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

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 L.Roth ( bryan_roth@med.unc.edu ) Materials Availability All the plasmids and cells generated from this study could be obtained directly from the Lead Contact with a completed Materials Transfer Agreement if there is potential for commercial application. All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

Data and Code Availability

All data generated or analyzed in this study are included in this article and the Supplementary Information . The cryo-EM density maps and corresponding coordinates have been deposited in the Electron Microscopy Data Bank (EMDB) and the Protein Data Bank (PDB), respectively, under the following accession codes: EMD-25403 and 7SRS (HTR2B/β-arrestin-1), EMD-25402 and 7SRR (HTR2B-LSD/miniGq) and EMD-25401 and 7SRQ (HTR2B-LSD). The mass spectrometry proteomics data are available via ProteomeXchange with identifier PXD030752.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Two eukaryotic cell lines, Spodoptera frugiperda ( Sf9 , Expression systems) cells and HEK293T cells (ATCC), were used in this study. Sf9 cells suspension in ESF-921 medium were purchased from Expression systems and used for protein expression of HTR2B for structural study. Sf9 cells were grown in ESF-921 medium at 27°C, 120 rpm without further validation. HEK293T cells were purchased from the American Type Culture Collection (ATCC, CRL-11268) and used for Gq and β-arrestin-1 recruitment assay in this study. HEK293T cells were grown in a humidified 37°C incubator with 5% CO 2 using DMEM medium (VWR, #45000) supplemented with 10% (v/v) fetal bovine serum (FBS, VWR, #89510–186) and 100 I.U./mL penicillin and 100 mg/mL streptomycin. Before cell plating, the DMEM medium were changed from 10% FBS to 1% (v/v) dFBS to remove serotonin. HEK293T cells were authenticated by the supplier (ATCC) using morphology, growth characteristics and STR profiling. METHODS DETAILS Constructs for structural studies To get the structure of HTR2B in complex with β-arrestin-1, the ICL3 and C-terminus of HTR2B were truncated based on BRET1 recruitment assay as noted in the main text. In detail, residues A248-V313 in the ICL3 of HTR2B were removed and replaced with a 7 residues linker “RLLSGSR”. For the C-terminus, the middle part (L408-S444) and the last 17 residues (L465-V481) were removed to make a C-tail chimera. Mutations K247 5.68 V and E319 6.30 L were added to the construct to increase the β-arrestin-1 recruitment by LSD. A modified thermostabilized apocytochrome b562RIL (BRIL) as a fusion partner was fused before T36 of HTR2B with M144 3.41 W mutation adapted from the previous construct ( Wacker et al., 2013 ). β-arrestin-1 isoform 2 with the R169E constitutive mutation was then fused to the C-terminus of HTR2B with a 4 × GSA linker. At last, an engineered scFv30 was directly fused after L368 of β-arrestin-1 to achieve a tandem expression. The HTR2B-β-arrestin-1-scFv30 chimera was cloned into a pFastBac1 vector containing a haemagglutinin (HA) signal sequence followed by FLAG-tag, His10-tag and TEV protease site at the N-terminus. The HTR2B construct used for transducer-free HTR2B has the same sequence as the receptor portion used for HTR2B-β-arrestin-1-scFv30 chimera, except that it does not have the E319 6.30 L mutation and the C-terminus is further truncated to C405 according to a previous crystallography construct ( Wacker et al., 2013 ). Fab P2C2 was cloned into a pFastBac-Dual vector with a honeybee signal peptide at the N-terminus of the light chain and a GP67 signal peptide at the N-terminus of the heavy chain ( Ishchenko et al., 2017 ). A 6 × His-tag was added to the C-terminus of heavy chain to facilitate protein purification. The HTR2B construct used for the HTR2B-Gq complex study contains residues 36–405 and a M144 3.41 W mutation. A N-terminal BRIL as a fusion partner was added before T36 of HTR2B. The N-bril fused receptor portion was then subcloned into a same vector used for HTR2B-β-arrestin-1-scFv30 complex. Constructs of scFv16 and the heterotrimeric mini-Gαq protein complex were the same as we used in our previous HTR2A-Gq structure ( Kim et al., 2020 ).

Expression and purification of Fab P2C2 and scFv16

Bac-to-Bac expression system was used to generate the baculovirus for all the protein expressions in this study. Before infection, virus titer was determined by flow-cytometric analysis using gp64-PE antibody (Expression systems) stained cells. For Fab P2C2, Sf9 cells at a density of 2 million cells per ml were infected with P1 virus at a multiplicity of infection (MOI) of 3. Supernatant containing the secreted P2C2 was collected at 96 h post-infection. Tris powder was then added to adjust the medium to pH 7.8. Chelating agents were quenched by addition of 1 mM nickel chloride and 5 mM calcium chloride and stirring in cold room for 1 hour. After another centrifugation, 1 ml His60 Ni Superflow Resin (Takara) was added to the supernatant for overnight binding at 4°C. The resin was collected next day and washed with 20 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 10 mM imidazole. The protein was eluted with a same buffer containing 250 mM imidazole and further purified by size exclusion chromatography using a Superdex 200 16/60 column (GE healthcare). The peak fractions were collected and concentrated to 1 mg ml-1 for future use. scFv16 was expressed and purified using the same protocol. Protein expression and purification of HTR2B-β-arrestin-1-scFv30 complex For HTR2B-β-arrestin-1-scFv30 chimera, Sf9 cells at a density of 2 million cells per were infected with P1 virus of HTR2B-β-arrestin-1-scFv30 and GRK2 at a multiplicity of infection (MOI) ratio of 3:1.5. Cells were harvested by centrifugation at 48 h post-infection. The cell pellet was then washed with a low-salt buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 and 20 mM KCl and proteinase inhibitor containing 500 mM AEBSF, 1 mM E-64, 1 mM Leupeptin and 0.15 mM Aprotinin. Subsequently, two rounds of high-salt wash were performed using a buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl and 500 mM NaCl to remove membrane associated proteins. Purified cell membranes were resuspended in low-salt buffer and incubated with 10 μM LSD and 200 μM TCEP for 30 min at room temperature. After additional 1 hour incubation in cold room, the cell membranes were solubilized using solubilization buffer at a final concentration of 30 mM HEPES, 100 mM NaCl, 10 % glycerol, 5 μM LSD, 100 μM TCEP and 0.6% (w/v) n-dodecyl-beta-D-maltopyranoside (DDM, Anatrace), 0.2% (w/v) cholesteryl hemisuccinate (CHS, Sigma) at 4°C for 2.5 h. The solubilized proteins in the supernatants were isolated by ultra-centrifugation at 40,000 rpm for 30 min, and then incubated at 4°C overnight with 1 ml TALON IMAC resin (Clontech) and 20 mM imidazole. The resin was collected next day and washed with 10 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% glycerol, 1 μM LSD and 100 μM TCEP. The resin was then incubated with 20 mM HEPES pH 7.5, 100 mM NaCl, 0.8% Lauryl Maltose Neopentyl Glycol (LMNG, Anatrace), 0.27% GDN (Anatrace), 0.08% CHS, 5% glycerol, 1 μM LSD and 100 μM TCEP to exchange the detergent from DDM/CHS to LMNG/GDN/CHS. The resin was washed with additional 15 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 0.01% LMNG, 0.033% GDN, 0.001% CHS, 5% glycerol, 1 μM LSD, 100 μM TCEP and 30 mM imidazole. The protein was then eluted with 4.5 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 0.01% LMNG, 0.033% GDN, 0.001% CHS, 5% glycerol, 5 μM LSD, 100 μM TCEP and 250 mM imidazole. Eluted protein was concentrated to 400 μl and incubated with 100 μl P2C2 Fab (1 mg/ml) at 4°C for 2 h before subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES pH 7.5, 100 mM NaCl, 1 μM LSD, 100 μM TCEP, 0.001% (w/v) MNG, 0.00033 (w/v) GDN and 0.0001% (w/v) CHS. Peak fractions were then collected and concentrated to 3.5 mg ml-1 to make the cryoEM grids. Protein expression and purification of HTR2B-P2C2 complex For the transducer-free HTR2B-P2C2 complex, Sf9 cells at a density of 2 million cells per were infected with receptor P1 virus at a multiplicity of infection (MOI) of 3. Cells were harvested by centrifugation at 48 h post-infection. Followed by a low-salt wash, the cell membranes were washed twice by high-salt buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl and 1 M NaCl. The membrane was solubilized and binding overnight using a same protocol as HTR2B-β-arrestin-1-scFv30, except TCEP was not added. The resin was collected next day and washed with 25 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% glycerol and 1 μM LSD. The protein was then eluted using a same buffer containing 250 mM imidazole. Eluted protein was concentrated to 350 μl and incubated with 150 μl P2C2 Fab (1 mg/ml) at 4°C for 2 h before subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES pH 7.5, 100 mM NaCl, 1 μM LSD, 0.001% (w/v) MNG, 0.00033 (w/v) GDN and 0.0001% (w/v) CHS. Peak fractions were then collected and concentrated to 6.5 mg ml−1 to make the cryoEM grids. Protein expression and purification of HTR2B-Gq complex The cell pellet of HTR2B and Gq complex from 2 L culture was thawed at room temperature and resuspended in law salt buffer containing 10 mM HEPES pH 7.5, 50 mM NaCl, 20 mM KCl, and protease inhibitor 500 μM AEBSF, 1 μM E-64, 1 μM Leupeptin and 0.15 μM Aprotinin. The HTR2B and Gq complexes were formed on the membrane in the presence of 5 μM LSD, followed by incubation for 1.5 h at room temperature. Cell membranes were collected by ultra-centrifugation at 40,000 rpm for 25 min. The membranes were then resuspended and solubilized in solubilization buffer containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl2, 20 mM imidazole, 0.1 mM TCEP, 10%(v/v) glycerol, 0.5% DDM, 0.1%(w/v) CHS, 10 μM LSD, and protease inhibitor cocktail at 4°C. After 3 h incubation, the supernatant was isolated by centrifugation at 70,000 rpm for 50 min and then incubated overnight with pre-equilibrated TALON resin (CLONTECH) at 4°C. The resin was collected and washed with 30 column volumes with washing buffer 20 mM HEPES, pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.1 %(w/v) LMNG, 0.01 % (w/v) CHS and 10 μM LSD. The protein was then eluted using the same buffer supplemented with 300 mM imidazole. 500 μl of protein sample was applied to PD MiniTrap G-25 columns (GE Healthcare) to remove imidazole and change the detergent to LMNG 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 μM TCEP, and 20 μM LSD. The N-terminal BRIL protein was removed by the addition of His-tagged PreScission protease (Genescript) and incubation overnight at 4 °C. Unnecessary his-tagged proteins such as protease, BRIL, and free his tag were removed by equilibrated TALON resin, and the flow-through was collected. The eluate protein was concentrated and subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES, pH 7.5, 100 mM NaCl, 10 μM LSD (agonist), 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN, and 0.000075% (w/v) CHS. The peak fractions of complex were pooled and concentrated to 8.5 mg/ml for electron microscopy experiments.

Cryo-EM data acquisition and processing for HTR2B-LSD complexes

Particles from two cryo-EM data collections of the same sample vitrified on UltrAuFoil holey gold (Quantifoil, Au300-R1.2/1.3) or Quantifoil holey carbon (R1.2/1.3) grids contributed to the HTR2B/β-arrestin-1 complex reconstruction presented here. For this, 3 μl of purified HTR2B/β-arrestin-1 complex at a concentration of 3.5 mg/mL were applied to the glow-discharged (45–50 seconds at 10–15 mA) grids in 100% humidity at 18°C. Samples were blotted for 1 second and plunged-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Cryo-EM 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 57,050 × (0.8521 Å/pixel) in counting mode using SerialEM ( Mastronarde, 2005 ). The collection on holey gold or carbon grids generated 6,217 movies and 1,533 movies, respectively, dose fractioned over 50 frames, recorded for 0.05 sec/frame for a total dose of 68 electrons/Å2 in super-resolution mode with a defocus range of 0.8–1.8 μm, for a total of 7,750 movies. Cryo-EM data processing was performed with cryoSPARC ( Punjani et al., 2017 ). 3 μl of purified HTR2B-LSD/miniGq or HTR2B-LSD/Fab complexes at concentrations of 8.5 mg/mL or 6.5 mg/mL, were applied to glow-discharged (45–50 seconds at 10–15 mA) UltrAuFoil holey gold grids (Quantifoil, Au300-R1.2/1.3) in 100% humidity at 18°C or 4°C, respectively. Samples were blotted for 1 second and plunged-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Cryo-EM 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 55,000 × (0.8677 Å/pixel) in counting mode using SerialEM ( Mastronarde, 2005 ). For the HTR2B-LSD/miniGq/i complex, 3,493 movies, dose fractioned over 57 frames, were recorded for 0.044 sec/frame for a total dose of 61 electrons/Å2 in super-resolution mode with a defocus range of 0.8–1.8 μm. For HTR2B-LSD/Fab, 12,303 movies dose fractioned over 50 frames, were recorded for 0.05 sec/frame for a total dose of 61.5 electrons/Å2 in super-resolution mode with a defocus range of 0.7–1.8 μm. Initial sets of 8,491,352 and 3,214,021 particles were selected and subjected to multiple 2D and 3D classification rounds for HTR2B-LSD/Fab or HTR2B-LSD/miniGq complexes, respectively, using cryoSPARC ( Punjani et al., 2017 ). Subsets of 665,475 and 772,614 particles contributing to the HTR2B-LSD/Fab and HTR2B-LSD/miniGq reconstructions, went through global CTF refinement, non-uniform and homogeneous refinement. The HTR2B-LSD/Fab map was further refined locally obtaining a 2.7 Å resolution map for the LSD bound receptor. The HTR2B-LSD/miniGq global map was refined locally obtaining separate 2.9 Å resolution maps for the LSD bound receptor and the G protein. Maps resulting from the local refinements for the HTR2B-LSD or HTR2B-LSD/miniGq complexes were sharpened using DeepEMhancer ( Sanchez-Garcia et al., 2021 ). A total of 4,415,425 particles, 3,544,745 from holey gold and 870,680 from holey carbon grids, were extracted from the corrected 7,750 micrographs for the HTR2B/β-arrestin-1 complex. After 2D and 3D classification, a subset of 126,485 particles were subjected to homogeneous refinement followed by local refinements of the LSD bound receptor (with an scFv fragment of the HTR2B Fab P2C2 used for particle alignment) and β-arrestin-1 at a resolution of 3.3 Å. Maps resulting from local refinements for HTR2B-LSD/miniGq or HTR2B/β-arrestin-1 were combined in Chimera ( Pettersen et al., 2004 ). Flowcharts describing data processing steps are presented in the Figures S2 – S4 .

Model building and refinement

Coordinates for HTR2B-LSD (PDB ID: 5TVN) ( Wacker et al., 2017 ), miniGq (PDB: 6WHA) ( Kim et al., 2020 ) and β-arrestin-1 (PDB: 6UP7) ( Huang et al., 2020 ) were used as an initial models for docking into the EM density maps using Chimera ( Pettersen et al., 2004 ). Models were subjected to iterative rounds of manual refinement in Coot ( Emsley et al., 2010 ) and real-space refinement in Phenix. Validation of cryo-EM maps and models was performed with Phenix ( Liebschner et al., 2019 ) comprehensive cryo-EM validation. Model statistics were validated with Molprobity ( Chen et al., 2010 ) and final refinement statistics are provided in Supplemental Table 1 . Map/model visualizations and figure preparation were done in UCSF Chimera ( Pettersen et al., 2004 ) and ChimeraX ( Pettersen et al., 2021 ). Bioluminescence resonance energy transfer assays (BRET1) For the HTR2B-mediated b-arrestin1 and miniGq recruitment assays, HEK293T cells were co-transfected with C-terminal RLuc8-tagged human HTR2B, and either N-terminal Venus-tagged β-arrestin-1 in a 1:8 ratio or N-terminal Venus-tagged miniGq in a 1:4 ratio. After 8 hours, transfected cells were plated in poly-lysine coated 96-well white clear bottom cell culture plates in plating media (DMEM + 1% (v/v) dialyzed FBS) at a density of 25–50,000 cells in 200 μl per well and incubated overnight. The next day, media was decanted and cells were washed with 60 μl of drug buffer (1xHBSS, 20 mM HEPES, 0.1% (w/v) BSA, 0.01% (w/v) ascorbic acid, pH 7.4). Afterward, 60 μl of drug buffer and 30 μl of the drug (3X) were added per well, and then the plates were transferred to the 37 °C incubator and kept for 20 minutes. Before reading, 10 μl of the coelenterazine h (Promega, final concentration is 5 mM) was added per well and the plate was incubated for an additional 10 minutes to allow for the substrate diffusion. Plates were read for both luminescence at 475 nm and fluorescent eYFP emission at 535 nm for 1 s per well using a PHERAstar FSX multimode microplate reader. The ratio of eYFP/RLuc was calculated per well and analyzed in Graphpad Prism 9 (Graphpad Software Inc., San Diego, CA). Mutagenesis data were normalized to WT stimulation and reanalyzed using nonlinear regression ‘log(agonist) vs. response’ in GraphPad Prism 9. The protein expression of WT and mutant HTR2B measured by the Rluc8 counts of each construct are shown in Figure S7 .

Mass spectrometry analysis

For the protein digestion, 100 μg HTR2B-β-arrestin-1-scFv30 protein was denatured in 2 M urea, 100 mM Tris-HCl (pH 8.0), followed by reduction in 5 mM dithiothreitol (DTT) at 37°C for 30min and alkylation in 10 mM iodoacetamide in the dark for 30 min. The protein mixture was digested with 2 μg Trypsin at 37°C for 6 h and additional 1 μg Asp-N for overnight digestion at 37°C. The digestion reaction was quenched by a final 0.5% trifluoroacetic acid (TFA) and desalted using SepPak C 18 cartridges (Waters, Milford, MA). To achieve Phosphopeptide enrichment, 100 μl (20 μL per sample) of Ni-NTA magnetics beads (QIAGEN) was washed with 3 × 100 μl H 2 O, incubated with 100 μl of 50 mM EDTA (pH 8.0) for 30 min, washed with 3 × 100 μl H 2 O, incubated with 100 μl of 50 mM FeCl 3 for 30 min, and washed with 3 × 100 μl 0.1% TFA in 80% acetonitrile (ACN). Beads were resuspended in 150 μl of 80% ACN/0.1% TFA. 100 μg protein digests were resuspended in 150 μl 80% ACN/0.1% TFA and incubated with equilibrated beads for 30 min. Beads were washed with 3 × 150 μl 80% ACN/0.1% TFA. Phosphopeptides were eluted with 50 μl of 50% ACN/0.75% ammonium hydroxide and acidified with 30 μl of 75% ACN/10% formic acid. The samples were then analyzed on an Orbitrap Exploris 480 mass spectrometry system (Thermo Fisher Scientific) equipped with an Easy nLC 1200 ultrahigh pressure liquid chromatography system (Thermo Fisher Scientific). Digested peptide samples were loaded onto a C18 reverse phase column (15 cm × 75 μm packed with BEH 1.7 μm particles). Mobile phase A consisted of 0.1% formic acid (FA) and mobile phase B consisted of 0.1% FA/80% ACN. Peptides were separated by an organic gradient from 4% to 16% mobile phase B over 30 minutes, followed by an increase to 28% B over 20 minutes and 44% B over 10 min, then held at 90% B for 8 minutes at a flow rate of 300 nl/min. FTMS survey scans of peptide precursors from 350 to 1250 m/z were performed in the Orbitrap at 120K resolving power with a normalized AGC target of 300%, an RF lens setting of 40%, and auto maximum injection time. Using a data-dependent acquisition mode, the 20 most abundant ions at charge states 2–6 were fragmented by higher energy collisional dissociation (HCD) with an isolation width of 1.6 Da, a normalized collision energy (NCE) of 30%, a normalized AGC target of 250% at a resolving power of 15,000 with a maximum injection time of 40 ms. The proteomic data was searched against the human UniProt database augmented with the sequence of beta-arrestin 1 fused to HTR2B. Peptide and protein identification were searched using the default setting in MaxQuant (version 1.6.12.0) with variable modification of methionine oxidation and phosphorylation of serine, threonine, and tyrosine, and static modification of cysteine carbamidomethylation. All peptide and protein identification were filtered to a 1% false discovery rate. The phosphorylation probabilities of HTR2B C-tail residues are shown in Table S5 . The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE ( Perez-Riverol et al., 2019 ) partner repository with the dataset identifier PXD030752. Molecular Dynamics (MD) Simulations Simulation setup: We performed simulations in 5 distinct conditions: (A) the HTR2B/β-arrestin-1/LSD complex (12 independent simulations, roughly 2 μs each), (B) the HTR2B/LSD transducer-free receptor (12 independent simulations, roughly 2 μs each), (C) the HTR2B/Gq/LSD complex (12 independent simulations, roughly 3 μs each), (D) the HTR2B/LSD receptor from the HTR2B/β-arrestin-1/LSD complex, but with β-arrestin-1 removed (12 independent simulations, roughly 2 μs each) and (E) the HTR2B/LSD receptor from the HTR2B/Gq /LSD complex, but with Gq removed (12 independent simulations, roughly 2 μs each). For all conditions, the initial structures were based on the three cryoEM structures reported in this paper and were prepared using Maestro (Schrödinger, LLC). Condition A was prepared as follows. The scFv30 and Fab nanobodies were removed from the initial structure. The HTR2B construct mutations M144 3.41 W, K247 5.68 V, and E319 6.30 L as well as the β-arrestin-1 construct mutation R169E were reverted to WT. The unresolved residues in the ICL3 replacement (RLLSGSR) of HTR2B construct G252 and S253, the unresolved residues that connect helix 8 with the resolved part of the C-tail in the HTR2B construct (NYRATKSVKTPMRLRSST) as well as the unresolved residue G137 in β-arrestin-1 were modeled using Maestro’s ‘crosslink’ tool. Condition B was prepared as follows. The Fab P2C2 nanobody used for structure determination was not included in the initial structure. The HTR2B construct mutations M144 3.41 W and K247 5.68 V were reverted to WT. The unresolved residues of ICL2 (I161 34.51 -Q162 34.52 -A163 34.53 -N164 34.54 -Q165 34.55 -Y166 34.56 -N167 34.57 ) were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN) ( Wacker et al., 2017 ). To achieve a smooth transition between the intracellular tips of TM3 and TM4 with ICL2, K159 3.56 -P160 34.50 of TM3 and S168 4.38 -R169 4.39 of TM4 in the transducer-free cryoEM structure were replaced by the corresponding residues in the HTR2B/LSD crystal structure (PDB ID: 5TVN). The unresolved residues of the intracellular tips of TM5 and TM6 as well as the connecting ICL3 replacement (RLLSGSR) were modeled as follows: L244 5.65 -Q245 5.66 -K246 5.67 and N318 6.29 -E319 6.30 -Q320 6.31 -R321 6.32 were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN). To achieve a smooth transition, A243 5.64 and A322 6.33 -S323 6.34 in the transducer-free cryoEM structure were replaced by the corresponding residues in the HTR2B/LSD crystal structure (PDB ID: 5TVN). The residues K247 5.68 -R248-L249-L250-S251-G252-S253-R254-Q314 6.25 -T315 6.26 -I316 6.27 -S317 6.28 , which include the ICL3 replacement (RLLSGSR) were modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure (including modeling G252 and S253 with Maestro’s ‘crosslink’ tool and reverting the construct mutation K247 5.68 V to WT). The unresolved C-terminal residues I395 8.58 -T396-C397 of helix8 were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN). Condition C was prepared as follows. The scFv16 nanobody was removed from the initial structure. The missing parts of the N-terminus of the Gα subunit and the C-terminus of the Gγ subunit were modeled in and lipidated as follows: G2 of the Gα subunit was myristoylated, C3 of the Gα subunit was palmitoylated, and C68 of the Gγ subunit was prenylated ( Wedegaertner et al., 1995 ). The unresolved residues G88-Q89 in the Gα subunit were modeled with Maestro’s ‘crosslink’ tool. The HTR2B construct mutation M144 3.41 W was reverted to WT. The HTR2B/LSD-Gq construct contains the WT ICL3, which is not resolved in the cryoEM map. For consistency, we included the ICL3 replacement also in the HTR2B/LSD-Gq complex for simulations and modeled it based on the HTR2B/LSD-β-arrestin-1 structure. More specifically, Q245 5.66 -K246 5.67 -K247 5.68 -R248-L249-L250-S251-G252-S253-R254-Q314 6.25 -T315 6.26 -I316 6.27 -S317 6.28 -N318 6.29 , were modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure (including modeling G252 and S253 with Maestro’s ‘crosslink’ tool and reverting the construct mutation K247 5.68 V to WT). Moreover, the C-terminal residues of helix 8 (G392 8.55 -R393 8.56 -Y394 8.57 -I395 8.58 -T396-C397) were also modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure to include the palmitoylated membrane anchor residue C397. Condition D was obtained from condition A by removing β-arrestin-1. Condition E was obtained from condition C by removing Gq. For all simulation conditions, a palmitoyl group was added to residue C397 of the receptor using Maestro (Schrödinger, LLC). Missing amino acid side chains were modeled using Prime (Schrödinger, LLC). Neutral acetyl and methylamide groups were added to cap the N- and C-termini, respectively, of the protein chains, except for the N-myristoylated N-terminus of the Gα subunit and the natural C-termini of the Gα and Gβ subunit, which were carboxylated. Titratable residues were kept in their dominant protonation state at pH 7, except for D100 2.50 and D152 3.49 , which were protonated to their neutral form, as studies indicate that these conserved residues are protonated in active class-A GPCRs ( Ghanouni et al., 2000 , Ranganathan et al., 2014 ). Histidine residues were modeled as neutral, with a hydrogen atom bound to either the delta or epsilon nitrogen depending on which tautomeric state optimized the local hydrogen-bonding network. Dowser ( Zhang and Hermans, 1996 ) was used to add water molecules to protein cavities. The LSD tertiary amine nitrogen was protonated, corresponding to the dominant protonation state at pH 7.0 and enabling formation of the conserved salt bridge with neighboring D135 3.32 . The receptor in the HTR2B/LSD transducer-free system and the HTR2B/LSD-β-arrestin-1 system was aligned on the receptor in the crystal structure of HTR2B/LSD (PDB ID: 5TVN) in the Orientation of Proteins in Membranes (OPM) database ( Lomize et al., 2006 ), the receptor of the HTR2B/LSD-Gq system was aligned on the receptor in the HTR2A/NBOH-Gq cryoEM structure (PDB ID: 6WHA) ( Kim et al., 2020 ) in the OPM database. The aligned structures were inserted into a pre-equilibrated palmitoyloleoyl-phosphatidylcholine (POPC) membrane bilayer using Dabble ( RM, 2017 ). Sodium and chloride ions were added to neutralize each system at a concentration of 150 mM. The final system of condition A comprised 218,598 atoms, including 365 lipid molecules and 52,858 water molecules (initial system dimensions: 120 Å × 120 Å × 150 Å). The final system of condition B comprised 62,370 atoms, including 150 lipid molecules and 12,505 water molecules (initial system dimensions: 99 Å × 69 Å × 93 Å). The final system of condition C comprised 290,075 atoms, including 464 lipid molecules and 70,953 water molecules (initial system dimensions: 139 Å × 123 Å × 166 Å). The final system of condition D comprised 59,036 atoms, including 141 lipid molecules and 11,797 water molecules (initial system dimensions: 79 Å × 84 Å × 92 Å). The final system of condition E comprised 54,146 atoms, including 126 lipid molecules and 10,839 water molecules (initial system dimensions: 74 Å × 82 Å × 92 Å). Simulation protocols: For each simulation, initial atom velocities were assigned randomly and independently. We employed the CHARMM36m force field for protein molecules, the CHARMM36 parameter set for lipid molecules and salt ions, and the associated CHARMM TIP3P model for water ( Huang et al., 2017 , Klauda et al., 2010 ). Parameters for LSD were taken from ( Wacker et al., 2017 ). Simulations were run using the AMBER20 software ( D.A. Case, 2021 ) under periodic boundary conditions with the Compute Unified Device Architecture (CUDA) version of Particle-Mesh Ewald Molecular Dynamics (PMEMD) on one GPU ( Salomon-Ferrer et al., 2013 ). After energy minimization, the systems were first heated over 12.5 ps from 0 K to 100 K in the NVT ensemble using a Langevin thermostat with harmonic restraints of 10.0 kcal∙mol −1 ∙Å −2 on the non-hydrogen atoms of the lipids, protein, and ligand. Initial velocities were sampled from a Boltzmann distribution. The systems were then heated to 310 K over 125 ps in the NPT ensemble. Equilibration was performed at 310 K and 1 bar in the NPT ensemble, with harmonic restraints on the protein and ligand non-hydrogen atoms tapered off by 1.0 kcal∙mol −1 ∙Å −2 starting at 5.0 kcal∙mol −1 ∙Å −2 in a stepwise manner every 2 ns for 10 ns, and finally by 0.1 kcal∙mol −1 ∙Å −2 every 2 ns for an additional 18 ns. All restraints were completely removed during production simulation. Production simulations were performed at 310 K and 1 bar in the NPT ensemble using the Langevin thermostat and Monte Carlo barostat. Lengths of bonds to hydrogen atoms were constrained using SHAKE, and the simulations were performed using a timestep of 4.0 fs while using hydrogen mass repartitioning ( Hopkins et al., 2015 ). Non-bonded interactions were cut off at 9.0 Å, and long-range electrostatic interactions were calculated using the particle-mesh Ewald (PME) method with an Ewald coefficient (β) of approximately 0.31 Å and B-spline interpolation of order 4. The PME grid size was chosen such that the width of a grid cell was approximately 1 Å. Snapshots of the trajectory were saved every 200 ps. Simulation analysis protocols The AmberTools17 CPPTRAJ package ( Roe and Cheatham, 2013 ) was used to reimage trajectories at 1 ns per frame, Visual Molecular Dynamics (VMD) ( Humphrey et al., 1996 ) was used for visualization and analysis, and PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC) was used for renderings. Plots of time traces from representative simulations were generated with Matplotlib ( Hunter, 2007 ) and show both original, unsmoothed traces (transparent lines) and traces smoothed with a moving average (thick lines), using an averaging window of 20 ns. All traces include the initial equilibration with harmonic restraints on the protein and ligand non-hydrogen atoms.

QUANTIFICATION AND STATISTICAL ANALYSIS

For the β-arrestin-1 and miniGq recruitment assay, the data was analyzed in Graphpad Prism 9.0 using nonlinear regression “log(agonist) vs. response”. 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.

Materials Availability

All the plasmids and cells generated from this study could be obtained directly from the Lead Contact with a completed Materials Transfer Agreement if there is potential for commercial application. All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Two eukaryotic cell lines, Spodoptera frugiperda ( Sf9 , Expression systems) cells and HEK293T cells (ATCC), were used in this study. Sf9 cells suspension in ESF-921 medium were purchased from Expression systems and used for protein expression of HTR2B for structural study. Sf9 cells were grown in ESF-921 medium at 27°C, 120 rpm without further validation. HEK293T cells were purchased from the American Type Culture Collection (ATCC, CRL-11268) and used for Gq and β-arrestin-1 recruitment assay in this study. HEK293T cells were grown in a humidified 37°C incubator with 5% CO 2 using DMEM medium (VWR, #45000) supplemented with 10% (v/v) fetal bovine serum (FBS, VWR, #89510–186) and 100 I.U./mL penicillin and 100 mg/mL streptomycin. Before cell plating, the DMEM medium were changed from 10% FBS to 1% (v/v) dFBS to remove serotonin. HEK293T cells were authenticated by the supplier (ATCC) using morphology, growth characteristics and STR profiling.

METHODS DETAILS Constructs for structural studies To get the structure of HTR2B in complex with β-arrestin-1, the ICL3 and C-terminus of HTR2B were truncated based on BRET1 recruitment assay as noted in the main text. In detail, residues A248-V313 in the ICL3 of HTR2B were removed and replaced with a 7 residues linker “RLLSGSR”. For the C-terminus, the middle part (L408-S444) and the last 17 residues (L465-V481) were removed to make a C-tail chimera. Mutations K247 5.68 V and E319 6.30 L were added to the construct to increase the β-arrestin-1 recruitment by LSD. A modified thermostabilized apocytochrome b562RIL (BRIL) as a fusion partner was fused before T36 of HTR2B with M144 3.41 W mutation adapted from the previous construct ( Wacker et al., 2013 ). β-arrestin-1 isoform 2 with the R169E constitutive mutation was then fused to the C-terminus of HTR2B with a 4 × GSA linker. At last, an engineered scFv30 was directly fused after L368 of β-arrestin-1 to achieve a tandem expression. The HTR2B-β-arrestin-1-scFv30 chimera was cloned into a pFastBac1 vector containing a haemagglutinin (HA) signal sequence followed by FLAG-tag, His10-tag and TEV protease site at the N-terminus. The HTR2B construct used for transducer-free HTR2B has the same sequence as the receptor portion used for HTR2B-β-arrestin-1-scFv30 chimera, except that it does not have the E319 6.30 L mutation and the C-terminus is further truncated to C405 according to a previous crystallography construct ( Wacker et al., 2013 ). Fab P2C2 was cloned into a pFastBac-Dual vector with a honeybee signal peptide at the N-terminus of the light chain and a GP67 signal peptide at the N-terminus of the heavy chain ( Ishchenko et al., 2017 ). A 6 × His-tag was added to the C-terminus of heavy chain to facilitate protein purification. The HTR2B construct used for the HTR2B-Gq complex study contains residues 36–405 and a M144 3.41 W mutation. A N-terminal BRIL as a fusion partner was added before T36 of HTR2B. The N-bril fused receptor portion was then subcloned into a same vector used for HTR2B-β-arrestin-1-scFv30 complex. Constructs of scFv16 and the heterotrimeric mini-Gαq protein complex were the same as we used in our previous HTR2A-Gq structure ( Kim et al., 2020 ).

Expression and purification of Fab P2C2 and scFv16

Bac-to-Bac expression system was used to generate the baculovirus for all the protein expressions in this study. Before infection, virus titer was determined by flow-cytometric analysis using gp64-PE antibody (Expression systems) stained cells. For Fab P2C2, Sf9 cells at a density of 2 million cells per ml were infected with P1 virus at a multiplicity of infection (MOI) of 3. Supernatant containing the secreted P2C2 was collected at 96 h post-infection. Tris powder was then added to adjust the medium to pH 7.8. Chelating agents were quenched by addition of 1 mM nickel chloride and 5 mM calcium chloride and stirring in cold room for 1 hour. After another centrifugation, 1 ml His60 Ni Superflow Resin (Takara) was added to the supernatant for overnight binding at 4°C. The resin was collected next day and washed with 20 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 10 mM imidazole. The protein was eluted with a same buffer containing 250 mM imidazole and further purified by size exclusion chromatography using a Superdex 200 16/60 column (GE healthcare). The peak fractions were collected and concentrated to 1 mg ml-1 for future use. scFv16 was expressed and purified using the same protocol. Protein expression and purification of HTR2B-β-arrestin-1-scFv30 complex For HTR2B-β-arrestin-1-scFv30 chimera, Sf9 cells at a density of 2 million cells per were infected with P1 virus of HTR2B-β-arrestin-1-scFv30 and GRK2 at a multiplicity of infection (MOI) ratio of 3:1.5. Cells were harvested by centrifugation at 48 h post-infection. The cell pellet was then washed with a low-salt buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 and 20 mM KCl and proteinase inhibitor containing 500 mM AEBSF, 1 mM E-64, 1 mM Leupeptin and 0.15 mM Aprotinin. Subsequently, two rounds of high-salt wash were performed using a buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl and 500 mM NaCl to remove membrane associated proteins. Purified cell membranes were resuspended in low-salt buffer and incubated with 10 μM LSD and 200 μM TCEP for 30 min at room temperature. After additional 1 hour incubation in cold room, the cell membranes were solubilized using solubilization buffer at a final concentration of 30 mM HEPES, 100 mM NaCl, 10 % glycerol, 5 μM LSD, 100 μM TCEP and 0.6% (w/v) n-dodecyl-beta-D-maltopyranoside (DDM, Anatrace), 0.2% (w/v) cholesteryl hemisuccinate (CHS, Sigma) at 4°C for 2.5 h. The solubilized proteins in the supernatants were isolated by ultra-centrifugation at 40,000 rpm for 30 min, and then incubated at 4°C overnight with 1 ml TALON IMAC resin (Clontech) and 20 mM imidazole. The resin was collected next day and washed with 10 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% glycerol, 1 μM LSD and 100 μM TCEP. The resin was then incubated with 20 mM HEPES pH 7.5, 100 mM NaCl, 0.8% Lauryl Maltose Neopentyl Glycol (LMNG, Anatrace), 0.27% GDN (Anatrace), 0.08% CHS, 5% glycerol, 1 μM LSD and 100 μM TCEP to exchange the detergent from DDM/CHS to LMNG/GDN/CHS. The resin was washed with additional 15 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 0.01% LMNG, 0.033% GDN, 0.001% CHS, 5% glycerol, 1 μM LSD, 100 μM TCEP and 30 mM imidazole. The protein was then eluted with 4.5 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 0.01% LMNG, 0.033% GDN, 0.001% CHS, 5% glycerol, 5 μM LSD, 100 μM TCEP and 250 mM imidazole. Eluted protein was concentrated to 400 μl and incubated with 100 μl P2C2 Fab (1 mg/ml) at 4°C for 2 h before subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES pH 7.5, 100 mM NaCl, 1 μM LSD, 100 μM TCEP, 0.001% (w/v) MNG, 0.00033 (w/v) GDN and 0.0001% (w/v) CHS. Peak fractions were then collected and concentrated to 3.5 mg ml-1 to make the cryoEM grids. Protein expression and purification of HTR2B-P2C2 complex For the transducer-free HTR2B-P2C2 complex, Sf9 cells at a density of 2 million cells per were infected with receptor P1 virus at a multiplicity of infection (MOI) of 3. Cells were harvested by centrifugation at 48 h post-infection. Followed by a low-salt wash, the cell membranes were washed twice by high-salt buffer containing 10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl and 1 M NaCl. The membrane was solubilized and binding overnight using a same protocol as HTR2B-β-arrestin-1-scFv30, except TCEP was not added. The resin was collected next day and washed with 25 column volumes buffer containing 20 mM HEPES pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.05% (w/v) DDM, 0.01% (w/v) CHS, 10% glycerol and 1 μM LSD. The protein was then eluted using a same buffer containing 250 mM imidazole. Eluted protein was concentrated to 350 μl and incubated with 150 μl P2C2 Fab (1 mg/ml) at 4°C for 2 h before subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES pH 7.5, 100 mM NaCl, 1 μM LSD, 0.001% (w/v) MNG, 0.00033 (w/v) GDN and 0.0001% (w/v) CHS. Peak fractions were then collected and concentrated to 6.5 mg ml−1 to make the cryoEM grids. Protein expression and purification of HTR2B-Gq complex The cell pellet of HTR2B and Gq complex from 2 L culture was thawed at room temperature and resuspended in law salt buffer containing 10 mM HEPES pH 7.5, 50 mM NaCl, 20 mM KCl, and protease inhibitor 500 μM AEBSF, 1 μM E-64, 1 μM Leupeptin and 0.15 μM Aprotinin. The HTR2B and Gq complexes were formed on the membrane in the presence of 5 μM LSD, followed by incubation for 1.5 h at room temperature. Cell membranes were collected by ultra-centrifugation at 40,000 rpm for 25 min. The membranes were then resuspended and solubilized in solubilization buffer containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 5 mM MgCl2, 20 mM imidazole, 0.1 mM TCEP, 10%(v/v) glycerol, 0.5% DDM, 0.1%(w/v) CHS, 10 μM LSD, and protease inhibitor cocktail at 4°C. After 3 h incubation, the supernatant was isolated by centrifugation at 70,000 rpm for 50 min and then incubated overnight with pre-equilibrated TALON resin (CLONTECH) at 4°C. The resin was collected and washed with 30 column volumes with washing buffer 20 mM HEPES, pH 7.5, 100 mM NaCl, 30 mM imidazole, 0.1 %(w/v) LMNG, 0.01 % (w/v) CHS and 10 μM LSD. The protein was then eluted using the same buffer supplemented with 300 mM imidazole. 500 μl of protein sample was applied to PD MiniTrap G-25 columns (GE Healthcare) to remove imidazole and change the detergent to LMNG 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 μM TCEP, and 20 μM LSD. The N-terminal BRIL protein was removed by the addition of His-tagged PreScission protease (Genescript) and incubation overnight at 4 °C. Unnecessary his-tagged proteins such as protease, BRIL, and free his tag were removed by equilibrated TALON resin, and the flow-through was collected. The eluate protein was concentrated and subjected to size-exclusion chromatography on a Superdex 200 Increase 10/300 column (GE Healthcare) that was pre-equilibrated with 20 mM HEPES, pH 7.5, 100 mM NaCl, 10 μM LSD (agonist), 0.00075% (w/v) LMNG, 0.00025% (w/v) GDN, and 0.000075% (w/v) CHS. The peak fractions of complex were pooled and concentrated to 8.5 mg/ml for electron microscopy experiments.

Cryo-EM data acquisition and processing for HTR2B-LSD complexes

Particles from two cryo-EM data collections of the same sample vitrified on UltrAuFoil holey gold (Quantifoil, Au300-R1.2/1.3) or Quantifoil holey carbon (R1.2/1.3) grids contributed to the HTR2B/β-arrestin-1 complex reconstruction presented here. For this, 3 μl of purified HTR2B/β-arrestin-1 complex at a concentration of 3.5 mg/mL were applied to the glow-discharged (45–50 seconds at 10–15 mA) grids in 100% humidity at 18°C. Samples were blotted for 1 second and plunged-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Cryo-EM 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 57,050 × (0.8521 Å/pixel) in counting mode using SerialEM ( Mastronarde, 2005 ). The collection on holey gold or carbon grids generated 6,217 movies and 1,533 movies, respectively, dose fractioned over 50 frames, recorded for 0.05 sec/frame for a total dose of 68 electrons/Å2 in super-resolution mode with a defocus range of 0.8–1.8 μm, for a total of 7,750 movies. Cryo-EM data processing was performed with cryoSPARC ( Punjani et al., 2017 ). 3 μl of purified HTR2B-LSD/miniGq or HTR2B-LSD/Fab complexes at concentrations of 8.5 mg/mL or 6.5 mg/mL, were applied to glow-discharged (45–50 seconds at 10–15 mA) UltrAuFoil holey gold grids (Quantifoil, Au300-R1.2/1.3) in 100% humidity at 18°C or 4°C, respectively. Samples were blotted for 1 second and plunged-frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Cryo-EM 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 55,000 × (0.8677 Å/pixel) in counting mode using SerialEM ( Mastronarde, 2005 ). For the HTR2B-LSD/miniGq/i complex, 3,493 movies, dose fractioned over 57 frames, were recorded for 0.044 sec/frame for a total dose of 61 electrons/Å2 in super-resolution mode with a defocus range of 0.8–1.8 μm. For HTR2B-LSD/Fab, 12,303 movies dose fractioned over 50 frames, were recorded for 0.05 sec/frame for a total dose of 61.5 electrons/Å2 in super-resolution mode with a defocus range of 0.7–1.8 μm. Initial sets of 8,491,352 and 3,214,021 particles were selected and subjected to multiple 2D and 3D classification rounds for HTR2B-LSD/Fab or HTR2B-LSD/miniGq complexes, respectively, using cryoSPARC ( Punjani et al., 2017 ). Subsets of 665,475 and 772,614 particles contributing to the HTR2B-LSD/Fab and HTR2B-LSD/miniGq reconstructions, went through global CTF refinement, non-uniform and homogeneous refinement. The HTR2B-LSD/Fab map was further refined locally obtaining a 2.7 Å resolution map for the LSD bound receptor. The HTR2B-LSD/miniGq global map was refined locally obtaining separate 2.9 Å resolution maps for the LSD bound receptor and the G protein. Maps resulting from the local refinements for the HTR2B-LSD or HTR2B-LSD/miniGq complexes were sharpened using DeepEMhancer ( Sanchez-Garcia et al., 2021 ). A total of 4,415,425 particles, 3,544,745 from holey gold and 870,680 from holey carbon grids, were extracted from the corrected 7,750 micrographs for the HTR2B/β-arrestin-1 complex. After 2D and 3D classification, a subset of 126,485 particles were subjected to homogeneous refinement followed by local refinements of the LSD bound receptor (with an scFv fragment of the HTR2B Fab P2C2 used for particle alignment) and β-arrestin-1 at a resolution of 3.3 Å. Maps resulting from local refinements for HTR2B-LSD/miniGq or HTR2B/β-arrestin-1 were combined in Chimera ( Pettersen et al., 2004 ). Flowcharts describing data processing steps are presented in the Figures S2 – S4 .

Model building and refinement

Coordinates for HTR2B-LSD (PDB ID: 5TVN) ( Wacker et al., 2017 ), miniGq (PDB: 6WHA) ( Kim et al., 2020 ) and β-arrestin-1 (PDB: 6UP7) ( Huang et al., 2020 ) were used as an initial models for docking into the EM density maps using Chimera ( Pettersen et al., 2004 ). Models were subjected to iterative rounds of manual refinement in Coot ( Emsley et al., 2010 ) and real-space refinement in Phenix. Validation of cryo-EM maps and models was performed with Phenix ( Liebschner et al., 2019 ) comprehensive cryo-EM validation. Model statistics were validated with Molprobity ( Chen et al., 2010 ) and final refinement statistics are provided in Supplemental Table 1 . Map/model visualizations and figure preparation were done in UCSF Chimera ( Pettersen et al., 2004 ) and ChimeraX ( Pettersen et al., 2021 ). Bioluminescence resonance energy transfer assays (BRET1) For the HTR2B-mediated b-arrestin1 and miniGq recruitment assays, HEK293T cells were co-transfected with C-terminal RLuc8-tagged human HTR2B, and either N-terminal Venus-tagged β-arrestin-1 in a 1:8 ratio or N-terminal Venus-tagged miniGq in a 1:4 ratio. After 8 hours, transfected cells were plated in poly-lysine coated 96-well white clear bottom cell culture plates in plating media (DMEM + 1% (v/v) dialyzed FBS) at a density of 25–50,000 cells in 200 μl per well and incubated overnight. The next day, media was decanted and cells were washed with 60 μl of drug buffer (1xHBSS, 20 mM HEPES, 0.1% (w/v) BSA, 0.01% (w/v) ascorbic acid, pH 7.4). Afterward, 60 μl of drug buffer and 30 μl of the drug (3X) were added per well, and then the plates were transferred to the 37 °C incubator and kept for 20 minutes. Before reading, 10 μl of the coelenterazine h (Promega, final concentration is 5 mM) was added per well and the plate was incubated for an additional 10 minutes to allow for the substrate diffusion. Plates were read for both luminescence at 475 nm and fluorescent eYFP emission at 535 nm for 1 s per well using a PHERAstar FSX multimode microplate reader. The ratio of eYFP/RLuc was calculated per well and analyzed in Graphpad Prism 9 (Graphpad Software Inc., San Diego, CA). Mutagenesis data were normalized to WT stimulation and reanalyzed using nonlinear regression ‘log(agonist) vs. response’ in GraphPad Prism 9. The protein expression of WT and mutant HTR2B measured by the Rluc8 counts of each construct are shown in Figure S7 .

Mass spectrometry analysis

For the protein digestion, 100 μg HTR2B-β-arrestin-1-scFv30 protein was denatured in 2 M urea, 100 mM Tris-HCl (pH 8.0), followed by reduction in 5 mM dithiothreitol (DTT) at 37°C for 30min and alkylation in 10 mM iodoacetamide in the dark for 30 min. The protein mixture was digested with 2 μg Trypsin at 37°C for 6 h and additional 1 μg Asp-N for overnight digestion at 37°C. The digestion reaction was quenched by a final 0.5% trifluoroacetic acid (TFA) and desalted using SepPak C 18 cartridges (Waters, Milford, MA). To achieve Phosphopeptide enrichment, 100 μl (20 μL per sample) of Ni-NTA magnetics beads (QIAGEN) was washed with 3 × 100 μl H 2 O, incubated with 100 μl of 50 mM EDTA (pH 8.0) for 30 min, washed with 3 × 100 μl H 2 O, incubated with 100 μl of 50 mM FeCl 3 for 30 min, and washed with 3 × 100 μl 0.1% TFA in 80% acetonitrile (ACN). Beads were resuspended in 150 μl of 80% ACN/0.1% TFA. 100 μg protein digests were resuspended in 150 μl 80% ACN/0.1% TFA and incubated with equilibrated beads for 30 min. Beads were washed with 3 × 150 μl 80% ACN/0.1% TFA. Phosphopeptides were eluted with 50 μl of 50% ACN/0.75% ammonium hydroxide and acidified with 30 μl of 75% ACN/10% formic acid. The samples were then analyzed on an Orbitrap Exploris 480 mass spectrometry system (Thermo Fisher Scientific) equipped with an Easy nLC 1200 ultrahigh pressure liquid chromatography system (Thermo Fisher Scientific). Digested peptide samples were loaded onto a C18 reverse phase column (15 cm × 75 μm packed with BEH 1.7 μm particles). Mobile phase A consisted of 0.1% formic acid (FA) and mobile phase B consisted of 0.1% FA/80% ACN. Peptides were separated by an organic gradient from 4% to 16% mobile phase B over 30 minutes, followed by an increase to 28% B over 20 minutes and 44% B over 10 min, then held at 90% B for 8 minutes at a flow rate of 300 nl/min. FTMS survey scans of peptide precursors from 350 to 1250 m/z were performed in the Orbitrap at 120K resolving power with a normalized AGC target of 300%, an RF lens setting of 40%, and auto maximum injection time. Using a data-dependent acquisition mode, the 20 most abundant ions at charge states 2–6 were fragmented by higher energy collisional dissociation (HCD) with an isolation width of 1.6 Da, a normalized collision energy (NCE) of 30%, a normalized AGC target of 250% at a resolving power of 15,000 with a maximum injection time of 40 ms. The proteomic data was searched against the human UniProt database augmented with the sequence of beta-arrestin 1 fused to HTR2B. Peptide and protein identification were searched using the default setting in MaxQuant (version 1.6.12.0) with variable modification of methionine oxidation and phosphorylation of serine, threonine, and tyrosine, and static modification of cysteine carbamidomethylation. All peptide and protein identification were filtered to a 1% false discovery rate. The phosphorylation probabilities of HTR2B C-tail residues are shown in Table S5 . The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE ( Perez-Riverol et al., 2019 ) partner repository with the dataset identifier PXD030752. Molecular Dynamics (MD) Simulations Simulation setup: We performed simulations in 5 distinct conditions: (A) the HTR2B/β-arrestin-1/LSD complex (12 independent simulations, roughly 2 μs each), (B) the HTR2B/LSD transducer-free receptor (12 independent simulations, roughly 2 μs each), (C) the HTR2B/Gq/LSD complex (12 independent simulations, roughly 3 μs each), (D) the HTR2B/LSD receptor from the HTR2B/β-arrestin-1/LSD complex, but with β-arrestin-1 removed (12 independent simulations, roughly 2 μs each) and (E) the HTR2B/LSD receptor from the HTR2B/Gq /LSD complex, but with Gq removed (12 independent simulations, roughly 2 μs each). For all conditions, the initial structures were based on the three cryoEM structures reported in this paper and were prepared using Maestro (Schrödinger, LLC). Condition A was prepared as follows. The scFv30 and Fab nanobodies were removed from the initial structure. The HTR2B construct mutations M144 3.41 W, K247 5.68 V, and E319 6.30 L as well as the β-arrestin-1 construct mutation R169E were reverted to WT. The unresolved residues in the ICL3 replacement (RLLSGSR) of HTR2B construct G252 and S253, the unresolved residues that connect helix 8 with the resolved part of the C-tail in the HTR2B construct (NYRATKSVKTPMRLRSST) as well as the unresolved residue G137 in β-arrestin-1 were modeled using Maestro’s ‘crosslink’ tool. Condition B was prepared as follows. The Fab P2C2 nanobody used for structure determination was not included in the initial structure. The HTR2B construct mutations M144 3.41 W and K247 5.68 V were reverted to WT. The unresolved residues of ICL2 (I161 34.51 -Q162 34.52 -A163 34.53 -N164 34.54 -Q165 34.55 -Y166 34.56 -N167 34.57 ) were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN) ( Wacker et al., 2017 ). To achieve a smooth transition between the intracellular tips of TM3 and TM4 with ICL2, K159 3.56 -P160 34.50 of TM3 and S168 4.38 -R169 4.39 of TM4 in the transducer-free cryoEM structure were replaced by the corresponding residues in the HTR2B/LSD crystal structure (PDB ID: 5TVN). The unresolved residues of the intracellular tips of TM5 and TM6 as well as the connecting ICL3 replacement (RLLSGSR) were modeled as follows: L244 5.65 -Q245 5.66 -K246 5.67 and N318 6.29 -E319 6.30 -Q320 6.31 -R321 6.32 were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN). To achieve a smooth transition, A243 5.64 and A322 6.33 -S323 6.34 in the transducer-free cryoEM structure were replaced by the corresponding residues in the HTR2B/LSD crystal structure (PDB ID: 5TVN). The residues K247 5.68 -R248-L249-L250-S251-G252-S253-R254-Q314 6.25 -T315 6.26 -I316 6.27 -S317 6.28 , which include the ICL3 replacement (RLLSGSR) were modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure (including modeling G252 and S253 with Maestro’s ‘crosslink’ tool and reverting the construct mutation K247 5.68 V to WT). The unresolved C-terminal residues I395 8.58 -T396-C397 of helix8 were modeled based on the HTR2B/LSD crystal structure (PDB ID: 5TVN). Condition C was prepared as follows. The scFv16 nanobody was removed from the initial structure. The missing parts of the N-terminus of the Gα subunit and the C-terminus of the Gγ subunit were modeled in and lipidated as follows: G2 of the Gα subunit was myristoylated, C3 of the Gα subunit was palmitoylated, and C68 of the Gγ subunit was prenylated ( Wedegaertner et al., 1995 ). The unresolved residues G88-Q89 in the Gα subunit were modeled with Maestro’s ‘crosslink’ tool. The HTR2B construct mutation M144 3.41 W was reverted to WT. The HTR2B/LSD-Gq construct contains the WT ICL3, which is not resolved in the cryoEM map. For consistency, we included the ICL3 replacement also in the HTR2B/LSD-Gq complex for simulations and modeled it based on the HTR2B/LSD-β-arrestin-1 structure. More specifically, Q245 5.66 -K246 5.67 -K247 5.68 -R248-L249-L250-S251-G252-S253-R254-Q314 6.25 -T315 6.26 -I316 6.27 -S317 6.28 -N318 6.29 , were modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure (including modeling G252 and S253 with Maestro’s ‘crosslink’ tool and reverting the construct mutation K247 5.68 V to WT). Moreover, the C-terminal residues of helix 8 (G392 8.55 -R393 8.56 -Y394 8.57 -I395 8.58 -T396-C397) were also modeled based on the HTR2B/LSD/β-arrestin-1 cryoEM structure to include the palmitoylated membrane anchor residue C397. Condition D was obtained from condition A by removing β-arrestin-1. Condition E was obtained from condition C by removing Gq. For all simulation conditions, a palmitoyl group was added to residue C397 of the receptor using Maestro (Schrödinger, LLC). Missing amino acid side chains were modeled using Prime (Schrödinger, LLC). Neutral acetyl and methylamide groups were added to cap the N- and C-termini, respectively, of the protein chains, except for the N-myristoylated N-terminus of the Gα subunit and the natural C-termini of the Gα and Gβ subunit, which were carboxylated. Titratable residues were kept in their dominant protonation state at pH 7, except for D100 2.50 and D152 3.49 , which were protonated to their neutral form, as studies indicate that these conserved residues are protonated in active class-A GPCRs ( Ghanouni et al., 2000 , Ranganathan et al., 2014 ). Histidine residues were modeled as neutral, with a hydrogen atom bound to either the delta or epsilon nitrogen depending on which tautomeric state optimized the local hydrogen-bonding network. Dowser ( Zhang and Hermans, 1996 ) was used to add water molecules to protein cavities. The LSD tertiary amine nitrogen was protonated, corresponding to the dominant protonation state at pH 7.0 and enabling formation of the conserved salt bridge with neighboring D135 3.32 . The receptor in the HTR2B/LSD transducer-free system and the HTR2B/LSD-β-arrestin-1 system was aligned on the receptor in the crystal structure of HTR2B/LSD (PDB ID: 5TVN) in the Orientation of Proteins in Membranes (OPM) database ( Lomize et al., 2006 ), the receptor of the HTR2B/LSD-Gq system was aligned on the receptor in the HTR2A/NBOH-Gq cryoEM structure (PDB ID: 6WHA) ( Kim et al., 2020 ) in the OPM database. The aligned structures were inserted into a pre-equilibrated palmitoyloleoyl-phosphatidylcholine (POPC) membrane bilayer using Dabble ( RM, 2017 ). Sodium and chloride ions were added to neutralize each system at a concentration of 150 mM. The final system of condition A comprised 218,598 atoms, including 365 lipid molecules and 52,858 water molecules (initial system dimensions: 120 Å × 120 Å × 150 Å). The final system of condition B comprised 62,370 atoms, including 150 lipid molecules and 12,505 water molecules (initial system dimensions: 99 Å × 69 Å × 93 Å). The final system of condition C comprised 290,075 atoms, including 464 lipid molecules and 70,953 water molecules (initial system dimensions: 139 Å × 123 Å × 166 Å). The final system of condition D comprised 59,036 atoms, including 141 lipid molecules and 11,797 water molecules (initial system dimensions: 79 Å × 84 Å × 92 Å). The final system of condition E comprised 54,146 atoms, including 126 lipid molecules and 10,839 water molecules (initial system dimensions: 74 Å × 82 Å × 92 Å). Simulation protocols: For each simulation, initial atom velocities were assigned randomly and independently. We employed the CHARMM36m force field for protein molecules, the CHARMM36 parameter set for lipid molecules and salt ions, and the associated CHARMM TIP3P model for water ( Huang et al., 2017 , Klauda et al., 2010 ). Parameters for LSD were taken from ( Wacker et al., 2017 ). Simulations were run using the AMBER20 software ( D.A. Case, 2021 ) under periodic boundary conditions with the Compute Unified Device Architecture (CUDA) version of Particle-Mesh Ewald Molecular Dynamics (PMEMD) on one GPU ( Salomon-Ferrer et al., 2013 ). After energy minimization, the systems were first heated over 12.5 ps from 0 K to 100 K in the NVT ensemble using a Langevin thermostat with harmonic restraints of 10.0 kcal∙mol −1 ∙Å −2 on the non-hydrogen atoms of the lipids, protein, and ligand. Initial velocities were sampled from a Boltzmann distribution. The systems were then heated to 310 K over 125 ps in the NPT ensemble. Equilibration was performed at 310 K and 1 bar in the NPT ensemble, with harmonic restraints on the protein and ligand non-hydrogen atoms tapered off by 1.0 kcal∙mol −1 ∙Å −2 starting at 5.0 kcal∙mol −1 ∙Å −2 in a stepwise manner every 2 ns for 10 ns, and finally by 0.1 kcal∙mol −1 ∙Å −2 every 2 ns for an additional 18 ns. All restraints were completely removed during production simulation. Production simulations were performed at 310 K and 1 bar in the NPT ensemble using the Langevin thermostat and Monte Carlo barostat. Lengths of bonds to hydrogen atoms were constrained using SHAKE, and the simulations were performed using a timestep of 4.0 fs while using hydrogen mass repartitioning ( Hopkins et al., 2015 ). Non-bonded interactions were cut off at 9.0 Å, and long-range electrostatic interactions were calculated using the particle-mesh Ewald (PME) method with an Ewald coefficient (β) of approximately 0.31 Å and B-spline interpolation of order 4. The PME grid size was chosen such that the width of a grid cell was approximately 1 Å. Snapshots of the trajectory were saved every 200 ps. Simulation analysis protocols The AmberTools17 CPPTRAJ package ( Roe and Cheatham, 2013 ) was used to reimage trajectories at 1 ns per frame, Visual Molecular Dynamics (VMD) ( Humphrey et al., 1996 ) was used for visualization and analysis, and PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC) was used for renderings. Plots of time traces from representative simulations were generated with Matplotlib ( Hunter, 2007 ) and show both original, unsmoothed traces (transparent lines) and traces smoothed with a moving average (thick lines), using an averaging window of 20 ns. All traces include the initial equilibration with harmonic restraints on the protein and ligand non-hydrogen atoms.

Simulation protocols: For each simulation, initial atom velocities were assigned randomly and independently. We employed the CHARMM36m force field for protein molecules, the CHARMM36 parameter set for lipid molecules and salt ions, and the associated CHARMM TIP3P model for water ( Huang et al., 2017 , Klauda et al., 2010 ). Parameters for LSD were taken from ( Wacker et al., 2017 ). Simulations were run using the AMBER20 software ( D.A. Case, 2021 ) under periodic boundary conditions with the Compute Unified Device Architecture (CUDA) version of Particle-Mesh Ewald Molecular Dynamics (PMEMD) on one GPU ( Salomon-Ferrer et al., 2013 ). After energy minimization, the systems were first heated over 12.5 ps from 0 K to 100 K in the NVT ensemble using a Langevin thermostat with harmonic restraints of 10.0 kcal∙mol −1 ∙Å −2 on the non-hydrogen atoms of the lipids, protein, and ligand. Initial velocities were sampled from a Boltzmann distribution. The systems were then heated to 310 K over 125 ps in the NPT ensemble. Equilibration was performed at 310 K and 1 bar in the NPT ensemble, with harmonic restraints on the protein and ligand non-hydrogen atoms tapered off by 1.0 kcal∙mol −1 ∙Å −2 starting at 5.0 kcal∙mol −1 ∙Å −2 in a stepwise manner every 2 ns for 10 ns, and finally by 0.1 kcal∙mol −1 ∙Å −2 every 2 ns for an additional 18 ns. All restraints were completely removed during production simulation. Production simulations were performed at 310 K and 1 bar in the NPT ensemble using the Langevin thermostat and Monte Carlo barostat. Lengths of bonds to hydrogen atoms were constrained using SHAKE, and the simulations were performed using a timestep of 4.0 fs while using hydrogen mass repartitioning ( Hopkins et al., 2015 ). Non-bonded interactions were cut off at 9.0 Å, and long-range electrostatic interactions were calculated using the particle-mesh Ewald (PME) method with an Ewald coefficient (β) of approximately 0.31 Å and B-spline interpolation of order 4. The PME grid size was chosen such that the width of a grid cell was approximately 1 Å. Snapshots of the trajectory were saved every 200 ps.

Simulation analysis protocols The AmberTools17 CPPTRAJ package ( Roe and Cheatham, 2013 ) was used to reimage trajectories at 1 ns per frame, Visual Molecular Dynamics (VMD) ( Humphrey et al., 1996 ) was used for visualization and analysis, and PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC) was used for renderings. Plots of time traces from representative simulations were generated with Matplotlib ( Hunter, 2007 ) and show both original, unsmoothed traces (transparent lines) and traces smoothed with a moving average (thick lines), using an averaging window of 20 ns. All traces include the initial equilibration with harmonic restraints on the protein and ligand non-hydrogen atoms.

Supplementary Material 1