Ch. III — Receptor biology and signaling bias

Chapter III of Post-2010 Psychedelics: An Expert-Panel Review. For the executive summary and full table of contents, start there.

This chapter develops the receptor-level pharmacology that underwrites the rest of the review: why 5-HT2A is the necessary target, what secondary receptors contribute, the unresolved question of whether biased signaling explains the hallucinogenic/non-hallucinogenic split, what the post-2020 cryo-EM structures have actually shown, and where the “non-hallucinogenic 5-HT2A agonist” frontier now stands. Systems-level and circuit consequences of receptor engagement are taken up in Chapter IV.

3.1 5-HT2A as the necessary target

The proposition that 5-HT2A receptor activation is necessary for the classical psychedelic experience is unusually well-supported, resting on three converging lines of evidence.

Human pharmacological blockade. Vollenweider, Vollenweider-Scherpenhuyzen, Bäbler, Vogel and Hell (1998, NeuroReport) administered psilocybin (250 µg/kg) to healthy volunteers with or without ketanserin (5-HT2A/2C antagonist; 5-HT2A-selective), risperidone (5-HT2A/D2 antagonist), or the D2 antagonist haloperidol. The psilocybin-induced psychosis-like syndrome was attenuated dose-dependently by ketanserin (40 mg ketanserin abolished subjective effects of ~20 mg psilocybin) and by risperidone, but not by haloperidol — the first demonstration in humans that classical-psychedelic action depends on 5-HT2A and is independent of dopamine.¹ Preller et al. (2018, eLife) extended this to LSD: ketanserin fully abolishes the subjective and resting-state fMRI connectivity signatures of 100 µg LSD, and LSD’s whole-brain connectivity-change pattern co-locates with cortical 5-HT2A gene expression.² Larsen et al. (2025) used [^11C]-Cimbi-36 PET to quantify dose-proportional ketanserin occupancy at 5-HT2A in human cortex.³

Animal behavioral correlates. Glennon and colleagues (1984) established the rat drug-discrimination paradigm: in rats trained to discriminate DOM from saline, ED50 for stimulus generalisation across 22 hallucinogens correlated with 5-HT2 binding affinity at r ≈ 0.94, with the same correlation against human hallucinogenic potency.⁴ The mouse head-twitch response (HTR) assay, refined by Halberstadt and colleagues, correlates with human hallucinogenic potency at r ≈ 0.94; HTR is abolished by selective 5-HT2A antagonists (M100907, ketanserin) and by 5-HT2A knockout, and is preserved by 5-HT2C knockout.⁵

Efficacy thresholds. Lisuride, an ergoline with high 5-HT2A binding affinity, is non-hallucinogenic in humans. Recent analyses (Wallach et al. 2023 Nat Commun; Ippolito et al. 2025 Br J Pharmacol) converge on Gq-signaling efficacy — not ligand bias direction — as the cleanest molecular discriminator between psychedelic and non-psychedelic 5-HT2A agonists.⁶⁷ LSD and psilocin are partial agonists in most assays (Emax ~30–35% in the Wallach panel) but clear a threshold that lisuride does not.

The cumulative case: 5-HT2A is necessary, but necessity is not sufficiency. A 5-HT2A agonist can be non-hallucinogenic if its efficacy fails to engage the downstream pathway producing subjective effects — the setup for §3.3 and §3.5.

3.2 Secondary receptor targets

Classical psychedelics are promiscuous; “5-HT2A agonist” is shorthand. Clinically or mechanistically relevant secondary targets:

• 5-HT2C. Substantial affinity across the class; contributes to autonomic effects (mydriasis, mild hyperthermia). 5-HT2C is structurally homologous to 5-HT2A and binding-pocket cross-reactivity is the rule.
• 5-HT1A. 5-MeO-DMT has unusually high 5-HT1A affinity — the 5-HT1A/5-HT2A ratio is ~1:10 for 5-MeO-DMT versus ~1:100 for psilocin and N,N-DMT.⁸ The 5-HT1A component is widely invoked to account for 5-MeO-DMT’s qualitatively distinct ego-dissolving, non-visual subjective profile.
• 5-HT2B. All five classical psychedelics bind 5-HT2B with affinity equal to or greater than 5-HT2A; all are partial 5-HT2B agonists (mescaline excepted). 5-HT2B agonism on cardiac valvular interstitial cells is the canonical valvulopathy mechanism (fenfluramine, pergolide). Tagen et al. (2023, J Psychopharmacol) concluded that microdose safety margins are larger than for established valvulopathogens but non-zero; the risk is theoretically real and uncharacterised by adequately-powered prospective studies.⁹ (Developed in Ch IX.)
• D2. LSD has notable D2 affinity, comparable to its 5-HT2A affinity in some assays. Marona-Lewicka & Nichols’ rodent data and Preller et al. (2017) suggest a biphasic LSD effect with a later dopamine-mediated phase. Lisuride (high D2 agonism, non-hallucinogenic) complicates the interpretation: D2 engagement alone does not produce a psychedelic state.
• TAAR1, sigma-1, others. DMT’s sigma-1 affinity is low-µM, near/above bolus brain concentrations; mechanistic role uncharted. 5-HT1B/5/6/7, α1/α2-adrenergic, and weak histaminic engagement is documented across the class with mostly speculative functional contributions.

Differences between psilocin, LSD, and 5-MeO-DMT subjective profiles are reasonably attributed in part to these secondary targets rather than to 5-HT2A engagement alone.

3.3 Biased agonism: Gq versus β-arrestin

The central pharmacological puzzle of the past decade is whether the hallucinogenic property of 5-HT2A agonists tracks a specific signaling pathway, and if so, which one.

5-HT2A couples to multiple intracellular transducers: canonically Gαq → PLC-β → IP3/DAG → Ca2+/PKC, and, in parallel, GRK-phosphorylated receptor recruits β-arrestins 1/2 for desensitisation and (debated) separate signaling. Ligand-biased agonism — different ligands favouring different transducers at the same receptor — opened the therapeutic possibility that hallucination might track one signaling axis while antidepressant/plasticity effects track another. González-Maeso et al.’s 2003–2008 work on cortical 5-HT2A signaling provided early framing.¹⁰

Three propositions complicate the simple bias story:

1. Mouse HTR is Gq/PLC-dependent. Selective 5-HT2A Gq disruption ablates HTR; β-arrestin-2 knockout does not eliminate it (and in some assays enhances it). The behavioral surrogate for hallucination tracks Gq, not arrestin.⁵
2. β-arrestin-biased agonists are non-hallucinogenic in HTR. Roth and colleagues’ IHCH-7086 (β-arrestin-biased 5-HT2A partial agonist from the lumateperone scaffold) is essentially silent on Gq, recruits β-arrestin, produces no HTR in mice, and shows antidepressant-like behavioral activity in chronic stress paradigms.¹¹
3. But hallucinogenicity may track Gq efficacy threshold, not bias per se. Wallach et al. (2023, Nat Commun) and Ippolito et al. (2025, Br J Pharmacol) report that classical psychedelics cluster at higher 5-HT2A Gq Emax than non-psychedelic agonists (lisuride, ergotamine: Emax < 70% in normalised assays). IHCH-7086 fits because its Gq efficacy is near-zero, not because β-arrestin recruitment cancels hallucinogenesis.⁶⁷

Current best synthesis: hallucinogenic activity at 5-HT2A is Gq-efficacy gated. Compounds below the threshold do not produce HTR or, by extrapolation, human hallucination. Whether the direction of bias matters beyond efficacy gating is the live empirical question. The framing also accommodates Kaplan et al.’s (2022, Nature) Gq-biased (R)-69/(R)-70 — Gq agonists with antidepressant-like activity but no HTR — if their in vivo brain exposure keeps effective Gq engagement below threshold.¹²

Caveat: HTR is a behavior, not the human subjective experience, and the HTR→human-hallucination extrapolation is empirically tight (r ≈ 0.94) but not perfect. Chapter VIII’s grey-market fluoroethyl lysergamides — for which no formal human pharmacology exists — exemplify the limits of extrapolating from HTR + receptor binding alone.

3.4 Cryo-EM structures of 5-HT2A bound to psychedelics

Structural biology of 5-HT2A advanced sharply between 2017 and 2025, moving from individual crystal structures to a comparative atlas.

Wacker et al. (2017, Cell) solved the X-ray crystal structure of LSD bound to 5-HT2B at 2.9 Å (PDB 5TVN) — the first high-resolution structure of a psychedelic-receptor complex. The defining insight was extracellular loop 2 (EL2) forming a “lid” over the orthosteric pocket and gating LSD’s slow on/off kinetics; EL2 engagement recurs across all subsequent psychedelic-receptor structures.¹³ Kim, Che, Panova, Roth and colleagues (2020, Cell) reported the first active-state cryo-EM structure of 5-HT2A bound to a hallucinogen (25CN-NBOH) in complex with mini-Gαq (PDB 6WHA), with X-ray structures of 5-HT2A bound to LSD and to methiothepin (inverse agonist).¹⁴

The comparative panel arrived with Cao, Yu, Wang and colleagues (2022, Science, Roth co-corresponding): cryo-EM structures of 5-HT2A bound to (1) serotonin, (2) psilocin, (3) LSD, (4) lisuride (non-hallucinogenic ergoline), (5) lumateperone, and (6) IHCH-7086 (engineered β-arrestin-biased), PDB codes 7WC4–7WC9.¹¹ The consequential observation: serotonin and psilocin display a second, non-canonical binding mode in which the indole shifts toward the TM5/TM6 interface and the ethylamine engages EL2 residues. This second mode is preferentially stabilised by ligands that recruit β-arrestin without activating Gq, and was exploited to design IHCH-7086 and IHCH-7113. (The Cao 2022 work is sometimes miscited as Nature; it is in Science.)

Gumpper, Jain, Kim and colleagues (2025, Nat Commun; Roth corresponding) reported seven cryo-EM structures spanning all major hallucinogenic and non-hallucinogenic chemotypes (tryptamine, ergoline, phenethylamine, NBOH), including the β-arrestin-biased RS130-180.¹⁵ The principal conclusion: EL2 engagement, the position of the W6.48 toggle switch, and TM6 cytoplasmic conformation vary systematically across compounds and correlate with G-protein vs arrestin coupling, but no single residue predicts bias from structure. Bias is conformational, not contact-based, and emerges from integrated TM-bundle rearrangement. Psychedelic 5-HT2A agonists stabilise a TM6-rotated-outward conformation accommodating Gαq, with rotation magnitude correlating with G-protein efficacy; lisuride stabilises an intermediate conformation that does not fully open the G-protein cavity — the structural underpinning of its low Gq Emax.

A complicating 2026 Nature paper (Xu, Wang, Yu, Shao and colleagues) reported that 5-HT2A-mediated non-canonical Gi signalling is essential for hallucinogenic effect, on the basis of five cryo-EM structures of 5-HT2A–Gi and 5-HT2A–Gq complexes bound to psychedelics or non-hallucinogenic analogues, in-vitro pharmacology, and in-vivo behavioural pharmacology in mice.¹⁶ The paper identifies a contact between non-hallucinogenic analogues and 5-HT2A that mediates signalling bias toward the non-hallucinogenic side. This complicates rather than overturns the Gq-efficacy-threshold story — Xu et al.’s data are most consistent with Gq + Gi co-efficacy distinguishing hallucinogenic from non-hallucinogenic 5-HT2A agonists — and the field has not yet integrated Gq/Gi/arrestin three-way bias into a single predictive framework. The Drewko, Habets and Brunt 2025 Molecular Psychiatry review (“Above the threshold, beyond the trip”) synthesises the post-2023 efficacy-threshold position as the current consensus, with the Xu Gi finding flagged as the next frontier.¹⁷

3.5 Non-hallucinogenic 5-HT2A agonists — the new frontier

The post-2020 development of compounds engaging 5-HT2A without hallucinogenic activity is the most consequential pharmacological development since Hofmann. Three classes by structural origin:

Structure-based β-arrestin-biased agonists. Cao et al.’s (2022, Science) campaign produced IHCH-7086 (pyridopyrroloquinoxaline from the lumateperone scaffold) and IHCH-7113 as the first rationally designed, structurally validated β-arrestin-biased 5-HT2A partial agonists. IHCH-7086: Ki = 12.6 nM, Emax ~13% (β-arrestin), undetectable Gq, antidepressant-like activity in mouse forced-swim and chronic stress without HTR.¹¹ Related analogues remain preclinical as of mid-2026.

Empirically discovered isotryptamines and ibogalogs (Olson lab).

• AAZ-A-154 (isotryptamine, identified via the PsychLight 5-HT2A biosensor): no HTR across doses, dendritic arbor complexity comparable to ketamine, antidepressant-like activity; ketanserin abolishes the plasticity effect, confirming 5-HT2A mediation.¹⁸ Licensed by Delix Therapeutics.
• Tabernanthalog (TBG; DLX-007). Cameron, Tombari, Olson and colleagues (2021, Nature) reported a water-soluble, non-hallucinogenic, non-cardiotoxic ibogaine analogue. TBG reduces alcohol/heroin seeking and produces antidepressant-like effects in rodents, with hERG potency 100-fold lower than ibogaine’s. Not yet in clinical trials.¹⁹
• DLX-001 (zalsupindole). Delix’s lead clinical candidate; isotryptamine; partial 5-HT2A agonist with low Gq Emax; 5-HT2B antagonist (addressing the §3.2 valvulopathy concern). Phase 1 in healthy volunteers (2–360 mg PO, reported 2025): no hallucinogenic effects, robust qEEG biomarkers, clean safety. FDA cleared a Phase 2 IND in October 2025 with an at-home administration design — a striking regulatory signal.²⁰

Enantiomer differentiation: R-(-)-LSD. MindMed’s MM402 (R-LSD enantiomer with reduced 5-HT2A affinity but retained activity at other monoamine targets) is in Phase 1 / early Phase 2 for autism-related social anxiety. Peer-reviewed preclinical data are sparse at mid-2026; the hypothesis is compelling but not structurally grounded to IHCH/DLX depth.

The open clinical question. Whether the therapeutic effects of classical psychedelics — TRD antidepressant action, anxiolysis, addiction-craving reduction — require the subjective psychedelic experience, or whether molecular plasticity-induction at cortical 5-HT2A is sufficient. The non-hallucinogenic 5-HT2A agonists are the experimental probe that, in principle, decides this. They produce psilocybin-comparable dendritic spine growth in rodents (Vargas 2023 Science; Shao 2021 Neuron; Ly 2018 Cell Rep) — but the dendritic-spine endpoint is not the clinical endpoint.²¹²²²³ Phase 2 readouts of DLX-001 and eventual translation of IHCH-series and TBG/AAZ-A-154 will be the actual data. Until then, both positions remain defensible: “plasticity is necessary and sufficient; subjective experience is epiphenomenal” (Olson framing) versus “subjective experience is mechanistically necessary; non-hallucinogenic agonists will not replicate psilocybin’s durability or effect size” (Yaden/Griffiths framing, developed in Ch IV). Receptor-level pharmacology cannot adjudicate.

3.6 What “hallucinogenic” maps to mechanistically — bridge to Chapter IV

To consolidate before Chapter IV’s systems-level treatment: at the receptor level, classical psychedelics produce hallucinogenic effects via 5-HT2A activation with Gq signaling efficacy above a threshold that non-hallucinogenic 5-HT2A agonists (lisuride; IHCH-7086; DLX-001; Kaplan’s Gq-biased but in-vivo-non-hallucinogenic compounds) fail to clear. The structural underpinning is a TM6 outward rotation and EL2 conformational shift that determine the geometry of the receptor’s intracellular face and therefore G-protein vs arrestin coupling efficiency. Mouse HTR predicts human hallucinogenicity well (r ≈ 0.94) but is not perfect, and the Gi/o signaling findings (Xu et al. 2026) are not yet integrated with the Gq/arrestin framework.

What non-hallucinogenic 5-HT2A engagement buys at the cellular level — dendritic spine growth, mTOR activation, BDNF upregulation — is developed in Chapter IV (the Vargas 2023 intracellular-5-HT2A model; entropic-brain/REBUS; post-acute connectivity changes). What it does not yet buy — durability and effect size comparable to subjectively-intense single-session psilocybin — is the open clinical question taken up in Chapter V. Receptor biology is necessary infrastructure for that discussion; it cannot pre-empt it.

References

← Ch. II · Overview · Ch. IV →