POAGAD2→TMN neurons are most active during REMs.

(A) Left, schematic of fiber photometry with simultaneous EEG and EMG recordings. Mouse brain figure adapted from the Allen Reference Atlas - Mouse Brain (atlas.brain-map.org). Center left, fluorescence image of POA in a GAD2-Cre mouse injected with AAVretro-FLEX-jGCaMP8s into the TMN. Scale bar, 1 mm. Center right, location of fiber tracts. Each colored bar represents the location of optic fibers for photometry recordings. Right, heatmaps outlining areas with cell bodies expressing GCaMP8. The green color code depicts how many mice the virus expression overlapped at the corresponding location (n = 10 mice).

(B) Example fiber photometry recording. Shown are EEG spectrogram, EMG amplitude, color-coded brain states, and ΔF/F signal.

(C) Non-normalized and z-scored ΔF/F activity during REMs, wake, and NREMs. Bars, averages across mice; lines, individual mice; error bars, ± s.e.m. One-way repeated measures (rm) ANOVA, p = 9e-4, 2e-6 for non-normalized ΔF/F and z-scored ΔF/F signals; pairwise t-tests with Bonferroni correction, non-normalized ΔF/F, p = 0.0056, 0.0039 for REMs vs. Wake and REMs vs. NREMs; z-scored ΔF/F, p = 6e-5, 1e-7. n = 10 mice.

(D) Average EEG spectrogram (top), z-scored ΔF/F activity (middle) and normalized EEG δ, θ and σ power (bottom) during NREMs→REMs transitions (left) and NREMs→Wake transitions (right). Shading, ± s.e.m. One-way rm ANOVA, p = 3.18e-49, 9.50e-8 for NREMs→REMs and NREMs→Wake; pairwise t-tests with Holm-Bonferroni correction, NREMs→REMs p < 0.0419 between -40 s and 30 s, NREMs→Wake p < 0.0106 between -10 s and 10 s. Gray bar, period when ΔF/F activity was significantly different from baseline (-60 to -50 s). n = 10 mice.

(E) ΔF/F activity during NREMs. The duration of NREMs episodes was normalized in time, ranging from 0 to 100%. Shading, ± s.e.m. Pairwise t-tests with Holm-Bonferroni correction p < 8.14e-9 between 20 and 100. Gray bar, intervals where ΔF/F activity was significantly different from baseline (0 to 20%, the first time bin). n = 566 events.

(F) Left, linear filter mapping the normalized EEG spectrogram onto the POAGAD2→TMN neural activity. Time point 0 s corresponds to the predicted neural activity. Right, coefficients of the linear filter for δ, θ and σ power band. Shading, ± s.e.m.

See Table S1 for the actual p values.

Inhibiting POAGAD2→TMN neurons reduces REMs.

(A) Left, schematic of optogenetic inhibition experiments. Center left, fluorescence image of POA in a GAD2-CRE mouse injected with AAVretro-DIO-SwiChR++-eYFP into the TMN. Scale bar, 1 mm. Center right, location of optic fiber tracts. Each colored bar represents the location of an optic fiber. Right, experimental paradigm for laser stimulation (2 s step pulses at 60 s intervals) in SwiChR++ and eYFP-expressing mice.

(B) Example recording of a SwiChR++ (left) and eYFP mouse (right) with laser stimulation. Shown are EEG power spectra, EMG amplitude, and color-coded brain states.

(C) Percentage of time spent in REMs, NREMs and wakefulness with and without laser in SwiChR++ and eYFP mice. Mixed ANOVA, virus p = 0.0629, laser p = 0.0003, interaction p = 0.0064; t-tests with Bonferroni correction, SwiChR-laser vs. SwiChR-w/o laser p = 3.00e-5, SwiChR-laser vs. eYFP-laser p = 0.0058.

(D) Normalized EEG δ, θ, and σ power during REMs in SwiChR++ and eYFP mice with laser. Unpaired t-tests, SwiChR vs. eYFP p = 0.0432, 0.0099 for δ and σ.

(E) Percentage of time spent in REMs, NREMs, and wakefulness during post laser sessions.

(C) Bars, averages across mice; lines, individual mice; error bars, ± s.e.m.

(D, E) Bars, averages across mice; dots, individual mice; error bars, ± s.e.m. SwiChR++: n = 12 mice; eYFP: n = 9 mice

POAGAD2→TMN neurons exhibit an increased number of calcium transients during REMs restriction.

(A) Schematic of REMs restriction/rebound and photometry recording experiments. The brain state was continuously monitored; once a REMs episode was detected, we used a vibrating motor (ZT 1.5-5.5) or pulled a string (ZT 5.5-7.5) attached to the mouse head to terminate REMs. Fiber photometry recordings were performed during REMs restriction (ZT 6.5-7.5) and rebound (ZT 7.5-8.5).

(B) Top, example fiber photometry recording. Shown are EEG spectrogram, EMG amplitude, color-coded brain states, and ΔF/F signal. Bottom, brain states, ΔF/F signal (green), low-pass filtered ΔF/F signal (orange), detected peaks (red) and pulls (arrows) during a selected interval (dashed box) at an expanded timescale.

(C) Number of calcium peaks during all states (left), NREMs (middle) and REMs (right). Bars, averages across mice; lines, individual mice; error bars, ± s.e.m. n = 8 mice.

Total: two-way rm ANOVA, treatment (baseline vs. manipulation) p = 0.0031, time p = 0.0613, interaction p = 0.0189; t-tests with Bonferroni correction, REM restriction (restr.) vs. baseline (ZT 6.5-7.5) p = 0.0033, restr. vs. REM rebound (reb.) p = 0.0341, restr. vs. baseline (ZT 7.5-8.5) p = 0.0049.

NREMs: two-way rm ANOVA, treatment p = 0.0233, time p = 0.0363, interaction p = 0.003; t-tests with Bonferroni correction, restr. vs. baseline (ZT 6.5-7.5) p = 0.0057, restr. vs. reb. p = 0.0046, restr. vs. baseline (ZT 7.5-8.5) p = 0.0115.

(D) Left, average NREMs calcium peaks during REMs restriction and baseline recordings. Right, average amplitude of the NREMs calcium peaks. The amplitude was calculated by subtracting the ΔF/F value 10 s before the peak from its value at the peak. Unpaired t-tests, p = 6.30e-10. n = 295 and 196 peaks during restriction and baseline recordings.

(E) Left, average REMs calcium peaks during REMs restriction and baseline recordings. Right, average amplitude of the REMs calcium peaks. Unpaired t-tests, p = 0.0437. n = 44 and 42 peaks during restriction and baseline recordings.

(F) Left, average NREMs calcium peaks during REMs rebound and baseline recordings. Right, average amplitude of the NREMs calcium peaks. Unpaired t-tests, p = 0.0002. n = 180 and 196 peaks during rebound and baseline recordings.

(G) Left, average REMs calcium peaks during REMs rebound and baseline recordings. Right, average amplitude of the REMs calcium peaks. n = 84 and 36 peaks during rebound and baseline recordings.

(D-G) Bars, averages across trials; error bars, ± s.e.m; shadings, ± s.e.m.

Inhibition of POAGAD2→TMN neurons during REMs restriction attenuates the REMs rebound.

(A) Schematic of REMs restriction/rebound and optogenetic inhibition experiments. During closed-loop REMs restriction (ZT 1.5-7.5), a vibrating motor attached to the mouse head was used to terminate REMs. REMs was restricted for 6 h (ZT 1.5-7.5). During the last 3 h of restriction (ZT 4.5-7.5), laser stimulation (2 s step pulses at 60 s intervals) was applied in SwiChR++ and eYFP mice.

(B) Example sessions from a SwiChR++ (top) and eYFP mouse (bottom) during REMs restriction with laser stimulation (left) and rebound (right). Shown are EEG spectrogram, EMG amplitude, motor vibration events, laser and color-coded brain states.

(C) Percentage of time spent in REMs, NREMs and wakefulness during the last 3 h of REMs restriction with laser stimulation (ZT 4.5-7.5) in mice expressing SwiChR++ and eYFP. Unpaired t-tests, p = 0.026 for REMs amount.

(D) Frequency of REMs episodes during the last 3 h of REMs restriction with laser stimulation (ZT 4.5-7.5). Unpaired t-tests, p = 0.0821.

(E) Normalized EEG δ, θ and σ power during the last 3 h of REMs restriction with laser stimulation (ZT 4.5-7.5). Unpaired t-tests, p = 0.0091, 0.0332, and 0.038 for REMs δ, NREMs θ and σ power.

(F) Percentage of time spent in REMs, NREMs and wakefulness during REMs rebound (ZT 7.5-8.5) in SwiChR++ and eYFP mice. Unpaired t-tests, p = 0.0205 for REMs amount.

(G) Frequency of REMs episodes during REMs rebound (ZT 7.5-8.5).

(H) Normalized EEG δ, θ and σ power during REMs rebound (ZT 7.5-8.5).

Bars, averages across mice; dots, individual mice; error bars, ± s.e.m. SwiChR++: n = 9 mice; eYFP: n = 7 mice