SIRT1 coordinates transcriptional regulation of neural activity and modulates depression-like behaviors in the nucleus accumbens

Abstract

Background

Major depression and anxiety disorder are significant causes of disability and socio-economic burden. Despite the prevalence and considerable impact of these affective disorders, their pathophysiology remains elusive. Thus, there is an urgent need to develop novel therapeutics for these conditions. We evaluated the role of SIRT1 in regulating dysfunctional processes of reward by using chronic social defeat stress (CSDS) to induce depression- and anxiety-like behaviors. CSDS induces physiological and behavioral changes that recapitulate depression-like symptomatology and alters gene expression programs in the nucleus accumbens, yet cell type-specific changes in this critical structure remain largely unknown.

Methods

We examined transcriptional profiles of D1-MSNs lacking deacetylase activity of SIRT1 by RNA sequencing (RNA-Seq) in a cell-type specific manner using the RiboTag line of mice. We analyzed differentially expressed genes using gene ontology tools including SynGO and EnrichR, and further demonstrated functional changes in D1-MSN specific SIRT1-KO mice using electrophysiological and behavioral measurements.

Results

RNAseq revealed altered transcriptional profiles of D1-MSNs lacking functional SIRT1 and showed specific changes in synaptic genes including glutamatergic and GABAergic receptors in D1-MSNs. These molecular changes may be associated with decreased excitatory and increased inhibitory neural activity in Sirt1-KO D1-MSNs, accompanied by morphological changes. Moreover, the D1-MSN-specific Sirt1-KO mice exhibited pro-resilient changes in anxiety- and depression-like behaviors.

Conclusions

SIRT1 coordinates excitatory and inhibitory synaptic genes to regulate GABAergic output tone of D1-MSNs. These findings reveal a novel signaling pathway that has the potential for the development of innovative treatments for affective disorders.

Introduction

One in five individuals will experience a major episode of clinical depression during their lifetime (1). Depression is a significant cause of disability and current medications to treat major depressive disorder (MDD) are largely ineffective, with fewer than 50% of patients responding to treatment (2–5). Anxiety disorder is another major condition in affective disorders, and shows high comorbidity with MDD (6). This high comorbidity not only underscores the complexity of MDD but also potentially contributes to its severity and the challenges in its management. Thus, there is a significant need to discover and develop novel therapeutics for depression. The etiology of depression increasingly focuses on the interaction between genetic and environmental factors, with chronic stress recognized as a trigger for major depressive episodes (7). Given the heterogeneity and complexity of brain circuits, identification of molecular mechanisms within specific cell-types and neural networks will advance development of therapies for this disorder.

SIRT1 is a class III NAD⁺-dependent histone deacetylase (HDAC), that serves as an epigenetic remodeler for physiological processes ranging from neurodevelopment to aging (8–14). SIRT1 is also involved in neuropathological conditions such as drug addiction (15–17) and mental disorders (18–23). Indeed, SIRT1 was identified as the first reproducible gene locus associated with major depression in humans by the CONVERGE Consortium (23). This finding was supported by several preclinical rodent models targeting brain regions implicated in reward dysfunction (18–22). Additionally, a separate study in humans revealed that a SNP (rs10997870) in the SIRT1 gene is associated with risk for anxiety disorders, and two genetic mutations in the N terminus (S14P and P37L) increase SIRT1 protein activity (20). These results were replicated using samples from the Virginia Adult Twin Study of Psychiatric and Substance Use Disorders (24). Moreover, another independent study of Japanese subjects found a significant association between a different SIRT1 SNP (rs10997875) and MDD (25). Taken together, SIRT1 is a critical epigenetic regulator in depression-related gene networks, yet its specific downstream actions are unclear. Addressing how SIRT1 affects anxiety and depression behaviors is crucial for linking genetic risks to brain circuitry mechanisms.

Chronic social defeat stress (CSDS) triggers depression- and anxiety-like physiological and behavioral changes (26–28) and modifies gene expression in the NAc (29–32). The NAc’s role in anhedonia, reduced motivation, and decreased energy levels, which are prevalent in depression, highlights its contribution to these symptoms (33–35). Our previous findings show that CSDS modulates SIRT1 levels in the NAc, where its overexpression heightens anxiety- and despair-like behaviors in a cell-type specific manner (19). The NAc comprises two types of medium spiny neurons (MSNs) distinguished by their dopamine D1 and D2 receptor enrichment, influencing rewarding or aversive behaviors, respectively. This balance is supported by evidence showing differential responses of NAc D1 and D2 MSNs to stress (36–38), with restraint stress reducing excitatory synaptic strength in D1-MSNs without affecting D2-MSNs (39). CSDS-susceptible mice showed reduced excitatory inputs to D1-MSNs, unlike D2-MSNs, and chronic chemogenetic suppression of D1-MSN activity induced depressive-like behaviors in previously resilient mice (40). These findings underline the distinct roles of NAc neuron types in depression but leave the specific molecular and cellular mechanisms by which SIRT1 impacts these pathways open for further investigation.

In this study, we identified the cell-type specific mechanisms through which SIRT1 mediates depression- and anixiety-like behaviors. Using cell-type specific approaches, our results revealed that SIRT1 regulates the expression of genes involved in synaptic transmission which in turn influences electrophysiological, morphological, and behavioral endpoints. These findings suggest that the SIRT1-signaling pathway may be targeted for the development of novel antidepressants.

Methods and Materials

Mice

Mice were housed on a 12 hr light-dark cycle with access to food and water ad libitum. Male CD1 retired breeders were obtained from Charles River Laboratories. Sirt1 floxed mice (41) crossed to D1-Cre BAC transgenic mice from GENSAT (42) on a C57BL/6J background were used for behavioral experiments (Sirt1^D1-KO). To induce Rpl22-HA in D1 neurons, we crossed RiboTag mice (43) with D1-Cre or Sirt1^D1-KO mice. To label D1 neurons for electrophysiological and morphological analyses, either viral (AAV-DIO-EYFP) or trangenic (Ai6 mice) (44) approaches were used for D1-Cre or Sirt1^D1-KO mice. All animal procedures were approved by the University of Arizona Institutional Animal Care and Use Committees.

Chronic social defeat stress

Social defeat stress was performed according to published protocols (45, 46). Test mice were exposed to an aggressive unfamiliar male CD1 mouse for 5 minutes per day for 10 days, and then placed in an adjacent compartment of the same cage for 24 hours with sensory but not physical contact. Twenty-four hours after the last defeat, animals were assayed for their social interaction, and classified into either susceptible or resilient phenotypes.

RNA-seq

The RiboTag approach was used for D1-MSN specific mRNA isolation. RNA-seq was performed by the University of Arizona Genetics Core. Data were analyzed by Rosalind (OnRamp BioInformatics).

Electrophysiological recording and morphological analysis

The D1-MSNs were visualized in Sirt1-fl^D1-KO mice crossed to Ai6 reporter (44) or in D1-Cre mice injected with AAV-DIO-EYFP virus and sacrificed after three weeks. In the NAc shell, EYFP⁺-D1-MSNs were identified and recorded for spontaneous EPSCs, IPSCs, and synaptic AMPA/NMDA receptor currents with local electric stimulation. The DiOListic labeling method (47) was used for analyzing dendritic morphology and spine shape in D1-MSNs.

Behavior tests and optogenetic stimulation

All subject mice were placed in the testing room one hour before testing for acclimation. Data were analyzed by Ethovision (Noldus) in time bins of 2.5 min and full time (5 min) and significant results were presented. We adapted and modified a previously established protocol for the optogenetic activation of D1-MSNs (40).

Statistical Analysis

We performed following statistical analyses to determine significance for conditions in which there were more than two groups or two factors: unpaired Student t-tests, or two-way ANOVAs with Tukey’s or Sidak’s post-hoc test. Statistical analysis was performed using Prism (GraphPad). All values represent means ± s.e.m.

Additional Details

All detailed procedures are included in Supplemental Information.

Results

Generation of D1-MSN specific Sirt1-KO mice

In our previous study (19), we established that SIRT1 mediates depression- and anxiety-like behaviors via cell-type-specific actions in the NAc. Overexpression of SIRT1 in D1-MSNs, but not in D2-MSNs, increased despair- and anxiety-like behaviors. To investigate whether loss of SIRT1 deacetylase activity in D1-MSNs contributes to despair or anxiety-like behaviors, we generated mice lacking SIRT1 in D1-MSNs (SIRT1^D1-KO) (Figure 1A). To confirm the KO of SIRT1 in D1 neurons, we further crossed the SIRT1^D1-KO mice with a Ribotag (RT) line (RT;Sirt1^D1-KO), which allows for cell-type specific isolation and enrichment of translating mRNAs (translatome) (32, 43). We found a specific reduction of the exon 4-retained transcript in SIRT1^D1-KO but not in D1-Cre littermate controls (Figure 1B, left panel). Importantly, non-floxed (exon 8–9) transcripts were unaffected in both WT and KO mice (Figure 1B, right panel). In the NAc of WT mice, SIRT1 protein is detected as distinct double bands (48), whereas in SIRT1^D1-KO mice, lower molecular weight bands appear on the blot which corresponds to the HDAC catalytic domain (exon 4 region) that was deleted in SIRT1^D1-KO mice (Figure 1C). Next, to determine if SIRT1 catalytic activity was diminished in SIRT1^D1-KO, we assessed the acetylation state of well-known SIRT1 canonical targets forkhead transcription factor (FKHR) and nuclear factor kappa light chain enhancer of activated of B cells (NF-kB), which show an increase in acetylation levels in KO relative to WT tissue (Figure 1D). Given the lack of cell-type specificity in the western blot analysis, we further employed immunohistochemistry to determine acetylated-FKHR levels in D1-MSNs. Acetylated-FKHR levels were specifically increased by ~50% in D1-MSNs of SIRT1^D1-KO mice but not in other cell-types in WT mice (Figure 1E). These results demonstrated that the cell-type specific SIRT1 knock-out strategy successfully reduced SIRT1’s catalytic activity in D1-MSNs.

Figure 1. Cell-type specific gene modulations in D1-MSN specific Sirt1 knock-out (KO) mice.

(A) Schematic diagram of D1-MSN specific Sirt1-KO mice, Sirt1-fl;D1-Cre (SIRT1^D1-KO), generation. (B) Validation of exon 4 excised Sirt1-KO transcripts using semi-quantitative PCR (upper) and qPCR (lower bar charts; RiboTag purified; ***p = 0.0008, t₍₉₎ = 7.910, unpaired t-test; WT, n = 5; KO, n = 6). (C) Western blot analyses of SIRT1 expression in the NAc (open arrowhead: wild-type SIRT1, closed arrowhead: exon 4-excised form of SIRT1 (KO)). (D) Western blot analyses of SIRT1 deacetylation targets, FKHR (left) and NF-kB (right) in wild-type and Sirt1-fl^D1-KO (KO) mice (**p = 0.0014; FKHR: t₍₅₎ = 6.205, p = 0.0016; NF-kB: t₍₅₎ = 6.431, unpaired t-test; WT, n = 3; KO, n = 4). (E) Immunohistochemical validation of D1-MSN specific altered acetylation of FKHR protein (scale bar, 20 μm; ***p = 0.0001, t₍₁₀₎ = 6.013, unpaired t-test; WT, n = 6; KO, n = 6). Graphs are represented as mean ± s.e.m. (**p < 0.01, ***p < 0.001, n.s., not significant; t-test, two-tailed).

SIRT1 regulates the expression of glutamatergic and GABAergic receptor subunits in D1-MSNs

SIRT1 influences despair- and anxiety-like behaviors and may act as a transcriptional regulator of gene networks implicated in depression. Using the approach of cell-type specific RNA-Sequencing (32), we evaluated how a perturbation in SIRT1 signaling influences gene networks in D1-MSNs. After isolating the D1-specific translatome from the NAc of RT;D1-Cre and RT;SIRT1^D1-KO mice (Figure 2A), RNA sequencing was performed according to established protocols (32) (Table S1). We show that RiboTag IP samples are enriched with mature mRNA, as evidenced by decreased read alignments in the intronic regions and increased 3’-UTR coverage, reflecting 3’ UTR lengthening in neuronal cells (Figure 2B) (49). Reads mapped to the Sirt1 locus clearly showed depletion of the exon 4 region of the Sirt1 gene of SIRT1^D1-KO relative to D1-Cre mice (Figure 2C). To validate D1-MSN-specific mRNA isolation, we assessed expressions of D1-MSN markers between IP and Input samples and found enrichment of gene markers for D1-MSNs in IP samples, results consistent with our previous study and others (32, 50) (Figure S1A). Next, normalized gene expression profiles were analyzed and plotted for each sample in a t-distributed stochastic neighbor embedding plot, which showed distinct clustering of transcriptional profiles of D1-MSNs for WT and SIRT1^D1-KO groups (Figure S1B). Differential gene expression analysis between WT IP and SIRT1^D1-KO IP transcripts identified 645 differentially expressed genes (DEGs), with 322 genes upregulated and 323 downregulated (p-adj<0.05) (Figure 2D, S2A; Table S2). Gene ontology (GO) analyses conducted with ShinyGO (51) highlighted that these DEGs are significantly associated with synaptic functions, including synaptic transmission and neurotransmitter transport, among the top 10 enriched terms (Figure 2E). Further analysis using DisGeNet through Enrichr (52) revealed significant associations of DEGs with MDD, unipolar depression, and mental depression, alongside drug perturbation gene sets for heroin and morphine (Figures S2B–C). Notably, DEGs also indicated enrichment for genes treated with resveratrol, a SIRT1 agonist, underscoring altered SIRT1 pathways in SIRT1^D1-KO (Figure S2C). which validates altered SIRT1 pathways in SIRT1^D1-KO. The gene set for “glutamatergic synapse,” crucial in depression (53–55), was most significantly enriched in SIRT1-KO D1-MSNs, alongside significant enrichment for terms such as “cell junction”, “postsynaptic density”, and “Abnormal excitatory postsynaptic currents” (Figures S2D–E). Using SynGO (56), a focused analysis on synaptic function showed that out of the 645 DEGs, 122 genes were mapped to SynGO annotated genes, with significant enrichment in cellular components and biological processes related to synapses (Figure 2F, Table S3).

Figure 2. Functional transcriptome analysis in D1-MSNs of Sirt1^D1-KO mice.

(A) Schematic diagram of D1-MSN specific Sirt1-KO crossed with RiboTag mice (RT;SIRT1^D1-KO). The HA-tagged polyribosome allows cell-type specific mRNA preparation from D1-MSNs. (B) Read mapping composition of each RNA-seq sample on the mouse genome (mm10) for RiboTag IP and input samples (two-way ANOVA; Intron, genotype: p = 0.2717, sample type: p < 0.0001, interaction: p = 0.4692; post hoc Tukey’s test, WT IP vs WT Input: ***p < 0.0001, KO IP vs KO Input: ***p < 0.0001; 3’ UTR, genotype: p = 0.0382, sample type: p < 0.0001, interaction: p = 0.4460; post hoc Tukey’s test, WT IP vs WT Input: ***p < 0.0001, KO IP vs KO Input: ***p < 0.0001). (C) Validation of Sirt1 KO transcripts using RNAseq read mapping on Sirt1 locus using Integrated Genomics Viewer (IGV) (read numbers for the exon 4: control = 26.75 ± 4.59; KO = 4.00 ± 0.71; **p = 0.0027, unpaired t-test). (D) Heatmap of 645 DEGs between wild-type and SIRT1^D1-KO (KO) mice comparison (increased in KO, 322 genes; decreased in KO, 323 genes; adjusted p-value < 0.05). (E) Gene ontology (GO) analysis of the 645 DEGs using ShinyGO (top 20 terms). (F) Synaptic component-based enrichment analysis by SynGO (cellular compartment) with terms and related genes. The top 3 cellular component ontology terms are “synapse” (gene count: 111, p-adj = 3.18e-22), “presynapse” (gene count: 67, p-adj = 1.32e-17) and “postsynapse” (gene count: 61, p-adj = 9.67e-12). Channel and receptor genes are indicated in bold (right panel). (G) Gene expression changes of glutamatergic and GABAergic signaling from the 645 DEGs: glutamate receptor ionotropic kainate 3 (Grik3), FC = 1.20, p-adj = 0.024; glutamate receptor ionotropic (Gria4) FC = −1.21, p-adj = 0.001; glutamate receptor ionotropic kainate 1 (Grik1), FC = −1.34, p-adj = 0.019; glutamate receptor metabotropic 1 (Grm1), FC = −1.36, p-adj = 5.45e-6; gamma-aminobutyric acid type B receptor subunit 2 (Gabbr2), FC = 1.32, p-adj = 0.016; gamma-aminobutyric acid type A receptor subunit alpha2 (Gabra2), FC = 1.37, p-adj = 3.57e-07.

These data (Figure 2D–F) raise the compelling possibility that SIRT1 regulates neural activity-related genes such as glutamatergic or GABAergic receptor gene expression. To further probe this hypothesis, we assessed if SIRT1-KO in D1-MSNs results in changes of glutamatergic receptor subunit expression in our RNA-seq dataset and found changes in Grik3, Gria4, Grik1, and Grm1. Interestingly, two GABAergic receptor genes, Gabbr2 and Gabra2, in D1-MSNs showed increased expressions in Sirt1^D1-KO (Figure 2G).

Sirt1-KO in D1-MSNs decreases excitatory neurotransmission

Results from the RNA sequencing study strongly suggest cell-type specific ablation of SIRT1 in D1-MSNs may alter excitatory and inhibitory synaptic transmissions. To further probe this hypothesis, we performed electrophysiological analyses of D1-MSNs in both Sirt1^D1-KO and wild-type littermates (Figure 3A). We found both AMPAR and NMDAR current were significantly decreased in D1-MSNs of Sirt1-KO mice, while the AMPAR/NMDAR ratio was also decreased (Figure 3B, C). We next measured spontaneous EPSC (sEPSC) activity (Figure 3D, E). sEPSC amplitude from D1-MSNs of Sirt1-KO slices distributed to the lower amplitude bins as shown in cumulative plot, the left-shifted curve compared to the control. Violin plot of the raw sEPSC amplitudes also showed a significant amplitude reduction in Sirt1-KO D1-MSNs. No change in sEPSC frequency was observed (Figure 3E). On the other hand, spontaneous IPSC (sIPSC) was increased in Sirt1-KO D1-MSNs while its frequency was not significantly changed (Figure 3F, G). Structurally, D1-MSNs of SIRT1^D1-KO showed significantly decreased dendritic arborization measured at 60–120 μm from the soma (Figure 3H). Interestingly, the number of mushroom spines of D1 neurons was significantly increased in SIRT1^D1-KO mice, while other types of spines were not changed (Figure 3I, S3). Taken together, we observed overall decreases in neural activity along with structural changes in D1-MSNs of SIRT1^D1-KO mice.

Figure 3. Electrophysiological and morphological changes in D1-MSNs of SIRT1^D1-KO mice.

(A) Schematics of the experimental plan of electrophysiological and morphological analyses. (B, C) Representative AMPAR- and NMDAR-EPSC traces (B), bar graphs (C; mean ± s.e.m.; AMPAR-EPSC: WT, 157.84 ± 8.34 pA, KO: 72.38 ± 6.22 pA, t₍₃₈₎ = 8.31, p < 0.001; NMDAR-EPSC: WT, 78.00 ± 4.27 pA; KO, 44.10 ± 3.73 pA, t₍₃₈₎ = 6.0, p < 0.001; AMPA/NMDA ratio: WT, 2.08 ± 0.11, KO: 1.71 ± 0.10, t₍₃₈₎ = 2.47, p = 0.018; WT, n = 19 cells/3 mice; KO, n = 21 cells/3 mice). (D, E) Representative sEPSC traces (D), bar graphs, cumulative distribution and violin plot of sEPSC amplitude and frequency (E) in D1-MSN from control versus SIRT1^D1-KO mice (cumulative distribution, p < 0.0001, Kolmogorov-Smirnov test; violin plot, median amplitude: WT, 16.4 ± 0.41 pA, KO, 14.1 ± 0.43 pA; frequency: WT, 3.01 ± 0.45 Hz; KO, 2.79 ± 0.28 Hz, t₍₂₁₎ = 1.02, p = 0.17; WT, n = 12 cells/6 mice; KO, n = 11 cells/5 mice). (F, G) Representative sIPSC traces (F), bar graphs (G) and cumulative distribution of sIPSC amplitude and frequency in D1-MSN from WT control versus Sirt1 KO mice (amplitude: WT, 24.19 ± 1.35 pA, KO: 35.93 ± 2.24 pA, t₍₃₃₎ = 4.13, p < 0.001; frequency: WT, 1.21 ± 0.23 Hz; KO, 1.53 ± 0.24 Hz, t₍₃₃₎ = 0.92, p > 0.05; WT, n = 15 cells/2 mice; KO, n = 20 cells/4 mice). (H, I) Morphological analysis of DiI-stained control or SIRT1-KO D1-MSNs. (H) Representative trace images and analysis of dendritic branch intersections (WT, n = 10; KO, n = 11). (I) Spine density analysis for mushroom-type spines (WT, n = 10; KO, n = 11). Graphs are represented as mean ± s.e.m. (*p < 0.05, **p < 0.01, ***p < 0.001, n.s., not significant; t-test, two-tailed (C, E, G, I); two-way ANOVA with Sidak’s post-hoc test (H)).

Sirt1-KO in D1-MSNs blocks the development of anxiety and social avoidance following CSDS

In a previous study (19), we demonstrated that overexpression of SIRT1 in D1-MSNs induces depression- and anxiety-like behaviors. To examine if the absence of functional SIRT1 in D1-MSNs is protective against the development of anxiety- and despair-like behaviors following stress, SIRT1^D1-KO and wild-type littermate male mice were subjected to chronic social defeat stress and results were compared to baseline controls (Figure 4A). Importantly, we found that SIRT1^D1-KO mice showed decreased susceptibility after CSDS while wild-type littermates showed around 60% susceptible ratio (Figure 4B). Although the average values for interaction ratio and time remained largely unchanged, there was a noticeable alteration in the distribution of data points, which showed the decrease of susceptible mice in the KO group (Figure S4A). While wild-type and SIRT1^D1-KO groups showed similar levels of locomotor activity in the baseline and CSDS conditions (Figure S4B), wild-type littermates showed increased anxiety-like behaviors after CSDS, as exemplified by decreases in time spent exploring the center zone of the open field maze (Figure 4C, upper) and decreases in time spent exploring the open arms of the elevated plus maze (Figure 4C, lower), respectively. As SIRT1^D1-KO mice showed decreased social avoidance and anxiety following CSDS, we further assessed whether SIRT1^D1-KO mice exhibit changes in additional despair-like behaviors using forced swim test (FST) and tail suspension test (TST). Both wild-type and SIRT1^D1-KO mice showed no significant change in immobility time after CSDS while performing FST (Figure 4D, upper). However, in TST, both groups showed significant increases in immobility (Figure 4D, lower). To examine consistency in the depression- and anxiety-related behaviors for individual animals, we performed Pearson’s correlation analysis (Figure S4C). Wild-type mice data revealed significant correlations among SI ratio, EPM, FST, and TST, except in the open field test. Conversely, SIRT1^D1-KO mice showed weaker, altered correlations, significant only between open field and TST, and FST and TST.

Figure 4. Anxiety- and depression-like behaviors of SIRT1^D1-KO mice in baseline and chronic social defeat stress (CSDS) conditions.

(A) Experimental timeline and schematic of baseline behavioral testing and chronic social stress paradigm of Sirt1^D1-KO mice. (B) The susceptibility of WT control and Sirt1^D1-KO mice upon CSDS (WT: 55.6%, SIRT1^D1-KO: 29.2%, t₍₁₆₎ = 2.73, p = 0.0149; unpaired t-test; WT, n = 12 cohorts; KO, n = 6 cohorts). (C) Analysis of anxiety-like behaviors using open field (upper panel; two-way ANOVA, genotype: p = 0.0359, stress: p = 0.0024, interaction: p = 0.0614; post hoc Tukey’s test, WT baseline vs WT social defeat: p = 0.0028, WT social defeat vs SIRT1^D1-KO baseline: p = 0.0021, WT social defeat vs SIRT1^D1-KO social defeat: p = 0.0265) and elevated plus maze (lower panel; two-way ANOVA, genotype: p = 0.0035, stress: p = 0.0444, interaction: p = 0.0178; post hoc Tukey’s test, WT baseline vs WT social defeat: p = 0.0084, WT social defeat vs SIRT1^D1-KO baseline: p = 0.0033, WT social defeat vs SIRT1^D1-KO social defeat: p = 0.0014). (D) Analysis of depression-like behaviors using forced swim test (upper panel; two-way ANOVA, genotype: p = 0.7380, stress: p = 0.0012, interaction: p = 0.9606; post hoc Tukey’s test, not significant for all comparisons) and tail suspension test (lower panel; two-way ANOVA, genotype: p = 0.4692, stress: p < 0.0001, interaction: p = 0.1830; post hoc Tukey’s test, WT baseline vs WT social defeat: p < 0.0001, WT baseline vs SIRT1^D1-KO social defeat: p < 0.0001, WT social defeat vs SIRT1^D1-KO baseline: p < 0.0001, SIRT1^D1-KO baseline vs SIRT1^D1-KO social defeat: p = 0.0005). (E) Experimental timeline and schematic of optogenetic manipulation of D1-MSNs (F-H) Behavioral assessments at pre- and post-optogenetic activation by SI test (interaction zone time w/ social target) (F), open field (G), and elevated plus maze (H) (SI test (F), genotype: p = 0.8727, treatment: p < 0.0001, interaction: p = 0.0059; WT pre vs WT post: p = 0.3034, KO pre vs KO post: p < 0.0001; open field (G); genotype: p = 0.0037, treatment: p = 0.1606, interaction: p = 0.3478; WT pre vs WT post: p = 0.7082, KO pre vs KO post: p = 0.1255; EPM (H); genotype: p = 0.0120, treatment: p = 0.0002, interaction: p = 0.0087; WT pre vs WT post: p = 0.3632, KO pre vs KO post: p = 0.0002; post hoc Sidek’s test). Graphs are represented as mean ± s.e.m. ((B) *p < 0.05; WT, n = 12; KO, n = 6; t-test, two-tailed; (C, D) baseline: WT, n = 23 from 3 cohorts; KO, n = 20 from 3 cohorts; social defeat: WT, n = 21 from 4 cohorts; KO, n = 21 from 4 cohorts; two-way ANOVA; (F-H) WT, n = 12; KO, n = 9; ***p < 0.001, **p < 0.01, *p < 0.05; two-way repeated measures ANOVA).

To validate the causal relationship between D1-neuronal activity and behavioral alterations in SIRT1^D1-KO mice, we applied optogenetic techniques for high-frequency stimulation of D1-neurons (40), as shown in Figure 4E. Repetitive stimulations at 50 Hz effectively elevated D1-neuronal activity in SIRT1^D1-KO mice to levels comparable with wild-type mice (Figure S5). Chronic activation of D1-neurons in the KO mice significantly reduced their social interaction time, in contrast to the unaltered interaction times in wild-type mice (Figure 4F). Without a social target, both KO and WT mice exhibited no changes in interaction zone times (Figure S6A). Interestingly, the chronic stimulation of D1-neurons selectively impacted anxiety-like behaviors: while there was no notable alteration in the time spent in the center zone during open field tests (Figure 4G, S6B), a significant decrease was observed in the open arm times of the elevated plus maze tests (Figure 4H, S6C). Taken together, these results demonstrate that the enhanced D1-neuronal activity in SIRT1^D1-KO mice is associated with an exacerbation of depressive and anxiety-like phenotypes, closely resembling the behavioral manifestations observed in wild-type mice.

Discussion

Sirtuins, particularly SIRT1, play a critical role in cellular metabolism and mitochondrial function (57, 58), and affective disorders such as MDD (23) by influenceing despair- and anxiety-like behaviors, as evidenced in animal models (18–22). Our current study builds on previous findings that SIRT1 overexpression in D1-MSNs enhances depression- and anxiety-like behaviors (19). Here we show that SIRT1 in D1-MSNs of the NAc mediates depression- and anxiety-like behaviors via coordinated regulation of synaptic genes, and balanced excitatory and inhibitory inputs onto D1-MSNs.

To investigate SIRT1’s role in NAc’s D1-MSNs, we generated Sirt1^D1-KO mice and validated the lack of SIRT1 deacetylase activity by showing increased acetylation levels of canonical SIRT1 targets, FKHR and NF-kB, in D1-neurons. This setup enabled us to further explore unbiased gene expression profiles by crossing with RiboTag mice and performing RNA sequencing. We identified unique gene expression profiles regulated by SIRT1 and used GO analyses with bioinformatics tools to assess if SIRT1 controls depression-related gene networks in D1-MSNs. The GO analysis revealed that SIRT1 significantly modifies genetic programs associated with mood disorders, showing a notable alignment with gene sets for major mood disorders. We found genes linked to both excitatory and inhibitory synaptic currents are enriched, highlighting important molecular pathways modulated by SIRT1 in D1-MSNs. Specifically, we observed decreased expression of ionotropic and metabotropic glutamate receptor genes in the DEGs, in contrast, GABA receptor subunits were increased in the KO D1-MSNs. These changes could collectively decrease the GABAergic output of D1-MSNs.

Electrophysiological assessments in D1-MSNs of SIRT1^D1-KO mice revealed decreased excitatory neural activity, with reduced AMPAR and NMDAR currents, aligning with observed gene expression changes of decreased glutamatergic receptor and increased GABA receptor genes. In this regard, it is noteworthy that Robinette et al. reported that an NMDAR receptor subunit, Grin2A, was decreased in brain-specific SIRT1-KO mice (59). Interestingly, GABA synthesizing enzymes, GAD1 and GAD2, were also significantly decreased in D1-MSNs of SIRT^D1-KO, which suggests an overall decrease of GABAergic output of D1-MSNs to various projecting areas including ventral tegmental area (VTA), substantia nigra (SNr), and ventral pallidum (VP) in SIRT1^D1-KO mice. Accordingly, the current literature suggests that SIRT1 positively regulates neuronal excitability and plasticity. For example, in the NAc; pharmacological modulation of SIRT1 was positively correlated with neuronal excitability (60), while in the hippocampus, SIRT1 increased dendritic complexity (18, 61–63) and synaptic plasticity (62–64). Decreased neuronal excitability in Sirt1-KO mice was also observed in mPFC pyramidal neurons (21) and AgRP neurons (65). These findings align with our observation of Sirt1’s role in D1-MSNs.

Emerging evidence suggests dysfunction of excitatory synaptic transmission correlates with the onset of depression (53–55). Indeed, enhanced AMPA receptor function within the NAc was noted in both susceptible mice (32, 66) and depressed patients (66). Additionally, neural activity changes in D1-MSNs are strongly associated with depression. Francis and colleagues reported increased excitability of D1-MSNs in susceptible mice (40), whereas our study shows decreased excitability in D1-MSNs of Sirt1^D1-KO mice, which exhibit a pro-resilient phenotype. This suggests elevated SIRT1 expression in susceptible mice could explain the increased excitability observed in the previous report.

In our morphological analysis, dendritic branching was decreased in D1-MSNs of Sirt1^D1-KO mice. Similar reductions were observed in hippocampal neurons lacking SIRT1 (61, 62), while SIRT1 activation increased dendritic arborization (18). Regarding spine morphology, we observed increased mushroom spines in SIRT1-KO D1-MSNs, without an total spine increase. Despite being a minor category, the significant increase in mushroom spines without affecting the overall spine count suggests conversion from other spine types to the mushroom type. This contrasts with the absence of spine shape changes in the hippocampus (62), implicating brain region-specific effects of SIRT1. Interestingly, previous studies reported that mushroom spine currents in D1-MSNs were increased in resilient mice (67), while no apparent spine changes were observed following CSDS (68), suggesting SIRT1 ablation may exert pro-resilient effects via different mechanisms to the resilient, which needs to be explored in the future.

Importantly, we validated that the ablation of SIRT1 deacetylase activity in D1-MSNs is behaviorally protective as SIRT1^D1-KO mice did not show increases in anxiety and social avoidance after CSDS. These results are consistent with the viral vector-mediated manipulation of SIRT1 expression in our previous study (19), which confirms the pro-susceptible effects of SIRT1 actions in D1-MSNs. The absence of significant differences in average interaction ratio and time between wild-type and SIRT1 knock-out mice, could result from data aggregation across both susceptible and resilient individuals. Nonetheless, the distribution shift in the knock-out group clearly indicates the pro-resilient impact of SIRT1 knock-out on social behavior and stress response. It also suggests the SIRT1 knockout’s protective role against CSDS is not consistent across all individuals, requiring further investigation. While no significant difference was observed between WT and KO in FST and TST, the previous study (19) also showed inconsistent results from FST, which highlights the limitations and variability of such immobility measures in depression studies (69).

Indeed, the results from our optogenetic study on SIRT1^D1-KO mice demonstrate a clear causal link between D1-neuronal activity and behavioral changes. High-frequency stimulation of D1-neurons led to significant behavioral alterations, notably a reduction in social interaction and increased anxiety-like behaviors, as evidenced in the elevated plus maze tests. Interestingly, the selective impact of D1-neuron stimulation on anxiety-like behaviors, evidenced by unchanged open field test behavior but a pronounced decrease in open arm times in the elevated plus maze, suggestively indicated by correlation data (Figure S4C, WT), highlights the complex interplay between D1-neuronal activity and behavioral manifestations in SIRT1^D1-KO mice, meriting further exploration. These results suggest that SIRT1-mediated D1-neuronal hyperactivity contributes to depressive and anxiety-like phenotypes, which provides compelling evidence of connections between SIRT1, neural activity and depressive behaviors.

Considering our findings, it would be interesting to determine which neuronal circuits connect D1 neuronal activity governed by SIRT1 to the behavioral outcomes. As previously reported, D1-MSNs in the NAc have various projections to the VTA, SNr, VP, and even among MSNs (37, 38, 70). Among them, the NAc-VTA circuits are a well-established mediator of depression, reward, and motivational processes (71–74). Increased SIRT1 expression in the NAc correlates with despair-like behaviors in socially defeated mice (19), suggesting that D1-MSNs may increase GABAergic tone to the VTA, potentially altering reward circuitry. Future research should explore how SIRT1 modulation affects D1-MSN activity and its impact on dopaminergic and other neuronal activities. This could involve fiber-photometry with GCaMPs (75, 76) and the novel dopamine biosensor (dLight) (77) to measure dopamine release in the NAc (78). Bdnf is reported as another important signaling pathway in the NAc-VTA circuit regarding depression (73). Recently, using the RiboTag approach, to target the TrkB in D1-MSNs, Engeln et al. reported that ablation of Bdnf signaling exerts distinct transcriptional changes in D1-MSNs of the dorsal striatum related to stereotypy phenotypes (79). While our study is not directly comparable with the data, both studies contribute to our understanding of cell-type specific signaling in the striatum, with our findings underscoring the distinct role of SIRT1 in neural activity and depression-related behaviors in NAc D1-MSNs.

While our focus was on the specific targeting of Sirt1 in D1-MSNs, we cannot rule out potential off-target effects, arising both from the irreversible ablation of Sirt1 enzymatic activities and the involvement of D1-positive cells beyond the NAc region. This complexity underscores the challenge of dissecting gene functions within intricate neuronal networks and emphasizes the importance of cautious interpretation in our genetic analyses. This nuanced understanding leads us to explore the diverse regulatory roles of SIRT1 on excitatory and inhibitory receptor genes. Considering the role of SIRT1 as a higher-order chromatin remodeler (80, 81), Rec8, the most downregulated gene in SIRT1-KO D1-MSNs is noteworthy (Table S2), as the protein is a key component of the cohesin complex, although its postmitotic function is still elusive (82). Future studies using ATAC-Seq or Hi-C will facilitate elucidating the precise molecular mechanism of SIRT1 actions in regulating gene networks linked to depression. Irrespective of SIRT1’s role in transcriptional regulation, we cannot rule out non-genomic actions of SIRT1 in influencing neural activity as SIRT1 directly modulates BK channel acetylation and subsequent membrane localization (22).

Our study, focusing solely on male subjects, is limited by the well-documented fact that women are more prone to depression and anxiety disorders (83). Research on mouse depression models has suggested sex-specific differential mechanisms in the brain (21, 84, 85). Additionally, sex-specific differential gene expression profiles in D1- and D2-MSNs of the NAc were explored through the RiboTag method (86), underscore this point. Thus, it is highly plausible that Sirt1’s role in the NAc may also exhibit cell-type and sex specificity, similar to its demonstrated effects in mPFC neurons where Sirt1 ablation induces depression behaviors in male, but not in female mice (21). Further research is essential to elucidate Sirt1’s functions across sexes.

In conclusion, we found that SIRT1 regulates genes mediating neuronal transmission in D1-MSNs, which in turn influences electrophysiological, morphological, and behavioral endpoints. These data raise the compelling possibility that SIRT1 influences the expression of genes associated with synaptic transmission in brain reward regions, which are implicated in anxiety and depressive behaviors.

KEY RESOURCES TABLE

Resource Type	Specific Reagent or Resource	Source or Reference	Identifiers	Additional Information
Add additional rows as needed for each resource type	Include species and sex when applicable.	Include name of manufacturer, company, repository, individual, or research lab. Include PMID or DOI for references; use “this paper” if new.	Include catalog numbers, stock numbers, database IDs or accession numbers, and/or RRIDs. RRIDs are highly encouraged; search for RRIDs at https://scicrunch.org/resources.	Include any additional information or notes if necessary.
Antibody	rabbit anti-HA	Abcam	Abcam Cat# ab9110, RRID:AB_307019
Antibody	rabbit anti-SIRT1	Cell Signaling	Cell Signaling cat #2028S
Antibody	rabbit anti-Ac-FKHR	Santa Cruz Biotechnology	Santa Cruz Biotechnology Cat # sc-49437
Antibody	rabbit anti-Ac-NF-KB	Abcam	Abcam Cat # ab52175–100
Antibody	Dynabeads anti-rabbit IgG	Thermo Fisher Scientific	TFS Cat# 11203D, RRID:AB_2783009
Organism/Strain	Mouse: C57BL/6J, male	The Jackson Laboratory	RRID:IMSR_JAX:000664
Organism/Strain	RiboTag	The Jackson Laboratory	RRID:IMSR_JAX:029977
Organism/Strain	Sirt1-floxed	The Jackson Laboratory	RRID:IMSR_JAX:008041
Organism/Strain	Ai6	The Jackson Laboratory	RRID:IMSR_JAX:007906
Organism/Strain	D1-Cre	GENSAT	line FK150, RRID:MMRRC_037156-JAX
Organism/Strain	CD1 retired breeder mice	Charles River Laboratories	RRID:IMSR_CRL:022
Bacterial or Viral Strain	AAV-DIO-EYFP	Addgene	Addgene Cat # 27056-AAV5
Bacterial or Viral Strain	AAV-DIO-ChR2-ChETA	Addgene	Addgene Cat # 26968-AAV5
Software; Algorithm	Neurolucida 360	MicroBrightField	RRID:SCR_001775
Software; Algorithm	Prism v8.3.0	GraphPad Software	RRID:SCR_002798
Software; Algorithm	Orange v3.20.1	Orange Data Mining	RRID:SCR_019811
Software; Algorithm	Rosalind	OnRamp BioInformatics	RRID:SCR_006233
Deposited Data; Public Database	GSE147640	NCBI GEO DataSets	RRID:SCR_005012
Commercial Assay Or Kit	supersignal dura ECL	Thermo Fisher Scientific	TFS Cat # 34075
Commercial Assay Or Kit	Direct-zol miniprep kit	Zymo Research	Zymo Cat # R2052
Commercial Assay Or Kit	KAPA Stranded mRNA-Seq Kit	Roche	Roche Cat # 07962169001