Psilocybin

Psilocybin induces rapid and persistent growth of dendritic spines in frontal cortex in vivo

This cell study shows that brain cells, specifically the layer five pyramidal neurons in mice, grew by 10% after the introduction of psilocybin. The effects were still present 30 days later, providing more evidence for brain plasticity as an underlying mechanism of psychedelic-assisted therapies' long-lasting effects.

Authors

Delagarza, K.
Gregg, I.
Kwan, A. C.

Published

June 25, 2021

Neuron

individual Study

Links

Read Paper DOI Google Scholar

Abstract

Psilocybin is a serotonergic psychedelic with untapped therapeutic potential. There are hints that the use of psychedelics can produce neural adaptations, although the extent and timescale of the impact in a mammalian brain are unknown. In this study, we used chronic two-photon microscopy to image longitudinally the apical dendritic spines of layer 5 pyramidal neurons in the mouse medial frontal cortex. We found that a single dose of psilocybin led to ∼10% increases in spine size and density, driven by an elevated spine formation rate. The structural remodeling occurred quickly within 24 h and was persistent 1 month later. Psilocybin also ameliorated stress-related behavioral deficit and elevated excitatory neurotransmission. Overall, the results demonstrate that psilocybin-evoked synaptic rewiring in the cortex is fast and enduring, potentially providing a structural trace for long-term integration of experiences and lasting beneficial actions.

Unlocked with Blossom Pro

Research Summary of 'Psilocybin induces rapid and persistent growth of dendritic spines in frontal cortex in vivo'

Introduction

Serotonergic psychedelics produce atypical conscious states with altered perception, cognition and mood, and have long been investigated for therapeutic potential in disorders such as depression, obsessive-compulsive disorder and addiction. Psilocybin in particular has shown rapid and sustained antidepressant effects in humans and received FDA "Breakthrough Therapy" designation, while structural neuroplasticity in the frontal cortex has been implicated in antidepressant action. Prior work has linked psychedelics to markers of synaptic plasticity in gene expression, to transient spine enlargement and branch proliferation in neuronal cultures, and to presynaptic changes in large animals, but no study had directly demonstrated psilocybin-induced structural remodelling of dendritic spines at cellular resolution in a mammalian brain nor established the time course of such changes in vivo. Shao and colleagues set out to test whether a single dose of psilocybin produces rapid and persistent changes in dendritic spines of medial frontal cortex pyramidal neurons in mice, and whether any structural changes are associated with altered excitatory synaptic transmission and behavioural outcomes. The investigators combined longitudinal in vivo two-photon imaging in Thy1 GFP mice, confocal imaging in a separate cohort, whole-cell electrophysiology, behavioural assays (head-twitch response and learned helplessness), and pharmacological manipulation with the 5-HT2A antagonist ketanserin to probe receptor involvement and persistence of new spines over up to 34 days.

Methods

The study used adult mice across several coordinated experiments. Behavioural pharmacology (head-twitch and learned helplessness) employed C57BL/6J mice, whereas structural imaging used Thy1 GFP (line M) mice that express GFP sparsely in deep-layer pyramidal neurons. Psilocybin was administered intraperitoneally at doses of 0, 0.25, 0.5, 1 and 2 mg/kg to define a dose–response for the head-twitch assay; 1 mg/kg was selected as the primary dose because it marked the inflection point for head-twitch behaviour. For ketanserin pretreatment experiments, ketanserin 1 mg/kg i.p. was given 10 minutes before psilocybin or saline. Longitudinal structural imaging used chronic cranial windows over the right medial frontal cortex (Cg1/M2). Two-photon imaging sessions were performed on days −3, −1, 1, 3, 5 and 7 relative to treatment, with a subset of animals re-imaged at day 34. In total, the longitudinal two-photon dataset comprised 1,820 dendritic spines on 161 branches from 12 mice (6 male, 6 female). Imaging parameters and motion-correction procedures were described; spine inclusion required a protrusion extending > 0.4 mm from the shaft according to the extracted protocol (note: the extracted text may contain unit artefacts). Structural endpoints were spine density, spine head width and spine protrusion length, together with spine formation and elimination rates calculated between consecutive sessions. Confocal imaging replicated the structural analysis in a separate cohort sacrificed at 24 h post-injection; analyses were extended to multiple brain regions (Cg1/M2, PrL/IL and M1) and both apical and basal compartments, yielding 23,226 spines on 1,885 branches from 12 mice (6 male, 6 female). Electrophysiology involved acute coronal slices containing Cg1/M2 prepared 24 h after psilocybin or saline injection. Whole-cell recordings of putative layer 5 pyramidal neurons measured miniature excitatory postsynaptic currents (mEPSCs) in the presence of tetrodotoxin and picrotoxin to isolate spontaneous glutamatergic events. Behavioural assays included high-speed video-recorded head-twitch monitoring and a five-day learned helplessness protocol with two induction sessions of inescapable footshock followed by pre- and post-treatment active-avoidance tests. Statistical analysis relied primarily on linear mixed-effects models for longitudinal imaging (treatment, sex, time and interactions; random effects for dendrites nested within mice), two-way ANOVAs for electrophysiology and confocal data, and appropriate non-parametric tests for some behavioural comparisons. Blinding was used for spine scoring, electrophysiology and image analyses, and post hoc tests applied Bonferroni correction for multiple comparisons.

Results

Behaviourally, head-twitch responses showed a sharp increase at 1 mg/kg psilocybin; at that dose head-twitch frequency peaked 6–8 minutes after injection and subsided by about 2 hours. In the learned helplessness paradigm, psilocybin (1 mg/kg) reduced within-individual escape failures (post hoc Bonferroni-corrected t test, p = 0.004); among animals classified as susceptible, 15 of 16 (94%) showed a decrease or no change in escape failures after psilocybin. However, an across-subjects mixed model comparing saline, ketamine and psilocybin did not show a significant main effect of treatment (p = 0.09), which the authors attribute to inter-individual variability. Longitudinal two-photon imaging in Cg1/M2 demonstrated that a single 1 mg/kg dose of psilocybin produced rapid increases in dendritic spine density and morphology. Spine density increased by 7% ± 2% on day 1 and 12% ± 3% on day 7 compared with baseline (main effect of treatment, p = 0.011, mixed-effects model). Average spine head width rose by 11% ± 2% on day 1 and 5% ± 1% on day 7 (main effect p = 0.013), and spine protrusion length was also increased. Analysis of turnover revealed that the density gain was driven by an elevated spine formation rate with no detectable change in the elimination rate. Females showed an increase in formation rate of 8% ± 2% (absolute formation rate from ≈7% ± 1% on day −1 to ≈15% ± 2% on day 1; main effect p = 0.034), whereas males showed a smaller increase of 4% ± 2% (≈6% ± 1% to ≈10% ± 2%). Persistence analyses indicated that roughly half of the spines formed on day 1 remained on day 7 (49% ± 10% for females, 52% ± 12% for males). In a subset re-imaged at day 34 (n = 4 mice), a fraction of the new spines remained stable at 34 days (34% ± 10% females; 37% ± 12% males). The authors note heterogeneity between branches, with some branches retaining many new spines and others losing them almost entirely. Ketanserin (1 mg/kg) pretreatment abolished psilocybin-induced head-twitch behaviour but did not fully block structural effects. In ketanserin-pretreated mice (1,443 spines on 120 branches from 8 animals), the psilocybin-associated spine density increases were no longer statistically significant (+5% ± 2% day 1; +8% ± 2% day 7; main effect p = 0.09), yet spine head width increased (+12% ± 1% on day 1 and day 7; p = 0.01), and increases in protrusion length and formation rate were still detectable. The authors point out that 1 mg/kg ketanserin produces only ~30% 5-HT2A receptor blockade in rodent cortex, which may explain the preserved structural effects. Electrophysiologically, whole-cell recordings 24 h after treatment showed a substantial increase in mEPSC frequency in psilocybin-treated animals compared with saline controls (main effect p = 0.0002, two-way ANOVA), consistent with an increase in the number of functional excitatory synapses; there was a trend toward increased mEPSC amplitude (p = 0.06). The confocal replication across multiple frontal regions confirmed psilocybin’s ability to promote spine growth in Cg1/M2, particularly in females (apical spine density: 0.46 ± 0.02 mm−1 in controls versus 0.50 ± 0.01 mm−1 after psilocybin; treatment × sex p = 0.013). Additional region- and compartment-specific effects were observed: increased spine protrusion length in PrL/IL (p = 0.026), increased spine density and head width in M1 in females (treatment × sex p = 0.021 and p = 0.008 respectively), and significant effects on basal dendrites in Cg1/M2 (spine density p = 0.0004; protrusion length p = 0.012).

Discussion

Shao and colleagues interpret their findings to show that a single dose of psilocybin evokes rapid and durable growth of dendritic spines on layer 5 pyramidal neurons in the mouse medial frontal cortex. They suggest two non-mutually exclusive ways this structural remodelling could relate to therapeutic effects: first, by restoring synaptic connections lost in depression, and second, by increasing the brain’s capacity for experience-dependent learning and integration of psychological interventions. The observed time course—an early surge in spine formation that yields persistent new spines weeks later—parallels reports for other rapid-acting antidepressants such as ketamine, implying that synaptic rewiring may be a shared downstream mechanism among compounds with rapid behavioural effects. The authors discuss receptor-level uncertainty. Although ketanserin pretreatment fully suppressed the acute head-twitch response, it did not abolish psilocybin-induced structural changes; this dissociation implies that the psychedelic-like behavioural effects and structural plasticity may be separable. Nevertheless, the authors caution that ketanserin at the chosen dose achieves only partial 5-HT2A blockade (~30% in rodents), so residual receptor activity could still mediate the spine changes. They note species differences in 5-HT2A receptor kinetics and occupancy thresholds for intense subjective effects in humans (50%–70%), and therefore emphasise that the role of 5-HT2A and other serotonin receptor subtypes remains unresolved; cell-type and region-specific knockouts are proposed as a path to more decisive evidence. Limitations acknowledged include the uncertain generalisability to humans and the incomplete receptor blockade in the ketanserin experiments. The authors also highlight inter-branch heterogeneity in responsiveness and variability in behavioural outcomes across animals as factors that complicate interpretation. Finally, they call for future work to elucidate the molecular pathways by which agents with different primary targets (for example, psilocybin versus ketamine) converge on similar structural and functional synaptic changes, noting that clarifying these mechanisms will be important for understanding the neurobiology of rapid-acting antidepressants.

View full paper sections

INTRODUCTION

Serotonergic psychedelics are compounds that produce an atypical state of consciousness characterized by altered perception, cognition, and mood. It has long been recognized that these compounds may have therapeutic potential for neuropsychiatric disorders, including depression, obsessive-compulsive disorder, and addiction. Among serotonergic psychedelics, psilocybin is recently shown to relieve depression symptoms rapidly and with sustained benefits for several months. This progress led to a ''Breakthrough Therapy'' designation by the United States Food and Drug Administration (FDA) in 2019 and the initiation of multi-site clinical trials to test psilocybin as a treatment for major depressive disorder. It is well established that structural neuroplasticity in the frontal cortex is key to the action of antidepressants. Synaptic atrophy is found in the prefrontal cortex of patients with depression. Likewise, synaptic deficiencies, including loss of dendritic arborization, reduced spine density, and damped neurotransmission, are present in the frontal cortex of rodent chronic stress models. By contrast, compounds with fast-acting antidepressant effects promote structural plasticity to reverse the synaptic deficits caused by chronic stress. For instance, a single dose of ketamine leads to higher spine density in the medial frontal cortex of rodents, which is due to an increase in spine formation rate, likely involving elevated calcium signaling in the dendritic compartment. What is the current evidence that serotonergic psychedelics such as psilocybin can alter synaptic architecture? A few studies have shown that the expression of genes involved in synaptic plasticity is elevated after administration of serotonergic psychedelics in rats. In neuronal cultures, bath application of serotonergic psychedelics induces transient increases in spine sizeand proliferation of dendritic branches. A recent study showed that an analog of ibogaine, a psychedelic with different molecular targets than psilocybin, increases spine formation rate in mice. Finally, in the pig, psilocybin administration was associated with higher binding of a presynaptic protein tracer in positron emission tomography. Although these studies provided clues linking serotonergic psychedelics to structural and functional neuroplasticity, significant gaps remain. In particular, there has been no direct demonstration of psilocybininduced structural plasticity at cellular resolution in a mammalian brain. Importantly, the timescale in which such synaptic rewiring may occur in vivo is unknown.

RESULTS

A single dose of psilocybin leads to long-lasting increases in spine density and spine head width in the mouse medial frontal cortex To test the potency and dose dependence of psilocybin in mice, we measured the head-twitch response, a classic assay for characterizing psychedelic compounds in rodents. We observed that mice would exhibit high-frequency headshakes intermittently after administration of psilocybin (Video S1). We characterized 82 C57BL/6J mice (40 males and 42 females) with five doses of psilocybin (0, 0.25, 0.5, 1, and 2 mg/kg, intraperitoneally [i.p.]; range, 7-10 per sex per dose). A sharp rise of elicited headtwitch responses occurred at 1 mg/kg (Figure), consistent with prior reports. Thus, we chose to use 1 mg/kg (the inflection point of the dose-dependence curve) to assess psilocybin's effect on structural plasticity. At this dose, the rate of head-twitch responses peaked at 6-8 min after administration and then gradually declined until they ceased at $2 h (Figure). Next, to determine if this dose is associated with mitigation of stress-related phenotypes in mice, we tested the effect of 1 mg/ kg psilocybin in a learned helplessness paradigm. Mice received prolonged stress in the form of repeated, inescapable footshocks over two induction sessions, and then were tested for active avoidance behavior with escapable footshocks 1 day before and 1 day after treatment (Figure). Susceptible animals are mice in a learned-helpless state characterized by reduced attempts to escape from the footshocks. We used 68 C57BL/6J mice to compare psilocybin (1 mg/kg, i.p.) against saline and ketamine (10 mg/kg, i.p.), which served as negative and positive controls. Within individuals, psilocybin reduced the proportion of escape failures (p = 0.004, post hoc Bonferroni-corrected t test; Figure). For susceptible animals, the psilocybin group had a decrease or no change in escape failures in all but 1 of the 16 mice tested (94%; Figuresand). Across individuals, comparison among saline, ketamine, and psilocybin did not reveal a main effect of treatment (p = 0.09, two-way ANOVA; Figure), presumably because of the across-subject variability in the behavioral responses. Nevertheless, overall, these results indicate that psilocybin can ameliorate maladaptive behavior induced by uncontrollable stress in mice. In the body, psilocybin is dephosphorylated to psilocin, an agonist of 5-HT 2A receptors that are densely expressed in apical dendrites of layer 5 pyramidal neurons in the medial frontal cortex of primates and rodents. We therefore hypothesize that psilocybin may modify the dendritic architecture in the medial frontal cortex. We used chronic two-photon microscopy to track apical dendritic spines in the cingulate/premotor (Cg1/M2) region of the medial frontal cortex of Thy1 GFP mice (line M), in which a sparse subset of infragranular (layer 5 and 6) pyramidal neurons express GFP(Figuresand). We imaged before and after administering psilocybin (1 mg/kg, i.p.) or saline at 2-day intervals and then again $1 month later for a total of seven imaging sessions (Figuresand). In total, we tracked 1,820 dendritic spines on 161 branches from 12 animals (6 males and 6 females). Spine morphology was analyzed blind to experimental conditions using standardized procedures. We took advantage of the longitudinal data to normalize the change in spine density as fold change in individual dendritic segments. For statistical analyses, we used a mixed-effects model including treatment, sex, and days as factors, as well as all interaction terms. Variation within mouse and dendrite across days was accounted by including random effects terms for dendrites nested by mice. Our results indicate that a single dose of psilocybin induces a significant elevation in spine density (+7% ± 2% on day 1 and +12% ± 3% on day 7; main effect of treatment, p = 0.011, mixed-effects model; Figures), an increase in the width of spine heads (+11% ± 2% on day 1 and +5% ± 1% on day 7; main effect of treatment, p = 0.013; Figures), and higher spine protrusion lengths (Figures). Details for all statistical tests, including sample sizes, are provided in Table. Psilocybin elevates the formation rate of dendritic spines in vivo Increased spine density could be due to a higher formation rate, a lower elimination rate, or both. To distinguish between the possibilities, we analyzed the same dendritic segments across adjacent imaging sessions to determine the turnover of dendritic spines. In females, the spine formation rate increased by 8% ± 2% after psilocybin (absolute values for the formation rate:. See also Figure. 7% ± 1% on day À1 and 15% ± 2% on day 1; main effect of treatment, p = 0.034, mixed-effects model; Figuresand). Likewise, the spine formation rate was higher by 4% ± 2% in males after psilocybin (absolute values for the formation rate: 6% ± 1% on day À1 and 10% ± 2% on day 1). By contrast, there was no change in the elimination rate of spines (Figure). The increase in spine formation rate was highest shortly after psilocybin administration and then diminished in subsequent days to return to baseline level and in equilibrium with the elimination rate. These data therefore support the view that the long-term increase in spine density is due to an initial boost of enhanced spine formation. A fraction of the psilocybin-induced spines is persistent for at least 1 month A key question was whether the new spines formed after psilocybin administration would persist, because nascent dendritic spines can take 4 days to mature into functional synapses. For this reason, we tracked the new spines formed after psilocybin on day 1 and found that approximately half of them remained stable on day 7 (49% ± 10% for females and 52% ± 12% for males; Figure). This suggests that a portion of the new dendritic spines induced by psilocybin would become functional synapses. Furthermore, because clinical trials indicated that psilocybin may provide long-term benefits for up to several months, for a subset of four mice, we imaged yet again at a further time point at 34 days after administration to find. See also Figure. that a fraction of the psilocybin-evoked new spines remained persistent (34% ± 10% for females and 37% ± 12% for males; Figuresand). Psilocybininduced spines were not significantly different and therefore were no less stable than spines formed in control conditions (main effect of treatment, p = 0.9, two-way repeated-measures ANOVA). Intriguingly, select individual dendritic branches appeared to retain all of the new spines, while other branches lost them almost completely, suggesting heterogeneity and potentially responsive and nonresponsive subpopulations of pyramidal neurons. Altogether, these results demonstrate that a single dose of psilocybin induces rapid and long-lasting dendritic remodeling in layer 5 pyramidal neurons in the mouse medial frontal cortex. Ketanserin pretreatment, sufficient to abolish headtwitch responses, does not block psilocybin-induced structural plasticity Multiple lines of evidence demonstrated that 5-HT 2A receptors are essential for serotonergic psychedelics' psychotomimetic effects in humansand head-twitch responses in mice. To study whether the effects of psilocybin on structural plasticity may involve 5-HT 2A receptors, we reduced the number of available 5-HT 2A receptors in the brain by pretreating animals with the 5-HT 2A receptor antagonist ketanserin (1 mg/kg, i.p.), 10 min prior to the administration of psilocybin (1 mg/kg, i.p.) or saline. Behaviorally, the ketanserin pretreatment abolished completely the psilocybin-induced head-twitch responses (Figure). Next, we repeated the two-photon imaging experiments in ketanserin-pretreated mice and tracked 1,443 dendritic spines on 120 branches from 8 animals (4 males and 4 females) (Figure). We found that although the enhancing effect of psilocybin on spine density was no longer statistically significant (+5% ± 2% on day 1 and +8% ± 2% on day 7; main effect of treatment, p = 0.09, mixed-effects model; Figures, and S3B), there were still detectable increases in spine head width (+12% ± 1% on day 1 and +12% ± 1% on day 7; main effect of treatment, p = 0.01; Figures, and S3D), spine protrusion length (Figures S3E-S3G), and spine formation rate (absolute values for the formation rate: 5% ± 1% on day À1 and 10% ± 2% on day 1 for female mice; 8% ± 1% on day À1 and 14% ± 2% on day 1 for male mice; Figuresand). It was previously determined that 1 mg/kg ketanserin led to only an $30% blockade of 5-HT 2A receptors in the rat neocortex, likely due to limited transport into the brain for rodents. Therefore, in agreement with a recent study in the. See also Figure. hippocampus, our results demonstrate that while a moderate knockdown of 5-HT 2A receptor function eliminates head-twitch responses, it is not sufficient to abolish the psilocybin-induced structural remodeling in mice. Psilocybin elevates excitatory neurotransmission in medial frontal cortex Most, but not all, dendritic spines are functional glutamatergic synapses. To elaborate on the effects of psilocybin on synaptic function, we performed whole-cell recordings in brain slices to measure miniature excitatory postsynaptic currents (mEPSCs) from putative layer 5 pyramidal neurons, identified based on morphology, in Cg1/M2 (Figure). The results showed that 24 h after treatment, we could detect an increase in mEPSC frequency in psilocybin-treated animals compared to saline controls (main effect of treatment, p = 0.0002, two-way ANOVA; Figure). We also report a moderate effect of psilocybin on mEPSC amplitude (main effect of treatment, p = 0.06, twoway ANOVA; Figure). Because mEPSC frequency and amplitude reflect the number and strength of synapses, these results demonstrate that the psilocybin-induced structural remodeling is accompanied by enhanced excitatory neurotransmission.

DEPENDENCE OF PSILOCYBIN-INDUCED STRUCTURAL REMODELING ON BRAIN REGION AND DENDRITE TYPE

To further support the conclusions, we tried to replicate the findings in a completely separate cohort of animals using a different approach. We administered Thy1 GFP mice with psilocybin (1 mg/ kg, i.p.) or saline, sacrificed them 24 h later, and imaged coronal brain sections using confocal microscopy. We expanded analyses to six areas of the brain, including two zones that encompass apical and basal dendrites and three regions of the frontal cortex: Cg1/M2, prelimbic/infralimbic (PrL/IL), and primary motor cortex (M1) (Figures). The results, consisting of 23,226 dendritic spines counted on 1,885 branches from 12 animals (6 males and 6 females), reaffirmed the ability of psilocybin to promote the growth of new dendritic spines in Cg1/M2 in female mice (spine density: 0.46 ± 0.02 versus 0.50 ± 0.01 mm À1 ; Figure). Effects of psilocybin on spine density were more pronounced in female animals than in male animals (treatment 3 sex, p = 0.013, two-way ANOVA; Figure). We did not detect differences in spine protrusion length and spine head width (Fig-and), which may be due to the across-subjects design, as we could not normalize the changes to the same dendritic branch, and therefore, this approach had less power than the within-subjects design of the chronic imaging experiment. We detected select morphological differences in PrL/IL and M1, including increases in spine protrusion length in PrL/IL (main effect of treatment, p = 0.026, two-way ANOVA), spine density in M1 in females (treatment 3 sex, p = 0.021, two-way ANOVA), and spine head width in M1 in females (treatment 3 sex, p = 0.008, two-way ANOVA), suggesting that the plasticity-promoting impact may not be unique to Cg1/M2 (Figures). Furthermore, psilocybin had significant impact on basal dendrites in Cg1/M2, leading to higher spine density and spine protrusion length (spine density: main effect of treatment, p = 0.0004; spine protrusion length: main effect of treatment, p = 0.012, two-way ANOVA; Figuresand). Overall, the two sets of data converge to show that psilocybin promotes the growth of dendritic spines in layer 5 pyramidal neurons in the medial frontal cortex.

DISCUSSION

This study demonstrates that a single dose of psilocybin evokes growth of dendritic spines in the medial frontal cortex of the mouse. The persistence of the neural modifications is notable and may relate to the compound's therapeutic effects for at least two reasons. First, depression is associated with a loss of synapses in the frontal cortex. Restoring the number of neuronal connections may correct such deficit, providing a biological mechanism for alleviating symptoms of depression. Second, structural remodeling is integral to learning and facilitates the storage of lifelong memories. Psilocybin-induced neural plasticity could prime the brain for integrating new psychological experiences. Regardless of the relative importance of these mechanisms, which are not mutually exclusive, our results indicate that the underlying structural trace in the brain is enduring and can be observed a long time after the initial drug exposure. There is an ongoing debate over whether the hallucinogenic effects of serotonergic psychedelics are dissociable from the therapeutic effects. Consistent with another new study, our results indicate that structural remodeling in the medial frontal cortex is undeterred by a moderate knockdown of 5-HT 2A receptor availability. The possibility to disrupt psilocybin's acute behavioral effects without abolishing structural plasticity actions has clear implications for treatment in the clinic. However, it is not yet clear if the results will extrapolate to humans, because 5-HT 2A receptors have species-dependent differences in dissociation kinetics with serotonergic psychedelics. Moreover, our results do not rule out the involvement of 5-HT 2A receptors, because this dose of ketanserin only blocks $30% of 5-HT 2A receptors in rodents, and the unaffected receptors might be enough to drive the dendritic remodeling. This number may be compared to the 50%-70% 5-HT 2A receptor occupancy level required for the more intense psilocybin-induced psychological experience in humans. Future studies with region-and cell-type-specific knockout of serotonin receptor subtypes are needed to produce more decisive evidence on the role of 5-HT 2A and other receptors in mediating the effects of psilocybin on dendritic plasticity. By showing that the time course for psilocybin-induced structural remodeling is rapid and persistent in vivo, our study suggests that synaptic rewiring may be a mechanism shared by compounds with rapid antidepressant effects. Of note, the timing of psilocybin's effect on the neural architecture is reminiscent of ketamine, which at subanesthetic dose causes similar rapid increase in spine density and elevation of spine formation rate in the medial frontal cortex. However, still unknown is how drugs with disparate molecular targets may yield comparable modifications on neural architecture and behavior. Elucidating the mechanisms will be crucial toward unraveling the neurobiology of rapid-acting antidepressants. Head-twitch response Head-twitch response was evaluated using 40 male and 42 female C57BL/6J mice. Upon arrival, animals habituated at the housing facility for > 2 weeks before behavioral testing. Behavioral testing took place between 10:00 AM and 4:00 PM. Animals were weighed and injected intraperitoneally with saline or psilocybin (0.25, 0.5, 1, or 2 mg/kg). For ketanserin pretreated groups, animals received ketanserin (1 mg/kg, i.p.; S006, Sigma-Aldrich) 10 min before administration of saline or psilocybin (1 mg/kg, i.p.). Meanwhile, a group of animals received saline (10 mL/kg, i.p.) 10 min before administration of psilocybin (1 mg/kg, i.p.) as positive controls. We tried a higher dose of ketanserin (4 mg/kg, i.p.; n = 8 mice), however animals became visibly ill. We measured head-twitch response in groups of two animals: after injections, the two animals were immediately placed into separate chambers, made by inserting a plastic divider to halve an open-field-activity box (12'' W x 6'' H x 10'' D). The box was within a sound attenuating cubicle with a built-in nearinfrared light source and a white light source (interior: 28'' W x 34'' H x 22'' D, Med Associates Inc.). Videos were recorded by a highspeed (213 fps), near-infrared camera (Genie Nano M1280, Teledyne Dalsa) mounted overhead above the open-field-activity box. Typical recordings were 30 minutes long and, for a subset of mice (2 males and 2 females), extended to > 150 minutes. Between each measurement, the open-field activity box was thoroughly cleaned with 70% ethanol. The videos were scored for head twitches by an experienced observer blind to the experimental conditions.

LEARNED HELPLESSNESS

Learned helplessness was evaluated using 34 male and 34 female C57BL/6J mice. Upon arrival, animals habituated at the housing facility for > 2 weeks before behavioral testing. Behavioral testing took place between 7:00 AM and 4:00 PM. All animals underwent a 5-day protocol, adapted from previously described procedures for mice. An animal was placed in a shuttle box separated by a guillotine door which, when open, allowed the animal to shuttle between two compartments (16'' x 6.5'' x 8.5,'' Med Associates Inc.). On Day 1, the mouse received an induction session which involved 360 inescapable footshocks (0.15 mA) with variable duration (1-3 s) and variable inter-shock interval (1-15 s). The guillotine door was open throughout the induction session. On Day 2, the animal received another induction session with the same parameters. On Day 3, the animal underwent Test 1. The test session began with the guillotine door opening. Each test session involved a series of 30 footshocks (0.15 mA). The animal would receive a footshock from the grid floor of the compartment it was presently in. Footshock was terminated if the animal shuttled to the other compartment (''escape'') or at the end of the 10 s if it did not shuttle (''escape failure''), whichever occurred earlier. Each footshock was followed by an inter-shock interval (30 s), during which the guillotine door was closed. Escape latency was defined as the time elapsed from onset of footshock to time crossing the guillotine door, measured by infrared photobeams, for escape trials. Escape latency was set to 10 s for escape failure trials. On Day 4, 24 hr after Test 1, the animal was weighed and injected with saline (10 mL/kg, i.p.), ketamine (10 mg/kg, i.p.), or psilocybin (1 mg/kg, i.p.) and then immediately returned to their home cages. On Day 5, 24 hr after treatment, the animal underwent Test 2 which followed the same procedures as Test 1. At the end, data from all animals were collated, and mice were classified as ''resilient / non-learned helpless'' or ''susceptible / learned helpless'' based on their performance in Test 1. Escape failures and escape latencies were used as indicators of learned helplessness, and a k-means (k = 2) clustering algorithm was applied for classification.

SURGERY

Prior to surgery, the mouse was injected with carprofen (5 mg/kg, s.c.; 024751, Henry Schein Animal Health,) and dexamethasone (3 mg/kg, i.m.; 002459, Henry Schein Animal Health). During surgery, the mouse was anesthetized with isoflurane (3 -4% for induction and 1 -1.5% for the remainder of surgery) and fixed in a stereotaxic apparatus (David Kopf Instruments). The body of the mouse rested on a water-circulating heating pad (Stryker Corp) set to 38 C. Petrolatum ophthalmic ointment (Dechra) was used to cover the animal's eyes. The hair on the head was shaved, and the scalp was wiped and disinfected with ethanol pad and betadine. An incision was made to remove the skin and the connective tissue above the skull was removed. Subsequently, a dental drill was used to make a $3-mm-diameter circular craniotomy above the right medial frontal cortex (center position: +1.5 mm anterior-posterior, AP; +0.4 mm medial-lateral, ML; relative to bregma). Artificial cerebrospinal fluid (ACSF, containing (in mM): 135 NaCl, 5 HEPES, 5 KCl, 1.8 CaCl2, 1 MgCl2; pH 7.3) was used to irrigate the exposed dura above brain. A two-layer glass window was made from two round 3-mmdiameter, #1 thickness glass coverslip (64-0720 (CS-3R), Warner Instruments), bonded by UV-curing optical adhesive (NOA 61, Norland Products). The glass window was carefully placed over the craniotomy and, while maintaining a slight pressure, adhesive (Henkel Loctite 454) was used to secure the glass window to the surrounding skull. A stainless steel headplate was affixed on the skull with C&B Metabond (Parkell) centered on the glass window. Carprofen (5 mg/kg, s.c.) was given to the mouse immediately after surgery and on each of the following 3 days. The mouse would recover for at least 10 days after the surgery before the start of imaging experiments.

TWO-PHOTON IMAGING

The two-photon microscope (Movable Objective Microscope, Sutter Instrument) was controlled by ScanImage 2019 software. The laser excitation was provided by a tunable Ti:Sapphire femtosecond laser (Chameleon Ultra II, Coherent) and focused onto the mouse brain with a water-immersion 20X objective (XLUMPLFLN, 20x/0.95 N.A., Olympus). The laser power measured at the objective was % 40 mW. To image GFP-expressing dendrites, the laser excitation wavelength was set at 920 nm, and a 475 -550 nm bandpass filter was used to collect the fluorescence emission. During an imaging session, the mouse was head fixed and anesthetized with 1 -1.5% isoflurane. Body temperature was controlled using a heating pad and DC Temperature Controller (40-90-8D, FHC) with rectal thermistor probe feedback. Each imaging session did not exceed 2 hr. We imaged apical tuft dendrites at 0 -200 mm below the dura. To target Cg1/M2 region, we imaged within 0 -400 mm of the midline as demarcated by the sagittal sinus. Multiple fields of view were imaged in the same mouse. For each field of view, 10 -40-mm-thick image stacks were collected at 1 mm steps and at 1024 3 1024 pixels at 0.11 mm per pixel resolution. We kept the same set of imaging parameters for the different imaging sessions. For longitudinal imaging, we would return to the same fields of view across imaging sessions by locating and triangulating from a landmark on the left edge of the glass window. Each mouse was imaged on days À3, À1, 1, 3, 5 and 7 relative to the day of treatment. A subset of mice (2 males and 2 females) was imaged additionally on day 34. On the day of treatment (day 0), there was no imaging, and the mouse was injected with either psilocybin (1 mg/kg, i.p.) or saline (10 mL/kg, i.p.). For ketanserin pretreated groups, animals received ketanserin (1 mg/kg, i.p.) 10 min before administration of psilocybin (1 mg/kg, i.p.) or saline (10 mL/kg, i.p.). After injection, the mouse was placed in a clean cage under normal room lighting to visually inspect for head-twitch responses for 10 minutes, before returning the mouse to its home cage.

CONFOCAL IMAGING

Each mouse was injected with either psilocybin (1 mg/kg, i.p.) or saline (10 mL/kg, i.p.). At 24 hr after injection, the mouse was deeply anesthetized with isoflurane and transcardially perfused with phosphate buffered saline (PBS, P4417, Sigma-Aldrich) followed by paraformaldehyde (PFA, 4% in PBS). The brains were fixed in 4% PFA for 24 hr at 4 C, and then 50-mm-thick coronal brain slices were sectioned using a vibratome (VT1000S, Leica) and placed on slides with coverslip with mounting medium. The brain slices were imaged with a confocal microscope (LSM 880, Zeiss) equipped with a Plan-Apochromat 63x/1.40 N.A. oil objective for dendritic spine imaging and a Plan-Apochromat 20x/0.8 N.A. objective for stitching images of an entire brain slice.

BRAIN SLICE PREPARATION

Female and male mice were randomly selected to receive either psilocybin (1 mg/kg, i.p.) or saline (10 mL/kg, i.p.) 24 hr before the experiment. The experimenter performing the electrophysiological recordings and analysis was blinded to the treatment condition. Coronal brain slices containing Cg1/M2 were prepared following procedures in a prior study. Briefly, mice were deeply anesthetized with isoflurane and rapidly decapitated. The brain was quickly isolated into ice-cold slicing solution containing (in mM): 110 choline, 25 NaHCO 3 , 11.6 sodium ascorbate, 7 MgCl 2 , 3.1 sodium pyruvate, 2.5 KCl, 1.25 NaH 2 PO 4 , 0.5 CaCl 2 , and 20 glucose. Acute coronal slices (300 mm thick) were cut with a vibratome (VT1000 S, Leica Biosystems). The vibratome chamber was surrounded by ice and filled with oxygenated slicing solution. Slices were incubated in artificial cerebral spinal fluid (aCSF) containing (in mM): 127 NaCl, 25 NaHCO 3 , 2.5 KCl, 2 CaCl 2 , 1.25 NaH 2 PO 4 , 1 MgCl 2 , and 20 glucose for 30 min at 34 C. The slices were then maintained at room temperature for a minimum of 30 min before recording. The slicing solution and aCSF were prepared with deionized water (18.2 MU-cm), filtered (0.22 mm), and bubbled with 95% O 2 and 5% CO 2 for at least 15 min prior to use and throughout the slice preparation and recording.

WHOLE-CELL RECORDING

Slices were placed into an open bath chamber and perfused constantly with aCSF (2-3 mL/min) supplemented with tetrodotoxin (0.5 mM; Abcam) and picrotoxin (50 mM) to block Na + currents and GABA A receptors for isolating miniature excitatory post-synaptic currents (mEPSCs). aCSF was warmed and maintained at 34 C via an inline heater with closed-loop feedback control. Recording pipettes were pulled from borosilicate glass (BF-150-86-10, Sutter Instruments) to a resistance of 2-4 MU with a puller (P97, Sutter Instruments) and filled with double-filtered (0.22 mm) internal solution containing (in mM): 100 CsMeSO 4 , 25.5 CsCl, 10 Glucose, 10 HEPES, 8 NaCl, 4 Mg-ATP, 0.3 Na 3 -GTP, and 0.25 EGTA (pH 7.3, adjusted with 1M CsOH). Liquid junction potential was calculated to be 12.1mV and was not corrected for in recordings. Slices were visualized using differential interference contrast in a microscope (BX51W, Olympus) with a CCD camera (Retiga Electro, QImaging). Putative layer 5 pyramidal neurons were targeted for recording based on morphological features including large cell body, prominent apical dendrite, and distance from the pia. Cells were targeted with a depth of at least 30 mm below the surface of the slice. Electrophysiological recordings were performed on neurons that initially formed a stable seal and subsequently broke in successfully to the whole-cell configuration. Recordings were amplified (MultiClamp 700B, Molecular Devices) and digitized at 20 kHz (Digidata 1550, Molecular Devices). Neurons were held at À70 mV during recording. Recordings were excluded if the holding current > 200 pA when held at À70 mV or if the access resistance increased by > 10% from baseline during the recording o if the access resistance exceeds 25 MU at any point of the recording. Analysis of mEPSCs was conducted offline using the Easy Electrophysiology software (Easy Electrophysiology Ltd), with a template search algorithm. All drugs and regents were obtained from Sigma-Aldrich or Tocris unless otherwise noted.

QUANTIFICATION AND STATISTICAL ANALYSIS

Analysis of the imaging data Analyses of the two-photon and confocal imaging data were mostly similar, with an additional pre-processing step for motion correction of the two-photon imaging data using the StackReg plug-in The ´venaz et al.in ImageJ. Structural parameters such as spine head width and spine protrusion length were quantified based on a standardized protocol. Briefly, if a protrusion extended for > 0.4 mm from the dendritic shaft, a dendritic spine was counted. The head width of a dendritic spine was measured as the width at the widest part of the head of the spine. The protrusion length of a dendritic spine referred to the distance from its root at the shaft to the tip of the head. The line segment tool in ImageJ was used to measure the distances. Change in spine density, spine head width and spine protrusion length across imaging sessions were shown as foldchange from the value measured on the first imaging session (day À3) for each dendritic segment. The spine formation rate was calculated as the number of dendritic spines newly formed between two consecutive imaging sessions divided by the total number of dendritic spines observed in the first imaging session. The spine elimination rate was calculated as the number of dendritic spines lost between two consecutive imaging sessions divided by the total number of dendritic spines observed in the first imaging session. To assess the long-term dynamics of the spine formation and elimination rates across imaging sessions, we calculated the difference from the baseline rate, which was the spine formation or elimination rate of the same dendritic segment before psilocybin and saline injection (i.e., from day À3 to day À1). To quantify the persistence of newly formed spines, we calculated the number of dendritic spines newly formed on day 1 that are still present on day 7 and day 34, and divided by the total number of newly formed dendritic spines on day 1.

STATISTICS

Sample sizes and statistical analyses for each experiment are listed in Table. Sample sizes were selected based on previous experiments reported in related publications. GraphPad Prism 8 and R were used for statistical analysis. In the figures, data are presented as the mean ± SEM per dendritic branch. For learned helplessness, we used a mixed effects model to test how proportion of escape failures (dependent variable) was impacted by fixed effects of treatment (saline versus ketamine versus psilocybin), test number (Test 1 versus Test 2), and sex (female versus male), including all second and higher-order interaction terms. Within-mouse variation was included as a random effects term. Post hoc pairedsamples t tests were used to analyze the change in Day 1 and Day 2 proportion of escape failures for the three treatment conditions, using Bonferroni correction for multiple comparisons. For ketanserin pretreatment experiments, a Kruskal-Wallis test (non-parametric oneway ANOVA) was used to test the difference in 10-minute head twitch responses across treatment groups (Saline + Psilocybin versus Ketanserin + Psilocybin versus Ketanserin + Saline), followed by Dunn's multiple comparisons test for post hoc pairwise comparisons. For in vivo two-photon imaging, dendritic spine scoring was performed while blind to treatment and time. Longitudinal measurements of dendrite structure were analyzed with mixed effects models for repeated-measures using the lme4 package in R. Linear mixed effects models were preferred to the commonly used repeated-measures analysis of variance (ANOVA) due to fewer assumptions being made about the underlying data (e.g., balanced sampling, compound symmetry). Separate mixed effects models were created for each of five dependent variables: fold-change in spine density, fold-change in average spine head width, fold-change in average spine protrusion length, spine formation rate, and spine elimination rate. Each model included fixed effects for treatment (psilocybin versus saline), sex (female versus male), and time (Day 1, 3, 5, and 7) as factors, including all second and higher-order interactions between terms. Importantly, variation within mouse and dendrite across days was accounted by including random effects terms for dendrites nested by mice. Visual inspection of residual plots revealed no deviations from homoscedasticity or normality. P-values were calculated by likelihood ratio tests of the full model with the effect in question against the model without the effect in question. Post hoc t-tests were used to contrast psilocybin and saline groups per day, with and without splitting the sample by sex, applying Bonferroni correction for multiple comparisons. Spine persistence from two-photon imaging was analyzed with separate repeated-measures ANOVAs for male and female mice, using fixed effects of treatment (psilocybin versus saline), time (day 7 versus day 34), and their interaction as independent predictors within dendrite. The same statistical analysis was applied to two-photon imaging data following ketanserin pretreatment, where the treatment groups were ketanserin + psilocybin versus ketanserin + saline. For electrophysiology data, blinding procedures involved one person inject psilocybin or saline, another person performing recording and measurements blind to treatment. Data were unblinded after all the measurements were completed. Two-way ANOVAs were used for mEPSC frequency and amplitude statistics. Treatment (psilocybin versus saline), sex (female versus male), and their interaction were included as independent predictors. Post hoc t-tests were used to contrast psilocybin and saline groups within sex, applying Bonferroni correction for multiple comparisons. For confocal imaging data, blinding procedures involved one person performing imaging, another person scrambling the image file names, and a third person performing dendritic structural measurements blind to sex, treatment, and brain region. Data were unblinded after all of the measurements were completed. For each brain region in the confocal dataset (Cg1/M2, PrL/IL, and M1), separate two-way ANOVAs were constructed for apical and basal dendrites using spine density, spine head width, or spine protrusion length as the dependent variable. Treatment (psilocybin versus saline), sex (female versus male), and their interaction were included as independent predictors. Post hoc t-tests were used to contrast psilocybin and saline groups within sex, applying Bonferroni correction for multiple comparisons.