The polypharmacology of psychedelics reveals multiple targets for potential therapeutics

Classical psychedelics (tryptamines, phenethylamines, lysergamides) are known to produce profound subjective effects and have renewed interest for therapeutic use in several neuropsychiatric conditions. Earlier work has established 5-HT2A receptor (5-HT2AR) activation as central to psychedelic effects, but other targets including various serotonin subtypes, dopamine receptors, trace amine receptors, σ receptors and even TrkB kinase have been proposed to contribute to either therapeutic or side-effect profiles. The precise molecular pharmacology across the broader chemotype diversity of psychedelics therefore remains incompletely defined. Jain and colleagues set out to characterise, in an unbiased and systematic way, the polypharmacology of a representative library of psychedelics. They selected 41 compounds spanning tryptamines, phenethylamines and lysergamides and profiled them against a large GPCR panel (318 non-sensory GPCRs), performed radioligand and functional characterisation across serotonin, dopamine and adrenergic receptors using TRUPATH and BRET platforms, screened LSD across >450 human kinases, and determined an active-state cryo-EM structure of LSD bound to the D2 dopamine receptor. The stated aim was to reveal the receptor and signalling pathways engaged by psychedelics and to clarify candidate mechanisms for both therapeutic actions and safety liabilities.

Jain and colleagues began by compiling 255 molecules annotated as having psychedelic-like effects from public databases then used chemical fingerprinting (Morgan fingerprints) and pairwise Tanimoto coefficients to choose 31 representatives that spanned the chemical space; 10 additional analogues were added to make a final set of 41 compounds. The similarity ensemble approach (SEA) was applied to predict targets across a CHEMBL-derived library of 6,679 protein targets, filtering predicted interactions for p < 1e-10 and potency ≤1 μM. Primary experimental screening used the PRESTO-Tango β-arrestin recruitment platform (the ‘‘GPCRome’’) in HTLA cells to test the 41 compounds at single concentrations (mainly 10 μM, with 3 or 1 μM for some compounds) against 318 GPCRs. Hits (fold-over basal ≥2.5) underwent concentration–response confirmation. Targets with >50% inhibition in primary radioligand binding were followed up in full competition binding assays to generate pKi values; the authors used pKi > 5 as a relevant threshold. Functional signalling was characterised using TRUPATH biosensors and BRET1/2 assays to quantify receptor coupling to multiple G proteins (including Gαq, Gα11, Gα15, Gαz, Gαi1) and β-arrestin recruitment (β-Arr1, β-Arr2). Calcium-flux (FLIPR) assays were applied to 5-HT2A, 5-HT2B and 5-HT2C to assess canonical Gαq-mediated Ca2+ mobilization, and a cAMP GloSensor assay was used for β3-adrenergic readouts. Transduction coefficients (Δlog10(τ/KA)) were calculated by fitting concentration–response data to the Black and Leff operational model; ligand bias factors were derived according to Kenakin and Christopoulos by comparing pathways to reference ligands. To test kinase interactions, LSD was submitted to a KINOMEscan platform covering >450 human kinases (primary screen at 1 μM with follow-up concentration–dependent counterscreens). TrkB-specific effects were evaluated in both a cell-free TrkB ADP-Glo assay and a TrkB AP1-Luc reporter assay in transfected HEK293T cells, including co-application experiments with BDNF. Kinetic ligand-binding experiments used custom [3H]-lisuride to measure association (Kon) and dissociation (koff) rates at 5-HT2AR. For structural studies, the D2 receptor was expressed as a D2R–miniGαoA complex, purified and subjected to cryo-EM with LSD present; model building and validation followed standard refinement and MD simulation pipelines to examine ligand conformations and dynamics. Statistical analyses included nonlinear regression for concentration–response fits, ANOVA with Dunnett corrections for some comparisons, and Pearson correlations between in vitro transduction coefficients and available in vivo head-twitch response (HTR) ED50s or mean recreational human doses.

SEA predictions indicated substantial pharmacological promiscuity among prototypical psychedelics, suggesting engagement of multiple targets beyond 5-HT2AR. Experimental profiling of 41 compounds across the 318-GPCR panel confirmed broad activity concentrated in serotonin, dopamine and α-adrenergic receptor families. Radioligand competition binding and secondary concentration–response assays showed that most compounds bound many 5-HT receptor subtypes; a number of ligands also showed measurable affinity for dopamine and adrenergic receptors and a few for σ, opioid, melatonin, histamine and other GPCRs. Lysergamides were the most promiscuous class (mean ≈21.3 targets per ligand) followed by tryptamines (mean ≈13.8) and phenethylamines (mean ≈12.0), and lysergamides exhibited modestly higher mean affinity (mean pKi ≈6.97) than the other classes. Functional TRUPATH and BRET assays showed that every tested psychedelic activated 5-HT2A, 5-HT2B and 5-HT2C receptors, with generally greater potency and efficacy at 5-HT2A and 5-HT2B. Arrestin recruitment to 5-HT2AR (β-Arr1 and β-Arr2) was observed for most compounds, though a subset (e.g., allylmescaline, escaline, isoproscaline, BL-3912A, CPM, 5-MAPDB, MDMA, methylone, lisuride) did not recruit β-Arr1, and 5-MAPDB, MDMA and methylone failed to recruit β-Arr2. Lysergamides showed strong activation of dopamine receptors (D2, D3, D4), whereas most tryptamines and phenethylamines activated D2–D4 but were generally inert at D1 and D5. Adrenergic receptor engagement varied by chemotype: several tryptamines activated α1A with specific G-protein pathways, while lysergamides activated multiple α1 subtypes and some β subtypes. When comparing hallucinogenic versus non-hallucinogenic analogues, hallucinogens engaged a larger number of receptor–transducer pathways and had higher relative transduction coefficients at 5-HT2AR Gαq, β-Arr1 and β-Arr2 pathways. Non-hallucinogenic compounds such as lisuride and BL-3912A activated fewer receptor–G-protein pathways. Bias analysis showed that most psychedelics were Gαq-biased at 5-HT2AR relative to β-Arr2, with some exceptions (e.g., 5-MeO-DALT, 5-MeO-DiPT, 5-MeO-EiPT were β-Arr2-biased; 4-AcO-MALT, N,N-DMT, psilocin and lisuride were relatively Gαz-biased). All tested psychedelics activated 5-HT2BR Gαq (and often Gα11 and Gα15), identifying 5-HT2BR agonism as a universal feature in this library and a potential cardiac safety liability. Connecting in vitro signalling to in vivo measures, the authors correlated pathway-specific transduction coefficients with mouse head-twitch response (HTR) ED50s (available for 12 compounds) and with mean human recreational doses. The strongest correlations were with 5-HT2A family pathways: 5-HT2AR Gαq showed r = 0.97 (p = 3.2e-7) versus HTR ED50, and 5-HT2AR Gαq/Gα11/Gα15 had correlations ≈0.96 versus recreational dose. Across the 100 pathways examined, the highest Pearson R was for 5-HT2AR. KINOMEscan profiling of LSD at 1 μM revealed only weak primary hits (BTK, DYRK1B, LKB1, PRKCE, RAF1, TGFBR1) that did not validate in concentration–response counterscreens; no potent interactions with tested kinases including TrkB were observed. Complementary TrkB reporter and cell-free kinase assays showed no agonist or positive allosteric modulation by LSD or psilocin on BDNF–TrkB signalling in the employed assays. Kinetic binding experiments showed that lisuride, a non-hallucinogen, has a slow dissociation rate at 5-HT2AR (koff ≈0.006 min−1; t1/2 ≈125 min at 37 °C) comparable to LSD (koff ≈0.005 min−1; t1/2 ≈221 min), suggesting ligand residence time alone does not explain hallucinogenic potential. Using a 5-HT2A–κOR chimeric nanobody biosensor, time-dependent conformational data were consistent with these slow kinetics. Structurally, a 2.3 Å cryo-EM structure of D2R bound to LSD and miniGαoA placed LSD in the orthosteric pocket with the diethylamide in a trans conformation; molecular dynamics supported this conformation. Comparison with D2R–bromocriptine and 5-HT2AR–LSD structures revealed conserved interactions (salt bridge to the conserved aspartate and hydrogen bonding to a conserved serine) and differences in hydrophobic contacts (e.g., F1103.28 in D2R versus W1513.28 in 5-HT2AR) that may account for higher potency/efficacy at 5-HT2AR. Mutational perturbation of D2R orthosteric pocket residues (D114A, F110A, V115A, W386A, Y416A) altered LSD-mediated activation as measured by D2R–Gαi1 BRET assays.

Jain and colleagues interpret their results as demonstrating extensive polypharmacology across classical psychedelics: the molecules profiled interact potently with a range of biogenic amine GPCRs, most prominently multiple 5-HT receptor subtypes, but also dopamine and adrenergic receptors. They emphasise that 5-HT2AR activation across multiple transducer pathways (particularly Gαq, Gα11, Gα15 and β-arrestins) correlates strongly with the canonical mouse behavioural proxy (HTR) and with reported recreational doses, supporting a central role for 5-HT2AR signalling in psychedelic-like actions. At the same time, the dataset shows that other receptors and pathways may contribute to pharmacology and thus to therapeutic or adverse effects. The authors found no supporting evidence in their assays for direct, high-affinity interactions between LSD and human kinases including TrkB, nor for LSD or psilocin potentiating BDNF–TrkB signalling in the reporter systems used. They therefore argue against a simple model in which direct TrkB binding by psychedelics explains their neuroplasticity effects. A major safety-related conclusion is that every psychedelic tested activated 5-HT2B receptor pathways; because 5-HT2BR agonism is linked to valvular heart disease, the authors note that subtype-selectivity screening (and regulatory scrutiny) is necessary, particularly for chronic or repeated dosing regimens. The discussion acknowledges key limitations the authors raise themselves: all functional pharmacology was measured in engineered, reconstituted systems (BRET/TRUPATH biosensors, heterologously expressed GPCRs) in non-neuronal cell lines, which may not capture the organisation and signalling complexity of native neural tissue. Consequently, in vitro bias, efficacy and pathway engagement may differ in vivo. The authors present their comprehensive dataset as a resource to guide candidate selection, mechanistic studies, and de-risking (predicting potential adverse events) as psychedelic-based therapeutics progress, and they encourage further work to map which of the many signalling outcomes are causally related to therapeutic benefit versus side effects.