Nucleus accumbens medium spiny neuron subtypes differentially regulate stress-associated alterations in sleep architecture

Abstract

Background:

Stress is implicated in the pathophysiology of major depression and posttraumatic stress disorder. These conditions share core features, including motivational deficits, heighted anxiety, and sleep dysregulation. Chronic stress produces these same features in rodents, with some individuals being susceptible or resilient, as seen in humans. While stress-induced neuroadaptations within the nucleus accumbens (NAc) are implicated in susceptibility-related dysregulation of motivational and emotional behaviors, their effects on sleep are unclear.

Methods:

We used chemogenetics (DREADDs) to examine the effects of selective alterations in activity of NAc medium spiny neurons (MSNs) expressing dopamine-1 receptors (D1-MSNs) or dopamine-2 receptors (D2-MSNs) on sleep-related endpoints. Mice were implanted with wireless transmitters enabling continuous collection of electroencephalography (EEG) data to quantify vigilance states over a 20-day test period. Parallel cohorts were examined in behavioral tests assessing stress susceptibility.

Results:

D1- and D2-MSNs play dissociable roles in sleep regulation. Stimulation of inhibitory or excitatory DREADDs expressed in D1-MSNs exclusively affects rapid eye movement (REM) sleep, whereas targeting D2-MSNs affects slow wave sleep. The combined effects of D1-MSN inhibition and D2-MSN activation on sleep resemble those of chronic social defeat stress. Alterations in D1-MSN function also affect stress susceptibility in social behavior tests. Elevation of CREB (cAMP response element binding protein) within D1-MSNs is sufficient to produce stress-like effects on REM sleep.

Conclusions:

In addition to regulation of motivational and emotional behaviors, the NAc also influences sleep, an endpoint with high translational relevance. These findings provide a neural basis for comorbidity in key features of stress-related illness.

INTRODUCTION

Sleep dysregulation is a core feature of major depressive disorder (MDD) and posttraumatic stress disorder (PTSD), and frequently comorbid with signs of diminished motivation (e.g., anhedonia) and heightened anxiety (1–4). Up to 90% of depressed patients report problems with sleep (1), and clinical studies have identified sleep disturbance as a core feature of MDD that is predictive of future episodes of illness (5). Sleep alterations that forecast depressive illness include increased total time in rapid eye movement (REM) sleep, decreased latency to first REM bout, and increased REM density (2,3,5–8). Conversely, standard antidepressants have REM-suppressive effects, and REM deprivation has anti-depressant effects (7,9). These observations raise the possibility that sleep dysregulation represents a biomarker for risk of depressive illness (2,3,5). Stress—a major precipitating factor for MDD and PTSD—causes profound sleep dysregulation in humans and laboratory animals (2,3,10–12), suggesting that studies of interactions between sleep and stress have high translational relevance. Mice have important differences in their sleep patterns compared to humans, including polyphasic sleep throughout the day; however, changes in sleep in response to stress and changes in stress in response to sleep disturbances are observed across species (2,3,11). Indeed, sleep as a metric has many characteristics, including the ability to measure the same endpoints across species, that enhance alignment of clinical and neuroscience research on psychiatric illness (13,14).

The nucleus accumbens (NAc) plays a well-characterized role in motivation and is increasingly considered a key node in the neural circuitry that mediates mood (15–20). The NAc integrates inputs from regions including ventral tegmental area (VTA), hippocampus, amygdala and prefrontal cortex to regulate motivated behavior (18,21). Changes in excitatory inputs to the NAc, as well as the intrinsic properties of NAc neurons themselves, produce alterations in motivation ranging from elevated reward to depressive-like states (17,19,21,22). One method of triggering altered motivation in rodents is chronic social defeat stress (CSDS), which induces depression- and anxiety-like phenotypes such as anhedonia and social avoidance (16,23). Use of CSDS to study the neurobiology of mood and anxiety disorders has numerous advantages: as examples, it is an ethological form of stress that produces sustained changes in behavior, and these changes are mitigated by translationally-relevant regimens of antidepressants (24,25). While CSDS is most thoroughly validated in male mice, it can produce similar outcomes in females (26).

Recent studies have provided new insights on the NAc cell subtypes that regulate the persistent behavioral effects of CSDS, as well as the neural mechanisms of resilience and susceptibility to stress (27). Within the NAc, two major populations of medium spiny neurons (MSNs) comprise 98% of local neurons (28,29): those predominantly expressing dopamine-1 receptors (D1-MSNs) or dopamine-2 receptors (D2-MSNs). These populations have different inputs and outputs as well as distinct electrophysiological profiles, and have been ascribed divergent roles across numerous modalities (19,28,30–32). Stressors including CSDS cause dramatic changes in the intrinsic electrophysiological properties of NAc MSNs and their synaptic inputs (17,33,34–36). Pre-stress differences in synaptic activity of D1- but not D2-MSNs are associated with stress resilience or susceptibility (37), and optogenetic or chemogenetic regulation of D1-MSN activity following CSDS modifies these phenotypes (27). Additionally, stress triggers molecular changes in the NAc that reflect, at least in part, cell-type specific transcriptional changes leading to stress resilience or susceptibility (36,38,39,40). As examples, MSN cell-type specific differences in expression of dynorphin (DYN), ΔFosB, ß-catenin, and other gene products have been implicated in stress sensitivity (17,36,38–43). These findings indicate that cell-type specific processes differentially regulate the development of depressive-like phenotypes in response to stress.

We recently characterized the effects of a CSDS regimen that produces social avoidance and anhedonia (25) on sleep and diurnal rhythms in mice (11). We found that CSDS increases REM sleep (also called paradoxical sleep [PS]) and slow wave sleep (SWS). The depression-like increases in PS were blocked by a kappa-opioid receptor (KOR) antagonist, suggesting the involvement of cyclic AMP response element binding protein (CREB) induction of DYN within D1-MSNs—which has previously been linked to depressive-like behaviors (20)—in sleep regulation (11,43). The effects of CSDS on PS have been replicated (12) recently, and other studies have begun to implicate the mesolimbic system in sleep (44–48). While these recent studies suggest that circuits typically associated with motivation are also involved in regulating sleep, the contributions of NAc cell types remains uncharacterized.

The current studies were designed to provide insight into the neural mechanisms by which stress affects sleep in mice. We used chemogenetics to selectively alter the function of D1- or D2-MSNs in NAc and examined real-time effects on sleep-related endpoints. In parallel, we examined the effects of the same manipulations in a battery of other behavioral tests to identify differences in stress resilience and susceptibility. Considering accumulating evidence for the importance of D1-related mechanisms in stress susceptibility, we also examined if alterations in CREB function within NAc D1-MSNs are sufficient to alter sleep.

METHODS (see also: Supplement)

Mice

Drd1-Cre (Tg(Drd1-cre)FK150Gsat/Mmucd) or Drd2-Cre (Tg(Drd2-Cre)ER44Gsat/Mmucd) heterozygotes (GENSAT) were crossed with wild-type C57BL/6J mice (Jackson Laboratories), and Cre-positive males were used as subjects. Adult male CD1 mice (retired breeders; Charles River Laboratories) served as aggressors. Mice were housed in a temperature-controlled vivarium on a 12-hour light:dark cycle; Zeitgeber time (Z-hour) 0 was defined as 07:00 (lights-on). Mice weighed >25g at the time of surgery and had unrestricted access to food and water except during testing. Procedures were approved by McLean Hospital and conformed to National Institute of Health guidelines.

Surgery

For viral vector infusion and transmitter implantation, mice were anesthetized with a 100 mg/kg ketamine/10 mg/kg xylazine mixture. Viral vectors (0.3 μl, diluted to 5.0 × 10⁹ viral genomes (vg)/μl in sterile PBS) were infused bilaterally into the NAc (relative to bregma: A/P=+1.6, M/L=+/−1.0, D/V=−4.4) at 0.1μl/min. AAV vectors had DIO cassettes encoding mCherry, hM4(Gi)-mCherry, or hM3(Gq)-mCherry (serotype 5) (49,50), or green fluorescent protein (GFP), wild-type CREB-GFP, or dominant-negative mutant (m)CREB-GFP (serotype 8.2) (51), and were infused according to validated procedures (52). Transmitters were implanted as decribed (53,54); briefly, an incision through the abdominal wall was made, and the device (HD-X02; DSI) was placed inside abdominal cavity. Electroencephalography (EEG) wires were attached to skull screws contacting dura (relative to bregma: A/P=1.0, M/L=1.0; and A/P=−3.0, M/L=−3.0) and secured with dental cement, and electromyography (EMG) lead wires were sutured into trapezius muscle. Incisions were sutured and covered with triple antibiotic cream, and antibiotic (trimethoprim-sulfamethoxazole 0.8 mg/mL) was provided in water. Mice were given 3 weeks for recovery before testing.

DREADD activation

We administered clozapine, which can serve as a DREADD ligand in vivo (23,55,56), via drinking water as described (57). Baseline water intake was measured for 3 days prior to starting treatment; clozapine (Sigma-Aldrich) was solubilized with a few drops of acetic acid, diluted into a larger volume of drinking water, and administered at 0.3 mg/kg/day.

Behavior

CSDS was performed as described (11,25). Behavioral experiments were performed during the light-phase, beginning at the same time (09:00) and using the same sequence of mice. CD1 mice were pre-screened for aggression before the experiments. CSDS was performed on 10 consecutive days, with social interaction tests (11,25) on the morning of Days 4 and 11. Each test comprised two 150-sec trials in an open arena (44×44×30 cm). In the first trial, mice were tested in the presence of an empty wire cage; in the second, a CD1 mouse was placed in the wire cage. Open field tests were performed in the same arena. Mice were initially placed against a side to start 5-min test sessions. Elevated plus maze (EPM) tests were performed using a standard apparatus (30-cm arms, 5.5×5.5-cm center area, 5.5-cm walls, elevated 80 cm); mice were placed in center to start 5-min tests. Procedures were videotaped and analyzed using Ethovision.

Physiological Recordings

Mice implanted with transmitters lived in home cages (30x16x14 cm) that rested on receiver platforms (RPC-1; DSI), as described (11). Data quantifying sleep, locomotion, and body temperature were analyzed as described (11,53).

Real Time PCR

Reverse transcription of mRNA isolated from NAc punches (1.0-mm tissue corer) dissected from snap-frozen brain was performed using SuperScript4 (Invitrogen) as described (58). Validated primers were identified from an institutional (Massachusetts General Hospital, Boston MA) primer data bank. For analysis, CT values were converted to fold change and ΔΔCT compared to ITM2B control.

Statistics

Analyses were performed using Prism 8 (GraphPad). Data were analyzed via t-tests or ANOVAs that included repeated measures when mice were tested across numerous time points. In a small number of cases, mixed-effects model analyses (REML) were used when individual data points were missing due to transient transmitter malfunction. Following ANOVAs, post hoc analyses were performed using Holm-Sidak tests or Dunnett’s multiple comparison tests (when repeated measures were used). A small number of mice (N=3) were excluded from the analyses due to misplaced (non-bilateral) infusions.

RESULTS

There were two parallel cohorts of mice: one to assess sleep, and the other to assess stress susceptibility in a battery of behavioral assays. The wireless transmitters enabled continuous measurement of EEG, EMG, body temperature, and locomotor activity without tethering that might restrict free movement or posture (Figure 1A). Following recovery and 5 days of baseline measurements, effects of activation or inhibition of NAc MSN subtypes were examined over a 10-day period, enabling comparisons with a 10-day CSDS regimen that produces increases in time in PS and SWS with decreases in active wake (AW) patterns in control mice (P’s<0.05-<0.01, Holm-Sidak post hoc tests) (Figure 1B) (11). DREADDs were selectively expressed in Cre-positive NAc cells via DIO vectors (Figure 1C). Following vector infusions, expression of the transgene reporter (mCherry) was predominantly restricted to the NAc; there was negligible expression outside the infusion area or within NAc efferents (e.g, prefrontal cortex). To activate DREADDs, clozapine (0.3 mg/kg/day) was delivered via drinking water (23,55–57) (Figure 1D). In dose-finding studies, acute treatment with this clozapine dosage produced cFos activation in mice expressing the excitatory (Gq-linked) DREADD on D1-MSNs but not control (mCherry only) mice (Supplemental Figure 1A–B).

Figure 1:

Telemetric EEG/EMG recording enables continuous monitoring of sleep in mice expressing DREADDs. (A) Mouse with transmitter (upper) and implantation schematic (lower). (B) Experimental design of a 10-day CSDS regimen (upper) that produces increases in time spent in PS and SWS while decreasing time spent in AW (Treatment X Vigilance State: F[2,57]=11.46, P=0.00007; *P<0.05, **P<0.01, Holm-Sidak tests) (n=10-11/group) (lower). Data are a novel re-analysis of previous work (11). (C) Mouse lines and viral constructs used (upper), and schematic representation of area targeted for viral-mediated gene transfer (lower). (D) Chemogenetic strategy for examining the effects of modulating D1- or D2-MSN activity on sleep. Green boxes indicate 24-hour periods during which Day 5 of “Baseline”, Day 10 of “Clozapine”, and Day 5 of “Washout” were collected.

Chronic stimulation of inhibitory (Gi-) or excitatory (Gq-) DREADDs expressed in NAc D1-MSNs (Figure 2A) had profound effects on sleep. For clarity, experimental epochs were consolidated into 3 periods: Baseline (5 days), Treatment (clozapine, 10 days), and Washout (cessation of Treatment, 5 days); data from Baseline Day 5, Treatment Day 10, and Washout Day 5 are presented. Inhibition of D1-MSNs via the Gi-linked DREADD produced significantly lower latencies to PS on Day 10 (P<0.05) (Figure 2B). This effect did not persist upon treatment cessation (Washout), suggesting rapid recovery. Shortened PS latencies reflect a depressive-like phenotype: humans with MDD have lower latencies to the first REM bout, indicative of more rapid cycling through sleep stages (59). In contrast, there were no differences in PS latencies in mice expressing the excitatory DREADD or the control protein (mCherry) on D1-MSNs. Chronic D1-MSN activation or inhibition differentially affected time spent in PS: D1-MSN inhibition caused large increases in time spent in PS, whereas activation produced decreases (P’s<0.05) (Figure 2C). Both effects remained evident following washout (P’s<0.05), suggesting induction of long-lasting neuroadaptations that maintain these phenotypes. Neither activation nor inhibition of D1-MSNs affected time spent in SWS or AW, reflecting the fact that mice normally only spend a small proportion of time (~5%) in PS, such that large increases in PS may not be accompanied by statistically significant decreases in other states. D1-MSN inhibition also produced increases in PS bouts (P<0.05) that remained detectable (P=0.06) following washout (Figure 2D), whereas activation of D1-MSNs produced nominal decreases in bouts. Bout length of SWS and AW were not affected by either treatment. Inhibition or activation of D1-MSNs did not affect PS, SWS, or AW bout length (Figure 2E), nor did they affect EEG spectral power (Supplemental Figure 2). The proportion of time in PS, SWS, or AW during baseline measurements did not differ among groups, indicating that expression of Gi or Gq DREADDs in D1-MSNs did not affect sleep in the absence of clozapine (Supplemental Figure 3A). These findings suggest that chronic DREADD-mediated inhibition of D1-MSNs recapitulates the broad effects of CSDS on PS with great selectivity, but without producing CSDS-like effects on SWS (11,12), and that activation of D1-MSNs produces effects that are opposite in direction.

Figure 2:

Selective modulation of D1-MSNs via excitatory (Gq) or inhibitory (Gi) DREADDs affects Paradoxical Sleep (PS) without affecting Slow Wave Sleep (SWS) or Active Wakefulness (AW). (A) Representative image of viral transduction of DREADDs, indicated by expression of mCherry reporter, in the NAc of D1-Cre mice (scale bar=200 μm). (B) Chronic D1-MSN inhibition decreased latencies to first PS bout (Main effect of DREADD: F[2,16]=4.842 P=0.0227; *P<0.05 versus baseline, #P<0.05 versus mCherry, Dunnett’s tests) (n=6-7/group). (C) Chronic D1-MSN inhibition caused increases in time spent in PS that persisted following washout, whereas chronic activation caused complementary decreases in time spent in PS (DREADD X Treatment interaction: F[4,32]=6.144, P=0.0009; *P<0.05 versus baseline, #P<0.05, ##P<0.01 versus mCherry, Dunnett’s tests). Neither treatment affected SWS or AW. (D) Chronic D1-MSN inhibition increased the number of PS bouts (Main effect of DREADD: F[2,16]=12.86 P=0.0005; *P<0.05 versus mCherry, Dunnett’s tests) that remained detectable but not statistically significant (P=0.06) following washout. Neither treatment affected SWS or AW. (E) D1-MSN modulation did not affect the length of sleep bouts.

Stimulation of Gi or Gq DREADDs on D1-MSNs did not reliably affect diurnal rhythm of body temperature (Supplemental Figure 4A–C) or average daily body temperature (Supplemental Figure 4D). Chronic clozapine treatment caused small increases in the amplitude (difference between peak and trough) of diurnal body temperature in all groups, including the mCherry controls, suggesting an effect of the drug per se (Supplemental Figure 5A). However, upon cessation of clozapine treatment and washout, this amplitude was decreased relative to control (mCherry) in mice that received chronic D1-MSN inhibition (P<0.05) (Supplemental Figure 4E). Similar reductions—reflecting a “flattening” of diurnal core body temperature rhythms—are observed in mice following CSDS (11) and in humans with MDD (60–62), suggesting a depressive-like phenotype. In contrast, amplitude was increased in mice that received chronic D1-MSN activation (P<0.05) (Supplemental Figure 4E), representing another sleep/rhythm-related metric that is modulated in opposite directions by activation and inhibition of NAc D1-MSNs.

Chronic stimulation of Gi or Gq DREAADs in D2-MSNs (Figure 3A) also altered sleep, although the effects were qualitatively different and involved other features. Whereas inhibition of D1-MSNs decreased latencies to PS, activation of D2-MSNs tended to increase latencies, although this effect did not reach statistical significance on Day 10 or during Washout (Figure 3B). Chronic D2-MSN activation or inhibition had no effects on PS metrics: time, bouts, and bout length were unaffected (Figure 3C–E). Rather, D2-MSN activation affected SWS and AW, causing significant increases in time spent in SWS (P<0.05) with corresponding decreases in AW (P<0.05) (Figure 3C), although both of these effects recovered following washout. These effects on SWS and AW resemble those produced by CSDS (11). Increases in SWS time were not accompanied by changes in the number of SWS or AW bouts (Figure 3D). Bouts of SWS and AW were both decreased during washout (P’s<0.05). Chronic D2-MSN activation significantly increased SWS bout length (Ps<0.05) without affecting AW bout length (Figures 3E). During washout, SWS and AW bout lengths were increased (P’s<0.05–0.01). There were no significant effects of D2-MSN inhibition on any of these endpoints. Likewise, neither activation nor inhibition of D2-MSNs affected EEG spectral power (Supplemental Figure 6). The proportion of time in PS, SWS, or AW during baseline did not differ among groups, indicating that expression of Gi or Gq DREADDs on D2 MSNs did not affect sleep in the absence of clozapine (Supplemental Figure 3B). These findings suggest that the effects of CSDS on sleep are mimicked by a combination of inhibition of NAc D1-MSNs—which produces the PS-related effects without affecting SWS or AW—and activation of NAc D2-MSNs—which produces SWS- and AW-related effects without affecting PS.

Figure 3:

Selective modulation of D2-MSNs via excitatory (Gq) or inhibitory (Gi) DREADDs affects Slow Wave Sleep (SWS) and Active Wakefulness (AW) without affecting Paradoxical Sleep (PS). (A) Representative image of viral transduction of DREADDs, indicated by expression of mCherry reporter, in the NAc of D2-Cre mice (scale bar=200 μm). (B) Chronic modulation of D2-MSNs alters latency to first PS bout (Main effect of DREADD: F[2,18]=5.536 P=0.0134) (n=8/group). (C) Chronic D2-MSN activation caused increases in time spent in SWS accompanied by complementary decreases in time spent in AW (SWS: DREADD X Treatment interaction: F[4,58]= 3.983, P=0.0064; Wake: DREADD X Treatment interaction: F[4,58]=3.396, P=0.0145; *P<0.05, **P<0.01 versus baseline and #P<0.05, ##P<0.01 versus mCherry, Dunnett’s tests). Neither treatment affected PS. (D) D2-MSN modulation did not alter the number of sleep bouts. (E) Chronic D2-MSN activation increased SWS and AW bout length (SWS: Main effect of DREADD: F[2,18]=5.185, P=0.0167; Wake: DREADD X Treatment interaction: F[4,36]=2.678, P=0.047; *P<0.05 versus baseline and #P<0.05 versus mCherry, Dunnett’s tests).

Diurnal rhythms of core body temperature were altered by stimulation of Gq DREADDs on D2-MSNs only (Supplemental Figure 7A–C). This change was reflected by a reduction in average daily body temperature (P<0.05) (Supplemental Figure 7D), without corresponding effects on amplitude of daily body temperature (Supplemental Figure 5B). Effects recovered upon washout. As was the case in the D1-MSN experiments, chronic clozapine treatment caused small increases in the amplitude of diurnal body temperature in all groups, again consistent with a pharmacological effect of clozapine itself (Supplemental Figure 5 A–B). However, upon washout, this amplitude was increased relative to control (mCherry) in mice that received chronic D2-MSN inhibition (P’s<0.05) (Supplemental Figure 7E). D2-MSN activation-induced decreases in average body temperature are broadly consistent with increases in SWS time.

To determine if conditions that altered sleep metrics would also affect anxiety-like behavior at baseline, we examined the effects of activation or inhibition of NAc MSNs in a battery of standard tests used to examine this domain in mice (Figure 4A). Mice were tested in all procedures, but the test sequence progressed from less to more stressful. In the D1-Cre mice, neither inhibition nor activation of D1-MSNs affected time spent in the center of an open field (Figure 4B), an index of anxiety-related behavior (63), although both treatments caused increases in activity that were evident following initial placement in the arena (P’s<0.05-0.01) but not the home cage (Supplemental Figure 8A–B). Likewise, neither treatment affected time spent in the open arms of an elevated plus maze (EPM) (Figure 4C), or either baseline (pre-social defeat) distances traveled (Figure 4D) or time spent in the interaction zone during social interaction (Figure 4E).

Figure 4:

Selective modulation of D1-MSNs via excitatory (Gq) or inhibitory (Gi) DREADDs does not alter baseline stress-related behaviors. (A) Experimental design depicting test order. Neither chronic activation nor inhibition of D1-MSNs altered (B) time spent in the center of an open field, (C) time spent in the open arms of an elevated plus maze, (D) distance traveled upon placement in the social interaction arena, or (E) the amount of time in the interaction zone in the presence of a novel CD-1 mouse (n=6-9/group).

Following cessation of clozapine treatment, mice were exposed to CSDS to determine if chronic alterations in the activity of MSN cell populations affects stress susceptibility. When mice were tested for social interaction behavior in the presence of a novel aggressor (CD-1) at an intermediate time point following subchronic (3-day) exposure to CSDS, large increases in social avoidance behaviors were observed only in mice that had previously received chronic inhibition of D1-MSNs, as reflected by decreases in time spent in the interaction zone (P’s<0.05-0.01) (Figure 5A) and increases in the time spent in corners (Figure 5B). Because the majority of D1-MSN inhibition mice already qualified as “stress susceptible” (i.e., social interaction ratios <1.0) after only 3 defeat sessions (25), whereas this phenotype was rarely seen in controls (Figure 5C), they were not tested further. When the remaining groups were tested after a full 10-day CSDS regimen shown previously to produce social avoidance (25), D1-MSN activation mice showed evidence of stress resilience, as reflected by more time sent in the interaction zone (P<0.05) (Figure 5D) and less time spent in corners (P<0.01) (Figure 5E), as well as a low proportion of mice qualifying as stress-susceptible (Figure 5F; Supplemental Figure 9A–D). These data suggest that chronic inhibition of D1-MSNs produces signs of stress susceptibility, consistent with previous reports (27), whereas chronic activation produces stress resilience.

Figure 5:

Inhibition of D1-MSNs via Gi DREADDs produces a stress-susceptible phenotype in social interaction tests. (A) Prior chronic D1-MSN inhibition together with 3-days of CSDS decreased time spent in the interaction zone when a novel CD1 is present (F[2,20)]=9.90, P=0.001; **P<0.01, Dunnett’s tests) (n=6-9/group), (B) increased time spent in corner (F[2,20]=3.2 P=0.05; #P=0.05 versus mCherry, Dunnett’s tests), and (C) yielded a large fraction of mice qualifying as stress-susceptible (interaction ratio<1.0). (A-C) Prior chronic D1-MSN activation did not affect behavior after 3 days of CSDS, but after 10 days of CSDS this treatment (D) increased time spent in the interaction zone when a novel CD-1 mouse was present (t[12]=2.25, *P<0.05) (n=6-8/group), (E) decreased time spent in corner when a novel CD-1 mouse present (t[12]=2.98, *P<0.05), and (F) yielded a small fraction of mice qualifying as susceptible.

Parallel testing in the D2-Cre mice indicated no reliable effects of activation or inhibition of D2-MSNs on these endpoints (Figure 6A–E). There were no effects on time spent in the center of an open field (Figure 6B); upon initial placement in the open field, both treatments (activation and inhibition) caused decreases in activity that reached significance in the D2-MSN inhibition group (P<0.05), whereas in the home cage D2-MSN activation caused slight decreases (P<0.05) and inhibition caused slight increases (P<0.05) in activity (Supplemental Figure 10A–B). Neither treatment affected time spent in the open arms of an EPM (Figure 6C), baseline (pre-social defeat) distances traveled (Figure 6D) or time spent in the interaction zone during social interaction (Figure 6E). Following 3 days of CSDS, no differences were observed in the presence of a novel aggressor in time spent in the interaction zone (Figure 7A) or time spent in corners (Figure 7B). Because there were no systematic patterns of stress susceptibility after 3 days of CSDS (Figure 7C), testing continued for all mice. After the full 10-day CSDS regimen, there were no statistically significant differences among groups, although there were nominal increases in resilience signs in mice that had received chronic inhibition of D2-MSNs (Figure 7D–F; Supplemental Figure 11A–D). Although D2-Cre control (mCherry) mice showed more evidence of stress-susceptibility when compared to the D1-Cre controls (compare Figure 5F and Figure 7F), these data suggest that neither chronic activation nor inhibition of D2-MSNs produces systematic alterations in anxiety-like behaviors.

Figure 6:

Selective modulation of D2-MSNs via excitatory (Gq) or inhibitory (Gi) DREADDs does not alter baseline stress-related behaviors. (A) Experimental design depicting test order. Neither chronic activation nor inhibition of D2-MSNs altered (B) time spent in the center of an open field, (C) time spent in the open arms of an elevated plus maze, (D) distance traveled upon placement in the social interaction test arena, or (E) the amount of time in the interaction zone in the presence of a novel CD-1 mouse (n=6-8/group).

Figure 7:

Modulation of D2-MSNs does not alter phenotypes in social interaction tests. Neither prior activation nor inhibition of D2-MSNs affected social interaction behaviors in tests conducted after 3 (A-C) or 10 days of CSDS (D-F) (n=5-8/group); 3 mice were removed between Day 3-10 due to CSDS-related injuries.

Because these tests indicated that the signs of persistent stress susceptibility are associated with D1-MSNs, we examined cellular mechanisms in the D1 -line. A 10-day regimen of CSDS elevated mRNA levels of Creb (P<0.05) and the CREB target prodynorphin (Pdyn) (P<0.05) (Figure 8A–B) within the NAc, both of which have been previously associated with depressive- and anxiety-like effects (20,43). To determine if alterations in CREB function specifically within NAc D1-MSNs play a role in regulation of sleep architecture, we used DIO vectors to express CREB, mCREB, or GFP only (control) (Figure 8C–F). Elevation of CREB in NAc D1-MSNs significantly increased time spent in PS (P<0.05) (Figure 8G) without affecting time spent in SWS (Figure 8H) or AW (Figure 8I), recapitulating the effects of chronic D1-MSN inhibition on these metrics. Expression of mCREB did not affect these metrics, consistent with previous findings in studies of conditioned fear (64).

Figure 8:

Roles for nucleus accumbens (NAc) CREB in stress-like effects on sleep. Chronic social defeat stress (CSDS) elevated (A) Creb1 (t[9]=2.99, *P<0.05) and (B) prodynorphin (Pdyn) (t[9]=2.16, *P<0.05) mRNA levels in the NAc (n=5-6/group). (C) Schematic of experimental strategy for examining the effect of CREB over-expression in D1-MSNs on sleep, and representative images of viral-mediated transduction of (D) green fluorescent protein (GPF) control, (E) CREB, and (F) dominant-negative CREB (mCREB) in the NAc of D1-Cre mice (scale bar=200 μm). (G) Elevated CREB expression in D1-MSNs increases time spent in PS (F[2,18]=8.51, P=0.0025; *P<0.05 versus mCherry, Dunnett’s tests), whereas mCREB was without effect (n=6-8/group). Neither CREB nor mCREB expression affected time spent in (H) SWS or (I) AW.

DISCUSSION

Alterations in the function of NAc MSNs are sufficient to produce changes in sleep architecture that mimic those produced by a potent form of stress (CSDS) in mice. This work complements our previous findings that alterations in NAc function affect motivation and emotion, producing changes in behaviors that reflect reward, despair, anxiety, and fear (64–66). Moreover, use of cell-specific expression of inhibitory and excitatory DREADDs reveals striking differences in the consequences of selectively altering the activity of D1- and D2-expressing MSN populations on sleep architecture. Alterations in D1-MSN activity exclusively regulate PS—without affecting SWS or wakefulness—whereas alterations in D2-MSN activity exclusively affect SWS and wakefulness without affecting PS. Inhibition of D1-MSNs produces CSDS-like effects on PS, causing increases in PS time and bouts. These effects are persistent and are accompanied by reductions in body temperature amplitude that emerge following cessation of treatment. Changes in these metrics align with those previously reported in humans with MDD, including reductions in latency to REM, more time spent in REM, more REM bouts, and flattening of the diurnal amplitude of body temperature (1,6,8,59–62,67,68). Conversely, activation of D1-MSNs produces the opposite effects on PS— reductions in time spent in PS—also without affecting SWS or wakefulness. Combining the individual effects of D1-MSN inhibition and D2-MSN activation fully recapitulates the effects of CSDS on sleep, suggesting that alterations in the activity of both cell subtypes are needed to replicate the broad effects of this form of stress, and that CSDS normally engages NAc circuits in ways that produce this pathophysiological outcome.

Exposure to CSDS produces persistent increases in numerous anxiety- and depressive-like behaviors, including social avoidance, reductions in sucrose preference and cocaine reward, reductions in the impact of rewarding brain stimulation, metabolic disturbances, and disruption of sleep architecture (11,16,25,69,70). On these endpoints and others, some individuals show heightened resilience whereas others show susceptibility, despite equivalent stress exposure (71). We found that chronic changes in the activity of NAc MSNs had no effects on pre-CSDS tests of anxiety-related behavior, but subsequently rendered mice more or less susceptible to CSDS. Our observation that chronic inhibition of D1-MSNs produces a stress-susceptible phenotype is consistent with previous reports (27,35,69) and suggests a “2-hit” mechanism whereby underlying stress susceptibility requires additional perturbation for detection (59,72,73). By establishing that stress susceptibility can be modified before stress exposure by modulation of D1-MSN function, our new findings extend previous work showing that modulating D1-MSN function after stress also modifies susceptibility (31,35). Furthermore, they align with previous reports indicating that pre-defeat differences in the excitability of D1-MSNs predict post-defeat stress-susceptible or resilient phenotypes (37). The fact that changes in sleep architecture can occur without producing spontaneous increases in anxiety-like behaviors—at least those we evaluated—raises the possibility that they contribute to the relative risk exerted by additional stressors.

The mechanisms by which stress triggers the persistent alterations in behavior and physiology that characterize illnesses such as MDD are not fully understood. Broad (non cell-specific) activation of CREB in the NAc has been implicated in the pro-depressive and anxiogenic effects of stress (64,66,74,75). The finding that elevation of CREB exclusively in D1-MSNs produces stress-like effects on PS is consistent with a complex molecular cascade that involves increased expression of DYN, an endogenous KOR ligand (76), as a critical mediator of these effects (43). According to this model, stress activates NAc CREB in D1-MSNs, which leads to elevated DYN (77), which acts upon inhibitory KOR receptors expressed on the NAc terminals of VTA dopamine neurons (78). Increased KOR inhibitory tone reduces dopamine efflux in the NAc (79), resulting in less activation of local D1Rs and a net reduction of direct pathway output that is mimicked by DREADD-mediated inhibition of D1-MSNs. Indeed, inhibition of D1-MSNs is previously implicated in anhedonia (27,31,69). The ability of KOR antagonists to mitigate stress effects on sleep (11), as well as their antidepressant-like effects in species ranging from rodents to humans (52,80–82), strengthens evidence that this mechanism plays a key role in regulation of numerous diagnostic features of depressive illness.

An important caveat is that our sleep data were examined in 24-hour bins whereas circadian fluctuations (e.g., dopamine release in striatum (45)) occur on shorter time scales. As such, the effects of the DREADD manipulations may depend on variables including the timing of the stressor. Another caveat is that the D2-Cre mouse line we used may induce transgene expression in NAc cholinergic interneurons (70). Although we cannot rule out the possibility that alterations in cholinergic cells contribute to outcomes reported for this line, concerns are mitigated by previous reports of consistent outcomes in the D2-Cre line and the A2A-Cre line, which does not express Cre in cholinergic neurons (48). Future studies should directly compare these lines on sleep-related endpoints. Likewise, future studies in females will help to determine if the cellular processes involved in producing the effects on stress resilience and susceptibility, or the relative prominence of sleep disruption in response to stress exposure, differ between sexes.

The observation that alterations in NAc function affect sleep in addition to behaviors reflecting motivation and emotion provides important new insight into why these symptoms are frequently co-morbid in humans with conditions such as MDD and PTSD (4). Classically, sleep is most often associated with areas including brainstem, thalamic, and hypothalamic nuclei (2,3,83). However, it was recently shown that acute optogenetic activation of D1-MSNs or their projections to the midbrain and lateral hypothalamus promotes arousal, whereas activation of D2-MSNs promotes sleep (46–48). Our work builds importantly upon these findings by providing new information on how stress changes these circuits and triggers numerous diagnostic features of psychiatric illness, and raises the possibility that a common pathology within the NAc can explain an assortment of frequently co-morbid symptoms (20). Considering that sleep is defined and measured in the same way in mice and humans, and that sleep metrics are becoming increasingly available in humans, studies focused on this endpoint may have particular translational relevance (13,14). An improved understanding of the neural mechanisms of these symptoms may sharpen diagnoses and enable the development of new treatments that, by virtue of targeting specific NAc cell populations, could dramatically improve outcomes because they relieve numerous features of stress-related illness simultaneously.

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	r anti-c-Fos	Santa Cruz	SC-52
Bacterial or Viral Strain	AAV5-DIO-mCherry	ADDGENE	50459-AAV5
Bacterial or Viral Strain	AAV5-hsyn-DIO-hM4(Gi)-mCherry	ADDGENE	44362-AAV5
Bacterial or Viral Strain	AAV5-hsyn-DIO-hM3(Gq)-mCherry	ADDGENE	44361-AAV5
Bacterial or Viral Strain	AAV8.2-GFP	VIROVEK
Bacterial or Viral Strain	AAV8.2-CREB-GFP	VIROVEK
Bacterial or Viral Strain	AAV8.2-mCREB-GFP	VIROVEK
Chemical Compound or Drug	Clozapine	Sigma-Aldrich	C6305
Commercial Assay Or Kit	SuperScript4	Thermo Fischer	18091050
Commercial Assay Or Kit	SYBR Green RT-PCR Master Mix	Thermo Fischer	A25742
Commercial Assay Or Kit	RPC-1	DSI
Commercial Assay Or Kit	HD-X02	DSI
Organism/Strain	Tg(Drd1-cre)FK150Gsat/Mmucd, mouse, male	GENSAT	FK159
Organism/Strain	Tg(Drd2-Cre)ER44Gsat/ Mmucd, mouse, male	GENSAT	ER44
Organism/Strain	C57BL/6J, mouse, male	Jackson Laboritories	#000664
Organism/Strain	Adult CD1 Retired Breeder, mouse, male	Charles River Laboratories	#022, retired breeder
Primer	Itm2B F		AGACCTACAAACTGCAGCGCC
Primer	Itm2B R		AAAGGGGCAGGGTATGCTGTGG
Primer	Creb F		AGC AGC TCA TGC AAC ATC ATC
Primer	Creb R		AGT CCT TAC AGG AAG ACT GAA CT
Primer	Pdyn F		AGCTTGCCTCCTCGTGATGC
Primer	Pdyn R		CAAGTCATCCTTGCCACGGAGC
Software; Algorithm	Ethovision 7	Noldus
Software; Algorithm	Ponema v. 6	DSI
Software; Algorithm	Clocklabs	Actimetrics
Software; Algorithm	Prism 8	GraphPad