5-MeO-DMT may cause mixed waking and sleeping states in mice

Following the intraperitoneal (i.p.) injection of 5 mg/kg 5-MeO-DMT in freely behaving mice, we observed the emergence of clear-cut slow oscillations on the cortical EEG (Fig. 1a, b). While brain signals were characteristic of NREM sleep, no other sleep-related oscillations such as spindles or sharp wave ripples were observed, and the animals were unequivocally awake, as confirmed by direct observations and video recording (Supplementary Movie 1) (Fig. 1.b).

Plotting the EEG spectrograms following vehicle and 5-MeO-DMT revealed that marked effects of the compound on cortical EEG activity lasted less than an hour in all cases (representative example: Fig. 1.c). Specifically, we observed that while the animals were awake and moving after the injection of 5-MeO-DMT, theta-frequency activity was replaced by slow frequencies for approximately 45 min before returning to levels comparable to vehicle (Fig. 1c). Further quantitative analyses determined that in the frontal derivation, the injection of 5-MeO-DMT resulted in a significant increase in EEG slow wave activity (SWA, 0.5–4 Hz) and EEG spectral power in the 15–20 Hz frequency range (Fig. 1d). In the occipital derivation, there was a significant increase in SWA and a significant decrease in theta frequency power (Fig. 1e). No difference was observed in the spindle frequency range, consistent with the absence of visually detected spindle events. As non-oscillatory scale-free activity is considered an additional reliable marker of brain state^34,35, we compared the spectral slope calculated for the 20–120 Hz frequency range between the conditions. We found that in both derivations, the slope after 5-MeO-DMT treatment was comparable to normal wakefulness (Supplementary Table 1). We also observed that EEG power in faster frequencies ( > 30 Hz) was comparable to the awake state, despite the concomitant occurrence of slow waves, further suggesting that the state induced by 5-MeO-DMT has features of both waking and NREM sleep (Supplementary Fig. 1.a-d).

The differences in wake EEG power between 5-MeO-DMT and vehicle were observed primarily within one hour after the injection (Supplementary Fig.). To further interrogate the effects of the compound on cortical activity, we calculated the modulation index across frequencies. No clear-cut changes were found in the coupling of slow waves and high frequencies, although the modulation index for theta/gamma coupling in the occipital derivation was slightly decreased after 5-MeO-DMT (Supplementary Fig.), consistent with the strong attenuation of theta oscillations. 1e-f 2

Both NREM and REM sleep were initially suppressed after 5-MeO DMT (Supplementary Fig.), and the animals were behaviourally awake during the first 1 h interval after injection. While no major changes were observed in the subsequent sleep architecture and sleep stability (Supplementary Fig.), 48 h monitoring in a subset of animals revealed an increase in cumulative time spent in REM sleep over the first 24 h, with a further modest build up over the following 24 h period (Supplementary Fig.), resulting in overcompensation of REM sleep initially lost after the drug administration. 3.a-c 3.d-j 3.k-n

In all animals, behaviour after 5-MeO-DMT administration remained largely normal, with no evident locomotor deficits. Specifically, we never observed hyperactivity, flat body posture, catalepsy, hind limb abduction, or body tremors, and instead animals were engaged in exploratory behaviour, grooming, and nesting (Supplementary Movie, 01:31; 03:23). It is important to note that the injection of 5-MeO-DMT did induce head twitches, as characteristic of psychedelics (Supplementary Movie, 01:53) (Supplementary Fig.) and produced persistent exploration of the bedding immediately under the animal using their forelimbs (Supplementary Movie). To further characterise provoked behaviour and the motivational components of wakefulness, as well as to rule out behavioural abnormalities, we presented the animals with a plastic cup of sugar pellets and assessed their performance on a running wheel. 5-MeO-DMT increased the latency to cup interaction and pellet eating (Supplementary Fig.; Supplementary Movie), and shifted predominant behaviour to grooming or exploration. This might indicate a reduction in appetitively motivated behaviours, characteristic of the acute psychedelic state. We did not find any major change in the spontaneous wheel running behaviour that the control animals readily exhibited (Supplementary Fig.; Supplementary Movie), although, as the sample size in this experiment was low, no strong conclusions can be drawn at this stage. 1 1 4.a 1 4.b-c 2 4.d-g 3

As coexistence of active behaviours and global sleep-like brain signals is unusual, we next sought to rule out the possible occurrence of abnormal or artefactual brain signals, unrelated to physiological slow-wave activity³⁶. To this end, we carefully inspected the local field potentials (LFP) and corresponding multi-unit activity (MUA) in the primary visual cortex (V1) recorded in a subset of animals. It is well established that slow waves during physiological sleep are accompanied by the occurrence of OFF-periods–generalised periods of synchronised neuronal silence, typically lasting 100-200 ms, when the recorded populations of neurons do not generate action potentials^37,38.

As expected, in the vehicle condition, LFPs and MUA showed typical signatures of both wakefulness and NREM sleep, with the latter characterised by frequent OFF-periods during LFP slow waves (Fig. 2.a-c, Supplementary Movie 4). 5-MeO-DMT injection also resulted in the occurrence of prominent LFP slow waves during wakefulness, which were invariably accompanied by neuronal OFF-periods, normally rare during wakefulness (Fig. 2d, e; Supplementary Movie 4). Plotting the distribution of slow waves as a function of their amplitude and duration revealed a marked redistribution towards higher amplitude values during wake after 5-MeO-DMT treatment as compared to the vehicle condition (Fig. 2.f), with no difference found in wave duration between NREM sleep, wake with vehicle and wake with 5-MeO-DMT (Fig. 2.g). However, as observed on the EEG, these slow waves coexisted with faster frequencies, consistent with the awake rather than the NREM sleep state.

Increased SWA both in waking and sleep is typical for elevated homeostatic sleep pressure³⁹. Therefore, next we addressed the effects of sleep deprivation on 5-MeO-DMT-generated slow waves. We hypothesised that if slow waves after 5-MeO-DMT were reflecting increased sleep drive, they would be enhanced when the animals are experiencing high sleep pressure. To this end, we kept a group of animals awake for four hours, followed by an injection of the same dose of 5-MeO DMT as used before^32,39–41.

As expected, we observed that NREM sleep SWA after sleep deprivation was significantly increased compared to baseline in the vehicle condition (Fig. 3.a). Interestingly, elevated sleep pressure did not result in a further increase of SWA in awake animals following 5-MeO-DMT (Fig. 3b, c), suggesting that slow-waves induced by the drug correspond to an altered brain state rather than reflecting increased homeostatic sleep drive. However, unlike injections at light onset, 5-MeO-DMT administration at ZT4, following undisturbed sleep, resulted in significantly lower NREM SWA (Fig. 3d). Moreover, when administered immediately after sleep deprivation, 5-MeO-DMT markedly attenuated SWA rebound during subsequent NREM sleep (Fig. 3e), suggesting an effect on the expression of sleep homeostasis.

The occurrence of slow waves has been linked to decreased levels of attention and alertness^42,43. To measure the effects of 5-MeO-DMT on global arousal, we developed a device we called an oculometer. This was based on a miniature camera previously used in birds⁴⁴, that was fitted to the head-stage in order to track the state of the pupil in freely behaving mice. Previous studies in mice showed that movement or exploration - behaviours typically associated with increased levels of arousal are consistently accompanied by pupil dilation^45–47. In our study, unrestrained mice wearing the oculometer exhibited normal spontaneous behaviours, including exploration, grooming, nesting, and sleeping (Fig. 4a; Supplementary Movie 5), while the quality of video recordings from the eye was sufficient for reliable tracking of the pupil and subsequent image analysis with DeepLabCut (Fig. 4b; Supplementary Movie 6).

Firstly, we replicated the effects of the injection of 5-MeO-DMT in animals wearing the oculometer (Supplementary Fig. 5), although the suppressing effects of the drug on theta-activity outlasted the effects on SWA in this cohort (Fig. 4c-e). The injection of 5-MeO-DMT led to a marked increase ( + 75%) in pupil size lasting for 10–15 minutes (Fig. 4f; Supplementary Movie 7), which correlated with the increase in frontal SWA and decrease in occipital theta-power (Fig. 4g-i). The latter was unexpected, as pupil size and elevated levels of arousal during wake behaviours have traditionally been associated with elevated hippocampal and cortical theta activity^48,49.

To begin addressing the origin of the changes in cortical and autonomic arousal induced by systemic administration of 5-MeO-DMT, we next performed intracortical infusions of the compound. With this experiment, we tested the hypothesis that the changes in cortical neural activity were driven by local 5-HT-mediated signalling vs global changes in neuromodulatory tone. The animals were implanted with cannulae, and 500 nL of 5-MeO-DMT (4.3 mg/mL) was infused in the primary somatosensory cortex (Supplementary Fig. 6), as has been done in an earlier study with a dose of 5-HT_2A/2c agonist 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane (DOI) that induced head twitches⁵⁰. Contrary to those earlier findings, we did not observe any marked behavioural or EEG alterations, although a minor increase in EEG SWA was observed after the infusion (Supplementary Fig. 6). Further studies are necessary to clarify whether the prominent effects of systemic 5-MeO-DMT on brain state and arousal require the involvement of global, brain-wide circuitry, or depend on direct actions in other cortical or subcortical areas beyond the primary somatosensory cortex.

Unlike many serotonergic psychedelics, which preferentially target 5-HT_2A receptors, 5-MeO-DMT is a non-specific 5-HT receptor agonist with a 100-fold higher affinity for 5-HT_1A versus 5-HT_2A receptors^51,52. To investigate the role of 5-HT_1A receptor-mediated neurotransmission in the effects of 5-MeO DMT on brain activity, a separate group of animals was injected with 1 mg/kg i.p. of the 5-HT_1A receptor antagonist WAY-100635 (WAY) 15 minutes before the injection of 5-MeO-DMT. We observed that the animals injected with 5-MeO-DMT + WAY showed less activity and were often found staying immobile in one spot, exhibiting strong head twitch behaviour (Supplementary Movie 6). Furthermore, we observed that wake EEG SWA was further increased in the 5-MeO-DMT + WAY condition compared to both 5-MeO-DMT alone and vehicle conditions in both frontal (Fig. 5a) and occipital derivations (Fig. 5b), and this increase lasted longer frontally than in the 5-MeO-DMT alone condition (Fig. 5c), but not in the occipital derivation (Fig. 5d). Interestingly, the occipital theta activity, which was strongly suppressed after 5-MeO-DMT alone, was no longer different from the vehicle condition after 5-MeO-DMT + WAY (Fig. 5e), as was also the case for pupil diameter, whose increase was abolished by the 5-HT_1A antagonist (Fig. 5f). Thus, these data dissociate 5-MeO-DMT-induced changes in frontal SWA activity and pupil diameter, suggesting that only the latter effect is mediated by 5-HT_1A receptors (Fig. 5g). These results indicate that different 5-HT receptor subtypes have unique and distinct contributions to different characteristics of the altered state of vigilance induced by 5-MeO DMT.

Adult male C57BL/6 J mice (n = 42, 9 to 15 weeks old) were kept singly housed in individual Plexiglas cages (20.3 cm × 32 cm × 35 cm) placed inside ventilated sound-attenuated Faraday chambers (Campden Instruments, Loughborough, UK), under a 12–12 h light-dark cycle (9 am–9 pm). The recording room was maintained at 22 ± 1 °C and 50 ± 20% humidity. Food and water were provided ad libitum throughout the experiment. A subset of animals was not implanted and were used to monitor head twitches (n = 4), to study the effects of injection on running wheel activity (n = 4), or to study feeding behaviour (n = 4). Sample sizes were determined based on previous publications¹⁰³. All animals were monitored daily for unexpected weight loss or visible suffering as a marker of a humane endpoint. We have complied with all relevant ethical regulations for animal use. All procedures were performed under a UK Home Office Project License and conformed to the Animals (Scientific Procedures) Act 1986.

Procedures were performed based on established protocols for device implantation in mice^103,107. Prior to surgeries, mice (n = 30) were habituated to mash and jelly food and housed in individually ventilated cages. Surgeries were performed under isoflurane anaesthesia (4% induction, 1–2% maintenance). EEG screws were implanted above the right frontal cortex (2 mm anteroposterior, 2 mm mediolateral), right occipital cortex (anteroposterior 3.5 mm, mediolateral 2.5 mm), and left cerebellum for reference (Supplementary Fig. 7a).

In a subset of animals (n = 10), laminar probes (A1x16-3 mm-100-703-Z16, NeuroNexus) were implanted in addition to EEG electrodes as above, in the primary visual cortex (-3.4 mm anteroposterior, -2 mediolateral) and referenced to the cerebellum screw (Supplementary Fig. 7b).

In another subset of animals (n = 4), cannulae (C315G-005T GDE/ELE 26GA/.005” TUNG 8637 RECEPTACLE, Bilaney Consultants Ltd) affixed with two tungsten electrodes (4.8 mm cut below pedestal; 4.7 mm and 5 mm electrode Projection) were implanted unilaterally in the primary somatosensory cortex (-1.6 mm anteroposterior; +3.3 mm mediolateral; −0.5 mm dorsoventral from brain surface, with a 30° angle) (Supplementary Fig. 7.c). The location of the cannulae was then visually assessed by an experienced anatomist with cresyl violet coloration (Supplementary Fig. 8). EMG wires were inserted into the left and right nuchal muscles. Dental acrylic (Super Bond, Prestige Dental, Bradford, UK) was used to fix the implanted electrodes to the skull and to protect the exposed wires (Simplex Rapid, Kemdent, Swindon, UK).

A group of animals (n = 12) were also implanted frontally with a stainless-steel hexagonal nut (M2, DIN 934, RS Components) holder for the oculometer, a subset of which were also implanted with EEG and EMG electrodes (n = 6). Analgesics were administered immediately before surgery (5 mg/kg Metacam and 0.1 mg/kg Vetergesic, subcutaneous) and for at least three days following surgery (0.07 mL Metacam, oral, mixed in 3 mL of jelly). Mice were kept in individual, ventilated cages and monitored at least twice a day until baseline levels of well-being were scored for three consecutive days. They were then moved to their home Plexiglass cages for a week of habituation.

To monitor levels of arousal, we built a device called an oculometer consisting of an aluminium holder with a miniature camera directed at the eye (NE2D B\&W V90F2.7\_2m, NanEye, Austria) and a far-red LED (Luxeon Rebel LED 720 nm LXML-PF-1, Luxeon Star Leds, Canada) (Supplementary Fig.), maintained together with Sugru mouldable silicone rubber (Black, Sugru, Germany). The LED was powered through a 700 mA BuckPuck DC driver (3023-D-E-700, Luxeon Star Leds) and connected to the BioAmp Processor (RZ 2, Tucker-Davis Technologies) with an AV 9-way cable (6126423, RS Components) for activation through user input. The camera was connected to a PC via a wired board (985-NanoUSB 2.2, Mouser UK) and the video was recorded with the NanEye Viewer software (NanEye, v6.0.3, default configuration) for NanEye 2D and NanoUSB2 boards at a frame rate between 45 frames per second (fps) and 55 fps. Exposure, gain, and offset were automatically calibrated through the software to minimise loss of data due to poor lighting conditions, which can sometimes occur during some behaviours such as digging or nest building. The device was secured to the head of the mouse with a slot cheese nylon machine screw (M2 x 6 mm, DIN 84, RS Components). 7.d

To test the oculometer, the first recorded images were performed under anaesthesia (n = 3). The videos of freely moving mice (n = 12) were recorded following one 1-hour session of habituation to the device during which all mice were fitted with their respective oculometer, and the arms of the oculometer angled to record the most optimal view of the eye. During recording days, all animals wore the device at least 10 minutes before the start of experiments. The recordings started at light onset and were followed by the injections of either vehicle or 5-MeO-DMT as described below. As the device is quite cumbersome and could impact normal behaviour of the animals, it was decided to only record 60 min of video per subject per session.

To investigate the effects of psychedelics not only on vigilance states but also on behaviour and arousal, we used 5-MeO-DMT supplied by Beckley Psytech (BPL-003, Beckley Psytech, UK). 5-MeO-DMT was selected for its strong, fast-acting properties, inducing dream-like experiences in humans and behavioural changes in rodents within 5 min of administration. Following an i.p injection in mice, 5-MeO-DMT has a maximum concentration in plasma after 5 min and a terminal half-life of 12–19 min⁵⁵. On the day of the injection, 1 mg of 5-MeO-DMT was diluted into 1 mL of 0.9% sterile saline, and syringes containing either enough for an injection of 5 mg/kg 5-MeO-DMT or the vehicle (5 mL/kg sterile 0.9% saline) were prepared. The dose of 5-MeO-DMT was selected based on previous work reporting increased head twitch responses at this dose^51,57.

After one day of baseline electrophysiological recording, the mice were subjected to a counterbalanced, crossover design whereby they received a first intraperitoneal injection at light onset (Zeitgeber time (ZT) = 0) with either 5-MeO-DMT or vehicle and were then left undisturbed unless mentioned otherwise (Supplementary Fig. 7.e). Mice were randomly allocated to a specific experimental group. 72 h following the first injections, the mice received an intraperitoneal injection of the substance they did not originally receive (Supplementary Fig. 7.e). The majority of injected animals (n = 26) followed the above injection protocol. A similar version with the injections at a different ZT was performed for another group of animals (n = 16).

To investigate the contribution of 5-HT_1A receptors, a group of mice (n = 10) were injected with 1 mg/kg i.p. 5-HT_1A receptor antagonist WAY-100635, following previously established protocols⁵⁷.

The mice implanted with cannulae (n = 4) received an intracortical injection of 5-MeO-DMT with a total volume of 500nL, at the rate of 50nL per minute. The cannulae were connected to Hamilton syringes (Microliter Syringe 700 series, 22 gauge 80621, Hamilton Company) with a connector cannula (0/SPR 40 cm, Bilaney Consultants Ltd) and thin PE50 clear vinyl tubing (C232CT, Bilaney Consultants Ltd). The infusion rate of the Hamilton syringe was controlled by an automated pump (Harvard Apparatus Pump Controller, Harvard Apparatus), connected to the infuser (Nanomite Injector, Harvard Apparatus). As this method was novel, a concentration of 4.3 mg/mL was used based on previous work in rats using DOI and comparing the ED50 of DOI to the ED50 of 5-MeO-DMT in rats (Willins & Meltzer, 1997).

To assess the effects of the compound on exploratory behaviour and feeding, animals (n = 4) received a bowl of 12 sugar pellets (Sucrose Tab/Fruit Punch 14MG, 5TUT - 1811324, TestDiet). On day 0, the bowl was introduced without prior habituation 10 min after dark onset (ZT12). On day 1, the animals received either a vehicle (saline) or a 5 mg/kg 5-MeO-DMT (1 mg/ml saline) injection at dark onset (ZT = 12) at random, followed by two days of recovery, then the other injection on day 4. The bowl of sugar pellets was placed on the cage floor 10 min after the injections. Video recording was acquired for one hour following the injection and analysed offline. All latencies are expressed from the time the bowl is deposited in the environment and the hand of the experimenter has left the environment.

To investigate the effects of 5-MeO-DMT on behaviour, animals (n = 4) were given access to running wheels as an assay for locomotor activity. Installation of the setup and habituation was performed based on established protocols¹⁰⁸. In a crossover design, the animals received either vehicle (saline) or a 5 mg/kg 5-MeO-DMT (1 mg/ml saline) injection at dark onset (ZT12) at random on day 1, followed by two days of recovery, then the other injection on day 4. The animals were placed in the running wheels 10 minutes following each injection and were left otherwise undisturbed to assess spontaneous activity. Video recording was acquired for one hour following the injection and analysed offline.

To investigate the interaction between sleep-wake history and the effects of 5-MeO-DMT, mice (n = 8) were subjected to sleep-deprivation from ZT0 to ZT4 (Supplementary Fig. 7.f). At light onset, the chambers were opened and nesting material removed from the home cages. Removing the nesting material introduces changes to the environment, which kept the animals awake for around 30 min, during which the animals explored their environment or exhibited nest-building behaviours using the bedding of the cage. Repeated nest-building behaviour was immediately followed by a manual introduction of novel objects in the cage. Novel objects included small wooden blocks, folded gloves, and various toys, which were added to the cage or replaced as required. This procedure is known to increase homeostatic sleep pressure and subsequent EEG slow-wave activity during subsequent sleep and relies on the natural exploratory behaviour of mice^103,109,110. The mice were continuously monitored by the experimenters, and novel objects were introduced into the environment as soon as the animal was immobile or as slow waves became visible on the EEG.

Electrophysiological signals were acquired using a multichannel neurophysiology recording system (Tucker-Davis Technologies Inc., Florida, USA). Signals were acquired and processed online using the software package Synapse (Tucker-Davis Technologies Inc., Florida, USA). All signals were amplified (PZ5 NeuroDigitizer preamplifier, Tucker-Davis Technologies Inc., Florida, USA), filtered online (0.1–128 Hz) and stored with a sampling rate of 305 Hz. Signals were read into MATLAB (MathWorks) and filtered with a zero-phase Chebyshev type filter (0.5–120 Hz for EEG, 10–45 Hz for EMG), then resampled at 256 Hz. Exclusion criteria were based on a derivation basis after visual estimation of the signal quality. All analyses were done blind to the treatment.

Vigilance states were manually scored using the signals in European Data Format using SleepSign (Kissei Comtec, Nagano, Japan). The signals were divided into epochs of 4 s. Each epoch was scored based on visual inspections of frontal, occipital, and EMG signals (Supplementary Fig.). If recorded, LFP signals were also used. Epochs with recording artefacts due to gross movements, chewing, or external electrostatic noise were assigned to the respective vigilance state but not included in the electrophysiological analysis. Wake was scored in epochs containing high EMG activity. NREM sleep was scored in epochs containing low EMG and visual slow activity (0.5 Hz to 4 Hz) of high amplitude in the frontal or in both frontal and occipital derivations. REM sleep was scored when the occipital signal was mostly defined by theta activity (7 Hz to 12.5 Hz) with low EMG. Epochs containing mixed frequencies of lower amplitude and high EMG between two sleep epochs (NREM–NREM or REM–NREM) were scored as micro-arousals, unless 5 or more consecutive similar epochs were scored, in which case it was classified as wake. After sleep scoring, the power spectra of the signal were computed using a fast Fourier transform routine (Hanning window) with a 0.25 Hz resolution and exported in the frequency range between 0 and 120 Hz for spectral analysis. 7.g

Fast Fourier transform signals were stored into .mat files using a custom MATLAB script containing the spectrum of each epoch and the associated vigilance state. The spectral extraction script also corrected the epoch mistakenly scored as micro-arousal when they should have been scored as wake, and vice versa.

For the analysis of pupil diameter video recordings were first converted into AVI format and analysed using DeepLabCut (Mathis et al., 2018; Nath et al., 2019). The model trained for scoring the pupil diameter used the following parameters: resnet = 50, Shuffle 6; 5,000,000 trials. Four points were used to track the pupils: the northernmost, the southernmost, the easternmost and westernmost points (Fig. 3.b). Each point was marked only if it was clearly in view and not hidden by the eyelid or any light artefact. Once scored, the DeepLabCut model was trained on a randomly generated sample of images. The trained model then automatically scored all remaining videos, and only points labelled by the model with a confidence above 99% were kept. Pupil diameter was estimated as the average of the distance between the Northern and Southern point and the distance between the Eastern and Western point.

The time course of vigilance states is always analysed in equivalent time intervals between the drug and vehicle conditions. This enables comparison to previous studies and allows us to take into account the relatively fast kinetics of the substance, which mostly affects vigilance states in the initial intervals, naturally resulting in different amounts of time spent in a specific state early on after drug administration. However, we always ensured that, for spectral analyses, appropriate time intervals were chosen to ensure that all animals contributed to the comparisons with a representative number of artefact-free epochs.

The aperiodic analysis was performed on fast Fourier transform (FFT)–derived EEG power spectra for NREM sleep, baseline wake, and wake following an injection. Spectral slopes were calculated in linear–log space by fitting a straight line to the spectral power between 20 and 120 Hz, in order to minimise the influence of state-specific oscillatory activities prominent during waking and sleep at lower frequencies, such as sigma (10–15 Hz), theta (6–9 Hz), and delta (0.5–4 Hz). To assess changes in cross-frequency coupling, we calculated the modulation index for phase frequencies between 1.5 Hz and 10.5 Hz and amplitude frequencies between 0.25 Hz and 100.5 Hz using previously established MATLAB toolboxes¹¹¹.To investigate the incidence of local field potential (LFP) or EEG slow waves and corresponding neuronal activity, LFP signals in n = 4 mice were analysed as previously³⁸. Briefly, the LFP signal was first bandpass filtered between 0.5–4 Hz (stopband edge frequencies 0.2–8 Hz) with MATLAB filtfilt function exploiting a Chebyshev Type II filter design, and waves were detected as positive deflections of the filtered LFP signal between two consecutive negative deflections below the zero-crossing. Only LFP waves with a peak amplitude larger than the mean plus two standards deviations of the amplitude across all detected waves were included in MUA analyses, while all waves were utilised for plotting the distribution of amplitudes and durations. For MUA analysis, slow waves were aligned to their positive peak, and the corresponding average profile of neuronal spiking was computed (Supplementary Fig. 7.h-i).

Statistical analyses were computed using Prism (GraphPad Software, Boston, Massachusetts USA, version 10.2.0 (392)). All data sets were tested for normality with the Shapiro-Wilk test, and sphericity tested with Spearman’s rank correlation test. If all assumptions were met, a parametric test was conducted; otherwise, its non-parametric equivalent was selected for α = 0.05. Each figure will detail which tests were used. On figures, a significant difference is plotted as a straight black or coloured line in spectrogram analyses, as plotting the difference for each data bin would make the figure less readable. For all other figures, significance levels are indicated with black asterisks: * for p < [0.05; 0.01], ** for p < [0.01; 0.001], *** for p < [0.001; 0.0001], **** for p < 0.0001. Figures were generated using Prism and show group means with SEM for parametric data, unless mentioned otherwise in the figure legend, such as for plotting representative raw signals of one animal. All boxplots show the median, interquartile ranges, and the minimum and maximum values.

Further information on research design is available in thelinked to this article. Nature Portfolio Reporting Summary

Communications Biology thanks Francesca Siclari, Yuer Wu and the other anonymous reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Joao Valente.

All data presented in this manuscript are available on figshare (10.6084/m9.figshare.30058069). All raw data are available upon reasonable request to the corresponding authors.

All codes are available upon reasonable request to the corresponding authors.