Cryo-electron microscopy structure of the SADS-CoV spike glycoprotein provides insights into an evolution of unique coronavirus spike proteins

Hongxin Guan; Youwang Wang; Abdullah F.U.H. Saeed; Jinyu Li; Syed Sajid Jan; Vanja Perculija; Yu Li; Ping Zhu; Songying Ouyang

(H-I). Structures of S1-NTD (cyan) and S1-CTD (green), the putative RBM The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.04.976258 doi: bioRxiv preprint CTDs are in the "lying down" position, it is not possible to bind ACE2 due to 221 the partial binding sites are hidden and steric clashes between binding 222 factors. This kind of conformation represents an inactive state (Fig. 2H) . The 223 CTDs in our structure keep a "lying down" state, and the receptor-binding 224 moieties are partially concealed. It needs to "stand up" on the spike trimer and 225 release the steric clash for efficient receptor binding. The linker between 226 CTDs and S1 subdomains workes as a hinge to facilitate the conformation 227 change of CTDs from "lying down" to "stand up", furthermore, transit the 228 receptor-binding inactive state to active state (Fig. 2F) . 229 It is noteworthy that stronger interactions in the SADS-CoV S trimer may 230 become obstacle conformational change and dissociation of S1-CTD in the 231 prefusion process (Fig. S5) . Moreover, to probe the intrinsic mobility of the (Fig. S6) indicated that all the three CTDs/RBDs were stabilized 238 in a "lying down" state similar to the structure observed in this study, i.e. the 239 distance differences calculated from simulations and crystal structure were 240 less than 5 Å. Although we did not find apparent expending of the CTDs/RBDs 241 from a "lying down" state to a "stand up" state during the simulations, it did not 242 . CC-BY 4.0 International license author/funder. It is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.04.976258 doi: bioRxiv preprint rule out the co-existence of the two states in the apo form. Because such a 243 large conformational transition may happen over long time scales (ms to s), 244 far beyond the simulations performed here. Taken together, these simulation 245 results may explain the homogeneity of SADS-CoV S observed in our 3D 246 classification and that the different S1-CTD conformations are invisible in our 247 results (Fig. 1B) . 248 For the NTDs, the core structure consists of two six-stranded antiparallel 249 β -sheet layers stacked together, which takes the same galectin fold as human 250 galectins and the NTDs from the other genera. Besides the core structure, 251 NTD of SADS-CoV S1 also has a loop (residues 133-150) formed as a partial CoV which has a reinforced ceiling-like structure ( Fig. 2I-M) . Based on the 255 structural similarity between the NTDs from four different CoVs genera, the 256 sugar-binding site in SADS-CoV S1-NTD might also be located in the pocket 257 formed between the core structure and the partial ceiling (Fig. 2I) . Compared 258 with other CoVs, the structure of the subdomain under the core domain of 259 SADS-CoV S1-NTD has the same situation as the partial ceiling-like structure 260 ( Fig. 2I-M) . The previous study gave the idea that NTDs from the four genera The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.04.976258 doi: bioRxiv preprint the structural evolution of SADS-CoV S1-NTD is more likely to located 265 between the other α -CoVs, δ -CoVs, and γ -CoVs. 266 For the CTDs, despite there are dramatic sequential differences between 267 SADS-CoV S1-CTD and other CTDs, all of them share similar structural 268 topology (Fig. 2N-