Median preoptic GABA and glutamate neurons exert differential control over sleep behavior

Summary

Previous studies suggest that the median preoptic nucleus (MnPO) of the hypothalamus plays an important role in regulating the wake-sleep cycle and in particular homeostatic sleep drive. However, the precise cellular phenotypes, targets and central mechanisms by which the MnPO neurons regulate the wake-sleep cycle remain unknown. Both excitatory and inhibitory MnPO neurons innervate brain regions implicated in sleep promotion and maintenance, suggesting that both cell types may participate in sleep control. Using genetically-targeted approaches, we investigated the role of the MnPO GABAergic (MnPO^Vgat) and glutamatergic (MnPO^Vglut2) neurons in modulating wake-sleep behavior of mice. We found that both neuron populations differentially participate in wake-sleep control, with MnPO^Vgat neurons being involved in sleep homeostasis, and MnPO^Vglut2 neurons promoting sleep during allostatic (stressful) challenges.

In brief

Machado et al. show that GABA neurons in the median preoptic nucleus (MnPO) mediate sleep homeostatic drive, whereas glutamate neurons modulate the effects of acute stress on sleep.

Graphical Abstract

Introduction

Sleep is necessary for both physical and mental wellbeing. The harmful effects of sleep deficiency are clearly demonstrated when humans or other animals are subjected to acute or chronic sleep deprivation, which results in deleterious effects on metabolism, cardiovascular function, mental health and cognition^1–4.

Sleep is regulated by circadian, homeostatic, and allostatic drives². The circadian input to wake-sleep regulation is relayed from the suprachiasmatic nucleus to the subparaventricular zone, and from there to the dorsomedial nucleus of the hypothalamus and on to wake-promoting systems, such as the orexin neurons, and sleep-promoting systems, including the ventrolateral preoptic nucleus (VLPO)⁵. The homeostatic drive for sleep is evident from the increase in both duration and depth of sleep after a prolonged period of wakefulness, but its origin is poorly understood. There is evidence for buildup of metabolic byproducts of energy metabolism such as accumulation of adenosine or prostaglandin D2 during prolonged wakefulness causing sleepiness^6,7. However, it is not clear on what neural targets these humoral mediators work. One line of evidence suggests that firing of the neurons of the median preoptic nucleus (MnPO) may increase during sleep deprivation (SD) and thus represent homeostatic sleep drive, but it is not clear how this would affect wake-sleep circuitry⁸. Allostatic loads, such as psychogenic stressors, also affect sleep. Stress-induced insomnia is common in both humans and experimental animals, and involves activation of wake-promoting systems in the brain, causing delay of sleep onset, more frequent arousals, and early awakening^9,10. Chronic insomnia is also commonly observed in patients with psychiatric disorders such as anxiety, depression and post-traumatic stress disorder^9,11. There is evidence that both the medial prefrontal cortex and the extended amygdala contribute to the hyperarousal associated with stress ¹², but the downstream circuitry that modulates stress-induced insomnia is not understood. The evidence suggesting a role for the MnPO neurons in homeostatic drive for sleep derives from studies showing that SD increases the firing rate as well as causing cFos expression (a marker for neural activation) in MnPO neurons¹³. Many of the cFos-positive MnPO neurons during SD also contain glutamic acid decarboxylase (GAD), suggesting that they may be GABAergic^8,14. By contrast, VLPO neurons, which are thought to inhibit the arousal system to promote sleep ^15,16, do not show cFos expression during SD, but only when the animal actually is permitted to fall asleep ¹⁷. Interestingly, cFos activation of MnPO neurons has also been observed after a period of wakefulness caused by psychogenic stress ^10,18, raising the question of whether the activation of these neurons represents a direct response to the stress or homeostatic drive due to sleep loss.

In a recent study, Moffitt and colleagues found that many classes of preoptic neurons that express GAD1, GAD2, or both do not express the vesicular GABA transporter (Vgat). Surprisingly, they instead express a vesicular transporter for glutamate (Vglut2). Thus, these GAD-positive neurons are actually glutamatergic (excitatory) rather than GABAergic (inhibitory) ¹⁹. These results raise the question whether MnPO GAD+ neurons that show cFos expression during SD and psychogenic stress are inhibitory, excitatory or perhaps a mixture of the two. The idea that both excitatory and inhibitory MnPO neurons may regulate the wake-sleep cycle is also supported by anatomical evidence: MnPO neurons expressing both Vgat and Vglut2 innervate several brain regions critically involved with wake-sleep regulation²⁰, including the sleep-active galaninergic neurons in the VLPO ²¹ and the orexinergic field located in the lateral hypothalamic area (LH)²⁰.

Strong evidence for the involvement of MnPO GABAergic neurons in sleep control comes from a recent study using selective activation of Vgat-expressing neurons in the MnPO (MnPO^Vgat neurons). Chemogenetic activation of MnPO^Vgat neurons produced an increase on non-rapid eye movement (NREM) sleep and a reduction of REM sleep during the light phase²². However, mice are usually asleep during the light phase, and the ability of these neurons to drive sleep during the active (dark) phase was not tested. On the other hand, activation of glutamatergic MnPO neurons has been shown to cause a profound hypothermia^23–26, which itself affects sleep, so the effect of these neurons on NREM sleep is difficult to assess by this method. These results point to a possible role of both inhibitory (Vgat) and excitatory (Vglut2) neurons in the MnPO in the regulation of wake-sleep. However, it is not clear which of these MnPO neurons are involved in the homeostatic or allostatic regulation of sleep.

To study the precise cellular phenotypes and mechanisms by which MnPO neurons may regulate the wake-sleep cycle, we have employed genetically-targeted methods to determine the distinct roles of MnPO^Vgat and MnPO^Vglut2 neurons in sleep homeostasis and allostasis.

Results

Activation of MnPO^Vgat neurons during the active phase produces an increase in NREM sleep, reduction of REM sleep and decrease in body temperature.

The median preoptic GABAergic neurons are thought to regulate sleep homeostasis ^8,27,28. To test the ability of these neurons to promote sleep onset during the active period, we used a chemogenetic approach to promote selective activation of MnPO^Vgat neurons. For these experiments, we injected AAV8-hSyn-DIO-hM3D(Gq)-mCherry (hM3Dq) in the MnPO of Vgat-IRES-cre mice (hM3Dq_Vgat mice). Then, 15 minutes before dark onset, these mice were treated in a randomized order with a specific ligand for hM3Dq, clozapine-N-oxide (CNO, 0.3 mg/kg), or saline as a control. We observed that during the first three hours of the dark phase (ZT12-ZT14), activation of MnPO^Vgat neurons induced a statistically significant increase in NREM sleep (26.09 % ± 12.5 SEM) at the expense of both REM (−52.16 % ± 18.7 SEM) and wake (−6.46 % ± 4.4 SEM), although there is a statistically significant reduction only for REM sleep (Figure 1a). The effect size of the response to CNO treatment was calculated for each mouse relative to their control (saline) condition over the ZT12-ZT14 period. These results were consistent with previous observations using a similar approach ²² during the light phase, but indicate that MnPO^Vgat neurons can drive sleep during the adverse circadian phase. Activation of MnPO^Vgat neurons induced the first NREM sleep episode with a shorter latency (Figure 1b). Importantly, the increase in NREM sleep is not caused by CNO itself, as CNO injection does not promote changes in wake-sleep of Vgat-IRES-cre mice that do not express hM3Dq ²². We also examined EEG during NREM sleep (for the first 3 hours after treatment with CNO) and found no change in relative power in the delta (0.5–4Hz), theta (4–8 Hz) or alpha (8–13 Hz) bands compared to saline treatment (Figure 1c).

Figure 1. Activation of MnPO^Vgat neurons causes earlier onset and increased amount of NREM sleep.

(a) Vgat-IRES-Cre mice expressing AAV-DIO-hM3Dq were injected with CNO (0.3mg/kg, red) or saline (black) 15 minutes before ZT12. CNO injection induced an increase in percent time in NREM sleep (34.1 ± 3.4 CNO vs. 27.1 ± 2.5 SAL, n=8 each group, p=0.03, paired t-test) and reduced REM sleep (0.5 ± 0.1 CNO vs. 1.9 ± 0.6 SAL, n=8 each group, p=0.03, paired t-test) during the first three hours of the dark phase after drug administration. (b) The treatment with CNO also reduced the latency for the first NREM episode (19 ± 2.7 minutes after CNO; red) when compared to the same mice injected with saline (black; 44 ± 10 minutes, p=0.03, paired t-test n=8 each group). (c) After CNO treatment, mice showed no changes in the EEG power during NREM sleep when compared to saline treatment. (d) Beginning about 30 minutes after treatment with CNO (red), these mice also show a 1.2 °C reduction of the stress hyperthermia usually seen over the next hour after handling, when compared to treatment with saline (black) (36.7 ± 0.15 °C CNO vs. 37.92 ± 0.12 °C SAL, n=7, p=0.008, t-test multiple comparisons using the Sidak-Bonferroni). (e) A brain map of the location of the injection sites of hM3Dq in Vgat-IRES-cre mice. (f) A representative image showing native mCherry (red) expression in the hM3Dq-expressing Vgat+ neurons in the MnPO of a Vgat-IRES-cre mouse.

The CNO treatment was also associated with a reduction in the typical hyperthermia due to handling the animals from about 1.8° C in saline treated animals to 1.2 °C in CNO treated animals at 30 minutes after the injection (asterisk in Figure 1d). However, the body temperature (Tb) of the animals was otherwise unaffected. Mapping the hM3Dq injection sites in Vgat-IRES-cre mice used in these experiments (Figures 1e,f) demonstrated that they accurately targeted the MnPO. Examination of cFos expression in hM3Dq-expressing MnPO^Vgat neurons in a subset of animals showed a dramatic increase in the number of cFos –positive MnPO neurons after CNO (752.2 ± 242 cFos+ after CNO, n=3, vs. 52.2 ± 2.52 in the control animals after saline, n=2).

Ablation of MnPO^Vgat neurons produces an increase in wakefulness and reduction of NREM during the light phase

To investigate whether MnPO^Vgat neurons directly regulate sleep homeostasis, we then injected a Cre-dependent AAV-mCherry-DIO-DTA (AAV-DTA) that produces mCherry in all transduced neurons and diphtheria toxin A in Cre-expressing neurons (which kills them) (Figure 2a) in the MnPO of Vgat-IRES-cre mice or Vgat-IRES-cre-L10-GFP mice (MnPO^Vgat-ablated, DTA_Vgat mice) or WT (DTA_WT) littermates as controls. As we previously showed with this method, there is deletion of Vgat neurons at the injection site, which can be identified by the absence of Vgat+ neurons that express GFP (in Vgat-IRES-cre-GFP mice) in the area labeled with mCherry (Figure 2a), which is expressed in non-Vgat neurons at the injection site ²⁹. We started the recordings between 4–6 weeks after the AAV-DTA brain injections. During the baseline (undisturbed) condition, we found that ablation of MnPO^Vgat neurons increased the mean time spent in wake by approximately 22.2% compared to DTA_WT_Baseline and reduced NREM sleep by about 11.6 % during the light phase (ZT0-ZT10) (Figures 2b and c). These changes were particularly pronounced during the first three hours of the light phase (when sleep pressure is usually highest; Figure 2b). However, no statistically significant changes in REM sleep were observed during the light period (Figures 2b and c). During the subsequent dark phase, no significant changes were seen in the time spent in wake, NREM or REM sleep (Figures 2b and d). We also compared the sleep behavior of DTA_Vgat and DTA_WT mice with Vgat-IRES-cre mice expressing hM3Dq in the MnPO (hM3Dq_Vgat) at least 12h after saline injection, as a control for the mouse strain (Figure S1). As expected, we found that hM3Dq_Vgat mice spent less time awake when compared to DTA_Vgat mice during the light phase (ZT0-ZT10), but no differences were found between the two controls (WT mice injected with AAV-DTA and Vgat-IRES-cre mice injected with hM3Dq) in wakefulness, NREM sleep or REM sleep. A brain map of the injection sites (Figure 2e) showed that they accurately targeted the MnPO.

Figure 2. Ablation of MnPO^Vgat neurons produces an increase in wakefulness and a reduction of NREM sleep during the light phase.

(a) A representative injection site for AAV-mCherry-DIO-DTA in the MnPO of a Vgat-IRES-cre-L10-GFP reporter mouse. mCherry expression (red) in the upper panel marks the non-Vgat neurons at the injection site, while within that region (middle and lower panels) there are no remaining green Vgat+ neurons. (b) Time course of wakefulness, NREM sleep and REM sleep in the baseline condition (ZT0-ZT22) in DTA_Vgat mice (red) and intact DTA_WT littermate mice (black). DTA_Vgat mice have more wake and less NREM and REM sleep during the first 3 hrs after light onset. The loss of REM but not NREM sleep is made up during the remainder of the day. (c) Bar graphs showing total wake, NREM and REM sleep time after injecting AAV-DTA in WT and Vgat-Cre mice. During the first 10 hrs of the light cycle, mice with ablation of MnPO^Vgat neurons show increased time spent awake (41.8 ± 1.5% SEM DTA_Vgat_Baseline WAKE vs. 34.2% ± 2.2 SEM DTA_WT_Baseline WAKE, n=11,8 respectively, p=0.01, unpaired t-test), and decreased time spent in NREM sleep (50.5 ± 1.4% SEM DTA_Vgat_Baseline NREM vs. 57.1% ± 2.0 SEM DTA_WT_Baseline NREM, n=11,8 respectively, p=0.01, unpaired t-test), but no change in REM sleep (8.0 ± 0.4% SEM DTA_Vgat_Baseline REM vs. 8.0% ± 0.4 SEM DTA_WT_Baseline WAKE, n=11,8 respectively, p=0.9, unpaired t-test) (see also Figure S1). Two-way ANOVA, Bonferroni’s multiple comparisons test also reveals significant difference in NREM sleep (F _{(10, 198)} = 2.4, p=0.007) and time spent in wake (F _{(10, 187)} = 2.2, p=0.01) at ZT2 and ZT6. (d) During the first ten hours of the subsequent dark phase, the same mice showed no statistically significant changes in the time spent in wake, NREM or REM sleep (i.e., did not recover the sleep lost in the first hours of the light phase). (e) The injection sites of the AAV-DTA in Vgat-IRES-cre are shown in a brain map.

Mice with ablation of MnPO^Vgat neurons do not show as much sleep after sleep deprivation as WT mice.

Because the loss of sleep primarily in the early light phase in animals with deletion of MnPO^Vgat neurons suggested a role in regulating sleep homeostasis, we performed 4h of sleep deprivation (SD) at the onset of the light (sleep) phase (ZT0-ZT3, 7am-11am), by using novel objects or nesting materials to keep mice awake (Figure 3a) ^30,31. To test the efficacy of our SD protocol, we sleep-deprived 2 Vgat-IRES-cre mice injected with AAV-DTA in the MnPO and 3 that had not been injected with AAV-DTA, and found that mice spent about 90–95% of the time awake during the SD protocol time (95.35 Vgat-cre_SD ± 1.46 vs. 91.95 ± 3.38 DTA_Vgat_SD). We then tested the effect of SD on a cohort of 8 DTA_WT_SD and 8 DTA_Vgat_SD mice by examining EEG power spectrum and cumulative sleep over 19h. We analyzed the EEG power spectrum during NREM sleep for the 3 hours immediately after SD or during the baseline condition of both groups (ZT4-6), but found no statistically significant difference in power in the delta, theta, or alpha band (Figure 3d). We also examined cumulative sleep over the 19h period after SD (ZT4-ZT22) to determine if there was sleep recovery. Although DTA_Vgat_SD mice showed less NREM (about 47 min) than DTA_WT_SD mice (about 61 min of NREM sleep recovery) (Figure 3d), this did not reach statistical significance. However, the DTA_Vgat_SD group had about the same amount of NREM as the DTA_WT_Baseline mice, and substantially less NREM sleep than the DTA_WT_SD group (NREM sleep: 537.25 ± 11.06 minutes DTA_WT_SD vs. 464.62 ± 12.57 minutes DTA_Vgat_SD vs. 476.05 ± 19.12 minutes DTA_WT baseline vs. 417.55 ± 19.47 DTA_Vgat_Baseline, n= 8 each group, F_(54,504)=4.77, p<0.0001, Two-way ANOVA, Bonferroni’s multiple comparisons test). The REM sleep recovery was similar in both DTA_WT_SD (about 12 minutes) and DTA_Vgat_SD (about 16 minutes) groups (REM sleep: 65.39 ± 4.17 minutes DTA_WT_SD vs. 75.96 ± 8.18 minutes DTA_Vgat_SD vs 53.18 ± 3.77 minutes DTA_WT_Baseline vs. 60.21 ± 5.97 minutes DTA_Vgat_Baseline, n= 8 each group, F_(54,532)=0.67, p=0.96, Two-way ANOVA, Bonferroni’s multiple comparisons test). Thus, the DTA_Vgat mice have residual homeostatic sleep drive, but require a much higher degree of sleep pressure to produce the same amount NREM of sleep. These data suggest that the MnPO^Vgat neurons are only a component of homeostatic sleep drive, but that the other components cannot compensate fully for their loss.

Figure 3. Ablation of MnPO^Vgat neurons reduces NREM sleep recovery after sleep deprivation.

(a) To test the role of MnPO^Vgat neurons in regulating sleep homeostasis, we performed 4 hours of sleep deprivation (SD) at the onset of the light phase (ZT0-ZT3, 7am-11am) by using novel objects or nesting materials. (b) Time course of Wake, NREM sleep and REM sleep after SD (ZT4-ZT22) in DTA_Vgat_SD mice (red) and intact DTA_WT_SD littermate mice (black), compared to baseline wake-sleep for each group. (c) There were no statistically significant changes in EEG power in DTA_WT (n=8) or DTA_Vgat (n=8) after SD (ZT4-6) when compared to their baseline condition at the same time. (d) During the 19 h following SD, both the DTA_WT (61.2 ± 22 SEM) and DTA_Vgat (47.0 ± 23 SEM) mice recovered sleep compared to their baseline, and the difference was not statistically significant (Unpaired t-test, p=0.2). However, the total amount of NREM sleep for the DTA_Vgat mice (464.62 ± 12.57 SEM) was less than the DTA_WT mice (537.25 ± 11.06 SEM; p<0.0001) under both SD and at baseline (476.05 ± 19.12 SEM minutes DTA_WT baseline vs. 417.55 ± 19.47 SEM DTA_Vgat_Baseline, p=0.001, n= 8 each group, F_(54,504)=4.77, p<0.0001, Two-way ANOVA, Bonferroni’s multiple comparisons test). The amount of NREM sleep for the DTA_Vgat mice after SD was almost identical with the DTA_WT mice at baseline, showing that they can sleep equal amounts but require a higher degree of sleep pressure to do so. REM sleep responses were qualitatively similar, but due to large variances, only the difference between baseline and SD for the DTA_Vgat animals reached statistical significance (75.96 ± 8.18 SEM minutes of cumulative REM sleep in DTA_Vgat_SD vs 60.21 ± 5.97 SEM minutes of cumulative REM sleep in DTA_Vgat_Baseline, F_(54,504)=2.65, p<0.0001, Two-way ANOVA, Bonferroni’s multiple comparisons test).

Ablation of MnPO^Vglut2 neurons does not change wake-sleep patterns during the light phase, but prevents sleep recovery after sleep deprivation.

We did not attempt to activate the MnPO^Vglut2 neurons with hM3Dq and CNO because in our previous work we had found that this causes profound hypothermia²⁵, which itself can affect the amount of sleep. Instead, we used deletion of the MnPO^Vglut2 neurons using the DTA vector (Figure 4a), as our previous work had shown that this does not affect baseline thermoregulation²⁹. Ablation of MnPO^Vglut2 neurons did not alter the baseline time spent in wake, NREM sleep or REM sleep during the light phase (ZT0-ZT10) (Figures 4b and c) or during the dark phase (Figure 4d). We then used novel objects to sleep deprive mice for 4h (ZT0-ZT3) (Figure 4f). DTA_Vglut2_SD mice showed a small but statistically significant increase in delta power during NREM sleep when compared to their baseline condition, indicating that SD caused an increase in sleep pressure in DTA_Vglut2_SD mice (Figure 4g) that was at least as great as in DTA_WT mice. During the entire 19 hr period after SD, the amount of sleep recovery for the DTA_Vglut2 mice (n=9) was nearly identical with their DTA_WT littermates (n=6; Figure 4h). However, during the first 3 hr after SD, the DTA_WT_SD mice showed a robust (22.3 ± 5.3 SEM minutes, p=0.03) increase in sleep compared to baseline which was statistically significant. DTA_Vglut2_SD mice had a less robust sleep recovery during this period (10.5 ± 6.6 SEM minutes, p=0.15) which was not statistically different from baseline during these hours (ZT4-6). Our study was powered to detect a 15 minutes increase in NREM sleep to be statistically significant in the DTA_WT_SD group compared to their baseline condition, and 21 minutes in the DTA_Vglut2_SD compared to their baseline condition. However, the difference between the groups subjected to SD was not statistically significant due to the unexpectedly large variance (10.5 minutes ± 19.9 standard deviation DTA_Vglut2 vs 22.3 minutes ± 13 standard deviation DTA_WT, unpaired t-test p=0.1). The increase in delta power during NREM sleep and eventual recovery of amount of sleep after sleep deprivation indicates that the DTA_Vglut2 mice do not have a deficit in sleep homeostasis.

Figure 4. Ablation of MnPO^Vglut2 neurons does not change baseline wake-sleep patterns or long-term sleep recovery after SD, but reduces rebound to SD during the light phase.

(a) A representative injection site for AAV-mCherry-DIO-DTA in the MnPO of a Vglut2-IRES-cre-L10-GFP reporter mouse. mCherry expression (red) in the upper panel marks the non-Vgat neurons at the injection site, while within that region (middle and lowes panel) there are no remaining green Vgat+ neurons. No colocalization of Vglut2+ neurons (green) and mCherry (red) was seen. (b) Graphs showing the time course of wakefulness, NREM sleep and REM sleep in the baseline condition (ZT0-ZT22) in DTA_Vglut2 mice (blue) vs. intact DTA_WT littermate mice (black). No differences were seen. (c) Ablation of MnPO^Vglut2 neurons did not alter the time spent during the light phase in wake (44.7 ± 1.2%SEM DTA_Vglut2_Baseline vs. 41.8 ± 2.5%SEM DTA_WT_Baseline, n=10,6 respectively, p=0.3, unpaired t-test), NREM sleep (47.22 ± 1.1% SEM DTA_Vglut2_Baseline vs. 50.4% ± 2.1 SEM DTA_WT_Baseline, n=10,6 respectively, p=0.21, unpaired t-test) or REM sleep (8.0 ± 0.3%SEM DTA_Vglut2_Baseline vs. 7.8 ± 0.4%SEM DTA_WT_Baseline, n=10,6 respectively, p=0.66, unpaired t-test). (d) During the active (dark) phase, DTA_Vglut2 mice (blue) showed no difference in the amount of time spent in the wake state, NREM or REM compared to DTA_WT mice (black). (e) A map showing injection sites of the AAV-DTA in Vglut2-IRES-cre mice. (f) We used novel objects to sleep deprive mice for 4 hours (ZT0-ZT3). (g) EEG power analyses showed an increase in delta power in DTA_Vglut2 (n=9) mice during the 3 hr after SD (ZT4-6) when compared to their baseline condition (ZT4-6) (F_(6,78)= 5.9, p=0.007, Two-way ANOVA, Tukey’s multiple comparisons test. (h) DTA_Vglut2 mice (blue) did not show a statically significant increase in NREM sleep 3 hr after SD (10.5 ± 6.6 SEM minutes, p=0.15), but DTA_WT mice (black) did (22.3 ± 5.3 SEM minutes, p=0.03). DTA_Vglut2_SD mice caught up on total sleep so that for the entire 19 hour period after sleep deprivation, DTA_WT_SD (n=6) and DTA_Vglut2_SD (n=9) mice showed similar amounts of time spent in NREM (463.32.9 ± 32.9 SEM minutes DTA_WT_SD vs. 464.35 ± 16.59 SEM minutes DTA_Vglut2_SD) and REM sleep (72.77 ± 1.03 SEM minutes DTA_WT_SD vs. 75.8 ± 4.86 SEM minutes DTA_Vglut2_SD). Both groups subjected to SD show a lower amount of time spent in wake when compared to their baseline conditions (ZT22: 567.8 ± 36.0 minutes of cumulative wake in DTA_WT_SD vs. 598.15 ± 16.16 DTA_Vglut2_SD vs. 661.61 ± 12.01 DTA_WT_Baseline vs. 656.78 ± 20.14 DTA_Vglut2_Baseline, F_(54,486)=2.71, n=6,6,9 and 10, p<0.0001, Two-way ANOVA, Bonferroni’s multiple comparisons test). Thus, ablation of MnPO^Vglut2 neurons attenuated sleep recovery for a few hours immediately after SD that it is compensated for in the subsequent dark phase.

We then wondered whether the DTA_Vglut2 mice showed a reduction of NREM sleep recovery immediately following the SD and both strains showed an unexpectedly high variance of sleep recovery due to an emotional response to the novel object SD treatment. This would be similar to humans who, after staying awake to watch an exciting movie, may find it difficult to fall asleep immediately, but have an intact homeostatic drive and eventually do catch up on their sleep. Although the novel objects method of sleep deprivation was designed to be less stressful than methods that involve physical manipulation of the animals, it may not be stress-free. We therefore wanted to examine whether the deletion of the MnPO^Vglut2 neurons might reduce the ability of the animals to cope with stress.

MnPO^Vglut2 neurons, but not MnPO^Vgat neurons, limit stress-induced insomnia

To test the hypothesis that MnPO^Vglut2 neurons modulate wake-sleep responses after stressors, we used a cage exchange stress paradigm that typically causes insomnia accompanied by an elevation of body temperature of about 2° C (stress hyperthermia) ^29,32. Vglut2-cre, Vgat-cre male mice and their WT littermates injected with AAV-mCherry-DIO-DTA in the MnPO were permitted to sleep during the first 3–4h of the light phase before being transferred to a cage previously occupied by another male mouse for several days (dirty cage) (Figure 5a).

Figure 5. MnPO^Vglut2 neurons but not MnPO^Vgat neurons limit stress-induced insomnia.

(a) Schematic figure of the cage exchange stress protocol in which WT, Vglut2-cre and Vgat-cre mice injected with AAV-DTA into the MnPO are transferred at ZT4 to a cage previously occupied by another male mouse. (b) DTA_Vglut2_STRESS mice (dark blue) show an increase in EEG delta power during the first three hours of stress (ZT4-6) when compared to their baseline condition (light blue) (F_(6,87)= 3.5, p=0.003, DTA_Vglut2 (n=10) DTA_WT (n=7) determined using a two-way ANOVA, Bonferroni’s multiple comparisons test and during ZT7-9 when compared to the DTA_WT_STRESS group (black) (F_(6,84)= 3.1, p=0.03, DTA_Vglut2_STRESS (n=10) DTA_WT_STRESS (n=7) determined using the two-way ANOVA, Bonferroni’s multiple comparisons test), indicating increased homeostatic drive for sleep. (c) However, despite this homeostatic response, DTA_Vglut2_STRESS mice (blue) show greater wakefulness between ZT5 and ZT7 when compared to the DTA_WT mice (black) or Vglut2-cre mice that had not been injected with AAV-DTA (Vglut2-cre STRESS) as controls (ZT6: 69.2 ± 5.2% SEM DTA_Vglut2_STRESS vs. 44.5 ± 5.9% SEM DTA_WT_STRESS vs. 41.98 ± 9.5% SEM Vglut2-cre STRESS, n=10,7,5 respectively, p=0.01 determined using the two-way ANOVA, Bonferroni’s multiple comparisons test). There is a concomitant reduction in NREM sleep (ZT6: 27.8 ± 4.5% SEM DTA_Vglut2_STRESS vs. 49.4 ± 5.0% SEM DTA_WT_STRESS vs. 60.5 ± 7.0% SEM Vglut2-cre STRESS, n=10,7,5 respectively, * DTA_Vglut2 vs. DTA_WT ^#DTA_Vglut2 vs. Vglut2-cre, p<0.001 determined using the two-way ANOVA, Bonferroni’s multiple comparisons test). We did not find any difference between the two control groups (DTA_WT_STRESS and Vglut2-cre STRESS). Similar changes were seen in REM sleep (see also Table S1). (d) Ablation of the GABAergic MnPO neurons (red, n=6) did not change the time spent in wake when compared to WT animals (black, n=9) also exposed to cage exchange stress and no changes were seen in NREM sleep or REM sleep (see also Table S2).

WT littermates of DTA_Vglut2-Cre (DTA_WT_STRESS) mice showed a dramatic increase in wake time, a decrease in NREM sleep and complete loss of REM sleep for the first hour after cage exchange (ZT4) when compared to their baseline condition (see Figures 5b–d), but then quickly fell back into their normal sleep behavior. DTA_Vglut2_STRESS mice showed a similar increase in wakefulness at ZT4 in response to cage exchange stress but had a more sustained wakefulness when compared DTA_WT_STRESS mice, which lasted for about 3h (ZT4-ZT6). This response was associated with a concomitant reduction of NREM sleep and REM sleep (see Figure 5c). This sleep loss was not due to a reduction in homeostatic sleep pressure, because when the DTA_Vglut2_STRESS mice did sleep, EEG power analysis revealed an increase in delta power during NREM sleep in DTA_Vglut2_STRESS mice when compared to their baseline condition (ZT4-6), and when compared to DTA_WT_STRESS mice in subsequent hours (ZT7-9) (Figure 5b).

As a further control, we also evaluated the changes in sleep behavior induced by stress in Vglut2-IRES-cre mice that had not been injected with AAV-DTA (strain control). As expected, the Vglut2-Cre mice showed essentially the same pattern of excess wakefulness during the stress protocol as did the DTA_WT_STRESS mice. We also evaluated the role of MnPO^Vgat neurons in modulating the wake-sleep cycle during a stressful stimulus. Ablation of the GABAergic MnPO neurons did not change the time spent in wake, NREM sleep, or REM sleep when compared to WT animals also exposed to cage exchange stress (Figure 5d).

In summary, the DTA-Vglut2 mice had significantly more prolonged loss of sleep in response to cage exchange stress than the DTA_WT_STRESS mice, DTA_Vgat_STRESS mice, or Vglut2-cre control mice (Table S1 and Table S2). The exacerbation of stress-induced insomnia in MnPO^Vglut2-ablated mice when compared to these controls (Figure 5) suggests that MnPO^Vglut2 neurons modulate the effects of a psychogenic stressful stimulus (cage exchange stress) on subsequent wakefulness. We have previously demonstrated that ablation of MnPO^Vglut2 neurons potentiated hyperthermia and hyperactivity induced by cage exchange stress ²⁹. Thus, the role of the MnPO^Vgat neurons in recovery from sleep deprivation appears to be limited to the homeostatic response, whereas the MnPO^Vglut2 neurons may be promoting recovery sleep by minimizing the allostatic insomnia caused by stress.

Discussion

MnPO^Vgat neurons and sleep homeostasis

Sleep is a homeostatically regulated behavior with an increase of sleep need accumulating when the animal is awake. Several studies suggest that MnPO neurons are active during the increase in the homeostatic drive for sleep ^8,14,27,28, and that the neurons that are more active during prolonged wakefulness are GABAergic. However, there has been no evidence that these neurons are necessary for homeostatic accumulation of sleep drive, and only recently has it been reported that driving these neurons during the light phase, when mice ordinarily sleep most of the time, could promote NREM sleep at the expense of REM sleep²². There has been no previous study of the effect of these neurons during the dark phase, when the animals are ordinarily awake. In this study, we first used a chemogenetic approach to promote activation of MnPO^Vgat neurons during a time of the day when sleep pressure is low (beginning of the dark phase) and found that activation of MnPO^Vgat neurons reduced the latency for the first NREM sleep episode and produced an increase in NREM sleep while REM sleep was reduced. Thus, the MnPO^Vgat neurons can drive NREM sleep not only during the light phase when animals are usually sleeping anyway, but also during the adverse phase, when animals are ordinarily awake most of the time. This is important for homeostatic sleep drive, which typically comes into play during the wake phase after there is sleep disruption during the prior sleep phase. In addition, activating the MnPO^Vgat neurons causes a small, brief fall in Tb (Figure 1d, see also Hrvatin et al., 2020 extended Figure 8b ³³). The body cooling induced by activation of the MnPO^Vgat neurons, which is likely to be caused by cutaneous vasodilation and consequent heat loss through the skin, may play a role in promoting initiation and maintenance of sleep ^34,35.

In agreement with these results, we found that during the light phase, MnPO^Vgat-ablated mice showed a decrease in the percentage of time spent in NREM sleep and an increase in wakefulness. This response was more evident during the earlier part of the light phase, when sleep pressure is highest, suggesting that it was due to a reduction of homeostatic sleep drive. In order to test the role of MnPO^Vgat neurons in sleep homeostasis, we performed 4 hours of SD and found that MnPO^Vgat-ablated mice were able to recover a portion of the lost sleep, but that their amount of sleep was still less than intact WT controls. In fact, after 4 hr of SD, the total NREM sleep in the next 19 hrs was only equal to the amount of sleep that WT mice accumulated during this period at baseline. These findings indicate that, while there are other homeostatic mechanisms functioning in MnPO^Vgat-ablated mice, they require a much higher degree of sleep pressure to produce the same amount of sleep as WT mice.

One limitation of this approach is that the DTA_Vgat mice usually have less NREM sleep in the early light phase, so that they lost about 125 min of NREM sleep during SD (52% of 240 min), while DTA_WT mice lost about 144 min of sleep (60% of 240 min) during the same period (Figure 2b). We found that DTA_WT_SD mice showed about 61 minutes NREM sleep recovery over 19 h (42.4% of the lost sleep), while DTA_Vgat_SD mice had about 47 minutes of NREM sleep recovery (37.6% of lost sleep) (Figure 3d). Although this difference was not statically significant, it does indicate that the reduced sleep after SD in the MnPO^Vgat-ablated mice was not due to having lost less sleep.

Another concern is that MnPO neurons are also involved in regulating body temperature, drinking, osmolality, and the cardiovascular system ^20,28,29. We previously showed that ablation of MnPO^Vgat neurons does not affect baseline body temperature or locomotor activity²⁹, and now found that MnPO^Vgat neurons are not involved with the regulation of sleep behavior after a psychogenic stressor (Figure 5d). However, measuring other physiological variables and the extent to which they influence sleep regulation, needs to be addressed in future studies.

Role of MnPO^Vglut2 neurons in stress-induced sleep loss.

We also investigated the role of the excitatory MnPO^Vglut2 neurons in regulating the wake-sleep cycle. We and others had previously shown that activation of MnPO^Vglut2 neurons causes deep hypothermia, which itself affects sleep, so we did not repeat that approach here^23–25,33. However, ablation of MnPO^Vglut2 neurons does not alter baseline thermoregulation, so we chose that approach. We found that ablation of MnPO^Vglut2 neurons did not affect the wake-sleep cycle during baseline conditions. However, after 4 hours of SD, MnPO^Vglut2-ablated mice showed a smaller amount of recovery sleep during the remainder of the light phase with a greater variance in their sleep rebound compared to control SD mice, such that the increase in NREM sleep during sleep recovery did not reach statistical significance when compared to their baseline sleep, despite the fact that they had increased delta power during NREM when they did sleep. However, over the entire 19 hrs after SD, the sleep recovery of the MnPO^Vglut2-ablated mice was identical to that of the WT mice. Thus, although the MnPO^Vglut2 neurons do not appear to play a role in sleep homeostasis, we hypothesized that the difficulty in sleep recovery after ablating the MnPO^Vglut2 neurons might be due to their role in modulating stress response to the SD procedure. This possibility was reinforced by our previous observation that mice in which the MnPO^Vglut2 neurons had been deleted had greater hyperthermia and increased locomotion during stress induced by moving them to the cage that had been previously occupied by another male mouse²⁹. Consistent with this hypothesis, ablation of MnPO^Vglut2 neurons exacerbated insomnia produced by cage-exchange stress causing an increase in time spent in wake and reduction of NREM sleep when compared to non-ablated controls. This evidence suggests that the role of the MnPO^Vglut2 neurons may essentially be anxiolytic, limiting the effects of stress on sleep, thermoregulation, and locomotor activity.

Conclusions

Both MnPO^Vgat and MnPO^Vglut2 neurons project with about equal intensity to sites that modulate wake-sleep, including the VLPO, lateral hypothalamic area (LH), tuberomammillary nucleus, ventrolateral periaqueductal gray matter, parabrachial nucleus, and locus coeruleus ^20,21. Our data suggest that both sets of neurons play a role in sleep regulation, but that they are fundamentally different in nature. The MnPO^Vgat neurons appear to accumulate sleep need under baseline condition and to play a role in driving sleep after SD ^27,36–39. In their absence, animals have less baseline sleep, particularly in the early part of the light phase when sleep pressure is usually highest, and they also have proportionally less sleep after SD. Hence, while there clearly must be other factors contributing to longer term homeostatic sleep drive, in the absence of MnPO^Vgat neurons, it takes a higher degree of sleep pressure to produce a given amount of sleep. However, animals lacking MnPO^Vgat neurons have the same amount of sleep loss after stress, indicating that they do not play a role in the allostatic regulation of sleep.

MnPO^Vglut2 neurons, on the other hand, appear to play little role in sleep homeostasis. Our data suggest that instead they are involved in regulating the sleep response to stress. We have previously used the cage-exchange model to induce stress in rodents. Male rats and mice are territorial and attack intruders into their space. Thus, placing a male mouse in cage in which another male has lived for several days causes a stress response, including elevated body temperature and locomotor activity²⁹. We show here that cage exchange also causes insomnia in mice (as we previously showed in rats¹⁰), and that this insomnia is exacerbated by ablation of MnPO^Vglut2 neurons.

It is interesting that both MnPO^Vglut2 and MnPO^Vgat neurons directly innervate neurons in the VLPO^{20,21 21 40}. Because both types of MnPO neurons promote sleep, we hypothesize that the MnPO^Vglut2 neurons are likely to innervate sleep-promoting VLPO galanin neurons ⁴⁰ directly, while the MnPO^Vgat neurons probably innervate inhibitory interneurons in the VLPO thus promoting sleep by disinhibition. On the other hand, it has been proposed that the MnPO^Vgat neurons may induce sleep onset when sleep pressure is high by inhibiting orexin neurons ^27,36–39. It is also possible that the MnPO^Vglut2 neurons may inhibit arousal pathways by activating inhibitory interneurons. These hypotheses will have to be tested experimentally. Such studies will require genetically targeted approaches to elucidate the central mechanisms by which MnPO inhibitory neurons mediate sleep homeostatic drive and MnPO excitatory neurons modulate the effects of acute stress on sleep, autonomic and some behavioral responses.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests should be directed to and will be fulfilled by the Lead Contact, Clifford Saper (csaper@bidmc.harvard.edu)

Materials availability

This study did not generate new unique reagents.

Data and code availability

Data reported in this manuscript will be shared by the lead contact upon request. This study did not generate any unique datasets or code. Additional information required about the data reported in this study is available upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Animals

All animal care and experimental procedures were approved by the Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee. We used male Vgat-IRES-Cre mice, some of which were mated to Cre-dependent GFP reporter mice to generate Vgat-IRES-Cre;R26-loxSTOPlox-L10-GFP mice, and male Vglut2-IRES-Cre and Vglut2-IRES-Cre-L10-GFP (Vglut2-IRES-Cre;R26-loxSTOPlox-L10-GFP) male mice^32,41.

The age of mice at the time of experimentation ranged between 12–30 weeks. Mice were individually housed in standard plastic cages with standard corn cob bedding with nesting materials on a 12 h light: 12 h dark cycle (7am light - 7pm dark) at ambient temperatures ranging between 22 ± 2°C. Mouse chow (Teklad F6 Rodent Diet 8664) and water were provided ad libitum.

METHOD DETAILS

Surgery

All surgeries were performed under aseptic conditions. Mice were anesthetized with ketamine/xylazine (100 and 10 mg/kg, respectively, i.p.) with additional doses of 10% of the initial dose throughout surgery as needed to eliminate the withdrawal reflex. Stereotaxic microinjections of AAVs were made into the MnPO (coordinates: AP = +0.43 mm, L = 0.0, DV = −4.6 mm from Bregma) as previously described ⁴². Mice were implanted with four skull screws attached to a 6-pin connector for electroencephalographic (EEG) and two flexible electrodes for electromyographic (EMG) recordings. At the same time, mice were implanted with a radiotelemetry temperature and locomotor activity sensor (TA-F10, Data Sciences International, DSI) in the intra peritoneal space via laparotomy. Meloxicam treatment, for analgesia, was administered prior to surgery. Mice were allowed to recover at least 10 days prior to experimentation. Following recovery, mice showed no signs of discomfort and gained weight normally.

Viral Vectors

An AAV conditionally expressing subunit A of diphtheria toxin in a Cre-dependent manner (AAV-lox-mCherry-lox-DTA-lox2; or AAV-Flex-DTA) was used at a dose of 30–50 nl per injection. The AAVs were prepared by Dr. Patrick Fuller and packaged by Dr. Caroline Bass ^29,42.

AAV8-hSyn-DIO-hM3D(Gq)- mCherry (AAV-hM3Dq) produced at the University of North Carolina virus core was used in the volume of 30–50 nL and Clozapine-N-oxide (CNO) was used at the dose of 0.3mg/kg (sigma, catalog #34233-69-7), as a specific ligand for the hM3Dq receptor.

All experiments were conducted after appropriate time for the transfection of the AAVs (3–4 weeks).

Body temperature (core) temperature and locomotor activity recordings

Body temperature (Tb) and locomotor activity (LMA) was recorded using the radiotelemetry DSI system. The signal was sent from the telemetry probes previously implanted to receivers and converted using the PhysioTel HD and PhysioTel (DSI) hardware which provides the mean of Tb and total LMA counts every 5 minutes.

Sleep recordings and sleep analyses

Sleep recordings were done using a preamplifier connected to both the 6-pin connector implanted in the skull of all mice and to the data acquisition system (8200-K1-SE, Sirenia Software, from Pinnacle Technology). Digitized polygraphic data were analyzed off-line in 10 s bins using Sleep Sign software (Kissei). Using this software, we autoscored each epoch using an algorithm that identified three behavioral states (wake, NREM and REM sleep) based on EEG and EMG. Then, the auto-scored data were manually checked to confirm or correct automatic state classification as previously described ⁴². To perform an EEG power spectrum analysis, the scored recordings were processed using a Fast Fourier Transform (FFT) using Kissei software. The data were normalized relative to the total power considering the delta (0.5–4Hz), theta (4–8Hz) and alpha (8–13Hz) bands.

For the mice expressing AAV-DTA, the baseline was recorded between 4–6 weeks from the brain injections, and the experiments on sleep deprivation and stress were performed over the subsequent 2 weeks.

Sleep Deprivation (SD)

All mice were kept undisturbed in their home cages for at least 48 hours before the SD protocol. At light onset, mice were kept awake for 4 hours (ZT0-ZT3) by providing new nesting material, novel objects or gently tapping the cage or gently stroking the mouse with a cotton swab when necessary. During this period, EEG and EMG signals were monitored and behavioral responses were carefully observed to guarantee that mice were fully awake throughout the protocol. Mice were not disturbed when they were spontaneously awake, but at times when a mouse EEG signals displayed slow waves (indicating that an entry to NREM was imminent), a novel object was provided in the mouse’s cage. The first item provided was always nesting materials. Additional toys were designed to elicit exploring behavior and chewing ^30,31.

Cage exchange stress protocol

Singly housed mice were switched to an empty cage that was previously occupied for at least 7 days by another male mouse who was not a littermate (cage exchange stress). Stress protocols were performed at ZT4 and food and water were provided during the protocols ad libitum ^29,32.

Perfusion and brain sectioning

After the recordings, mice were deeply anesthetized with chloral hydrate (1.5% BW of a 7% solution) and transcardially perfused with 30ml phosphate buffer saline and then 30ml 10% pH neutral formalin (Fisher). Brains were extracted and post fixed overnight in 10% formalin and then stored in 20% sucrose until sectioning using a freezing microtome (40 μm coronal sections into 3 series). Following sectioning tissue was stored at 4°C in PBS containing the preservative sodium azide until processed for histology. For mice expressing AAV-hM3Dq, clozapine-N-oxide (CNO) was used at a dose of 0.3mg/kg (sigma, catalog #34233-69-7) as a specific ligand for hM3Dq or saline as control about 2h before the perfusions for cFOS analyses.

Immunohistochemistry

To confirm injection sites, brain tissue sections were processed free floating at room temperature. For cFos and mCherry labelling, the brain sections were rinsed and incubated in a blocking solution of 0.1M neutral phosphate buffer containing 10% horse serum, then rinsed and incubated in the same solution containing primary antibody overnight. mCherry protein was detected using a cross-reacting rabbit polyclonal anti-DSRed antibody (1:5,000, Clontech, catalog #14088015). This antibody did not stain anything in brains that had not been injected with an AAV expressing mCherry. cFos was detected using rabbit polyclonal antiserum against residues 4–17 from human cFos (AB5, 1:20,000; Oncogene Sciences, catalog #484). This antibody stained only the nuclei of cells in a pattern consistent with previous studies in control animals ⁴³. Sections were then rinsed and incubated with biotinylated secondary antibody for 120 minutes, and rinsed and incubated in ABC solution (1:500, Vector) for 60 minutes, followed by rinsing and incubation in a solution containing 1% DAB (Sigma) plus 0.02% H₂O₂. The sections were stained brown with DAB only or black by adding 0.05% cobalt chloride and 0.01% nickel ammonium sulfate to the DAB solution. For dual DAB staining, the sections were first stained with black DAB, then rinsed overnight and the process repeated with a different antibody using the brown DAB reaction. Sections were finally rinsed and mounted on glass slides.

QUANTIFICATION AND STATISTICAL ANALYSES

Quantitative analyses of histology

To determine the number of cFOS and mCherry-imunnoreactive neurons, a blinded observer counted only neurons with a clear nucleus in cells that were labeled with mCherry. An Abercrombie correction based on nuclear diameter was applied to all cell counts ⁴⁴. cFos stained only the nuclear compartment, while mCherry stained only the cytoplasm, which permitted dual labeling detection.

For mice injected with AAV-DTA, we used one series of the mice’s sectioned brains to map the injection sites. Mice that expressed native fluoresce expression of the AAV-lox-mCherry-lox-DTA-lox2 in the MnPO of Vgat-IRES-cre, Vgat-IRES-cre-L10-GFP, Vglut2-IRES-cre or Vglut2-IRES-cre-L10-GFP were included in the experimental groups.

Statistical analysis

Statistical analysis was performed using GraphPad Prism v7 (GraphPad Software). Significant differences were determined using a repeated measures (RM) one-way ANOVA or Two-Way ANOVA followed by Bonferroni corrected paired t-test or unpaired t-test; some of the t-tests were corrected for multiple comparisons using the Holm-Sidak method as indicated in the figure legends. For data not satisfying D’Agostino-Pearson or Shapiro-Wilk normality tests, we used Mann-Whitney or Friedman tests. One-way ANOVA was followed by Bonferroni’s correction for repeated measures when indicated. Data are presented as mean ± SEM. The statistical test used, statistical significance and number of animal subjects per group are reported in results, figure legends or tables, data were considered to be statistically significant when p<0.05 for single comparisons or at the indicated threshold for Bonferroni corrections. All animals used in our experiments were euthanized after completion of the study, the viral expression and the injection sites were verified histologically as criteria for assignment of animals in experimental groups. These decisions were made blinded to the physiological results for the animals. Analysis of the physiology was done blinded to the anatomical results of the animals. We performed power calculations using www.biomath.info. For these analyses, we used the means and SDs derived from our data and a significant level of power of 80%.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit anti-DSRED	Clontech	Cat#14088015
Rabbit polyclonal antiserum against residues 4–17 from human c-Fos, AB5	Oncogene Sciences	Cat#484
Donkey anti-rabbit biotinylated-IgG	Jackson ImmunnoResearch Laboratories	Cat#711065152
Bacterial and virus strains
AAV-hSyn-DIO-hM3D(Gq)-mCherry	UNC	N/A
AAV-lox-mCherry-lox-DTA-lox2	BIDMC/Harvard University, P Fuller29	N/A
Chemicals, peptides, and recombinant proteins
Clozapine-N-oxide	Sigma-Aldrich	Cat#34233-69-7
Experimental models: Organisms/strains
Mouse: Vglut2-IRES-cre	BIDMC, Harvard University, Dr. B Lowell41	N/A
Mouse: Vglut2-IRES-cre;R26-loxSTOPlox-L10-GFP	BIDMC, Harvard University, Dr. B Lowell41	N/A
Mouse: Vgat2-IRES-cre	BIDMC, Harvard University, Dr. B Lowell41	N/A
Mouse: Vgat-IRES-cre;R26-loxSTOPlox-L10-GFP	BIDMC, Harvard University, Dr. B Lowell41	N/A
Software and algorithms
GraphPad Prism v7	GraphPad Software	https://www.graphpad.com/scientific-software/prism/

Highlights.

• Activation of MnPO GABA neurons drive NREM sleep during the dark (active) phase

• MnPO/Vgat-ablated mice show less time spent in NREM sleep during the light phase

• MnPO/Vgat-ablated mice need a higher degree of sleep pressure to recover NREM sleep

• Ablation of MnPO/Vglut2 neurons exacerbates stress-induced insomnia