A cluster of mesopontine GABAergic neurons suppresses REM sleep and curbs cataplexy

Abstract

Physiological rapid eye movement (REM) sleep termination is vital for initiating non-REM (NREM) sleep or arousal, whereas the suppression of excessive REM sleep is promising in treating narcolepsy. However, the neuronal mechanisms controlling REM sleep termination and keeping sleep continuation remain largely unknown. Here, we reveal a key brainstem region of GABAergic neurons in the control of both physiological REM sleep and cataplexy. Using fiber photometry and optic tetrode recording, we characterized the dorsal part of the deep mesencephalic nucleus (dDpMe) GABAergic neurons as REM relatively inactive and two different firing patterns under spontaneous sleep-wake cycles. Next, we investigated the roles of dDpMe GABAergic neuronal circuits in brain state regulation using optogenetics, RNA interference technology, and celltype-specific lesion. Physiologically, dDpMe GABAergic neurons causally suppressed REM sleep and promoted NREM sleep through the sublaterodorsal nucleus and lateral hypothalamus. In-depth studies of neural circuits revealed that sublaterodorsal nucleus glutamatergic neurons were essential for REM sleep termination by dDpMe GABAergic neurons. In addition, dDpMe GABAergic neurons efficiently suppressed cataplexy in a rodent model. Our results demonstrated that dDpMe GABAergic neurons controlled REM sleep termination along with REM/NREM transitions and represented a novel potential target to treat narcolepsy.

Fig. 1. Activity of dDpMe^GABA neurons across different brain state transitions.

a Schematic showing the fiber photometry used to assess GCaMP6f fluorescence of the dDpMe in GAD2-IRES-Cre mice with simultaneous polysomnographic recordings. Drawings of irregular-shaped superimposed viral expressing sites with circle-shaped micropipette tips (n = 5). b Raw traces of GCaMP6f fluorescence changes that were associated with different brain states. c Mean normalized fluorescence (z-score) during wake, NREM sleep, and REM sleep (n = 5 mice), taking the mean of 8 sessions per mouse; one-way ANOVA between brain states: F (1.324, 5.295) = 17.12, P = 0.0065; *P < 0.05, **P < 0.01, followed by Tukey’s multiple comparison test, P_(REM–Wake) = 0.022 and P _(REM–NREM) = 0.018. d Ca²⁺ normalized signals (z-score) associated with transitions among brain states. Top, individual transitions with color-coded fluorescence intensity; bottom, z-score of fluorescence changes from all transitions expressed as means (blue trace) ± SEM (shading). e Schematic showing the multichannel recording in vivo used to assess signal units by optic tetrodes with simultaneous polysomnographic recordings (n = 35). Typical fluorescent images representing the dDpMe labeled by ChR2-mCherry near the ChAT-labeled PPT (Green) and recording sites identified by electrolytic lesions stained by Nissl. f Example unit. Top, spike raster showing multiple trials of 470 nm laser stimulation. Bottom, raw trace showing spontaneous and laser-evoked spikes. Red, mean spontaneous firing rate of Type 1 neurons; yellow-green, mean spontaneous firing rate of Type 2 neurons; blue, laser-evoked spikes. g The firing rate of the neuron with the EEG spectrogram, EMG trace, and brain states (color coded). Freq, Frequency. There are two types of neurons among all identified GABAergic units in the dDpMe. Red, Type 1; yellow-green, Type 2. h The variation of identified dDpMe GABAergic neuronal firing rates at state transitions. To display the whole process of each phase, we picked up 30 s before and after each time point of state transitions, and then the firing rates averaged during REM sleep, wake, or NREM sleep were within each segments. Shading indicates ± SEM. i REM-NREM (R-N) vs REM-wakefulness (R-W) activity difference. j Mean firing rate of two-type dDpMe GABAergic neurons among brain states. Assessed by one-way ANOVA followed by Tukey’s multiple comparison test, *P < 0.05, **P < 0.01. Type 1: n = 22, P_(REM–Wake) = 2.0 × 10⁻⁴, P_(REM–NREM) = 9.2 × 10⁻³, and P _{(Wake–NREM)} = 2.6 × 10⁻³; Type 2: n = 13, P_(REM–Wake) = 0.028 and P _{(Wake–NREM)} = 2.6 × 10⁻³. dDpMe dorsal part of the deep mesencephalic nucleus, PAG periaqueductal gray, PPT pedunculopontine tegmental nucleus.

Fig. 2. dDpMe^GABA neurons are essential for REM sleep suppression and REM-to-NREM transitions.

a Schematic of optogenetic experiments with EEG/EMG recordings in a GAD2-IRES-Cre mouse with an AAV carrying Cre-dependent ChR2-mCherry injected into the dDpMe. Drawings of superimposed ChR2-mCherry-expressing sites with micropipette tips (n = 5, circles indicate tips and irregular shapes indicate expressing sites) and a typical fluorescent image representing the dDpMe labeled by ChR2-mCherry, dorsomedial to the ChAT-labeled PPT. b All trials (n = 172) and the probabilities of different brain states from five mice before, during, and after blue-light delivery (30 Hz, 120 s). c An example recording shows the EEG power spectrograms, EEG/EMG traces, and hypnograms after bilateral laser onset above the dDpMe in terms of REM sleep. Right, the power density of dDpMe^GABA ChR2 and dDpMe^GABA mCherry groups during different phases of laser stimulation from REM sleep (pre-stimulation, stimulation, and post-stimulation). d The probabilities of different brain states in GAD2-IRES-Cre mice after laser onset falling on REM sleep. The red lines indicate a statistical difference between the ChR2-mCherry and mCherry group (F_(1,637)(REM) = 301.10, P < 0.001; F_{(1,637)(NREM)} = 92.97, P < 0.001). e The pie charts and histograms show changes in brain states during laser stimulation initiated from different brain states (REM → NREM, P = 2.3 × 10⁻¹²; Wake→NREM, P = 7.9 × 10⁻⁹; ChR2-mCherry group, n = 5; mCherry group, n = 4). f Schematic of optogenetic experiments with EEG/EMG recordings of a GAD2-IRES-Cre mouse with Cre-dependent eNpHR-eYFP AAV injected into the dDpMe. Drawings of superimposed eYFP-expressing sites with the position of micropipette tips (n = 6) and typical fluorescent images in the dDpMe are shown, dorsomedial to the ChAT-labelled PPT (red). g All trials (n = 163) and the probabilities of different brain states from six mice before, during, and after constant 593-nm laser exposure. h Example recording of EEG power spectrograms, EEG/EMG traces, and hypnograms after bilateral laser stimulation above the dDpMe (593 nm, 60 s) in terms of NREM sleep. The power density of dDpMe^GABA eNpHR and dDpMe^GABA eYFP groups during different phases of laser onset falling on NREM sleep (pre-stimulation, stimulation, and post-stimulation). i The probabilities of different brain states in GAD2-IRES-Cre mice after laser onset falling on NREM sleep. The red lines indicate a significant difference between the eNpHR-eYFP or eYFP group (in NREM: F_(1,608)(REM) = 174.4, P < 0.001; F_{(1,608)(NREM)} = 130.2, P < 0.001). j Pie charts and histograms showing major states during laser stimulation initiated from different brain states, which were assessed by two-tailed t-tests: eNpHR-eYFP group, P_NREM→REM = 9.0 × 10⁻⁶; eNpHR-eYFP group, n = 6; eYFP group, n = 4. Data represent means ± SEM. *P < 0.05, **P < 0.001. The red or yellow horizontal lines in power density analysis indicate a statistical difference between pre-stimulation, stimulation, or post-stimulation (P < 0.05, red for pre-stimulation and stimulation; yellow for pre-stimulation and post-stimulation, two-way ANOVA with posthoc of the Sidak’s multiple comparisons test). DR dorsal raphe.

Fig. 3. Activation of dDpMe^GABA axonal terminals in the SLD/MPB or LH suppresses REM sleep and promotes REM-to-NREM transitions.

Schematic for optogenetic experiments and polygraphic recordings of the dDpMe-SLD/MPB (a) and dDpMe-LH pathway (f) with dDpMe neurons expressing ChR2-mCherry in GAD2-IRES-Cre mice. Example recordings of EEG power spectrograms, EEG/EMG traces, and hypnograms in GAD2-IRES-Cre mice after bilateral blue-light stimulation (bottom, 470 nm, 30 Hz, 5 ms, 120 s) or yellow-light stimulation (top, 593 nm, 30 Hz, 5 ms, 120 s) of dDpMe^GABA axonal terminals in the SLD/MPB (b) LH (g). Time courses of REM sleep (top) or NREM sleep (bottom) in GAD2-Cre mice after 593-nm or 470-nm laser stimulation of axonal terminals in the dDpMe or ChR2-mCherry-expressing neurons in the SLD/MPB (c), LH (h) initiated from REM sleep. The red lines indicate a significant difference (P < 0.05) compared to the control group, as assessed by two-way ANOVA (posthoc Sidak’s multiple comparison tests). F_(1,728)(REM) = 149.7, P < 0.001; F_{(1,728)(NREM)} = 123.8, P < 0.001; F_{(1,728)(wake)} = 0.69, P = 0.41 (c). F_(1,542)(REM) = 65.22, P < 0.001; F_{(1,542)(NREM)} = 10.98, P < 0.001; F_{(1,524)(wake)} = 25.43, P < 0.001(h). Pie charts and histograms showing major states during laser stimulation initiated from REM sleep (d) ChR2-dDpMe-SLD, n = 6; ChR2-dDpMe-LH (i), n = 4. Diagram showing the injection of AAV-ChR2 into the dDpMe of GAD2-Cre mice, and the response recorded in the SLD (e) or LH (j). Photo-stimulation (5 ms) evoked IPSCs in an SLD neuron, which was abolished by SR 95531 (a GABA_A antagonist). The latency of photo-stimulation-evoked IPSCs was counted from 12 responded SLD neurons in four mice (e) or 5 responded LH neurons in three mice (j). The mean latency of SLD neurons responding to the laser was 2.99 ± 0.31 ms (e) and that of LH neurons was 2.92 ± 0.54 ms (j). Number and proportion of recorded SLD or LH neurons showing a positive or negative response to the photo-stimulation of dDpMe neural terminals. Data represent means ± SEM. *P < 0.05, **P < 0.01. PnO oral part of pontine reticular nucleus, Mo5 motor trigeminal nucleus, ZI zona incerta.

Fig. 4. dDpMe^GABA→SLD^Glu neural pathway is necessary for REM sleep termination and REM-to-NREM transitions.

a, b Sagittal diagram for optical inhibition of the dDpMe by eNpHR after shRNA interference by shCtrl or shVglut2. shRNA interference of Vglut2 in the SLD was identified by in situ hybridization labelling Vglut2-mRNA. c, d Typical examples of EEG power spectrograms, EEG/EMG traces, and the mean percentage of power densities from GAD2-Cre mice with bilateral 593-nm laser delivery (bottom) or 470-nm laser delivery (top) to dDpMe eNpHR-expressing neurons. The horizontal bars indicate a significant difference (P < 0.01, two-way ANOVA with post-hoc Sidak’s multiple-comparison tests compared to 60-s pre-stimulation conditions; the red bars indicate pre-stimulation vs stimulation, and the orange bars represent pre-stimulation vs post-stimulation). Time courses of REM sleep (top) or NREM sleep (bottom) in GAD2-Cre mice with control or SLD^Glu interference after 593-nm or 470-nm laser stimulation of dDpMe eNpHR-eYFP-expressing neurons initiated from NREM sleep. The red lines indicate a significant difference (P < 0.05, two-way ANOVA with posthoc Sidak’s multiple comparison tests: shCtrl, F_(1,608)(REM) = 128.1, P < 0.001, F_{(1,608)(NREM)} = 232.5, P < 0.001; shVglut2, F_(1,608)(REM) = 0.56, P = 0.46, F_{(1,608)(NREM)} = 0.02, P = 0.89). e, f Pie chart and histograms showing major brain states during laser stimulation initiated from NREM sleep in the shCtrl group (e) or shVglut2 group (f), as assessed by two-way ANOVA with post-hoc Sidak’s multiple comparison tests: SLD-shCtrl group, F_(1,16) = 64.74, P_(NREM→REM) = 8.45 × 10^-9; SLD-shVglut2 group, F_(1,16) = 2.71, P_(NREM→REM) = 0.33. shVglut2 group_, n = 5; shCtrl group, n = 5. Data represent means ± SEM. *P < 0.05, **P < 0.01. PnC caudal part of pontine reticular nucleus.

Fig. 5. Activation of the dDpMe shuts down cataplexy attacks in mice with orexin neuronal lesions.

a Schematic of a celltype-specific lesion combined with optogenetic stimulation and poly-graphic recordings of the dDpMe in mice. Left-most, orexin-stained neurons in the LH are present in the control group but absent in the caspase-3 lesion group. Right-most, Images of dDpMe AAV-Vgat-ChR2-mCherry neurons co-labelled with GAD2 mRNA. b Example recordings of a wake-to-cataplexy transition, which contains EEG power spectrograms, EEG/EMG traces, and hypnograms. c The time, total amount, counts, and mean duration of cataplexy in wild-type mice between the control group and lesion group over a 12-h night period. Regarding the amount of time, significant differences were assessed by two-way ANOVA for time course (posthoc Sidak’s multiple comparisons tests): lesion group vs control group, F_{1, 72} = 104.7, P < 0.001. The difference in total amount of 12-h night time, counts, and mean duration of cataplexy between the two groups were assessed by Student’s t-test, P < 0.001. d Example recordings of EEG power spectrograms, EEG/EMG traces, and hypnograms in wild-type mice, with bilateral blue-light stimulation (bottom, 470 nm, 30 Hz, 5 ms, 120 s) or yellow-light stimulation (top, 593 nm, 30 Hz, 5 ms, 120 s) of dDpMe^GABA neurons. e Time courses of cataplexy, theta ratio, the probability of major stage, and EMG integral in wild-type mice after 2 min control (593-nm) or 470-nm laser stimulation of dDpMe ChR2-mCherry-expressing neurons initiated from cataplexy. The red lines indicate a statistical difference between the ChR2-mCherry and mCherry groups (P < 0.01). EMG integral assessed by Student’s t-test (P < 0.01). f EEG power spectrograms, EEG/EMG traces, and hypnograms of wild-type mice with orexin neuronal lesions following the application of 3-h light stimulation (30 Hz, 5 s per 30 s) in dDpMe^GABA ChR2 neurons. g Difference in the time and total amount of cataplexy in wild-type mice between the control group and lesion group during 3-h light stimulation. Regarding the time, significant differences were assessed by two-way ANOVA for time course (post-hoc Sidak’s multiple comparisons tests): *lesion group vs control group, F_{1, 72} = 7.117, P = 0.0094. Differences in the total amount of each brain state during 3-h light stimulation were assessed by Student’s t-test. P = 0.0036. Data represent means ± SEM. *P < 0.05, **P < 0.01. ns nigrostriatal bundle, f forn.