Cryo-EM Structure of the Human PAC1 Receptor Coupled to an Engineered

bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 TITLE 2 Cryo-EM structure of the human PAC1 receptor coupled to an 3 engineered heterotrimeric G protein. 4 Kazuhiro Kobayashi1*, Wataru Shihoya1*‡, Tomohiro Nishizawa1*, Francois Marie 5 Ngako Kadji2, Junken Aoki2, Asuka Inoue2, Osamu Nureki1‡. 6 7 Affiliations: 8 1 Department of Biological Sciences, Graduate School of Science, The University of 9 Tokyo, Bunkyo, Tokyo 113-0033, Japan. 10 2 Graduate School of Pharmaceutical Sciences, Tohoku University, 6-3, Aoba, Aramaki, 11 Aoba-ku, Sendai, Miyagi 980-8578, Japan. 12 13 *These authors contributed equally to this work. 14 ‡To whom correspondence should be addressed. E-mail: [email protected] (W.S.) 15 and [email protected] (O.N.) 16 17 This article is a preprint version and has not been certified by peer review. 18 19 20 21 22 23 24 25 26 27

– 1 –

28 Abstract

29 Pituitary adenylate cyclase-activating polypeptide (PACAP) is a pleiotropic 30 neuropeptide hormone functioning in the central nervous system and peripheral tissues. 31 The PACAP receptor PAC1R, which belongs to the class B G-protein-coupled receptors 32 (GPCRs), is a drug target for mental disorders and dry eye syndrome. Here we present a 33 cryo-electron microscopy structure of human PAC1R bound to PACAP and an 34 engineered Gs heterotrimer. The structure revealed that TM1 plays an essential role in 35 PACAP recognition. The ECD (extracellular domain) of PAC1R tilts by ~40° as 36 compared to that of the glucagon-like peptide-1 receptor (GLP1R), and thus does not 37 cover the peptide ligand. A functional analysis demonstrated that the PAC1R-ECD 38 functions as an affinity trap and is not required for receptor activation, whereas the 39 GLP1R-ECD plays an indispensable role in receptor activation, illuminating the 40 functional diversity of the ECDs in the class B GPCRs. Our structural information will 41 facilitate the design and improvement of better PAC1R agonists for clinical 42 applications. 43 44 Main text

45 Introduction

46 Pituitary adenylate cyclase-activating polypeptide (PACAP), a 38-amino acid 47 linear peptide discovered in extracts of ovine hypothalamus1, is a multi-functional 48 peptide hormone that acts as a neurotrophic factor, neuroprotectant, neurotransmitter, 49 immunomodulator, and vasodilator2. PACAP is distributed mainly in the central nervous 50 system (CNS), but is also detected in the testis, adrenal gland, digestive tract, and other 51 peripheral organs. PACAP shares 68% amino acid sequence homology with vasoactive 52 intestinal polypeptide (VIP). PACAP and VIP stimulate three different PACAP 53 receptors: PAC1R3, VPAC1R, and VPAC2R, with different affinities. These receptors 54 share about 50% sequence identity. The affinity of PAC1R for PACAP is higher than

– 2 –

55 that for VIP4, indicating that PAC1R is relatively selective for PACAP. 56 PAC1R belongs to the class B G-protein-coupled receptors (GPCRs), and 57 predominantly activates the adenylyl cyclase stimulatory G protein Gs. PAC1R is 58 widely expressed in the CNS and peripheral tissues2. PACAP/PAC1R signaling has been 59 implicated in playing essential roles in several cellular processes, including circadian 60 rhythm regulation, food intake control, glucose metabolism, learning and memory, 61 neuronal ontogenesis, apoptosis, and immune system regulation. Furthermore, 62 perturbations in the PACAP/PAC1R pathway cause abnormal stress responses 63 underlying posttraumatic stress disorder (PTSD)5, and thus PAC1R has been studied as 64 a drug target for numerous disorders. PACAP and PAC1R are expressed in lacrimal 65 glands, and induce tear secretion by increasing the aquaporin 5 (AQP5) levels in the 66 plasma membrane6. Therefore, PAC1R is also a drug target for dry eye syndrome. 67 However, the design of small molecule agonists for PAC1R has not yet been achieved, 68 limiting the clinical applications targeting PAC1R. 69 PAC1R comprises two distinct domains: an N-terminal extracellular domain 70 (ECD) and a transmembrane domain (TMD), as in the other class B GPCRs. A 71 two-step/two-domain model has been proposed for ligand binding and receptor 72 activation in the class B GPCRs7: the ECD is responsible for the initial and high-affinity 73 binding of peptide ligands, and the TMD plays a key role in both ligand binding and 74 receptor activation. A previous study suggested that PAC1R follows this model, and the 75 PAC1R-ECD is not required for receptor activation8. However, in glucagon-like peptide 76 1 receptor (GLP1R), the ECD also plays an indispensable role in receptor activation, 77 suggesting the divergent role of the ECD in the activation of class B GPCRs. Although 78 the crystal structure of the PAC1R-ECD was determined in a ligand-free conformation9, 79 little is known about the mechanism of the ligand recognition and signal transduction by 80 PAC1R. Here we present a cryo-electron microscopy (Cryo-EM) structure of the human 81 PAC1R, bound to the endogenous ligand PACAP and coupled to an engineered Gs 82 heterotrimer. The structure, combined with complementary functional analyses, 83 revealed the unique interaction between PACAP and the PAC1R-TMD and the 84 structural basis for the functional divergence between the PAC1R-ECD and 85 GLP1R-ECD. – 3 –

86 Results

87 Overall structure

88 To facilitate expression and purification, we truncated the C-terminal residues 89 418–468 of the human PAC1R. This truncation did not alter the Gs-coupling activity, as 90 measured by a NanoBiT-G-protein dissociation assay10 (Supplementary Fig. 1a, b, and 91 Table 1). For the cryo-EM analysis, we used the mini-Gs protein, an engineered 92 minimal G protein developed for structural studies11. The C-terminal truncated PAC1R 93 was purified in the presence of PACAP and subsequently incubated with the mini-Gs 94 heterotrimer (mini-Gs, β1, and γ2) and the nanobody Nb35, which stabilizes the 95 GPCR-Gs complex. The reconstructed complex was purified by gel filtration. 96 Vitrified complexes were imaged using a Titan Krios microscope equipped 97 with a VPP (Supplementary Fig. 2). The 3D classification revealed two different classes, 98 one containing a single complex (monomer class) and the other containing two 99 complexes with inverted molecular packing (dimer class), which probably formed 100 during the sample preparation. The structures of these two classes were determined at 101 4.5 Å and 4.0 Å resolutions, respectively, with the gold-standard Fourier shell 102 correlation (FSC) criteria. Since the cryo-EM density suggested almost identical 103 conformations in these classes, we built the atomic model of the receptor, ligand, and 104 G-protein based on the higher resolution dimer class cryo-EM map. The local resolution 105 of the map reached about 3.7 Å in the core region, including the TM helices of the 106 receptor and the α5 helix of the Gαs Ras-like domain (Fig. 1a, b, Table 2, and 107 Supplementary Fig. 3a). The molecular packing of the two complexes in the dimer class 108 is solely mediated through a weak hydrophobic contact between V3185.48 and M3225.52 109 (Wooten numbering in superscript) in TM5, and the ECD and G-protein are not engaged 110 in this interaction (Supplementary Fig. 3b), indicating that the dimerization minimally 111 affects the conformation of the Gs-complexed PAC1R structure. 112 The PAC1R-TMD adopts the typical architecture of the activated class B 113 GPCR conformation12–16, characterized by a sharp kink at TM6 (Fig. 1c). One notable

– 4 –

114 difference is observed in TM7, which is kinked at the highly conserved G3937.50 in the 115 other class B GPCRs. PAC1R has an additional glycine G3897.46 near G3937.50, and thus 116 TM7 unwinds and bends around G3897.46 in the current structure (Fig. 1d). However, 117 the G3897.46A mutation, which would facilitate the α-helical formation of the unwound 118 TM7, did not alter the Gs-coupling activity (Supplementary Fig. 1a, b, and Table 1). 119 This result suggests that this unwinding in TM7 is not related to the PAC1R function. 120 121 Interaction between PACAP and PAC1R-TMD

122 We observed an unambiguous density extending from the TMD, which allowed 123 us to assign the secondary structure and side-chain orientations of PACAP (Fig. 2a). The 124 N-terminus of the peptide ligand PACAP is directed toward the TMD core, as in the 125 other class B GPCR structures. The H1 to L27 residues of PACAP form a continuous 126 α-helix and protrude from the transmembrane binding pocket. By contrast, the residues 127 after G28 are disordered, consistent with the fact that the C-terminal truncated variant

128 PACAP1-27 has the same affinity as PACAP. Notably, the N-terminal four residues (H1 129 to G4) form a continuous α-helix, together with the I5 to L27 residues, while these 130 residues were disordered in the previous nuclear magnetic resonance (NMR) structure 17 131 of PACAP1-27 bound to detergent micelles . These residues are recognized by 13 132 residues of the receptor, and G4 of PACAP closely contacts the W3065.36 side chain of 133 the receptor (Fig. 2b, c). These interactions stabilize the α-helical structure at the 134 N-terminus of PACAP, which is essential for receptor activation. 135 The N-terminal 17 residues of PACAP create an extensive interaction network 136 with TM 1-3, 5, 7, and ECL2 of the receptor (Fig. 2b, c). The details are summarized in 137 Supplementary Table 1. Notably, PACAP forms numerous interactions with the 138 extracellular portion of TM1, involving the aromatic residues of PACAP (F6, Y10, and 139 Y13), PAC1R (Y1501.36, Y1571.43, and Y1611.47), and four hydrogen-bonding 140 interactions (D3-Y1611.47, S9-Y1501.36, Y10-K1541.40, and Y13-D1471.33) (Fig. 2b-e). 141 These close interactions with TM1 are not observed in the other class B GPCR 142 structures (Supplementary Fig. 4a-d), and are a unique feature of the PACAP-PAC1R 143 structure. – 5 –

144 PACAP can activate three types of PACAP receptors, PAC1R, VPAC1R, and 145 VPAC2R with similar affinities4. To investigate the similarity in their ligand recognition, 146 we mapped the conserved residues on the current structure (Fig. 2d, e, and 147 Supplementary Fig. 5). Notably, the residues involved in the ligand recognition are 148 highly conserved in TM1, suggesting that TM1 plays a critical role in the PACAP 149 recognition by the receptors. 150 151 Structural insight into G-protein activation

152 In the class B GPCRs, ligand binding induces the rearrangement of the central 153 polar interaction network, followed by the unwinding of TM6 at the highly conserved 154 P6.47-X-X-G6.50 motif and the opening of the intracellular cavity of the receptor for 155 G-protein coupling12,13,14. In the central region of PAC1R, we observed a similar polar 156 interaction network and the unwinding of TM6 (Fig. 3a, b). The polar interaction 157 network comprises D3 of PACAP and Y1611.47, R1992.60, N2403.43, Y2413.44, P3606.47, 158 G3636.50, H3656.52, Y3666.53, and Q3927.49 of the receptor. Notably, D3 forms a hydrogen 159 bond with Y1611.47 and an electrostatic interaction with R2602.60. R2602.60 in turn forms 160 a hydrogen bond with Y2413.44. Y3666.53, in the extracellular portion of TM6, is directed 161 toward the receptor core and participates in this network. Overall, this polar interaction 162 network extends from D3 to the carbonyl oxygens of P3606.47 and G3636.50 in TM6. A 163 previous SAR study showed that the substitution of D3 with alanine reduces both the 4 164 Emax value to 70% and the affinity for the receptor . Therefore, PACAP binding directly 165 induces the rearrangement of the polar interaction network in the central region and 166 plays a key role in receptor activation, by unwinding TM6. 167 TM6 is kinked at P3606.47 and G3636.50 in the P6.50-X-X-G6.53 motif, as in the 168 other Gs-complexed class B GPCR structures. Notably, the kink at G3636.50 is sharp (~ 169 90°), whereas that at P3606.47 is to a less extent (Fig. 3b). Previous mutational studies of 170 the calcitonin receptor family members suggested the functional importance of P6.47 in 171 receptor activation18,19. However, the P3606.47A mutation to PAC1R did not alter the 172 Gs-coupling activity (Fig. 3c and Table 1). By contrast, the G3636.50A mutation 173 completely abolished the activity, suggesting that G3636.50, rather than P3606.47, is – 6 –

174 responsible for the TM6 unwinding upon receptor activation, consistent with the 175 structural observations. 176 The intracellular cavity of the receptor closely contacts the α5-helix of Gs, 177 which is the primary determinant for the G-protein coupling (Fig. 3a, d). Specifically, 178 S3546.41 in TM6 directly hydrogen bonds with the carbonyl oxygen of L393. K3345.64 in 179 TM5 forms a salt bridge with D381, and the carbonyl oxygens of L255 and V256 in 180 ICL2 hydrogen bond with Q384 and K380, respectively. These interactions are also 181 observed in other Gs-complexed class B GPCR structures13,16 (Fig. 3e, f), suggesting 182 that they are conserved structural features of the Gs-coupling in class B GPCRs. 183 184 Diverged functional role of ECDs in class B GPCRs

185 The class B GPCRs have an ECD (about 120 amino acids) at the N-terminus, 186 which is commonly important for the initial, high-affinity binding to peptide hormones. 187 Although the ECD is less well resolved in our EM map, probably due to its flexibility, 188 we could fit the ECD region of the previous PAC1R-ECD crystal structure (PDB code: 189 3N94)9 onto the map by a rigid body. This model can facilitate discussions about the 190 interactions between PACAP and the ECD (Fig. 4a). The PAC1R-ECD adopts a 191 three-layer α–β–βα fold, which is conserved in the class B GPCRs. The C-terminal 192 portion of PACAP (Q16, V19, Y22, L23, and L27) interacts with the loops connecting 193 β1-β2 and β3-β4, and the N-terminal ends of α-helix 1 and α-helix 2 in the PAC1R-ECD, 194 as in other class B GPCRs (Supplementary Fig. 4e-g). Furthermore, the PAC1R-ECD 195 has an additional α-helix, α-helix 3, in the loop connecting β3-β4, which closely 196 contacts PACAP. A previous study showed that the N-terminal splice variant 197 PAC1R-short20, which lacks residues 89-109 between the α-helix 3 and β4, exhibits 198 increased affinity for PACAP. While residues 89-110 are not modeled in our EM map, 199 we suggest that the truncation affects the conformation of the α-helix 3 and enhances 200 the interaction with PACAP. 201 The ECD in class B GPCRs also plays a key role in receptor activation. Previous 202 functional analyses demonstrated that the ECD-truncated GLP1R does not respond to 203 GLP121. In the GLP1R structure, the ECD covers the top of the C-terminal portion of – 7 –

204 GLP1, to facilitate the interactions between the N-terminal portion of GLP1 and TMD 205 (Fig. 4b). However, the PAC1R-ECD tilts by ~40° as compared with the GLP1R-ECD 206 (Fig. 4c, d), and thus it only interacts with the side of the C-terminal portion of PACAP, 207 suggesting the different role of the PAC1R-ECD. 208 To investigate the function of the PAC1R-ECD, we truncated the C-terminal 209 portion of PACAP (residues 18-38) that interacts with the ECD (Fig. 5a). The truncated

210 peptide PACAP1-17 activated the receptor at the same level, as compared with PACAP in

211 the NanoBiT-G-protein dissociation assay, while its EC50 was significantly increased by

212 about 6000-fold (Fig. 5b and Table 3), suggesting that PACAP1-17 is capable of

213 functioning as a full agonist for PAC1R. Moreover, PACAP and PACAP1-17 also 214 activated the ECD-truncated PAC1R to mostly the same level (Fig. 5c, Table 3, and 215 Supplementary Fig. 1b). These results indicate that the PAC1R-ECD functions merely 216 as an affinity trap to bind and precisely localize the peptide hormone to the receptor, 217 whereas the interaction between PACAP and the PAC1R-TMD is necessary and 218 sufficient for receptor activation. This observation is consistent with the previous study,

219 which showed that the PAC1R-TMD covalently linked to the PACAP1-12 at the 21 220 N-terminus constitutively activates the G-protein . By contrast, GLP17-23, which lacks 221 the C-terminal portion of GLP (Fig. 5d), completely lost the agonist activity for GLP1R 222 (Fig. 5e and Table 3). Furthermore, the ECD-truncated GLP1R was poorly expressed 223 and lacked receptor activity (Fig. 5f and Supplementary Fig. 1c). These results 224 confirmed that the GLP1R-ECD plays an indispensable role in receptor activation. 225 While the ECDs are commonly essential for ligand recognition in the class B GPCRs, 226 their contributions to receptor activation diverge among the receptors. 227 228 Discussion

229 We determined the PAC1R structure in complex with PACAP and the Gs-protein, 230 which revealed a unique interaction between PACAP and the PAC1R-TMD, involving 231 the aromatic residues in PACAP and TM1. Structural observations and functional 232 analyses indicated that the interaction between PACAP and the TMD is necessary and 233 sufficient for receptor activation, while the ECD is only required for the high-affinity – 8 –

234 binding. Our structural information will help the design of novel peptide-mimetic 235 agonists for PAC1R, to treat dry eye syndrome and mental disorders. 236 The class B GPCRs include 15 receptors in humans and are commonly activated 237 by peptide ligands. The class B receptors share similar characteristics, such as the 238 N-terminal ECD, and are distinct from the class A GPCRs, which are activated by 239 diverse ligands (e.g., peptides, amines, purines, and lipids)22. The structures of the class 240 B GPCRs suggested two different types of ligand recognition. In the structures of 241 calcitonin receptor12 and calcitonin receptor-like receptor (CLR)15, the N-terminal 242 portions of the peptide ligands bind to the TMD in α-helical conformations, while the 243 C-terminal portion binds to the ECD in an extended conformation (Supplementary Fig. 244 4d, h). The function of the calcitonin receptor family is modified by receptor 245 activity-modifying proteins (RAMPs). Essentially, CLR can receive calcitonin 246 gene-related peptide (CGRP) by the interaction between its ECD and RAMP1, 247 suggesting that the ECD plays a key role in both ligand binding and receptor activation. 248 In the structures of the glucagon receptor family members, GLP1R13,14 and glucagon 249 receptor, GCGR23,24, the peptide ligands adopt continuous α-helices and their ECDs 250 cover the ligands to facilitate the interactions with the TMDs, thus playing an 251 indispensable role in receptor activation. Although PACAP also adopts a continuous 252 α-helix, the PAC1R-ECD has no functional role in receptor activation, because PACAP 253 can interact with the TMD without the aid of the ECD. The PAC1R-ECD functions 254 merely as an affinity trap for the high-affinity binding of PACAP. Despite the structural 255 similarities in the class B GPCRs, the functional roles of the ECD are diverse. 256 257 Acknowledgements

258 We thank R. Danev and M. Kikkawa for setting up the cryo-EM infrastructure, 259 K. Ogomori for technical assistance, and K. Yamashita for model building. We also 260 thank Ayumi Inoue (Tohoku University, Japan) for technical assistance. This work was 261 supported by grants from the Platform for Drug Discovery, Informatics and Structural 262 Life Science by the Ministry of Education, Culture, Sports, Science and Technology 263 (MEXT), JSPS KAKENHI grants 16H06294 (O.N.), 17J30010 (W.S.), 30809421 (W.S.), – 9 –

264 17K08264 (A.I.), 17H05000 (T.N.) and the Japan Agency for Medical Research and 265 Development (AMED) grants: the PRIME JP18gm5910013 (A.I.) and the LEAP 266 JP18gm0010004 (A.I. and J.A.), and the National Institute of Biomedical Innovation. 267 268 Author contributions

269 K.K. expressed and purified the mini-Gs heterotrimer, and performed the 270 complex formation, grid-preparation, and cryo-EM observation. W.S. designed the 271 experiments, purified the receptor, established the preparation method for the mini-Gs 272 heterotrimer and Nb35, and refined the structure. T.N. performed the cryo-EM data 273 collection and single particle analysis. A.I., F.M.N.K., and J.A. performed and oversaw 274 the cell-based assays. The manuscript was mainly prepared by W.S., K.K., and A.I., 275 with assistance from T.N. and O.N. 276

277 Competing interests

278 The authors declare no competing interests.

279 Figures

280 Fig. 1. Overall structure of the PAC1R-mini-GSβ1γ2-Nb35 complex.

281 a, Sharpened cryo-EM map with variably colored densities (PAC1R TM: cyan, PACAP: 282 yellow, mini-Gs heterotrimer: green, red, and purple, Nb35: light blue). b, Structure of 283 the complex determined after refinement in the cryo-EM map. The model is shown as a 284 ribbon representation with the transparent map. c, Superimposition of the TMD 285 structures of PAC1R (cyan) and the other class B GPCRs determined to date (gray). d, 286 TM7 unwinding in the PAC1R structure. Residues 387 to 393 are shown as sticks. 287

– 10 –

288 Fig. 2. PACAP binding site in TMD.

289 a, Sharpened map of PACAP and the TMD, viewed from the extracellular side. PACAP 290 and TMD are shown as ribbon representations with the transparent map. b, c, Detailed 291 interactions between PACAP and the TMD, shown as ribbon representations colored as 292 in Fig. 1. Contact residues are shown as sticks. The interactions with TM1, 6, and 7 are 293 shown in (b), while those with TM2, 3, and 5 are shown in (c). Hydrogen-bonding 294 interactions are indicated by black dashed lines. d, e, Sequence conservation of the 295 PACAP binding site between three types of PACAP receptors (PAC1R, VPAC1R, and 296 VPAC2R), mapped onto the PAC1R structure. Conserved and non-conserved residues 297 are colored magenta and cyan, respectively. The conserved hydrogen-bonding 298 interactions are shown in (d), and all of the conserved residues are shown in (e). 299 300 Fig. 3. Mechanism of receptor activation and Gs coupling.

301 a, Ribbon representation of the PAC1R-TMD, PACAP, and α5-helix of mini-Gs, viewed 302 from the membrane plane and colored as in Fig. 1. b, The residues involved in the 303 central polar interaction network are represented by sticks. Hydrogen bonding 304 interactions are indicated by black dashed lines. c, PAC1R-mediated Gs activation, 305 measured by the NanoBiT-G-protein dissociation assay. Cells transiently expressing the 306 NanoBiT-Gs along with the indicated PAC1R construct were treated with PACAP 307 (1-38), and the change in the luminescent signal was measured. d-f, Cytoplasmic views 308 of the PAC1R (cyan) with the C-terminal α5 helix of Gαs (yellow) (d), compared to the 309 GLP1-GLP1R:Gs complex (orange, PDB 5VAI) (e) and the LA-PTH-PTH1R:Gs 310 complex (pink, PDB 6NBF) (f). Hydrogen bonding interactions are indicated by black 311 dashed lines. 312 313 Fig. 4. Structural comparison of PAC1R and GLP1R.

314 a, Interaction between PACAP and the PAC1R-ECD. The PAC1R-ECD is shown as a 315 ribbon representation with the transparent sharpened map. C54 is shown as a stick

– 11 –

316 model. b, c, Surface representations of GLP1R (b) and PAC1R (c). The receptors are 317 shown as ribbon representations with transparent surfaces. PACAP and PAC1R are 318 colored as in Fig. 1a. GLP1 and GLP1R are colored orange and light-green, respectively. 319 d, Superimposition of PAC1R and GLP1R, viewed from the membrane plane. 320 321 Fig. 5. Characterization of the truncated analogs of PACAP and GLP1.

322 a, Overall structure of the PACAP-bound PAC1R, viewed from the membrane plane. 323 PACAP and PAC1R are shown as ribbon representations, colored as in Fig. 1a. The 324 C-terminal portion of PACAP (18-38) is shown with increased transparency. b, c, 325 PACAP-induced Gs activation measured by the NanoBiT-G-protein dissociation assay. 326 Cells transiently expressing the NanoBiT-Gs along with PAC1R (b) or PAC1R∆ECD 327 (c) were stimulated by the indicated PACAP peptides, and the change in the luminescent 328 signal was measured. d, Overall structure of the GLP1-bound GLP1R, viewed from the 329 membrane plane. GLP1 and GLP1R are shown as ribbon representations, colored as in 330 Fig. 4b. The C-terminal portion of GLP1 (24-37) is shown with increased transparency. 331 e, f, GLP-1-induced Gs activation measured by the NanoBiT-G-protein dissociation 332 assay. Cells transiently expressing the NanoBiT-Gs along with GLP1R (e) or 333 GLP1R∆ECD (f) were stimulated by the indicated GLP-1 peptides, and the change in 334 the luminescent signal was measured. 335 336 Table 1. Pharmacological characterization of mutant PAC1Rs.

PAC1R--WT ΔC G389A P360A G363A

n = 4 3 4 4 4

EC50 (nM) 0.73 0.38 1.2 0.83 > 1 μM

pEC50 (mean ± SEM) 9.14 ± 0.22 9.42 ± 0.03 8.92 ± 0.20 9.08 ± 0.19 < 6 337

– 12 –

338 Table 2. Data collection, processing, model refinement, and validation.

Data collection and processing FEI Titan Krios

Magnification 96,000

Voltage (kV) 300

Electron exposure (e− Å−2) 64

Defocus range (μm) −0.8 to -1.6

Pixel size (Å)a 0.861

Symmetry imposed C2 Initial particle images (no.) 980,964 Final particle images (no.) 132,808 Map resolution (Å) 4.0 FSC threshold 0.143

Map resolution range (Å) 3.7–4.9

Refinement

Initial model used (PDB code) 3N94, 6B3J

Model resolution (Å) 4.0 FSC threshold 0.143

Model resolution range (Å)

Map sharpening B factor (Å2) -162.683 Model composition Non-hydrogen atoms 8842

Protein residues 1081

– 13 –

R.m.s. deviations

Bond lengths (Å) 0.010

Bond angles (°) 1.162

Validation

Clashscore 6.46

Rotamer outliers (%) 1.88

Ramachandran plot

Favored (%) 92.14

Allowed (%) 7.86

Disallowed (%) 0

339 340 Table 3. Pharmacological characterization of truncated analogs of PACAP 341 and GLP1

PAC1R-WT PAC1R-TMD

n = 6 4

PACAP1-38 EC50 (nM) 1.3 580

pEC50 (mean ± SEM) 8.90 ± 0.10 6.24 ± 0.08

PACAP1-17 EC50 (μM) 8.1 19

pEC50, (mean ± SEM) 5.09 ± 0.09 4.72 ± 0.17 GLP1R-WT GLP1R-TMD n = 4 3

GLP17-37 EC50 (nM) 1.2 >1000

pEC50 (mean ± SEM) 8.93 ± 0.17 < 6

GLP17-23 EC50 (μM) >100 >100

pEC50, (mean ± SEM) <4 <4 342

– 14 –

343 344 Expression and purification of the human PAC1R

345 The N-terminal signal sequence in human PAC1R (Genbank ID: AK290046) 346 was replaced with the haemagglutinin signal peptide. The C-terminus was truncated 347 after S417. The modified receptor was subcloned into a modified pFastBac vector25, 348 with the resulting construct encoding a TEV cleavage site followed by a GFP-His10 tag 349 at the C-terminus. The recombinant baculovirus was prepared using the Bac-to-Bac 350 baculovirus expression system (Invitrogen). Sf9 insect cells were infected with the 351 virus at a cell density of 4.0 × 106 cells per milliliter in Sf900 II medium, and grown 352 for 48 h at 27 °C. The harvested cells were disrupted by sonication, in buffer 353 containing 20 mM Tris-HCl, pH 7.5, and 20% glycerol. The crude membrane fraction 354 was collected by ultracentrifugation at 180,000g for 1 h. The membrane fraction was 355 solubilized in buffer, containing 20 mM Tris-HCl, pH 7.5, 200 mM NaCl, 1% LMNG, 356 0.1 %CHS, 20% glycerol, and 1 μM PACAP38, for 2 h at 4 °C. The supernatant was 357 separated from the insoluble material by ultracentrifugation at 180,000g for 20 min, 358 and incubated with TALON resin (Clontech) for 30 min. The resin was washed with 359 ten column volumes of buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 360 0.01% LMNG, 0.001% CHS, 0.1 μM PACAP38, and 15 mM imidazole. The receptor 361 was eluted in buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 0.01% 362 LMNG, 0.001% CHS, 0.1 μM PACAP38, and 200 mM imidazole. The eluate was 363 treated with TEV protease and dialyzed against buffer (20 mM Tris-HCl, pH 7.5,

364 500 mM NaCl). The cleaved GFP–His10 tag and the TEV protease were removed with 365 Co2+-NTA resin. The receptor was concentrated and loaded onto a Superdex200 366 10/300 Increase size-exclusion column, equilibrated in buffer containing 20 mM 367 Tris-HCl, pH 7.5, 150 mM NaCl, 0.01% LMNG, 0.001% CHS, and 0.1 μM PACAP38. 368 Peak fractions were pooled, concentrated to 5 mg ml−1 using a centrifugal filter 369 device (Millipore 50 kDa MW cutoff), and frozen in liquid nitrogen. 370

– 15 –

371 Expression and purification of the mini-Gs heterotrimer

372 The gene encoding mini-Gs26, with codons optimized for an E. coli expression 373 system, was synthesized (GeneArt) and subcloned into a modified pET21a(+)-vector,

374 with the resulting construct encoding a His6 tag followed by a TEV cleavage site at the 375 N-terminus. The protein was expressed in E. coli BL21 cells. Protein expression was 376 induced by 1 mM isopropyl β-D-thiogalactopyranoside (IPTG) for 20 h at 25 °C. The 377 harvested cells were disrupted by sonication, in buffer containing 20 mM Tris-HCl, 378 pH 7.5, 20% glycerol, 10 μM GDP, and 10 mM imidazole. The cell debris was removed 379 by centrifugation at 25,000g for 30 min. The supernatant was incubated with Ni-NTA 380 resin (Qiagen) for 30 min. The resin was washed with ten column volumes of buffer, 381 containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 10 μM GDP, and 30 mM imidazole. 382 The protein was eluted in buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 10 383 μM GDP, and 200 mM imidazole. The eluate was treated with TEV protease and 384 dialyzed against buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 10 μM GDP). 385 The TEV protease was removed by Ni-NTA resin. The protein was concentrated and 386 loaded onto a Hiload Superdex200 10/300 Increase size-exclusion column, equilibrated 387 in buffer containing 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 1 μM GDP). Peak 388 fractions were pooled, concentrated to 8 mg ml−1 using a centrifugal filter device 389 (Millipore 10 kDa MW cutoff), and frozen in liquid nitrogen.

390 His6-rat Gβ1 and bovine Gγ2 were subcloned into the pFastBac Dual vector. 391 The recombinant baculovirus was prepared using the Bac-to-Bac baculovirus expression 392 system (Invitrogen). Sf9 insect cells were infected with the virus at a cell density of 393 4.0 × 106 cells per milliliter in Sf900 II medium, and grown for 48 h at 27 °C. The 394 harvested cells were disrupted by sonication, in buffer containing 20 mM Tris-HCl, 395 pH 7.5, 150 mM NaCl, 10 mM imidazole, and 20% glycerol, and clarified by 396 ultracentrifugation at 180,000g for 30 min. The supernatant was incubated with 397 Ni-NTA resin (Qiagen) for 30 min. The resin was washed with ten column volumes of 398 buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 30 mM imidazole. The 399 protein was eluted in buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 400 200 mM imidazole. The protein was concentrated and loaded onto a Superdex200 – 16 –

401 10/300 Increase size-exclusion column, equilibrated in buffer containing 20 mM 402 Tris-HCl, pH 7.5, and 150 mM NaCl. Peak fractions were pooled, concentrated to 403 8 mg ml−1 using a centrifugal filter device (Millipore 10 kDa MW cutoff), and frozen 404 in liquid nitrogen.

405 The purified mini-Gs and Gβ1Gγ2 were mixed and incubated overnight on ice. The 406 sample was concentrated and loaded onto a Superdex200 10/300 Increase 407 size-exclusion column, equilibrated in buffer containing 20 mM Tris-HCl, pH 7.5, 408 150 mM NaCl, and 1 μM GDP. The fractions containing the mini-Gs heterotrimer 409 were pooled, concentrated to 8 mg ml−1 using a centrifugal filter device (Millipore 410 10 kDa MW cutoff), and frozen in liquid nitrogen. 411 412 Expression and purification of Nb35

413 The gene encoding the C-terminally His6-tagged nanobody-35 (Nb35), with 414 codons optimized for an E. coli expression system, was synthesized (GeneArt) and 415 subcloned into the pET22b(+)-vector. The protein was expressed in the periplasm of E. 416 coli C41(Rosetta) cells. Protein expression was induced by 1 mM isopropyl 417 β-D-thiogalactopyranoside (IPTG) for 20 h at 25 °C. The harvested cells were disrupted 418 by sonication, in buffer containing 20 mM Tris-HCl, pH 7.5, and 20% glycerol. The cell 419 debris was removed by centrifugation at 25,000g for 30 min. The supernatant was 420 incubated with Ni-NTA resin (Qiagen) for 30 min. The resin was washed with ten 421 column volumes of buffer, containing 20 mM Tris-HCl, pH 7.5, 500 mM NaCl, and 30 422 mM imidazole. The protein was eluted in buffer, containing 20 mM Tris-HCl, pH 7.5, 423 500 mM NaCl, and 200 mM imidazole. The eluate was dialyzed against buffer (20 mM 424 Tris-HCl, pH 7.5, 150 mM NaCl). The protein was concentrated to 3 mg ml−1 using a 425 centrifugal filter device (Millipore 10 kDa MW cutoff), and frozen in liquid nitrogen. 426

427 Formation and purification of the PAC1R-mini-GSβ1γ2-Nb35 complex

428 Purified PAC1R was mixed with a 1.2-fold molar excess of mini-Gsβ1γ2 and a 429 1.5-fold molar excess of Nb35 in the presence of apyrase (0.1 U/ml) and the mixture

– 17 –

430 was incubated on ice overnight. The sample was loaded onto a Superdex200 10/300 431 Increase size-exclusion column, equilibrated in buffer containing 20 mM HEPES-Na, 432 pH 7.5, 150 mM NaCl, 0.0075% LMNG, 0.0025% GDN, and 0.00025%CHS. Peak

433 fractions of the PAC1R-mini-Gsβ1γ2-Nb35 complex were pooled and concentrated to 8 434 mg/ml. 435 436 Sample vitrification and cryo-EM data acquisition

437 The purified complex was applied onto a freshly glow-discharged Quantifoil 438 holey carbon grid (R1.2/1.3, Cu/Rh, 300 mesh), blotted for 4 s at 4 °C in 100% humidity, 439 and plunge-frozen in liquid ethane by using a Vitrobot Mark IV. The grid images were 440 obtained with a 300kV Titan Krios G3i microscope (Thermo Fisher Scientific), 441 equipped with a GIF Quantum energy filter (Gatan), a Volta phase plate (Thermo Fisher 442 Scientific), and a Falcon III direct electron detector (Thermo Fisher Scientific). A total 443 of 2,895 movies were obtained in the electron counting mode, with a physical pixel size 444 of 0.861 Å. The data set was acquired with the EPU software, with a defocus range of 445 −0.8 to −1.6 μm. Each image was dose-fractionated to 64 frames at a dose rate of 6– 446 8 e− pixel−1 per second, to accumulate a total dose of 64 e− Å−2. In total, 2,895 447 super-resolution movies were collected. 448 449 Image processing

450 The movie frames were aligned in 5 × 5 patches, dose weighted, and binned by 2 451 in MotionCor227. Defocus parameters were estimated by CTFFIND 4.128. First, 452 template-based auto-picking was performed with the two-dimensional class averages of 453 a few hundred manually picked particles as templates. A total of 980,964 particles were 454 extracted in 3.24 Å pixel−1. These particles were subjected to three rounds of 455 two-dimensional classification in RELION 3.0. The initial model was generated in 456 RELION-3.028. Subsequently, 980,964 particles were further classified in 3D without 457 symmetry. Two stable classes showed detailed features for all subunits. One contained a 458 single complex (monomer class). The other contained two complexes in an inverted

– 18 –

459 molecular packing with C2 symmetry (dimer class). The particles of the monomer and 460 dimer classes were 282,622 and 132,808 particles, respectively, were then re-extracted 461 with the original pixel size of 1.35 Å pixel, and subsequently subjected to 3D refinement. 462 The resulting 3D models and particle sets were subjected to per-particle defocus 463 refinement, Bayesian polishing, and 3D refinement. The final 3D refinement and 464 postprocessing yielded maps of the monomer and dimer classes with global resolutions 465 of 4.5 Å and 4.0 Å, respectively. The comparison of the density maps of the two classes 466 suggested almost identical conformation, and therefore, we built an atomic model onto 467 the higher resolution map of the dimer class. All density maps were sharpened by 468 applying the temperature-factor, which was estimated using the post-processing in 469 RELION-3.1. The local resolution was estimated by RELION-3.1. The processing 470 strategy is described in Supplementary figure 2. 471

472 Model building and refinement

473 The initial template for the PAC1R transmembrane regions, PACAP, G-protein, 474 and Nb35 was derived from the structure of human GLP1R in complex with a 475 dominant-negative Gαs (PDB code: 6B3J), followed by extensive remodeling using 476 COOT29. Owing to the discontinuous and/or variable density in the ECD region, we 477 assigned the high-resolution X-ray crystal structure of the PAC1R (PDB code: 3N94)9 478 by a rigid body fit, and the model was rebuilt using Rosetta30 against the density, 479 manually readjusted using COOT, and refined using phenix.real_space_refine31. 480 Validation was performed in MolProbity32. The potential overfitting of the refined 481 models was tested by using a cross-validation method, as described previously. Briefly, 482 the final models were ‘shaken’ by introducing random shifts to the atomic coordinates 483 with an rms of 0.5 Å, and were refined against the first half map. These shaken refined

484 models were used to calculate the FSC against the same first half maps (FSChalf1 or

485 work), and the second half maps (FSChalf2 or free) that were not used for the refinement,

486 using phenix.mtriage. The small differences between the FSChalf1 and FSChalf2 curves 487 indicated no severe overfitting of the models. The curves representing model vs. full

– 19 –

488 map were calculated, based on the final model and the full, filtered and sharpened map. 489 The statistics of the 3D reconstruction and model refinement are summarized in Table 2. 490 All molecular graphics figures were prepared with CueMol (http://www.cuemol.org) 491 and UCSF Chimera33. 492

493 NanoBit G-protein dissociation assay

494 PAC1R- and GLP1R-induced Gs activation was measured by a 495 NanoBiT-G-protein dissociation assay10, in which the interaction between a Gα subunit 496 and a Gβγ subunit was monitored by a NanoBiT system (Promega). Specifically, a 497 NanoBiT-Gs protein consisting of a large fragment (LgBiT)-containing Gαs subunit and

498 a small fragment (SmBiT)-fused Gγ2 subunit, along with the untagged Gβ1 subunit, was 499 expressed with a test GPCR, and the ligand-induced luminescent signal change was 500 measured. We used the N-terminal FLAG (DYKDDDK) tagged constructs of the human 501 PAC1R, PAC1RΔECD (148-468), GLP1R, and GLP1RΔECD (140-463). HEK293 cells 34 5 502 deficient for Gq/11 were seeded in a 6-well culture plate at a concentration of 2 x 10 503 cells ml-1 (2 ml per well in DMEM (Nissui Pharmaceutical) supplemented with 10% 504 fetal bovine serum (Gibco), glutamine, penicillin, and streptomycin), one day before 505 transfection. The transfection solution was prepared by combining 4 µl (per well 506 hereafter) of polyethylenimine solution (Polysciences, 1 mg ml-1) and a plasmid mixture

507 consisting of 100 ng LgBiT-containing Gαs subunit, 500 ng Gβ1, 500 ng SmBiT-fused

508 Gγ2, and 200 ng test GPCR (or an empty plasmid) in 200 µl of Opti-MEM 509 (ThermoFisher Scientific). To prepare a larger volume of transfected cells, 10-cm 510 culture dishes (10 ml culture volume) were used with 5-fold scaling of the 6-well plate 511 contents. After an incubation for one day, the transfected cells were harvested with 0.5 512 mM EDTA-containing Dulbecco’s PBS, centrifuged, and suspended in 2 ml of HBSS 513 containing 0.01% bovine serum albumin (BSA fatty acid–free grade, SERVA) and 5 514 mM HEPES (pH 7.4) (assay buffer). The cell suspension was dispensed in a white 515 96-well plate at a volume of 80 µl per well, and loaded with 20 µl of 50 µM 516 coelenterazine (Carbosynth), diluted in the assay buffer. After 1 h incubation at room

– 20 –

517 temperature, the titrated antagonist (Atropine, NMS, or Tiotropium), diluted in the assay 518 buffer at 10X of the final concentration, was added at a volume of 10 µl per well. After 519 2 h incubation, the plate was measured for baseline luminescence (Spectramax L, 520 Molecular Devices) and 20 µl portions of 6X test compound, diluted in the assay buffer, 521 were manually added. After an incubation for 3-5 minutes at room temperature, the 522 plate was read for the second measurement. The second luminescence counts were 523 normalized to the initial counts, and the fold-changes in the signals over the vehicle 524 treatment were plotted for the G-protein dissociation response. Using the Prism 8 525 software (GraphPad Prism), the G-protein dissociation signals were fitted to a

526 four-parameter sigmoidal concentration-response curve, from which the pEC50 values

527 (negative logarithmic values of EC50 values) were used to calculate the mean and SEM.

528 The pEC50 values for PACAP-17 were calculated by restraining the “Shared values for 529 all datasets” for the “Top” and “Bottom” parameters, using both PACAP-17 and 530 PACAP-38.

531 Flow cytometry analysis

34 532 Gq/11-deficient HEK293 cells were seeded in a 12-well culture plate at a 533 concentration of 2 x 105 cells ml-1 (1 ml per well), one day before transfection. The 534 transfection solution was prepared by combining 2 µl of the polyethylenimine solution 535 (1 mg ml-1) and 500 ng of a plasmid encoding the FLAG epitope-tagged GPCR in 100 536 µl of Opti-MEM. One day after transfection, the cells were collected by adding 100 μl 537 of 0.53 mM EDTA-containing Dulbecco’s PBS (D-PBS), followed by 100 μl of 5 mM 538 HEPES (pH 7.4)-containing Hank’s Balanced Salt Solution (HBSS). The cell 539 suspension was transferred to a 96-well V-bottom plate and fluorescently labeled with 540 an anti-FLAG epitope (DYKDDDDK) tag monoclonal antibody (Clone 1E6, FujiFilm 541 Wako Pure Chemicals; 10 μg/ml diluted in 2% goat serum- and 2 mM EDTA-containing 542 D-PBS (blocking buffer)) and a goat anti-mouse IgG secondary antibody conjugated 543 with Alexa Fluor 488 (ThermoFisher Scientific, 10 μg/ml diluted in the blocking buffer). 544 After washing with D-PBS, the cells were resuspended in 200 μl of 2 mM 545 EDTA-containing-D-PBS and filtered through a 40-μm filter. The fluorescent intensity

– 21 –

546 of single cells was quantified by an EC800 flow cytometer equipped with a 488 nm 547 laser (Sony). The fluorescent signal derived from Alexa Fluor 488 was recorded in an 548 FL1 channel, and the flow cytometry data were analyzed with the FlowJo software 549 (FlowJo). Live cells were gated with a forward scatter (FS-Peak-Lin) cutoff at the 390 550 setting, with a gain value of 1.7. Values of mean fluorescence intensity (MFI) from 551 approximately 20,000 cells per sample were used for analysis.

552

553 Data Availability

554 The raw image of the PAC1R-miniGsβ1γ2-Nb35 complex after motion 555 correction has been deposited in the Electron Microscopy Public Image Archive, under 556 accession code XXXX. The cryo-EM density map and atomic coordinates for the 557 PAC1R-mini-Gs-Nb35 complex have been deposited in the Electron Microscopy Data 558 Bank and the PDB, under accession codes XXXX and ZZZZ, respectively.

559

560 Supplementary Figures

561 Supplementary Fig. 1. Functional characterization of mutant PAC1

562 receptors.

563 a, PACAP-induced Gs activation, measured by the NanoBiT-G-protein dissociation 564 assay. Cells transiently expressing the NanoBiT-Gs, along with the indicated PAC1R 565 construct, were treated with PACAP (1-38) and the change in the luminescent signal 566 was measured. b, c, Cell surface expression of the PAC1R and the GLP1R constructs. 567 Cells transiently expressing the indicated FLAG epitope-tagged PAC1R constructs (b) 568 or the GLP1R constructs (c) were labeled with an anti-FLAG tag antibody along with 569 an Alexa488-conjugated secondary antibody, and the fluorescent signals from individual 570 cells were measured by a flow cytometer.

– 22 –

571

572 Supplementary Fig. 2. Cryo-EM analysis.

573 Flow chart of the cryo-EM data processing for the PACAP–PAC1R–Gs complex, 574 including particle projection selection, classification, and 3D density map reconstruction. 575 Details are provided in the Methods section. 576

577 Supplementary Fig. 3. Map/model quality.

578 a, The cryo-EM density map and model are shown for PACAP, including all seven 579 transmembrane α-helices, ECD, and α5 of Gαs. b, Dimer interface. The complexes are 580 shown as ribbon representations, colored as in Fig. 1a. The side chains of V3185.48 and 581 M3225.52 are shown as sticks. Two complexes form an anti-parallel dimer with C2 582 symmetry in the detergent micelles. This dimer does not reflect the physiological 583 condition, but is produced during the sample preparation. The molecular packing of the 584 two complexes in the dimer class is mediated through only a weak hydrophobic contact 585 between V3185.48 and M3225.52. Therefore, the dimerization minimally affects the 586 conformation of the Gs-complexed PAC1R structure.

587 Supplementary Fig. 4. Comparison of peptide binding interactions in class

588 B GPCRs.

589 a-d, Ligand binding interactions with the TMDs in the class B GPCR structures (a, 590 PAC1R, b, GLP1R, c, PTH1R, and d, CGRP). Hydrogen bonding interactions are 591 indicated by black dashed lines. PACAP forms extensive hydrogen-bonding interactions 592 with TM1, whereas GLP1 forms only a hydrophobic contact with Y1451.40. PTH also 593 interacts with TM1; however, these interactions are mainly hydrophobic. d-f, Relative 594 positions of the peptide ligands and the ECDs in the class B GPCR structures (e, 595 PAC1R, f, GLP1R, g, PTH1R, and h, CGRP).

– 23 –

596

597 Supplementary Fig. 5. Sequence alignment of PAC1R, VPAC1R, and

598 VPAC2R.

599 Amino-acid sequences of the transmembrane domains of the human PAC1R (UniProt 600 ID: P41586), VPAC1R (P32241), and VPAC2R (P25101). Secondary structure elements 601 for α-helices and β-strands are indicated by cylinders and arrows, respectively. 602 Conservation of the residues between the PACAP receptors is indicated as follows: red 603 panels for completely conserved, red letters for partly conserved, and black letters for 604 not conserved. The residues involved in the PACAP binding are indicated by squares. 605

606 Supplementary Table 1. Interactions of the PACAP N-terminal helix with the

607 PAC1R-TMD.

608 Residues within 4.0 Å are shown.

609

610 Reference.

611 1. Miyata, A. et al. Isolation of a novel 38 residue-hypothalamic polypeptide which 612 stimulates adenylate cyclase in pituitary cells. Biochem. Biophys. Res. Commun. 613 164, 567–74 (1989).

614 2. Vaudry, D. et al. Pituitary adenylate cyclase-activating polypeptide and its 615 receptors: 20 years after the discovery. Pharmacol. Rev. 61, 283–357 (2009).

616 3. Hashimoto, H., Ishihara, T., Shigemoto, R., Mori, K. & Nagata, S. Molecular 617 cloning and tissue distribution of a receptor for pituitary adenylate 618 cyclase-activating polypeptide. Neuron 11, 333–342 (1993). – 24 –

619 4. Doan, N.-D. et al. Design and in vitro characterization of 620 PAC1/VPAC1-selective agonists with potent neuroprotective effects. Biochem. 621 Pharmacol. 81, 552–561 (2011).

622 5. Ressler, K. J. et al. Post-traumatic stress disorder is associated with PACAP and 623 the PAC1 receptor. Nature 470, 492–497 (2011).

624 6. Nakamachi, T. et al. PACAP suppresses dry eye signs by stimulating tear 625 secretion. Nat. Commun. 7, 12034 (2016).

626 7. Castro, M., Nikolaev, V. O., Palm, D., Lohse, M. J. & Vilardaga, J.-P. Turn-on 627 switch in parathyroid hormone receptor by a two-step parathyroid hormone 628 binding mechanism. Proc. Natl. Acad. Sci. U. S. A. 102, 16084–9 (2005).

629 8. Zhao, L.-H. et al. Differential Requirement of the Extracellular Domain in 630 Activation of Class B G Protein-coupled Receptors. J. Biol. Chem. 291, 15119 631 (2016).

632 9. Kumar, S., Pioszak, A., Zhang, C., Swaminathan, K. & Xu, H. E. Crystal 633 Structure of the PAC1R Extracellular Domain Unifies a Consensus Fold for 634 Hormone Recognition by Class B G-Protein Coupled Receptors. PLoS One 6, 635 e19682 (2011).

636 10. Inoue, A. et al. Illuminating G-Protein-Coupling Selectivity of GPCRs. Cell 177, 637 1933-1947.e25 (2019).

638 11. García-Nafría, J., Lee, Y., Bai, X., Carpenter, B. & Tate, C. G. Cryo-EM 639 structure of the adenosine A2A receptor coupled to an engineered heterotrimeric 640 G protein. Elife 7, (2018).

641 12. Liang, Y.-L. et al. Phase-plate cryo-EM structure of a class B GPCR–G-protein 642 complex. Nature 546, 118–123 (2017).

643 13. Zhang, Y. et al. Cryo-EM structure of the activated GLP-1 receptor in complex 644 with a G protein. Nature 546, 248–253 (2017).

– 25 –

645 14. Liang, Y.-L. et al. Phase-plate cryo-EM structure of a biased agonist-bound 646 human GLP-1 receptor–Gs complex. Nature 555, 121–125 (2018).

647 15. Liang, Y.-L. et al. Cryo-EM structure of the active, Gs-protein complexed, 648 human CGRP receptor. Nature 561, 492–497 (2018).

649 16. Zhao, L.-H. et al. Structure and dynamics of the active human parathyroid 650 hormone receptor-1. Science 364, 148–153 (2019).

651 17. Inooka, H. et al. Conformation of a peptide ligand bound to its G-protein coupled 652 receptor. Nat. Struct. Biol. 8, 161–165 (2001).

653 18. Bailey, R. J. & Hay, D. L. Agonist-dependent consequences of proline to alanine 654 substitution in the transmembrane helices of the calcitonin receptor. Br. J. 655 Pharmacol. 151, 678–687 (2007).

656 19. Conner, A. C. et al. A key role for transmembrane prolines in calcitonin 657 receptor-like receptor agonist binding and signalling: implications for family B 658 G-protein-coupled receptors. Mol. Pharmacol. 67, 20–31 (2005).

659 20. Dautzenberg, Mevenkamp, Wille & Hauger. N-Terminal Splice Variants of the 660 Type I PACAP Receptor: Isolation, Characterization and Ligand 661 Binding/Selectivity Determinants. J. Neuroendocrinol. 11, 941–949 (2001).

662 21. Zhao, L.-H. et al. Differential Requirement of the Extracellular Domain in 663 Activation of Class B G Protein-coupled Receptors. J. Biol. Chem. 291, 15119– 664 30 (2016).

665 22. Venkatakrishnan, A. J. et al. Molecular signatures of G-protein-coupled receptors. 666 Nature 494, 185–194 (2013).

667 23. Haonan Zhang, et al. Structure of the glucagon receptor in complex with a 668 glucagon analogue. Nature 553, 106–110 (2018).

669 24. Hilger, D. et al. Structural insights into ligand efficacy and activation of the 670 glucagon receptor. bioRxiv 660837 (2019). doi:10.1101/660837

– 26 –

671 25. Shihoya, W. et al. Activation mechanism of endothelin ET B receptor by 672 endothelin-1. Nature 537, 363-368, (2016).

673 26. García-Nafría, J., Lee, Y., Bai, X., Carpenter, B. & Tate, C. G. Cryo-EM 674 structure of the adenosine A2A receptor coupled to an engineered heterotrimeric 675 G protein. Elife 7, (2018).

676 27. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion 677 for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017).

678 28. Zivanov, J. et al. New tools for automated high-resolution cryo-EM structure 679 determination in RELION-3. Elife 7, (2018).

680 29. Emsley, P. & Cowtan, K. Coot : model-building tools for molecular graphics. 681 Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2126–2132 (2004).

682 30. Wang, R. Y.-R. et al. Automated structure refinement of macromolecular 683 assemblies from cryo-EM maps using Rosetta. Elife 5, (2016).

684 31. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for 685 macromolecular structure solution. Acta Crystallogr. D. Biol. Crystallogr. 66, 686 213–21 (2010).

687 32. Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular 688 crystallography. Acta Crystallogr. D. Biol. Crystallogr. 66, 12–21 (2010).

689 33. Pettersen, E. F. et al. UCSF Chimera--A visualization system for exploratory 690 research and analysis. J. Comput. Chem. 25, 1605–1612 (2004).

691 34. Schrage, R. et al. The experimental power of FR900359 to study Gq-regulated 692 biological processes. Nat. Commun. 6, 10156 (2015).

693

– 27 –

a b ECD

PACAP

PAC1R TMD

miniGαs

Gγ Nb-35

Gβ

c d

TM1 S3907.47 G3897.46

F3917.48 G3636.50 E3927.49 6.47 TM1 6.49 F362 P360 G3937.50

TM7 L3616.48 TM7 TM6 TM6

Kobayashi et al. Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a b c M17 TM3 PACAP K15 M299 1.32 TM4 N146 R14

Y13 1.33 D147 S11 TM2 N300 Y10 D8 5.52 2.70 I5 5.36 D298 1.36 W306 Y211 90° Y150 1.40 45° K154 G4 T7 F6 D3 I5 1.39 5.40 V153 K310 PAC1R TMD D3 2.67 1.43 K206 7.35 K378 Y157 5.39 3.36 L3827.39 I309 F233 TM1 S2 5.43 H1 7.43 V313 2.64 TM7 L386 1.47 3.40 V203 R3817.38 Y161 V237 3.44 7.42 Y241 E385 TM1 TM5 TM5 TM2 TM6 TM7 TM4 d e

S11 Y10 TM1 45° S11 D2985.52 Y10 T7 D2985.52 S10 Y1501.36 D8 T7 K1541.40 1.36 Y150 F6 S2 5.36 W306 K2062.67 K2062.67 7.42 D3 G4 2.64 E385 D3 V203 Y1571.43 S2 7.43 I3095.39 L386 Y1611.47 H1 7.42 E385 Y1611.47 TM2 TM7 TM1 TM5 TM6

Kobayashi et al. Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a b c

D3 basal ) S2 1.0 G363A P360A er basal ) WT 7.37 WT E385 er 0.9 R1992.60 3.44 Y241 0.8

6.53 1.47 ov hange P360A Y366 Y161 N2403.43 Gs dissociatio n 0.7 (RLU c G3636.50 ov hange 7,49 G363A TM5 Q392 -9 -8 -7 -6 -10 0 Gs dissociatio n 10 10 10 10 10 TM6 P3606.47 PACAP (1-38) [M]

TM7 (RLU c d e f PAC1R GLP1R PACPTH1RAP-38 (M)

ICL2 ICL2 ICL2

3.59 3.59 A256 V256 2.46 TM3 TM3 TM3 R380 R176 R380

3.58 Q384 3.58 L255 3.58 L255 I310 a5H Q384 a Q390 D381 5H a5H R385 L393 Q384 7.61 R385 N406 L393 5.64 H8 D381 H8 L393 K334 H8 L394 E392 5.64 S4096.41 6.41 K334 7.60 S354 V405 6.41 5.64 S352 L394 K388 L394 TM5 TM5 TM5 TM6 TM6 TM6

Kobayashi et al. Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. pEC50 (± SEM) EC50 n = a b GLP1R-GLP1 c PAC1R-PACAP d

9.14 ± 0.22 0.73 nM 4 40° α1 9.08 ± 0.19 0.83 nM 4 β4 β3 β5 β1 <6 β2 >1 M 4

L27

89-110 V23 α2 Y22 V19 α3

Q16

Kobayashi et al. Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Mock FL N C a b PAC1R c PAC1RΔECD (148-468) basal ) 1.0 basal ) 1.0 1.0 1.0 PACAP (1-38) er er PACAP (1-17) 0.9 M17 0.9 0.9 0.9

0.8 ov hange 0.8 0.8 0.8 NanoBiT-G-proteinGs dissociatio n dissociation assay

H1 (RLU c 0.7 0.7 0.7 0.7 ov hange

-9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 0 -10 0 -10 0 -10 0 -10 Gs dissociatio n Moc10 10 10 10k10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 PACAP [M] (RLU c FL N d e GLP1R PACAPf (MGLP1RΔECD) (140-463) basal )

1.0 PACAP-38 basal ) 1.0 1.0 GLP1 (7-37) er

er pEC50 (± SEM) 8.90 ± 0.10 6.24 ± 0.08 8.95GLP-1 ± 0.10 (7-37) 0.9 0.9 0.9 GLP1 (7-23) Q23 EC50 1.3 nM 580 nM 1.1 nM

0.8 ov hange 0.8 0.8

Gs dissociatio n GLP-1 (16)

0.7 H7 PACAP-17 (RLU c 0.7 0.7

ov hange pEC50 (± SEM) 5.09 ± 0.09 4.72 ± 0.17 4.86 ± 0.06

-9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 0 -10 EC50 0 -118-10.1 M 0 -1119-10 M 14 M Gs dissociatio n 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10

(RLU c n = 5 4 4 GLP-1GLP1 [M] (M)

GLP-1 (7-37)

basal ) pEC50 (± SEM) 8.93 ± 0.17 <6 pEC50 (± SEM) EC50 n = 1.0 Mock er EC50 1.2 nM >1 M 0.9 WT 9.14 ± 0.22 0.73 nM 4 0.8 GLP-1 (16) <4 <4 0.7 pEC50 (± SEM) C 9.42 ± 0.03 0.38 nM 3 ov hange EC50 >100 M >100 M

Gs dissociatio n -9 -8 -7 -6 8.92 ± 0.20 1.2 nM 4 0 -10 G389A n 10= 10 10 10 10 4 3 (RLU c PACAP-38 (M)

pEC50 (± SEM) EC50 n = basal )

er WT 9.14 ± 0.22 0.73 nM 4 P360A 9.08 ± 0.19 0.83 nM 4

ov hange G363A <6 >1 M 4 Gs dissociatio n (RLU c PACAP-38 (M)

Kobayashi et al. Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a pEC50 (± SEM) EC50 n = 1.0 pEC50 (± SEM) EC50 n =

Mock

) basal

er 0.9 4 WT 9.14 ± 0.229.14 ± 00.73.22 nM 0.73 nM 4

0.8 ov

hange C 9.42 ± 0.03 0.38 nM 3

ov hange

n dissociatio Gs

0.7 nM 3 c

(RLU G389A 9.42 ± 0.03 0.38 n

Gs dissociatio Gs 8.92 ± 0.20 1.2 nM 4

c (RLU

-10 -9 -8 -7 -6 8.92 ± 0.20 1.2 nM 4 0 10 10 10 10 10

PACAP (1-38) (M) b c 8000 2500

6000 pEC50 (± SEM) 2000EC50 n = ) basal 1500

4000 1.0

er WT 9.14 ± 0.22 0.73 nM 4 0.9 1000 2000 pEC50 (± SEM) EC50 n = 0.8 P360A 9.08 ± 0.19 5000.83 nM 4 GPCR expression leve l

GPCR expression leve l

) basal

0.7 (mean fluorescence unit, a.u. )

(mean fluorescence unit, a.u. ) 1.00 0 ov

hange 4

er G363AWT <6 9.14 ± >10. 2M2 0.73 nM 4

0.9 Mock -9 -8 -7 -6

-10 P360A G363A G389A n

Gs dissociatio Gs 0 10 10 10 10 10

0.8 PAC1R-WT ΔC (1-417) 9.08 ± 0.GLP1R-WT19 0.83 nM 4 c (RLU P360A ΔECD (148-468) ΔECD (140-463)

0.7 PACAP-38 (M) ov hange G363A <6 >1 M 4

-9 -8 -7 -6

-10 n

Gs dissociatio Gs 0

10 10 10 10 10

c (RLU PACAP-38 (M)

Kobayashi et al. Supplementary Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Motion Correction, CtfFind

2,895 movies

Autopick, Extract (3.24 Å/pix)

980,964 particles

Class3D

Monomer class Dimer class Monomer class Dimer class

Refine3D Refine3D

PostProcess PostProcess

1.0 1.0

0.8 0.8

282,622 particles 132,808 particles 0.6 4.5 Å 0.6 4.0 Å

0.4 0.4

0.2 0.2

FSC = 0.143 Fourier Shell Correlation FSC = 0.143

Å Å 0.0 0.0 Extract (1.35 /pix) Extract (1.35 /pix) Fourier Shell Correlation

-0.2 -0.2 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 Resolution (1/Å) Resolution (1/Å)

Refine3D, CtfRefine Refine3D, CtfRefine 0.9 Polish, Refine3D Polish, Refine3D half1 vs model half2 vs model 0.8 sum vs model

0.7 4.05 Å

0.6 FSC = 0.5 0.5

0.4

0.3

0.2

0.1 Fourier Shell Correlation

-0.1 0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 Resolution (1/Å)

4.9

4.6

90° 4.3

4.0

3.7

Kobayashi et al. Supplementary Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a TM1 (147-176) TM2 (183-211) TM3 (224-256) TM4 (265-291) TM5 (306-336) TM6 (346-371) TM7 (376-400)

Helix8 (404-416) Gαs Ras α5 (360-384) PACAP (1-28)

TM5 M3185.48

TM5 V3225.52

V3225.52

TM5 M3185.48 TM5

Kobayashi et al. Supplementary Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

a b c d

Y1451.40

TM4 TM7 TM7 TM5 TM5 TM6TM7 TM1TM7 TM5 TM4 TM5 TM6TM1 TM5 TM4 TM1TM5 TM5 TM4 TM7 TM1

e β4 f g h β3 RAMP1 β5 β4 β1 β1 β2 β2 α1 β1 β1 β4 α2 β4 β2 α1 β3 β3 α1 α2 β2

α1 α3 β3

TM7

TM1 TM2 TM7 TM1 TM2 TM7 TM1 TM2 TM7 TM1 TM2

Kobayashi et al. Supplementary Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 10 20 30 40 PAC1 MAGVVHVS LAALL L LP...... M A PA MHS D CI FKKEQAMCLEKIQR ANE LMGF N VIPR2 M RTLLPPA L LTCW LL A...... P V NS I HP EC R F HLEIQE..EETKC A EL L RSQ T VIPR1 MRPPSPLP ARWLC V LAGALAWALGPAGGQ A AR LQE E CD YVQMIEV..QHKQC LEE AQLE N

50 60 70 80 90 100 PAC1 D SSPG C PG MWDN ITCW KPAHV G EM V L V SCP E LF RI F NPDQVWETETIGESDFGDSNSLDL VIPR2 E KHKA C SG VWDN ITCW RPANV G ET V T V PCP K VF SN F Y...... VIPR1 E T.IG C SK MWDN LTCW PATPR G QV V V L ACP L IF KL F SS......

TM1

110 120 130 140 150 160 PAC1 SDMGV V S RN CT E D GW SEPF P.H YFD ACG FD EYESETGD.QDY Y YLS VK AL YT V GY ST SL V VIPR2 SKAGN I S KN CT S D GW SETF P.D FVD ACG YS DPEDES...KIT F YIL VK AI YT L GY SV SL M VIPR1 I QGRN V S RS CT D E GW THLE PGP YPI ACG LD DKAASLDEQQTM F Y GS VK TG YT I GY GL SL A

TM2

170 180 190 200 210 220 PAC1 T L TTAM V IL CR FRKLHCTRN F IH M N LF V SF M LRA I S V FI KD WI L Y AEQD S NH C ...FIS T VIPR2 S L ATGS I IL CL FRKLHCTRN Y IH L N LF L SF I LRA I S V LV KD DV L Y SSSG T LH C PDQPSSW VIPR1 T L LVAT A IL SL FRKLHCTRN Y IH M H LF I SF I LRA A A V FI KD LA L F DSGE S DQ C S...EGS

TM3 TM4

230 240 250 260 270 280 PAC1 VE CK AV M VF FH YC VV S N Y FWL F I EGLYL F TLL VETFFPE R R Y FYW YT I IGWG T P TVCVT V VIPR2 V GCK L S LVF LQ YC IM AN FFWL L VEGLYL HTLL V AMLPP. R R CF LA Y L LIGWG LP T VCIG A VIPR1 VG CK AA M VF FQ YC VM A N F FWL L V EGLYL Y TLL AVSFFSE R K Y FWG YI L IGWG V P STFTM V

TM5

290 300 310 320 330 340 PAC1 WAT L R LYF D DT GCWD MNDS TAL WW V I KG P VV G SI M VNF V LFI G II V IL V QKL QS PD MGG N VIPR2 W TA AR L YL ED TGCWD TNDH S VP WW VI R IP IL ISI IVNF VLFI SII RIL LQKL TS PD V GG N VIPR1 WTI A R IHF E DY GCWD TIN. SSL WW I I KG P IL T SI L VNF I LFI C II R IL L QKL RP PD IRK S

TM6 TM7

350 360 370 380 390 400 PAC1 ES S IY LRLA RSTLLLIPLFG IHY T VFA FS P E N VSKRER LV FEL G LGSFQG FVVA VLYCFL VIPR2 DQ SQ YK RLA K STLLLIPLFG V HY M V FA VF P I SISSKYQ IL FEL C LGSFQG L VVA VLYCFL VIPR1 DS S PY SRLA RSTLLLIPLFG VHY I MFA FF P D N FKPEVK MV FEL V VGSFQG FVVA ILYCFL

410 420 430 440 450 460 PAC1 NG EVQ A E IK RKWR SWKV NRYFAVDF K HRHP S LAS S G VNGGT Q LSIL SK S SSQ I RMSGLPA VIPR2 NS EVQ C E LK RKWR SRCP TPSASRDY R VCGS S FSR N G SEGAL Q FHRG SR A QSF L QTETSVI VIPR1 N GEVQ AE LR RKWR RWHL Q GVLGWNP K YRHP S GGS N G ATCST Q VSML TR V S PG A RRSSSFQ

PAC1 DNLAT. VIPR2 ...... VIPR1 AEVSLV

Kobayashi et al. Supplementary Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/2019.12.23.887737; this version posted December 27, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. PACAP PAC1R Interaction Concerved His1 Val237 Tyr241 Trp306 Hydrophobic interaction Ile309 C Lys310 Val313 Ser2 Glu385 Hydrogen bond Leu382 Hydrophobic interaction Leu386 C Asp3 Tyr161 Hydrogen bond C Val203 C Phe233 Hydrophobic interaction Leu386 C Gly4 Asn300 Hydrophobic interaction Trp306 C Ile5 Lys378 Arg381 Hydrophobic interaction Leu382 Phe6 Tyr150 C Val153 Lys154 C Hydrophobic interaction Tyr157 C Leu382 Leu386 C Thr7 Lys206 C Tyr211 Hydrophobic interaction Asp298 C Asp8 Asp298 C Met299 Hydrophobic interaction Asn300 Ser9 Tyr150 C Hydrogen bond Lys378 Tyr10 Lys154 Hydrogen bond C Tyr150 C Hydrophobic interaction Tyr211 Ser11 Lys206 C Tyr211 Hydrophobic interaction Asp298 C Met299 Arg12 Asp301 Electrostatic interaction Tyr13 Gln146 Hydrophobic interaction Asp147 Hydrogen bond Arg14 Leu210 Hydrophobic interaction Tyr211 Lys15 Met299 Hydrophobic interaction Met17 Asp147 Hydrophobic interaction

Kobayashi et al. Supplementary Table 1