Structure of a Hallucinogen-Activated Gq-Coupled 5-HT2A Serotonin Receptor
This study reveals structurally how psychedelics, including LSD, psilocin, mescaline, and various N-BOH analogs, mediate their therapeutic and hallucinogenic effects by binding to and activating their molecular target, the serotonin (5-HT) 2A receptor coupled with G-protein Gaq.
Abstract
Hallucinogens like lysergic acid diethylamide (LSD), psilocybin, and substituted N-benzyl phenylalkylamines are widely used recreationally with psilocybin being considered as a therapeutic for many neuropsychiatric disorders including depression, anxiety, and substance abuse. How psychedelics mediate their actions-both therapeutic and hallucinogenic-are not understood, although activation of the 5-HT2A serotonin receptor (HTR2A) is key. To gain molecular insights into psychedelic actions, we determined the active-state structure of HTR2A bound to 25-CN-NBOH-a prototypical hallucinogen-in complex with an engineered Gαq heterotrimer by cryoelectron microscopy (cryo-EM). We also obtained the X-ray crystal structures of HTR2A complexed with the arrestin-biased ligand LSD or the inverse agonist methiothepin. Comparisons of these structures reveal determinants responsible for HTR2A-Gαq protein interactions as well as the conformational rearrangements involved in active-state transitions. Given the potential therapeutic actions of hallucinogens, these findings could accelerate the discovery of more selective drugs for the treatment of a variety of neuropsychiatric disorders.
Research Summary of 'Structure of a Hallucinogen-Activated Gq-Coupled 5-HT2A Serotonin Receptor'
Introduction
Psychedelic substances such as LSD, psilocybin and N-benzyl phenethylamine derivatives have a long history of human use and recent clinical interest because of their rapid and sometimes durable therapeutic effects in conditions including depression, anxiety and substance-use disorders. Earlier work implicates agonism of the 5-HT2A serotonin receptor (HTR2A) as essential for their psychedelic effects, but how different hallucinogens engage and stabilise distinct receptor states and downstream signalling cascades remains incompletely understood. Kim and colleagues set out to define molecular details of HTR2A activation by hallucinogens. They determined an active-state structure of HTR2A bound to the selective phenethylamine hallucinogen 25CN-NBOH in complex with an engineered mini-Gαq–βγ heterotrimer stabilised by a single-chain antibody fragment using single-particle cryo-electron microscopy (cryo-EM). To provide structural contrasts, the authors also solved X-ray crystal structures of HTR2A bound to the ergoline hallucinogen LSD and to the inverse agonist methiothepin, enabling comparisons between an active G protein–coupled state and ligand-stabilised inactive-like states. The objective was to identify ligand and receptor features that determine agonism, selectivity and G protein coupling, with implications for structure-guided drug design at HTR2A.
Methods
The investigators used a combination of cryo-EM, X-ray crystallography, mutagenesis, binding and functional assays, and molecular docking to characterise ligand–receptor interactions and receptor–G protein interfaces. For structural work, HTR2A constructs were engineered by truncating the N terminus (residues 1–65) and C terminus (405–471) to enhance expression; crystallographic constructs included a BRIL fusion in ICL3 and two thermostabilising point mutations (L247A and L371A). For cryo-EM, wild-type HTR2A (with the same truncations) was assembled with an engineered mini‑Gαq chimera, Gβ1 and Gγ2 expressed together in Sf9 insect cells and stabilised by the single‑chain antibody fragment scFv16. Ligands used were 25CN‑NBOH for the cryo-EM active-state complex and LSD or methiothepin for lipidic cubic phase crystallography. Protein expression, purification and complex assembly were performed in Sf9 cells with standard detergent solubilisation and immobilised metal affinity purification followed by size-exclusion chromatography. For the cryo-EM complex, purified HTR2A/25CN‑NBOH was mixed with a 1.2 molar excess of mini‑Gαq heterotrimer, treated with apyrase to remove GDP, and incubated with scFv16 before final purification. Cryo-EM data were collected on a Titan Krios/K3, yielding 2.7 million particle picks; a final subset of 168,570 particles produced a map refined to a global nominal resolution of ~3.27 Å. Lipidic cubic phase crystallisation of HTR2A–LSD and HTR2A–methiothepin complexes produced crystals diffracting to ~3.4 Å. Structures were solved by molecular replacement and refined with standard packages. Functional characterisation included radioligand binding with [3H]-LSD to determine affinities and dissociation kinetics, and bioluminescence resonance energy transfer (BRET) assays to measure receptor-driven G protein recruitment and β‑arrestin2 translocation in HEK293T cells. Alanine and selected point mutations were introduced to probe ligand interactions and the receptor–Gα interface; effects on binding, potency (pEC50 changes) and efficacy (Emax) were assessed. Molecular docking of 5‑HT and 25CN‑NBOH into the active structure was performed using established DOCK protocols to support ligand pose interpretation.
Results
Cryo-EM of the HTR2A–25CN‑NBOH–mini‑Gαq–βγ–scFv16 complex produced a near‑atomic active-state model at ~3.27 Å. Preliminary functional assays (BRET) indicated the engineered mini‑Gαq construct reproduces wild-type Gαq recruitment and that 25CN‑NBOH was the most efficacious agonist tested in those assays. The full complex was assembled from individually expressed receptor and G protein components in Sf9 cells and stabilised by scFv16, which binds between the αN helix of mini‑Gαq and the β‑propeller of Gβ1. Complementary X-ray crystal structures of HTR2A bound to LSD and to the inverse agonist methiothepin were solved at ~3.4 Å using an identical BRIL‑fused receptor construct with the two thermostabilising mutations. The LSD structure shows extracellular loop 2 (ECL2) forming a lid over LSD, consistent with prior data and with slow dissociation kinetics. Binding-site solvent-accessible volumes differed markedly across structures: methiothepin ~188.1 Å3, LSD ~153.1 Å3 and 25CN‑NBOH ~287.0 Å3, reflecting ligand-specific pocket rearrangements. Comparison of active and inactive-like states revealed canonical activation movements: outward tilting of TM5 and TM6 (measured displacements of ~4.3 Å at TM5 and ~7.1 Å at TM6) and inward shift of TM7 (~2.0 Å), together with rearrangements in conserved microswitches (NPxxY inward shift, disruption of the ionic lock involving R1733.50 and E3186.30, and rotation of the toggle residue W3366.48 with subsequent movement of F3326.44). 25CN‑NBOH adopts a distinct binding pose relative to LSD and methiothepin. Its 2‑hydroxyphenyl moiety projects into a previously unrecognised sub‑pocket between TM3 and TM6, making an edge‑to‑face π–π interaction with W3366.48 and a hydrogen bond with S1593.36; S159A markedly reduced 25CN‑NBOH potency. G3697.42 shifts inward to accommodate the ligand, and G369A dramatically reduces agonist potency without substantially altering binding affinity, suggesting a role for this residue in signal transduction rather than simple binding. Sequence and mutational data implicate G2385.42 (serine or threonine in many other biogenic amine receptors) as a determinant of 25CN‑NBOH selectivity, since G238S reduced agonist potency. LSD occupies the orthosteric pocket with similarities to earlier HTR2B–LSD structures; in HTR2A the indole‑NH of LSD hydrogen‑bonds to S2425.46 and ECL2 forms a lid that likely contributes to LSD's prolonged residence time. Mutation S242A accelerated LSD dissociation but did not substantially change binding affinity. Docking of serotonin (5‑HT) into the active HTR2A model predicted the expected salt bridge between its primary amine and D1553.32 and stabilising interactions with F339/F340 and hydrogen bonds with N3436.55 and S2425.46; N343A reduced 5‑HT potency but did not affect LSD. At the receptor–Gα interface, the authors mapped multiple polar and hydrophobic contacts between HTR2A and the C‑terminal α5 helix of Gαq. Hydrogen bonds involve receptor residues N1072.37, D1723.49, N3176.29 and N3848.47 with Gαq residues E242, Y243, Q237 and N244, respectively, while a hydrophobic core interfaces with several leucine and valine residues on both sides. Mutagenesis showed that N384A8.47 and L325A6.37 dramatically reduce receptor–Gαq coupling, R185A (ICL2) and N107A2.37 decreased 25CN‑NBOH efficacy by ~50%, and Q237A (Gαq H5.17) abolished agonist activity when mutated on the G protein side. N244A (Gαq H5.24) also diminished potency and efficacy. A notable conformational rearrangement of R185 in ICL2 brings it into backbone engagement with the Gα α5 helix. Functional coupling profiling across 14 Gα subunits showed that HTR2A couples robustly to Gq‑family members, minimally to conventional Gi or Gs families, and yields a reproducible but smaller response with Gz. Mutations at I181 (ICL2) — I181A and I181E — abolished Gq activation by 25CN‑NBOH while potentiating β‑arrestin recruitment, indicating a single residue can shift signalling bias from G protein to arrestin pathways.
Discussion
Kim and colleagues interpret their structural and functional findings as providing a mechanistic basis for how hallucinogens engage and activate HTR2A and how the receptor selectively couples to Gq‑family G proteins. The cryo‑EM active‑state structure reveals ligand-specific engagement of conserved microswitches and a distinct 25CN‑NBOH binding mode that perturbs W3366.48 and associated motifs to promote the outward movement of TM6 and accommodation of the Gα α5 helix. The authors emphasise that key contacts between HTR2A and residues on the tip of the Gαq α5 helix, together with supporting hydrophobic and backbone interactions outside the α5 tip (notably involving ICL2), underlie productive Gq coupling in vitro. These results contrast with earlier work using chimeric G proteins that suggested broad coupling of HTR2A to many Gα subtypes; the authors point to specific direct interactions (for example involving a conserved Q352‑cognate residue) that are critical for Gq activation and that may not be preserved in chimaeric constructs. The mutational data showing that I181 substitutions abolish Gq signalling while enhancing arrestin recruitment are highlighted as evidence that individual residues at the receptor–G protein interface can determine coupling specificity and signalling bias. The authors also note that the unique 25CN‑NBOH pose and its interactions with S1593.36, W3366.48 and G3697.42 may explain its high potency and selectivity for HTR2A; however, they acknowledge that more active‑state HTR2A structures will be needed to establish whether this pose is ligand‑specific or a general feature of active HTR2A. Finally, the study's implications for neuropsychiatric drug discovery are outlined: detailed maps of ligand recognition and receptor–effector coupling should support structure‑guided efforts to develop more selective HTR2A agonists with therapeutic potential. The authors also recognise limitations implicit in the present work, including that comparisons are made between a single active‑state structure and several mainly inactive structures and that in vitro coupling profiles may not fully capture the complexity of signalling in native tissues.
View full paper sections
INTRODUCTION
Naturally occurring psychedelics from plants as diverse as the mescaline-containing cactus Lophophora williamsii, the psilocybin-containing mushrooms Psilocybe sp., and the lysergamidecontaining seeds from ''Morning Glories'' (Ipomoea sp.) have been reportedly used for millennia for religious purposes and shamanism. Beginning in the 1950s and 1960s, semi-synthetic hallucinogens like lysergic acid diethylamide (LSD) and synthetic hallucinogens related to mescaline, including substituted amphetamines such as 2,5-dimethoxy-4methylamphetamine, were available and used recreationally. More recently so-called ''designer hallucinogens'' with a scaffold related to N-benzyl-2,5-dimethoxy-phenethylamine (NBOMe) (Figure) have become popularalbeit with scattered reports of toxicity. Of the NBOMe series, 25CN-NBOH is reported to be among the most potent and selective in vitro and in vivo. Hallucinogens like psilocybin and LSD have been described to have potential therapeutic actions for many neuropsychiatric diseases. Psilocybin, for instance, has shown efficacy in two phase II clinical trials of depression and anxiety. Remarkably, psilocybin's effects appear to be both rapid and enduring. Likewise, LSD has been reported in pilot studies to have efficacy in cluster headaches, to alleviate anxiety in terminal illness, and a number of uncontrolled and anecdotal reports have suggested its therapeutic actions in other neuropsychiatric disorders. Although the precise mechanisms of action of hallucinogens remain largely unclear, 5-HT 2A serotonin (5-hydroxytryptamine [5-HT]) receptor agonism is essential for their psychedelic effects in humans. The key to an understanding of psychedelic actions, then, is an appreciation of how they interact with and activate 5-HT 2A serotonin receptors (HTR2A). To elucidate hallucinogen actions at the molecular level, we report here the active-state structure of 5-HT 2A bound to 25CN-NBOH in complex with a mini-Gaq-bg heterotrimer stabilized by a single-chain variable fragment (scFv16) obtained by cryoelectron microscopy (cryo-EM). For comparison, we also report X-ray structures of HTR2A complexed with the prototypical hallucinogen LSD and the inverse agonist methiothepin. Collectively, these results provide insights into how a model hallucinogen stabilizes a specific HTR2A transducercoupled state. These findings should accelerate structureguided efforts for discovering more selective HTR2A agonists with potential therapeutic activity for many neuropsychiatric diseases.
RESULTS
A Gaq-Coupled HTR2A Structure Bound to the Hallucinogen 25CN-NBOH For structural studies with a transducer-coupled HTR2A, we initially evaluated the HTR2A non-selective hallucinogen LSDand the HTR2A-selective agonist 25CN-NBOH. In preliminary experiments, we found that 25CN-NBOH provided more effective stabilization of the final complex than LSD (not shown). We used an engineered Gaq that was previously demonstrated to faithfully recapitulate wild-type Gaq-coupled G-proteincoupled receptor (GPCR) activation for several GPCRs, including HTR2A. Here, 35 amino acids at the N terminus of engineered mini-Gaq were replaced by the corresponding sequence in Gai2 to facilitate binding of a single-chain variable fragment (scFv16), thereby further stabilizing the HTR2A-Gaq protein complex for structural studies (Figure). An analogous approach was recently used to obtain a structure of the M1-for fitted parameter values. (F) Ligand-mediated mini-Gaq recruitment to HTR2A. Several hallucinogen agonists were compared in BRET 1 assays. See Tablefor fitted parameter values. See also Figureand Table. muscarinic receptor complexed with Ga11. To verify the functionality of the complex we performed bioluminescence resonance energy transfer (BRET 1 and 2) (seefor details) (Figure) studies similar to those previously reported, which revealed that the functional activity of this engineered mini-Gaq construct is comparable to the wild-type Gaq (Figure; Table). The studies also showed that 25CN-NBOH was the most efficacious agonist tested in our Gaq BRET recruitment assays (Figure; Table). To improve the expression of HTR2A, parts of the N-and C-terminal of HTR2A were truncated and then cloned into a pFASTBac 1 vector (Figure). To obtain a stable complex of HTR2A-Gaq, the sequences of the engineered Gaq iN , Gb1, and Gg2 were integrated into a pFast dual expression vector to afford simultaneous expression of all three components and pre-formation of the Gaq iN -Gb1-Gg2 heterotrimer (Figure). We expressed HTR2A and G protein complex (Gaq iN -Gb1-Gg2) in sf9 insect cells and purified them individually. The full complex of HTR2A-Gaq iN -Gb1-Gg2 was then assembled in the presence of the agonist 25CN-NBOH and further stabilized by the single-chain antibody scFv16 (Figuresand). The structure of G protein-coupled HTR2A complexed with 25CN-NBOH was determined by single-particle cryo-EM at a global nominal resolution of 3.27 A ˚, enabling near-atomic resolution modeling of the complex (Figuresand; Table). The 25CN-NBOH pose was further confirmed and validated through the GemSpot pipelineand was virtually identical to the pose predicted by independent molecular docking studies (Figuresand). The overall structure of 25CN-NBOH-stabilized HTR2A-Gaq/b1/g2 structure is consistent with the fully active-state receptor conformation that has been observed in several other GPCRs bound to a G-protein ternary complex. We found that scFv16 binds to the crevice between the aN helix of mini-Gaq iN and the b-propeller of Gb1 similar to that observed in the MOR-Gai complex, thereby providing additional stabilization of the complex (Figure).
STRUCTURES OF LSD-AND METHIOTHEPIN-BOUND HTR2A
To obtain a more thorough understanding of how hallucinogens might interact with HTR2A, we determined crystal structures of two HTR2A-ligand complexes. LSD is the prototypical hallucinogen that exerts effects its psychedelic primarily via HTR2A, and we previously reported the structure of the 5-HT2B (HTR2B) receptor complexed with LSDand related ergolines. These structures provided important insights into mechanisms of biased agonism at HTR2Band hints regarding how LSD might interact with HTR2A. To obtain a better understanding of the actions of LSD on HTR2A, we solved the X-ray crystal structures of HTR2A complexed with LSD (Figure), and, for comparison, the potent inverse agonist methiothepin (Figure; Table). Both structures were obtained at 3.4 A ˚resolution using an identical HTR2A construct with apocytochrome b562RIL (BRIL) fused to the third intracellular loop 3 (ICL3) of the receptor and two thermostabilizing HTR2A mutations (L247A 5.51 and L371A 7.44 ) (Figures S2A-S2D). Binding assays showed that this engineered construct maintained WT ligand-binding affinities (Figure). The asymmetric unit (ASU) contains three molecules, and crystal contacts are mediated by TM1, TM4, and H8 (Figure). The superposition of these three molecules shows that the orientation of the receptors is nearly identical, although slight differences are observed in the BRIL fusion protein. We focused on HTR2A molecule ''chain A'' for subsequent analysis (Figure). The electron density maps for LSD and methiothepin were well resolved (Figure). The overall differences between LSD bound to HTR2A and the previous LSD-HTR2B structure are relatively subtle, as predicted. In the HTR2B/LSD structure, ECL2 forms a ''lid'' over LSD, prolonging LSD's residence time. In confirmation of our prior mutagenesis, kinetic, and molecular modeling studies, we find that ECL2 in the LSD-HTR2A complex is located in a similar position, occluding LSD (Figure). Using a previously described constitutively active HTR2A mutant, we found that methiothepin (a non-selective serotonin, dopamine, and adrenergic receptor antagonist) was a potent and efficacious HTR2A inverse agonist (Figure). Methiothepin's binding pose in HTR2A displays unique features that are detailed below.
STRUCTURAL COMPARISONS OF HTR2A INACTIVE-AND ACTIVE-STATE STRUCTURES
With these available structures in hand, we first assessed the overall conformational rearrangements between the inactive and active state structures. A surface view from the extracellular side also shows that the Gaq-coupled state is considerably more open than the LSD-bound state (Figures). We calculated the size of the ligand-binding pockets for all three structures using the CASTp 3.0 serverand found that the solvent-accessible volumes of the ligand-binding pockets were 188.1 A 3 for methiothepin, 153.1 A 3 for LSD, and 287.0 A 3 for 25CN-NBOH (Figure; Table). Typically, based on analysis of other class A active-state GPCR/G protein complex structures (e.g., b2-adrenergic/Gas, M1-muscarininc/Ga11 and M2-muscarinic/Go, CB1/ Gai1 [Krishna, A1-adenosine/Gai, and the k-opioid-nanobody), receptor activation leads to a contraction of the extracellular binding pocket and expansion of the intracellular end, thereby providing space for the engagement of transducers (e.g., G proteins or arrestins). Conceivably, the observed expansion of the ligand-binding pocket in the 25CN-NBOH-bound complex is ligand-specific, coinciding with its unique binding pose compared to LSD and methiothepin, although additional Gaq-coupled structures will be needed to test this hypothesis. In the cryo-EM map, the density of the ligand 25CN-NBOH, the transmembrane domains, and the G protein binding interface are all well resolved, thus allowing us to examine conformational changes upon agonist binding and transducer coupling (Figures). Consistent with previous G protein-and arrestincomplexed GPCR structures, the intracellular ends of TM5 and TM6 in HTR2A-Gaq complex undergo the largest displacement and are tilted outward by 4.3 A ˚(Ca of A265 5.69 ) and 7.1 A (Ca of I315 6.27 ), respectively, while TM7 in the HTR2A-Gaq complex is shifted inward by 2.0 A ˚(Ca of F383 7.56 ) (Figure). It is noteworthy that the structure of the muscarinic receptor M1 bound to Ga11, which is also a Gaq family subtype, shows an outward movement of TM6 by 8 A ˚that displays a slightly different tilt of the a5 helix of G11. Because of the stabilization provided by Gaq coupling, the overall structure of HTR2A resides in a rigid conformation, and ICL2 now displays helical turns reminiscent of that predicted by computational studiesthat are not resolved in the inactive structures (Figure). Other features related to receptor activation were observed in conserved motifs including (1) an inward shift of residues in NPXXY motif, (2) breaking of the ionic lock between R173 3.50 and E318 6.30 (Shapiro et al., 2002) due to side chain rearrangement of R173 3.50 in the E/DRY motif, and (3) rotation of the side chain of the toggle switch residue, W336 6.48 , and subsequent movement of the side chain of F332 6.44 in the P-I-F motif (Figure). Although these residues or motifs have been implicated in playing important roles in receptor activation and signaling transduction, the initiation and sequence of these events are still understudied.
CN-NBOH DISPLAYS A DISTINCT BINDING MODE AT HTR2A
We next examined the ligand-binding pocket of the three structures as they represent unique subfamilies (25CN-NBOH as an N-benzyl phenethylamine full-agonist, LSD is an ergoline arrestin-biased partial agonist, and methiothepin is an inverse agonist). Conserved interactions observed in all three structures reveal a shared mechanism for HTR2A binding (Figures). A salt bridge is observed between D155 3.32 and a positivelyfor volume estimates. charged nitrogen in each ligand, which is a critical interaction for ligand binding in serotonin and other monoamine receptors. Not surprisingly, mutation of D155 3.32 leads to the loss of function for almost all HTR2A ligands (Figure). Other hydrophobic interactions have also been observed in all three liganded structures, such as V156 3.33 , V235 5.39 , and, as predicted previously, W336 6.48 , F339 6.51 , and F340 6.52 (Figures 3A-3C). Mutagenesis and functional experiments suggested that many of these residues are essential for ligand binding and subsequent receptor activation (Figures, and S5B; Tableand). Quite recently, inactive-state structures of HTR2A bound to risperidone or zotepine were reported. Thus, a total of five ligand-HTR2A complexes were examined (risperidone and zotepine, LSD, 25CN-NBOH, and methiothepin here), thereby providing the opportunity to examine ligand-specific features related to binding. Alignment of all five ligand-complexed HTR2A structures shows that the HTR2A-selective agonist 25CN-NBOH displays a unique binding pose compared with the non-selective partial agonist LSD and the non-selective inverse agonists methiothepin, risperidone, and zotepine (Figure). We note that these following comparisons represent a single active state with four mainly inactive state structures. It will be important going forward to obtain more active-state HTR2A structures to determine if the binding pose of 25CN-NBOH is unique. The 2-hydroxyphenyl moiety of 25CN-NBOH dives deep down into a previously undescribed pocket between TM3 and TM6, forming hydrophobic interactions with the indole ring of W336 6.48 (Figure). For many years, this highly conserved W 6.48 in class A GPCRs has been proposed as a ''toggle-switch'' to control GPCRs transitioning between ''on'' and ''off'' states during signaling transduction. The close interaction between 25CN-NBOH and W336 6.48 coincides with a large displacement of the side chain of W336 6.48 , acting as a pivot for the outward movement of TM6 (Figure). This displacement of W336 accompanied by a movement of F332 6.44 that is part of the conserved P 5.50 -I 3.40 -F 6.44 motif (Figure). Both W 6.48 and the P 5.50 -I 3.40 -F 6.44 motif have been shown to undergo conformational changes upon receptor activation. Here, the dynamic coupling between W 6.48 and the P 5.50 -I 3.40 -F 6.44 motif may partially explain how allosteric changes are transmitted in the active state similar to what was observed in the cryo-EM structure of the CB1 cannabinoid-Gi in complex with fubinaca (Krishna. A comparison of all available serotonin-receptor structures, including active and inactive, reveals that the displacement of W336 6.48 of HTR2A is also greater than those observed in other structures (Figure). In addition to an edge-to-face p-p interaction with W336 6.48 , the 2-OH of 25-CN-NBOH is the only observed ligand to form a hydrogen bond with S159 3.36 , and this results in a 2.9 A ˚side-chain movement. As anticipated, the S159A 3.36 mutation dramatically attenuated 25CN-NBOH's agonist potency (Figure). The 2-hydroxyphenyl moiety is also accommodated by the conserved G369 7.42 which moves ''in'' upon agonist binding to form hydrophobic interactions. Slightly increasing the size of the side chain through mutation G369A 7.42 dramatically attenuated 25CN-NBOH's agonist potency (Figure; Tableand Table). However, the binding affinity of G369A 7.42 does not significantly change compared to the wild-type Figureand Table), indicating that this residue may play a role in signal transduction in HTR2A. These interactions, as will be shown below, are not observed in either the LSD-or methiothepin-bound structures. Given that LSD and methiothepin are non-selective HTR ligands, the observed binding pose between 25CN-NBOH and HTR2A could contribute to the selectivity of 25CN-NBOH at 5HTR2A over many other biogenic amine receptors (Figuresand). As reportedand summarized (Figure), 25CN-NBOH has negligible affinity for all other human biogenic amine receptors while preferring HTR2A over HTR2C and HTR2B (Figure). An alignment of binding site residues among biogenic amine receptors discloses that G238 5.42 is replaced by Ser 5.42 or Thr 5.42 in most other biogenic amine receptors (Figure). As predicted from this alignment, a G238 5.42 S mutation diminished 25CN-NBOH's agonist potency (Figure), and this likely represents one of the structural determinants essential for 25CN-NBOH's selective pharmacology. Although the phenethylamine group of 25CN-NBOH is located in the orthosteric pocket similar to LSD and other antagonists, 25CN-NBOH does not, like LSD, have substantial interactions with the conserved serines S239 5.43 and S242 5.46 . In particular, S242 5.46 is unique to HTR2A among the serotonin family (Figure). A comparison of LSD-bound HTR2A with HTR2B structures shows that the overall orientation of LSD is similar (Figure). However, the steric extrusion of L229 ECL2 in HTR2A and the H-bond interaction of S242 5.46 together may cause a slight rotation and leftward shift of the ergoline ring. (Figure). In the LSD-HTR2A structure, LSD forms a hydrogen bond between the indole-NH and side chain of S242 5.46 , and we next preformed additional experiments to gain insight into this interaction. We found that mutation of S242 5.46 to an alanine does not significantly change the binding affinity of LSD, although this mutation accelerates its dissociation rate, implying that this unique residue may contribute to LSD's unusually long binding kinetics (Figure). As expected, because the S242 5.46 does not directly interact with 25CN-NBOH, the S242A 5.46 mutation has no effect on 25CN-NBOH's potency or efficacy in BRET (Figure; Table). Guided by this structure, we also evaluated potential binding modes of the endogenous agonist 5-HT using molecular docking. In the docked conformation (Figure), the positively charged primary amine of serotonin forms a saltbridge with the anionic D155 3.32 (distance 2.8 A ˚) as predicted by mutagenesis studies. In further agreement with many prior mutagenesis and molecular modeling studies) 5-HT is also predicted to be stabilized by aromatic interactions with F340 6.52 and F339 6.51 and hydrogen bonds with N343 6.55 and S242 5.46 . N343 6.55 was previously implicated from molecular modeling studiesand was here found to be essential for 5-HT's agonist potency (Figure; Table). As predicted from the HTR2A-LSD structure, the N343 6.55 A mutation did not affect LSD's agonist potency, efficacy, or binding affinity (Figuresand).
THE AGONIST-GAQ COMPLEX BINDING INTERFACE
A major interface between HTR2A and the Gaq subunit is mediated by the C-terminal helix (a5 helix) of Gaq as suggested by G protein-bound structures. In detail, HTR2A residues N107 2.37 , D172 3.49 , N317 6.29 , and N384 8.47 form H-bonds with Gaq residues E242 H5.22 , Y243 H5.23 , Q237 H5.17 , and N244 H5.24 , respectively (superscription is CGN numbering system). Additionally, the residues A321 6.33 , L261 5.65 , I177 3.54 , L325 6.37 , and V324 6.36 of HTR2A form a hydrophobic core with L236 H5.16 , L240 H5.20 , and L245 H5.25 of Gaq (Figure). Considering these residues form the major interactions between HTR2A and Gaq, alanine mutagenesis studies were conducted, and BRET2 assays were performed to determine which might be key for Gaq recognition. Measurement of receptor expression levels showed that these mutants, for both receptors and Gaq (A-C) Specific residues in the binding pockets that interact with 25CN-NBOH (yellow) (A), LSD (light blue) (B), and methiothepin (brown) (C), respectively. Alternative 2D diagrams showing direct interactions with each ligand are also provided at the bottom of each panel. The salt bridge interaction, as well as hydrogen bond interactions, is shown as red dashed lines. (D) Mutagenesis studies showing the effects of orthosteric-site residues on ligand-binding affinity and functional activity. Heatmap of DpEC 50 (EC 50WT -EC 50mt ) (by BRET 2, HTR2A/Gaq) and DpKi (Ki wt -Ki mt ) (by binding assay, [ 3 H]-LSD) shows differences between HTR2A wild-type and mutants. See Figureand Tablesandfor fitted parameter values that represent mean ± SEM of n = 3 biological replicates. See also Figure. proteins, generally maintain robust whole cell (Figure) and receptor cell surface (Figure) expression, although some mutations slightly decreased receptor expression levels compared to wild-type (Figuresand). Among the HTR2A mutations, N384A 8.47 and L325A 6.37 dramatically reduced receptor-Gaq coupling compared to wild-type, whereas R185A ICL2 and N107A 2.37 decreased the efficacy of 25CN-NBOH by 50%, indicating that these residues probably play a crucial role in G protein signaling (Figure). Among the Gaq mutations, the agonist activity of 25CN-NBOH was completely abolished by Q237A H5.17 (Figure). Q237 H5.17 is a highly conserved residue in the Gaq family (Lys in Gai/o family and Arg in Gas family), indicating that this residue may be involved in Gaq specificity (Figure). N244 H5.24 is also a conserved residue existing only in the Gaq family (Gly in Gai/o, Met in G12/13, and Glu in Gas family), and expectedly, the N244A H5.24 mutant significantly decreases both the potency and efficacy of 25CN-NBOH (Figuresand), indicating that it may also contribute to receptor-Gaq coupling. An unexpected interaction was exemplified by R185 ICL2 that undergoes a dramatic10.5 A ˚rearrangement from being displaced upward toward the helical core to a close engagement via backbone interactions with R32 near the bottom of the a5 helix (Figuresand). The essential nature of this interaction was demonstrated by the R185A ICL2 mutation that significantly attenuated agonist potency (Figure; Table). In a similar manner, N384 8,47 transitions toward the a5 helix with an $3 A ˚movement where it engages N244 H5.24 near the tip of the a5 helix (Figure). We next compared the Gq interface with the recent M1muscarinic G11 interface to explore features responsible for Gaq subunit specificity. We found that the terminal hydroxyl of Y243, which is comparable to Y356 on the extreme terminus of the a5 helix and that has been implicated in receptor-G protein selectivity, was shifted 2.9 A ˚to create additional interactions with HTR2A that are not present in the M1 muscarinic receptor (Figure). Additionally the terminal N-L-V motif, which is conserved among all Gq-like Ga subunits, was displaced downward to achieve a previously unreported interacting surface with the terminus of TM7 of M1 (Figure). By contrast, in the Gq-HTR2A structure, the terminal V246 (that corresponds to V359 in M1-G11) is shifted upward 4.8 A ˚to avoid a potential clash with TM7 (Figure). Overall, this illustrates a striking rearrangement of the key terminal residues in HTR2A-Gq versus M1-G11 to afford productive interactions with the core of the respective Gq/11-coupled GPCRs (Figure). These results indicate that although the motifs used by different Gq-family subunits to afford receptor selectivity might be conserved, the interactions to mediate this selectivity are apparently distinct. Recent studies performed with chimeric G proteins have indicated that GPCRs may interact with a much larger diversity of Ga subunits than previously anticipated. These authorsprovided data indicating that HTR2A can activate not only Gq-family Ga subunits but also several members of the Gi (Gai1, Gai3, GaO, and Gaz) and, minimally, Gas family members (albeit with low relative intrinsic activity). Because these results are surprising, given the specific and non-conserved interactions we visualized in our 5-HT2A-Gq structure, we compared the structures of another 5-HT receptor in complex with a Gi-family member-5-HT1B-Gao (Garcı ´a-Nafrı ´a et al., 2018). A comparison of the two structures revealed that the terminal Y354 of Gao is displaced 10 A ˚down from the terminal V246 of Gaq where it cannot interact with any known residues of 5-HT2A (Figure). As well, the interaction surface formed by the final 5 residues of the a5 helix of Gao are not expected to form productive interactions with any of the closest residues in either TM7 or TM6 (Figure) in HTR2A. These results suggested to us that it was unlikely, based on an examination of these two structures, that HTR2A would productively interact with Gao or related Gai proteins. To test this hypothesis, we quantified the ability of HTR2A to productively interact with 14 distinct Ga subunits using our recently described BRET-based technology. As shown in Figuresand, HTR2A coupled robustly to Gqfamily members and minimally to conventional Gi-or Gs-family members. A diminished but reproducible response was seen for the pertussis-toxin-insensitive Gz. We verified the Gz interactions by showing a lack of response to the agonists in the presence of the HTR2A-selective antagonist M100907 (10 mM) and a lack of response in cells not transfected with HTR2A plasmid (Figure). Residue I181(ICL2) 34.51 is highly conserved among the GPCR family and has been reported to play a crucial role in Gas or Gaq coupling, but not Gai/o. In the HTR2A/Gaq structure, I181 34.51 interacts with L34 S1.02 , V79 S3.01 , F228 H5.08 , and I235 H5.15 of Gaq via hydrophobic interactions (Figure). We found that the I181A 34.51 and I181E 34.51 mutations completely abolished the Gaq activation by 25CN-NBOH, while potentiating its activity in arrestin recruitment (Figuresand). As shown in Figuresand, the I181A 34.51 and I181E 34.51 mutations similarly either attenuated (Figure) or abolished (Figure) the ability of LSD and 5-HT respectively, to activate Gaq (see Tablefor fitted parameters). (B) S159, W336, and G369 form a binding pocket that is important for 25CN-NBOH's agonist activity. (C) W336 6.48 acts as a pivot for the outward movement of TM6. (D) Conformational displacement of side-chain of W336 6.48 , followed by F332 6.44 in the P-I-F motif. See Figure. (E) The sequence alignment of the serotonin receptor family with an HTR2A specific residue S242 5.46 is highlighted. (F) The S242A 5.46 mutation accelerates LSD dissociation from HTR2A; data represent mean ± SEM of n = 3 biological replicates. (G) The overall structural comparison of HTR2A/LSD (pink/magenta color) and HTR2B/LSD (olive/lime color) and inset shows side view of LSD (magenta)-bound HTR2A (pink) crystal structure overlaid with the LSD (lime)-bound HTR2B (olive) structure. Hydrogen-bond interactions are highlighted by red dash lines. See also Figuresand. Finally, we observed that when most Family A active state structures are superimposed, the phenolic hydroxyl group of Y 5.58 is located in the inner core of TM6. It has been reported that Y 5.58 forms a water mediated-hydrogen bond network with R 3.50 (the middle part of the E/DRY motif) and Y 7.53 (the latter part of the NPXXY motif) as previously suggested. By contrast, in the case of HTR2A, HTR2B, and HTR2C, the phenolic hydroxyl of Y 5.58 is displaced outward from TM6 (Figure). Y 5.58 apparently interacts with F 6.41 by p-stacking interactions (Figure). The sequence alignment of all aminergic receptors shows that F 6.41 in the HTR2s is highly conserved (Figure), and this represents one of the possible reasons for the observed outward orientation of Y 5.58 in TM6.
DISCUSSION
In this paper, we determined the agonist-activated structure of the 5-HT2A serotonin receptor coupled to its canonical transducer Gaq, compared it with inactive state structures stabilized by both the partial agonist LSD and the inverse agonist methiothepin, and identified key determinants essential for agonist ac-tions and receptor-Gq coupling. Because the 5-HT2A receptor is essential for the actions of classical psychedelics including LSD, psilocin, mescaline, and various N-BOH analogs, these studies provide structural insights into the actions of psychedelics at their molecular target. Importantly, we also discovered that the specific interactions the 5-HT2A receptor achieves with Gq residues are essential for the apparently selective engagement of this receptor with Gq-family proteins in vitro. Some of these findings contrast with predictions from a recent study utilizing chimeric Ga subunits that indicated that HTR2A interact efficiently with all of the 11 tested Ga subunits, albeit with low efficacy at Gas. However, the Ga subtypes in that studywere created by replacing the terminal seven amino acids of Gq with the corresponding sequences from several other Ga subunits using the motif LXXXLX (where X = subtype selective sequences). Pertinent to this, we note that immediately prior to L353 is Q352 that is conserved among all of the chimeric G proteins employed in the study by. Here, we discovered that the Q352 cognate residue directly interacts with HTR2A TM6 residue N317 6.29 , and mutation of this residuefor fitted parameter values. (D) BRET validation of residues in the Gaq interface. See Tablefor fitted parameter values where data represent mean ± SEM of n = 3 biological replicates. See also Table. abolishes the ability of HTR2A to activate Gq. We also point out that in the only other Gq-GPCR structure available, the same Q352 interacts directly with the TM5 residue R 5.46. Importantly, using assays in which full-length heterotrimeric G proteins were used, we find that HTR2A productively couples weakly with only one Gi-family member, Gaz. Relevant to these findings, we also found important interactions outside the a5 helix that are essential for HTR2A-Gq interactions. Thus, R132 ICL2 interacts via an extended backbone interaction with N32 of the a-helical domain, and we further demonstrated via mutagenesis and functional assays that this interaction is essential for agonist-induced activation of Gq. This GPCR-Gq interaction is not seen in the M1-G11 structure, although R134 ICL2 in M1 could weakly interact with the backbone carbonyl of R32 in Gq based on an examination of that prior structure. These results underscore suggestions made decades ago that the selectivity determinants for Ga subunits may not be entirely specified by reside in their terminal 5-7 amino acids. We also discovered that a key hydrophobic residue essential for G protein coupling among various GPCRs-I181 ICL2when mutated, abolishes Gaq coupling while potentiating (A) Shown is an alignment of the HTR2A-Gq (green and olive, respectively) and the M1-G11 (lavender and pink, respectively) interface at the tip of the a5 helix of the respective Ga subunits. As depicted, V359 H5.26 undergoes a 4.8 A ˚shift in HTR2A compared with M1 while Y356 H5.23 is shifted 2.9 A ˚relative to M1. (B) Shows a comparison of HTR2A-Gq with the HTR1B-Go (yellow and gray, respectively) showing that the HTR2A Y354 H5.26 cognate residue V246 H5.26 is shifted up 10.9 A ˚. (C) Shows a heatmap of the relative efficacy for selected agonists at HTR2A versus a reference agonist for 14 distinct Ga subunits. (D) Shows concentration-response curves (N = 3 biological replicates each) for the 14 Ga subunits in 7C along with controls for the Gz study. See also Table. arrestin binding at HTR2A. This result was unexpected because there was no prior data with sufficiently high resolution of GPCR-arrestin complexes that could illuminate how the mutation of a single amino acid at the GPCR-Ga interface would switch the coupling of HTR2A from an unbiased to an arrestin-exclusive state. While this paper was in revision, however, a 3.3 A ˚resolution structure of the turkey b1-adrenergic receptor complexed with human b-arrestin1 was reported. A comparison of this structure with the Gs-coupled b2-adrenergic receptorand our HTR2A-Gq structure reveals that the cognate hydrophobic residue F139 ICL2 in b2-AR interacts with the Gs a5-helix in a manner similar to HTR2A-Gq. In the b1 adrenergic receptor-Arrestin structure, by contrast, F147 ICL2 is displaced upward where it would potentially clash with residues in the a5 helix of Gs (Figure) and where it is engaged in no productive interactions. A similar relative lack of extensive engagement of this cognate residue with arrestins can be seen in the lower resolution muscarinic-and neurotensin-receptor arrestin complexes) (Figure) This analysis provides a potential structural explanation for how loss of this hydrophobic interaction impairs Gq subunit coupling while preserving arrestin interaction-at least for HTR2A.
A
In addition to these fundamental insights into GPCR-Gq interactions, our findings have relevance for neuropsychiatric drug discovery. Psilocybin and LSD have emerged as potential therapeutics for a number of neuropsychiatric conditions including depression, anxiety, substance abuse, and cluster headaches. Because the 5-HT2A receptor is essential for the actions of psilocybin in humans, insights into the molecular details of both HTR2A ligand recognition and HTR2A-effector coupling are key to arriving at a molecular understanding of hallucinogen actions. Additionally, these studies will provide a framework for a structure-guided search to identify more selective and efficacious HTR2A agonists as potential innovative neuropsychiatric therapeutics as recently exemplified by us for other GPCRs.
STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:
RESOURCE AVAILABILITY
Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact,Bryan Roth (bryan_roth@med.unc.edu)
MATERIALS AVAILABILITY
Plasmids generated from this study may be obtained directly from the Lead Contact.
DATA AND CODE AVAILABILITY
The following structure files are available from the Protein Databank: 6WHA, 6WGT and 6WH4.
EXPERIMENTAL MODEL AND SUBJECT DETAILS
The following eukaryotic cell lines were used: Spodoptera frugiperda (Sf9) cells and HEK293T cells. The HEK293T cells were obtained directly from the American Type Culture Collection (CRL-11268) and have been validated by analysis of short tandem repeat (STR) DNA profiles and these profiles showed 100% match at the STR database from ATCC. The Sf9 cells were obtained as Sf-900 II SFM cells from Invitrogen and were not validated further. HEKT cells were grown in a humidified 37 C incubator with 5% CO 2 using media supplemented with 100 I.U./mL penicillin and 100 mg/mL streptomycin (Invitrogen, Cat#15140-122). The human cell lines HEK293T were maintained in DMEM (VWR, #45000) containing 10% fetal bovine serum (FBS, VWR, #89510-186). At 24-48 hr prior to studies, HEKT cells were cultured in the above media containing 1% dialyzed fetal bovine serum (Omega Scientific, #FB-03) instead of 10% fetal bovine serum to remove serotonin. For expression studies, Sf9 cells were grown at 27 C at a cell density of 2 X 10 6 cells/ml in ESF921 medium (Expression systems) with P1 virus at a multiplicity of infection (MOI) of 3.
METHOD DETAILS
Generation of HTR2A constructs for X-ray crystallography and cryoEM For crystallization, the modified thermostabilized apocytochrome b562RIL (BRIL) as a fusion partner was inserted into the receptor's third intracellular loop (ICL3) at A265 and T311 of the human 5-HT2A gene. The construct was further optimized by truncation of N-terminal residues 1-65 and C-terminal residues 405-471. The DN65-5HT2A-BRIL-DC405 DNA was subcloned into a modified pFASTBac1 vector for expression in Spodoptera frugiperda (Sf9) cells. All constructs also contained a hemagglutinin (HA) signal sequence followed by a FLAG tag at the N terminus and a PresCission protease site followed by a 10X histidine tag (His tag) at the C terminus to enable purification by immobilized metal affinity chromatography. To increase the thermostability and homogeneity of the 5-HT2A, two point mutations, Leu247Ala and Leu371Ala, were introduced in by standard Quick Change PCR (Figure). For the cryoEM study, wild-type human HTR2A was truncated in N-terminal residues 1-65 and C-terminal residues 405-471 to enhance expression levels, and a haemagglutinin (HA) signal sequence followed by FLAG-, His10-tag, TEV protease site, BRIL, HRV3C and linker ''GSGSG'' at the N terminus were indroduced (Figure). This construct also was subcloned into a modified pFASTBac1 vector.
EXPRESSION FOR HTR2A-XTAL AND -CRYOEM CONSTRUCTS
The Bac-to-Bac Baculovirus Expression System (Invitrogen) was used to generate high-titer recombinant baculovirus (> 10 9 viral particles per ml). Recombinant baculovirus was obtained by transfecting $5 mg of recombinant bacmid into 5X10 5 per well settled Spodoptera frugiperda (Sf9) in a 12-well plate (Corning) using 3 mL of Cellfectin II reagent (Invitrogen). After 5-12h, medium was exchanged for 1 mL of Sf-900 II SFM medium (Invitrogen), and the plates were incubated for 4-6d at 27 C. P0 viral stock was harvested as the supernatant and used to generate high-titer baculovirus stock by infection of 40 mL of 3X10 6 Sf9 cells/ml and incubation for 3d. Viral titers were determined by flow cytometry analysis of cells that were immobilized with phycoerythrin (PE)-conjugated gp64 antibody (Expression Systems). Expression of 5-HT2A was carried out by infection of sf9 cells at a cell density of 2 X 10 6 cells/ml in ESF921 medium (Expression systems) with P1 virus at a multiplicity of infection (MOI) of 3. Cells were harvested by centrifugation at 48 h after infection, washed in TN buffer (20 mM Tris-Cl, 100 mM NaCl, pH7.5) and stored at À80 C until use. Purification for HTR2A-XTAL and -cryoEM Thawed insect cell membranes were disrupted in a hypotonic buffer containing 10 mM HEPES (pH7.5), 10 mM MgCl 2 , and 20 mM KCl and protease inhibitors containing 500 mM AEBSF, 1 mM E-64, 1 mM Leupeptin and 0.15 mM Aprotinin. Subsequently, soluble and membrane associated proteins were removed in a high osmotic buffer containing 10 mM HEPES (pH7.5), 1,000 mM NaCl, 10 mM MgCl 2 , and 20 mM KCl. Purified membranes were incubated in the presence of desired ligands (50 mM 25CNNBOH, LSD or methiothepin) and protease inhibitor cocktail at 4 C for 2 h. The membranes were incubated with 2.0 mg/ml iodoacetamide (Sigma) for 30 min and were solubilized in the buffer containing 50 mM HEPES (pH 7.5), 1% (w/v) n-dodecyl-beta-D-maltopyranoside (DDM, Anatrace), 0.2% (w/v) cholesterol hemisuccnate (CHS, Sigma) and 150 mM NaCl, at 4 C for 2h. The solubilized HTR2A proteins in the supernatants were isolated by ultra-centrifugation in 40,000 rpm at 4 C for 50 min, and then incubated at 4 C overnight with TALON IMAC resin (Clontech), 800 mM NaCl and 20 mM imidazole as the final buffer concentration. The resin was washed with 10 column volumes of washing buffer I containing 50 mM HEPES (pH 7.5), 0.1% (w/v) DDM, 0.02% (w/v) CHS, 800 mM NaCl, 10% (v/v) glycerol, 20 mM imidazole and 50 uM of desired ligands (25CN-NBOH, LSD or methiothepin) and 10 column volumes of washing buffer II 50 mM HEPES (pH 7.5), 0.05% (w/v) DDM, 0.01% (w/v) CHS, 500 mM NaCl, 10% (v/v) glycerol and 50 mM of desired ligands (50 mM 25CN-NBOH, LSD or methiothepin) without imidazole. The protein was eluted using 3 column volumes of elution buffer containing 50 mM HEPES (pH 7.5), 0.05% (w/v) DDM, 0.01% (w/v) CHS), 500 mM NaCl, 10% (v/v) glycerol, 250 mM imidazole and 50 mM of desired ligands (25CN-NBOH, LSD or methiothepin) and concentrated in a Vivaspin 20 concentrator with a molecular weight cutoff of 100 kDa (Satorius Stedim) to 500 ml. The followed methods differed with crystal and CryoEM construct purifications. For the crystal construct purification, the 500 mL of 5HT2A protein sample was applied to PD MiniTrap G-25 columns (GE Healthcare) to remove imidazole. The C-terminal 10 X His tag was removed by addition of His-tagged PreScission protease (GeneScript) and incubation overnight at 4 C. Protease, cleaved His tag and uncleaved protein were trapped by equilibrated TALON IMAC resin (Clontech) and collecting the flow-through. 5-HT2A/LSD or methiothepin complexes were then concentrated to $30 mg/ml using a Vivaspin 500 centrifuge concentrator with a molecular weight cutoff of 100 kDa (Sartorius Stedim). Protein purity and monodispersity were tested by analytical size-exclusion chromatography column, SRT-300 (Sepax scientific) and Ultimate 3000 UHPLC systems (Thermo Scientific). For the CryoEM construct purification, the 500 mL of 5HT2A protein sample was applied to PD MiniTrap G-25 columns (GE Healthcare) to remove imidazole with buffer containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 0.5% (w/v) LMNG, 0.05% (w/v) CHS, 0.00025% (w/v) GDN, 100 uM TCEP, and 50 mM 25CN-NBOH. The N-terminal BRIL was removed by addition of His-tagged PreScission protease (GeneScript) and incubation overnight at 4 C. Protease, cleaved BRIL and uncleaved protein were trapped by equilibrated TALON IMAC resin (Clontech) and the flow-through was collected. The BRIL removed HTR2A was further purified by size exclusion chromatography on Superdex 200 10/300 gel filtration column (GE) with SEC buffer containing 20 mM HEPES (pH7.5), 100 mM NaCl, 0.001% (w/v) LMNG, 0.0001% (w/v) CHS, 0.00025% (w/v) GDN, 100 uM TCEP, 1 mM MgCl2 and 10 uM GDP, and collected until use.
GENERATION OF HETEROTRIMERIC MINI-GAQ PROTEIN COMPLEX
The mini-Gaq iN chimeric construct was designed for the binding of scFv16 and sub-cloned into a designed vector that help form heterotrimeric G complex in situ (Figure). For the expression, Sf9 insect cells were infected with one virus, encoding three subunits including the mini-Gaq iN subunit and Gb1/g2 subunits with histidine tag inserted at the amino terminus of the b subunit. Cells expressing the heterotrimeric G-protein were harvested 72 hours post infection. Cells were lysed in lysis buffer containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 30 mM imidazole, 5 mM b-mercaptoethanol, 0.1 mM GDP, 1 mM MgCl 2 , 0.2 % (v/v) Triton X-100 and protease inhibitors, and the soluble fraction was isolated by ultra-centrifugation at 40,000 rpm at 4 C for 50 min. The heterotrimer containing soluble fraction was purified using Ni-NTA chromatography. Human Rhinovirus 3C protease (HRV3C protease) was added and the histidine tag was cleaved at 4 C for overnight. The histidine tag removed heterotrimeric G protein was further purified by size exclusion chromatography on a Superdex 200 10/300 gel filtration column (GE) with SEC buffer containing 20 mM HEPES (pH 7.5), 100 mM NaCl, 0.001% (w/v) LMNG, 0.0001% (w/v) CHS, 0.00025% (w/v) GDN, 100 uM TCEP, 1 mM MgCl 2 , and 10 uM GDP, and collected and concentrated to $25 mg/ml, and stored at À80 C until use. Formation of HTR2A/mini-Gaq heterotrimer and scFv16 ScF16 was expressed from Spodoptera frugiperda (Sf9) insect cells as a secreted protein using baculovirus infection system and the purification process exactly as previously reported. Media expressing scFv16 from Spodoptera frugiperda (Sf9) was pH balanced to pH 8.0 by addition of Tris powder. Chelating agents were quenched by addition of 1 mM nickel and 5 mM calcium chloride and incubation with stirring for 1 hour at 25 C. The precipitants were removed by centrifugation at 16,263 g for 30 min and supernatant was incubated with His60 Ni Superflow Resin (Takara) for 5 hours. The resin was loaded over Poly-Prep Chromatography column (Bio-Rad) and washed with the washing buffer (20 mM HEPES pH7.5, 100 mM NaCl and 20 mM Imidazole). The protein was eluted with the elution buffer (20 mM HEPES pH 7.5, 100 mM NaCl and 300 mM Imidazole) and treated the HRC-3C protease to cleave the carboxy-terminal octa-histidine tag. Cleaved protein was further purified by size exclusion chromatography using a Superdex 200 16/60 column (GE healthcare). Monomeric fractions were pooled, concentrated, flash frozen in liquid nitrogen and stored at À80 C freezer until use. Purified scFv16 was concentrated, and flash frozen in liquid nitrogen until further use. Purified HTR2A/25CN-NBOH was mixed with a 1.2 molar excess of mini-Gaq iN heterotrimer. The coupling reaction was allowed to proceed at 24 C for 1 hr and was followed by addition of 0.2 U/ml (final concentration) of apyrase to catalyze hydrolysis of unbound GDP. After one hour, a 1.5 molar excess of scFv16 was added to the HTR2A /mini-Gaq mixture and incubated overnight at 4 C. HTR2A/mini-Gaq iN heterotrimer/scFv16 complex was further purified by size exclusion chromatography on a Superdex 200 10/300 column in 20 mM HEPES (pH 7.5), 100 mM NaCl, 0.001% (w/v) LMNG, 0.0001% (w/v) CHS, 0.00025% (w/v) GDN, 100 uM TCEP, 1 mM MgCl 2 , and 10 mM GDP. Peak fractions were concentrated to $17 mg/ml for electron microscopy studies.
CRYOEM DATA COLLECTION AND 3D RECONSTRUCTION
The sample (3.5 mL) was applied at a concentration of 17 mg/mL to glow-discharged holey carbon grids (Quantifoil R1.2/1.3) and vitrified using a Vitrobot Mark IV (FEI) at 22 C and 100% humidity, and plunged frozen into liquid ethane. CryoEM imaging was performed on a Titan Krios (ThermoFisher) electron microscope operated at 300 kV with a K3 Summit direct electron detector (Gatan) at a magnification of 57,050x in counting mode. 2368 movies of 50 frames each (4 s exposure) were obtained at a total dose of 82 electrons/A ˚2 and defocus ranging from À0.8 to À2.1 mm. Dose-fractionated image stacks were subjected to beam-induced motion correction and dose-weighting using MotionCor2. Contrast transfer function parameters for corrected micrographs were determined by Gctf. A total of 2.7 million particles were extracted from 2368 micrographs using semi-automated particle selection. 2D and 3D classification rounds were performed on a binned dataset (pixel size 1.704A ˚) using Relion 3.0. A subset of 168,570 particles were selected for the final map. The unbinned (with pixel size of 0.8421A ˚) particle set was subjected to Ctf Refinement and two rounds of Bayesian Polishing before the final refinement and sharpening was applied. The final map was postprocessed in Relion with a temperature factor of À90 A ˚2. The resolution of the final map at the 0.143 FSC threshold was estimated to be 3.27 A ˚by Relion and 3.23 A ˚using Mtriage in PHENIX (Table).
MODEL BUILDING AND REFINEMENT
Homology models of active-state HTR2A were built by SWISS-MODELusing HTR2C (PDB code 6BQG)for the receptor and M1R/G11 (PDB code 6OIJ)for the G-protein as template models, respectively. All models were docked into the EM density map using Chimerafollowed by iterative manual adjustment in COOTand phenix.real_space_refine in Phenix. The model statistics was validated using Molprobity. Structural figures were prepared by Chimera or Pymol (). The final refinement statistics were provided in Table.
LIPIDIC CUBIC PHASE CRYSTALLIZATION
The purified 5-HT2A protein in complex with LSD and methiothepin was screened for crystallization in lipidic cubic phase (LCP) with mixed molten lipid (90% (w/v) monoolein and 10% (w/v) cholesterol) at a protein:lipid ratio of 1:1.5 (v/v) using a mechanical syringe mixer. Crystallization was done on 96-well glass sandwich plates (Marienfeld) in 50 nL LCP drops that were dispersed from a 10 mL gas-tight pipette (Hamilton) using a handheld dispenser (Art Robbins Instruments) and overlaid with 1 mL of precipitant solution. After optimization, crystals were obtained in 100 mM Tris-HCl (pH 7.0), 380-430 mM potassium phosphate monobasic, 27%-32% (v/v) polyethylene glycol 400 (PEG400), 100 mM guanidine hydrochloride and 300 mM NDSB-195. Crystals grew to a maximum size of $30 mm X 20 mm X 20 mm within 2 weeks and were harvested directly from the LCP plates using MiTeGen micromounts and storage in liquid nitrogen. Data Collection and Structure Determination X-ray diffraction data of 5-HT2A/LSD and methiothepin crystals were collected at beam line 23ID-B (GM/CA CAT) of the Advanced Photon Source (Argonne, IL, USA) using a 10-mm minibeam at a wavelength of 1.0330 A ˚and an Eiger-16 m detector (Dectris) and beam line 17ID-2 (FMX) of the NSLS-II (Brookhaven, NY, USA). The data collection strategy was set up as an exposing the crystals for 0.2 s to an unattenuated beam using 0.2 oscillation. Diffraction images were indexed, integrated and scaled, and merged using XDSand further scaled using AIMLESS. Initial phase was obtained by molecular replacement (MR) method with Phaserusing the receptor and BRIL portion of 5-HT2B/LSD (PDB: 5TVN) as independent search models. Refinement was carried out with Phenixand REFMACfollowed by manual examination and adjustments of the refined structures in the program COOTwith both 2|Fo|-|Fc| and |Fo|-|Fc| maps. After refinement, we did not observe any Ramachandran outliers in any of the structure: 95.98% and 4.02% for LSD, and 93.73% and 6.27% for methiothepin of favored and allowed regions, respectively, as defined by Ramachandran statistics. We further observed Molprobity scores of 2.04 for LSD, and 2.1 for methiothepin (Table).
RADIOLIGAND BINDING ASSAYS
Competitive binding assays were performed using membrane preparations from HEK293 T cells transiently expressing HTR2A wt or mutants. Binding assays were set up in 96-well plates in the standard binding buffer (50 mM Tris, 0.1 mM EDTA, 10 mM MgCl 2 , 0.1% BSA, 0.01% ascorbic acid, pH 7.40). Saturation binding assays with 0.1-20 nM [ 3 H]-LSD in standard binding buffer were performed to determine equilibrium dissociation constant (Kd) and Bmax, whereas 10 mM final concentration of LSD was used to define nonspecific binding. For the competition binding, 50 mL each of 3 H-LSD (final 0.6 nM), drug solution (3X) and homogeneous HTR2A membrane solution was incubated in 96-well plates in the standard binding buffer. Reactions (either saturation or competition binding) were incubated for 2 h at room temperature in the dark and terminated by rapid vacuum filtration onto chilled 0.3% PEI-soaked GF/A filters followed by three quick washes with cold washing buffer (50 mM Tris HCl, pH 7.40) and read. Results were analyzed using the equation 'one-site fit Ki' in GraphPad Prism 8.0.
MOLECULAR DOCKING OF 5-HT
The agonist 5-HT was docked into the active HTR2A structure bound to 25-CN-NBOH (chain A, current work) using DOCK3.7. During the docking, 30280 complexes were sampled at the orthosteric binding site. Each pose was scored against three pre-calculated enegy grids which represent AMBER van der Waals potential, Poisson-Boltzmann electrostatic potential by QNIFFTT, and ligand desolvation, respectively. The best scoring 5-HT conformation received a docking score of À37.5 kcal/mol. The 3D dockable db2 file of 5-HT was downloaded from ZINC15 ().
MOLECULAR DOCKING OF 25-CN-NBOH
The 3D dockable db2 file of 25-CN-NBOH was generated using our standard protocol (). The protonation states at pH;), 3D structures (Corina 4.2), conformer ensembles (omega v.2.5.1.4,), partial charges and desolvation energies (AMSOL version 7.1) were calculated as previously described. The docking grids used were identical to those used for the 5-HT docking as follows. All steps in the standard protocol are identical to the steps in the ZINC15 pipeline: Jchem;) is used to enumerate the protonation states at pH 7.4 from an input SMILES. Each protonated SMILES is used to produce an initial 3D structure by Corina (version 4.2). Based on the 3D conformer, AMSOL (version 7.1) is used to compute partial charges and desolvation energies. OMEGA (version 2.5.1.4,) is used to generate conformer ensembles. At the final step, mol2db2 compresses all the information calculated above into a dockable db2 file. Over 7,500 orientations were explored in the orthosteric site, resulting 2,723,040 complexes sampled during the docking (orientations x conformations). The best scoring pose received a favorable docking score of À44.2 kcal/mol. The ligand symmetry accounted r.m.s.d (0.16A ˚) between the docked pose and the cryo-EM pose is calculated by the Hungarian algorithm in DOCK6.
LIGAND DISSOCIATION RADIONLIGAND BINDING ASSAY
Radioligand dissociation assays were performed in parallel with the competitive binding assays utilizing the same concentrations of radioligand, membrane preparations, and binding buffer (50 mM Tris, 10 mM MgCl 2 , 0.1 mM EDTA, 0.1% BSA, 0.01% ascorbic acid, pH 7.4). All assays utilized at least two concentrations of radioligand ([ 3 H]-LSD = 0.4 and 1.2 nM) (PerkinElmer). For dissociation assays, membranes were incubated with radioligand for at least two hours at room temperature before the addition of 10 mL of 10 mM excess cold ligand LSD to the 200 mL membrane suspension at designated time points. Time points spanned 2 minutes to 7 hours. Immediately at time = 0 min, plates were harvested by vacuum filtration onto 0.3% polyethyleneimine pre-soaked 96-well filter mats (Perkin Elmer) using a 96-well Filtermate harvester, followed by three washes of cold wash buffer (50 mM Tris pH 7.4). Scintillation (Meltilex) cocktail (Perkin Elmer) was melted onto dried filters and radioactivity was counted using a Wallac Trilux MicroBeta counter (PerkinElmer). Data were analyzed using ''Dissociation -One phase exponential decay'' in Graphpad Prism 8.0.
BIOLUMINESCENCE RESONANCE ENERGY TRANSFER ASSAYS (BRET)
To measure HTR2A-mediated b-arrestin2 recruitment, HEK293T cells were co-transfected in a 1:5 ratio with human 5-HT 2A R containing C-terminal Renilla luciferase (RLuc8), and Venus-tagged N-terminal b-arrestin2. After at least 16 hours, transfected cells were plated in poly-lysine coated 96-well white clear bottom cell culture plates in plating media (DMEM + 1% dialyzed FBS) at a density of 25-50,000 cells in 200 ml per well and incubated overnight. The next day, media was decanted and cells were washed twice with 60 mL of drug buffer (1 3 HBSS, 20 mM HEPES, 0.1% BSA, 0.01% ascorbic acid, pH 7.4), then 60 mL of drug buffer was added per well. For kinetic experiments, plates were incubated at 37 C at least 20 minutes prior to receiving drug stimulation. Afterward, 30 mL of drug (3X) was added per well and incubated for designated time points. Before reading, 10 mL of the RLuc substrate, coelenterazine h (Promega, 5 mM final concentration) was added per well, incubated an additional 5 minutes to allow for substrate diffusion, and plates were immediately read for both luminescence at 485 nm and fluorescent eYFP emission at 530 nm for 1 s per well using a Mithras LB940 multimode microplate reader. The ratio of eYFP/RLuc was calculated per well and the net BRET ratio was calculated by subtracting the eYFP/RLuc per well from the eYFP/RLuc ratio in wells without Venus-b-Arrestin present. The net BRET ratio was plotted as a function of drug concentration using Graphpad Prism 5 (Graphpad Software Inc., San Diego, CA). Data were normalized to % 5-HT stimulation and analyzed using nonlinear regression ''log(agonist) vs. response'' in GraphPad Prism 8.0. For measurement of arrestin translocation at other GPCRs an identical approach was used. For HTR2A-mediated G protein activation, HEK293T cells were plated either in six-well dishes containing 700-800,000 cells per well, or 10-cm dishes at approximately 7-8 million cells/dish. Cells were transfected 2-4 hours later, using a 1:1:1:1 ratio of the receptor:GarLuc8:Gb:Gg GFP DNA. Transit 2020 (Mirus biosciences) was used to complex the DNA at a ratio of 3 mL Transit/mg DNA, in OptiMEM (GIBCO) at a concentration of 10 ng DNA/mL OptiMEM. The next day, cells were harvested from the plate using Versene (0.1M PBS + 0.5 mM EDTA, pH 7.4), and plated in poly-D-lysine-coated 96-well white assay plates (Greiner) at a density of 25-50,000 cells per well. One day after plating in 96-well assay plates, white backings (Perkin Elmer) were applied to the clear bottoms of the plate, and media was carefully aspirated and replaced with 50 mM coelenterazine 400a (nanolight technology) in 60 mL of drug buffer (1 3 HBSS, 20 mM HEPES, 0.1% BSA, 0.01% ascorbic acid, pH 7.4). After a five-minute equilibration period, cells were treated with 30 mL of the drug for an additional 5 minutes. Plates were then read in an LB940 Mithras plate reader (Berthold Technologies) with a 395 nm (RLuc8-coelenterazine 400a) and 510 nm (GFP2) emission filters, at 1 s integration times. Plates were read six times, and measurements from the sixth read were used in all analyses. BRET ratio was computed as the ratio of the GFP2 emission to rLuc8 emission. Data were normalized to % 5-HT stimulation and analyzed using nonlinear regression ''log(agonist) vs. response'' in Graph-Pad Prism 8.0. For measurement of G protein activation with reference ligands the approach above was used identically as described except that neurotensin was used as a control for NTSR1-neurotensin.
SURFACE EXPRESSION ENZYME-LINKED IMMUNOSORBENT ASSAY
To confirm cell surface expression of HTR2A and its mutants, after 48hr transfected, immunocytochemistry was done using cells plated on 384-white plates at 10,000 cells/well. Cells were fixed with 20 ml/well of 4% paraformaldehyde (Fisher, #AAJ19943K2) for 10 minutes at room temperature. After fixation, cells were washed twice with 40 ml/well of PBS. Blocking was performed with 20 ml/well of 5% BSA (Akron, #AK8909) in PBS for 30 minutes at room temperature. After blocking, 20 ml/well of monoclonal ANTI-FLAG M2-Peroxidase (HRP) antibody (Sigma-Aldrich, A8592) diluted 1/10,000 in PBS was added and incubated for 1 hour at room temperature. This was followed by two washes with 80 ml/well of PBS. Then, 20 ml/well of SuperSignal enzyme-linked immunosorbent assay (ELISA) Pico Chemiluminescent Substrate (Thermo Fisher, #37070) was added, and luminescence was counted using a PHERAstar FSX (BMG Labtech). Data were plotted as relative luminescent units (RLU) in GraphPad Prism 8.0.
QUANTIFICATION AND STATISTICAL ANALYSIS
For radioligand binding assays data were analyzed using ''Dissociation -One phase exponential decay'' in Graphpad Prism 8.0 for the kinetic studies and the the equation 'one-site fit Ki' in GraphPad Prism 8.0 for the inhibition studies. For the BRET studies, the net BRET ratio was plotted as a function of drug concentration using Graphpad Prism 8 (Graphpad Software Inc., San Diego, CA) and data were normalized to % 5-HT or the reference agonist stimulation and analyzed using nonlinear regression ''log(agonist) vs. response'' in GraphPad Prism 8.0. For the cell surface expression studies Data were plotted as relative luminescent units (RLU) in GraphPad Prism 8.0. For the analysis of all fitted curves the goodness-of-fit was calculated by GraphPad Prism 8.0 and the data analyzed for parameter estimates was not evaluated for hetereskedasticity. Data in figures and tables are reported as mean ± standard error of the mean (SEM) with the number of biological and technical replicates indicated in the figure and table legends where ''n'' represents the number of biological replicates performed. EC 50 and Emax were analyzed by analysis of variance (ANOVA), and then a Dunnett's multiple comparison test using comparing each mutant to the wild-type receptor. The family wise-significance and confidence level was set at 0.05. For the depiction of the heatmap of agonist efficacies, the on-line program MORPHEUS () was used with the map paramters listed in the figure. Figure S7. Y 5.58 of HTR2A, HTR2B, and HTR2C Shows Distinct Orientation, Related to Figure(A) Superimposed active structures of HTR2A/Gq, HTR2B, HTR2C, M1/G11, HTR1B/Go, A1/Gi2, A2A/Gs, CB1/Gi1, CB2/Gi1, Mu/Gi1, M2/Go, and Rho/Gt. Y 5.58 , R 3.55 , and Y 7.53 are highlited with sticks. (B) F 6.41 and Y 5.58 fo HTR2s (2A, 2B, and 2C) forms pi-stacking interaction. (C) Sequence alignment of aminergic receptors in position 6.41. Only F 6.41 (green color) of HTR2s is highly conserved. (D) Sequence alignment of a5 helix of Ga subunits.