Ablation of Ventral Midbrain/Pons GABA Neurons Induces Mania-like Behaviors with Altered Sleep Homeostasis and Dopamine D2R-mediated Sleep Reduction.

Individuals with the neuropsychiatric disorder mania exhibit hyperactivity, elevated mood, and a decreased need for sleep. The brain areas and neuronal populations involved in mania-like behaviors, however, have not been elucidated. In this study, we found that ablating the ventral medial midbrain/pons (VMP) GABAergic neurons induced mania-like behaviors in mice, including hyperactivity, anti-depressive behaviors, reduced anxiety, increased risk-taking behaviors, distractibility, and an extremely shortened sleep time. Strikingly, these mice also showed no rebound sleep after sleep deprivation, suggesting abnormal sleep homeostatic regulation. Dopamine D2 receptor deficiency largely abolished the sleep reduction induced by ablating the VMP GABAergic neurons without affecting the hyperactivity and anti-depressive behaviors. Our data demonstrate that VMP GABAergic neurons are involved in the expression of mania-like behaviors, which can be segregated to the short-sleep and other phenotypes on the basis of the dopamine D2 receptors.

Introduction

Sleep problems are the core components of mood disorders, such as bipolar disorder characterized by periods of depression and mania (Benca et al., 1997). The clinical criteria for mania include persistently elevated mood or energy and decreased need for sleep (American Psychiatric Association, 2013). In humans, infarctions of the midbrain or pontine areas induces secondary mania (Antelmi et al., 2014, Caplan, 2010, Drake et al., 1990, Satzer and Bond, 2016, Shen et al., 2005). In the present study, we focused on the ventral medial midbrain/pons (VMP) area, a region that contains sleep-regulating GABAergic neurons (Chowdhury et al., 2019, Takata et al., 2018, Yu et al., 2019). To elucidate the physiologic functions of VMP GABAergic neurons, we selectively ablated these neurons in mice with Cre-dependent viral expression of the diphtheria toxin subunit A (DTA) in the VMP of vesicular GABA transporter (VGAT; a marker of GABAergic neurons) Cre mice (VGAT-CreDTA/VMP mice: Figures 1A and 1B). For controls, we prepared mice expressing humanized Renilla reniformis-derived green fluorescent protein (hrGFP) in the VMP GABAergic neurons (VGAT-CrehrGFP/VMP). These mice were subjected to a comprehensive behavioral test battery including various tasks that are reliably used to assess locomotor activity, depression-like behavior, sociability, anxiety, species-typical behavior, and learning and memory (Crawley, 2007, Crawley and Paylor, 1997).

Figure 1

Selective Ablation of the VMP GABAergic Neurons Induces Hyperactivity, Anti-depressive Behaviors, and Reduced Anxiety

(A) VGAT-Cre mice were injected with AAV-FLEX-hrGFP (VGAT-CrehrGFP/VMP mice; control) or AAV-FLEX-DTA (VGAT-CreDTA/VMP mice) into the VMP region.

(B) Brain sections stained against VGAT mRNA. VMP GABAergic neurons were selectively ablated in VGAT-CreDTA/VMP mice but not in VGAT-CrehrGFP/VMP mice. RN, red nucleus; VTA, ventral tegmental area; SN, substantia nigra; IP, interpeduncular nucleus. Scale bar, 500 μm.

(C) Representative path traces and total distance traveled in the open field test. Two-tailed unpaired t test, t = 4.418, df = 17, p = 0.0004.

(D) Immobility time in the tail suspension test. Two-tailed unpaired t test, t = 3.238 df = 17, p = 0.0048.

(E) Immobility time in the forced swim test. Two-tailed Mann-Whitney U test, U = 0, p < 0.0001.

(F) Sociability and social interaction test. Preference for the zone with mouse chamber compared with the empty chamber. Two-tailed unpaired t test, t = 2.319, df = 17, p = 0.0331.

(G) Time in open arms in the elevated plus maze. Two-tailed unpaired t test, t = 3.214, df = 16, p = 0.0054.

(H) Nest-building scores. Two-tailed unpaired t test, t = 4.52, df = 17, p = 0.0003.

(I) Exploratory preference toward familiar or novel objects in the novel object recognition task on the test day. VGAT-CrehrGFP/VMP: two-tailed unpaired t test, t = 4.742, df = 18, p = 0.0002. VGAT-CreDTA/VMP: two-tailed unpaired t test, t = 1.592, df = 16, p = 0.1308.

(J) Total number of entries into each arm and spontaneous alternation rates in the Y-maze task. Number of arm entries: two-tailed unpaired t test, t = 3.159, df = 17, p = 0.0057. Spontaneous alternation rate: two-tailed unpaired t test, t = 1.165, df = 17, p = 0.2601.

(C−J) n = 8–10 mice for each group. Individual values are plotted in the graphs. Data are presented as the mean ± SEM. n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Results

GABAergic VMP Ablation Induces Hyperactivity, Anti-depressive Behaviors, and Reduced Anxiety

Our behavioral test battery revealed that VGAT-CreDTA/VMP mice exhibited increased locomotor activity with a longer distance traveled in the open field test compared with control VGAT-CrehrGFP/VMP mice (p < 0.001, Figure 1C). VGAT-CreDTA/VMP mice also showed consistently decreased immobility in both the tail suspension test (p < 0.01, Figure 1D) and the forced swim test (p < 0.0001, Figure 1E), suggesting anti-depressive behaviors. In addition, VGAT-CreDTA/VMP mice exhibited increased preference for the zone with mice compared with control VGAT-CrehrGFP/VMP mice (p < 0.05, Figure 1F). In the elevated plus maze, VGAT-CreDTA/VMP mice spent more time in the open arms (p < 0.01, Figure 1G), indicating less anxiety and increased risk-taking behavior. In addition, VGAT-CreDTA/VMP mice had lower scores in the nest building test (p < 0.001, Figure 1H). We also assessed learning and memory. In the novel object recognition task, control VGAT-CrehrGFP/VMP mice showed increased exploratory preference for novel objects (p < 0.001, Figure 1I). On the other hand, VGAT-CreDTA/VMP mice demonstrated no preference for novel objects (not significant, n.s., p = 0.131, Figure 1I), suggesting impaired novel object recognition. Conversely, in the Y maze as a test for working memory, VGAT-CreDTA/VMP mice exhibited no difference from control mice in spontaneous alternation rate (n.s., p = 0.260, Figure 1J), but the number of entries into each arm was increased, consistent with their hyperactivity (p < 0.01, Figure 1J). Overall, the series of behavioral tests provided us landmark information to support the involvement of VMP GABAergic neurons in the various aspects of manic episodes described in both humans and rodent models (American Psychiatric Association, 2013, Logan and McClung, 2016, Perry et al., 2009, Young et al., 2007), including hyperactivity, reduced depression and anxiety, and risk-taking behavior. Together with these core components, one of the prominent symptoms of mania is a decreased need for sleep (American Psychiatric Association, 2013, World Health Organization, 2004), which helps to discriminate mania from other disorders with hyperactivity, e.g., attention-deficit/hyperactivity disorder (Marangoni et al., 2015). Therefore, we next performed a detailed analysis of the sleep/wake behaviors in VGAT-CreDTA/VMP mice.

GABAergic VMP Ablation Induces Reduced Daily Sleep Amounts and Loss of Rebound Sleep

Sleep consists of rapid eye movement sleep (REMS) and slow-wave sleep (SWS), also known as non-REM sleep, which are distinguished on the basis of electroencephalography (EEG) and electromyography (EMG) characteristics. It is commonly considered that sleep need is represented during SWS as EEG power in delta frequency (0.5–4 Hz), which increases in animals after prolonged wakefulness (Daan et al., 1984, Lazarus et al., 2019) as well as in mutant mice with prolonged daily sleep amount (Funato et al., 2016, Honda et al., 2018). We first measured the daily sleep amount of VGAT-CreDTA/VMP mice. The VGAT-CreDTA/VMP mice displayed reduced sleep amounts and increased wake amounts compared with control VGAT-CrehrGFP/VMP mice (SWS 24 h: p < 0.0001, SWS light: p < 0.0001, SWS dark: p < 0.001, wake 24 h: p < 0.0001, wake light: p < 0.0001, wake dark: p < 0.001, REMS 24 h: p < 0.001, REMS light: p < 0.0001, REMS dark: p < 0.05, Figures 2A and 2B), consistent with our previous study (Takata et al., 2018). Next, we calculated the delta power during daily SWS to evaluate the mean level of sleep need. In contrast to the severe sleep reduction in VGAT-CreDTA/VMP mice, there was no difference in the EEG power spectra (p = 0.650, Figure S1) and the delta power (p = 0.378, Figure 2C) during SWS over 24 h compared with control VGAT-CrehrGFP/VMP mice. This finding suggests that VGAT-CreDTA/VMP mice are able to maintain the same level of sleep need compared with control mice, despite lower daily sleep amounts. Alternatively, although the daily sleep amount is constantly reduced, it would be an adequate amount for VGAT-CreDTA/VMP mice such that it does not evoke changes in the mean level of the SWS delta power. Thus, VGAT-CreDTA/VMP mice may not be under higher sleep pressure or suffering from greater “sleepiness” despite the reduced sleep amounts. To further investigate sleep/wake phenotypes, we performed sleep deprivation (SD) experiments.

Figure 2

Selective Ablation of the VMP GABAergic Neurons Induces Reduced Daily Sleep Amounts and Loss of Sleep Rebound

(A) EEG, EMG, EEG delta power, and wake/REMS/SWS hypnogram. SD, sleep deprivation.

(B) Total time in SWS, wake, and REMS during 24 h, light and dark periods on the baseline day. SWS 24 h: two-tailed unpaired t test, t = 6.484, df = 12, p < 0.0001. SWS light: two-tailed unpaired t test, t = 7.579, df = 12, p < 0.0001. SWS dark: two-tailed unpaired t test, t = 5.06, df = 12, p = 0.0003. Wake 24 h: two-tailed unpaired t test, t = 6.503, df = 12, p < 0.0001. Wake light: two-tailed unpaired t test, t = 7.855, df = 12, p < 0.0001. Wake dark: two-tailed unpaired t test, t = 4.756, df = 12, p = 0.0005. REMS 24 h: two-tailed unpaired t test, t = 5.231, df = 12, p = 0.0002. REMS light: two-tailed unpaired t test, t = 6.736, df = 12, p < 0.0001. REMS dark: two-tailed unpaired t test, t = 2.2, df = 12, p = 0.0481.

(C) Average of delta power (%) during SWS across 24 h on the baseline day. Two-tailed unpaired t test, t = 0.9161, df = 12, p = 0.3777.

(D) Sleep rebound. Time in SWS during 4 h (20:00–24:00) ad libitum sleep (baseline) and after SD. VGAT-CrehrGFP/VMP: two-tailed paired t test, t = 4.266, df = 7, p = 0.0037. VGAT-CreDTA/VMP: two-tailed paired t test, t = 0.4298, df = 5, p = 0.6852.

(E) Latency to first SWS episode after SD. Two-tailed unpaired t test, t = 2.579, df = 11, p = 0.0256.

(F) Time course of delta power change during SWS after SD compared with baseline delta power. Baseline delta power is the averaged absolute value of delta power during SWS across 24 h on the baseline day. Two-way repeated measures ANOVA followed by Sidak's multiple comparisons. F (1, 11) = 0.00004835, p = 0.9946, the main effect of viral transduction. F (4, 44) = 12.87, p < 0.0001, the main effect of time. VGAT-CrehrGFP/VMP: 0–10 min versus 30–40 min, p = 0.0012; 0–10 min versus 40–50 min, p = 0.0002. VGAT-CreDTA/VMP: 0–10 min versus 30–40 min, p = 0.0323; 0–10 min versus 40–50 min, p = 0.0041.

(G) Time course of delta power change during wake after SD compared with baseline delta power. Baseline delta power is the averaged absolute value of delta power during wake across 24 h on the baseline day. Two-way repeated measures ANOVA. F (1, 9) = 0.3214, p = 0.5846, the main effect of viral transduction. F (4, 36) = 2.341, p = 0.0735, the main effect of time.

(B−G) n = 5–8 mice for each group. Individual values are plotted in the graphs. Data are presented as the mean ± SEM. n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. See also Figure S1.

Prolonged wakefulness with SD leads to an increase in the sleep period by so-called sleep homeostasis (Campbell and Tobler, 1984, Hendricks et al., 2000) among animals. In this study, a normal sleep rebound after SD was observed in control VGAT-CrehrGFP/VMP mice (p < 0.01, Figure 2D), whereas VGAT-CreDTA/VMP mice did not exhibit sleep rebound after 4-h SD (n.s., p = 0.685, Figure 2D), suggesting less sleep pressure after SD in VGAT-CreDTA/VMP mice.

Although the sleep latency after SD was extended in VGAT-CreDTA/VMP mice (p < 0.05, Figure 2E), we found that EEG delta power during SWS was increased after SD and dissipated along with SWS time (not total time) in VGAT-CreDTA/VMP mice (0–10 min versus 30–40 min: p < 0.05, 0–10 min versus 40–50 min: p < 0.01, Figure 2F), despite lacking sleep rebound (Figure 2D). These data suggest that the mechanisms regulating delta power during SWS and sleep amounts are dissociable, consistent with findings from a previous study (Suzuki et al., 2013). Remarkably, SWS delta power dissipation along with SWS time after SD was noticeably similar between VGAT-CreDTA/VMP and the control VGAT-CrehrGFP/VMP mice (n.s., p = 0.995, the main effect of viral transduction, ANOVA; Figure 2F). In contrast, the gradual decay of delta power was not observed during wakefulness (n.s., p = 0.074, the main effect of time, ANOVA; Figure 2G). These observations suggest that the neural mechanisms generate delta power after SD function mainly during SWS and pause during wakefulness. Overall, we revealed that VMP GABAergic neurons have critical roles in the regulation of both daily sleep amounts and homeostatic sleep as a response to SD.

Dopamine D2 Receptors Mediate Sleep Reduction Induced by GABAergic VMP Ablation

Our previous pharmacologic study suggested that D2 and/or D3 receptors mediate the wake-promoting effect induced by ablating or inhibiting VMP GABAergic neurons (Takata et al., 2018). To evaluate the involvement of D2 receptors (D2R) in notable behaviors in VGAT-CreDTA/VMP mice, we generated D2R knockout (KO) mice using fertilized egg donors of VGAT-Cre mice by the CRISPR-Cas9 technique (VGAT-Cre;D2R−/− mice; Figure 3A). We first histologically confirmed that immunostaining for D2R was only detected in VGAT-Cre;D2R+/+ mice and not in VGAT-Cre;D2R−/− mice (Figure 3B). To verify an in vivo functional D2R deficit, we administered a moderate dose (45 mg/kg) of modafinil, a dopamine transporter blocker that has an arousal effect and is largely ineffective in D2R-deficient mice (Qu et al., 2008). Consistent with the previous study, our newly generated VGAT-Cre;D2R−/− mice did not exhibit arousal after modafinil administration (p = 0.788, Figure 3C), in contrast to the significant arousal effect observed in VGAT-Cre;D2R+/+ mice (p < 0.001, Figure 3C).

Figure 3

Generation of D2R Knockout VGAT-Cre Mice by the CRISPR/Cas9 System

(A) Schematic of CRISPR/Cas9-mediated D2R gene knockout using the fertilized egg donor of VGAT-Cre mice. The CRISPR target including the PAM sequence (total 23 bp) contains the part of intron 2 and exon 3 of the D2R gene encompassing the splicing acceptor (SA) site. ssODN, single-stranded oligodeoxynucleotide.

(B) Histologic verification of D2R knockout. Immunostained D2R signals were only detected in D2R+/+ mice and not in D2R−/– mice. Ctx, cortex; CPu, caudate putamen (striatum). Scale bar represents 1,000 μm.

(C) In vivo functional verification of D2R knockout based on the loss of modafinil-induced arousal in D2R−/− mice. D2R+/+: two-tailed paired t test, t = 9.164, df = 4, p = 0.0008. D2R−/−: two-tailed paired t test, t = 0.2872, df = 4, p = 0.7882. n = 5 mice for each group. Individual values are plotted in the graphs.

Data are presented as the mean ± SEM. n.s., not significant, ∗∗∗p < 0.001.

To assess the D2R involvement in sleep behaviors in VGAT-CreDTA/VMP mice, we next injected AAV-FLEX-DTA or control AAV-FLEX-hrGFP virus into the VMP of VGAT-Cre;D2R+/+ mice or VGAT-Cre;D2R−/− mice. We first confirmed that VGAT-CreDTA/VMP;D2R+/+ mice exhibited reduced amounts of SWS over 24 h (p < 0.0001, the main effect of viral transduction, ANOVA; Figure 4A) in both light and dark periods (light: p < 0.001, dark: p < 0.01, Figure 4B) compared with control VGAT-CrehrGFP/VMP;D2R+/+ mice. Although VGAT-CreDTA/VMP;D2R−/− mice also showed reduced SWS amounts over 24 h (p < 0.001, the main effect of viral transduction, ANOVA; Figure 4C), strikingly, VGAT-CreDTA/VMP;D2R−/− mice showed no difference in SWS amounts during the light period compared with control VGAT-CrehrGFP/VMP;D2R−/− mice (light: n.s., p = 0.182, dark: p < 0.001, Figure 4D). In other words, the reduced sleep amounts observed in VGAT-CreDTA/VMP mice during the light period (Figure 2B) were functionally rescued by the loss of D2R. This finding indicates that the sleep reduction induced by ablating VMP GABAergic neurons is mainly mediated by D2R.

Figure 4

Dopamine D2R Mediate Sleep Reduction Induced by GABAergic VMP Ablation

(A) Hourly SWS amounts in D2R+/+ mice. Two-way repeated measures ANOVA followed by Sidak's multiple comparisons. F (1, 9) = 139.5, p < 0.0001, the main effect of viral transduction.

(B) Total SWS amounts in D2R+/+ mice. Light: two-tailed unpaired t test, t = 6.057, df = 9, p = 0.0002. Dark: two-tailed Mann-Whitney U test, U = 0, p = 0.0043.

(C) Hourly SWS amounts in D2R−/− mice. Two-way repeated measures ANOVA followed by Sidak's multiple comparisons. F (1, 9) = 24.86, p = 0.0008, the main effect of viral transduction.

(D) Total SWS amounts in D2R−/− mice. Light: two-tailed unpaired t test, t = 1.447, df = 9, p = 0.1817. Dark: two-tailed unpaired t test, t = 6.14, df = 9, p = 0.0002.

(E) Sleep rebound. Time in SWS during 4 h (20:00–24:00) ad libitum sleep (baseline) and after SD. VGAT-CrehrGFP/VMP;D2R+/+: two-tailed paired t test, t = 8.863, df = 5, p = 0.0003. VGAT-CreDTA/VMP;D2R+/+: two-tailed paired t test, t = 2.042, df = 4, p = 0.1107. VGAT-CrehrGFP/VMP;D2R−/−: two-tailed paired t test, t = 4.091, df = 5, p = 0.0094. VGAT-CreDTA/VMP;D2R−/−: two-tailed paired t test, t = 0.9074, df = 5, p = 0.4058.

(F) Latency to first SWS episode after SD. D2R+/+: two-tailed unpaired t test, t = 2.902, df = 9, p = 0.0175. D2R−/−: two-tailed unpaired t test, t = 2.287, df = 10, p = 0.0452.

(G) Total distance traveled in the open field test. D2R+/+: two-tailed unpaired t test, t = 3.411, df = 18, p = 0.0031. D2R−/−: two-tailed unpaired t test, t = 2.261, df = 18, p = 0.0364.

(H) Immobility time in the tail suspension test. D2R+/+: two-tailed unpaired t test, t = 3.064, df = 30, p = 0.0046. D2R−/−: two-tailed Mann-Whitney U test, U = 45.5, p = 0.0007.

(I) Immobility time in the forced swim test. D2R+/+: two-tailed unpaired t test, t = 4.952, df = 30, p < 0.0001. D2R−/−: two-tailed Mann-Whitney U test, U = 42.5, p = 0.0004.

(J) Schematic model. VMP GABAergic neurons involved in the genesis of mania-like behaviors, including hyperactivity, anti-depressive behaviors, and reduced sleep. The reduced sleep amounts due to dysfunction of VMP GABAergic neurons were mediated in a D2R-dependent manner.

(A−I) n = 5–17 mice for each group. Individual values are plotted in the graphs. Data are presented as the mean ± SEM. n.s., not significant, ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

We next examined the effect of SD, which resulted in the loss of rebound sleep by ablation of VMP GABAergic neurons regardless of D2R+/+ or D2R−/− conditions (p = 0.111, VGAT-CreDTA/VMP;D2R+/+ basal versus SD; p = 0.406, VGAT-CreDTA/VMP;D2R−/− basal versus SD; Figure 4E), suggesting that the impaired sleep homeostasis in VGAT-CreDTA/VMP mice (Figure 2D) is D2R independent. Similarly, the extended latency to SWS after SD (Figure 2E) was D2R-independent (p < 0.05, VGAT-CrehrGFP/VMP;D2R+/+ versus VGAT-CreDTA/VMP;D2R+/+; p < 0.05, VGAT-CrehrGFP/VMP;D2R−/− versus VGAT-CreDTA/VMP;D2R−/−; Figure 4F).

We then assessed VGAT-CreDTA/VMP mice in a set of behavioral tasks to examine the D2R dependency of hyperactivity and anti-depressive behaviors. In the open field test, both VGAT-CreDTA/VMP;D2R+/+ and VGAT-CreDTA/VMP;D2R−/− mice exhibited increased travel distance (p < 0.01, VGAT-CrehrGFP/VMP;D2R+/+ versus VGAT-CreDTA/VMP;D2R+/+; p < 0.05, VGAT-CrehrGFP/VMP;D2R−/− versus VGAT-CreDTA/VMP;D2R−/−; Figure 4G) demonstrating that the hyperactivity in VGAT-CreDTA/VMP mice (Figure 1C) is D2R independent. Consistently, a decreased immobility in both the tail suspension test and forced swim test was observed regardless of D2R+/+ or D2R−/− conditions (tail suspension: p < 0.01, VGAT-CrehrGFP/VMP;D2R+/+ versus VGAT-CreDTA/VMP;D2R+/+; p < 0.001, VGAT-CrehrGFP/VMP;D2R−/− versus VGAT-CreDTA/VMP;D2R−/−; Figure 4H; and forced swim: p < 0.0001, VGAT-CrehrGFP/VMP;D2R+/+ versus VGAT-CreDTA/VMP;D2R+/+; p < 0.001, VGAT-CrehrGFP/VMP;D2R−/− versus VGAT-CreDTA/VMP;D2R−/−, Figure 4I), suggesting that the phenotypes consistent with anti-depressive behaviors in VGAT-CreDTA/VMP mice (Figures 1D and 1E) are also D2R independent.

Discussion

In the present study, we found that ablating GABAergic neurons in the VMP induced various behavioral phenotypes, including hyperactivity, reduced depression and anxiety, loss of sleep rebound, and reduced sleep. The phenotypes in VGAT-CreDTA/VMP mice with hyperactivity in the open field (Figure 1C), reduced immobility in tail suspension (Figure 1D) and forced swim tests (Figure 1E), and increased social interaction (Figure 1F) may be consistent with the facets of mania symptoms in humans, such as increased goal-directed activity and psychomotor agitation, which is described in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (American Psychiatric Association, 2013, Johnson, 2005). Hyperactivity and reduced depression are common features in the reported genetic mouse models of mania, including Shank3-overexpression (Han et al., 2013) and GSK3-β-overexpression mutants (Prickaerts et al., 2006). Importantly, objectively measured high locomotor activity in manic patients quantitively bears a resemblance to the hyperactivity in hyperdopaminergic rodent models on the basis of cross-species translational studies (Perry et al., 2009, Young et al., 2007, Young et al., 2011a). Further, activation of dopaminergic neurons in the ventral tegmental area (VTA) induces hyperactivity in rodents (Boekhoudt et al., 2016). Because the VTA region is anatomically included in the VMP (Figure 1B), ablation of VMP GABAergic neurons may unlock inhibitory outputs on VTA dopaminergic neurons, which could lead to locomotor hyperactivity. To further dissect the detailed mechanisms underlying hyperactivity phenotypes in future studies, dopamine levels should be measured around the nucleus accumbens, a major target area of the VTA, in VGAT-CreDTA/VMP mice. The reduced anxiety and increased risk-taking behavior of VGAT-CreDTA/VMP mice in the elevated plus maze (Figure 1G) are presumably related to excessive involvement in activities with a high potential for a painful consequence in humans (American Psychiatric Association, 2013, Leahy, 1999). Increased risk-taking behaviors are also reported in both manic patients and dopamine transporter knockdown (DAT-KD) hyperdopaminergic mice using comparable tasks (van Enkhuizen et al., 2014, Young et al., 2011b). In addition, the low nest-building performance (Figure 1H) supports the aspect of distractibility described in the diagnostic criteria (American Psychiatric Association, 2013, Oltmanns, 1978, World Health Organization, 2004). VGAT-CreDTA/VMP mice exhibited impaired novel object recognition (Figure 1I), whereas alternation rate in the Y-maze was intact, suggesting that VGAT-CreDTA/VMP mice could remember and refer to the location of the arm they entered previously as working memory during the task. In contrast to the Y maze task, the novel object recognition task has a 24-h test interval (see Transparent Methods), allowing mice to have sleep states and suggesting the possible contribution of sleep to memory consolidation in this specific task. Sleep deprivation or disruption impairs the performance of novel object recognition in mice (Palchykova et al., 2006, Rolls et al., 2011). Thus, the reduced sleep in VGAT-CreDTA/VMP mice may account for their impaired performance in the novel object recognition task. The largely abolished daily sleep amounts in VGAT-CreDTA/VMP mice (Figures 2A and 2B) are consistent with decreased need for sleep in manic individuals (American Psychiatric Association, 2013, World Health Organization, 2004) and ClockΔ19 mutant mice (Naylor et al., 2000), a well-characterized genetic model of mania (Logan and McClung, 2016).

Although some genes, cells, or neural circuits in mammals are suggested to have roles in homeostatic sleep regulation (Halassa et al., 2009, Hayaishi et al., 2004, Ma et al., 2019), the detailed mechanisms remain largely unknown. On the other hand, some behaviors can overcome the homeostatic sleep pressure, even after long-lasting wakefulness. For example, SD induced by an environmental change such as cage exchange shows markedly less sleep rebound compared with gentle handling in mice (Suzuki et al., 2013). VGAT-CreDTA/VMP mice did not show sleep rebound after SD (Figure 2D), indicating that either the VMP GABAergic neurons are critical for producing sleep states in response to SD or the behaviors induced by the loss of the neurons overcame the homeostatic sleep pressure induced by SD.

Increasing evidence indicates that dopamine and D2R promote wakefulness to suppress sleep (Oishi and Lazarus, 2017, Oishi et al., 2017, Qu et al., 2008, Qu et al., 2010). In this study, using a constitutive genetic D2R KO (Figure 3), we clarified that sleep loss (promoting wake) in VGAT-CreDTA/VMP mice was dependent on D2R, at least during the light period (Figure 4). These data imply that the VMP GABAergic neurons suppress mainly dopaminergic systems and D2R to presumably prevent excessive amounts of wakefulness. It is unclear which dopaminergic systems are suppressed by the VMP. Anatomic studies (Jhou et al., 2009a, Jhou et al., 2009b, Omelchenko and Sesack, 2009, Taylor et al., 2014) indicate that the GABAergic VTA and rostromedial tegmental nucleus, both of which are included in the VMP, innervate dopaminergic neurons in the VTA and dorsal raphe nucleus (DRN), also known as the ventral periaqueductal gray matter, which regulate wakefulness (Cho et al., 2017, Eban-Rothschild et al., 2016, Lu et al., 2006, Oishi et al., 2017, Taylor et al., 2016). The functional relationship between VMP and these dopaminergic areas should be examined in the future. Serotonin (5-HT) is also involved in regulating both sleep/wake and locomotor activity (Correia et al., 2017, Saper et al., 2005). Studies in both humans and mice suggest that serotonin is involved in mania and its treatment (Maddaloni et al., 2018, Shiah and Yatham, 2000). In this sense, investigating the loss of VMP GABAergic innervation to DRN serotonergic neurons is another potential study direction.

The mechanisms related to hyperactivity, anti-depressive behaviors, and reduced sleep amount during the dark period in VGAT-CreDTA/VMP mice remain unclear. Recent studies revealed GABAergic projections from the VTA to the lateral hypothalamus (Taylor et al., 2014, Yu et al., 2019) and especially onto orexin neurons (Chowdhury et al., 2019), which are critical for maintaining wakefulness (Chemelli et al., 1999). Optogenetic activation of GABAergic VTA terminals in the lateral hypothalamus suppresses wakefulness (Chowdhury et al., 2019, Yu et al., 2019). Moreover, lateral hypothalamic kindling induces mania-like behaviors (Abulseoud et al., 2014). Therefore, the lateral hypothalamus might play a role in the mania-like behaviors induced by GABAergic VMP ablation.

In summary, as illustrated in Figure 4J, our findings revealed that VMP GABAergic neurons are involved in various aspects of mania-like behaviors, and the reduced sleep amounts are mediated through the D2R-dependent dopaminergic system. Further research using VGAT-CreDTA/VMP mice may help to elucidate the different aspects of mania as well as the detailed neural mechanisms of sleep homeostasis and contribute to the development of safer and more effective drugs.

Limitations of the Study

Pharmacological studies in VGAT-CreDTA/VMP mice, such as of lithium and valproic acid, which are the commonly used treatments for bipolar disorder (Chiu et al., 2013, Logan and McClung, 2016), are important for further investigation of the effects on individual behavioral phenotypes and D2R-dependency. Because all the experiments were performed using male mice in the present study, investigating the phenotypical difference with female mice provides further information about our findings.

Resource Availability

Lead Contact

Further information and requests for reagents and resources should be directed to and will be fulfilled by the Lead Contact, Yo Oishi (oishi.yo.fu@u.tsukuba.ac.jp).

Materials Availability

The mouse line generated in this study (D2R KO strain) will be made available with a completed Material Transfer Agreement.

Data and Code Availability

The data that support the findings of this study are available from the Lead Contact on reasonable request.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.