An analog of psychedelics restores functional neural circuits disrupted by unpredictable stress

Stress caused by unpredictable adverse events can produce widespread synaptic, circuit, and behavioural disturbances and is implicated in many psychiatric disorders. Previous rodent work has shown region-specific dendritic atrophy and spine loss (for example in hippocampus, medial prefrontal and somatosensory cortices), disruption of excitation–inhibition balance, and resultant impairments in anxiety regulation, sensory processing, and cognitive flexibility. Although classical psychedelics (LSD, psilocybin, DMT and related compounds) have returned to clinical research because of promising efficacy in depression, anxiety and addiction, their hallucinogenic properties limit therapeutic use. Tabernanthalog (TBG), a synthetic analogue of 5-MeO-DMT, was reported to have antidepressant and anti-addictive potential without producing the mouse head-twitch response, but its ability to reverse stress-induced structural and functional brain changes was not known. Lu and colleagues set out to test whether a single post-stress dose of TBG could reverse behavioural deficits produced by an unpredictable mild stress (UMS) protocol in mice, and to determine the underlying neural circuit mechanisms. Specifically, the study examined anxiety-like behaviour, sensory discrimination and cognitive flexibility, dendritic spine dynamics on cortical pyramidal neurons, mesoscopic and cellular calcium activity in barrel cortex during whisking and texture discrimination, and intrinsic excitability of parvalbumin-expressing inhibitory interneurons (PV+ INs). The goal was to link behavioural rescue with synaptic and network-level changes induced by TBG in the stressed brain.

Experimental animals were Thy1-GFP-M and C57BL/6J mice (both sexes, aged 1–2 months) obtained from Jackson Laboratory. Mice were group-housed on a 12 h light/dark cycle, randomly assigned to experimental groups, and experiments were performed under institutional animal care approvals. The stress model was a 7-day unpredictable mild stress (UMS) protocol; details of the daily stressors were reported in a supplementary table (not reproduced here). Behavioural assays included the elevated plus maze (EPM) to assess anxiety, a whisker-dependent texture discrimination (WTD) test conducted both in freely moving and head-fixed mice (using the Neurotar mobile home cage), and a four-choice odor discrimination and reversal task to probe learning and cognitive flexibility. Training and shaping details, criteria for task inclusion, and the discrimination/reversal session rules were specified in the Methods. Behavioural tracking used video systems with DeepLabCut for pose estimation and custom analysis scripts. Drug treatment was intraperitoneal administration of TBG (synthesised in the Olson lab) at 10 mg/kg; fluoxetine (10 mg/kg) and saline were used as comparators in some experiments. The 10 mg/kg dose was chosen based on prior pharmacokinetic data as likely to produce brain concentrations sufficient to engage 5-HT2 receptors. For structural and functional imaging, the investigators performed cranial window implantation and, where appropriate, AAV2/1-Syn-GCaMP6f injections to express GCaMP6f in cortical neurons. In vivo two-photon (2P) imaging of dendritic spines used thy1-GFP-M mice and repeated imaging of identified apical dendrites, with spine formation/elimination quantified from image stacks. Wide-field mesoscopic calcium imaging (through the cranial window) recorded whole-field dF/F0 at 10 fps while simultaneously filming whisker movements; whisking episodes were detected by optic-flow analysis of the whisker-pad video. Cellular-resolution 2P calcium imaging of L2/3 neurons was performed at 30 fps (down-sampled to 10 fps for analysis); ROIs were manually delineated and dF/F0 traces processed to detect transients and synchronous population events. Receiver operating characteristic (ROC) analysis was used to classify neurons as novel texture-selective (NTS), familiar texture-selective (FTS), or non-selective based on pre-contact vs contact activity, with permutation testing (1000 shuffles) to assess significance (p < 0.05 threshold). Acute slice whole-cell patch-clamp recordings targeted PV+ interneurons in S1BF L2/3 to quantify resting membrane potential, input resistance, capacitance, action potential threshold, rheobase, and firing in response to current steps. For spine imaging and electrophysiology, established image acquisition and analysis pipelines (ImageJ, custom Matlab/Python scripts) were used. Analysts were blinded to experimental groups. Statistical testing used parametric tests when assumptions were met (two-sided unpaired t-test, one-way ANOVA with Tukey's post hoc) or non-parametric alternatives when not (Wilcoxon signed-rank, Kruskal–Wallis with Dunn's post hoc); data are reported as mean ± s.e.m. The extracted text does not always state group sample sizes for each experiment in the body of the Methods.

Behavioural outcomes: After 7-day UMS, mice showed increased anxiety on the EPM (less time in open arms) despite similar locomotor distance to controls. UMS impaired tactile novelty preference in the WTD task and reduced cognitive flexibility in the four-choice odor reversal task (longer to reach reversal criterion), while initial discrimination learning was intact. A single post-stress injection of TBG (10 mg/kg) administered immediately after UMS rescued these deficits: anxiety levels were reduced on the EPM, preference for novel texture in WTD was restored, and reversal learning performance returned to control levels. A single dose of fluoxetine (10 mg/kg) or saline did not rescue the behavioural impairments according to the extracted text. Dendritic spine dynamics: In vivo 2P imaging of apical dendrites on L5 pyramidal neurons revealed that 7-day UMS increased spine elimination without changing formation rates. TBG given after stress nearly doubled spine formation within one day but did not alter spine elimination. A notable fraction of new spines formed within 2 μm of previously lost spines, paralleling observations made with ketamine. The authors report that TBG-induced spinogenesis compensated for approximately 20% of the spine loss incurred during UMS, about five times the amount observed during spontaneous recovery. Morphological categorisation showed similar distributions of new spine types (mushroom, stubby, thin) after TBG compared with UMS animals; filopodia dynamics were unchanged. The extracted text indicates a survival rate of newly formed spines over 12 days that was slightly higher than previously reported control values, but the exact percentage is not clearly reported in the extraction. Mesoscopic and cellular calcium activity: Wide-field calcium imaging across S1BF showed that whole-field Ca activity (Ca_WF) correlated with whisking in control mice; UMS reduced this correlation and TBG restored it. Quantifying whisking-modulation as the post-onset versus pre-onset Ca_WF difference, UMS decreased whisking modulation and TBG completely rescued it; response latency to whisking was unchanged across conditions. At cellular resolution, 2P calcium imaging of L2/3 neurons revealed that UMS increased baseline neuronal activity manifested as larger Ca transient size (sum dF/F0 per transient) without changing transient frequency; TBG restored transient size to control levels. Network synchrony measures showed that UMS increased the duration of synchronous Ca events, the proportion of time in synchrony, and the fraction of neurons participating per event; TBG normalised these metrics. The authors interpret these changes as an increase in baseline ‘‘noise’’ after UMS that TBG counteracts to restore signal-to-noise for sensory input. PV+ interneuron electrophysiology: Patch-clamp recordings from L2/3 PV+ interneurons showed that UMS produced a more hyperpolarised resting membrane potential and reduced input resistance; membrane capacitance, AP threshold and rheobase were reported as unaffected by UMS in the extracted text. TBG restored resting potential and input resistance to control levels and additionally decreased rheobase while restoring AP firing in response to current injections to control-like levels. The investigators conclude that TBG increases intrinsic excitability of PV+ INs in the stressed brain, which may contribute to changes in excitatory neuron baseline activity and synchrony. Texture-selective neuronal responses during behaviour: Using a head-fixed WTD paradigm with concurrent 2P imaging, the study found that in control mice a greater proportion of neurons responded to novel versus familiar textures; specifically, about 32% of neurons were classified as novel texture-selective (NTS) and 6% as familiar texture-selective (FTS). After UMS, the fraction of NTS neurons fell to 2% (FTS ~3%), indicating loss of novelty-specific responses. Post-stress TBG partially restored selectivity: NTS rose to ~18% and FTS to ~11%. Overall, the proportion of neurons responding to the novel texture (NTS plus non-selective responders) was dramatically reduced by UMS and rescued by TBG. The result that the first presentation of a texture elicits responses in many neurons but a second presentation elicits fewer remained present across groups, indicating that basic familiarity-related response decrement persisted despite UMS.

Lu and colleagues interpret their findings to mean that a single post-stress dose of TBG (10 mg/kg) can rapidly reverse multiple consequences of unpredictable mild stress at structural, cellular, network and behavioural levels. The study extends prior reports of TBG's pro-plasticity effects in unstressed animals to a stress-recovery context and suggests a broader therapeutic window for TBG than previously appreciated, given rescue one day after dosing at a lower dose than used in some earlier work. Mechanistically, the authors link TBG-induced spinogenesis on cortical pyramidal neurons to the restoration of circuit function and behaviour: new spines provide the substrate for synapse formation and experience-dependent reorganisation, which may improve the brain's capacity to respond adaptively following stress. They draw parallels with ketamine, which also promotes rapid spine formation and behavioural rescue albeit via different molecular targets. At the network level, UMS elevated baseline L2/3 activity and synchrony while reducing whiskingmodulated and novelty-dependent responses, thereby lowering signal-to-noise; TBG reversed these network alterations. Restored intrinsic excitability of PV+ interneurons after TBG is proposed as one contributor to normalising excitatory neuron activity and synchrony, given PV+ INs' role in feedforward inhibition and shaping sparse firing in cortex. On pharmacology, the authors note that TBG shows high selectivity for 5-HT2, 5-HT1B and α2A adrenergic receptors with additional affinity for SERT and MAO-A in prior screening. They point to evidence that the 5-HT2A antagonist ketanserin blocks TBG's plasticity and antidepressant effects, supporting a role for 5-HT2 receptors, but acknowledge that other targets cannot be excluded and that the precise molecular mediators require further clarification. Although TBG does not evoke rodent hallucinogen-typical behaviours such as the head-twitch, the authors caution that only human studies can confirm whether it is non-hallucinogenic in people. Limitations and implications are acknowledged: the extracted text indicates uncertainty about the exact survival fraction of new spines from the extraction and reiterates that pharmacological targets and the functional consequences of increased spine survival remain to be elucidated. The study team suggests that psychoplastogenic analogues like TBG could have advantages over classical psychedelics (for example, potential for non-hallucinogenic formulations and take-home dosing), but that clinical testing will be required to establish safety, subjective effects and therapeutic potential in humans.