Articles https://doi.org/10.1038/s41594-018-0068-y

Structural basis for signal recognition and transduction by platelet-activating-factor receptor

Can Cao1,4,12, Qiuxiang Tan2,3,4,12, Chanjuan Xu5, Lingli He1, Linlin Yang6, Ye Zhou1,4, Yiwei Zhou5, Anna Qiao3,4, Minmin Lu3,4, Cuiying Yi2, Gye Won Han7, Xianping Wang1, Xuemei Li1, Huaiyu Yang8, Zihe Rao1, Hualiang Jiang2,3,8, Yongfang Zhao 1, Jianfeng Liu5, Raymond C. Stevens 9,10, Qiang Zhao2,3,4,11, Xuejun C. Zhang1,4* and Beili Wu 2,4,10,11*

Platelet-activating-factor receptor (PAFR) responds to platelet-activating factor (PAF), a phospholipid mediator of cell-to-cell communication that exhibits diverse physiological effects. PAFR is considered an important drug target for treating asthma, inflammation and cardiovascular diseases. Here we report crystal structures of human PAFR in complex with the antagonist SR 27417 and the inverse agonist ABT-491 at 2.8-Å and 2.9-Å resolution, respectively. The structures, supported by molecular docking of PAF, provide insights into the signal-recognition mechanisms of PAFR. The PAFR–SR 27417 structure reveals an unusual conformation showing that the intracellular tips of helices II and IV shift outward by 13 Å and 4 Å, respectively, and helix VIII adopts an inward conformation. The PAFR structures, combined with single-molecule FRET and cell-based functional assays, suggest that the conformational change in the helical bundle is ligand dependent and plays a critical role in PAFR activa- tion, thus greatly extending knowledge about signaling by G-protein-coupled receptors.

AFR, a member of the G-protein-coupled receptor (GPCR) P217–N223 in the third intracellular loop (ICL3) of the receptor superfamily, is expressed on the surfaces of a variety of cells with residues 2–148 of a modified flavodoxin (P2A Y98W)7 from Pand tissues1 and is involved in numerous pathophysiologi- Desulfovibrio vulgaris to generate a PAFR-flavodoxin fusion con- cal activities related to inflammation and immune responses as struct, and we made a PAFR–minimal T4 lysozyme (mT4L) fusion well as cardiovascular, reproductive, respiratory and nervous- construct by inserting the mT4L8 between residues V218 and A224 system regulation2. After binding to its endogenous agonist, PAF in ICL3. To improve protein homogeneity, we truncated 26 amino (1-O-alkyl-2-acetyl-sn-glycero-3-phosphocholine), PAFR is cou- acids (residues C317–N342) at the C terminus of the receptor in pled to Gq protein and consequently activates various downstream both constructs. Additionally, we introduced four point muta- signaling pathways3. Many compounds with high structural diver- tions, F1163.51Y, A2306.33D, V2346.37A and D2897.49N (superscript sity have been characterized as PAFR antagonists and inverse ago- indicates residue numbering in Ballesteros–Weinstein nomencla- nists, and show differential effects on receptor conformation and ture9), into the PAFR gene to improve protein yield, homogeneity activity4. The highly potent and selective PAFR ligands SR 27417 and thermal stability (Supplementary Fig. 1). Another mutation, (N-(2-dimethylamino ethyl)-N-(3-pyridinyl methyl)[4-(2,4,6-tri- N169D in the second extracellular loop (ECL2), was introduced isopropylphenyl)thiazol-2-yl]amine) and ABT-491 (4-ethynyl-N,N to remove a heterogeneous glycosylation site. Our ligand-binding -dimethyl-3-[3-fluoro-4-[(2-methyl-1H-imidazo-[4,5-c]pyridine- assay showed that the flavodoxin and mT4L fusion proteins and 1-yl)methyl]benzoyl]-1H-indole-1-carboxamide hydrochloride) the five mutations had little effect on the binding affinity of PAFR inhibit PAF signaling both in vitro and in vivo, and have been sug- toward PAF, SR 27417 and ABT-491 (Supplementary Fig. 2 and gested as drug candidates for the treatment of diseases including Supplementary Table 1). asthma and cardiovascular diseases5,6. To reveal PAFR’s modes of The modified PAFR constructs were cloned into a pFastBac 1 ligand binding and to better understand the signaling mechanism of vector for expression in Spodoptera frugiperda (Sf9) insect cells, PAFR, we solved crystal structures of human PAFR in complex with thus generating PAFR-flavodoxin and PAFR-mT4L fusion proteins, SR 27417 and ABT-491 (Fig. 1 and Table 1). which were copurified with SR 27417 and ABT-491, respectively. Crystallization trials of the four PAFR fusion protein–ligand com- Results plexes were set up by reconstituting the proteins in lipidic cubic Structural determination of PAFR–SR 27417 and PAFR–ABT- phase (LCP) with cholesterol supplementation. Crystals of PAFR– 491 complexes. To obtain crystals of PAFR, we replaced residues flavodoxin in complex with SR 27417 and of PAFR-mT4L in com-

1National Laboratory of Biomacromolecules, National Center of Protein Science–Beijing, CAS Center for Excellence in Biomacromolecules, Institute of Biophysics, Chinese Academy of Sciences, Beijing, China. 2CAS Key Laboratory of Receptor Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, China. 3State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, China. 4University of Chinese Academy of Sciences, Beijing, China. 5College of Life Science and Technology, Collaborative Innovation Center for Genetics and Development, and Key Laboratory of Molecular Biophysics of the Ministry of Education, Huazhong University of Science and Technology, Wuhan, China. 6Department of Pharmacology, School of Basic Medical Sciences, Zhengzhou University, Zhengzhou, China. 7Department of Chemistry, Bridge Institute, University of Southern California, Los Angeles, CA, USA. 8Drug Discovery and Design Center, State Key Laboratory of Drug Research, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, China. 9iHuman Institute, ShanghaiTech University, Shanghai, China. 10School of Life Science and Technology, ShanghaiTech University, Shanghai, China. 11CAS Center for Excellence in Biomacromolecules, Chinese Academy of Sciences, Beijing, China. 12These authors contributed equally: Can Cao, Qiuxiang Tan. *e-mail: [email protected]; [email protected]

488 Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Nature Structural & Molecular Biology Articles

ab Table 1 | Data collection and refinement statistics N ECL2 ECL2 N PAFR–SR 27417 PAFP-ABT-491 SR 27417 ABT-491 ECL3 (PDB 5ZKP) (PDB 5ZKQ) ECL1 ECL3 ECL1 Data collectiona

Space group P3221 C2 Cell dimensions a, b, c (Å) 67.1, 67.1, 280.3 86.2, 166.5, 100.3 IV II IIIIII VII α, β, γ (°) 90.0, 90.0, 120.0 90.0, 100.0, 90.0 IV II I I IIIIII V Resolution (Å) 50.0–2.8 (2.95–2.8)b 50.0–2.9 (3.06–2.9) V VII VI c Rpim 0.087 (0.622) 0.159 (0.910)

VI I/σ(I) 16.9 (0.8) 5.7 (1.0)

CC1/2 0.987 (0.528) 0.992 (0.536) C Completeness (%) 99.4 (99.6) 98.5 (97.7) C Redundancy 7.6 (8.1) 4.2 (3.8) ICL1 VIIVIIII Refinement Resolution (Å) 50.0–2.8 50.0–2.9 No. reflections 17,489 30,664 cd Rwork / Rfree 0.222 / 0.259 0.203 / 0.235 ECL1ECL1 No. atoms Protein 3,391 5,316 ECL1ECL1 ECL2ECL2 Ligand 33 72 N Other 31 179 ECLECL22 N B factors (Å2) PAFR 97.8 88.5 Flavodoxin or mT4L 121.5 70.9 ECL3ECL3 ECL3ECL3 Ligand 74.6 82.2 Other 125.3 100.1 S1P PAFR 1 R.m.s. deviations Bond lengths (Å) 0.010 0.010 Fig. 1 | Structures of the PAFR–SR 27417 and PAFR–ABT-491 complexes. a, Structure of the PAFR–SR 27417 complex. PAFR is shown in orange Bond angles (°) 1.08 1.11 cartoon representation. The ligand SR 27417 is shown in sphere aDiffraction data from 36 PAFR–SR 27417 crystals and 52 PAFR–ABT-491 crystals were used to b c representation with green carbons. The disulfide bond is shown as red solve the structures. Values in parentheses are for the highest-resolution shell. Rmerge values are sticks. b, Structure of the PAFR–ABT-491 complex. PAFR is shown in shown. blue cartoon representation. The ligand ABT-491 is shown in sphere representation with magenta carbons. The disulfide bond is shown as yellow sticks. c, Extracellular view of the PAFR–SR 27417 structure. PAFR helical bend observed in the GPCR structures containing a proline at this position. In PAFR, similarly to the other known structures is shown in gray molecular surface and cartoon representation. The N 12–15 terminus, ECL1, ECL2 and ECL3 of the receptor are colored blue, magenta, of lipid receptors, S1P1, FFAR1, LPA1 and CB1 , the ligand-bind- orange and cyan, respectively. The ligand SR 27417 is shown as spheres. ing pocket is capped by the N terminus and extracellular loops of d, Extracellular view of the S1P –ML506 structure (PDB 3V2W). S1P the receptor, mainly ECL2 (Fig. 1c,d). Strong interactions between 1 1 ECL2 and the N terminus, ECL1 and the extracellular tips of helices is shown in gray molecular surface and cartoon representation. The N 3.25 terminus, ECL1, ECL2 and ECL3 of the receptor are colored blue, magenta, I, II and VI, together with the C90 -C173 disulfide bridge between orange and cyan, respectively. The ligand ML506 is shown as spheres. helix III and ECL2, stabilize the conformation of ECL2. The roof- like structures over the ligand-binding pockets in the structures of

PAFR, S1P1, FFAR1, LPA1 and CB1 may be a general structural fea- ture shared by the lipid receptors. plex with ABT-491 were obtained and diffracted to approximately 2.9 Å, whereas the crystallization trials of the other two complexes PAFR binding modes of SR 27417 and ABT-491. The ligands SR were not successful. Strong and unambiguous electron densities 27417 and ABT-491 bind to a pocket defined by residues from all are present for both SR 27417 and ABT-491 in the PAFR structures the 7TM helices and ECL2 of PAFR (Fig. 2a,c). Compared with

(Supplementary Fig. 3a,b). those in the structures of S1P1–ML506, FFAR1–TAK875, LPA1– 12–15 ONO9780307 and CB1–AM6538 complexes , the ligand-binding Overall architecture of the PAFR structures. The PAFR struc- sites of SR 27417 and ABT-491 in PAFR are similar to ML506’s bind- tures contain a typical GPCR architecture comprising a seven- ing site in S1P1 (Supplementary Fig. 4a). Interactions between PAFR transmembrane (7TM) helical bundle (helices I–VII) (Fig. 1a,b). and SR 27417 are predominantly hydrophobic, involving only one

As in the purinergic receptor P2Y12R and sphingosine 1-phosphate hydrogen bond between the nitrogen atom of the pyridyl group in 10–12 2.64 receptor 1 (S1P1) , the highly conserved class A GPCR residue SR 27417 and the side chain hydroxyl of Y77 (Fig. 2a,b). The pyri- P5.50 is substituted with V5.50 in PAFR. As a result, helix V of PAFR dyl ring fits into a small subpocket shaped by helices I, II, III and VII, exhibits a straight α​-helical conformation that lacks the canonical and forms a π​–π​ stacking interaction with W732.60 and hydrophobic

Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb 489 © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Articles Nature Structural & Molecular Biology

a c e IV II IV III III IV 3.37 Y102 3.33 3.36 F98 II 4.60 T101 3.33 F152 III 2.60 F98 4.60 W73 F1524.60 F152 F174 3.32 3.32 F983.33 F97 F97 F973.32 F174 2.60 Y1023.37 W73 W732.60 II I1915.43 PAF

Y221.39 1.39 5.43 Y22 I191 2.64 3.37 Y77 7.42 Y102 5.40 L282 6.51 H188 2.64 H248 Y77 5.40 2.64 L2797.39 H188 Y77 V Q2526.55 Y177 E175 L2797.39 V Q2526.55 7.35 I H275 V H2757.35 I VI 6.58 VII 6.58 H2486.51 W255 I VI VI W255 VII VII

b 4.60 F174 d f F152 2.60 W73 H1885.40 F174 3.36 H1885.40 T101 E175 F973.32 3.33 4.60 Y177 W732.60 F98 SR 27471 F152 2.64 Y77 VI Y221.39 I1915.43 VII V 7.39 3.37 Lipid L279 6.55 Y102 7.39 Q252 L279 ABT-491 PAF 6.51 H248 7.35 5.43 7.42 H275 I191 I L282 III 3.33 3.32 F98 F97 1.39 6.58 Y22 IV W255 2.64 Y77 Y1023.37 II

Fig. 2 | Ligand-binding modes of PAFR. a, PAFR ligand-binding pocket for SR 27417. PAFR is shown in gray cartoon representation. The ligand SR 27417 (green carbons) and PAFR residues (orange carbons) involved in ligand binding are shown in stick representation. The hydrogen bond is displayed as a blue dashed line. The disulfide bond is shown as yellow sticks. b, Schematic representations of interactions between PAFR and SR 27417, as analyzed in LigPlot+ (ref. 32). The stick drawings of PAFR residues and SR 27417 are colored orange and green, respectively. c, PAFR ligand-binding pocket for ABT-491. PAFR is shown in gray cartoon representation. The ligand ABT-491 (magenta carbons) and PAFR residues (blue carbons) involved in ligand binding are shown in stick representation. The hydrogen bonds are displayed as blue dashed lines. The disulfide bond is shown as yellow sticks. The lipid molecule interacting with the ligand is shown as cyan lines. d, Schematic representations of interactions between PAFR and ABT-491, as analyzed in LigPlot+ (ref. 32). The stick drawings of PAFR residues and ABT-491 are blue and magenta, respectively. e, Docking poses of PAF in the PAFR–SR 27417 structure. The PAFR structure is presented in gray cartoon. The three top-scoring docking conformations of PAF are shown as yellow, blue and green sticks, respectively. The PAFR residues that substantially affect PAF binding are shown as magenta sticks, and the residues that have weak effects on PAF binding are shown as cyan sticks. f, Side view of the PAF-binding pocket. PAFR is shown in gray surface and orange cartoon representation. interactions with Y221.39, Y772.64, F973.32 and L2797.39. The dimeth- of the methylbenzoyl spacer in ABT-491 have also been found to ylaminoethyl group of SR 27417 extends toward the extracellular be critical to the ability of the compound to bind PAFR16. In the surface and forms hydrophobic contacts with W2556.58, H2757.35 and PAFR–ABT-491 structure, the benzene ring within the methylben- L2797.39 on helices VI and VII. The ligand’s triisopropylphenyl group zoyl group traverses a narrow channel between F973.32 and F174, inserts into a binding crevice with helices III and IV on one side and and plays a key role in maintaining the required spatial relation- helices VI and VII on the other side. The benzene ring within the ship between the indole group and the imidazopyridine ring. The triisopropylphenyl group forms an aromatic edge-to-face interac- nature of the methylimidazopyridine group of ABT-491 has been tion with F983.33, and the three isopropyl substituents interact with reported to be important for receptor–ligand binding16. In the PAFR multiple residues on helices III–VII and ECL2, thus greatly contrib- structure, this heterocycle fits into a subpocket and forms multiple uting to the tight receptor–ligand binding. hydrophobic contacts with helices I, II, III and VII, and ECL2. Two The ligand ABT-491 occupies a binding pocket similar to that nitrogen atoms within this group are coordinated by two hydrogen occupied by SR 27417 in PAFR (Fig. 2c,d and Supplementary Fig. 4a). bonds with Y221.39 and Y772.64. Nitrogen atoms at different posi- The indole group of the ligand is surrounded by several aromatic tions in various regioisomers of the methylimidazopyridine show and hydrophobic residues contributed by helices III, IV and V of dramatic effects on binding potency16, which may be explained by PAFR, which provide a strong hydrophobic binding environment the importance of these two hydrogen bonds in retaining the PAFR for the indole ring. The N,N-dimethylcarbamoyl substituent is binding affinity of this chemical series. engaged in a hydrogen bond with H1885.40 and makes hydrophobic interactions with F174 and Y177 on ECL2. This finding is consis- Docking poses of the agonist PAF. To further deepen understand- tent with those from previous structure–activity relationship stud- ing of the signal-recognition mechanisms of PAFR, we performed ies showing that the indole nitrogen substitution is essential for full-atom docking of its endogenous ligand, PAF, into the crystal optimal in vitro and in vivo potency of this chemical series16. In structure of PAFR–SR 27417 with the ligand removed. The results structure–activity relationship studies, the length and orientation show that the sn-2 acetyl moiety and sn-3 phosphocholine group

490 Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Nature Structural & Molecular Biology Articles a cd e IV

V III ECL2 ECL2 F973.32 SR 27417 2.53 F66 SR 27417

VI III 2.3 Å II VII II II

VII IV I IV b VI f V VII 4.60 4 Å I VI F152 V III VI II II III III ICL1 1.6 Å 13 Å 13 Å II IV IV VIII 4 Å III IV

Fig. 3 | Conformational changes of the helical bundle in PAFR. a,b, Structural comparisons of PAFR–SR 27417 and PAFR–ABT-491 complexes. a, Extracellular view. b, Intracellular view. The receptors in the PAFR–SR 27417 and PAFR–ABT-491 structures are shown in cartoon representation and are colored orange and blue, respectively. The ligands SR 27417 and ABT-491 are shown as green and magenta sticks, respectively. The red arrows indicate the movements of helices II and IV in the PAFR–SR 27417 structure compared with the PAFR–ABT-491 structure. Helix VIII is not resolved in the PAFR–ABT-491 structure. c, Conformational change of helix II in PAFR. The ligand SR 27417 is shown in green sphere representation. Residues F662.53 and F973.32, which regulate the conformational change of helix II, are shown as sticks and are colored orange in the PAFR–SR 27417 structure and blue in the PAFR–ABT-491 structure. The red arrow indicates the movement of helix II in the PAFR–SR 27417 structure compared with the PAFR–ABT-491 structure. d, Close-up view of the key residues F662.53 and F973.32, which regulate the conformational change of helix II. The red arrows indicate the movements of the side chain of F973.32 and the Cα​ atom of F662.53 in the PAFR–SR 27417 structure compared with the PAFR–ABT-491 structure. e, Conformational change of helix IV in PAFR. Residue F1524.60, which regulates the conformational change of helix IV is shown as sticks and is colored orange in the PAFR–SR 27417 structure and blue in the PAFR–ABT-491 structure. The red arrow indicates the movement of helix IV in the PAFR–SR 27417 structure compared with the PAFR–ABT-491 structure. f, Close-up view of the key residue F1524.60, which regulates the conformational change of helix IV. The red arrow indicates the movement of the Cα ​atom of F1524.60 in the PAFR–SR 27417 structure compared with the PAFR–ABT-491 structure. of PAF occupy a subpocket bordered by helices I, II, VI and VII PAFR–SR 27417 complex has a distinct helical-bundle conformation and ECL2, and the substituent at either the sn-2 position or the sn-3 (Supplementary Fig. 4b), which has not previously been observed in position points to the extracellular surface of the receptor (Fig. 2e,f). the solved GPCR structures, to our knowledge. On the extracellular The sn-1 alkyl chain extends into a ‘tunnel’ formed by aromatic and side, only subtle structural differences are present between the two hydrophobic residues of helices III, IV, V and VI, and ECL2, and its PAFR structures (Fig. 3a). However, in the intracellular half, PAFR tail squeezes into the lipid bilayer through the gap between helices shows unique conformational features (Fig. 3b). In the PAFR–SR IV and V (Fig. 2f), thus suggesting that this route may allow PAF to 27417 structure, the intracellular tips of helices II and IV move away access the binding pocket from the membrane. The tunnel-like cav- from the central axis of the helical bundle by 13 Å and 4 Å, respec- ity for the sn-1 alkyl chain is consistent with previous data showing tively, as compared with the ABT-491-bound structure, and α​-helix that a variety of structurally related phospholipid molecules with VIII adopts a unique conformation in the plane of the cell mem- different carbon chains bind PAFR, and the chain length has modest brane perpendicular to the orientation of helix VIII in most of the effects on the biological potency of the ligands17. other known GPCR structures (Supplementary Fig. 4b). Our mutagenesis studies further supported the proposed bind- A closer inspection of the two PAFR structures revealed confor- ing mode between PAFR and PAF, showing that the mutations mational changes in the receptor core (Fig. 3 and Supplementary W732.60A, Y772.64A, H2486.51W, W2556.58A, H2757.35A and H2757.35W Fig. 3c–f). Within the ligand-binding pocket in the PAFR–ABT-491 within the subpocket for the sn-2 and the sn-3 groups substantially structure, residue F973.32 on helix III makes an aromatic edge-to-face decreased the ability of the receptor to bind [3H]PAF, whereas the contact with the phenyl ring of the ligand (Fig. 2c). In contrast, in the mutants F973.32A, F983.33A, Y1023.37A, F1524.60A, F1524.60W and SR 27417–bound structure, one of the isopropyl substituents within Q2526.55A, surrounding the tunnel for the alkyl chain, had no the triisopropylphenyl group of the ligand causes a spatial hin- effect or a weak effect on PAF binding to the receptor (Fig. 2e and drance, thus leading to a rotation of the side chain of F973.32 toward Supplementary Table 1). helix II and subsequently to an outward movement of the Cα ​ atom of F662.53 on helix II by approximately 2.3Å (Fig. 3c,d). Additionally, Conformational change of the helical bundle of PAFR. In the F1524.60 on helix IV forms another edge-to-face interaction with known structures of class A GPCRs bound to inhibitors, the heli- the indole group of the ligand in the ABT-491–bound structure cal bundles of the receptors exhibit similar conformations. The (Fig. 2c), whereas in the PAFR–SR 27417 structure, this residue is 7TM helical bundle of PAFR in the PAFR–ABT-491 structure is pushed away by another isopropyl substituent of SR 27417, by 1.6 Å in a conformation close to that of the other inactive GPCR struc- (Fig. 3e,f). Owing to the above conformational changes induced by tures (Supplementary Fig. 4c,d). Unexpectedly, the structure of the the ligands around the ligand-binding pocket, helices II and IV shift

Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb 491 © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Articles Nature Structural & Molecular Biology

abPAFR(Fl)–SR 27417 PAFR(Fl)–ABT-491 c PAFR(Fl)–PAF 9 8 n = 543 n = 767 n = 689 8 10 7 7 8 6 6 5 5 6 4 4 3 3 4 Occupancy (% ) Occupancy (% ) 2 Occupancy (% ) 2 2 1 1 0 0 0 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 FRET FRET FRET

dePAFR(mT)–SR 27417 PAFR(mT)–ABT-491 fg 10 n = 743 n = 791 12 V V VI 8 10 VI VII 6 8 VII III K532.40AzK 6 III 4 I IV 2.40 II 4 K53 AzK I Occupancy (%) Occupancy (%) IV 2 4.40 4.40 2 K132 AzK VIII II K132 AzK 0 0 PAFR–SR 27417 PAFR–ABT-491 0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0 structure structure FRET FRET

Fig. 4 | smFRET assay of PAFR, measuring the distance between the intracellular tips of helices II and IV. a–c, smFRET histograms from single-molecule traces of the PAFR-flavodoxin fusion protein (PAFR(Fl)) in complex with SR 27417, ABT-491 or PAF. The number of molecules (n) for each sample is shown. The smFRET efficiency for the smFRET peak of each sample is labeled with a red dashed line. d,e, smFRET histograms from single-molecule traces of the PAFR-mT4L fusion protein (PAFR(mT)) in complex with SR 27417 or ABT-491. The smFRET efficiency for the smFRET peak of each sample is labeled with a red dashed line. Data shown are from an experiment representative of three independent experiments. Error bars, s.d. of ≥100​ bootstrap samples. f,g, Fluorescence-labeling sites in the PAFR–SR 27417 and PAFR–ABT-491 complexes used in the smFRET assay. f, The PAFR–SR 27417 structure, shown as an orange cartoon. Residues R1314.39–I1414.49 were modeled as an extension of helix IV. Residues K532.40 and K1324.40, which were mutated to the unnatural amino acid AzK, are shown as orange sticks. The green circles indicate fluorescent labels. The green dashed line indicates the distance between the intracellular tips of helices II and IV. g, The PAFR–ABT-491 structure, shown in blue cartoon. Residues K532.40 and K1324.40, which were mutated into the unnatural amino acid AzK, are shown as blue sticks. outward by 13 Å and 4 Å, respectively, on the intracellular side of the tion similar to that observed in the PAFR–SR 27417 complex. The receptor in the SR 27417–bound PAFR structure compared with the conformation of the PAFR–SR 27417 complex was further analyzed ABT-491–bound structure. over the course of a 2-μ​s molecular dynamics (MD) simulation, To confirm the conformational changes of helices II and IV, which shows that the helical bundle of PAFR retains the same con- we used a single-molecule fluorescence resonance energy transfer formation as that observed in the crystal structure (Supplementary (smFRET) assay to measure the distance between the intracellular Fig. 6). Together, these data support the conformations of the helical tips of helices II and IV of the receptor in complex with SR 27417, bundle in the PAFR crystal structures and suggest that the confor- ABT-491 or PAF (Fig. 4). Because the protein yield and stability mational differences of helices II and IV in the PAFR–SR 27417 and of the wild-type (WT) PAFR protein were poor, we used both the PAFR–ABT-491 structures are probably stabilized by binding todif- PAFR-flavodoxin and PAFR-mT4L fusion proteins to eliminate the ferent ligands. effects of the fusion partners on the conformation of the helical In addition, we performed an inositol phosphate (IP) assay for bundle. For the complex of PAFR–SR 27417, the FRET efficiency HEK293 cells expressing WT PAFR in the presence of SR 27417, for the FRET peak was 0.75 (Fig. 4a,d), whereas the FRET efficiency ABT-491 or PAF to test the effects of these ligands on PAFR- of the PAFR–ABT-491 complex was 0.9 (Fig. 4b,e). These data dem- mediated Gq-protein signaling (Fig. 5a). PAF strongly induced IP onstrate that the distance between the intracellular tips of helices production, whereas SR 27417 supported an IP production 15% that II and IV is much longer in the PAFR–SR 27417 complex than in the of the PAFR–PAF complex, a level similar to that produced by the PAFR–ABT-491 complex, in agreement with the two PAFR crys- apo receptor. Interestingly, the IP production greatly decreased in tal structures. Because we obtained similar results for both fusion the presence of ABT-491. These results demonstrate that SR 27417 proteins, the conformational changes of helices II and IV are not and ABT-491 behave differently in Gq-protein-mediated signaling: likely to have been caused by protein engineering. Furthermore, SR 27417 acts as an antagonist that retains the basal activity of the we measured the distance between the intracellular tips of heli- receptor, whereas ABT-491 acts as an inverse agonist that blocks ces I and V, which adopt similar conformations in the two PAFR basal signaling and presumably Gq-protein coupling. structures, as a negative control for the smFRET experiment. As We further tested the effects of residues F662.53, F973.32 and F1524.60, expected, the FRET efficiency remained the same in response to which play key roles in regulating the conformational changes of both SR 27417 and ABT-491 (Supplementary Fig. 5), thus further helices II and IV, on PAFR activation, by using IP-accumulation supporting the PAFR structures and indicating the reliability of the assays. ELISA analysis was performed to compare the cell-surface smFRET data. The distance between helices II and IV in the com- expression of the mutants with the WT PAFR, and none of the plex of PAFR–PAF was also measured, and the FRET efficiency was mutations decreased the cell-surface expression of the receptor determined to be 0.7 (Fig. 4c), thus suggesting that helices II and (Fig. 5b). The results of [3H]PAF binding assays showed that the IV in the agonist-bound receptor may adopt an outward conforma- three mutants retained similar binding affinity to PAF. The mutants

492 Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Nature Structural & Molecular Biology Articles a b I and II play an important role in stabilizing the conformation of 1,000 200 *** helix VIII (Fig. 6c). The interactions include another hydrophobic 1.61 800 150 core composed of T305, F308 and Y309 in helix VIII, and Y44 , 2.39 2.42 600 I52 and F55 in helices I and II. Two hydrogen bonds between NS 100 T305 and Y441.61, and Y309 and the main chain of I522.39 further 400

(% of WT) strengthen the interaction. ** 50 200 The results of the 2-μ​s MD simulation on the PAFR–SR 27417 IP1 production (nM)

Cell-surface expression crystal structure with the flavodoxin fusion protein removed show 0 0 WT F66A F97A F152A that helix VIII maintains the same conformation as that observed Apo PAF in the structure, and most of the interactions among helix VIII and ABT-491 SR 27417 helices I, II and VII remain intact throughout the length of simula- tion (Supplementary Fig. 6). These results suggest that the observed –8 cdSR 27417 (10–5 M) PAF (10 M) conformation of helix VIII is stable in the absence of fusion protein * 150 150 ** and crystal lattice. Furthermore, although helix VIII is not resolved ** *** * NS in the PAFR–ABT-491 structure, helices II and IV are in a close conformation and thus would form extensive spatial clashes with 100 100 helix VIII if it were in the same conformation as that observed in the PAFR–SR 27417 structure. Thus, helix VIII may adopt a different 50 50 conformation in the inverse agonist ABT-491–bound PAFR. Previous studies have shown that helix VIII is involved in recep- 18,19 IP1 production (% of WT) 0 IP1 production (% of WT) 0 tor activation in some GPCRs . We performed mutagenesis of WT F66A F97A F152A WT F66A F97A F152A helix VIII in PAFR by individually substituting residues F300–R315 to alanine in the WT receptor, then tested the behavior of these 2+ Fig. 5 | IP-accumulation assays of PAFR. a, IP accumulation of WT PAFR mutants in Gq protein signaling by using a PAF-induced Ca -flux in the apo state and in the presence of ABT-491 (at 10−5 M concentration), assay. The results showed that two of these mutations, F300A and SR 27417 (at 10−6 M concentration) or PAF (at 10−8 M concentration). R301A, in the N-terminal portion of helix VIII, decreased the max- Data shown are mean ±​ s.e.m. from a representative of three independent imum Ca2+ flux by 30% and 50%, respectively. Furthermore, the experiments using independently transfected cells and performed in mutation F308A decreased the half-maximal effective concentra- technical triplicate. **P <​ 0.01; ***P <​ 0.001; NS, nonsignificant by unpaired tion of the agonist PAF by approximately tenfold relative to that of two-tailed t test, compared with the IP1-production level of apo receptor. the WT protein (Fig. 6d). To further confirm the role of helix VIII b–d, Cell-surface expression and IP production of WT and mutant PAFRs, in activation, we performed [3H]PAF binding assays, which showed F662.53A, F973.32A and F1524.60A. b, Cell-surface expression. c, IP production that the three mutants retained a PAF binding affinity similar to that in the presence of SR 27417 alone at 10−5 M concentration. d, IP production of the WT receptor (Supplementary Table 1). Together, these data in the presence of PAF alone at 10−8 M concentration. Data shown are suggest that these mutations decrease the ability of PAFR to activate mean ±​ s.e.m. from a representative of three independent experiments Gq protein. Structurally, F300 and F308 play key roles in stabiliz- using independently transfected cells and performed in technical triplicate. ing the conformation of helix VIII in the structure of the PAFR–SR *P <​ 0.1; **P <​ 0.01; ***P <​ 0.001; NS, nonsignificant by unpaired two-tailed 27417 complex, which showed basal activity. These results sug- t test, compared with the IP1 production of the WT receptor. gest that this observed conformation of helix VIII may facilitate G-protein signaling of PAFR. Structural superimposition analysis between the PAFR–SR 27417 and β2​ adrenergic receptor–Gs protein F662.53A and F1524.60A displayed a SR 27417 binding ability similar complexes revealed that helix VIII of PAFR is located in the G pro- to that of the WT receptor, whereas the mutation F973.32A decreased tein-binding interface in close contact with, but not overlapping, the the binding ability (Supplementary Table 1). To ensure 100% occu- N-terminal α-helix​ of the Gα ​subunit (Supplementary Fig. 7). This pancy of SR 27417 in the receptor, we used a ligand concentration result suggests that the conformation of helix VIII may potentially −5 at least 100-fold higher (10 M) than the Ki values of the mutants to influence G-protein coupling to PAFR. measure the IP production. The results showed that the three muta- tions F662.53A, F973.32A and F1524.60A decreased PAFR-mediated IP Discussion production in the presence of either the antagonist SR 27417 or the Structural studies of GPCRs have made tremendous progress in the agonist PAF (Fig. 5c,d). Among the three mutations, F973.32A exhib- past decade. The GPCR structures, combined with molecular stud- ited the greatest effect on Gq-protein signaling, showing an approxi- ies, reveal distinct mechanisms through which diverse ligands mod- mately 70% loss of IP production in the presence of SR 27417 and ulate GPCR function20. The structures of several different GPCRs an approximately 95% decrease in IP accumulation induced by PAF. bound to agonists have provided essential insights into GPCR Together, these data suggest a correlation between the conforma- activation mechanisms. Comparison of the active-state structures tional changes of helices II and IV and PAFR-mediated Gq-protein of these GPCRs with their inactive conformations clearly indicates signaling. that the rearrangement of helices V, VI and VII plays an important role in GPCR activation, whereas the conformation of helices I, II, A unique conformation of helix VIII. In addition to the large lat- III and IV appears to be less mobile and thus presumably plays a eral deviations of helices II and IV, α​-helix VIII in the PAFR–SR lesser role in receptor signaling21–23. However, the previously solved 27417 structure adopts a unique conformation, which is not present structure of the purinergic receptor P2Y1R suggests that the confor- in all the other known GPCR structures. It extends across the helical mational plasticity of helices I–IV is in fact equally critical for recep- bundle flanked by ICL1 and ICL2, and its C terminus is wedged into tor activation as helices V–VII24. The structural features of PAFR a gap between helices II and IV (Fig. 6a). Residue F300 on the N ter- described here demonstrate that helices II and IV of the receptor minus of helix VIII, together with L304, forms a hydrophobic-pack- adopt distinct conformations dependent on the type of ligand in the ing core with F2957.55 and L2967.56 in helix VII, thus constraining binding pocket. The conformational change of the helical bundle is the conformation of helix VIII to depart in a different orientation supported by the results of smFRET assays and a 2-μs​ MD simula- (Fig. 6b). Additionally, interactions between helix VIII and helices tion, thus raising the question of what effect this conformational

Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb 493 © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Articles Nature Structural & Molecular Biology

a b I ICL1

VI II V R301 II T305 I VII VII Y309 F2957.55 VIII L2967.56 VI

F300 K307 - IV L304 III ICL3 F300 ICL2 VIII ------V

cd

125 WT III VII F300A I II V 100 R301A F308A 2.42 VI F55 75

VIII 50

------F308 I522.39 release (% of WT maximum) 25 2+ Ca ----- Y309 0 1.61 Y44 T305 –12 –10 –8 –6 log (PAF concentration (M))

Fig. 6 | Helix VIII in the PAFR–SR 27417 structure. a, Electron densities of helix VIII in the PAFR–SR 27417 structure. The receptor is shown in gray cartoon representation. Residues on helix VIII and the C terminus of helix VII are displayed as sticks with orange carbons. Electron densities are contoured at 1.0σ​

from a 2Fo – Fc map and are colored blue. The gray dashed lines indicate ICL2 and ICL3, which are not resolved in the structure. b, Interactions between helices VII and VIII in the PAFR–SR 27417 structure. Key residues involved in the interactions are shown as sticks with orange carbons. c, Interactions between helices I and II and helix VIII in the PAFR–SR 27417 structure. Key residues are shown as sticks with orange carbons. Hydrogen bonds are displayed as blue dashed lines. d, PAF-induced Ca2+-flux assay for WT PAFR and PAFR mutants on helix VIII, normalized to the surface expression of each protein. Cell-surface expression of the three mutants was measured with ELISA and compared with that of WT PAFR; the expression level of the mutants was close to that of the WT receptor (F300A, 66.1% of WT; R301A, 93.9% of WT; F308A, 105.6% of WT). Dose–response curves of fluorescence signals of Ca2+–Fluo-4 were generated from three independent experiments performed in technical triplicate. Data are shown as mean ±​ s.e.m. change might have on regulating receptor activity. From the PAFR play a role in the initial interaction with G protein27. The recently structures together with mutagenesis and G-protein-signaling stud- determined structures of class B GPCRs bound to Gs protein reveal 28–30 ies, we propose a correlation between the conformational change of direct interactions of ICL1 and helix VIII with Gs protein , and the helical bundle and the activation of PAFR. The conformations of further highlight the importance of these two regions in GPCR sig- helices II and IV observed in the PAFR–SR 27417 structure appear naling. In PAFR, the outward conformation of helices II and IV to facilitate G-protein signaling, thereby supporting the basal activ- on the intracellular side provides more space for the movements of ity of the receptor. Upon binding to the agonist, the outward con- the intracellular loops and helix VIII, which probably increase the formation of helices II and IV is likely to be retained or further binding ability of the receptor to G protein. The observed confor- stabilized, thereby amplifying signaling. In contrast, the inverse mation of the helical bundle and helix VIII in the structure of the agonist ABT-491 stabilizes the receptor in an inactive state by lock- PAFR–SR 27417 complex may represent an intermediate confor- ing helices II and IV in an inward conformation, thus decreasing mational state in the process of PAFR activation, which may initiate the receptor’s ability to signal via G protein. This finding further G-protein coupling and the associated conformational changes of highlights the importance of the conformational rearrangement of helices V, VI and VII to activate downstream signaling pathways. In helices I–IV in GPCR signaling. the recently published crystal structure of the angiotensin II recep- Several studies have suggested that helix VIII in some GPCRs tor, helix VIII also adopts a noncanonical conformation: the helix is involved in direct interactions with G protein25,26. However, no flips over to interact with the intracellular ends of helices III, V and interaction between helix VIII and G protein is seen in the struc- VI, and it may sterically block G-protein and β​-arrestin binding in 21 31 ture of the β2​ adrenergic receptor–Gs protein complex . One pos- a self-inhibitory manner . The structures of PAFR and the angio- sible explanation is the existence of multiple conformational states tensin II receptor suggest that helix VIII may modulate receptor of the GPCR–G protein complex27. A possible allosteric pathway activation by justifying its conformation and interacting with vari- of receptor activation from the agonist-binding pocket to ICL1 ous binding partners. and helix VIII has been proposed in previous NMR studies of the In summary, the PAFR structures reveal an unusual conforma- μ​-opioid receptor, thus suggesting that ICL1 and helix VIII may tion of the 7TM helical bundle, which was found to be involved in

494 Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Nature Structural & Molecular Biology Articles

G protein signaling. These results suggest that helices II and IV of 23. Xu, F. et al. Structure of an agonist-bound human A2A adenosine receptor. PAFR may play an important role in receptor activation. In addi- Science 332, 322–327 (2011). 24. Zhang, D. et al. Two disparate ligand-binding sites in the human P2Y1 tion, helix VIII adopts an unexpected conformation in the structure receptor. Nature 520, 317–321 (2015). of the PAFR–SR 27417 complex and consequently may facilitate the 25. Ernst, O. P. et al. Mutation of the fourth cytoplasmic loop of rhodopsin G-protein coupling process. Thus, the PAFR structures offer new afects binding of transducin and peptides derived from the carboxyl-terminal insights into the molecular mechanisms of GPCR signaling. sequences of transducin alpha and gamma subunits. J. Biol. Chem. 275, 1937–1943 (2000). 26. Delos Santos, N. M., Gardner, L. A., White, S. W. & Bahouth, S. W. Methods Characterization of the residues in helix 8 of the human β​1-adrenergic Methods, including statements of data availability and any asso- receptor that are involved in coupling the receptor to G proteins. J. Biol. ciated accession codes and references, are available at https://doi. Chem. 281, 12896–12907 (2006). org/10.1038/s41594-018-0068-y. 27. Sounier, R. et al. Propagation of conformational changes during μ​-opioid receptor activation. Nature 524, 375–378 (2015). Received: 11 January 2018; Accepted: 20 April 2018; 28. Liang, Y. L. et al. Phase-plate cryo-EM structure of a class B GPCR-G-protein complex. Nature 546, 118–123 (2017). Published online: 28 May 2018 29. Zhang, Y. et al. Cryo-EM structure of the activated GLP-1 receptor in complex with a G protein. Nature 546, 248–253 (2017). References 30. Liang, Y. L. et al. Phase-plate cryo-EM structure of a biased agonist-bound 1. Hwang, S. B. Specifc receptors of platelet-activating factor, receptor human GLP-1 receptor-Gs complex. Nature 555, 121–125 (2018). heterogeneity, and signal transduction mechanisms. J. Lipid Mediat. 2, 31. Zhang, H. et al. Structural basis for selectivity and diversity in angiotensin II 123–158 (1990). receptors. Nature 544, 327–332 (2017). 2. Braquet, P., Touqui, L., Shen, T. Y. & Vargafig, B. B. Perspectives 32. Laskowski, R. A. & Swindells, M. B. LigPlot+: multiple ligand-protein in platelet-activating factor research. Pharmacol. Rev. 39, 97–145 (1987). interaction diagrams for drug discovery. J. Chem. Inf. Model. 51, 3. Chao, W. & Olson, M. S. Platelet-activating factor: receptors and signal 2778–2786 (2011). transduction. Biochem. J. 292, 617–629 (1993). 4. Dupré, D. J., Le Gouill, C., Rola-Pleszczynski, M. & Stanková, J. Inverse Acknowledgements agonist activity of selected ligands of platelet-activating factor receptor. This work was supported by the National Key R&D Program of China, 2018YFA0507000; J. Pharmacol. Exp. Ter. 299, 358–365 (2001). CAS Strategic Priority Research Program grant XDB08020000 (B.W., X.C.Z., Z.R. and 5. Herbert, J. M. et al. Biochemical and pharmacological activities of SR 27417, X.L.); the Key Research Program of Frontier Sciences, CAS, grants QYZDB-SSW- a highly potent, long-acting platelet-activating factor receptor antagonist. SMC024 (B.W.) and QYZDB-SSW-SMC054 (Q.Z.); the National Science Foundation of J. Pharmacol. Exp. Ter. 259, 44–51 (1991). China, grants 81525024 (Q.Z.) and 31301163 (C.X.); and the Program of Introducing 6. Albert, D. H. et al. Pharmacology of ABT-491, a highly potent platelet- Talents of Discipline to the Universities of the Ministry of Education (grant B08029) activating factor receptor antagonist. Eur. J. Pharmacol. 325, 69–80 (1997). (J.L.). The authors thank M. Hanson, V. Cherezov and V. Katritch for careful review and 7. Chun, E. et al. Fusion partner toolchest for the stabilization and scientific feedback on the manuscript; H. Zhang for guidance in handling radiolabeled crystallization of G protein-coupled receptors. Structure 20, 967–976 (2012). chemicals; and J. W. Chin (Medical Research Council Laboratory of Molecular 8. Torsen, T. S., Matt, R., Weis, W. I. & Kobilka, B. K. Modifed T4 lysozyme Biology, Cambridge) for providing plasmids (U6-PylT*)4/EF1α​-PylRS, (U6-PylT*)4/ fusion proteins facilitate G protein-coupled receptor crystallogenesis. EF1α-sfGFP(TAG)​ and peRF1 (E55D). The synchrotron radiation experiments were Structure 22, 1657–1664 (2014). performed at BL41XU of SPring-8 with approval from the Japan Synchrotron Radiation 9. Ballesteros, J. & Weinstein, H. Integrated methods for the construction Research Institute (proposal nos. 2014B1057, 2015A1026, 2015A1027, 2015B2026 and of three-dimensional models and computational probing of structure- 2015B2027). We thank the BL41XU beamline staff members K. Hasegawa, H. Okumura function relations in G protein-coupled receptors. Methods Neurosci. 25, and H. Murakami for help with X-ray data collection. 366–428 (1995). 10. Zhang, K. et al. Structure of the human P2Y12 receptor in complex with an antithrombotic drug. Nature 509, 115–118 (2014). Author contributions 11. Zhang, J. et al. Agonist-bound structure of the human P2Y12 receptor. Nature C.C. and Q.T. optimized the construct; developed the purification procedure and 509, 119–122 (2014). purified the PAFR proteins for crystallization; and performed crystallization trials and 12. Hanson, M. A. et al. Crystal structure of a lipid G protein-coupled receptor. optimized crystallization conditions. C.X. and Yiwei Zhou designed, performed and 2+ Science 335, 851–855 (2012). analyzed Ca -flux and IP-accumulation assays of WT and mutant PAFRs. L.H. and 13. Chrencik, J. E. et al. Crystal structure of antagonist bound human C.C. designed, performed and analyzed smFRET assays. L.Y. performed and analyzed lysophosphatidic acid receptor 1. Cell 161, 1633–1643 (2015). MD simulations and docking assays. C.C. and Ye Zhou designed, performed and 14. Srivastava, A. et al. High-resolution structure of the human GPR40 receptor analyzed ligand-binding assays of WT and mutant PAFRs. A.Q. and M.L. assisted in bound to allosteric agonist TAK-875. Nature 513, 124–127 (2014). construct and crystal optimization. C.Y. expressed the PAFR proteins. G.W.H. assisted 15. Hua, T. et al. Crystal structure of the human cannabinoid receptor CB1. in structure refinement. X.W. and X.L. helped to develop the initial expression and Cell 167, 750–762.e14 (2016). purification protocol for PAFR. H.Y. oversaw computational assays. Z.R. oversaw 16. Curtin, M. L. et al. Discovery and evaluation of a series of 3-acylindole structure analysis and interpretation. H.J. oversaw computational assays and structure 2+ imidazopyridine platelet-activating factor antagonists. J. Med. Chem. 41, analysis and interpretation. Y. Zhao oversaw smFRET assays. J.L. oversaw Ca -flux and 74–95 (1998). IP-accumulation assays, and edited the manuscript. R.C.S. assisted in structure analysis 17. Ryan, S. D., Harris, C. S., Carswell, C. L., Baenziger, J. E. & Bennett, S. A. and interpretation, and edited the manuscript. Q.Z. oversaw construct design, collected Heterogeneity in the sn-1 carbon chain of platelet-activating factor crystal diffraction data, solved the PAFR structures and assisted with manuscript glycerophospholipids determines pro- or anti-apoptotic signaling in primary preparation. X.C.Z. and B.W. initiated the project, planned and analyzed experiments, neurons. J. Lipid Res. 49, 2250–2258 (2008). solved the structures, supervised the research and wrote the manuscript. 18. Wess, J., Han, S. J., Kim, S. K., Jacobson, K. A. & Li, J. H. Conformational changes involved in G-protein-coupled-receptor activation. Trends Pharmacol. Competing interests Sci. 29, 616–625 (2008). The authors declare no competing interests. 19. Kuwasako, K., Kitamura, K., Nagata, S., Hikosaka, T. & Kato, J. Structure- function analysis of helix 8 of human calcitonin receptor-like receptor within the adrenomedullin 1 receptor. Peptides 32, 144–149 (2011). Additional information 20. Wacker, D., Stevens, R. C. & Roth, B. L. How ligands illuminate GPCR Supplementary information is available for this paper at https://doi.org/10.1038/ molecular pharmacology. Cell 170, 414–427 (2017). s41594-018-0068-y. 21. Rasmussen, S. G. et al. Crystal structure of the ​2 adrenergic receptor–Gs β Reprints and permissions information is available at www.nature.com/reprints. protein complex. Nature 477, 549–555 (2011). 22. Park, J. H., Scheerer, P., Hofmann, K. P., Choe, H. W. & Ernst, O. P. Crystal Correspondence and requests for materials should be addressed to X.C.Z. or B.W. structure of the ligand-free G-protein-coupled receptor opsin. Nature 454, Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in 183–187 (2008). published maps and institutional affiliations.

Nature Structural & Molecular Biology | VOL 25 | JUNE 2018 | 488–495 | www.nature.com/nsmb 495 © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Articles Nature Structural & Molecular Biology

Methods residues A2–I148 of flavodoxin. Extra electron densities are present around the Cloning and expression of engineered PAFR proteins. Te human PAFR cDNA intracellular part of helix IV, thus suggesting that the conformation of this region (ProteinTech) was cloned into a modifed pFastBac 1 vector containing an HA is dynamic. However, modeling of dual conformations caused strong positive signal peptide at the N terminus and a PreScission protease site followed by a densities in the Fo – Fc map on the previously modeled outward conformation, thus decahistidine tag and FLAG tag at the C terminus. Residues 2–148 of a modifed indicating that it is the major conformation of helix IV. The minor conformation favodoxin (P2A Y98W)7 and mT4L8 were inserted into ICL3 of PAFR by overlap- was therefore removed in the final model. extension PCR. Twenty-six amino acids (residues C317–N342) were truncated The structure of the PAFR–ABT-491 complex was solved through a similar at the C terminus of PAFR. Five point mutations, F1163.51Y, N169D, A2306.33D, protocol to that described above. Diffraction data from the 52 best-diffracting V2346.37A and D2897.49N, were introduced into the PAFR gene through standard crystals of the PAFR–ABT-491 complex were processed. The receptor portion of QuikChange PCR. the PAFR–SR 27417 structure and the mT4L of the structure of M3 (PDB 4U15) High-titer recombinant baculovirus (>​108 viral particles per milliliter) was were used as search models for MR. There were two PAFR–ABT-491 complexes in obtained with the Bac-to-Bac Baculovirus Expression System (Invitrogen). each asymmetric unit. The structure was carefully refined, and the Ramachandran- Recombinant expression of the PAFR-flavodoxin and PAFR-mT4L fusion proteins plot analysis indicated 100% of the residues in favorable (95.5%) or allowed was performed through infection of Sf9 cells (Invitrogen) at a density of 2 ×​ 106 (4.5%) regions, with no outliers. The final model contains 277 residues (S6–L43, cells/ml with high-titer viral stock at a multiplicity of infection of 5. Cells were F49–P122, Q127–P217 and A224–T297) of PAFR in molecule A, and 283 residues routinely tested for mycoplasma contamination. The ligand SR 27417 (Sigma) or (S7–P45, F49–I123, Q127–V218 and A224–F300) of PAFR and residues N2–Y118 ABT-491 (Santa Cruz Biotechnology) was added at 1 μ​M concentration during of mT4L in molecule B. expression. Cells were harvested at 48 h post infection by centrifugation and stored at –80 °C until use. Molecular dynamics simulations. To monitor the conformational stability of helices II, IV and VIII, we conducted a 2-μs​ all-atom MD simulation on the Purification of PAFR proteins for crystallization. Insect cells were thawed on ice antagonist SR 27417–bound PAFR structure. The fusion protein flavodoxin in and disrupted by resuspension in a hypotonic buffer containing 10 mM HEPES, the crystal structure was removed, and the missing parts of ICL2 and helix IV pH 7.5, 10 mM MgCl2, 20 mM KCl and EDTA-free protease-inhibitor cocktail (K124–L137) and ICL3 (Q217–A223) in the structure were built in MODELER (Roche). To remove membrane-associated proteins and nucleic acids, extensive in Discovery Studio 2.6 (Accelrys Software). Then this PAFR model (S6–R315) washing of membranes was performed by repeated centrifugation and Dounce was used as the starting structure for the MD simulation, which was embedded homogenization. To purify the PAFR–SR 27417 complex, the membranes from in a 90 Å ×​ 90 Å palmitoyl oleoyl phosphatidylcholine (POPC) bilayer, and the cells expressing the PAFR-flavodoxin fusion protein were solubilized in 30 mM lipids located within 1 Å of the receptor were removed. The system was solvated 3 HEPES, pH 7.5, 0.5% (wt/vol) n-dodecyl-β​-d-maltopyranoside (DDM, Anatrace), in a box (90 ×​ 90 ×​ 110 Å ) with TIP3P waters and 0.15 M NaCl, including 80,204 39 0.1% (wt/vol) cholesterol hemisuccinate (CHS, Sigma), 150 mM NaCl, 5 mM atoms. The MD simulation was performed in the GROMACS 4.6.1 package with the isothermal–isobaric ensemble and periodic boundary conditions. We used MgCl2, 10 mM KCl, 7.5% (vol/vol) glycerol, 25 μ​M SR 27417, 1 mg/ml 40 iodoacetamide (Sigma) and EDTA-free protease-inhibitor cocktail (Roche). The the CHARMM36-CAMP force field for the protein, POPC phospholipids, ions PAFR–SR 27417 complex was purified with TALON IMAC resin (Clontech), and and water molecules, and ligand parametrization was done with the CHARMM 41,42 imidazole in the purification buffer was then removed with a PD MiniTrap G-25 generalized force field (CGenFF) . Energy minimizations were first performed to column (GE Healthcare). The protein was then treated overnight with 30 μ​l relieve unfavorable contacts in the system, and this was followed by equilibration histidine-tagged PreScission protease (custom made) to remove the C-terminal steps of 50 ns in total to equilibrate the lipid bilayer and the solvent with restraints histidine tag. The cleaved histidine tag and PreScission protease were removed by on the ligand and the main chain of PAFR. Subsequently, one 2-μ​s production run passage of the protein solution through Ni–NTA superflow resin (Qiagen). The was performed. The temperature of the system was maintained at 310 K with the 43 protein was then concentrated to 30–40 mg/ml with a Vivaspin concentrator with v-rescale method with a coupling time of 0.1 ps. The pressure was kept at 1 bar 44 −5 −1 a 100-kDa molecular-weight cutoff (Sartorius Stedim Biotech). The PAFR–ABT- with a Berendsen barostat with τ​p =​ 1.0 ps and a compressibility of 4.5 ×​ 10 bar . 45 46 491 complex was purified from membranes of cells expressing the PAFR-mT4L SETTLE constraints and LINCS constraints were applied on the hydrogen- fusion protein through the same protocol described above, then concentrated to involving covalent bonds in water molecules and in other molecules, respectively, 40–45 mg/ml before crystallization. More details of protein purification can be and the time step was set to 2 fs. Electrostatic interactions were calculated with the 47 found in the Supplementary Note. particle-mesh Ewald algorithm with a real-space cutoff of 1.4 nm.

Crystallization of PAFR–SR 27417 and PAFR–ABT-491 complexes in lipidic Molecular docking of PAF. To obtain a molecular model of the PAFR–PAF cubic phase. Purified protein samples were reconstituted into LCP through complex, flexible dockings were performed with the induced-fit docking module in mixture with molten monoolein/cholesterol (10:1 by mass) lipids at a weight Maestro v9.0 (Schrodinger). PAF was first processed with LigPrep v2.5 to produce ratio of 2:3 (protein/lipid) with the canonical two-syringe reconstitution method. the corresponding low-energy three-dimensional structure and the correct Crystallization trials were set up with a Gryphon LCP robot (Art Robbins ionization state (pH 7.0). The crystal structure of PAFR–SR 27417 was taken as Instruments). The protein–LCP mixture was dispensed onto 96-well glass the receptor structure and prepared with the Protein Preparation Wizard module. sandwich plates (Shanghai FAstal BioTech) in 20- to 40-nl drops and overlaid Hydrogen atoms and protein charges were added during a brief relaxation period. with 700–900 nl of precipitant solutions. Protein reconstitution in LCP and Restrained partial minimization was terminated when the r.m.s. deviation reached crystallization trials was performed at room temperature (19–22 °C). Plates were a maximum value of 0.30 Å. The receptor-grid file was generated with an enclosing incubated and imaged at 20 °C with an automated incubator/imager (RockImager box centered on the antagonist SR 27417 in the crystal structure. Because PAF is a 1000, Formulatrix). Data-collection-quality crystals of the PAFR–SR 27417 larger ligand than SR 27417, the maximum length of the ligand to be docked was complex were obtained in precipitant condition containing 100 mM HEPES, pH set to 28 Å to define an appropriate box size. The processed PAF was then docked 7.0, 14–20% (vol/vol) PEG 400, 110–160 mM NaSCN, 0.1 M Na citrate and 50 μ​M into the assumed binding pocket of PAFR in both Glide-SP (standard-precision) SR 27417. The PAFR–ART-491 crystals grew in 100 mM HEPES, pH 7.0, 28–32% mode and Glide-XP (extra-precision) mode, and the top 18 conformations were output. Side chains of residues within 8 Å of ligand poses were treated as flexible, (vol/vol) PEG 400, 220–260 mM MgSO4 and 50 μ​M ABT-491. The PAFR crystals were harvested directly from LCP matrix with 30–50 μ​m MiTeGen micromounts and default settings were used for the refinement and scoring. The most reliable and were flash frozen in liquid nitrogen. binding poses were then selected according to the favorable interaction energy and our visual inspection. Data collection and structure determination. X-ray diffraction data were collected at the SPring-8 beam line 41XU, Hyogo, Japan, with a Pilatus3 6 M smFRET assays. To measure the distance between the intracellular tips of helices detector (X-ray wavelength 1.0000 Å). The crystals were exposed with a 10 μ​m ×​ 8 II and IV, residues K532.40 and K1324.40 of PAFR were mutated to the unnatural μ​m mini-beam for 0.2 s and 0.2° oscillation per frame. Diffraction data from the amino acid AzK in the PAFR-flavodoxin and PAFR-mT4L constructs used for 36 best-diffracting crystals of the PAFR–SR 27417 complex were integrated and crystallization48. The genes were cloned into modified pE363 vectors for expression scaled with XDS33 and Scala34. The dataset was truncated to 2.9 Å, 2.9 Å and 2.8 Å of the decahistidine- and FLAG-tagged double-amber-codon PAFR mutants. The along the a, b and c axes, respectively. Initial phase information of the PAFR–SR plasmids were then cotransformed with pE323 and peRF1 (ref. 49) into adherent 27417 complex was obtained through molecular replacement (MR) in Phaser35, HEK293T cells (American Type Culture Collection). Cells were routinely tested with the receptor portion of purinergic P2Y1 receptor (PDB 4XNV) converted to for mycoplasma contamination. Cells were harvested after 48 h. The PAFR mutants polyalanines and flavodoxin (PDB 1F4P) as search models. There was one were copurified with SR 27417, ABT-491 or PAF with the same purification PAFR–SR 27417 complex in each asymmetric unit. The resulting model was protocol used for crystallization, then labeled with DBCO-sulfo-Cy3 and DBCO- improved through iterative rounds of refinement with REFMAC5 (ref. 36) and sulfo-Cy5 on the AzK amino acids. For the negative control measuring the distance autoBUSTER37 and subsequent manual examination and rebuilding of the refined between the intracellular tips of helices I and V, residues R421.59 and R2115.63 38 coordinates in COOT with both 2Fo – Fc and Fo – Fc maps. The structure was were mutated to AzK in the PAFR-flavodoxin construct. The mutants were then carefully refined, and the Ramachandran-plot analysis indicated 100% of the expressed, purified and labeled as described above. residues in favorable (94%) or allowed (6%) regions, with no outliers. The final Imaging chambers passivated with a mixture of PEG and biotin–PEG were model contains 289 residues (S6–I123, S138–Q216 and A224–R315) of PAFR and incubated with 0.045 mg/ml streptavidin. Protein samples were immobilized to

Nature Structural & Molecular Biology | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Nature Structural & Molecular Biology Articles the streptavidin-coated surface through monoclonal anti-FLAG BioM2 antibody The cell-associated radioactivity was measured with a scintillation counter (TriLux, (Sigma, F9291). All experiments were performed in buffer containing 25 mM MicroBeta 1450). Data were analyzed in Prism 6 (GraphPad). HEPES, pH 7.5, 150 mM NaCl, 10% (vol/vol) glycerol, 0.05% (wt/vol) DDM and 0.01% (wt/vol) CHS with an oxygen-scavenging system (0.1% glucose, Protein stability assays. Protein thermostability was tested through microscale 1 U/ml glucose oxidase, 1 U/ml catalase and 1 mM cyclooctatetraene). The fluorescence thermal stability assays with the thiol-specific fluorochrome CPM ligands SR 27417, PAF and ABT-491 were added at 25 µ​M concentration to the (N-[4-(7-diethylamino-4-methyl-3-coumarinyl)-phenyl]-maleimide), which imaging buffer. Images were taken at 50 ms/frame. Data analysis was performed in reacts with the native cysteines embedded in the protein interior as a sensor of SPARTAN50. The Cy3 and Cy5 channels were mapped with TetraSpeck fluorescent the overall integrity of the folded state. Experimental details can be found in the microsphere beads (Invitrogen, 0.1µ​m). At least ten beads were selected to obtain Supplementary Note. Protein homogeneity was tested with analytical size-exclusion the transformation matrix used in mapping in MatLab. chromatography with a 1260 Infinity HPLC system (Agilent Technologies). Photobleaching events in each trace were detected as a substantial decrease (three or more times the s.d. of background noise) in the median-filtered Reporting Summary. Further information on experimental design is available in

(window size, nine frames) total fluorescence intensity (Itotal =​ Icy3 + Icy5) without the Nature Research Reporting Summary linked to this article. returning to the previous average level. Signal-to-background-noise ratios were calculated as total intensity relative to the s.d. of background noise: Itotal/ Data availability. Atomic coordinates and structure factors have been deposited in [stdev(Icy3) +​ stdev(Icy5)]. Traces were selected automatically to meet the following the Protein Data Bank under accession codes PDB 5ZKP (for PAFR–SR 27417) and criteria: a single catastrophic photobleaching event and at least 8:1 signal to noise. PDB 5ZKQ (for PAFR–ABT-491). Source data for Figs. 4a–e and 5 are available Spectral bleed-through of Cy3 intensity on the acceptor channel was corrected by with the paper online. All other data are available from the corresponding authors subtraction of 7.5% of donor signal from the acceptor. FRET traces were calculated upon request. as: FRET =​ ICy5/(ICy3 +​ ICy5), where ICy3 and ICy5 are the instantaneous Cy3 and Cy5 fluorescence intensities, respectively. Contribution of the photophysical zero-FRET state in FRET histograms was removed by fitting the data to a two-state model References (E =​ 0.1 ±​ 0.1 and E =​ 0.4 ±​ 0.1) with the segmental k-means algorithm. 33. Kabsch, W. Xds. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010). All conditions were repeated three times on different days and showed no 34. Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D Biol. significant differences. Crystallogr. 62, 72–82 (2006). 35. McCoy, A. J. et al. Phaser crystallographic sofware. J. Appl. Crystallogr. 40, Intracellular calcium release and IP measurements. HA-tagged WT and mutant 658–674 (2007). PAFRs were expressed in HEK293 cells (American Type Culture Collection). Cells 36. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. Refnement of were routinely tested for mycoplasma contamination. For ELISAs, at 20 h post macromolecular structures by the maximum-likelihood method. Acta transfection, cells were washed twice with PBS, fixed in 4% paraformaldehyde Crystallogr. D Biol. Crystallogr. 53, 240–255 (1997). and blocked with PBS plus 5% FBS. After a 30-min reaction, anti-HA monoclonal 37. Smart, O. S. et al. Exploiting structure similarity in refnement: automated antibody conjugated with horseradish peroxidase (clone 3F10, Roche Bioscience, NCS and target-structure restraints in BUSTER. Acta Crystallogr. D Biol. 11867423001) was applied for 30 min, and cells were then washed. Bound Crystallogr. 68, 368–380 (2012). antibody was detected by chemiluminescence with Super Signal ELISA Femto 38. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and maximum-sensitivity substrate (Thermo Scientific, 37074) and a Flexstation reader development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, (Molecular Devices). 486–501 (2010). The intracellular calcium-flux assay was performed as described below. Briefly, 39. Hess, B., Kutzner, C., van der Spoel, D. & Lindahl, E. GROMACS 4: transfected cells were washed with a buffer containing 20 mM HEPES, pH 7.4, algorithms for highly efcient, load-balanced, and scalable molecular

1 mM MgSO4, 3.3 mM Na2CO3, 1.3 mM CaCl2, 0.1% bovine serum albumin and simulation. J. Chem. Teory Comput. 4, 435–447 (2008). 2.5 mM probenecid, then loaded with 1 mM Fluo-4 AM (Molecular Probes) for 40. Klauda, J. B. et al. Update of the CHARMM all-atom additive force feld for 1 h at 37 °C and incubated with PAF at different concentrations (1 pM–300 nM). lipids: validation on six lipid types. J. Phys. Chem. B 114, 7830–7843 (2010). Fluorescence signals (excitation/emission 485/525 nm) were measured with 41. Vanommeslaeghe, K. & MacKerell, A. D. Jr. Automation of the CHARMM a Flexstation reader (Molecular Devices) at 1.5-s intervals for 60 s. Data were general force feld (CGenFF) I: bond perception and atom typing. J. Chem. analyzed in Soft Max Pro software (Molecular Devices), and dose–response curves Inf. Model. 52, 3144–3154 (2012). were generated by maximum–minimum analysis of the signals and fitted in 42. Vanommeslaeghe, K., Raman, E. P. & MacKerell, A. D. Jr. Automation of the GraphPad Prism software. CHARMM General Force Field (CGenFF) II: assignment of bonded

IP accumulation in HEK293 cells was measured with an IP-One Gq kit parameters and partial atomic charges. J. Chem. Inf. Model. 52, (CisBio Bioassays). Transfected cells were plated in 96-well plates and, at 24 h 3155–3168 (2012). after transfection, were treated with the indicated ligands diluted in CisBio kit 43. Bussi, G., Donadio, D. & Parrinello, M. Canonical sampling through velocity stimulation buffer for 30 min at 37 °C. Cryptate-labeled anti-IP1 monoclonal rescaling. J. Chem. Phys. 126, 014101 (2007). antibody (CisBio Bioassays, 62IPAPEC) and d2-labeled IP1 in lysis buffer were 44. Berendsen, H. J. C., Postma, J. P. M., Vangunsteren, W. F., Dinola, A. & Haak, then added to the wells. After a 1-h incubation at room temperature, the plates J. R. Molecular-dynamics with coupling to an external bath. J. Chem. Phys. were read in a PHERAstar FS reader with excitation at 337 nm and emission 81, 3684–3690 (1984). at 620 nm and 665 nm. The accumulation of IP1 was calculated according to a 45. Miyamoto, S. & Kollman, P. A. Settle: an analytical version of the shake and standard dose–response curve. rattle algorithm for rigid water models. J. Comput. Chem. 13, 952–962 (1992). 46. Hess, B., Bekker, H., Berendsen, H. J. C. & Fraaije, J. G. E. M. LINCS: Ligand-binding assays. For saturation-binding experiments, 25 μ​l [3H]PAF C-16 a linear constraint solver for molecular simulations. J. Comput. Chem. 18, (American Radiolabeled Chemicals, 5–20 Ci/mmol) in a concentration range from 1463–1472 (1997). 0.2 nM to 1.6 μ​M was incubated with 100 μ​l of membrane preparations (15 μ​g per 47. Essmann, U. et al. A smooth particle mesh Ewald method. J. Chem. Phys. tube) from HEK293 cells expressing WT or mutant PAFRs in a total assay volume 103, 8577–8593 (1995). of 125 μ​l binding buffer containing 150 mM choline chloride, 10 mM Tris-HCl, 48. Chatterjee, A., Sun, S. B., Furman, J. L., Xiao, H. & Schultz, P. G. A versatile pH 7.5, 10 mM MgCl2 and 0.25% lipid-free bovine serum albumin. 10 μ​M ABT- platform for single- and multiple-unnatural amino acid mutagenesis in 491 or 50 μ​M PAF was used to determine the nonspecific binding. To measure Escherichia coli. Biochemistry 52, 1828–1837 (2013). the binding ability of WT and mutant PAFRs to SR 27417 or ABT-491, 100 μ​ 49. Schmied, W. H., Elsässer, S. J., Uttamapinant, C. & Chin, J. W. Efcient l of membrane preparations (15–25 μ​g per tube) from HEK293 cells expressing multisite unnatural amino acid incorporation in mammalian cells via WT or mutant PAFRs was incubated in a total volume of 125 μ​l of binding buffer optimized pyrrolysyl tRNA synthetase/tRNA expression and engineered eRF1. containing 50 nM [3H]PAF and SR 27417 or ABT-491 at different concentrations J. Am. Chem. Soc. 136, 15577–15583 (2014). (0.064–3,000 nM) at 4 °C for 4 h. Reactions were stopped by centrifugation at 50. Juette, M. F. et al. Single-molecule imaging of non-equilibrium molecular 10,000 g. Cells were washed twice with ice-cold binding buffer (500 μ​l). ensembles on the millisecond timescale. Nat. Methods 13, 341–344 (2016).

Nature Structural & Molecular Biology | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. nature research | reporting summary

Corresponding author(s): Beili Wu, Xuejun C. Zhang

Reporting Summary Nature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

Statistical parameters When statistical analyses are reported, confirm that the following items are present in the relevant location (e.g. figure legend, table legend, main text, or Methods section). n/a Confirmed The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement An indication of whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section. A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistics including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Clearly defined error bars State explicitly what error bars represent (e.g. SD, SE, CI)

Our web collection on statistics for biologists may be useful.

Software and code Policy information about availability of computer code Data collection No software was used.

Data analysis Data of calcium flux assay were analyzed using Soft Max Pro software (Molecular Devices) and dose-response curves were generated by max-min analysis of the signals and fitted using Prism GraphPad software (San Diego, CA, USA). Data of ligand-binding assays were analyzed using Prism 6 (GraphPad, San Diego, USA).

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers upon request. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information. April 2018

1 Data nature research | reporting summary Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

Atomic coordinates and structure factors for the PAFR-SR 27417 and PAFR-ABT-491 structures have been deposited in the Protein Data Bank with accession codes 5ZKP and 5ZKQ. Source data for Figure 4a-e and 5 are available with the paper online. All other data are available from the corresponding author upon request.

Field-specific reporting Please select the best fit for your research. If you are not sure, read the appropriate sections before making your selection. Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences For a reference copy of the document with all sections, see nature.com/authors/policies/ReportingSummary-flat.pdf

Life sciences study design All studies must disclose on these points even when the disclosure is negative. Sample size Due to radiation damage, X-ray diffraction data collection of the protein crystals was limited to 5-10 degree per crystal. To collect a complete data set for structure determination, diffraction data from multiple crystals were integrated and scaled using XDS. By calculating completeness of the data set, diffraction data from 36 PAFR-SR 27417 crystals and 52 PAFR-ABT-491 crystals were used to ensure the completeness was close to 100%.

Data exclusions No data were excluded from the analyses.

Replication All attempts at replication were successful.

Randomization Randomization is not relevant to this study, as protein and crystal samples are not required to be allocated into experimental groups in protein structural studies, and no animals or human research participants are involved in this study.

Blinding Blinding is not relevant to this study, as protein and crystal samples are not required to be allocated into experimental groups in protein structural studies, and no animals or human research participants are involved in this study.

Reporting for specific materials, systems and methods

Materials & experimental systems Methods n/a Involved in the study n/a Involved in the study Unique biological materials ChIP-seq Antibodies Flow cytometry Eukaryotic cell lines MRI-based neuroimaging Palaeontology Animals and other organisms Human research participants

Antibodies April 2018 Antibodies used ANTI-FLAG BioM2 antibody produced in mouse: Sigma, Cat#F9291; clone number: M2, monoclonal; lot number, SLBF5390. Anti-HA monoclonal antibody produced in rat: Roche, Cat#11867423001; clone number: 3F10, monoclonal; lot number: 16888000. Cryptate-labelled anti-IP1 monoclonal antibody: CisBio Bioassays, Cat#62IPAPEC; lot number: 12A

Validation ANTI-FLAG BioM2 antibody produced in mouse: validation statements are available on the manufacturer’s website (https:// www.sigmaaldrich.com/catalog/product/sigma/f9291?lang=en®ion=US#). Anti-HA monoclonal antibody produced in rat: validation statements available on the vendor’s website (https://

2 www.sigmaaldrich.com/catalog/product/roche/roahaha?lang=en®ion=US).

Cryptate-labelled anti-IP1 monoclonal antibody: validation statements are available in the papers titled “IP-3/IP-1 Assays - Assay nature research | reporting summary Guidance Manual” (Pubmed ID 22553873) and “Development of an Improved IP1 Assay for the Characterization of 5-HT2C Receptor Ligands” (Pubmed ID: 19922239).

Eukaryotic cell lines Policy information about cell lines Cell line source(s) Sf9 cells were obtained from Invitrogen. HEK293 cells were obtained from American Type Culture Collection (ATCC).

Authentication n/a

Mycoplasma contamination The cell lines are negative for mycoplasma contamination.

Commonly misidentified lines No commonly misidentified cell lines were used. (See ICLAC register) April 2018

3