Psychedelic drug action at dendrites is gated by behavioral state and serotonin receptors

Psychedelics such as psilocybin are of interest because they can produce rapid and durable improvements in depressive symptoms, yet the mechanism by which a drug that is cleared from the body within hours can have effects lasting weeks or months remains uncertain. A prominent idea is that psychedelics promote long-lasting structural plasticity in cortical pyramidal neurons, including increased dendritic spine density, but the acute dendritic processes that might initiate this remodelling have not been well characterised. Previous work has also suggested that dendritic calcium signalling is central to synaptic plasticity, making it a plausible intermediate between immediate drug effects and later structural change. Savalia and colleagues set out to test how psilocybin affects dendritic calcium dynamics in vivo, how these acute effects relate to subsequent spine changes, and whether behavioural state and serotonin receptor subtype shape these responses. In particular, they asked whether psilocybin-evoked dendritic activity depends on 5-HT2A receptors, whether it is altered by quiet wakefulness versus locomotion, and whether acute dendritic calcium signalling predicts later structural plasticity under psilocybin as it does under control conditions.

The researchers studied layer 5 pyramidal tract neurons in the mouse medial frontal cortex using sparse genetic labelling. Adult C57BL/6J mice, and conditional knockout mice lacking either 5-HT2A receptors (Htr2a f/f) or 5-HT1A receptors (Htr1a f/f) in the imaged PT neurons, received cortical AAV expressing mRuby2 and GCaMP6f together with retrograde AAV-Cre injected into the pons to target PT cells. Mice were implanted with chronic glass windows and imaged longitudinally while head-fixed on a treadmill after habituation. Calcium imaging focused on apical tuft dendrites over three days around a single injection of psilocybin (1 mg/kg, i.p.) or saline. The investigators collected baseline imaging on day -1, pre-treatment imaging on day 0, then 90 minutes after injection on the same day, and additional imaging 24 hours later on day 1. Structural z-stacks were also collected on day -1 and day 1 to quantify dendritic spines on the same branches. In a separate pharmacokinetic experiment, brain psilocin concentrations were measured at multiple time points after psilocybin using liquid chromatography-tandem mass spectrometry. Behavioural state was inferred from treadmill speed, with rest defined as 0 cm/s and run as >0 cm/s. Calcium signals were motion-corrected, extracted from manually traced dendritic regions of interest, corrected for neuropil contamination, and converted to ΔF/F. Calcium events were inferred with the CASCADE spike-inference toolbox. Spine density was counted on matched dendritic segments using blinded structural analysis. The authors analysed the hierarchical data with linear mixed-effects models, including random effects for dendrites nested within cells and mice, and used Bonferroni-corrected post hoc tests. Pearson correlations were used to relate drug pharmacokinetics, dendritic activity, and spine changes.

Psilocybin produced a transient increase in spontaneous dendritic calcium activity in apical tuft branches of frontal cortical PT neurons. Across 174 dendritic branches from 28 neurons in 21 mice, the rate of calcium events rose acutely after psilocybin but returned to baseline by 24 hours. The treatment-by-time interaction was significant (P = 0.006), and the acute change was about 10% above baseline for psilocybin-treated branches (8.9±2.4%) versus a slight decrease under saline (-1.8±1.1%; Bonferroni-corrected P = 9 × 10^-5). Averaging within neurons gave similar results. The time course of dendritic activation closely tracked brain psilocin exposure. Brain psilocin rose rapidly after psilocybin, peaked at about 33 minutes, had an estimated elimination half-life of 56 minutes, and was undetectable by 24 hours. The peak brain concentration was reported as 0.16 ng/mg, corresponding to approximately 0.81 μM. The increase in dendritic calcium events showed a significant positive correlation with psilocin levels over time (P = 0.02). By contrast, the dendritic response lagged behind the previously reported head-twitch response, which peaks earlier. Cell type-specific deletion of 5-HT2A receptors abolished the acute psilocybin-evoked increase in dendritic calcium events. In Htr2a f/f PT neurons, psilocybin no longer altered acute dendritic calcium signalling (psilocybin: 3.1±1.6%; saline: 0.3±1.6%; Bonferroni-corrected P = 0.5). The same knockout was associated with reduced dendritic calcium signalling 24 hours later after psilocybin (psilocybin: -6.6±1.8%; saline: 1.9±2.5%; P = 0.014), producing a significant treatment-by-time interaction (P = 2 × 10^-4). In contrast, deletion of 5-HT1A receptors did not block the acute psilocybin effect; the increase remained present (psilocybin: 9.3±1.6%; saline: 1.2±1.8%; treatment-by-time P = 0.011; post hoc P = 0.004), and there was no 24-hour effect. A three-way interaction of treatment, time, and genotype was significant (P < 0.001), supporting the conclusion that the acute dendritic response depends critically on 5-HT2A receptors, not 5-HT1A receptors. Behavioural state strongly shaped the dendritic response. Psilocybin reduced locomotion in wild-type head-fixed mice overall, although this suppression was not seen in the receptor knockout lines. When calcium events were analysed separately during rest and running, psilocybin increased dendritic activity selectively during quiet wakefulness: during rest, the change was 10.5±2.5% versus -1.9±1.2% for saline, with a significant treatment-by-locomotion interaction (P = 0.019). During running, the difference was not detectable. Baseline dendritic activity did not differ between rest and locomotion before treatment, suggesting that the state effect was not simply due to pre-existing differences in activity. The state-dependent response was eliminated by 5-HT2A receptor deletion but preserved after 5-HT1A receptor deletion. The pattern was not observed 24 hours later. For structural plasticity, the investigators tracked 113 dendritic branches in 21 C57BL/6J mice with matched calcium and spine imaging for 105 branches. Contrary to the authors’ previous work, spine density decreased over time in both groups: by 6.9% after psilocybin and by 5.2% after saline. Thus, the expected psilocybin-induced spine gain was not observed in this experiment. When the researchers related acute dendritic calcium changes to later spine changes, saline-treated branches showed a significant positive association between greater calcium activity and subsequent spine increase (r = 0.34, P = 0.034). No such relationship was detected after psilocybin (r = -0.24, P = 0.16), and a mixed-effects model showed a significant interaction between treatment and calcium change (P = 0.008).

Savalia and colleagues interpret the findings as showing that psilocybin engages dendritic calcium signalling in frontal cortical PT neurons through a 5-HT2A receptor-dependent mechanism, and that this acute response is gated by behavioural state, appearing preferentially during quiet wakefulness. They argue that this provides a cellular correlate for the long-recognised influence of ‘set and setting’ on psychedelic effects. They also conclude that, although physiological spine formation is normally linked to stronger dendritic calcium activity, this coupling was not preserved after psilocybin, suggesting that acute dendritic signalling and longer-term structural plasticity diverge downstream of shared initial events. The authors place these results alongside earlier work showing that some acute pharmacological readouts, such as immediate early gene induction or glutamate efflux, do not necessarily predict later therapeutic outcomes. They also note that the state dependence of psilocybin’s neural effects is consistent with recent human findings and with clinical dosing practices that use a low-stimulation environment. The main limitation they acknowledge is that, unlike in their previous study, they did not observe psilocybin-induced dendritic spinogenesis. They suggest that this discrepancy may relate to differences in behavioural context, because the mice in the present study remained head-fixed for imaging rather than returning to the home cage. They say future experiments should manipulate behavioural context more systematically. They also caution that calcium transients can reflect multiple forms of dendritic electrical activity, not just one mechanism, and that GCaMP6f likely captured widespread dendritic electrogenesis associated with somatic burst firing in many cases. The authors state that further work will be needed to disentangle how psilocybin modulates these different dendritic processes and how they contribute to structural plasticity. Overall, they present the results as evidence that acute dendritic responses to psilocybin and longer-term structural changes may be at least partly dissociable.