Convergent neuroimmune signaling underlying rapid antidepressant response to ketamine and psychedelics

Rapid-acting antidepressants such as ketamine and psilocybin have changed the treatment landscape for treatment-resistant depression because they can produce clinical improvement within a day, but their effects can be short-lived and difficult to deliver at scale. Although these drugs act on different primary receptors, earlier research has suggested they share downstream biological effects, including changes in glutamate signalling, network activity, plasticity and inflammatory markers. The authors argue that the common molecular mechanisms behind these rapid responses remain incompletely understood, particularly in relation to neuroimmune signalling. Jones and colleagues set out to identify conserved biomarkers and pathways linked to rapid antidepressant response across ketamine and serotonergic psychedelics. They focused on ketamine, psilocybin, the ketamine metabolite (2 R,6 R)-hydroxynorketamine, and LSD, and asked whether immune-related signals seen in iPSC-derived neurons, CSF proteomics, blood transcriptomics, plasma cytokines and MEG would converge on shared mechanisms. The study was designed to find biomarkers that might help explain therapeutic response and point towards more durable or accessible treatments for treatment-resistant depression.

The study combined several datasets and experimental platforms. First, the researchers generated induced pluripotent stem cell-derived cortical neurons from peripheral blood mononuclear cells from five participants with treatment-resistant depression and fibroblasts from five healthy volunteers, all female, and treated the neurons for 24 hours with vehicle, ketamine, (2 R,6 R)-hydroxynorketamine, LSD or psilocybin. Bulk RNA sequencing was then performed, and differential expression was analysed using mixed-effects approaches with treatment as a fixed effect and donor-related factors as random effects. These neuronal data were compared descriptively with serial cerebrospinal fluid proteomics from nine healthy volunteers who had received intravenous ketamine in a previous study, with six CSF samples per participant collected over 24 hours. Overlapping differentially expressed genes and proteins were used for network analysis with STRING and Cytoscape to identify hub pathways. The extracted text states that the iPSC-CSF comparison was presented descriptively rather than as a formal statistical test because of the differences between datasets. The authors then performed post-hoc analyses of a prior double-blind, placebo-controlled, randomised crossover ketamine study in treatment-resistant depression and healthy volunteers. In this clinical cohort, participants received a single intravenous infusion of ketamine, 0.5 mg/kg, or saline placebo. Whole-blood RNA sequencing was available at baseline and four hours post-infusion for 16 participants with treatment-resistant depression and 11 healthy volunteers; plasma cytokines were measured at baseline, four hours, one day and three days in a larger subset; and resting-state MEG was acquired at baseline and 6–9 hours post-infusion in 30 participants with treatment-resistant depression and 25 healthy volunteers. The authors also used cell-type deconvolution methods, including xCell and BayesPrism, to estimate immune-cell composition, and fitted linear mixed-effects models for cytokine-response and cytokine-MEG associations, adjusting for relevant covariates such as treatment order, age, sex, BMI and baseline depression severity. Multiple testing correction was applied to proteomic, cytokine and MEG analyses, while exploratory transcriptomic contrasts used less stringent thresholds for hypothesis generation. No formal power calculation was performed because these were post-hoc analyses of completed trials.

Across iPSC-derived neurons, the four rapid-acting antidepressant drugs produced highly correlated transcriptional changes, with pairwise correlations reported between 0.7 and 0.88 and all adjusted p values below 2 × 10^-16. When these neuronal signatures were compared with serial CSF proteomics after ketamine, there was substantial overlap. Among iPSC differentially expressed genes with corresponding Olink-detected proteins, 46.5% for psilocybin, 44.4% for (2 R,6 R)-hydroxynorketamine, 40.3% for LSD and 33.3% for ketamine also appeared as differentially expressed proteins in CSF. The authors aggregated 103 proteins that were altered in CSF after ketamine and in iPSC neurons after at least one rapid-acting antidepressant. Network analysis suggested two modules, with a dominant immune-signalling hub centred on IL-15, MCP-1/CCL2, CXCL1 and several cell-adhesion molecules, and a smaller module involving insulin signalling, synaptic plasticity and folate metabolism. Pathway analysis showed strong enrichment for immune and inflammatory pathways. In whole-blood transcriptomics from the ketamine crossover study, 616 genes were differentially expressed between responders and non-responders at baseline, and 1,061 genes differed after ketamine; 84% of post-treatment differentially expressed genes were unique to the post-treatment contrast. At baseline, responders showed reduced IL-15 pathway activity and increased B-cell signalling relative to non-responders, alongside lower insulin-secretion signalling. After ketamine, these IL-15/B-cell pathway differences largely reversed in responders. A readthrough transcript involving ADORA2A, SPECC1L-ADORA2A, was the top post-ketamine transcript in responders versus non-responders, and the authors report that this finding was confirmed by alignment checks and in external postmortem brain data. The network also identified adenosine-related mediators NT5E, DPP4 and GPR37, although these were not shared by (2 R,6 R)-hydroxynorketamine in iPSC neurons. Cell-type deconvolution suggested that post-treatment samples from responders clustered together, driven mainly by natural killer T-cell proportions. B-cell subsets, including mature, memory, naïve and plasma B cells, were significantly reduced in responders after correction for multiple testing. Natural killer T-cell proportions increased after ketamine in both responders and non-responders, but did not differ clearly by responder status. These deconvolution findings were broadly supported by complete blood count correlations with lymphocyte estimates. In plasma, IL-7 emerged as a central biomarker. Baseline IL-7 levels were positively associated with MEG gamma power across the brain, with the strongest associations in treatment-resistant depression and particularly in the thalamus, insula and striatum. After ketamine, this relationship inverted: higher IL-7 was associated with reduced gamma power, especially in default mode network regions. IL-15 was mostly below the detection limit in plasma, but detectable samples showed increases after ketamine, consistent with the transcriptomic findings. Cytokine ratios linked to the IL-15/IL-7 axis predicted clinical outcome. Higher baseline MCP-1/IL-7 ratios predicted worse antidepressant response to ketamine versus placebo, while higher baseline IL-4/IFN-γ ratios predicted greater response. Similar associations were seen when ratios across the full study period were analysed. In the cytokine models, the core immune markers in the proposed framework were among the strongest predictors of antidepressant response or non-response. A sensitivity analysis for inflammatory comorbidity and anti-inflammatory medication use gave largely similar results.

The authors interpret their findings as evidence for a convergent neuroimmune signature across pharmacologically distinct rapid-acting antidepressants, including ketamine, (2 R,6 R)-hydroxynorketamine, psilocybin and LSD. They argue that the data support an immune-centred network in which IL-15, MCP-1/CCL2 and related adhesion molecules sit at the core of broader synaptic, insulin and folate signalling changes. In the clinical ketamine data, the baseline pattern of IL-15 suppression and B-cell activation in responders, together with its reversal after treatment, is presented as consistent with this framework. The authors also view the IL-7–gamma power relationship and the predictive cytokine ratios as evidence that peripheral neuroimmune markers are linked to brain network changes and treatment response. The paper places these findings alongside earlier research showing that ketamine and psilocybin affect glutamate, plasticity and network connectivity, and suggests that immune signalling may be a shared downstream mechanism despite different primary targets and subjective effects. The authors also discuss a possible adenosine-signalling contribution, but they describe this as more exploratory and note that some adenosine-related findings were compound-specific. Several limitations are acknowledged. The biomarker strategy was hypothesis-generating and not pre-specified or preregistered. The multimodal clinical measures were collected at different times after dosing, which may not have been the optimal window for cytokine-MEG relationships. Some transcriptomic analyses used uncorrected thresholds to preserve sensitivity, increasing the chance of false positives, although the authors say they attempted to reduce this risk by corroborating findings across datasets. The iPSC-CSF comparison was limited by small sample sizes, all-female iPSC donors, different cell sources for healthy volunteer and depression lines, and the fact that CSF proteomics came only from healthy volunteers. The defined proteomic panel also covered only part of the transcriptome. The authors therefore present the iPSC-CSF findings as candidate signals requiring replication in larger, source-matched cohorts. Despite these caveats, Jones and colleagues state that the study is the first to combine whole-blood transcriptomics, plasma cytokines and MEG within the same clinical rapid-acting antidepressant cohort, with corroboration from CSF proteomics and iPSC-derived neurons. They suggest that the approach provides a broader human framework for understanding rapid antidepressant mechanisms and for prioritising biomarkers and targets for future validation.

The authors conclude that ketamine and psychedelic rapid-acting antidepressants may share critical neuroimmune mechanisms, particularly involving restoration of IL-15/IL-7 balance, with downstream effects on B-cell homeostasis and neuronal activity. They also state that the adenosine-related signals offer additional, but still preliminary, evidence for a conserved antidepressant mechanism. Overall, they present the multimodal approach as a way to identify clinically relevant biomarkers and novel therapeutic targets for rapid antidepressant research.