Psilocybin enhances sociability only in altered behavioural state and does not need an awake experience to be effective

Abstract

Psilocybin, a psychoactive compound contained in certain mushrooms, can provoke profound psychedelic experiences and also exerts established and controversial therapeutic efficacy in clinical depression and other neuropsychiatric conditions respectively. Here we report that a single systemic administration of the serotonergic psychedelic psilocybin can durably promote social behaviour in the Cntnap2-knockout mouse model of autism spectrum disorder (ASD), while no increase in sociability is observed in isogenic control mice. Addressing the concern that the therapeutic actions of psilocybin may involve potentially adverse psychedelic effects, we tested drug application in anesthetised mice and found that awake brain states are not required for psilocybin’s sociability enhancing actions in ASD mice.

Psilocybin shows therapeutic effects in the treatment of neuropsychiatric conditions but it can also cause strong hallucinations. Here, the authors show that psilocybin can improve social symptoms in a mouse model of autism, an effect that does not depend on wakefulness during the acute drug effects, suggesting that it might be possible to develop treatments that keep the benefits while avoiding the mind-altering side effects.

Introduction

Recent preclinical and clinical research on psychedelics such as psilocybin has uncovered their lasting influences on brain structure and function both in humans^1–5 and in mouse models^6–8 of neuropsychiatric disorders^9–11. Advanced clinical studies have already demonstrated psilocybin’s clear efficacy for treatment-resistant depression^1,3,12–15, while intense investigations are underway for several other neuropsychiatric conditions where psilocybin may exert therapeutic efficacy. These less established clinical indications include anxiety⁴, various forms of addiction¹⁶ and compulsive behaviours¹⁷ (extending to anorexia nervosa¹⁸), and reduced social cognition^19,20. We wondered whether psilocybin treatment may also ameliorate reduced sociability⁹, a form of social cognition and behavioural trait that has been well characterised in animal models such as mice as well as in humans^21,22. Sociability is a common characteristic of all types of autism under the classification of autism spectrum disorder (ASD), and reduced sociability is a core symptom replicated in several mouse models that mimic the genetic or environmental risk factors identified in ASD^23–25. Reduced sociability has also been suggested as a characteristic that may be associated with several other neuropsychiatric conditions beyond ASD, including, for instance, schizophrenia²⁶, major depressive disorder²⁷, and dementia²⁸. Here, our motivation for examining psilocybin’s effect on sociability is driven by sociability being a measurable behavioural trait reflecting an altered state of brain function where higher cognitive areas are involved, rather than to preclinically examine psilocybin’s effect on ASD (or to suggest that ASD should be treated) or other neuropsychiatric conditions. We therefore use an established mouse model of ASD here as a case of altered (social) behaviour to examine psilocybin’s effect on altered brain state.

It has been consistently reported that a single dose of psilocybin can intriguingly induce lasting influences on brain structure and function. For example, studies in humans revealed that a single dose of psilocybin can reduce depressive symptoms in treatment-resistant patients for several weeks, outlasting the acute psychedelic effect^3,15,29,30. In an apparent disagreement with the body of supportive findings of psilocybin’s lasting effects, a recent large-scale multicentre study reported no lasting effects of psilocybin in normal C57BL/6 mice³¹, contradicting several earlier reports^9,11,32,33. Similarly, mood- or mind-enhancing effects of psychedelics in human probands without neuropsychiatric diagnoses have been unclear. This ostensible controversy may be reconciled by the concept that psilocybin treatment re-adjusts neuronal mechanisms only in an suboptimal (disordered) state and accordingly provides no beneficial effects in the absence of a non-disordered “normal” state.

To address the question if psilocybin induces lasting effects on sociability, here we use the well-established Cntnap2 knockout ASD mouse model²⁵ as a model of altered social behaviour (and altered state) to examine the potential of a single systemic injection of psilocybin in restoring diminished social behaviour and whether such potential is durable. Our observations may also give indications to the possible scenario of beneficial effects only for altered states.

Psilocybin being a serotonergic psychedelic, the psychedelic experiences of humans treated with psilocybin have caused some concern, notably due to these experiences being challenging to manage³⁴ in clinical settings^35,36 and potentially can even cause adverse neuropsychiatric symptoms such as psychotic episodes^37–39. Psilocybin’s hallucinogenic effects communicated by human subjects are read out in rodents as the “head-twitch response”, which is an established proxy for hallucinogenic activity⁴⁰, and these effects are mediated by serotonin 2 A receptors (5HT2ARs)^40,41. To overcome the concern of possible adverse effects of the active psychedelic experience, potential solutions have been put forward, including both the development of non-hallucinogenic 5HT2AR agonist alternatives^42–44 and drug administration strategies such as microdosing^45,46 to mitigate the psychedelic experience. Nevertheless, the lasting effect of a drug, much beyond its bioavailability, points toward a mode of action outlasting the simple acute drug-receptor interactions. In clinical studies with patients, these more complex actions may involve subjective effects of psychedelics¹, such as mystical experiences⁵ or psychological interventions^3,15.

However, fundamentally, it is unclear whether an active, awake psychedelic experience is at all required for psilocybin’s lasting and potential therapeutic effects. Indeed, this has been a matter of intense debate in the psychedelic field^1,47,48. While we cannot straightforwardly read out psychedelic experiences in mice beyond the head-twitch response, to provide some experimental evidence for this ongoing debate, we can examine whether suppressing the awake subjective experience can still result in a behavioural adjustment effect that would have been achievable if psilocybin were given during the fully awake state. Therefore, we administered psilocybin to Cntnap2 knockout mice under light anaesthesia and maintained the mice under anaesthesia throughout the duration of psilocybin’s acute behavioural effect as per the awake condition. We found that wakefulness (and hence, an awake experience) at the time of psilocybin administration is not required for psilocybin’s induction of lasting sociability-enhancing effects in the Cntnap2 knockout mice.

Results

Psilocybin durably improves sociability in Cntnap2^-/- homozygous KO mice

We previously found that a single systemic injection of psilocybin can rescue altered sociability in the prenatal valproate exposure teratogenic mouse model of ASD at 1 day after treatment, beyond the phase of acute psychedelic effect⁹. To examine this effect further, we first asked whether a single systemic injection of psilocybin is able to improve the sociability of a genetic risk factor mouse model of ASD, and whether any improvement is durable. For this, we used the well-established genetic mouse model of the ASD risk factor gene CNTNAP2, a genetic mouse model with altered social behaviour that has been employed by many separate labs^25,49,50.

We administered a single systemic injection of 1 mg/kg psilocybin or saline to adult Cntnap2 homozygous (Cntnap2^-/- KO) mutant mice at 8 weeks of age. To test the durability of the potential sociability-enhancing effect, we assessed the social behaviour of these mice using the three-chamber social behaviour test at 1 day, and also at 1 week and 2 weeks after the single injection, corresponding to ages of postnatal week 8, 9, and 10 respectively (Fig. 1a, b). We quantified sociability as the amount of time a mouse approaches and explores a cup containing another mouse versus an empty cup, with body position near the cup and head directed towards the cup (Fig. 1b).

Fig. 1. Psilocybin durably increases sociability in Cntnap2^-/- homozygous KO mice.

a Schematic experiment timeline. b Example position maps from the different experimental conditions for the sociability testing. Dark shade dots: position locations across observation time where the test mouse displays interest in the cup or social partner. Light shade dots: positions where the test mouse did not display interest in cup or social partner. c A single injection of psilocybin ameliorates the social deficits in Cntnap2^-/- KO mice for up to two weeks. (N = 15 for Cntnap2^-/-_Psilocybin, 15 for Cntnap2^-/-_Vehicle). d Psilocybin’s sociability-rescuing effect is observed in Cntnap2^-/- KO mice tested for the first time at two weeks after a single treatment. (N = 10 for Cntnap2^-/-_Psilocybin, 9 for Cntnap2^-/-_Vehicle). Mean ± SEM shown. Individual data points shown.

First, we considered whether psilocybin treatment influenced social behaviour, combining sociability, total interaction time, and locomotion of the test mice during the testing session. We found a significant effect of psilocybin treatment on these measures (RMANOVA, F2/15 = 544.501, p < 0.001, effect size (partial eta squared) = 0.986) and a significant interaction between the measures and test session (F4/13 = 8.763, p = 0.001, effect size = 0.729), with no significant detectable effect of sex in our dataset (F1/16 = 0.347, p = 0.564). We should note that the effect of sex was not a focus of or a hypothesis in our study; as such, we simply tested for the presence of such an effect within our dataset and found none. Further, visualising our key dataset separately by sex (Supplementary Fig. 1) showed overlapping data spread between the two sexes. Therefore, we did not pursue additional data collection to further power the effect of sex, in view of the effect of sex not being central to our study.

Follow-up univariate analyses revealed that sociability was significantly affected by the psilocybin treatment (F1/16 = 9.144, p = 0.008, Bonferroni-corrected, effect size = 0.364, observed power = 0.810; sphericity not violated (Mauchly’s W = 0.961, p = 0.610)), while no significant difference was observed in our dataset for total interaction time (F1/16 = 0.104, p = 0.751, Bonferroni-corrected; sphericity not violated (Mauchly’s W = 0.839, p = 0.111)) or locomotion (F1/16 = 0.021, p = 0.887, Bonferroni-corrected; sphericity not violated (Mauchly’s W = 0.862, p = 0.328)). This indicated that psilocybin increased sociability in the Cntnap2^-/- KO mice, and that this effect was due to a specific improvement in social drive rather than a variability in the drive to explore. The significant increase in sociability of psilocybin-treated Cntnap2^-/- KO group compared to the vehicle-treated Cntnap2^-/- KO control group was sustained for at least two weeks following the single psilocybin administration (Fig. 1c; at 1 day: F1/26 = 6.826, p = 0.015, 95% CI [0.043 0.357]; at 1 week: F1/26 = 27.366, p < 0.001, 95% CI [0.273 0.625]; at 2 weeks: F1/26 = 4.217, p = 0.050, 95% CI [0.000 0.561]; all Bonferroni-corrected).

Further, to confirm that the sustained increase in sociability was not influenced by repeated testing, we assessed an independent cohort of Cntnap2^⁻/⁻ KO mice that was tested for the first time at two weeks after psilocybin or vehicle administration. Psilocybin-treated mice showed significantly higher sociability compared to vehicle-treated controls (Fig. 1d; T = 2.243, p = 0.019, T-test). This observation confirmed that psilocybin produces a lasting rescue of social behaviour independent of the testing schedule.

Psilocybin does not enhance social behaviour in isogenic control mice

A recent large-scale multicentre study reported no acute or lasting prosocial effects of psilocybin in mice³¹, while noting that this observation apparently contradicted our prior report of psilocybin’s sociability-rescuing effect in an established ASD mouse model for teratogenically-induced reduced sociability⁹. However, the large-scale multicentre report tested whether psilocybin affects the sociability of C57BL/6 mice that have normal social behaviour, not addressing the possible therapeutic potential in animal models of reduced sociability. We therefore asked if the robust sociability-increasing effect of psilocybin in the Cntnap2^-/- KO mice is also observed in wild-type control mice. To test this, we performed the same experiments described so far using isogenic Cntnap2^+/+ wild-type mice from the same colony (Fig. 2) as that of the Cntnap2^-/- KO mice.

Fig. 2. Psilocybin does not enhance sociability in Cntnap2^+/+ isogenic WT mice.

N = 7 for Cntnap2^+/+_Psilocybin, 11 for Cntnap2^+/+_Vehicle. Mean ± SEM shown. Individual data point shown.

In contrast to the effect of psilocybin in Cntnap2^-/- KO mice, the single administration of psilocybin did not enhance sociability in Cntnap2^+/+ wild-type mice (RMANOVA, F2/15 = 1.970, p = 0.174). We did not observe any significant statistical interaction (F = 1.790, p = 0.201), as such, further post-hoc analysis could not be reported by statistical convention. However, we note that, had we only tested for psilocybin’s effect at 1 day post-treatment instead of our repeated measure design, we would have detected a significant effect of psilocybin reducing sociability in Cntnap2^+/+ isogenic wild-type group at 1 day post-treatment (F1/16 = 22.118, p < 0.001, Bonferroni’s correction), similar to the trend we observed previously with teratogenic mouse model of ASD (the prenatal saline control group therein). We do not derive or conclude a negative effect of psilocybin on sociability in Cntnap2^+/+ WT mice from this observation, as this is not the main focus of the current study, but simply describe our observation for those who may be interested in investigating this further.

We considered whether psilocybin treatment of Cntnap2^-/- homozygous KO mice promoted social-related behaviours to the same level as that of the isogenic wild-type group. We observed no significant difference in sociability and total interaction time between Cntnap2^-/- mice that received psilocybin vs Cntnap2^+/+ mice that received saline vehicle (F1/22 = 1.096, p = 0.307), and no significant effect of sex was detected in our dataset (F1/22 = 0.926, p = 0.346). This indicated that the sociability of the psilocybin-treated Cntnap2^-/- group was restored to a level similar to that of the isogenic wild-type control group.

In line with the report that Cntnap2 is a recessive gene⁵¹, heterozygous Cntnap2^+/- mice did not show significantly altered sociability compared to the wild-type Cntnap2^+/+ group (RMANOVA, F1/17 = 0.678, p = 0.422). Cntnap2^+/- heterozygous mice displayed, like Cntnap2^+/+ mice, a tendency of reduced sociability after psilocybin treatment (Supplementary Fig. 2), even though not statistically significant (RMANOVA, F1/14 = 3.548, p = 0.081). These observations indicate that while the sociability-increasing effect of psilocybin has the potential to ameliorate abnormal behavioural traits, in the absence of behavioural deficits, psilocybin does not enhance and may even reduce sociability.

Our present observations corroborate our earlier report that psilocybin enhances sociability in mouse models that show sociability deficits, but in wild-type mice does not further enhance sociability (and may even have counter effects)⁹.

No detectable effects of psilocybin on two other ASD related phenotypes exhibited by Cntnap2^-/- KO mice

Then, we explored whether psilocybin also affected other aspects of behaviour that are not directly related to sociability but are possibly affected in the ASD model mice^25,52. For this, we examined repetitive behaviour and hyperactivity in our existing datasets, where psilocybin had been effective in enhancing sociability in the Cntnap2^-/- KO mice.

We examined repetitive behaviour by quantifying self-grooming activity during habituation phases (the final habituation prior to drug treatment and the habituation phase of post-treatment three-chamber tests prior to the introduction of a social partner). We observed that Cntnap2^-/- KO mice displayed a significantly higher number of self-grooming episodes compared to Cntnap2^+/+ WT mice during the final habituation session prior to drug treatments (Fig. 3a; U = 26, p = 0.005, Mann-Whitney). We then detected no significant difference in the number of self-grooming episodes displayed by Cntnap2^-/- KO mice at 1 day after psilocybin treatment (Fig. 3b, KW = 0.235, p = 0.628) or at 1 week after treatment (Fig. 3b, KW = 0.021, p = 0.886). We therefore concluded that repetitive behaviour, although increased in Cntnap2^-/- KO mice, was not affected by psilocybin under our treatment conditions.

Fig. 3. Psilocybin does not ameliorate two other ASD related phenotypes exhibited by Cntnap2^-/- KO mice.

a,b The significantly higher repetitive behaviour displayed by Cntnap2^-/- KO mice (a; N = 11 for Cntnap2^-/-, 14 for Cntnap2^+/+) was not altered by psilocybin treatment (b; N = 6 for Cntnap2^-/-_Psilocybin, 6 for Cntnap2^+/+_Vehicle). c,d The significant hyperactivity displayed by Cntnap2^-/- KO mice (c; N = 22 for Cntnap2^-/-, 15 for Cntnap2^+/+) was not altered by psilocybin treatment (d; N = 12 for Cntnap2^-/-_Psilocybin, 11 for Cntnap2^+/+_Vehicle). Mean ± SEM shown. Individual data point shown.

Further, we tracked the positions of the test mice during the habituation phases. In line with previous reports of hyperactivity in Cntnap2^-/- KO mice²⁵, we observed that Cntnap2^-/- KO mice displayed significantly higher distance travelled during the 10 min observation period compared to Cntnap2^+/+ isogenic wild-type mice (Fig. 3c, U = 52, p < 0.001, Mann-Whitney). Despite drug treatment, no significant difference was detected in the distance travelled by Cntnap2^-/- KO mice at 1 day after drug treatment (Fig. 3d, KW = 0.458, p = 0.498) or at 1 week after drug treatment (Fig. 3d, KW = 0.004, p = 0.951). We therefore concluded that psilocybin has no effect on spontaneous locomotor drive under our treatment conditions.

Psilocybin’s effect on sociability is mediated by 5HT2ARs

Psilocybin is known to act predominantly through the activation of serotonin 2A receptors (5HT2ARs)^53–55. To test whether our observed effects of psilocybin were indeed initiated by 5HT2AR activation, we next performed experiments using pretreatment with a specific 5HT2AR antagonist, volinanserin (MDL100907) at 15 min prior to treatment with psilocybin.

We observed that the 5HT2AR antagonist pretreatment significantly affected psilocybin’s effect on sociability (RMANOVA, effect of pretreatment: F1/34 = 7.400, p = 0.010, effect size = 0.179, observed power = 0.753; effect of treatment: F1/34 = 10.962, p = 0.002, effect size = 0.244, observed power = 0.895; pretreatment * treatment interaction: F1/34 = 14.751, p < 0.001, effect size = 0.303, observed power = 0.962). Post-hoc analysis showed that 5HT2AR antagonist pretreatment prevented the psilocybin-induced increase in sociability in Cntnap2^-/- KO mice (F1/34 = 21.524, p < 0.001, Bonferroni-corrected). MDL100907 pretreatment itself did not have any significant effect on sociability for Cntnap2^-/- KO mice that received vehicle instead of psilocybin (F1/34 = 0.628, p = 0.434, Bonferroni-corrected).

Further, we observed that the blocking effect of 5HT2AR antagonist pretreatment was effective at 1 day after treatment (Fig. 4a left; F1/17 = 15.985, p < 0.001, 95% CI [0.159 0.515], Bonferroni-corrected), and sustained at 1 week after treatment (Fig. 4a right; F1/17 = 7.197, p < 0.016, 95% CI [0.066 0.553], Bonferroni-corrected). This indicated that 5HT2AR activation is required for the sustained effect of psilocybin on sociability in Cntnap2^-/- KO mice.

Fig. 4. 5HT2AR antagonist pretreatment blocks the behavioural effects induced by psilocybin.

a Psilocybin’s sociability enhancing effect in Cntnap2^-/- KO mice was blocked by 5HT2AR antagonist pretreatment (N = 10 for Cntnap2^-/-_{Vehicle+Psilocybin}, 10 for Cntnap2^-/-_{Vehicle+Vehicle}, 9 for Cntnap2^-/-_{Ant-5HT2AR+Psilocybin}, 9 for Cntnap2^-/-_{Ant-5HT2AR+Vehicle}, 7 for Cntnap2^+/+ _{Vehicle+Vehicle}). b 5HT2AR antagonism blocked the head-twitch response, as expected (N = 10 for Cntnap2^-/-_{Vehicle+Psilocybin}, 10 for Cntnap2^-/-_{Vehicle+Vehicle}, 9 for Cntnap2^-/-_{Ant-5HT2AR+Psilocybin}, 9 for Cntnap2^-/-_{Ant-5HT2AR+Vehicle}). Mean ± SEM shown. Individual data point shown.

The acute effect of psilocybin does not depend on expression of the Cntnap2 gene

Psilocybin or its active metabolite psilocin, like other psychedelic 5HT2AR agonists, triggers an acute head twitch response (HTR)⁴⁰ that lasts for the duration of the agonist’s bioavailability⁵⁶. Immediately after psilocybin application, we assayed the HTR for a 20 min observation period to confirm the normal functioning of 5HT2ARs in psilocybin-treated mice (Fig. 4b).

Psilocybin significantly increased HTR in Cntnap2^-/- KO mice (F1/26 = 99.822, p < 0.001, Two-way ANOVA) with no significant effect of sex detected in our dataset (F1/26 = 2.482, p = 0.127, Two-way ANOVA). Variance was tested for and was homogeneous (Levene statistic = 0.752, p = 0.537). We also observed no significant difference in the level of psilocybin-induced HTRs between Cntnap2^-/- KO mice vs Cntnap2^+/+ isogenic WT mice (F1/18 = 0.352, p = 0.560, Two-way ANOVA), with no significant effect of sex detected in our dataset (F1/18 = 0.774, p = 0.391, Two-way ANOVA). 5HT2AR antagonist also effectively abolished the psilocybin-induced head-twitch response during the acute phase of psilocybin action (Fig. 4b; F1/15 = 144.126, p < 0.001, Two-way ANOVA; homogeneous variance: p = 0.219, Levene statistic = 2.061), confirming that the antagonist was fully effective. Even though sex-related differences in HTR have been reported for other psychedelic compounds and analogues^57,58, we did not detect any significant effect of sex on the HTR, which is similar to a lack of sex-specific effects on psilocybin-induced HTR as observed by others⁵⁹. Moreover, we observed no significant correlation between acute HTR and the subsequent sociability levels in the Cntnap2^-/- KO mice (At 1 day: rho = −0.397, p = 0.143; At 1 week: rho = 0.085, p = 0.763; At 2 weeks: rho = 0.084, p = 0.766; Spearman’s rho). This indicates that the reduced sociability in Cntnap2^-/- KO mice cannot be explained by an altered acute function of 5HT2ARs.

Sociability increasing effect of psilocybin in Cntnap2^-/- KO mice does not require wakefulness during treatment

While 5HT2AR activation has been established as being required for psychedelic experience of humans treated with psilocybin (and for the HTR as the proxy of psychedelic effect in mice), whether an awake active psychedelic experience—commonly known as a psychedelic “trip”—is itself necessary to attain the therapeutic benefits of psilocybin had long been debated but remained inconclusive^47,48. To address this important issue, we administered psilocybin to lightly anaesthetised mice to examine whether the sociability-rescuing effect of psilocybin is attainable absent an awake brain state.

We first lightly anaesthetised mice using isoflurane to achieve a stable breathing rate of 2–2.5 Hz commonly observed in mice under NREM and REM sleep states^60,61. Under this condition, mice showed no righting reflex, no active whisking, and no overt motor movement. After maintaining mice under this light anaesthesia steady state for 15 min, a single injection of psilocybin or vehicle was administered systemically via intraperitoneal injection. Mice were maintained for a further 2 h under steady-state light anaesthesia before being allowed to recover to an awake state. We observed no head-twitch response throughout the 2-hour light anaesthesia session.

We then tested the mice for social behaviour at 1 day, 1 week and 2 weeks after the treatment received under anaesthesia (Fig. 5). Psilocybin applied under anaesthesia also significantly increased sociability in the Cntnap2^-/- KO mice (RMANOVA, F1/10 = 10.629, p = 0.009, effect size = 0.515, observed power = 0.835; sphericity not violated (Mauchly’s W = 0.955, p = 0.812)). Also, in this dataset, we did not observe an effect of sex (F1/10 = 0.039, p = 0.847).

Fig. 5. Sociability enhancing effect of psilocybin in Cntnap2^-/- KO mice does not require wakefulness during treatment.

a Schematic of experiment timeline. b A single injection of psilocybin administered under isoflurane anaesthesia rescues the social deficits in Cntnap2^-/- KO mice (N = 8 for Cntnap2^-/-_Psilocybin, 6 for Cntnap2^+/+_Vehicle). Mean ± SEM shown. Individual data point shown.

Discussion

We demonstrated that psilocybin rescues social deficits in a mouse model of altered sociability in ASD, and this effect is sustained for at least two weeks and is mediated by 5HT2AR activation. Rather than generically increasing sociability, psilocybin tended to decrease sociability in isogenic control mice. This result directly addresses the recent conclusion of a multicentre study that did not find psilocybin exerting lasting effects of potential therapeutic benefits in wild-type C57BL/6 mice that did not have social deficits³¹. Based on our observations, both here and previously reported⁹, we propose that psilocybin treatment re-adjusts neuronal mechanisms only if out of balance (i.e., disordered state) and accordingly is not of therapeutic potential in the absence of a misbalance (i.e., non-disordered “normal” wild-type state). This is not a new concept raised only by us, but is also put forward by others. Indeed, experts of psychedelic science also suggested this during revision of this paper in a newly published review, that “the dramatic departure from a stable set point that is brought on by a psychedelic may cause brain activity and behaviour to become more plastic and, therefore, adaptable, even after the acute effects have resolved”¹. By logical deduction, this hypothesis implies that destabilising a setpoint that is already optimal will not drive an adaptation to a different setpoint.

Although the underlying molecular mechanism of this circuit re-adjustment is yet to be determined, one of psilocybin’s established acute effects in mice is to activate brain circuitry, and in particular those involving the prefrontal cortex^59,62. Stimulation of these circuits for a short time period (such as the half-life bioavailability following a single dose) may be comparable to other non-invasive stimulation techniques such as transcranial current stimulation, magnetic stimulation, and ultrasound stimulation, all of which have been shown to exert lasting effects in mouse and human brain functions. Acting through a defined serotonin receptor subtype, psilocybin may potentially mediate a more specific circuit activation than the abovementioned stimulation (activation) techniques.

The psychedelic state induced by 5HTR2AR agonists like psilocybin has a long history in the treatment of neuropsychiatric diseases and typically within a psychotherapeutic procedure⁶³. These therapeutic interventions are conceptualised, broadly speaking, as “mind-opening” effects of psychedelics. On the other hand, psychedelics such as lysergic acid diethylamide (LSD) and psilocybin have also been reported to exert an effect on the structure and function of neuronal circuits, including sprouting of dendrites and dendritic spines^6,59. In vivo, these effects are best documented in the frontal cortex, where 5HTR2ARs are prominently expressed. The psychological effects on perception and those on circuits differ in their timescale: a psychedelic trip lasts a few hours, while anatomical changes require more time to evolve. Our finding that psilocybin triggers long-lasting effects even when applied under slight anaesthesia separates these two modes of action further, although we cannot exclude the possibility that anaesthesia may influence the cellular mechanisms underlying psilocybin’s action, leading to its lasting effects even where similar therapeutic effects are observed behaviourally following psilocybin administration in awake versus anaesthetised states. Nevertheless, the therapeutic effects observed with psilocybin applied under anaesthesia are of potential translational significance because this implies that psychedelic experience and the associated potential risks of adverse effects triggered by them may not be required for psilocybin’s therapeutic potential.

Despite this promising therapeutic effect in socially deficient mice, we found that psilocybin does not improve sociability in mice that do not show behavioural deficits, such as the Cntnap2 isogenic wild-type or heterozygous groups here. Our results even indicate that psilocybin exerts opposing effects, reducing sociability in the absence of abnormal sociability traits. Thus, it appears that psilocybin is not a general sociability enhancer, but rather is only beneficial when there is a defect, a clinically relevant behavioural measure for brain disorders or mental health issues. When considering psilocybin administration as a pharmacological perturbation of neuronal circuit dynamics, its lasting effects suggest the existence of an adjustable setpoint that governs social behaviour. Considering this together with our observations, this experimental evidence collectively suggests that different setpoints (both at behavioural and at circuit dynamic levels) may exist for normal versus deficit states, which may be adjusted, at least experimentally. Exploration of the potential mechanism of such setpoint re-adjustment requires additional investigations.

Limitations of this study

This is a fundamental science study using a mouse model, which may inspire, but cannot replace, studies in humans. We used a well-established mouse model for ASD²⁵, but only as a model for altered sociability that was recapitulated in knockout mice carrying a genetic alteration that is an established ASD risk factor in humans⁵¹. We cannot and do not conclude on the potential therapeutic value of psilocybin in the treatment of humans diagnosed with ASD, nor do we suggest that such persons should necessarily be treated at all.

The 3-chamber test is often referred to as the “golden standard” for sociability testing in mice. But we acknowledge its limitations^64,65. For example, we do not know whether the test mice approaching either cup have an associated component of aggression (which is not a research question that we set out to examine). In the context of the present study, to our knowledge, neither knocking out the Cntnap2 gene nor psilocybin treatment is known to induce aggression; therefore, we considered that the main focus of our study is unlikely to be confounded by aggression.

Methods

Animals

Experimental procedures at Nanyang Technological University, Singapore, were performed in accordance with approved protocols following review by the Institutional Animal Care and Use Committee. Experimental procedures were performed at Imperial College London UK in accordance with the United Kingdom Animal Scientific Procedures Act (1986), under Home Office Personal and Project Licences following appropriate ethical review. Cntnap2^-/- homozygous KO²⁵ transgenic mice (JAX017482, Jackson Labs USA) were maintained on a C57BL/6 J background. Offsprings were weaned at P21 and housed in groups of up to five animals per cage after weaning. All animals were maintained in individually ventilated cages, on a 12/12 h light/dark cycle (light on between 07:00 and 19:00) at 21 °C ± 2 °C and 55 ± 10% humidity. Water and food were provided ad libitum. 8–10 weeks old mice of both genders were used for behavioural experiments. Mice were group-housed where possible, with the same littermate mice both before and after treatments. Mice in the same cage typically included treatments of different conditions and/or untreated companion mice. All experiments were conducted during the light phase.

Drug administration

Psilocybin (Usona Institute US) was dissolved in sterile saline solution (0.9% NaCl) to appropriate concentrations. 8-week old offspring were intraperitoneally injected either with saline or psilocybin (1 mg/kg) 24 hours prior to behavioural testing. We chose this dose of psilocybin based on our previous observations⁹ as well as observations from several other groups^32,59.

For mice treated with psilocybin (or vehicle) while awake, for the duration of the acute drug effect, each mouse was placed in a separate chamber to assess the head-twitch response.

For mice treated with psilocybin (or vehicle) while under anaesthesia, all mice were first lightly anaesthetised to reach a breathing rate of 2–2.5 Hz accompanied by a lack of active whisking and a loss of righting reflex, inside an anaesthesia induction chamber of 1% isoflurane carried by oxygen. All mice were maintained under this steady state for 15 min, then briefly removed to receive a single systemic injection and returned to the anaesthesia chamber within 5 s. All mice were then maintained in this steady state 1% isoflurane anaesthesia for 2 h.

For 5HT2A receptor antagonist pretreatment, we used the selective antagonist MDL100907 (Tocris, UK) to inhibit 5HT2A receptor activity in vivo. MDL was first dissolved in 10% DMSO then diluted using saline. One dose of 1 mg/kg MDL or vehicle was systemically administered via intraperitoneal injections per test animal at 15 min prior to psilocybin or saline treatment.

Head-twitch response

Head-twitch events were evaluated immediately after intraperitoneal administration of either saline or psilocybin for a period of 20 minutes. The mouse was placed into a customised behavioural box immediately after injection and was free to explore. The number of head-twitch events was counted by direct observation in 5-minute bins. The box is of size 40 ×20 X 20 cm [L x W x H] and the floor was covered with clean bedding. Experimenters were not blind to the treatment for head-twitch counts.

Behavioural testing

Social behaviour was tested in a custom-built Plexiglass three-chamber box with a matt white acrylic floor. The total arena (60 × 40 x 22 cm [L x W x H]) was divided in three smaller, evenly sized chambers (20 × 40 x 22 cm [L x W x H]) interconnected by 4 × 4 cm doors. Transparent cylinders (10.5 cm internal diameter, 11 cm external diameter, 16 cm length) with 1-cm slits were used as cups. The box was placed in a darkened and quite room, illuminated from above with infrared LEDs located 1 metre over the arena.

Age-matched, sex-matched non-littermate mice with no social behavioural deficits (i.e., no homozygous mutant KO mice) from the same genetic background were used as social partners.

Animals were acclimatised to the testing room for at least one hour before the start of experiment on each day. Behavioural testing took place over 4 consecutive days in the first week. Each test mouse was placed into the centre chamber and allowed to freely explore and habituate to the three-chamber setup for 10 minutes over 3 consecutive days; doors to the side chambers were kept open. Social partner mice were individually pre-habituated to the cups for 10 min over three consecutive days. On day 4, each test animal was allowed to freely explore all three empty chambers for 10 min, then driven to the central chamber with the doors closed while an age-, size-, and gender-matched unfamiliar non-littermate mouse (S) was placed into a cup in one side chamber, and an empty cup (C) was placed in the opposite side chamber. The side of the social partner chamber was randomised. The test mouse was then allowed to freely explore chambers for 10 min. Re-testing sessions at day 7 and day 14 after administration were performed using the day 4 protocol, with new stranger mice that the test mouse had never encountered previously. The sex of the test mice was randomised where possible. The testing apparatus was thoroughly cleaned after testing each mouse.

All behavioural sessions were video-recorded using a CMOS camera (Basler acA2000-165umNIR) and Basler software (Basler AG, Germany). Video recordings were used to track the position of the body of the animal in each chamber, as well as to manually time the duration of the test mouse’s social exploratory behaviour by a blinded, well-trained observer. We observed that the test mice frequently spent time in a side chamber while avoiding the interaction object; therefore, we manually quantified the actual interaction duration of the test mouse with the cup, quantified as either sniffing or exploring the cup that is either empty or containing another mouse with head and body directed toward the cup. We then derived a normalised sociability index using the equation:

$$Sociabilityindex=\frac{InteractiondurationwithS-InteractiondurationwithC}{InteractiondurationwithS+InteractiondurationwithC}$$

Sociability index values of > 0 indicate a preference for the social cup over the empty cup; values of <0 indicate the opposite.

Total interaction time is quantified as:

$$Totalinteractiontime=InteractiondurationwithS+InteractiondurationwithC$$

Position tracking and analysis

We used DeepLabCut^66,67 to track the positions of the randomly selected test mouse during the habituation phase. Each mouse was labelled by a trained network to identify at least four body parts, including the snout, two ears, and tail base. Body positions were then calculated from the label coordinates and analysed using custom codes in MATLAB. Movement speed is calculated as the difference between the body positions in two consecutive image frames.

Chamber occupancy during the habituation phase of the test was calculated. No consistent preference for the sites (left or right) of the social partner chamber and empty cup was detected either for an experimental group or for the same mouse on different test sessions.

Male and female mice of the same genotype were first analysed separately to assess potential sex-related differences in behaviours in our dataset (see Supplementary Fig. 1). Given that no differences were detected and the data spread was largely overlapping between sexes, male and female mice of the same genotype were grouped together for final analyses. No statistical methods were used to predetermine sample sizes, but our sample sizes are similar to those reported in previous publications. Post hoc power calculations were performed, which verified that the sample size for our main hypotheses and conclusions was sufficient.

Grooming analysis

Episodes of grooming behaviour were manually scored by an experimenter who was blind to the genotyping and treatment conditions.

Statistical analysis

Calculations were carried out using SPSS and MATLAB.

Head twitch response, grooming, and distance travelled data were analysed using the non-parametric Kruskal-Wallis test with alpha-adjusted Dunn’s multiple post-hoc comparisons, or the Mann-Whitney test as a priori specified.

Homogeneity of variance and sphericity were tested for all ANOVAs and were not violated. We reported partial eta squared as the effect sizes for ANOVAs, or 95% confidence intervals of differences for post-hoc paired comparisons.

For the awake treatment durability experiment, we first performed repeated-measures ANOVA using behavioural measures (social behaviour, combining sociability, total interaction time and locomotion) and time points as within-subject variables, sex and treatments as between-subjects factors. We performed Levene’s test for homogeneity of variance and did not detect any significant differences. We then performed univariate analyses, which revealed that sociability was significantly affected by psilocybin treatment but not the other two behavioural measures. We tested for normality using the Anderson-Darling test. We checked sphericity by way of Mauchly’s test and found that sphericity had not been violated, and therefore reported Bonferroni-corrected values for all post-hoc comparisons. For Fig. 1c and 2, some mice had brain tissue harvested for other use after the test session at 1 week post-treatment (timepoint 2, N = 4 for Cntnap2^-/-_vehicle, 5 for Cntnap2^-/-_psilocybin, 4 for Cntnap2^+/+_vehicle). ANOVA was performed on mice that were tested at all time points (N = 15 per each Cntnap2^-/- condition and 11 for Cntnap2^+/+_vehicle).

We performed similar repeated-measures ANOVAs using sociability and timepoint as within-subject variables for other durability experiments (on Cntnap2^+/+ WT and Cntnap2^+/- heterozygous groups, and treatment under anaesthesia for Cntnap2^-/- KO mice), and 5HT2AR pretreatment experiments. For durability experiments, we used treatments as between-subjects factors for the durability experiments. For 5HT2AR pretreatment experiments, we used pretreatment conditions as between-subjects factors. We also performed Levene’s test for homogeneity of variance and did not detect any significant differences. We also checked that sphericity is not violated by way of Mauchly’s test.

For a cross-sectional design where sociability was only measured at one time-point (two weeks after treatment), we used independent sample T test.

Reporting summary

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

Author contributions

C.S. and T.K. designed the study. C.S. and T.C. performed the experiments. C.S., T.K., and T.B. analysed the data. C.S. and T.K. supervised the project and wrote the manuscript with inputs from all authors.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by: US National Institute of Health BRAIN Initiative Grant (5U01NS099573) to TK, Hong Kong STEM Initiative Funds (Hong Kong Jockey Club Charities Trust) to TK, Research Matching Grant (Government of Hong Kong, TFD2023-P05-02) to TK, Lee Kuan Yew Postdoctoral Fellowship administered by Nanyang Technological University Singapore (022506-00001) to CS, and Open Fund Young Individual Research Grant (MOH-001720) administered by the Singapore Ministry of Health’s National Medical Research Council to CS. This work was supported by the Investigational Supply Programme of Usona Institute US.

Data availability

Source data is provided. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77238-0.