Circadian VIPergic Neurons of the Suprachiasmatic Nuclei Sculpt the Sleep-Wake Cycle

Abstract

Although the mammalian rest-activity cycle is controlled by a "master clock" in the suprachiasmatic nucleus (SCN) of the hypothalamus, it is unclear how firing of individual SCN neurons gates individual features of daily activity. Here, we demonstrate that a specific transcriptomically identified population of mouse VIP+ SCN neurons is active at the "wrong" time of day-nighttime-when most SCN neurons are silent. Using chemogenetic and optogenetic strategies, we show that these neurons and their cellular clocks are necessary and sufficient to gate and time nighttime sleep but have no effect upon daytime sleep. We propose that mouse nighttime sleep, analogous to the human siesta, is a "hard-wired" property gated by specific neurons of the master clock to favor subsequent alertness prior to dawn (a circadian "wake maintenance zone"). Thus, the SCN is not simply a 24-h metronome: specific populations sculpt critical features of the sleep-wake cycle.

Keywords: alertness; circadian; napping; optogenetics; siesta; sleep; vasoactive intestinal polypeptide; wake maintenance.

Figure 1

Identification of Night-Active SCN Neurons (A) Representative actogram of wild-type mouse RW activity in LD12:12. Note 2 peaks of RW activity separated by a period of quiescence and sleep, the siesta. (B) Average RW activity of 18 mice recorded for 7 days plotted in 30-min bins. Yellow bars indicate light; black bars indicate dark. (C) Left: representative immunostained SCN slices collected at indicated times (ZT, time relative to light onset). Top: DAPI (cell bodies). Bottom: c-FOS (marker of neuronal activity). Images were processed to remove background; only positive cells are shown. Inset: number of detected cells. Right: percentage of SCN DAPI+ cells co-expressing c-FOS (n = 3 animals per time point). (D) Left: unit activity detected across the day. Each row represents a unit active at any time point from one of 3 acute slices; totals are at bottom. Columns indicate time. Three trimmed, overlapping sets of measurements shown. Green, 0 Hz; Red, 3 Hz. Full data and quantification are in Figure S1. Right: Dorsal/ventral location and type of activity for each unit. Note that a higher proportion of ventral units show siesta-specific activity. (E) 40× maximum-projection images of SCN immunostained for c-FOS (green) and VIP (red) at ZT18. VIP+ axonal and dendritic staining is observed throughout the SCN, while VIP cell bodies are primarily ventral. Inset: arrow indicates neuron expressing VIP and c-FOS, Asterisk indicates neuron expressing only c-FOS. Pie chart: quantification of c-FOS and VIP cell body co-localization (red) versus c-FOS alone (gray); n = 6. (F) Patch clamp recordings of equal numbers of wild-type neurons recorded from nighttime and daytime slices. Blue, active neurons; red, inactive neurons. Axes indicate time, resting membrane potential, and firing rate. 1 h gap between daytime and nighttime units due to time of slice preparation. Further analysis is in Figure S2. Error bars represent SEM. ^∗p < 0.05, 2-tailed Student’s t test. See also Figures S1 and S2.

Figure 2

Characterization of Night-Active SCN Neurons by Patch-RNA-Seq (A) Ventral c-FOS::GFP+ neurons were recorded from ZT15–20 by patch clamp, then cytoplasm was collected for scRNA-seq (n = 20 neurons). (B) Heatmap showing expression of vip, avp, grp, and nms in neurons firing before and after ZT17. Color code shows Z-score normalized reads. Neurons firing after ZT17 express more vip compared to neurons firing before ZT17. (C) Pie chart of identified SCN neuronal classes, as defined by projecting each transcriptome to the SCN Drop-seq atlas (Wen et al., 2020). (D) Proportion of active (firing rate, >1 Hz) and silent neurons. (E) Electrophysiological properties of neurons from the two major classes: avp+ nms+ and vip+ nms+. Plots show firing frequency (in Hz; left) and resting membrane potential (RMP, in mV; right) over time. avp+ nms+ c-FOS::GFP+ neurons are mostly active in the early night (<ZT17) and vip+ nms+ c-FOS::GFP+ neurons from ~ZT17.

Figure 3

Neuronal Activity in VIP+SCN Neurons Regulates the Daily Siesta (A–C) RW activity of Vip-CRE and Avp-CRE mice injected at the SCN with AAV.Flex.TetLC. An average of 7 days of RW activity in 30 min bins is plotted for each genotype pre-injection (blue line), compared to 7 days of RW activity ≥2 weeks post-injection (green, orange, and red lines) when the virus should be fully expressed. Inset: ZT18.5–23.5 RW activity, 2-way ANOVA. (A) No effect of injection of TetLC virus into AVP-CRE mice on siesta activity, F(1, 168) = 2.776, n.s. (B) Injection of TetLC virus into VIP-CRE mice significantly increases RW activity during the siesta, F(1, 216) = 6.5, p < 0.05. (C) Quantification of the effect of VIP+SCN neurons upon siesta timing, defined as the point of lowest activity during the siesta. Comparisons by ANOVA with Tukey’s multiple comparison test for VIP>Tet: F(2, 35) = 5.892, p = 0.0062, otherwise Student’s t test. (D–F) RW activity of VIP-CRE mice injected at the SCN with the optogenetic probe AAV.ChETA. (D) RW activity of VIP>ChETA over 24 h before (baseline; blue line) or on day of stimulation (at 473 nm, which drives neurons to fire, via an optic fiber implanted between SCN lobes; red and blue bars). Stimulation for 1 h at 10 Hz at ZT14 (blue bars and blue arrow). Green arrow indicates new peak of RW activity that appears ~4 h post-stimulation (red bars, quantified in Figure S3). Inset: statistical summary before, during, and after stimulation; two-way ANOVA, F(1, 106) = 12.15, p < 0.005. (E) No effect of simulation for 1 h at 10 Hz at ZT14 on control-injected mice; F(1, 80) = 0.28, n.s. (F) Stimulation of VIP+SCN neurons for 1 h at 10Hz at ZT22 inhibits RW activity; F(1, 75) = 4.95, p < 0.05. In all panels, yellow: black bars represent 12-h:12-h LD cycles. Statistical comparisons are as specified. Error bars represent SEM. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.005. See also Figures S4 and S5.

Figure 4

VIP+SCN Neuronal Activity Regulates Nighttime Sleep (A) Total sleep as minutes per hour in 1 h bins in VIP>ChETA mice on the day of stimulation of VIP+SCN neurons at ZT14 (orange line) compared to baseline (blue line). Stimulation for 1 h at 10 Hz is indicated at blue bar. (B) Same experiment, but with control-injected mice. (C) Statistical summary before, during, and after stimulation. Left: VIP>ChETA; two-way ANOVA, F(1, 60) = 7.373, p < 0.01. Right: control; F(1, 66) = 0.4899. (D) The change in NREM sleep (NREM during stimulation − baseline NREM) is plotted for ZT14, ZT22, or ZT2 for control (black) and VIP>ChETA (blue). NREM is significantly increased in VIP>ChETA mice at ZT14. All stimulations of VIP>ChETA mice at ZT22 increased NREM sleep, with activation at 10 Hz at ZT22 reducing variability between mice. F test to compare variance: NREM sleep, F(10, 8) = 12.26, p < 0.01. There was no effect of VIP+SCN neuron activation on sleep at ZT2. (E–I) Amount of NREM (E) and REM (F) sleep in VIP>TetLC mice at baseline prior to injection and post-silencing, in 1-h bins. Blue, baseline (BL), prior to silencing; orange, silenced. REM sleep is specifically reduced during the siesta; quantified in Figure S6. (G) Length of NREM and REM sleep bouts in VIP>TetLC mice at baseline and post-silencing. (H) Number of NREM and REM sleep bouts for same as in (G). (I) Distribution of bout lengths for same. Error bars represent SEM. Statistical comparison by two-tailed Student’s t test unless otherwise stated. ^∗p < 0.05; ^∗∗p < 0.01. See also Figures S6 and S7.

Figure 5

VIP+SCN Activity Regulates RW Activity and Sleep Only at Night (A–D) Normalized RW activity in 30 min bins in VIP>eNpHR3.0 mice on the day of silencing of VIP+SCN neurons (orange bars) compared to baseline (blue line). VIP+SCN neurons were optogenetically silenced for 4 h in 2 min cycles of 1 min on, 1 min off (orange box). Silencing from (A) ZT0–4 (n = 5), (B) ZT8–12 (n = 8), and (C) ZT18–22 (n = 8). (D) Statistical summary showing a significant effect on RW activity only during silencing from ZT18–22. (E–H) Total sleep in 1 h bins in VIP>eNpHR3.0 mice on the day of silencing of VIP+SCN neurons for 4 h in 2 min on/off cycles (blue bars; 4 h silencing is indicated with orange bars) compared to baseline (blue line). (E) ZT0–4. (F) ZT8-12. (G) ZT18–22. (H) The difference in NREM and REM sleep for each individual VIP>eNpHR3.0 mouse between baseline and during 4 h silencing is shown for ZT18–22. Error bars represent SEM. Statistical comparison by two-tailed Student’s t test. ^∗p < 0.05; ^∗∗p < 0.01.

Figure 6

The Timing of the Siesta Is Regulated by the Molecular Clock RW activity was plotted in 30-min bins. Yellow:black bars represent 12 h:12 h LD cycles. Normalized RW activity is plotted in 30 min bins. (A) The tau allele of CKIe was used to shorten the period of VIP or AVP neurons. Top: AVP>CKIe^tau mice have a 22 h period in VIP neurons, 24 h in AVP neurons, and an advance in the timing of the siesta (orange arrow). Bottom: VIP>CKIe^tau mice with a 22 h period in AVP neurons but 24 h in VIP neurons have normal siesta timing (orange arrow). (B) VIP > Bmal^fl/fl mice lacking a functional clock in all VIP-CRE-expressing cells show increased RW activity during the siesta compared to sibling controls, F(1, 228) = 5.052, p < 0.05. (C) Quantification of the effect of VIP+SCN neurons upon siesta timing, defined as the point of lowest activity during the siesta. Comparisons by ANOVA with Tukey’s multiple comparison test for CKIε: F(2, 20) = 5.554, p = 0.0121 (Bmal^fl/fl by Student’s t test). Statistical comparisons as specified. Error bars represent SEM. ^∗p < 0.05; ^∗∗p < 0.01. See also Figure S3.

Figure 7

VIP+SCN Neuron-Driven Siesta Sleep Promotes Wake in the WMZ (A) Left: average RW activity of baseline (blue) and VIP>ChETA (red) is re-plotted from Figure 3D to show increased RW activity in mice ~4–5 h post-stimulation (green arrow). Blue box represents 1-h stimulation. Right: statistical analysis of same; RW activity is significantly increased from ZT18.5–19.5 in VIP>ChETA mice after induction of a “siesta” through stimulation at 10 Hz at ZT14, compared to baseline or control mice after same stimulation (one-way ANOVA). (B) RW activity of AVP>CK1e (purple) and VIP>CKIe (green) mice from Figure 6 re-plotted as normalized data to show the shift in timing of the second peak in AVP>CKIe mice. (C and D) Correlation between total amount of sleep in VIP>Tet mice at ZT18 and during the WMZ (ZT23–24), at baseline (C) and post-silencing (D). Note significant negative correlation at baseline, F(1, 8) = 6.87, p < 0.05; which is lost after silencing VIP+SCN neurons, F(1, 8) = 0.3928, n.s. Data are from Figure 3. (E) There is no significant correlation between siesta and WMZ sleep in VIP>eNpHR3.0 mice upon optogenetic silencing from ZT18–22, F(1, 14) = 4.58, n.s. Data are re-plotted from 14/18 mice in Figure 4 that showed reduced siesta sleep from ZT19–20 during silencing. (F) Correlation between total amount of sleep in VIP>ChETA mice at ZT14 at baseline (left) or during 10 Hz stimulation (right) with sleep at ZT18. Note negative correlation between total sleep at ZT14 and sleep at ZT18 post-stimulation (data are from Figure 5); F(1, 5) = 6.771, p < 0.05; no correlation without stimulation, F(1, 5) = 0.4305, n.s. (G) Model for the generation of siesta sleep and promotion of wakefulness in the wake maintenance zone (WMZ). VIP+SCN neurons signal at night to inhibit locomotor activity and/or directly promote sleep. As sleep accumulates during the siesta, consolidated by signals from VIP neurons, the amount of sleep required during the 2nd peak of RW activity (WMZ) is reduced. Thus, we propose that VIP+SCN neurons promote activity during the WMZ by increasing sleep during the siesta. Statistical comparisons are as specified. Error bars represent SEM. ^∗p < 0.05.