Orexin neurons inhibit sleep to promote arousal

Abstract

Humans and animals lacking orexin neurons exhibit daytime sleepiness, sleep attacks, and state instability. While the circuit basis by which orexin neurons contribute to consolidated wakefulness remains unclear, existing models posit that orexin neurons provide their wake-stabilizing influence by exerting excitatory tone on other brain arousal nodes. Here we show using in vivo optogenetics, in vitro optogenetic-based circuit mapping, and single-cell transcriptomics that orexin neurons also contribute to arousal maintenance through indirect inhibition of sleep-promoting neurons of the ventrolateral preoptic nucleus. Activation of this subcortical circuit rapidly drives wakefulness from sleep by differentially modulating the activity of ventrolateral preoptic neurons. We further identify and characterize a feedforward circuit through which orexin (and co-released glutamate) acts to indirectly target and inhibit sleep-promoting ventrolateral preoptic neurons to produce arousal. This revealed circuitry provides an alternate framework for understanding how orexin neurons contribute to the maintenance of consolidated wakefulness and stabilize behavioral state.

Fig. 1. Optogenetic stimulation of orexin terminals in the VLPO rapidly triggers arousals from both NREM and REM sleep.

a Orexin innervation of the POA and VLPO (scale bars: 1 and 0.5  mm). b AAV-DIO-ChR2-mCherry was injected bilaterally into the Ox field of Ox-IRES-Cre mice and optical fibers bilaterally placed above the VLPO. c Orexin neurons transduced with ChR2- mCherry (red; left) and optical fiber (*) above the VLPO (right; scale bars: 500  µm). Optical fiber placements (red * ChR2-mCherry and grey * WT; bottom). d ChR2-eYFP(+) neurons double-labeled for Ox-A (in red; scale bars: 200; 100, and 20  µm). Laser-induced arousals from NREM (e) and REM (f) sleep (10  s-long periods before and after 10  s light pulses, top). EMG and EEG recordings and continuous EEG wavelet transforms (bottom). Trial-averaged responses to laser stimulation in NREM (g) and REM (h) sleep. EEG wavelet power for trials containing arousals (top; n  = 5). Black contour lines: 5% significance level against sham trials (paired bootstrap confidence interval of the mean wavelet power difference). EMG mean responses (grey region: 95% bootstrap confidence interval; bottom). Mean differences in arousal probability, from NREM (i) and REM sleep (j) in ChR2-mCherry (n  =  5) and WT (n  =  3; top). On the bottom, the mean differences in arousal probability (stimulation vs. sham) within the ChR2-mCherry group (NREM: effect of genotype F_{(1, 24)}  =  22.92, p  =  0.003; interaction between genotype and stimulation frequency F_{(4, 24)}  =  1.66, p  =  0.192, two-way ANOVA one-repeated-measure. REM: interaction between genotype and stimulation frequency F_{(2, 12)}  =  7.02, p  =  0.010, two-way ANOVA one-repeated-measure). Mean difference in arousal probability from NREM (i, bottom; t  =  2.98 and 3.73, p  =  0.019 and 0.004 for 10 and 20  Hz stimulations) and REM sleep (j, bottom; t =  3.82 and 5.16, p  =  0.002 and 0.0007 for 1 and 10  Hz stimulations; post-hoc Holm–Šidák t-tests, two-way ANOVA one-repeated-measure; n  =  5; Source Data file). Error bars: SEM; ‡: significant for factor genotype (two-way ANOVA); #, significant interaction between the factors genotype and stimulation frequency (two-way ANOVA). Mean differences (stimulations vs sham within the ChR2-mcherry group) plotted as bootstrap sampling distributions (bottom). Dots: mean difference in arousal probability; vertical error bars: 95% confidence intervals; *p  <  0.05 post-hoc Holm–Šidák t-test. All testing was two-tailed, and n refers to the number of independent animals. 3V 3^rd ventricle, Opt optic tract, f fornix, Ac anterior commissure, CPu corpus striatum, BF basal forebrain and LH lateral hypothalamus. Atlas levels are per Franklin and Paxinos, 2001.

Fig. 2. Dual response of orexin in VLPO.

a Whole-cell recordings of VLPO neurons and post-hoc labeling of a biocytin-filled neuron (scale bars: 500 and 20 µm). b Map of the recorded neurons in VLPO (n = 29; scale bar: 500 µm). c Most of VLPO neurons exhibit LTS (80 out of 116). d Ox-A (0.3-1 µM) excites 35% of the VLPO neurons: it increases action potential firing frequency (n = 21; one-way ANOVA, F_{(2, 60)} = 30.49; p < 0.0001; CTRL vs Ox-A and Ox-A vs. Wash, adj-p < 0.0001) and depolarizes the membrane potential (n = 26; one-way ANOVA, F_{(2, 75)} = 49.89; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, adj-p < 0.0001). e Ox-A inhibits 60% of the VLPO neurons: it reduces action potential firing frequency (n = 37; one-way ANOVA, F_{(2, 108)} = 28.56; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, adj-p < 0.0001) and hyperpolarizes membrane potential (n = 44; one-way ANOVA, F_{(2, 129)}=63.18; p < 0.0001; CTRL vs Ox-A and Ox-A vs Wash, adj-p < 0.0001). ****p < 0.0001, Bonferroni’s post-hoc test. Ac, anterior commissure; 3V, 3^rd ventricle; Opt, optical chiasm. Panel d and e: data are represented as means ± SEM, n refers to the number of recorded neurons and Source Data are provided as a Source Data file.

Fig. 3. Pharmacological and molecular diversity of the VLPO GABAergic neurons.

We recorded from TdTomato labeled VLPO Vgat(+) (VLPO^Vgat) neurons in Vgat-IRES-Cre mice injected into the VLPO with AAV-DIO-TdTomato. VLPO^Vgat neurons (a top; TdTomato-labelled; sample #a18; scale bar: 20 μm) that are inhibited by NA (50 μM) express galanin. Single-cell RT-PCR results from 4 VLPO neurons inhibited by NA (b). The VLPO^Vgat neurons (c top; TdTomato-labelled; sample #a30; scale bar: 20 μm) that are excited by NA do not express galanin. Single-cell RT-PCR results from 5 VLPO^Vgat neurons excited by NA (d). Gal (182 bp), Gad1 (177 bp), Gad2 (248 bp), and Gapdh (171 bp; housekeeping gene), M marker ladder, nc negative control. e scRT-sqPCR results confirming the presence of galanin mRNA in 7 VLPO^Vgat neurons inhibited by NA and/or Ox-A (NA/Ox-A(−)) and the absence of galanin mRNA in 8 VLPO^Vgat neurons excited by NA and/or Ox-A (NA/Ox-A(+)). Mann–Whitney unpaired one-sided t-test, Gal NA/Ox-A(−) vs Gal NA/Ox-A(+); p = 0.0002, ***p < 0.001; values normalized to the mean Gapdh. e data are represented as means ± SEM. Panels b, d, and e Source Data are provided as a Source Data file.

Fig. 4. Orexin enhances the VLPO^GABA→VLPO^GABA/Gal circuit.

a To study the VLPO^Vgat→VLPO^GABA/Gal input, we injected a mix of AAV-fDIO-ChR2-eYFP and AAV-DIO-TdTomato into the VLPO of Vgat-Flp::Gal-IRES-Cre mice (n = 10) and recorded from TdTomato-labeled VLPO^GABA/Gal neurons. To study the VLPO^GABA/Gal→VLPO^GABA/Gal input, we injected AAV-DIO-ChR2-mCherry into the VLPO of Gal-IRES-Cre mice (n = 2) and recorded from mCherry-labeled VLPO^GABA/Gal neurons. VLPO^GABA neurons labelled in green (white arrow) and VLPO^GABA/Gal neurons double-labeled in green and red (magenta arrow; scale bar: 20 µm). b Restricted transduction of AAV-fDIO-ChR2-eYFP in the VLPO of Vgat-Flp::Gal-IRES-Cre mice (YFP immunolabeling in green, scale bar: 500 μm). Injection sites from 12 mice used for in vitro CRACM recordings. c Raster plots of IPSCs in VLPO^GABA/Gal neurons with photostimulation of VLPO^Vgat→VLPO^GABA/Gal input (left) and VLPO^GABA/Gal→VLPO^GABA/Gal input (right; bin duration: 50 ms). d IPSC probability in response to photostimulation of the VLPO^Vgat→VLPO^GABA/Gal input (black; n = 8) and the VLPO^GABA/Gal→VLPO^GABA/Gal input (grey; n = 6; means ± SEM). Photostimulation of VLPO^Vgat neurons evokes GABA_A-mediated oIPSCs in VLPO^GABA/Gal neurons (e; BIC 20 µM). Mean oIPSC amplitude and latency (f; n = 8; in red: means ± SEM). g Opto-evoked IPSCs recorded in TTX (1 µM + 4-AP 25–50 µM, n = 43; in red: mean ± SEM). h Ox-A (1 µM) increases oIPSC amplitude (recordings in TTX + 4-AP). i Mean oIPSC amplitude in CTRL, Ox-A and Wash (n = 9, one-way ANOVA), F_{(2, 24)} = 19.42, p < 0.0001, CTRL vs Ox-A and Ox-A vs Wash adj-p-value = 0.0002; means ± SEM). j Peak-scaled non-stationary fluctuation analysis showing the current/variance relationship for oIPSCs (left; control: N = 84.2, i = 0.84 pA and Ox-A: N = 139.6, i = 0.82 pA). Ox-A increases the numbers of activated GABA_A channels (center; n = 8, CTRL vs Ox-A, p = 0.0091, paired one-sided t-test; **p < 0.01) without affecting the GABA_A unitary current (right; n = 8, CTRL vs Ox-A, p = 0.2052, paired one-sided t-test; means ± SEM). Recordings at reversal potential of the ChR2-mediated current (Vh = −5 to −15 mV). Blue-light pulses (10 ms; blue bars). ***p < 0.001 Bonferroni’s post-hoc test. In grey: 30 individual oIPSCs; in black: average oIPSC. Opt, optical chiasm; 3V, 3^rd ventricle; Atlas levels are per Franklin and Paxinos, 2001. For panels d–j: n refers to the number of recorded neurons. Panels c, d, f, g, and i, j: Source Data are provided as a Source Data file.

Fig. 5. Differential expression profiles and orexin receptor expression in VLPO^GABA/Gal and VLPO^GABA neurons.

Heat maps representing the expression levels of the top 30 differentially expressed genes ranked by adj-p (panel a; top 15 hypo- and top 15 hyper-expressed) and expression levels of GABAergic genes (Slc32a1, Gad1, Gad2), Hcrtr1, Hcrtr2, Gal and Slc17a6 (panel b) in the VLPO^GABA (#0–3; 7–9; 12, 13, 15, 17, 19) and VLPO^GABA/Gal (#5, 10, 14) clusters. Hcrtr2 positive neurons were largely restricted to VLPO^GABA cluster #1. c Dot plot representing the average expression of Hcrtr2 gene (x axis) in VLPO^GABA and VLPO^GABA/Gal groups (y axis). Dot size: percentage of neurons that expresses a specific gene (bottom right). Color intensity: expression level (top right). d Heat map of Gal and Hcrtr2 gene expression levels in VLPO^GABA cluster #1 (orange; 489 neurons), in all the other VLPO^GABA clusters (#0, 2, 3; 7–9; 12, 13, 15, 17, 19; green; down sampled to 515 neurons) and merged VLPO^GABA/Gal clusters (#5, 10, 14; cyan; 515 neurons). Violin plots for Hcrtr2 (bottom left) and Gal (bottom right) differential gene expression: VLPO^GABA cluster #1 (orange), all the other VLPO^GABA clusters (#0, 2, 3; 7–9; 12, 13, 15, 17, 19; green) and VLPO^GABA/Gal clusters (#5, 10, 14; cyan). Hcrtr2 is expressed in VLPO^GABA #1 (VLPO^GABA #1 vs VLPO^GABA/Gal adj-p value = 4.98E−32; VLPO^GABA #1 vs. VLPO^GABA (#0, 2, 3; 7–9; 12, 13, 15, 17, 19) adj-p value = 1.63E−114). Gal is expressed in VLPO^GABA/Gal neurons (VLPO^GABA #1 vs VLPO^GABA/Gal adj-p value = 1.26E−144; VLPO^GABA (#0, 2, 3; 7–9; 12, 13, 15, 17, 19) vs VLPO^GABA/Gal adj-p value = 0). Heat map and Dot plot expression values are represented as z-scores. Expression levels is color- coded in the legend. In the heat maps, x-axis: individual clusters color-coded bar; y-axis: genes. As color dots: Cluster IDs (right). Test used: Wilcoxon Rank Sum two-sided Bonferroni-corrected Test. e Expression in VLPO of Gal (red), Slc32a1 (green) and Hcrtr2 (pseudo colored in blue) mRNAs. White arrows: Slc32a1(+), Hcrtr2(+) and Gal(−) neurons; magenta arrows: Slc32a1(+), Hcrtr2(−) and Gal(−) neurons. Scale bars: 500 and 20 µm (bottom right). 3V, 3^rd ventricle. Source Data are provided as a Source Data file.

Fig. 6. Orexin input excites VLPO^GABA neurons by glutamate release (Ox → VLPO^GABA).

a AAV-DIO-ChR2-eYFP injected into the Ox field of Ox-IRES-Cre mice and recordings from VLPO neurons. b Injection sites of 15 Ox-IRES-Cre mice used for in vitro CRACM (n = 10 mice) and in vivo optogenetics (n = 5 mice). c Photostimulation of Ox → VLPO input evokes AMPA-mediated oEPSCs in VLPO neurons (DNQX 200 µM; in grey: 30 individual oEPSCs; in black: average oEPSC). d Raster plot of EPSCs in a VLPO neuron before, during, and after photostimulation of the Ox → VLPO input (50 ms bins). e Average EPSC probability of VLPO neurons that responded (n = 13; black) and did not respond (n = 48; grey; means ± SEM) to the photostimulation of Ox input. f No changes in holding currents in response to photostimulation trains (60 s at 10 and 20 Hz; 10 ms light pulses) in neurons activated by the Ox → VLPO input with glutamate-mediated oEPSCs. The VLPO^GABA neurons are activated by the Ox → VLPO input. The VLPO neurons that responded to photostimulation of the Ox → VLPO input (n = 7) are excited by NA (g) and/or do not express galanin mRNA (h; scRT-sqPCR). Specifically 4 neurons were tested for both the responses to NA and for the presence of Gal mRNA, 2 only for NA, and 1 only for Gal mRNA. Mean effects of NA on the action potential frequency (g left; paired one-sided t-test, CTRL vs NA, p = 0.0033; n = 4) and membrane potential (g, right; paired one-sided t-test, CTRL vs NA, p = 0.0017; n = 6). Results from scRT-sqPCR for Gapdh and galanin (Gal) from 5 VLPO neurons Ox → VLPO input connected (conn) are compared to 13 controls (ctrl) VLPO^GABA/Gal neurons (h; ctrl: sampled from recorded VLPO^GABA/Gal TdTomato-labeled neurons from Gal-IRES-Cre mice injected with AAV-DIO-TdTomato). Mann-Whitney unpaired one-sided t-test, Gal (conn) vs Gal (ctrl); p = 0.0001. Values are normalized to the mean Gapdh. Means ± SEM. i oEPSC amplitude (left) and latency (right) in 15 VLPO neurons (blue dots: identified VLPO^GABA neurons; in red: means ± SEM). Recordings at V_h= −70 mV; 10 ms blue-light pulses (blue bars). ***p < 0.001 and **p < 0.01. 3V 3^rd ventricle, Opt optical chiasm, f fornix, mt mammillothalamic tract, LH lateral hypothalamus. Atlas levels are per Franklin and Paxinos, 2001. For panels e and g: n refers to the number of recorded neurons. Panel d–e and g–i: Source Data file.

Fig. 7. Orexin neurons promote wakefulness by inhibiting sleep-promoting VLPO^GABA/Gal neurons via feed-forward inhibition.

Orexin neurons promote arousal through their projections to the VLPO. Orexin input to the VLPO directly activates VLPO GABAergic neurons that do not express galanin (VLPO^GABA) via co-release of Ox and glutamate. In turn, VLPO^GABA neurons inhibit VLPO^GABA/Gal sleep-promoting neurons to produce wakefulness.