Signaling snapshots of a serotonin receptor activated by the prototypical psychedelic LSD

Lysergic acid diethylamide (LSD) is a broadly acting psychedelic that binds many biogenic amine G-protein-coupled receptors (GPCRs), with primary psychedelic effects attributed to 5-HT2A (HTR2A). However, LSD also activates 5-HT2B (HTR2B), and chronic activation of HTR2B is implicated in drug-induced valvular heart disease. GPCRs of the 5-HT2 family signal through both Gq proteins and b-arrestins, and some ligands stabilise receptor conformations that preferentially engage one downstream pathway over another (biased signaling). Despite prior X-ray structures of LSD-bound HTR2A and HTR2B in transducer-free states, the molecular bases for differential engagement of G proteins versus arrestins by the same ligand remain incompletely understood. To address this gap, Cao and colleagues determined cryo-electron microscopy (cryo-EM) structures of LSD-bound HTR2B in three states: transducer-free, Gq-coupled, and b-arrestin-1-coupled. Their aim was to capture conformational snapshots that reveal how the same ligand (LSD) can drive distinct transducer-specific receptor conformations and to combine structural work with functional assays, mass spectrometry and molecular dynamics (MD) to link structural features to signalling outcomes.

The investigators used single-particle cryo-EM to determine structures of LSD-bound HTR2B in three different states. Recombinant receptor constructs were expressed in Sf9 insect cells and purified with an extracellular Fab (P2C2) as a fiducial marker. For the Gq complex, LSD-bound HTR2B was assembled with a miniGq heterotrimer and purified; for the arrestin complex, the team engineered an HTR2B–b-arrestin-1 fusion that included targeted truncations and stabilising mutations and co-expressed GRK2 to promote receptor phosphorylation. Purified complexes were vitrified and imaged on Titan Krios microscopes; cryoSPARC was used for particle classification and refinement. Final map resolutions reported were ~2.7 Å for the transducer-free receptor, ~2.9 Å for the HTR2B–Gq complex, and ~3.3 Å for the HTR2B–b-arrestin-1 complex. Construct design and functional validation were integral to the methods. The b-arrestin complex required multiple engineering steps: ICL3 and portions of the C-tail were truncated based on bioluminescence resonance energy transfer (BRET1) assays; an ionic-lock disrupting double mutation (K2475.68V + E3196.30L) increased LSD-stimulated b-arrestin-1 recruitment and was incorporated; a constitutively active b-arrestin-1 variant (R169E) and an scFv30 single-chain Fab were fused to stabilise the complex for purification. The transducer-free and Gq constructs used related but distinct truncations and fusion partners (BRIL) to aid expression and stability. Functional assays involved BRET1 recruitment measurements of miniGq and b-arrestin-1 in HEK293T cells transfected with RLuc8-tagged HTR2B and Venus-tagged transducer partners. Mutagenesis (alanine substitutions and other point changes) was employed to test the importance of residues identified in the structures. Mass spectrometry (Orbitrap Exploris 480) after phosphopeptide enrichment was used to identify and estimate phosphorylation probabilities of C-tail residues. Finally, extensive MD simulations (multiple independent trajectories, each ~2–3 ms aggregate per condition) were performed for five system conditions—transducer-free, Gq-bound, arrestin-bound and 'transducer removed' derivatives—to probe dynamic conformational sampling and the propensity of the transducer-free receptor to adopt transducer-compatible states. Model building used published templates and iterative refinement with Coot and Phenix.

Structure determination produced high-resolution snapshots of LSD-bound HTR2B in three functional states. Cryo-EM reconstructions yielded a ~2.7 Å map for transducer-free HTR2B, ~2.9 Å maps for receptor and Gq in the HTR2B–Gq complex, and a ~3.3 Å map for the HTR2B–b-arrestin-1 assembly. The transducer-free cryo-EM structure differed from earlier crystal structures: intracellular tips of TM5 and TM6 were unresolved and no crystallographic lipid was observed between TM5 and TM6, and extracellular ends of several helices showed 0.9–2.5 Å outward displacements, producing a slightly larger orthosteric pocket for LSD. In the HTR2B–Gq complex, the Gq a5 helix inserts into the receptor cytoplasmic cavity formed by TM2, TM3, TM5, TM6, TM7 and ICL2. Hydrophobic residues on Gq (L236, L240, L245 in CGN numbering) engage the TM3–TM5–TM6 interface, while polar contacts (for example between Gq Q237/E242/N244 and receptor residues E3196.30, N16434.54 and N3848.47) further stabilise the interface. ICL2 residue I16134.51 inserts into a hydrophobic groove on Gq, a recurring feature seen in other GPCR–G protein complexes. Compared with HTR2A, ICL2 in HTR2B shows an inward displacement that shifts the Gq aN helix inward upon coupling. The HTR2B–b-arrestin-1 structure revealed arrestin engagement at three main regions: the receptor core, ICL2 and the phosphorylated C‑tail. The finger loop of b-arrestin-1 adopts a helical conformation and penetrates the cytoplasmic cavity, making hydrophobic contacts via residues L68, L71, L73 and F75 and electrostatic contacts via R65 and D67 with receptor residues N16434.54 and N16734.57. Mutations of finger-loop residues (L71, L73, R65) significantly reduced LSD-stimulated b-arrestin-1 recruitment, underlining the finger loop’s functional importance. ICL2 forms a helix and inserts into an arrestin hydrophobic cleft; an I16134.51A mutation reduced Emax of arrestin recruitment by ~70% relative to wild type. Cryo-EM density and mass spectrometry indicated phosphorylation at C-tail residues S455 and S456, with S457 showing lower probability; alanine substitutions at S455, S456 or S457 each reduced arrestin recruitment Emax by ~30–50%. Comparative analysis showed that many orthosteric pocket residues adopt similar conformations across Gq- and arrestin-bound states, but L3627.35 adopts a different rotamer in the arrestin-coupled structure to make closer hydrophobic contact with LSD’s diethylamide, consistent with prior findings linking this contact to arrestin recruitment. On the intracellular side, both transducer-bound structures are in active-like conformations with TM6 outward displacement, but the arrestin-coupled state exhibits additional outward movement of intracellular TM5 and TM6 (1.7 Å and 2.5 Å respectively) relative to the Gq-coupled state, likely accommodating the bulkier arrestin finger loop. Key microswitches also rearrange during transducer coupling: the ‘‘toggle switch’’ W3376.48 moves downward, the PIF motif adopts an active configuration, the DRY motif side chain R1533.50 rotates differently when engaging Gq versus arrestin, and the NPxxY motif Y3807.53 flips into a new position. The polar core interactions involving D1002.50 and N3767.49/S3737.46 differ between states; a D2.50–N7.49 interaction strengthens in the Gq-coupled structure but is weaker in the arrestin-coupled state, suggesting this polar interaction favors G protein activation. A specific residue, N3848.47 in helix 8, shows distinct roles: it hydrogen-bonds with Gq N244H5.24 in the Gq complex, whereas in the arrestin complex it rotates and stacks with hydrophobic finger-loop residues. Functional assays corroborate this structural observation: an N3848.47A mutation reduced Gq recruitment Emax by ~40% while increasing LSD-stimulated arrestin recruitment Emax. Other functional findings include that truncation construct CT464 (residues 1–464) increased arrestin assay Emax by ~50% compared with wild type in the BRET assay, and the double mutation K2475.68V + E3196.30L markedly increased basal and LSD-stimulated arrestin recruitment to levels comparable with wild-type receptor stimulated by full agonist 5-HT. MD simulations across multiple conditions showed that the LSD-bound transducer-free receptor samples conformations overlapping with those observed in Gq- and arrestin-bound states. The intracellular tips of TM5 and TM6 are dynamically mobile in the transducer-free state and transiently open to allow transducer engagement. When Gq or arrestin were removed in simulation, the receptor tended back toward the transducer-free conformation, supporting a model in which transducers synergistically stabilise fully active receptor states. The conversion of ICL2 from loop to helix upon transducer coupling was recurrent in simulations and appears important to position I16134.51 for productive transducer interactions.

Cao and colleagues present the first set of structures for the same GPCR–ligand pair (HTR2B–LSD) in transducer-free, G protein–bound and arrestin-bound states, providing mechanistic snapshots of how a single ligand can support distinct downstream signalling engagements. The authors highlight several key observations: arrestin coupling is associated with a larger TM6 outward movement than Gq coupling at HTR2B, structural rearrangements in conserved microswitch motifs (PIF, DRY, NPxxY and the polar core) differ between transducer states, and specific residues such as N3848.47 and L3627.35 contribute differentially to Gq versus arrestin engagement. Functional mutagenesis, BRET recruitment assays and mass spectrometry for C-tail phosphorylation are used to validate structural inferences, for example showing that C-tail phosphosites S455/S456 and ICL2 residue I16134.51 are important for arrestin recruitment, while N3848.47 promotes Gq recruitment. The investigators position these findings in the context of biased agonism. They suggest that the observed structural differences reveal receptor plasticity that can be exploited in structure-based drug design to favour one transducer over another. As an illustrative idea, the authors propose that designing partial agonists with lower efficacy might favour Gq-biased signalling at HTR2B, given the larger energy-driven TM6 displacement seen for arrestin coupling. They also note that intracellular conformational changes that determine transducer selectivity appear to differ among Gi-, Gs- and Gq-coupled receptors, implying that mechanisms of functional selectivity are receptor-family specific. The study authors acknowledge several limitations: the HTR2B–b-arrestin-1 complex required extensive engineering (truncations, stabilising mutations and direct fusion of a constitutively active arrestin isoform), which might influence how arrestin engages the receptor despite preservation of interface residues; scFv30 binding may bias the C-tail phosphorylation pattern and thus arrestin conformation; co-expression with GRK2 may have produced a phosphorylation pattern not representative of other GRKs or physiological contexts; and cryo-EM classification may have excluded minor conformations. The authors recommend that future advances in cryo-EM and complementary in vivo functional assays will be necessary to capture GPCR–arrestin complexes with greater physiological relevance.