Structural neural plasticity evoked by rapid-acting antidepressant interventions

Dendritic spines are the primary sites of excitatory input onto cortical pyramidal neurons, and their morphology correlates with the presence and strength of excitatory synapses and hence circuit wiring. Liao and colleagues note that although many psychoactive drugs change spine structure after chronic administration, recent longitudinal in vivo optical imaging in rodents has shown that single doses of certain compounds can produce rapid and long-lasting structural remodelling. Drugs highlighted for this property include ketamine and serotonergic psychedelics and analogues, several of which have either established or emerging clinical evidence as rapid-acting antidepressants. This Review sets out to synthesise preclinical work from roughly the past 15 years that tracks drug-evoked structural plasticity in vivo, to compare time courses and persistence across different psychoactive compounds, and to connect these findings to human evidence of synaptic deficits in depression. The authors aim to identify characteristics that distinguish putative rapid-acting antidepressant interventions, to discuss gaps in understanding of in vivo structural plasticity, and to consider translational opportunities for other interventions such as repetitive transcranial magnetic stimulation (TMS).

This paper is a narrative Review of human and preclinical literature rather than a new experimental study. The authors focused on longitudinal in vivo optical imaging experiments in mice that used laser-scanning two-photon-excited fluorescence microscopy to follow dendritic spines on cortical pyramidal neurons across days to weeks. They collated published studies that quantified spine turnover and density after administration of putative rapid-acting antidepressant drugs (notably ketamine, psilocybin and 5-MeO-DMT) and contrasted these with data from other psychoactive compounds (for example allopregnanolone, diazepam, zolpidem and cocaine). Where numerical comparisons were informative, the Review integrates values extracted from published figures and plots (see referenced Supplementary data) to place multiple studies on common axes. The authors describe the typical experimental targets (apical tufts of layer 5 pyramidal neurons, often in medial frontal cortex), note methodological heterogeneity across studies (imaged brain region, animal age, dosing regimen and number of administrations), and discuss technical constraints of in vivo imaging (spatial resolution, axial sampling, limits on measuring spine necks and small filopodia). Human evidence considered includes post-mortem electron microscopy, genomic/transcriptomic associations, structural MRI and in vivo PET imaging of synaptic vesicle glycoprotein 2A (SV2A), and the Review integrates these lines with the preclinical longitudinal data.

Human studies reviewed indicate convergent evidence for excitatory synaptic impairment in major depressive disorder (MDD). Post-mortem electron microscopy shows fewer spine synapses in layer 2/3 of dorsolateral prefrontal cortex in MDD, and transcriptomic and genomic analyses enrich for synapse-related gene sets and synaptic signalling categories. Structural MRI reports regionally reduced cortical thickness and grey matter, especially in hippocampus and subgenual prefrontal cortex, and magnetic resonance spectroscopy finds lower frontal glutamate and glutamine concentrations in individuals with MDD. PET imaging using SV2A radioligands detects diminished presynaptic signal in prefrontal, anterior cingulate and hippocampal regions that correlates with symptom severity; notably, ketamine administration increased SV2A signal in individuals with MDD who had low baseline SV2A, although PET cannot distinguish more synapses from stronger synapses. Preclinical longitudinal imaging studies in mice reveal drug-specific patterns of dendritic spine turnover, density and persistence. A single subanaesthetic (antidepressant) dose of ketamine, as well as single doses of psilocybin and 5-MeO-DMT, rapidly increased apical dendritic spine number density by roughly 12–20% in cortical pyramidal neurons. The kinetics of change are rapid: compounds with putative rapid antidepressant action elevated spine formation rates for about 1–3 days after single-dose administration, with ketamine-associated spine growth detectable at ~12 hours post-injection in some studies. Persistence differs across drugs: psilocybin-evoked new spines show higher survival (~50% at 7 days and ~35% at 34 days) whereas ketamine-evoked new spines display lower persistence (~20% survival at 15 days in one study). By contrast, some other compounds produce weaker or different effects: allopregnanolone and cocaine can raise spine density by ~5% but more gradually and typically after repeated dosing; diazepam (high-dose, chronic) robustly reduced spine density with effects lasting up to 2 months; zolpidem produced no detectable change in the studies cited. Spine size and number do not always change in concert. Psilocybin increased both spine number and head width acutely, but spine size tended to return to baseline while density remained elevated, whereas 5-MeO-DMT primarily increased spine number with minimal size changes. Dose and administration regimen strongly influence outcomes: ketamine at anaesthetic doses produced only transient filopodial changes without lasting architecture change, and higher sedative doses of diazepam produced greater spine loss than lower anxiolytic doses. Chronic stress models reduce spine density in medial frontal cortex primarily via increased spine elimination; ketamine can restore spines lost after chronic corticosterone and can prevent stress-induced elimination if given prophylactically, though prophylaxis may preclude further spinogenesis. Causal manipulation experiments provide the most direct link to behaviour: targeted disruption of ketamine-evoked new spines abolished the drug’s sustained benefit on motivated escape behaviour in mice while leaving the acute behavioural action intact, supporting a role for spine formation in long-term behavioural effects. Additional findings include methodological caveats and regional specificity: many two-photon studies imaged layer 5 pyramidal neuron apical tufts in medial frontal cortex, but other cortical regions and neuronal subtypes have been less studied. Preclinical evidence for TMS-induced structural remodelling is preliminary, and translation of clinically relevant TMS protocols to animal models is constrained by differences in brain geometry and stimulation scaling.

The authors interpret these converging lines of evidence to propose that structural neural plasticity—measured as changes in dendritic spine formation, elimination, density, size and persistence—offers a plausible neurobiological mechanism linking several rapid-acting antidepressant interventions to restoration of excitatory synaptic function. They emphasise that rapid onset of increased spine formation followed by sustained changes in spine density is a shared characteristic among several putative rapid-acting antidepressants, and that differences in persistence of new spines may map onto differences in clinical durability between treatments such as ketamine and psilocybin. At the same time, the Review stresses important uncertainties and limitations. Species and regional differences complicate translation from mouse imaging to human depression, and existing in vivo imaging methods miss some small or axially oriented protrusions and cannot fully resolve synaptic ultrastructure. Methodological heterogeneity across preclinical studies (brain region imaged, mouse age, dose and regimen) and the reliance on extracted values from disparate reports limit comparability. The authors note that spine number and spine size may reflect distinct plasticity processes, measurement of spine size can be imprecise in vivo, and PET SV2A imaging in humans cannot disambiguate synapse number from synapse strength. The possibility that new spines may not be uniformly beneficial is emphasised: the functional impact depends on presynaptic source, postsynaptic cell type and circuit context, and stress or pathology can differentially alter regions such as prefrontal cortex versus amygdala. Regarding implications, Liao and colleagues propose multiple translational opportunities: in vivo structural plasticity could serve as a drug-discovery assay to screen compounds with synaptogenic potential, and reverse translation could help elucidate mechanisms of non-pharmacological treatments such as TMS. They advocate for greater circuit and cell-type specificity in future studies (for example, dual-colour imaging and monosynaptic tracing) and for improved methods to measure synapse-level change in humans longitudinally, with SV2A PET and advanced imaging approaches highlighted as promising paths. Finally, the authors call for further causal experiments to determine which aspects of structural remodelling are necessary for therapeutic effects and for studies that delineate downstream mechanisms by which new or reactivated synapses influence behaviour.

Dendritic spines serve as a tractable proxy for excitatory synapses, and multiple strands of evidence indicate excitatory synaptic deficits in depression. In vivo longitudinal imaging studies in rodents demonstrate that certain rapid-acting antidepressant drugs, notably ketamine, psilocybin and related serotonergic agonists, can rapidly promote spine formation and in some cases yield persistent increases in spine density. Different structural metrics—formation rate, persistence, density and size—offer distinct biological insight. The authors conclude that studying structural neural plasticity provides valuable mechanistic information for antidepressant interventions and presents both opportunities and challenges for drug development and translational neuroscience.