pearson SE, Gomez R, Patel PM, Hall DA, Kasbekar AV, Baguley DM (2018) Vestibulotoxicity Associated With Platinum-Based Chemotherapy in Survivors of Cancer: A Scoping Review. Front Oncol 8:363. +Sadeghi SG, Beraneck M (2020) Task-Specific Differentiation of Central Vestibular Neurons and Plasticity During Vestibular Compensation. In: The Senses: A Comprehensive Reference, pp 290–308. Elsevier. Available at: https://linkinghub.elsevier.com/retrieve/pii/B978012809324524145X. +Sadeghi SG, Minor LB, Cullen KE (2007) Response of Vestibular-Nerve Afferents to Active and Passive Rotations Under Normal Conditions and After Unilateral Labyrinthectomy. J Neurophysiol 97:1503–1514. +Sadeghi SG, Minor LB, Cullen KE (2012) Neural Correlates of Sensory Substitution in Vestibular Pathways following Complete Vestibular Loss. J Neurosci 32:14685– 14695. +Sayyid ZN, Wang T, Chen L, Jones SM, Cheng AG (2019) Atoh1 Directs Regeneration and Functional Recovery of the Mature Mouse Vestibular System. Cell Rep 28:312–324.e4. +Sedó-Cabezón L, Boadas-Vaello P, Soler-Martín C, Llorens J (2014) Vestibular damage in chronic ototoxicity: A mini-review. NeuroToxicology 67. +Sedó-Cabezón L, Jedynak P, Boadas-Vaello P, Llorens J (2015) Transient alteration of the vestibular calyceal junction and synapse in response to chronic ototoxic insult in rats. Dis Model Mech:dmm.021436. +Seoane A, Demêmes D, Llorens J (2001) Pathology of the rat vestibular sensory epithelia during subchronic 3,3′-iminodipropionitrile exposure: hair cells may not be the primary target of toxicity. Acta Neuropathol (Berl) 102:339–348. +Shin S-L, Zhao GQ, Raymond JL (2014) Signals and Learning Rules Guiding Oculomotor Plasticity. J Neurosci 34:10635–10644. +Shinder ME, Perachio AA, Kaufman GD (2005) VOR and Fos response during acute vestibular compensation in the Mongolian gerbil in darkness and in light. Brain Res 1038:183–197. +Simon F, Pericat D, Djian C, Fricker D, Denoyelle F, Beraneck M (2020) Surgical techniques and functional evaluation for vestibular lesions in the mouse: unilateral labyrinthectomy (UL) and unilateral vestibular neurectomy (UVN). J Neurol 267:51– 61. +Simon F, Tissir F, Michel V, Lahlou G, Deans M, Beraneck M (2021) Implication of Vestibular Hair Cell Loss of Planar Polarity for the Canal and Otolith-Dependent Vestibulo-Ocular Reflexes in Celsr1–/– Mice. Front Neurosci 15:750596. +Soler-Martin C, Diez-Padrisa N, Boadas-Vaello P, Llorens J (2006) Behavioral Disturbances and Hair Cell Loss in the Inner Ear Following Nitrile Exposure in Mice, Guinea Pigs, and Frogs. Toxicol Sci 96:123–132. +Sousa AD, Andrade LR, Salles FT, Pillai AM, Buttermore ED, Bhat MA, Kachar B (2009) The Septate Junction Protein Caspr Is Required for Structural Support and Retention of KCNQ4 at Calyceal Synapses of Vestibular Hair Cells. J Neurosci 29:3103–3108. +Straka H, Vibert N, Vidal PP, Moore LE, Dutia MB (2005) Intrinsic membrane properties of vertebrate vestibular neurons: Function, development and plasticity. Prog Neurobiol 76:349–392. +Straka H, Zwergal A, Cullen KE (2016) Vestibular animal models: contributions to understanding physiology and disease. J Neurol 263:10–23. +Welgampola MS, Young AS, Pogson JM, Bradshaw AP, Halmagyi GM (2019) Dizziness demystified. Pract Neurol 19:492–501. +Wibble T, Pansell T, Grillner S, Pérez-Fernández J (2022) Conserved subcortical processing in visuo-vestibular gaze control. Nat Commun 13:4699. +Wilkerson BA, Artoni F, Lea C, Ritchie K, Ray CA, Bermingham-McDonogh O (2018) Effects of 3,3′-Iminodipropionitrile on Hair Cell Numbers in Cristae of CBA/CaJ and C57BL/6J Mice. J Assoc Res Otolaryngol 19:483–491. +Yang X, Zhou S, Wu J, Liao Q, Wang C, Liu M, Qu L, Zhang Y, Cheng C, Chai R, Zhang K, Yu X, Huang P, Liu L, Xiong W, Chen S, Chen F (2019) Surgery-free video-oculography in mouse models: enabling quantitative and short-interval longitudinal assessment of vestibular function. Neurosci Lett 696:212–218. +Zakir M, Dickman JD (2006) Regeneration of Vestibular Otolith Afferents after Ototoxic Damage. J Neurosci 26:2881–2893. +Zeng S, Ni W, Jiang H, You D, Wang J, Lu X, Liu L, Yu H, Wu J, Chen F, Li H, Wang Y, Chen Y, Li W (2020) Toxic Effects of 3,3 ′ -Iminodipropionitrile on Vestibular System in Adult C57BL/6J Mice In Vivo. Neural Plast 2020:1–11. +Zennou-Azogui Y, Xerri C, Harlay F (1994) Visual sensory substitution in vestibular compensation: neuronal substrates in the alert cat. Exp Brain Res 98 Available at: http://link.springer.com/10.1007/BF00233983. +Zwergal A, Schlichtiger J, Xiong G, Beck R, Günther L, Schniepp R, Schöberl F, Jahn K, Brandt T, Strupp M, Bartenstein P, Dieterich M, Dutia MB, la Fougère C (2016) Sequential [18F]FDG µPET whole-brain imaging of central vestibular compensation: a model of deafferentation-induced brain plasticity. Brain Struct Funct 221:159–170. + + +
diff --git a/test/fixtures/2023.07.30.551185/2023.07.30.551185.xml b/test/fixtures/2023.07.30.551185/2023.07.30.551185.xml new file mode 100644 index 00000000..d5d6f2cd --- /dev/null +++ b/test/fixtures/2023.07.30.551185/2023.07.30.551185.xml @@ -0,0 +1,445 @@ + + +
+ + +BIORXIV + +bioRxiv +bioRxiv + + +Cold Spring Harbor Laboratory + + + +10.1101/2023.07.30.551185 +1.3 + + +Regular Article + + +New Results + + +Neuroscience + + + +Smith-Magenis syndrome protein RAI1 regulates body weight homeostasis through hypothalamic BDNF-producing neurons and neurotrophin downstream signalling + + + +JavedSehrish +1 +2 + + +ChangYa-Ting +1 +2 + + +ChoYoobin +1 +2 + + +LeeYu-Ju +1 +2 + + +ChangHao-Cheng +1 +2 + + +HaqueMinza +1 +2 + + +LinYu Cheng +1 +2 + + +http://orcid.org/0000-0002-8411-9433 +HuangWei-Hsiang +1 +2 +* + +Department of Neurology and Neurosurgery, Centre for Research in Neuroscience, McGill University, Montréal, QC H3G 1A3, Canada +Brain Repair and Integrative Neuroscience Program, The Research Institute of the McGill University Health Centre, Montréal, QC H3G 1A3, Canada + + +Corresponding author. Email: wei-hsiang.huang@mcgill.ca + +2023 +2023.07.30.551185 + + +30 +7 +2023 + + +23 +10 +2023 + + +23 +10 +2023 + + + +© 2023, Posted by Cold Spring Harbor Laboratory +2023 +This pre-print is available under a Creative Commons License (Attribution 4.0 International), CC BY 4.0, as described at http://creativecommons.org/licenses/by/4.0/ + + + +Abstract

Retinoic acid-induced 1 (RAI1) haploinsufficiency causes Smith-Magenis syndrome (SMS), a genetic disorder with symptoms including hyperphagia, hyperlipidemia, severe obesity, and autism phenotypes. RAI1 is a transcriptional regulator with a pan-neural expression pattern and hundreds of downstream targets. The mechanisms linking neural Rai1 to body weight regulation remain unclear. Here we find that hypothalamic brain-derived neurotrophic factor (BDNF) and its downstream signalling are disrupted in SMS (Rai1+/-) mice. Selective Rai1 loss from all BDNF-producing cells or from BDNF-producing neurons in the paraventricular nucleus of the hypothalamus (PVH) induced obesity in mice. Electrophysiological recordings revealed that Rai1 ablation decreased the intrinsic excitability of PVHBDNF neurons. Chronic treatment of SMS mice with LM22A-4 engages neurotrophin downstream signalling and delayed obesity onset. This treatment also partially rescued disrupted lipid profiles, insulin intolerance, and stereotypical repetitive behaviour in SMS mice. These data argue that RAI1 regulates body weight and metabolic function through hypothalamic BDNF-producing neurons and that targeting neurotrophin downstream signalling might improve associated SMS phenotypes.

+ + + + + + + +Competing Interest Statement

The authors have declared no competing interest.

 + +Summary of Updates: +

provided source images for western blotting

 + + + + + +Introduction +

Obesity is an increasing global concern: 20% of the world’s adult population is expected to become obese by 2025 if current trends persist (Collaboration, 2017). However, the complex environmental, behavioural, and genetic contributions to obesity make this a complex condition to model and address. Monogenic obesity disorders provide a consistent system where the pathophysiology, genetic predisposition, and neurobiology underlying phenotypes can potentially provide insights that translate to polygenic obesity (Loos & Yeo, 2022).

+

One such monogenic disorder is Smith-Magenis syndrome (SMS; OMIM #182290) (Smith et al, 1986), a neurogenetic condition frequently associated with obesity. 80% of SMS cases are caused by interstitial deletions of the 17p11.2 chromosomal region containing a dosage-sensitive gene, retinoic acid-induced 1 (RAI1) (Slager et al, 2003). Most SMS patients are overweight (40% > 95th percentile weight, 80% > 85th percentile weight) and exhibit over-eating tendencies by adolescence (Alaimo et al, 2015; Burns et al, 2010; Gandhi et al, 2022). 20% of SMS patients do not have 17p11.2 deletions but instead carry RAI1 point mutations that are still frequently associated with obesity (Boot et al, 2021; Edelman et al, 2007). These clinical data indicate that RAI is critical for body weight regulation.

+

The human RAI1 protein shares >80% sequence identity with the mouse RAI1 protein, allowing us to study RAI1 function in vivo (Huang et al, 2016; Javed et al, 2020). We previously found that Rai1 encodes a transcriptional regulator that binds to the promoter or enhancer regions of its target genes, most of which are involved in neurodevelopment and synaptic transmission (Huang et al., 2016). SMS (Rai1+/-) mice exhibit obesity associated with hyperphagia and increased adiposity (Burns et al., 2010; Huang et al., 2016). Interestingly, RAI1 may have a sexually dimorphic function, as female SMS mice exhibit more pronounced obesity than males (Burns et al., 2010; Huang et al., 2016).

+

Using a homozygous Rai1 conditional knockout (Rai1-cKO) mice, we found that Rai1 deletion in either subcortical excitatory neurons (vGlut2+) or single-minded homolog 1 (Sim1+) neurons induced obesity in mice (Huang et al., 2016). Sim1+ neurons reside in several brain regions including the nucleus of the lateral olfactory tract, the supraoptic nucleus, the medial amygdala, and the paraventricular nucleus of the hypothalamus (PVH), a region that is critical for body weight regulation (Balthasar et al, 2005). Sim1+ neurons in the PVH are composed of several molecularly defined cell types, including those that express oxytocin, vasopressin, corticotropin-releasing hormone, somatostatin, melanocortin-4-receptor (MC4R), growth hormone-releasing hormone (GHRH), or brain-derived neurotrophic factor (BDNF) (An et al, 2015). Deleting Bdnf from the PVH results in obesity due to increased food intake, reduced energy expenditure, and reduced locomotion (An et al., 2015). While PVH neuronal activity is reduced during feeding (Xu et al, 2019), activation of PVHBDNF neurons results in weight loss (Wu & Xu, 2022). We recently found that RAI1 promotes Bdnf expression in the hypothalamus throughout life (Javed et al, 2021). We hypothesized that RAI11 may be connected to BDNF-producing neurons in PVH, explaining how a neural transcription factor regulates body weight homeostasis.

+

BDNF regulates energy metabolism by binding to its cognate receptor tropomyosin receptor kinase B (TRKB) (An et al., 2015) and may thus provide a potential therapeutic target for SMS. However, the link between RAI1, BDNF, and SMS symptoms is unclear. Patients carrying monogenic mutations in either BDNF (Gray et al, 2006) or NTRK2 (encoding TRKB) (Yeo et al, 2004) experience severe hyperphagic obesity. Moreover, obesity and associated metabolic dysfunctions such as impaired glucose and insulin sensitivity are correlated with impaired PI3K-AKT pathway, a TRKB downstream target (Su et al, 2021). Given that (1) Bdnf overexpression during early embryonic development or (2) targeting Bdnf overexpression to the PVH of adolescent SMS (Rai1+/-) mice protected them from weight gain and metabolic defects (Javed et al., 2021), and that (3) a point mutation in RAI1 disrupts its ability to transcriptionally enhance BDNF expression in vitro (Abad et al, 2018), pharmacological targeting of neurotrophin downstream signalling is an attractive and yet unexplored therapeutic avenue for SMS.

+

Building on our previous discoveries, here we examine the neurobiological functions of RAI1 in BDNF-producing neurons, specifically those that reside in the PVH. Importantly, we evaluated the involvement of the BDNF-TRKB pathways in SMS pathogenesis using a small molecule drug (LM22A-4) to enhance neurotrophin downstream signalling. Our work indicates that PVHBDNF neurons depend on RAI1 to maintain normal neuronal activity and regulate body weight homeostasis, and dysfunction of BDNF downstream signalling contributes to obesity associated with SMS mice.

+ + +Results + +<italic>Rai1</italic> deficiency dysregulates specific downstream targets of neurotrophin in the mouse hypothalamus +

Multiple lines of evidence indicated that SMS mice show decreased Bdnf mRNA expression in the hypothalamus (Burns et al., 2010; Huang et al., 2016; Javed et al., 2021). To determine if BDNF protein levels were consequently altered in the hypothalamus, we performed an enzyme-linked immunosorbent assay (ELISA) and found that BDNF protein levels were significantly reduced in 7-week-old pre-symptomatic SMS mice relative to controls (Fig 1A). To quantify if molecular pathways associated with BDNF downstream signalling were altered, we subjected hypothalamic tissue lysates from 7-week-old pre-obese SMS and control (Ctrl) mice to reverse phase protein array (RPPA) (Fig 1B). Control and SMS samples are clustered separately in the principal component analysis (Fig 1C).

+ + +<italic>Rai1</italic> haploinsufficiency disrupts hypothalamic proteomic profile in mice

(A) ELISA showing that BDNF protein levels in hypothalamic tissues were significantly reduced in SMS mice compared to controls (n=4/genotype). *p<0.05, unpaired t-test, two-tailed.

(B) A schematic diagram showing the experimental strategy. Hypothalamic tissue lysates from 7- week-old control (n = 5, technical triplicates/sample) and SMS (n=8, technical triplicates/sample) mice were subjected to RPPA that utilizes a cocktail of 240 validated antibodies to probe several key intracellular signalling pathways by quantifying the expression levels of total and phosphorylated proteins

(C) Principal component analysis segregates the SMS and Ctrl groups where PC1 explains 47% of the variance, and PC2 explains 25% of the variance.

(D) Heat map showing hierarchical clustering of differentially expressed phospho-proteins: most showed reduced levels in SMS samples.

(E) Protein-protein interaction (PPI) analysis showing differentially expressed proteins and phospho-proteins in a crosstalk network. PPI enrichment p-value: < 1.0e-16. The nodes represent the proteins/phospho-proteins and the lines indicate physical interactions. Associations with specific molecular pathways are shown as color-coded outer rings.

(F-G) The protein levels of two neurotrophin pathway-related nodes, PIK3CA (F) and phospho-NF-kB (G), were significantly reduced in SMS samples. ****p<0.0001, unpaired t-test, two-tailed

+ + +

Hierarchical clustering analysis indicated that most differentially expressed phosphorylated proteins (Fig 1D) and total proteins (Figure 1—figure supplement 1) were downregulated in SMS mice. We mapped total proteins, as well as phospho-proteins, that were differentially expressed in Ctrl and SMS samples onto the interconnected biological network and identified a subset of connected nodes downregulated in SMS (FDR < 10e-5) belong to the neurotrophin, PI3K-AKT, and mTOR signalling cascades (Fig 1E). These molecules are known to play a critical role in regulating body weight homeostasis and can be activated by BDNF either directly (neurotrophin pathway) or indirectly (PI3K-AKT and mTOR pathways) (Cota et al, 2006; Schultze et al, 2012). The expression levels of PIK3CA, a common node in these pathways, were significantly reduced in the SMS mice compared to controls (Fig 1F). Moreover, the expression levels of phosphorylated NF-κB (nuclear factor-kappa B), a downstream target of the neurotrophin pathway, were significantly reduced in the SMS group (Fig 1G). Together, these results suggest that Rai1 loss impairs specific pathways, including neurotrophin signalling, and indicate that dysfunction of BDNF-producing cells could contribute to SMS pathogenesis.

+ + +BDNF-producing cells depend on RAI1 to regulate body weight homeostasis +

To target BDNF-producing cells, we used a BdnfCre/+ mouse line that carries a viral 2A oligopeptide and Cre recombinase immediately upstream of Bdnf’s endogenous STOP codon (Tan et al, 2016). We crossed the BdnfCre/+ mice with tdTomato-Ai9 reporter mice and found that Ai9 signals faithfully reflected endogenous BDNF protein expression in the PVH (Figure 2— figure supplement 1A-E). We also found that RAI1 is expressed in PVHBDNF cells that are distinct from larger magnocellular cells, such as the PVH oxytocin neurons (Figure 2—figure supplement 2A-I). Specifically, 55% of PVHBDNF neurons express RAI1, whereas only 2% of PVHOxytocin neurons express RAI1 (Figure 2—figure supplement 2I). These data suggest that RAI1 is enriched in BDNF-producing neurons in the mouse PVH.

+

To determine if BDNF-producing cells depend on RAI1 to regulate body weight, we deleted one or both copies of Rai1 using the BdnfCre/+ and the conditional Rai1fl/fl mice (Huang et al., 2016) (Fig 2A, Figure 2—figure supplement 3A). Conditional knockout mice (cKO: BdnfCre/+; Rai1fl/fl) showed a significant reduction in Rai1 expression in PVHBDNF neurons compared to controls (Fig 2B-C, Figure 2—figure supplement 3B). The total number of BDNF-producing cells in the PVH remained similar in Ctrl and cKO mice (Figure 2—figure supplement 3C), indicating that RAI1 loss did not affect PVH neurogenesis or survival.

+ + +RAI1 is required in BDNF-producing cells to regulate energy metabolism +

(A) A schematic diagram showing selective deletion of Rai1 from BDNF-producing cells in mice.

(B) Representative images showing that in Ctrl mice (BdnfCre/+; Ai9), many PVHBDNF neurons (magenta) express RAI1 (cyan, double-positive cells were indicated with yellow arrows). By contrast, PVHBDNF neurons lack RAI1 expression in the cKO group (BdnfCre/+; Rai1fl/fl; Ai9). Scale bars = 20µm.

(C) Percentage of PVHBDNF neurons co-expressing RAI1 in Ctrl (n=3) and cKO (n=3) mice. unpaired t-test, two-tailed, ****p<0.0001.

(D) Female cKO mice (n = 12) showed a significant weight gain when compared to female Ctrl mice (n = 10). Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

(E) Body composition was measured with Echo-MRI, showing an increased fat mass in 26- week-old mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. ****p<0.0001.

(F) Fat mass of brown adipocytes (BAT), subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) in Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. **p<0.01, ****p<0.0001.

(G) Representative images showing eWAT adipocyte hypertrophy of the cKO mice (left). Scale bar = 500µm. Frequency distribution of adipocytes at each cellular size (right). Two-way ANOVA with Šidák’s multiple comparisons test. ***p<0.001, ****p<0.0001.

(H) Female cKO mice showed significantly increased blood leptin levels. unpaired t-test, two-tailed, **p<0.01.

(I) Female cKO mice showed reduced energy expenditure during the dark phase. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(J) Female cKO mice showed reduced locomotor activity during the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(K) Female cKO mice showed similar food intake as control mice. ns indicates not significant. unpaired t-test, two-tailed.

(L) Glucose tolerance test showing that female cKO mice became hyperglycemic 20 minutes after intraperitoneal glucose administration, suggesting glucose intolerance. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

Data are shown as mean±SEM.

+ + +

We monitored body weight weekly and found that male and female cKO mice showed a significant increase in body weight beginning at 15 weeks of age compared to controls (Ctrl: BdnfCre/+) (Fig 2D, Figure 2—figure supplement 4A). However, conditional Rai1 heterozygous mice (BdnfCre/+; Rai1fl/+) did not show a pathological increase in body weight (Figure 2—figure supplement 3D, Figure 2—figure supplement 4B), suggesting that non-BDNF-producing cells also contribute to SMS pathogenesis. EcoMRI found that 26-week-old cKO male and female mice showed increased fat mass but not lean mass (Fig 2E, Figure 2—figure supplement 4C). Further dissection of fat tissues from different organs showed an increased mass of subcutaneous inguinal (S.C. Ing.) fat in cKO mice of both sexes and an increased mass of epididymal adipose tissues (eWAT) in female cKOs (Fig 2F, Figure 2—figure supplement 4D).

+

In contrast, the total mass of brown adipose tissue (BAT) remained unchanged between groups (Fig 2F, Figure 2—figure supplement 4D). In both male and female cKO mice, we found hypertrophic eWAT adipocytes (Fig 2G, Figure 2—figure supplement 4E) and elevated blood leptin levels (Fig 2H, Figure 2—figure supplement 4F). However, there was no significant alteration in blood lipids such as triglycerides (TG), high-density lipoprotein (HDL), low-density lipoproteins (LDL), and very low-density lipoprotein (VLDL) in either male or female cKO mice (Figure 2—figure supplement 3E-G, Figure 2—figure supplement 4G-I). These data suggest that obesity caused by ablating Rai1 from BDNF-producing cells is associated with increased fat mass, fat deposition, and leptin levels.

+

Obesity could result from increased food intake, reduced energy expenditure, or both. Similar to SMS mice, male and female cKO mice did not have defects in respiratory exchange rates when measured by indirect calorimetry (Figure 2—figure supplement 3H, Figure 2—figure supplement 4J). Interestingly, cKO mice showed sex-specific differences in energy expenditure. During the dark phase, female cKO mice showed reduced energy expenditure (Fig 2I) and decreased total locomotor activity (Fig 2J). In contrast, total food intake remained unchanged (Fig 2K). Male cKO mice became hyperphagic (Figure 2—figure supplement 4K), showed increased energy expenditure during the dark phase (Figure 2—figure supplement 4L), and exhibited normal locomotor activity when compared to controls (Figure 2—figure supplement 4M). These results suggest that obesity in male cKO mice is attributable to overeating despite increased energy expenditure.

+

To determine if Rai1 loss from BDNF-producing cells impacts metabolic function, a glucose tolerance test (GTT) was performed. Interestingly, cKO mice of both sexes showed normal blood glucose levels right before the test but became hyperglycemic 20 minutes (female cKO) and 45 minutes (male cKO) after glucose administration (Fig 2L, Figure 2—figure supplement 4N). This change was accompanied by a significantly higher area under the curve (AUC) for both groups (Figure 2—figure supplement 3I, Figure 2—figure supplement 4O). While the male cKO mice remained hyperglycemic throughout the GTT (Figure 2—figure supplement 4N), the glucose levels in female cKO mice returned to normal towards the end (Fig 2L). We also found increased blood insulin levels during the GTT in both sexes (Figure 2—figure supplement 3J, Figure 2—figure supplement 4P). In the insulin tolerance test (ITT), female cKO mice showed moderately elevated glucose levels 20 minutes into the test (Figure 2—figure supplement 3K). By contrast, male cKO mice showed considerably higher glucose levels than controls in the ITT (∼6.5 ng/ml for cKO and ∼2.0ng/ml for Ctrl) during the 60- to 120-minute intervals (Figure 2— figure supplement 4Q). These experiments demonstrate that BDNF-producing cells depend on RAI1 to regulate body weight, adiposity, glucose and insulin tolerance in mice.

+ + +<italic>Rai1</italic> deletion reduces PVH<sup>BDNF</sup> neuronal excitability +

BDNF-producing neurons are widely distributed in the hypothalamus, being found in the dorsomedial hypothalamus (Unger et al, 2007), ventromedial hypothalamus (Xu et al, 2003), and PVH (An et al., 2015). To understand how Rai1 loss affects intrinsic neuronal excitability and synaptic transmission of PVHBDNF neurons that underlie energy homeostasis (An et al., 2015), we fluorescently labelled them with Cre-dependent tdTomato (Fig. 3A). We found that a hyperpolarized step followed by increasing the holding voltage at resting membrane potentials (Rm) was able to initiate rebound firing or rebound repetitive firing in most PVHBDNF neurons (Figure 3—figure supplement 1A). Hyperpolarization-induced IH currents and low depolarization events were observed in some PVHBDNF neurons (Figure 3—figure supplement 1A). cKO PVHBDNF neurons showed reduced action potential (AP) firing frequency at a holding voltage of -60mV compared to Ctrl PVHBDNF neurons (Fig 3B-C). The resting membrane potential (Rm) of cKO (-52mV) neurons showed a 10mV positive shift compared to Ctrl neurons (-62mV) (Fig 3D). We also found that fewer cKO PVHBDNF neurons showed spontaneous firing at Rm (Fig 3E, Ctrl:10/13; cKO:5/12) and at a holding voltage of -70mV (Fig 3F, Ctrl:6/12; cKO:1/12) compared to Ctrl PVHBDNF neurons. The threshold of the first AP initiated by a ramp current injection (1 nA/sec) from Rm was significantly increased in cKO neurons (Fig 3G-H). Because a wider shoulder and lower AP amplitude were observed when held at more depolarized potentials (Figure 3—figure supplement 1B), we compared the AP waveforms at a holding voltage of -60mV, close to the Rm of both genotypes. The AP of the cKO group exhibited a lower peak amplitude (measuring from the threshold to peak) (Fig 3I), a normal threshold (Fig 3J), and a normal half-width (Figure 3—figure supplement 1C). In summary, we found that both passive and AP properties of cKO PVHBDNF neurons were altered due to RAI1 loss. The Rm of cKO neurons shifted to a more positive voltage than the gating threshold of voltage-gated sodium channels (6/12 were more positive than -55mV), which was unfavourable for their transition from the inactivation state to the closed state. Combined with the higher AP threshold of cKO neurons, these changes contribute to a decreased excitability in cKO PVHBDNF neurons.

+ + +<italic>Rai1</italic> loss reduces intrinsic neuronal excitability and enhances inhibitory synaptic transmission of PVH<sup>BDNF</sup> neurons

(A) A representative image showing a patched PVHBDNF neuron labelled by BdnfCre-dependent tdTomato fluorescence signals.

(B-C) Representative traces of spontaneous firing of control (black) and cKO (red) neurons at a holding voltage of -60mV (B). The average spontaneous AP firing frequencies are shown in (C). *p<0.05, unpaired student’s t-tests.

(D) The resting membrane potentials of control (black) and cKO (red) PVHBDNF neurons. **p<0.01, unpaired student’s t-tests.

(E-F) The number of control (black) and cKO (red) PVHBDNF neurons that showed spontaneous firing (open) or are silent (solid) at resting membrane potentials (E) and a holding voltage of - 70mV (F).

(G) Representative traces of elicited APs in control (black) and cKO (red) PVHBDNF neurons responding to an inject current (bottom, blue), which ramps up from the resting membrane potentials at 1 nA per second for 500ms.

(H) The threshold of the first AP initiated by a current ramp in control (black) and cKO (red) PVHBDNF neurons. *p<0.05, unpaired student’s t-tests.

(I-J) The amplitude (I) and threshold (J) for the first AP firing at the holding voltage of -60mV in control (black) and cKO (red) PVHBDNF neurons. *p<0.05, unpaired student’s t-tests.

(K) The cumulative fractions of mIPSC frequency were measured in control (black) and cKO(red) PVHBDNF neurons. The average frequency of events from a 2-minute stable recording of each neuron is shown in the inset, unpaired student’s t-tests.

(L) The cumulative fractions of mIPSC amplitude of control (black) and cKO (red) PVHBDNF neurons. The average amplitude of events from a 2-min stable recording of each neuron is normalized to its capacitance and shown in the inset, unpaired student’s t-tests.

(M) The area under the curve of each mIPSC event from a 2-minute recording is averaged. Controls are shown in black, and cKOs are shown in red. unpaired t-tests, two-tailed. *p<0.05.

Data are shown as mean±SEM. Statistics: neuron numbers are in brackets (more than 3 mice/genotype)

+ + +

PVN neurons receive glutamatergic and GABAergic inputs and are tonically inhibited by GABAergic neurons (Colmers & Bains, 2018; Gao et al, 2017). We observed a slight rightward shift of the probability of miniature inhibitory postsynaptic current (mIPSC) frequency in cKO PVHBDNF neurons, although the average frequency (Fig 3K) was not significantly different between groups. The probability of mIPSC amplitude also showed a right shift without a significant change (Fig 3L, Figure 3—figure supplement 1D). However, we observed a significantly increased area under the curve (Fig 3M). Interestingly, we found that cKO PVHBDNF neurons had a smaller cellular diameter (Figure 3—figure supplement 1E). Overall, our data argue that Rai1 deletion decreases the excitability and AP firing of PVHBDNF neurons without significant changes in synaptic inhibitory transmission.

+ + +RAI1 is necessary for PVH<sup>BDNF</sup> neurons to regulate energy homeostasis +

Reduced PVHBDNF neuronal activity in cKO mice suggested that RAI1 loss in BDNF-producing PVH neurons might be sufficient to induce defective energy homeostasis in vivo. To test this hypothesis by deleting Rai1 only in PVHBDNF neurons but not other BDNF-producing cells, we performed PVHBDNF-specific Rai1 deletion using a CRISPR/Cas9 system packaged into AAV9. Two viruses were delivered: one encoding a reporter molecule (GFP) and three single guide RNAs targeting the Rai1 gene (sgRai1), and the other carrying Cre-dependent Staphylococcus aureus CRISPR-associated protein 9 (saCas9) (Fig 4A). Hereafter, our experiments focus on female SMS mice since they showed more consistent defects in fat distribution (Burns et al., 2010) and body weight regulation (Huang et al., 2016) than males. Bilateral administration of two AAVs into the PVH of female BdnfCre/+ mice (Exp group) at 3 weeks of age resulted in a significant loss of Rai1 protein expression in PVHBDNF neurons when compared to BdnfCre/+ mice injected with an AAV encoding the sgRai1 and GFP (Ctrl group) (Fig 4B-C, Figure 4—figure supplement 1A). Quantification of AAV-infected (GFP+) BDNF neurons showed that while the total number of BDNF+ (Figure 4—figure supplement 1B) and BDNF+ GFP+ (Figure 4— figure supplement 1C) cells were not different in the two groups, the percentage of AAV- infected BDNF+ cells that express RAI1 was significantly reduced in the Exp group (Fig 4C, Figure 4—figure supplement 1D-E). Moreover, the virus expression did not spread to the non-PVH neighboring nuclei (Figure 4—figure supplement 1F). Deleting Rai1 from the PVHBDNF neurons resulted in a significant increase in body weight beginning at 17 weeks of age (Fig 4D). Similar to the cKO (BdnfCre/+; Rai1fl/fl) mice, obesity in 26-week-old Exp mice was due to increased fat mass (Fig 4E) and more specifically, increased subcutaneous inguinal (S.C. ing) and epididymal white adipose tissues (eWAT) but not BAT (Fig 4F). We also found that eWAT adipocytes became hypertrophic (Fig 4G). Similar to the cKO mice, food intake did not differ between Exp and Ctrl groups (Fig 4H), likely due to the high level of satiety signals in the brain caused by increased blood leptin levels (Fig 4I). We also found that energy expenditure (Fig 4J), respiratory exchange rate (Figure 4—figure supplement 1G), and blood lipid parameters, including TG (Figure 4—figure supplement 1H), HDL (Figure 4—figure supplement 1I), LDL and VLDL levels (Figure 4—figure supplement 1J), were similar in Exp and Ctrl groups, consistent with the cKO data (Figure 2). Interestingly, Exp mice showed significantly less locomotor activity during the dark phase than the Ctrl mice (Fig 4K). A GTT indicated that Exp mice have a hyperglycemic profile (Fig 4L), accompanied by a slightly higher area under the curve for blood glucose levels (p=0.05) (Fig 4M). Insulin levels in Exp mice were also significantly elevated 15 minutes after glucose administration (Fig 4N).

+ + +Selective <italic>Rai1</italic> ablation in PVH<sup>BDNF</sup> neurons induces obesity +

(A) A schematic showing stereotaxic injection of adeno-associated viruses (AAVs) expressing Cas9 protein and single-guide RNAs targeting Rai1 (sgRai1) to delete Rai1 from the PVHBDNF neurons of BdnfCre/+ female mice. Scale bar = 50µm.

(B) Representative images showing that RAI1 immunoreactivity (cyan) was dramatically reduced in BdnfCre/+; Ai9 mice injected with both sgRai1 and DIO-saCas9 viruses (Exp group n=3, bottom row) into the PVH. In contrast, many PVHBDNF neurons in BdnfCre/+; Ai9 mice injected only with the sgRai1 virus showed RAI1 expression (Ctrl group n =3, top row).

(C) The percentage of PVHBDNF neurons co-expressing RAI1 and GFP (virus) was significantly reduced in the Exp group, indicating a successful Rai1 deletion. unpaired t-test, two-tailed, ****p<0.0001

(D) PVHBDNF neuron-specific Rai1 deletion induced body weight gain in Exp mice (Ctrl group: n=5, Exp group: n =5). Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, **p < 0.01, ****p < 0.0001.

(E) Body composition was measured with Echo-MRI, showing an increased fat mass deposition in 26-week-old Exp mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. ***p<0.001, ****p < 0.0001.

(F) Fat deposition analysis shows that the Exp group has significantly more subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) mass (grams). ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ****p < 0.0001.

(G) Representative images showing eWAT adipocyte hypertrophy of the Exp group (left). Scale bar = 500µm. Frequency distribution of adipocytes at each cellular size (right). ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, ***p<0.001.

(H) Exp and control mice showed similar food intake. ns indicates not significantly different, unpaired t-test, two-tailed.

(I) Blood leptin levels were significantly increased in the Exp mice. unpaired t-test, two-tailed, *p<0.05.

(J) Exp and control mice showed similar energy expenditure. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(K) Exp mice showed reduced locomotor activity in the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(L) Exp mice became hyperglycemic during the glucose tolerance test. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05.

(M) Measurement of the area under the curve (AUC) in Exp and Ctrl mice during the glucose tolerance test. unpaired t-test, two-tailed, p=0.05.

(N) Exp mice showed increased plasma insulin levels during the glucose tolerance test. Two-way ANOVA with Šidák’s multiple comparisons test, **p<0.01.

Data are shown as mean±SEM.

+ + +

While Exp mice showed a trend towards increased glucose levels in insulin tolerance tests, the change was not statistically different (Figure 4—figure supplement 1K). Altogether, these experiments indicate that conditional deletion of Rai1 from PVHBDNF neurons during adolescence impairs the regulation of body weight, adiposity, glucose tolerance, and locomotion. These changes recapitulate most phenotypes observed in cKO mice with Rai1 deleted in BDNF- producing cells from early development.

+ + +Pharmacologically enhancing neurotrophin downstream signalling partially alleviates obesity in SMS mice +

Our findings so far indicate that multiple pathways, including neurotrophin signalling, are dysregulated in SMS mice, and that Rai1 loss from BDNF-producing cells, including PVHBDNF neurons, impairs energy homeostasis in vivo. Next we sought to examine the contribution of BDNF dysfunction in SMS pathogenesis by pharmacologically activating downstream targets of the neurotrophin pathway.

+

We systematically treated mice with LM22A-4, a potential small molecule BDNF loop domain mimetic (Massa et al, 2010). To measure the bioavailability of LM22A-4 in the SMS brain, we simultaneously administered LM22A-4 intranasally (5 mg/kg body weight) and intraperitoneally (50 mg/kg body weight), using the highest doses used in previous studies (Canals et al, 2004; Chang et al, 2006; Gu et al, 2022; Kron et al, 2012; Ogier et al, 2007). We then harvested forebrain and hypothalamic tissues one or three hours after drug administration. The spectrometric analysis found a significantly higher LM22A-4 concentration in both brain regions one-hour post-treatment, while its concentration decreased three hours after treatment (Figure 5—figure supplement 1A). Consistent with the reduced neurotrophin signalling observed in our proteomic analysis, saline-treated SMS mice showed reduced phosphorylated-AKT (p-AKT; Fig 5A, Figure 5—figure supplement 1B, Figure 5—source data 1, Figure 5—figure supplement 1 source data 1) levels in the hypothalamus. LM22A-4 treatment significantly increased the levels of p-AKT in the SMS mice 1-hr post-treatment (Fig 5A, Figure 5—figure supplement 1B, Figure 5—source data 1, Figure 5—figure supplement 1 source data 1), indicating increased signalling downstream of the BDNF pathway.

+ + +LM22A-4 treatment partially alleviates obesity and stereotypic behaviour in adult SMS mice +

(A) Representative western blot showing p-AKT, total AKT, Histone 3 (H3, a loading control) levels in the hypothalamus of saline-treated Ctrl, saline-treated SMS and LM22A-4-treated SMS mice.

(B) Quantification showing deficits in p-AKT levels of the hypothalamus were reversed by LM22A-4 treatment in SMS mice. Unpaired t-test, two-tailed, *p<0.05.

(C-E) Body weight of saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. Saline or LM22A-4 treatment periods were highlighted in orange. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test. *p<0.05, **p < 0.01, ****p < 0.0001.

(F) Food intake (in grams) of saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05, **p < 0.01.

(G) Locomotor activity (number of beam breaks) for saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t- test, two-tailed, *p<0.05.

(H) Fat deposition analysis shows that the saline-injected SMS group has significantly more subcutaneous inguinal (S.C.ing) and epididymal white adipose tissue (eWAT) mass (grams) than saline-injected Ctrl mice. S.C.ing and eWAT mass in LM22A-4 injected mice is comparable to the saline-injected mice. Two-way ANOVA with Šidák’s multiple comparisons test. **p < 0.01, ***p<0.001, ****p < 0.0001.

(I) High-density lipoprotein (HDL) levels were restored to normal in LM22A-4 injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed, **p < 0.01, ***p<0.001.

(J) Insulin levels in saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice right before the start of the insulin tolerance test (time point 0). ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05.

(K) Insulin tolerance test performed on saline-injected control, saline-injected SMS, and LM22A-4 injected SMS mice, two weeks post daily administration of LM22A-4 or saline, shows reductions in the blood glucose levels of LM22A-4 treated SMS mice. * Shows significance between saline-injected Ctrl and saline-injected SMS group. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test. **p < 0.01.

(L) Stereotypic rearing behaviour was fully rescued in LM22A-4 treated SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed. *p<0.05, **p < 0.01.

Data are shown as mean±SEM.

+ + +

We next examined if chronic LM22A-4 treatment was sufficient to reverse or modify the progression of the obesity phenotype in adult SMS mice. Specifically, we treated SMS mice daily from 8 to 14 weeks of age, during which SMS mice began to gain significantly more weight. Saline-treated SMS mice (Fig 5C) were significantly heavier than saline-treated control mice from 9 weeks of age onward. Interestingly, LM22A-4-treated SMS mice showed a similar body weight to saline-treated control littermates until 13 weeks of age but became obese by 14 weeks of age (Fig 5D). Consistent with a partial rescue, LM22A-4-treated SMS mice were significantly less obese than saline-treated SMS mice throughout the treatment period (Fig 5E). Control mice treated with LM22A-4 did not show a further reduction in body weight (Figure 5—figure supplement 1C), indicating that the therapeutic effects were specific to SMS pathogenesis.

+

We measured food intake and found that LM22A-4 did not significantly rescue increased food intake in SMS mice (Fig 5F). SMS mice showed normal energy expenditure (Figure 5—figure supplement 1D) and respiratory exchange rate (Figure 5—figure supplement 1E), neither of which was altered by LM22A-4 treatment. Interestingly, we did observe a partial rescue of locomotor activity in LM22A-4-treated SMS mice that was not significantly different from saline-treated Ctrl or SMS mice (Fig 5G). We also found that the LM22A-4 treatment partially rescued fat deposition (Fig 5H) and completely rescued HDL levels (Fig 5I) in SMS mice. During an insulin tolerance test, we found that saline-treated SMS mice showed increased baseline insulin levels (Fig 5J) and blood glucose levels during the test (Fig 5K), both of which were partially rescued by LM22A-4 treatment.

+

In addition to a deficit in energy homeostasis, SMS mice showed two highly reproducible behavioural deficits: increased stereotypic rearing in the open field and decreased social dominance in the tube test (Huang et al, 2018; Rao et al, 2017). Saline-treated SMS mice showed significantly more rearing behaviour when compared to the saline-treated controls (Fig 5L). Interestingly, the LM22A-4 treatment reduced the repetitive rearing behaviour in SMS mice (Fig 5L). LM22A-4 treatment did not change repetitive behaviours in control mice (Figure 5— figure supplement 1F), suggesting the effect is specific to Rai1 haploinsufficiency. However, we did not observe a rescue of social dominance deficits in SMS mice (Figure 5—figure supplement 2A-E). Together, these data indicate that LM22A-4 treatment delayed the onset of obesity and fully rescued repetitive behaviour in SMS mice, suggesting that neurotrophin downstream signalling dysfunction contributes to changes in the mouse SMS brain linked to obesity and behavioural phenotypes.

+ + + +Discussion +

The hypothalamus and neurotrophin signalling in this region control body weight by regulating the intricate balance between energy intake and expenditure (Xu & Xie, 2016). The role of BDNF-producing neurons in human obesity disorders and how the activity of these cells is regulated remained poorly studied. We found that Rai1 deficiency, linked to Smith-Magenis syndrome (SMS), regulates BDNF expression and maintains intrinsic excitability in PVHBDNF neurons.

+

Consistent with this mechanism, we found that removing RAI1 from BDNF-producing neurons or PVHBDNF neurons results in imbalanced energy homeostasis and obesity. A small molecule drug LM22A-4 led to increased AKT phosphorylation and partially rescued locomotor activity, insulin intolerance, and HDL levels, while delaying the onset of obesity in SMS mice. These findings highlight a potential role for hypothalamic BDNF-producing neurons, specifically PVHBDNF neurons, in SMS pathogenesis and highlight BDNF signalling as a potentially druggable pathway to improve energy homeostasis defects and repetitive behaviours commonly observed in SMS patients (Burns et al., 2010; Moss et al, 2009).

+

The involvement of different PVH subtypes in human obesity is only beginning to be elucidated. Dysfunction of oxytocin PVH neurons is linked to the Prader-Willi syndrome (Lee et al, 2012), another genetic disorder associated with obesity. Since Rai1 expression is enriched in PVHBDNF neurons but not PVHOxytocin neurons, our data suggest that distinct pathophysiological mechanisms underlie SMS and Prader-Willi syndrome.

+

The conditional knockout analysis allowed us to precisely map the cell-specific origin of neuropathological phenotypes, such as SMS-like symptoms in mice. Our previous work found that deleting Rai1 from cortical excitatory neurons using an Emx1Cre allele (Gorski et al, 2002), from the GABAergic neurons using a Gad2Creallele (Taniguchi et al, 2011), or from astrocytes using a GfapCre allele (Garcia et al, 2004) did not induce SMS-like obesity (Huang et al., 2016). In contrast, obesity could be induced by ablating Rai1 from Sim1-lineage neurons and, to a lesser extent, SF1-lineage neurons (Huang et al., 2016). This analysis suggests that Rai1 expression in PVH is more important than in the ventromedial nucleus of the hypothalamus (VMH) for body weight regulation. Interestingly, Although BDNF is required in the VMH and DMH to regulate body weight (Unger et al., 2007), embryonic deletion of Bdnf from the SF1-lineage populations including the VMH did not result in obesity (Kamitakahara et al, 2016; Yang et al, 2016). By contrast, deleting Bdnf from Sim1-lineage neurons including the PVH resulted in a more severe obesity phenotype (An et al., 2015). This is consistent with the idea that BDNF expression in PVH is critical for energy homeostasis. We found that Rai1 expression in BDNF-producing cells regulates energy homeostasis in a sexually dimorphic manner. Specifically, female cKO mice showed normal food intake, decreased energy expenditure, and reduced locomotor activity. In contrast, male cKO mice showed increased food intake with no changes in locomotor activity. Intriguingly, energy expenditure was increased in male cKO mice, likely due to the higher demand to maintain normal activity levels in obese mice. In both sexes, cKO mice showed an increased fat mass and adiposity with no change in lean mass, highlighting an essential role for Rai1 in regulating fat deposition. Our data also indicate that BDNF-producing cells are only partially responsible for obesity in SMS: male and female cKO mice become overweight by 15 weeks, in contrast to SMS mice which become obese by 9 weeks of age, suggesting that non-BDNF neurons contribute to SMS phenotypes.

+

Notably, mice lacking one copy of Rai1 in the BDNF-producing cells do not exhibit obesity, whereas SMS patients and SMS mice show pronounced obesity (Burns et al., 2010; Huang et al., 2016; Smith et al, 2005). This indicates that although reduced Bdnf expression and BDNF-producing neurons contribute to regulating body weight, additional molecular changes and other hypothalamic populations also play important roles in regulating body weight homeostasis in SMS. Our RPPA data suggest that mTOR signalling is also misregulated in addition to the reduced activation of the neurotrophin downstream cascades. Hypothalamic mTORC1 is crucial to regulate glucose release from the liver, peripheral lipid metabolism, and insulin sensitivity (Burke et al, 2017; Caron et al, 2016; Smith et al, 2015), while mTORC2 regulates glucose tolerance and fat mass (Kocalis et al, 2014). How the impaired mTOR signalling contributes to energy homeostasis defects in SMS and the therapeutic potential of targeting this pathway to treat SMS-related obesity remains unclear and warrants future investigation.

+

What additional RAI1-dependent hypothalamic cell types residing in brain regions other than PVH regulate obesity in SMS? Other important cell types such as TRKB neurons within the PVH (An et al, 2020) and several RAI1-expressing hypothalamic nuclei including the arcuate nucleus, ventromedial nucleus of the hypothalamus (VMH), and lateral hypothalamus all play important roles in regulating energy homeostasis. POMC- and AGRP-expressing neurons within the arcuate nucleus are known to regulate food intake and glucose and insulin homeostasis (Quarta et al, 2021; Vohra et al, 2022). Therefore, Rai1 function in these neurons could contribute to obesity in SMS, a topic that awaits future investigation.

+

We found the metabolic phenotypes induced by deleting Rai1 from BDNF-producing cells were specifically due to Rai1 deletion in PVHBDNF neurons. Therefore, our study focused on PVHBDNF neurons. Electrophysiological experiments showed that most wild-type PVHBDNF neurons display high-frequency spontaneous AP firing (1-10 Hz) at the resting membrane potential. Most PVHBDNF neurons fire sharp APs held at a hyperpolarized potential and show repetitive overshoot during the current ramp test. Loss of Rai1 significantly reduced the intrinsic neuronal excitability of PVHBDNF neurons. Because GABAergic Rai1 loss did not induce obesity (Huang et al., 2016) and most BDNF-producing neurons are glutamatergic (Canals et al, 2001). The ionic changes in PVHBDNF neurons that precisely explain these excitability changes await future investigations. Different groups of PVH neurons show phasic bursting or slowly inactivating delayed rectifier potassium conductance, which are properties mediated by different potassium channels (Lee et al., 2012). Given that our previous RNA-seq experiment found that hypothalamic Rai1 loss induces misexpression of potassium channel subunits Kcnc2 and Kcnj11 (Huang et al., 2016), RAI1-dependent potassium channels are likely involved in altered neuronal activity. Furthermore, we found fewer low threshold depolarization events in cKO PVHBDNF neurons during the hyperpolarized voltage step, suggesting a potential involvement of HCN channels.

+

Our previous work found that Rai1 loss in the dentate gyrus resulted in neuronal hyperexcitability (Chang et al, 2022b). Here we demonstrated that Rai1 loss in the hypothalamus resulted in neuronal hypoactivity, suggesting that RAI1 regulates neuronal excitability in a cell type-specific manner. Reduced PVHBDNF neuronal activity upon Rai1 loss could potentially explain obesity, consistent with a previous report that found chemogenetic activation of PVHBDNF neurons promotes negative energy balance by decreasing feeding and increasing energy expenditure (Wu & Xu, 2022).

+

It is plausible that RAI1 regulates the expression of genes encoding inward rectifier K+ channels, which regulate neuronal activity and potentially energy homeostasis. For instance, KIR6 (a family of ATP-sensitive potassium channels, KATP) is widely expressed in the hypothalamus. Deleting the hypothalamic KIR6.2 subunit impairs KATP channel function and glucose tolerance (Miki et al, 2001; Parton et al, 2007). Moreover, reduced expression of hypothalamic GIRK4 (encoding an inwardly rectifying potassium channel) causes obesity (Perry et al, 2008). GABAergic neurotransmission from arcuate AGRP-expressing neurons to the PVH neurons is important to increase appetite by favouring hyperphagia (Atasoy et al, 2012). Disrupting the composition of these ion channels could contribute to reduced PVHBDNF neuronal firing, which awaits further investigations.

+

Female mice with a conditional knockout of Rai1 from BDNF-producing neurons do not display a noteworthy difference in food intake. Conversely, their male counterparts exhibit a significant increase in food intake. Although SMS individuals of both genders tend to overeat, male patients who are obese show significantly higher food consumption than their female counterparts (Gandhi et al., 2022). This observation raises the possibility that Rai1 regulates eating behaviours through multiple cell types in the hypothalamus and that a male-specific involvement of BDNF-producing neurons in regulating food intake, potentially provides a neurobiological basis for the observed pattern in SMS patients (Gandhi et al., 2022). In future work, it would be interesting to dissect the sex-specific role of RAI1 in regulating the functional development of PVHBDNF neurons. In addition to a role for RAI1 in regulating neuronal activity, RAI1 may regulate other aspects of PVHBDNF neuronal function, including projections to other brain regions to regulate energy homeostasis. In this regard, it would be interesting to dissect how RAI1 controls the circuit connectivity of PVHBDNF neurons and how Rai1 loss in PVHBDNF neurons affects nearby PVHTRKB neurons that receive BDNF signalling and are linked to body weight regulation (An et al., 2020).

+

At the molecular level, our RPPA data show that Rai1 depletion disrupted multiple intracellular signalling pathways and resulted in the hypoactivation of specific phospho-signalling proteins. Multiple early studies support the notion that Rai1 directly promotes Bdnf expression. First, hypothalamic Bdnf mRNA levels were reduced in SMS mice (Burns et al., 2010; Javed et al., 2021) or mice with Rai1 conditional deletions (Huang et al., 2016). Second, the RAI1 protein directly binds to a promoter region of Bdnf (Huang et al., 2016). Finally, when Rai1 expression was induced using CRISPR activation, Bdnf mRNA levels were significantly elevated (Chang et al, 2022a).

+

To lend further support, here we show that SMS mice have reduced hypothalamic BDNF protein levels, as well as reduced activation of a TRKB downstream target AKT. Reduced AKT activity is frequently associated with insulin resistance and obesity in mice (Shao et al, 2000) and children (Su et al., 2021). We also found that obesity, increased HDL levels, and insulin insensitivity in SMS mice could be partially alleviated using a small molecule drug (LM22A-4) that promotes AKT phosphorylation. LM22A-4 penetrates the blood-brain barrier reasonably well and has been shown to ameliorate symptoms of other neurological disorders associated with BDNF dysfunction including Rett syndrome, Huntington’s disease, Dravet syndrome, and refractory neonatal seizures (Canals et al., 2004; Chang et al., 2006; Gu et al., 2022; Kron et al., 2012; Ogier et al., 2007). We recognize that while several in vivo studies have demonstrated the potential of LM22A-4 in targeting neurotrophin downstream signalling (Kron et al, 2014; Li et al, 2017), an in vitro analysis failed to demonstrate the ability of LM22A-4 to activate TRKB directly (Boltaev et al, 2017). Therefore, the precise mechanism by which LM22A-4 enhances AKT cascades in the mammalian brain remains unclear and awaits further investigations. In the hypothalamus of SMS mice, LM22A-4 could indirectly engage neurotrophin downstream PI3K- AKT pathway through the G protein-coupled receptor-dependent transactivation of the TRKB receptor (Domeniconi & Chao, 2010) or other unknown mechanisms. Moreover, while LM22A-4 may have potential side effects, we found that wild-type mice treated with LM22A-4 did not show a further decrease in body weight, suggesting limited side effects regarding body weight regulation.

+

The partial rescue of SMS phenotypes that we observed contrasts with our previous findings that found (1) AAV-mediated BDNF overexpression in PVH at 3 weeks of age and (2) genetically overexpressing BDNF from early development could fully prevent the development of obesity in SMS mice (Javed et al., 2021). Therefore, further increasing TRKB signalling activation and earlier intervention might improve therapeutic outcomes. Alternatively, our RPPA data showed that in addition to disruption of neurotrophin signalling, multiple pathways including mTOR were downregulated in SMS. Increasing hypothalamic mTOR activity has been shown to decrease food intake and body weight (Cota et al., 2006). Therefore, it is possible that enhancing both mTOR and TRKB signalling could achieve a more robust rescue of obesity in SMS mice. Nevertheless, these data support the concept that drugging TRKB signalling may improve stereotypical repetitive behaviour and partially alleviate metabolic functions in a mouse model of SMS. Delayed symptom onset could provide a window of opportunity for other interventions to further modify disease progression, an area requiring further investigation.

+

This study has several limitations. First, whether the PVHBDNF neurons depend on Rai1 to regulate adaptive thermogenesis awaits future analysis. It is known that a subset of posterior PVHBDNF neurons project to the spinal cord and regulate thermogenesis (An et al., 2015). However, we found the weights of brown adipocytes were similar in control and cKO mice. Moreover, SMS mice have normal respiratory exchange rates and there are currently no reports of defective adaptive thermogenesis in SMS patients. This suggests that adaptive thermogenesis is not impacted in SMS. Another limitation is that we did not test different doses, durations, and time points for LM22A-4 treatment. The precise mechanisms by which LM22A-4 improves body weight, metabolic function, and repetitive behaviours remain unknown. Nevertheless, our work identifies RAI1 as a novel regulator of the neuronal excitability of PVHBDNF neurons. This regulation is critical for regulating food intake and energy expenditure. Our evidence suggests that BDNF-producing neurons are involved in SMS pathogenesis and that enhancing BDNF- TRKB signalling should be explored as a future therapeutic avenue for obesity and metabolic conditions associated with reduced neurotrophin signalling (Xu & Xie, 2016).

+ + +Materials and Methods + +Animal Management +

The animal protocol (MUHC-8127) was approved by the animal care committee of the Montreal General Hospital and was performed according to the guidelines of the Canadian Council on Animal Care. Mice were group-housed in 12-hour light/12-hour dark cycles with an ad libitum supply of regular chow and water. The exception was in the fasting studies when the animals were food deprived for 5 hours and in CALMS and food intake studies when animals were single-housed in a 12-hour light/12-hour dark cycle with an ad libitum supply of regular chow and water. All mice were maintained in a C57Bl/6J background. BdnfCre/Cre (RRID: IMSR_JAX:030189), Rosa26Ai9/Ai9(RRID: IMSR_JAX:007909), and ActbCre/Cre (RRID: IMSR_JAX:019099) mice were purchased from the Jackson laboratory. The Rai1fl/fl (RRID: IMSR_JAX:029103) strain was previously generated and characterized by us (Huang et al., 2016). SMS mice (Rai1+/-: whole-body germline Rai1 heterozygous knockout that genetically mimics SMS) were generated by crossing germline ActbCre/Cre mice with Rai1fl/+. To generate conditional knockout mice, BdnfCre/+; Rai1fl/+ mice were crossed with Rai1fl/+ to create control and experimental groups from the same littermate. 7 mice with malocclusion or lesions were excluded from the study: 4 with malocclusion (2 females and 2 males) and 3 with lesions (3 females).

+ + +Metabolic profiling +

Metabolic profiling was conducted in three experimental conditions: male and female conditional knockout mice, female virus-injected BdnfCre/+ mice, and female drug- or saline-injected SMS and control mice. Metabolic profiling included body weight and body composition analysis, food intake measurement, indirect calorimetry for locomotion activity, energy expenditure (EE), respiratory exchange rate (RER), measurement of serum lipids and leptin levels, adipose tissue histology analysis, and glucose and insulin tolerance tests. All experiments were performed as previously described (Javed et al., 2021). A brief description is detailed below.

+ +Body weight and body composition analysis +

The body weight of animals was measured weekly from week 3 (weaning age) onward. After the 25th week, mice were subjected to body composition analysis. Body composition, including total fat and lean mass, was assessed using a nuclear EchoMRI whole-body analyzer. Toward the end of the experiments, animals were euthanized to collect brown adipose, visceral, subcutaneous (inguinal) and ependymal white adipose tissues and weighed using an analytical scale (Sartorius).

+ + +Food Intake Measurement +

Mice were single-housed for food intake measurement. Food intake was measured over a one-week period during which food was weighed prior to placement in the cage with minimal bedding. Each morning, the remaining food inside the cage was weighed. The amount of total food consumed was averaged over 7 days.

+ + +Indirect Calorimetry +

A Comprehensive Lab Animal Monitoring System (CLAM) (Columbus instruments) was used to measure the RER and EE (normalized over lean mass). Animals were singled-housed in the CLAMS apparatus at 21◦C (70◦F) in a light–dark cycle matching their housing conditions for 24 hours (acclimation), followed by 48 hours of measurement.

+ + +Serum lipid level determination +

A fluorescence-based assay kit (Abcam ab65390) was used to measure serum lipid levels. The manufacturer’s instructions were followed except that we used half-area black 96-well microplates. Fluorescence measurements at 535/587 nm (Ex/Em) were recorded with the Ensight instrument from PerkinElmer. A new standard curve ranging from 0 to 10 μg/ml was produced for each microplate to accurately determine lipid levels in serum samples. Standards and samples were loaded and analyzed twice.

+ + +Histology of adipose tissue +

Ependymal white adipose tissues (WAT) were excised immediately after the animals were euthanized and fixed with 10% formalin in PBS. The fixed tissues were dehydrated, embedded in paraffin, and sectioned. Two 5-micrometre thick sections were taken from each animal and stained with hematoxylin and eosin. Images of sections were then captured using an Aperio ScanScope CS slide scanner at 20X optical magnification, and the resulting images were saved in jpeg format using the ImageScope software. The images were manually cleaned to remove non-adipose tissues such as lymph nodes. An automated method using the Adiposoft plugin version 1.16 was utilized to determine adipocyte size and number from the jpeg images in Fiji version 2.1.0/1.53c. The Adiposoft settings used excluded cells on edges and set a minimum and maximum cell diameter of 20 and 300 μm, respectively. An experimental value of 2.49 μm per pixel was determined using the Aperio scale bar and used for Adiposoft. To ensure accuracy, two images from each animal were analyzed blindly, examining more than 1600 cells per mouse and more than 9100 cells per condition. Adipocyte size frequencies were calculated using the frequency function in Excel.

+ + +Glucose and insulin tolerance tests +

For the glucose tolerance test, experimental mice were food deprived for 5 hours with ad libitum water supply. A bolus of glucose (1.5g/kg of lean body weight) was then administered intraperitoneally and glycemia was measured from tail vein blood samples using the Accu-chek Performa glucometer at T0 (before the injection) as well as at 15, 30, 45, 60, 90 and 120 min (after the injection). Tail vein blood samples were collected via a capillary for insulin assays at T0, 15 and 30 min. For insulin tolerance tests, experimental mice were food-deprived for 5 hours with ad libitum access to water. A bolus of insulin (1 U/kg of lean body weight) was administered via an IP injection and glycemia was measured from blood sampled at the tail vein using an Accu-chek Performa glucometer at T0 (before the injection), as well as at 15, 30, 45, 60, 90 and 120 min (after the injection). Tail vein blood samples were collected via a capillary for insulin assays at T0.

+ + + +Stereotaxic Injections +

BdnfCre/+ mice were anesthetized using isoflurane, and the ears were barred onto a stereotaxic device (David Kopf Instruments). Adeno-associated viruses (AAVs, volume = 250 nl) were bilaterally injected into the PVH using the following coordinates relative to the Bregma: -0.6 mm anteroposterior (AP), ±0.35 mm mediolateral (ML) and -5.52 mm dorsoventral (DV). A description of the AAV constructs and titre used is detailed below:

+

AAV9-CMV7-DIO-saCas9 (Viral title: >1 × 10^13), Vector Biolab

+

AAV9-CMV7-sgRai1-GFP (Viral title: >1 ×10^9), Applied Biological Materials Inc. Three sgRNAs were pooled to target Rai1 (sgRNA1: TTCCTCGCCAGAGTAGCGCC, sgRNA2: CCCAGCCTCATGATAGGCCG, sgRNA3: CCCAGCCTCATGATAGGCCG).

+ + +Immunostaining of mouse brain tissues +

Immunostaining experiments were performed as previously described (Huang et al, 2012). Briefly, mice were anesthetized and perfused with 1X PBS and 4% formaldehyde, and the brains were dissected and fixed in 4% formaldehyde overnight. Fixed brains were transferred to 30% sucrose solution and subsequently embedded in the OCT solution. The brains were then sectioned (40µm), washed in PBS, permeabilized with 0.5% Triton X-100, and incubated with 10% normal donkey serum in PBS for 2 hours. The sections were then incubated overnight with primary antibodies (Rai1: 1:500 dilution, RRID:AB_2921229 (Huang et al., 2016); BDNF: 1:500, RRID:AB_10862052, Abcam ab108319; p-TRKB: 1:500 dilution, RRID:AB_2721199, Millipore ABN1381; oxytocin: 1:1000 dilution, RRID:AB_11212999, Millipore MAB5296) at 4°C on a shaker. The following day, sections were washed with 1X PBS twice for 10 minutes and incubated with a secondary antibody (1:1000 dilution) for 3 hours at room temperature. Sections were washed with 1X PBS once before mounting on glass slides with the DAPI- mounting (DAPI-Fluoromount-G, SouthernBiotech) and then imaged using a confocal microscope (Olympus FV-1000 confocal laser scanning microscope) with 20x or 40x objectives (NA 0.85 and 1.3, respectively). Z-stacks were taken with a step size of 1.0 µm with a 1024 x 1024 resolution. Images were acquired as a stack with at least 13 optical sections. Image analysis was performed using the Fiji software.

+ + +Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) +

qRT-PCR was performed using mRNAs isolated from hypothalamic tissue samples. After the isolation of total RNA, mRNAs were reverse-transcribed with the SuperScript III First-Strand Synthesis System (Thermo-Fisher). Quantitative PCR reactions were conducted using SsoFast EvaGreen Supermix on a Bio-Rad qPCR system.

+ + +Enzyme-linked immunosorbent assay (ELISA) +

ELISA was performed using hypothalamic tissues from 7-week-old SMS mice to assess the mature BDNF protein levels using a commercially available BDNF ELISA kit (Mature BDNF Rapid ELISA Kit: Human, Mouse, Rat).

+ + +Western blotting analysis +

Mice were quickly decapitated, brains were rapidly removed, and hypothalamic tissues were collected, weighed, and snap-frozen with liquid nitrogen. Hypothalamic tissues were lysed using a tissue lysis buffer containing RIPA buffer (5M NaCl, 1M PH 8.0 Tris HCl, 1% NP.40, 10% C24H39NaO4, and 20% SDS) and the Halt™ Protease and Phosphatase Inhibitor Cocktail (Thermo-Fisher). Lysates were first sonicated on ice for 15 seconds and centrifuged at 10000 RPM at 4°C for 10 minutes. The supernatant was collected in a separate Eppendorf tube, and a BCA assay was performed to quantify protein levels. Proteins were then separated using SDS- PAGE gel and transferred onto nitrocellulose membranes. To determine the phosphorylation levels of AKT, membranes were cut and blocked in 5% milk in tris-buffered saline (TBS) for 1 hour and then incubated overnight at 4°C with anti-p-AKT (Ser473) antibody (1:1000 dilution, RRID: AB_2315049, Cell Signaling Technology 4060T) and anti-H3 antibody (internal control: 1:1000 dilution, RRID: AB_302613, Abcam ab1791). The next day, membranes were washed with TBST three times for five minutes before incubating with an HRP-conjugated secondary antibody (1:10000 dilution, Thermo Fisher 200-403-095S). Membranes were later stripped by incubating in 0.5M NaOH for 10 min and then re-probed for total Akt levels by incubating with anti-Akt (C67E7) antibody (1:1000 dilution, Cell Signaling Technology 4691T). The density of each immunoreactive phospho-protein band was expressed as a fraction of the band for the total AKT protein in the same lane.

+ + +Reverse-phase protein array (RPPA) +

A reverse-phase protein array (RPPA) was used for precise high-throughput quantification of total and phospho-protein levels in different signalling pathways. Protein lysates from hypothalamic tissues were extracted and transferred to 384-well microarray plates to print individual slides. The slides were then labelled with validated antibodies recognizing total protein and specific phosphorylation sites. SYPRO Ruby staining was used to measure total protein levels. The data obtained from RPPA were normalized for individual samples. Each antibody’s raw signal intensity (SI) was determined by subtracting the background from the antibody-specific signals during image analysis. The normalized antibody SI was expressed using the formula N = A - C/T x M, where N is the normalized SI, A is the raw antibody SI, C is the negative control SI, T is the SI of total protein, and M is the median SI of total protein from a spot within the same sample group. Antibodies with SI less than 200 were filtered out, and quality control (QC) scores were established to discard poor-quality reactions. The coefficient of variation (CV) score was determined for each antibody and defined as the percentage of samples on a slide with a CV < 20%. Antibodies with a median technical triplicate CV >20% were also excluded from further analysis. T-tests were performed for each antibody to identify differentially expressed proteins in the SMS compared to control groups. PCA plots were generated for all samples, annotated with their experimental group, and hierarchical clustering was performed using a Python analysis tool. The vertical axis of the dendrogram represents the dissimilarity (measured as distance) between protein expressions, and the horizontal axis represents the individual test samples. The colour code (ranging from red to yellow to green) specifies the expression levels of different proteins, where red indicates low expression, yellow indicates intermediate expression, and green indicates high expression. Enrichment analysis on the differentially expressed proteins was performed using an online data resource string: version 11.5.

+ + +Electrophysiology of acute brain slices +

Mouse brains were removed and placed in ice-cold carbonated slicing artificial cerebrospinal fluid (aCSF), which contained (in mM) 124 NaCl, 3 KCl, 1.4 NaH2PO4, 26 NaHCO3,1.3 Mg SO4, 10 D-glucose, 0.5 CaCl2, and 3.5 MgCl2. Coronal sections (300 µm) were cut on a vibratome. Slices were allowed to recover at 32 °C for 40 minutes and then at 30 °C for 30 minutes to 6 hours in the recording medium, carbonated aCSF (124 NaCl, 3 KCl, 1.4 NaH2PO4, 26 NaHCO3,1.3 Mg SO4, 10 D-glucose, 2.4 CaCl2, in mM, ∼300 mOsm). Signals were recorded with a 5× gain, low-pass filtered at 2 kHz, digitized at 10 kHz (Molecular Devices Multiclamp 700B) and analyzed with pClamp 11 (Molecular Devices). Whole-cell recordings were made using 3–5 MΩ pipettes filled with an internal solution that contained (in mM) 131 potassium gluconate, 8 NaCl, 20 KCl, 2 EGTA, 10 HEPES, 2 MgATP and 0.3 Na3GTP (pH 7.2, with KOH, 300–310 mOsm with sucrose). For miniature current recording, the internal solution contained (in mM) 130 gluconate acid, 8 CsCl, 1 NaCl, 2 EGTA, 0.8 CsOH, 10 HEPES, 2 MgATP and 0.3 Na3GTP (pH 7.2, with CsOH, 300–310 mOsm adjusted with sucrose). mEPSCs were recorded in the presence of 1 µM TTX at a holding potential of - 70 mV. mIPSCs were recorded at a holding potential of +10 mV. The mIPSCs were analyzed from events during a 2-minute continuous recording that contained > 15 events/neuron. Additional calculations, statistical analysis, and plotting were performed in Microsoft Excel or Prism (GraphPad).

+ + +Spectrometric analysis of LM22A-4 levels in the brain +

We assessed the efficacy of LM22A-4, a small molecule pharmacological agent, in 8-week-old SMS and control mice. LM22A-4, chemical name N1, N3, N5tris (2hydroxyethyl) 1, 3, 5 benzenetricarboxamide (Molecular formula: C15H21N3O6, Formula weight: 339.3), was manufactured by Cayman Chemical with purity specification >98%. A stock solution was made by first dissolving LM22A-4 in DMSO (solubility =30mg/ml) and then diluting it with 0.9% isotonic sterilized saline solution (Sigma). To maximize brain concentrations, LM22A-4 was administered both intranasally (5mg/kg) and intraperitoneally (50mg/kg), as previously reported (Simmons et al, 2013). The bioavailability of this drug in the hypothalamus using these doses and routes of administration was assessed with the reverse phase LC triple-quadrupole tandem mass spectroscopic detection (LC-MS/MS) by Absorption Systems. Mice were euthanized, and hypothalamic tissue was collected at either 1 h or 3 h post-injection (n= 4/timepoint). Known amounts of LM22A-4 were added to the blank sample to generate a standard curve to verify test accuracy.

+ + +LM22A-4 treatment +

Mice were divided into four groups: Saline-injected control, Saline-injected SMS, LM22A-4- injected SMS, and LM22A-4-injected control mice. Mice were injected daily, intra-nasally (5mg/kg) and intraperitoneally (50mg/kg), for 7 weeks. Body weight was measured between 8 and 14 weeks of age. Food intake was measured from 14 to 15 weeks of age. After 15 weeks of age, mice were euthanized to collect brown adipose, visceral, subcutaneous (inguinal) and ependymal white adipose tissues and weighed using an analytical scale (Sartorius). Hypothalamic brain tissues were collected for western blot analysis. Metabolic profiling was performed on a separate cohort of mice injected for two weeks prior to the start of experiments as well as at specific weeks during the experiment.

+ + +Vertical rearing +

To measure vertical repetitive rearing, mice were taken into the testing room and allowed to habituate for 1 h prior to testing. During the test, mice were placed in the center of a 45 cm (W) × 45 cm (D) × 40 cm (H) square arena, and their movement was recorded for 10 minutes.

+ + +Statistical analysis +

The sample sizes used were based on our previous studies (Huang et al., 2016; Javed et al., 2021). The data were subjected to statistical analysis for significance by using GraphPad Prism 0.9. Significance was defined as having a p-value < 0.05. The significance levels were as follows: * p< 0.05, **p<0.01, ***p<0.0001, **** p<0.0001. The statistical tests used for each analysis (e.g., Student’s t-test, analysis of variance [ANOVA]) are indicated in the text and legends.

+ + + +Data Availability +

This study includes no data deposited in external repositories.

+ + + + +Acknowledgements +

Sehrish Javed was supported by a Fonds de Recherche du Québec - Santé (FRQS) doctoral award. This work was supported by the Rare Diseases: Models & Mechanisms Network and the Smith-Magenis syndrome Research Foundation.

+ + +References +Abad C, Cook MM, Cao L, Jones JR, Rao NR, Dukes-Rimsky L, Pauly R, Skinner C, Wang Y, Luo F et al (2018) A Rare De Novo RAI1 Gene Mutation Affecting BDNF-Enhancer-Driven Transcription Activity Associated with Autism and Atypical Smith-Magenis Syndrome Presentation. Biology (Basel) 7 +Alaimo JT, Barton LV, Mullegama SV, Wills RD, Foster RH, Elsea SH (2015) Individuals with Smith-Magenis syndrome display profound neurodevelopmental behavioral deficiencies and exhibit food-related behaviors equivalent to Prader-Willi syndrome. Res Dev Disabil 47: 27–38 +An JJ, Kinney CE, Tan JW, Liao GY, Kremer EJ, Xu B (2020) TrkB-expressing paraventricular hypothalamic neurons suppress appetite through multiple neurocircuits. Nat Commun 11: 1729 +An JJ, Liao GY, Kinney CE, Sahibzada N, Xu B (2015) Discrete BDNF Neurons in the Paraventricular Hypothalamus Control Feeding and Energy Expenditure. Cell Metab 22: 175–188 +Atasoy D, Betley JN, Su HH, Sternson SM (2012) Deconstruction of a neural circuit for hunger. Nature 488: 172–177 +Balthasar N, Dalgaard LT, Lee CE, Yu J, Funahashi H, Williams T, Ferreira M, Tang V, McGovern RA, Kenny CD et al (2005) Divergence of melanocortin pathways in the control of food intake and energy expenditure. Cell 123: 493–505 +Boltaev U, Meyer Y, Tolibzoda F, Jacques T, Gassaway M, Xu Q, Wagner F, Zhang YL, Palmer M, Holson E et al (2017) Multiplex quantitative assays indicate a need for reevaluating reported small-molecule TrkB agonists. Sci Signal 10 +Boot E, Linders CC, Tromp SH, van den Boogaard MJ, van Eeghen AM (2021) Possible underreporting of pathogenic variants in RAI1 causing Smith-Magenis syndrome. Am J Med Genet A 185: 3167–3169 +Burke LK, Darwish T, Cavanaugh AR, Virtue S, Roth E, Morro J, Liu SM, Xia J, Dalley JW, Burling K et al (2017) mTORC1 in AGRP neurons integrates exteroceptive and interoceptive food-related cues in the modulation of adaptive energy expenditure in mice. Elife 6 +Burns B, Schmidt K, Williams SR, Kim S, Girirajan S, Elsea SH (2010) Rai1 haploinsufficiency causes reduced Bdnf expression resulting in hyperphagia, obesity and altered fat distribution in mice and humans with no evidence of metabolic syndrome. Hum Mol Genet 19: 4026–4042 +Canals JM, Checa N, Marco S, Akerud P, Michels A, Perez-Navarro E, Tolosa E, Arenas E, Alberch J (2001) Expression of brain-derived neurotrophic factor in cortical neurons is regulated by striatal target area. J Neurosci 21: 117–124 +Canals JM, Pineda JR, Torres-Peraza JF, Bosch M, Martin-Ibanez R, Munoz MT, Mengod G, Ernfors P, Alberch J (2004) Brain-derived neurotrophic factor regulates the onset and severity of motor dysfunction associated with enkephalinergic neuronal degeneration in Huntington’s disease. J Neurosci 24: 7727–7739 +Caron A, Labbe SM, Lanfray D, Blanchard PG, Villot R, Roy C, Sabatini DM, Richard D, Laplante M (2016) Mediobasal hypothalamic overexpression of DEPTOR protects against high-fat diet-induced obesity. Mol Metab 5: 102–112 +Chang HC, Lee YJ, Javed S, Haque M, Chang YT, Lin YC, Oram C, Huang WH (2022a) rAAV-CRISPRa therapy corrects Rai1 haploinsufficiency and rescues selective disease features in Smith-Magenis syndrome mice. J Biol Chem: 102728 +Chang Q, Khare G, Dani V, Nelson S, Jaenisch R (2006) The disease progression of Mecp2 mutant mice is affected by the level of BDNF expression. Neuron 49: 341–348 +Chang YT, Kowalczyk M, Fogerson PM, Lee YJ, Haque M, Adams EL, Wang DC, DeNardo LA, Tessier-Lavigne M, Huguenard JR et al (2022b) Loss of Rai1 enhances hippocampal excitability and epileptogenesis in mouse models of Smith-Magenis syndrome. Proc Natl Acad Sci U S A 119: e2210122119 +Collaboration NCDRF (2017) Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet 390: 2627–2642 +Colmers PLW, Bains JS (2018) Balancing tonic and phasic inhibition in hypothalamic corticotropin-releasing hormone neurons. J Physiol 596: 1919–1929 +Cota D, Proulx K, Smith KA, Kozma SC, Thomas G, Woods SC, Seeley RJ (2006) Hypothalamic mTOR signaling regulates food intake. Science 312: 927–930 +Domeniconi M, Chao MV (2010) Transactivation of Trk receptors in spinal motor neurons. Histol Histopathol 25: 1207–1213 +Edelman EA, Girirajan S, Finucane B, Patel PI, Lupski JR, Smith AC, Elsea SH (2007) Gender, genotype, and phenotype differences in Smith-Magenis syndrome: a meta-analysis of 105 cases. Clin Genet 71: 540–550 +Gandhi AA, Wilson TA, Sisley S, Elsea SH, Foster RH (2022) Relationships between food-related behaviors, obesity, and medication use in individuals with Smith-Magenis syndrome. Res Dev Disabil 127: 104257 +Gao H, Molinas AJR, Miyata K, Qiao X, Zsombok A (2017) Overactivity of Liver-Related Neurons in the Paraventricular Nucleus of the Hypothalamus: Electrophysiological Findings in db/db Mice. J Neurosci 37: 11140–11150 +Garcia AD, Doan NB, Imura T, Bush TG, Sofroniew MV (2004) GFAP-expressing progenitors are the principal source of constitutive neurogenesis in adult mouse forebrain. Nat Neurosci 7: 1233–1241 +Gorski JA, Talley T, Qiu M, Puelles L, Rubenstein JL, Jones KR (2002) Cortical excitatory neurons and glia, but not GABAergic neurons, are produced in the Emx1-expressing lineage. J Neurosci 22: 6309–6314 +Gray J, Yeo GS, Cox JJ, Morton J, Adlam AL, Keogh JM, Yanovski JA, El Gharbawy A, Han JC, Tung YC et al (2006) Hyperphagia, severe obesity, impaired cognitive function, and hyperactivity associated with functional loss of one copy of the brain-derived neurotrophic factor (BDNF) gene. Diabetes 55: 3366–3371 +Gu F, Parada I, Yang T, Longo FM, Prince DA (2022) Chronic partial TrkB activation reduces seizures and mortality in a mouse model of Dravet syndrome. Proc Natl Acad Sci U S A 119 +Huang WH, Guenthner CJ, Xu J, Nguyen T, Schwarz LA, Wilkinson AW, Gozani O, Chang HY, Shamloo M, Luo L (2016) Molecular and Neural Functions of Rai1, the Causal Gene for Smith-Magenis Syndrome. Neuron 92: 392–406 +Huang WH, Tupal S, Huang TW, Ward CS, Neul JL, Klisch TJ, Gray PA, Zoghbi HY (2012) Atoh1 governs the migration of postmitotic neurons that shape respiratory effectiveness at birth and chemoresponsiveness in adulthood. Neuron 75: 799–809 +Huang WH, Wang DC, Allen WE, Klope M, Hu H, Shamloo M, Luo L (2018) Early adolescent Rai1 reactivation reverses transcriptional and social interaction deficits in a mouse model of Smith-Magenis syndrome. Proc Natl Acad Sci U S A +Javed S, Lee YJ, Xu J, Huang WH (2021) Temporal dissection of Rai1 function reveals brain-derived neurotrophic factor as a potential therapeutic target for Smith-Magenis syndrome. Hum Mol Genet 31: 275–288 +Javed S, Selliah T, Lee YJ, Huang WH (2020) Dosage-sensitive genes in autism spectrum disorders: From neurobiology to therapy. Neurosci Biobehav Rev 118: 538–567 +Kamitakahara A, Xu B, Simerly R (2016) Ventromedial hypothalamic expression of Bdnf is required to establish normal patterns of afferent GABAergic connectivity and responses to hypoglycemia. Mol Metab 5: 91–101 +Kocalis HE, Hagan SL, George L, Turney MK, Siuta MA, Laryea GN, Morris LC, Muglia LJ, Printz RL, Stanwood GD et al (2014) Rictor/mTORC2 facilitates central regulation of energy and glucose homeostasis. Mol Metab 3: 394–407 +Kron M, Howell CJ, Adams IT, Ransbottom M, Christian D, Ogier M, Katz DM (2012) Brain activity mapping in Mecp2 mutant mice reveals functional deficits in forebrain circuits, including key nodes in the default mode network, that are reversed with ketamine treatment. J Neurosci 32: 13860–13872 +Kron M, Lang M, Adams IT, Sceniak M, Longo F, Katz DM (2014) A BDNF loop-domain mimetic acutely reverses spontaneous apneas and respiratory abnormalities during behavioral arousal in a mouse model of Rett syndrome. Dis Model Mech 7: 1047–1055 +Lee SK, Lee S, Shin SY, Ryu PD, Lee SY (2012) Single cell analysis of voltage-gated potassium channels that determines neuronal types of rat hypothalamic paraventricular nucleus neurons. Neuroscience 205: 49–62 +Li W, Bellot-Saez A, Phillips ML, Yang T, Longo FM, Pozzo-Miller L (2017) A small-molecule TrkB ligand restores hippocampal synaptic plasticity and object location memory in Rett syndrome mice. Dis Model Mech 10: 837–845 +Loos RJF, Yeo GSH (2022) The genetics of obesity: from discovery to biology. Nat Rev Genet 23: 120–133 +Massa SM, Yang T, Xie Y, Shi J, Bilgen M, Joyce JN, Nehama D, Rajadas J, Longo FM (2010) Small molecule BDNF mimetics activate TrkB signaling and prevent neuronal degeneration in rodents. J Clin Invest 120: 1774–1785 +Miki T, Liss B, Minami K, Shiuchi T, Saraya A, Kashima Y, Horiuchi M, Ashcroft F, Minokoshi Y, Roeper J et al (2001) ATP-sensitive K+ channels in the hypothalamus are essential for the maintenance of glucose homeostasis. Nat Neurosci 4: 507–512 +Moss J, Oliver C, Arron K, Burbidge C, Berg K (2009) The prevalence and phenomenology of repetitive behavior in genetic syndromes. J Autism Dev Disord 39: 572–588 +Ogier M, Wang H, Hong E, Wang Q, Greenberg ME, Katz DM (2007) Brain-derived neurotrophic factor expression and respiratory function improve after ampakine treatment in a mouse model of Rett syndrome. J Neurosci 27: 10912–10917 +Parton LE, Ye CP, Coppari R, Enriori PJ, Choi B, Zhang CY, Xu C, Vianna CR, Balthasar N, Lee CE et al (2007) Glucose sensing by POMC neurons regulates glucose homeostasis and is impaired in obesity. Nature 449: 228–232 +Perry CA, Pravetoni M, Teske JA, Aguado C, Erickson DJ, Medrano JF, Lujan R, Kotz CM, Wickman K (2008) Predisposition to late-onset obesity in GIRK4 knockout mice. Proc Natl Acad Sci U S A 105: 8148–8153 +Quarta C, Claret M, Zeltser LM, Williams KW, Yeo GSH, Tschop MH, Diano S, Bruning JC, Cota D (2021) POMC neuronal heterogeneity in energy balance and beyond: an integrated view. Nat Metab 3: 299–308 +Rao NR, Abad C, Perez IC, Srivastava AK, Young JI, Walz K (2017) Rai1 Haploinsufficiency Is Associated with Social Abnormalities in Mice. Biology (Basel) 6 +Schultze SM, Hemmings BA, Niessen M, Tschopp O (2012) PI3K/AKT, MAPK and AMPK signalling: protein kinases in glucose homeostasis. Expert Rev Mol Med 14: e1 +Shao J, Yamashita H, Qiao L, Friedman JE (2000) Decreased Akt kinase activity and insulin resistance in C57BL/KsJ-Leprdb/db mice. J Endocrinol 167: 107–115 +Simmons DA, Belichenko NP, Yang T, Condon C, Monbureau M, Shamloo M, Jing D, Massa SM, Longo FM (2013) A small molecule TrkB ligand reduces motor impairment and neuropathology in R6/2 and BACHD mouse models of Huntington’s disease. J Neurosci 33: 18712–18727 +Slager RE, Newton TL, Vlangos CN, Finucane B, Elsea SH (2003) Mutations in RAI1 associated with Smith-Magenis syndrome. Nat Genet 33: 466–468 +Smith AC, Magenis RE, Elsea SH (2005) Overview of Smith-Magenis syndrome. J Assoc Genet Technol 31: 163–167 +Smith AC, McGavran L, Robinson J, Waldstein G, Macfarlane J, Zonona J, Reiss J, Lahr M, Allen L, Magenis E (1986) Interstitial deletion of (17)(p11.2p11.2) in nine patients. Am J Med Genet 24: 393–414 +Smith MA, Katsouri L, Irvine EE, Hankir MK, Pedroni SM, Voshol PJ, Gordon MW, Choudhury AI, Woods A, Vidal-Puig A et al (2015) Ribosomal S6K1 in POMC and AgRP Neurons Regulates Glucose Homeostasis but Not Feeding Behavior in Mice. Cell Rep 11: 335–343 +Su X, Gu D, Xu L, Liang Z, Luo X, Yang P, Yang J (2021) PI3K/Akt pathway expression in children with different obesity degrees and its relationship with glucolipid metabolism and insulin resistance. Am J Transl Res 13: 6592–6598 +Tan CL, Cooke EK, Leib DE, Lin YC, Daly GE, Zimmerman CA, Knight ZA (2016) Warm-Sensitive Neurons that Control Body Temperature. Cell 167: 47–59 e15 +Taniguchi H, He M, Wu P, Kim S, Paik R, Sugino K, Kvitsiani D, Kvitsani D, Fu Y, Lu J et al (2011) A resource of Cre driver lines for genetic targeting of GABAergic neurons in cerebral cortex. Neuron 71: 995–1013 +Unger TJ, Calderon GA, Bradley LC, Sena-Esteves M, Rios M (2007) Selective deletion of Bdnf in the ventromedial and dorsomedial hypothalamus of adult mice results in hyperphagic behavior and obesity. J Neurosci 27: 14265–14274 +Vohra MS, Benchoula K, Serpell CJ, Hwa WE (2022) AgRP/NPY and POMC neurons in the arcuate nucleus and their potential role in treatment of obesity. Eur J Pharmacol 915: 174611 +Wu SW, Xu B (2022) Rapid and Lasting Effects of Activating BDNF-Expressing PVH Neurons on Energy Balance. eNeuro 9 +Xu B, Goulding EH, Zang K, Cepoi D, Cone RD, Jones KR, Tecott LH, Reichardt LF (2003) Brain-derived neurotrophic factor regulates energy balance downstream of melanocortin-4 receptor. Nat Neurosci 6: 736–742 +Xu B, Xie X (2016) Neurotrophic factor control of satiety and body weight. Nat Rev Neurosci 17: 282–292 +Xu Y, Lu Y, Cassidy RM, Mangieri LR, Zhu C, Huang X, Jiang Z, Justice NJ, Xu Y, Arenkiel BR et al (2019) Identification of a neurocircuit underlying regulation of feeding by stress-related emotional responses. Nat Commun 10: 3446 +Yang H, An JJ, Sun C, Xu B (2016) Regulation of Energy Balance via BDNF Expressed in Nonparaventricular Hypothalamic Neurons. Mol Endocrinol 30: 494–503 +Yeo GS, Connie Hung CC, Rochford J, Keogh J, Gray J, Sivaramakrishnan S, O’Rahilly S, Farooqi IS (2004) A de novo mutation affecting human TrkB associated with severe obesity and developmental delay. Nat Neurosci 7: 1187–1189 + + +Figures and Figure legends + + +Differentially expressed proteins in the hypothalamus of SMS mice

A heat map showing hierarchical clustering of differentially expressed proteins in the hypothalamus of SMS mice.

+ + + + +The <italic>Bdnf<sup>Cre</sup></italic> allele labels BDNF-producing neurons in the paraventricular nucleus of the hypothalamus +

(A) Schematic showing a transgenic mouse expressing BdnfCre-dependent td-Tomato (Ai9) signals in BDNF-expressing cells.

(B-E) Colocalization of BdnfCre-dependent td-Tomato signals with endogenous BDNF protein. Black: DAPI, Magenta: BDNF-producing cells; Green: endogenous BDNF. Scale bar: 50 µm, yellow arrowheads indicate Ai9 marked BDNF cells coexpressing endogenous BDNF.

+ + + + +RAI1 expression in BDNF-expressing but not oxytocin-expressing magnocellular PVH neurons +

(A-H) Representative images of DAPI expression (blue, A), oxytocin (green, B), BDNF- producing cells (Ai9 reporter, magenta, C), and RAI1 protein (cyan, D) in the PVH. Note that BDNF-producing neurons are distinct from the oxytocin-expressing neurons (E) and that RAI1 is more selectively expressed in the BDNF-expressing neurons (F-H).

(I) Quantification showing the percentage of RAI1+ cells that co-express either BDNF (left bar, magenta) or oxytocin (right bar, green) in the PVH. n=3 mice/group. Data are shown as mean±SEM.

+ + + + +Ablating <italic>Rai1</italic> from the BDNF-producing cells induces body weight gain and defective energy homeostasis +

(A) Schematic representation of the conditional knockout mice carrying either normal Rai1 alleles and the Ai9 reporter in BDNF-producing cells (top) or homozygous Rai1 deletion and the Ai9 reporter in the BDNF-producing cells.

(B) Representative images showing the expression of DAPI (white), RAI1-expressing cells (green), and BDNF-expressing cells (Ai9 reporter, magenta) in the PVH. The top panel shows the Ctrl group with RAI1-Ai9 co-expression (yellow arrowheads) and the bottom panel shows reduced RAI1 and Ai9 colocalization (yellow arrowheads). Note that RAI1 expression was still detected in non-BDNF neurons in the PVH.

(C) Quantification showing the percentage of BDNF-producing cells was not altered by Rai1 deletion, indicating that RAI1 loss did not impair the generation or survival of PVHBDNF neurons. ns indicates not significantly different, unpaired t-test, two-tailed.

(D) The BdnfCre/+ and BdnfCre/+; Rai1fl/+ female mice had similar body weights. ns indicates not significantly different. Two-way ANOVA with Šidák’s multiple comparisons test.

(E-G) Blood parameter analysis of triglycerides (TG), high-density lipoprotein (HDL), low-density lipoprotein and very low-density lipoprotein (LDL+VLDL) did not differ between Ctrl and cKO mice. ns indicates not significantly different, unpaired t-test, two-tailed.

(H) The respiratory exchange rate did not differ between Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(I) In the glucose tolerance test, the cKO mice showed an increased area under the curve (AUC). ***p<0.001, unpaired t-test, two-tailed.

(J) In the glucose tolerance test, cKO mice showed increased insulin levels at the beginning of the test and 30 minutes after glucose administration. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons tests, *p < 0.05, **p<0.01.

(K) In the insulin tolerance test, blood glucose level is measured after insulin injection. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05.

Data are shown as mean±SEM.

+ + + + +Metabolic profiles of male cKO mice lacking RAI1 expression in the BDNF-producing cells +

(A) Male cKO mice gained significantly more weight than Ctrl mice beginning at 15 weeks of age. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05, **p<0.01, ****p < 0.0001.

(B) Male BdnfCre/+ and BdnfCre/+; Rai1fl/+ mice had similar body weights. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(C) Echo-MRI analysis showed that 26-week-old male cKO mice had a significantly increased body weight due to increased fat but not lean mass. ns indicates the difference is not significant. unpaired t-test, two-tailed, **p<0.01, ****p<0.0001.

(D) Fat disposition analysis showing the weight of brown adipocytes (BAT), subcutaneous inguinal (S.C.ing), and epididymal white adipose tissues (eWAT). ns indicates the difference is not significant. unpaired t-test, two-tailed, *p<0.05, **p<0.01.

(E) Male cKO mice showed eWAT cell hypertrophy. Top: Representative images of the eWAT tissues in Ctrl and cKO mice. Scale bar = 500µm. Bottom: Frequency distribution of cellular sizes. Two-way ANOVA with Šidák’s multiple comparisons test, *p < 0.05, **p<0.01, ***p<0.001, ****p < 0.0001.

(F-I) Blood parameter analysis showing that male cKO mice showed significantly increased leptin levels (F) without alterations in TG (G), HDL (H), and LDL+VLDL (I). ns indicates the difference is not significant. unpaired t-test, two-tailed. *p< 0.05.

(J) No differences were found in the respiratory exchange rate in male Ctrl and cKO mice. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(K) Male cKO mice showed a significantly increased food intake compared to Ctrl littermates., unpaired t-test, two-tailed, *p< 0.05.

(L) Male cKO mice showed increased energy expenditure at the dark phase. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test, *p<0.05.

(M) Male Ctrl and cKO mice showed similar locomotor activities. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons test.

(N-P) Glucose tolerance test shows that cKO male mice are glycemic 45 minutes post intraperitoneal glucose administration (N) and had increased AUC (unpaired t-test, two-tailed) (O), despite significantly higher insulin levels before and 30 minutes after glucose administration (P), suggesting potential insulin insensitivity. ns indicates the difference is not significant. Two-way ANOVA with Šidák’s multiple comparisons tests, *p<0.05, **p < 0.01, ***p<0.001.

(Q) Insulin tolerance test showing that male cKO mice showed significantly higher blood glucose levels 60 minutes after intraperitoneal insulin administration. Two-way ANOVA with Šidák’s multiple comparisons test, *p<0.05, **p < 0.01.

Data are shown as mean±SEM.

+ + + + +The cellular and synaptic properties of control and <italic>Rai1</italic>- deficient PVH<sup>BDNF</sup> neurons +

(A) Representative traces of rebound and rebound repetitive firing of PVHBDNF neurons in control (black) and cKO (red).

(B) Representative AP waveforms at holding voltages of -80mV, -60mV, and -50mV in control (left) and -50mV and -45mV in cKO (right).

(C) The half-width of AP initiated at a holding voltage of -60mV (neuronal numbers in brackets, ns indicates the difference is not significant, unpaired t-test, two-tailed).

(D) Representative traces of miniature recordings in control (black) and cKO (red) PVHBDNF neurons. The traces were recorded with a Cs-based internal at a holding voltage of -70mV in aCSF (top), spontaneous mEPSCs were recorded in aCSF containing 1uM TTX at a holding voltage of -70mV (middle), and mIPSCs were recorded at holding voltage of +10mV (bottom).

(E) Quantification showing that the cellular size of cKO PVHBDNF neurons was significantly smaller than Ctrl neurons. unpaired t-test, two-tailed. ****p<0.0001.

+ + + + +Metabolic profile of mice lacking <italic>Rai1</italic> in PVH<sup>BDNF</sup> neurons +

(A) Representative images showing co-expression of virus-infected cells (GFP/green), BdnfCre- dependent Ai9 signals (BDNF cells, magenta), and endogenous RAI1 (cyan). Yellow arrowheads indicate BDNF neurons infected with AAV and in these cells, RAI1 signals were diminished in Exp but not Ctrl mice. Scale bar = 50µm.

(B-C) The total number of PVHBDNF neurons (B) and total number of PVHBDNF neurons infected with AAVs (C) per slice show that Rai1 deletion or virus infection did not alter the total number of PVHBDNF neurons. ns indicates the difference is not significant. unpaired t-test, two-tailed.

(D) Rai1 expression was significantly reduced in PVHBDNF neurons infected with both AAVs compared to PVHBDNF neurons infected with sgRai1-GFP AAV alone. unpaired t-test, two-tailed, **p<0.01.

(E) The number of PVHBDNF neurons that co-express RAI1 was significantly reduced in the Exp group. unpaired t-test, unpaired, **p<0.01.

(F) Representative image showing lack of viral expression in the PVH neighbouring nuclei. ARC: arcuate nucleus. VMH: ventromedial nucleus of the hypothalamus. 3V: third ventricle. Scale bar = 50µm.

(G) The respiratory exchange rate remained similar in Ctrl and Exp mice. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(H-J) Blood parameter analysis showing TG, HDL, and LDL+VLDL levels were similar in Ctrl and Exp mice. ns indicates the difference is not significant. unpaired t-test, two-tailed.

(K) In the insulin tolerance test, the Exp group showed a trend towards increased blood glucose levels after insulin injection. Two-way ANOVA, Šídák’s multiple comparisons test.

Data are shown as mean±SEM.

+ + + + +LM22A-4 treatment does not disrupt energy homeostasis and repetitive rearing in the Ctrl mice +

(A) LM22A-4 concentration in the hypothalamus and forebrain of SMS mice 1 hour (black dots) and 3 hours (red dots) after simultaneous intranasal and IP injections.

(B) Four independent sets of western blotting analyses showing reduced p-AKT levels in the hypothalamus of saline-treated SMS mice, which were increased by LM22A-4 treatment. Total AKT protein was used as a control.

(C) Similar body weight of Ctrl mice treated with either saline or LM22A-4. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(D) Energy expenditure was not different between groups in the light vs. dark phase. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(E) Respiratory exchange rates were not different between groups in the light vs. dark phase. ns indicates the difference is not significant. Two-way ANOVA, Šídák’s multiple comparisons test.

(F) Repetitive rearing behaviour in Ctrl mice was not altered by LM22A-4 treatment. ns indicates the difference is not significant. unpaired t-test, two-tailed.

Data are shown as mean±SEM.

+ + + + +LM22A-4 treatment in adult SMS mice is insufficient to improve social interaction deficit +

(A) Schematic showing the tube test that assesses social interaction between stranger mice of different genotypes.

(B) 7-week-old Ctrl mice won significantly more than SMS mice before the LM22A-4 treatment. unpaired t-test, two-tailed. ****p<0.0001.

(C) 10-week-old saline-injected Ctrl mice won significantly more than the saline-injected SMS mice (two weeks post-LM22A-4 treatment). unpaired t-test, two-tailed, ***p<0.001.

(D) 10-week-old saline-injected Ctrl mice won significantly more than LM22A-4 injected SMS mice. unpaired t-test, two-tailed, *p<0.05.

(E) 10-week-old saline-injected SMS mice showed a similar winning rate to LM22A-4-injected SMS mice. ns indicates the difference is not significant. unpaired t-test, two-tailed.

Data are shown as mean±SEM.

+ + + + + + + + + + + + + + + + + + + + + +