key: cord-346670-34wfy52f
authors: Gobeil, Sophie M-C.; Janowska, Katarzyna; McDowell, Shana; Mansouri, Katayoun; Parks, Robert; Manne, Kartik; Stalls, Victoria; Kopp, Megan; Henderson, Rory; Edwards, Robert J; Haynes, Barton F.; Acharya, Priyamvada
title: D614G mutation alters SARS-CoV-2 spike conformational dynamics and protease cleavage susceptibility at the S1/S2 junction
date: 2020-10-12
journal: bioRxiv
DOI: 10.1101/2020.10.11.335299
sha: 
doc_id: 346670
cord_uid: 34wfy52f

The SARS-CoV-2 spike (S) protein is the target of vaccine design efforts to end the COVID-19 pandemic. Despite a low mutation rate, isolates with the D614G substitution in the S protein appeared early during the pandemic, and are now the dominant form worldwide. Here, we analyze the D614G mutation in the context of a soluble S ectodomain construct. Cryo-EM structures, antigenicity and proteolysis experiments suggest altered conformational dynamics resulting in enhanced furin cleavage efficiency of the G614 variant. Furthermore, furin cleavage altered the conformational dynamics of the Receptor Binding Domains (RBD) in the G614 S ectodomain, demonstrating an allosteric effect on the RBD dynamics triggered by changes in the SD2 region, that harbors residue 614 and the furin cleavage site. Our results elucidate SARS-CoV-2 spike conformational dynamics and allostery, and have implications for vaccine design. Highlights SARS-CoV-2 S ectodomains with or without the K986P, V987P mutations have similar structures, antigenicity and stability. The D614G mutation alters S protein conformational dynamics. D614G enhances protease cleavage susceptibility at the S protein furin cleavage site. Cryo-EM structures reveal allosteric effect of changes at the S1/S2 junction on RBD dynamics.

The severe acute respiratory coronavirus 2 (SARS-CoV-2) belongs to the b-coronavirus family of enveloped, positive-sense single stranded RNA viruses, and has one of the largest genomes among RNA viruses (de Wit et al., 2016) . Of the seven known coronaviruses that infect humans, four (HCov-229E, HCoV-OC43, HCoV-NL63, CoV-HKU1) circulate annually causing generally mild respiratory symptoms in otherwise healthy individuals, while the SARS-CoV-1 and Middle East respiratory syndrome coronavirus (MERS-CoV) , that are closely related to SARS-CoV-2, have resulted in the 2002 -2003 SARS and 2012 MERS epidemics (Zumla et al., 2016 , respectively. The ongoing pandemic of coronavirus disease of 2019 , is a global public health emergency with more than 37 million cases and 1 million deaths recorded worldwide (Dong et al., 2020) (https://coronavirus.jhu.edu).

The surface of the SARS-CoV-2 is decorated with the spike (S) glycoprotein Turonova et al., 2020) that is the target of most current vaccine development efforts Sempowski et al., 2020) . In its prefusion conformation the SARS-CoV-2 S protein is a large homo-trimeric glycoprotein forming a crown (from the Latin corõna) at the surface of the virus capsid. Each S protomer is subdivided into two domains, S1 and S2, which are delimited by a furin cleavage site at residue 682-685 (Figure 1) . The S1 domain comprises the N-terminal domain (NTD), an NTD-to-RBD linker (N2R), the receptor binding domain (RBD), and subdomains 1 and 2 (SD1 and SD2). The S2 domain contains a second protease cleavage site (S2') followed by the fusion peptide (FP), heptad repeat 1 (HR1), the central helix (CH), the connector domain (CD), heptad repeat 2 (HR2), the transmembrane domain (TM) and a cytoplasmic tail (CT) (Figure 1) . The S1 domain is responsible for recognition and binding to the host-cell angiotensin-converting enzyme 2 (ACE2) receptor. The S2 domain is responsible for viral-host-cell membrane fusion and undergoes large conformational changes (Hoffmann et al., 2020a) , but only upon furin cleavage and further essential processing by cleavage at the S2' site by TMPRSS2 and related proteases (Bestle et al., 2020; Hoffmann et al., 2020b; Matsuyama et al., 2020) . Previous reports have demonstrated the central role of the dynamics of the RBD domains between a "closed" (or all RBD-down receptor inaccessible conformation) and "open" (or RBD-up) conformations for recognition and binding to the host cell ACE2 receptor (Gui et al., 2017; Shang et al., 2020; Yuan et al., 2017) .

Since the early stages of the COVID-19 pandemic, virus evolution has been followed by large-scale sequencing of the virus genomes isolated from patients, and several mutations that arose and propagated within different populations have been identified even though the virus has genetic proofreading mechanisms (Elbe and Buckland-Merrett, 2017; Korber et al., 2020) . The D614G mutation in particular has attracted attention since it has quickly become the dominant strain of SARS-CoV-2 circulating worldwide (Korber et al., 2020) . The D614G mutation of the S protein has been associated in numerous reports with increased fitness and/or infectivity of the virus (Korber et al., 2020; Li et al., 2020; Weissman et al., 2020) . Cryo electron microscopy (cryo-EM) structures of the S glycoprotein ectodomain have revealed that D614 is a surface residue in the vicinity of the furin cleavage site. Mutation of this residue to a glycine is expected to disrupt a critical interprotomer hydrogen bond involving T859 of the S2 domain (Korber et al., 2020) and resulting in a shift in the observed equilibrium between the open and closed state of the S protein ectodomain (Johnson et al., 2020; Weissman et al., 2020; Yurkovetskiy et al., 2020) ( Figure 1 ).

Most structures of the SARS-CoV-2 S ectodomain currently available include two mutations, one to disrupt the furin cleavage site (RRAR to GSAS = S-GSAS), and a double proline mutation (PP) of residues 986-987, designed to prevent conformational change to the post-fusion state (Wrapp et al., 2020) . Originally designed for the MERS S protein (Pallesen et al., 2017) , insertion of two consecutive Pro mutations at the junction of the HR1 and CH regions stabilized the pre-fusion conformation of the MERS, SARS and HCoV-HKU spikes, increased protein expression, and immunogenicity for the MERS S protein (Pallesen et al., 2017) . Based on these prior data, introduction of two consecutive proline residues at the beginning of the central helix was postulated as a general strategy for retaining b-coronavirus S proteins in the prefusion conformation. Thus, the PP mutations were carried over to the SARS-CoV-2 ectodomain (Wrapp et al., 2020) that is currently widely used in the field for vaccine and structural studies, and is also the component of a vaccine candidate .

Although shown to stabilize the pre-fusion conformation of other coronaviruses, the effect of the PP insertion has not been systematically studied for the SARS-CoV-2 S ectodomain.

With the goal of investigating the biophysical and structural consequences of the D614G mutation, and to prevent the engineered PP mutations from confounding our observations, we produced two SARS-CoV-2 S ectodomain constructs with the native K986 and V987 residues, incorporating either a D or a G at position 614 (Figure 1) . The RRAR sequence in the furin cleavage site was replaced by a GSAS sequence thus rendering the S constructs furin-cleavage deficient. To probe the effect of the D614G substitution on furin cleavage of the S protein, we either reinstated the native furin sequence or replaced it with an exogeneous HRV3C proteolysis signal. We determined the cryo-EM structures of the uncleaved D614 and G614 S ectodomains, as well as the structure of the fully cleaved G614 S ectodomain of the currently globally dominant SARS-CoV-2. Our results demonstrate the effect of the D614G substitution on the conformational dynamics and furin cleavage susceptibility of the S ectodomain, and reveal insights into the allostery between the RBD and distal regions of the S protein.

While the SARS-CoV-2 S ectodomain construct that includes mutations of residues K986 and V987, between the HR1 and CH subdomains (S2 domain), to prolines (PP) (named S-GSAS/PP in this study) (Figure 1 ) is widely used in the field, the origin of this PP construct was based upon the stabilization of the pre-fusion conformation of other coronavirus spikes (Pallesen et al., 2017; Walls et al., 2020; Wrapp et al., 2020) . Here, we generated an analogous S ectodomain construct that had the native K986 and V987 residues (named S-GSAS) (Figure 1 ).

In our 293F expression system (see methods for details), both the S-GSAS/PP and S-GSAS constructs expressed at similar levels, yielding about 3 mg final protein per L of culture.

Both proteins also showed similar migration profiles on SDS-PAGE and by size exclusion chromatography (SEC) on a Superose 6 column (Figure 2A, B) .

Similar to the S-GSAS/PP construct Henderson et al., 2020; Wrapp et al., 2020) , S-GSAS showed 80-90% intact prefusion spike trimers by negative stain electron microscopy (NSEM) (Figure 2C ). This finding is in contrast to previous observations for MERS and the SARS-CoV-1 ectodomains, which showed a mixture of the prefusion and postfusion conformations unless the PP mutation was included (Pallesen et al., 2017) . Binding of S-GSAS and S-GSAS/PP measured by ELISA to ACE2 and CR3022 , both requiring an RBD-up conformation, Ab712199 and Ab511584, two antibodies isolated from a COVID-19 convalescent donor with epitopes mapping to the ACE2 binding site and S2 domain respectively, and 2G12, binding to a quaternary S2 glycan epitope, were all nearly identical demonstrating that both constructs showed similar antigenic behavior ( Figure 2D ). Using differential scanning fluorimetry to measure the spike thermostability, we found the S-GSAS and S-GSAS/PP ectodomains showed similar melting temperatures ( Figure 2E ).

We next solved cryo-EM structures of the S-GSAS ectodomain (Figures 2F-H , S1-S2, Table S1 ), to compare with the S-GSAS/PP structures (Walls et al., 2020; Wrapp et al., 2020) and to visualize the impact that the engineered PP mutations had on the structure of the SARS-CoV-2 spike ectodomain. Two populations of the S-GSAS S ectodomain were identified in the cryo-EM dataset -a 1-RBD-up (or open) and a 3-RBD-down (or closed) conformation ( Figure   2F and Table S1 ). Both structures were similar to the corresponding structures of S-GSAS/PP (Walls et al., 2020) , with overall RMSDs of 0.45 Å and 0.54 Å for the 1-up and 3-down structures, respectively. In the region around the PP mutations, we found the S-GSAS structures to be similar to the corresponding S-GSAS/PP structures ( Figure 2H ). In the S-GSAS 1-RBD-up structure, we observed that the K986 side chain was appropriately positioned to make an interprotomer salt bridge with the D427 residue of the RBD of the adjacent protomer, an interaction that would be abrogated in the PP construct. The corresponding residues in the MERS S protein, V1060 and L1061 are non-polar, and the adjacent protomers too far to interact with these residues (Figure S3 ). In the SARS-CoV-1 S protein cryo-EM structure (PDB 5XLR, 5X5B) the residues D414-D415 (equivalent to SARS-CoV-2 D427-D428) lie further from K986 suggesting that this putative salt bridge interaction may be more transient in SARS-CoV-1.

Overall, our data show that for the SARS-CoV-2 S ectodomain, the S-GSAS construct showed similar structural, antigenic and stability behavior as the S-GSASS/PP construct that included the K986P and V987P mutations at the junction of the CH and HR1 regions. While these and analogous mutations had proved beneficial for the expression and stability of other CoVs (Pallesen et al., 2017) , for the SARS-CoV-2 S protein other compensating interactions may help confer stability to the pre-fusion form in the absence of the PP mutations. For the rest of this study we have used the S-GSAS construct as the platform for introducing mutations and other modifications of interest.

To understand the molecular details of the spike D614G mutation that arose and quickly dominated circulating SARS-CoV-2 isolates globally, we sought to assess the impact of the D614G mutation on the structure and antigenicity of the SARS-CoV-2 S ectodomain. The D614G mutated S-GSAS construct (S-GSAS/D614G), yielded an average of ~ 2 mg of purified protein per L of culture (n = 4). The SDS-PAGE, SEC and DSF profiles of the S-GSAS/D614G ( Figure 3A ) were similar to that of the S-GSAS S ectodomain (Figure 2A, B) . NSEM of the S-GSAS/D614G S ectodomain revealed typical and well-dispersed pre-fusion S particles ( Figure   3B ).

To visualize structural details at higher resolution, we determined the cryo-EM structures of S-GSAS/D614G construct ( Figure 3C -E, Table S1, Figures S4-S7) . Two major populations of the S ectodomain were identified in the cryo-EM dataset -one population with one RBD in the "up" or ACE2 receptor accessible conformation and the other with all three RBDs in the Figure 3C ). This is consistent with our previous observations made with NSEM data that showed an increase in the RBD-up population for the S-GSAS/D614G S ectodomain (Weissman et al., 2020) . Our results show that the D614G mutation in the SD2 domain, even though distal from the RBD region, has an allosteric effect on RBD dynamics leading to alteration of up/down RBD dispositions.

To understand the nature of this allostery, we examined changes in the S protein that accompany the up and down RBD transition (Figure 4 ) by comparing the RBD-up chain in the 1-RBD-up structure to the down chains in the 1-up and the 3-down structures ( Figure 3C ). In each S protein protomer, the polypeptide chain folds into domains as it traverses the length of the S1 subunit and enters the S2 subunit i.e. the NTD (residues 27-305) followed by the RBD (residues 335-521), the SD1 (residues 529-591) and SD2 (residues 592-697) domains ( Figure   4A ). The NTD and RBD are connected via a 28-residue linker spanning residues 306-334 (named N2R) that stacks against the SD1 and SD2 domains ( Figure 4A-D) , as it makes its way from the NTD to the RBD, essentially connecting all the individual domains in the S1 subunit, and forming "super" subdomains SD1' and SD2', respectively . Upon overlaying the protomers with the RBD in the up position with the protomers with their RBDs in the down position by using the S2 subunit residues 908-1035 for superpositions, we found that the down-to-up RBD motion is accompanied by a rigid body movement of the SD1' domain resulting in a shift of up to ~4.5 Å of the SD1 domain ( Figure 4D ), relative to its position in the RBD-down protomers and a shift of up to a 7 Å in the N2R linker as it hinges to enter into the RBD. This results in a ~20° tilt of residues 324-328 of the N2R linker region that forms part of the SD1' super subdomain, while residues 311-319 of the linker that associate with the SD2 subdomain remained virtually unmoved, showing only a slight tilt in the b beta strand that accompanied large movements in the RBD and adjoining SD1' domain ( Figure 4D ). Indeed, the SD2' super subdomain that harbors the D614G mutation appears to form a conformationally invariant anchor with the mobile RBD and NTD domains at either end ( Figure 4D) .

Additionally, the S2 subunit remains invariant between the different protomers showing that the large movements that occur in the S1 subunit are effectively arrested by the SD2' super subdomain conformationally invariant anchor.

These observations are mirrored in the difference distance matrices (DDM) comparing the RBD-up and down chains ( Figure 4E and Figure S8 ). DDM analyses (Richards and Kundrot, 1988) provide superposition-free comparisons between a pair of structures by calculating the differences between the distances of each pair of Ca atoms in a structure and the corresponding pair of Ca atoms in the second structure. The DDM analysis not only shows the large movement in the RBD region and the movement in the NTD, it also captures the movement in the N2R linker and the SD1 domain observed in the structures. Overall, these analyses show that the D614G mutation is acquired within a key region encompassing the SD2 domain and an additional b-strand contributed by residues 311-319 of the N2R linker that forms a region of relative structural stillness separating the mobile NTD and RBD, as well isolating the motions in S1 from the S2 subunit. This distal mutation altering RBD conformational dynamics shows that small changes in this region can translate into large allosteric effects, and suggests a role for the SD2 domain in modulating RBD dynamics.

In addition to the D614G mutation, the SD2 subdomain also harbors a furin cleavage site (residues 682-685) that separates the S1 and S2 subunits (Figure 1) . Cleavage of the S protein by furin at this site is essential for virus transmission (Shang et al., 2020) . The proximity of the D614G mutation to the furin cleavage site and the increased flexibility observed in the cryo-EM dataset of the S-GSAS/D614G ectodomain ( Figure 3C -E), prompted us to examine the effect of the D614G substitution on furin cleavage. Since our expression system (i.e. 293Freestyle cells) endogenously expresses furin, in order to obtain uncleaved spike that we could then test for protease cleavage in vitro, we engineered a HRV3C site (8 amino acids long) to replace the furin cleavage site (4 amino acids long) at the S1/S2 junction, resulting in the S-HRV3C and S-HRV3C/D614G S ectodomain constructs ( Figure 1A ). Both proteins expressed in 293F cells but at lower yields compared to the S-GSAS constructs (36 µg/L and 410 µg/L for the S-HRV3C and S-HRV3C/D614G proteins, respectively). SEC and SDS-PAGE profiles were similar to the S-GSAS and S-GSAS/D614G proteins confirming well-folded and homogeneous spike preparations ( Figure 5A , B). NSEM micrographs showed characteristic kite-shaped particles for the pre-fusion S protein, and 2D-classification of particles from NSEM revealed well folded spikes, further confirming that S-HRV3C spikes retained the overall fold and structure of the S-GSAS spikes (Figure 5C, D) .

To test the susceptibility of the HRV3C site engineered at the junction of the S1 and S2 subunits to protease cleavage, we incubated the purified S-HRV3C and S-HRV3C/D614G spikes with the HRV3C enzyme and followed the digestion by analyzing aliquots taken at different time-points by SDS-PAGE ( Figure 5E -G). We found that the digestion of the S-HRV3C/D614G spike ( Figure 5F -G) proceeded at a faster rate than that of the S-HRV3C spike ( Figure 5E-G) with the S-HRV3C/D614G spike almost 100% digested within the first 10 minutes of incubation, whereas, the S-HRV3C constructs only achieved 50% of cleavage after 24 hours, and a substantial portion remained uncleaved even upon addition of more enzyme followed by 4 additional hours of incubation. These results suggested that the D614G mutation increased the susceptibility of protease cleavage at the S1/S2 junction.

To study the effect of the D614G substitution on protease cleavage at the S1/S2 junction with the native furin site, we next generated spike ectodomains constructs where the furin site was restored to the native sequence, resulting in two constructs named S-RRAR and S-RRAR/D614G ( Figure 1A) . The proteins were expressed and purified using our usual methodology for the furin cleavage-deficient constructs (see methods). The SEC profiles ( Figure   6A ) showed a higher proportion of the first higher molecular weight peak. A second peak eluting at a similar molecular weight as the S-GSAS spike (at ~13.8 mL elution volume) was used for further characterization. The SEC profile of the S-RRAR spike preparation showed small populations of lower molecular weight peaks that were not observed for the S-RRAR/D614G protein ( Figure 6A ). On SDS-PAGE (Figure 6B) , the peak corresponding to the S ectodomain showed the S-RRAR construct as having one major band at the molecular weight corresponding to the spike monomer and some fainter bands corresponding to the S1 and S2 subunits while the S-RRAR/D614G protein showed a band corresponding to the spike monomer and the two bands corresponding to the molecular weights of the S1 and S2 subunits. The smaller molecular weight bands corresponding to the S1 and S2 subunits were in higher proportions in the S-RRAR/D614G spike preparation compared to the S-RRAR preparation. In summary the SEC and SDS-PAGE profiles showed that, although both the S-RRAR and S-RRAR/D614G constructs were cleaved by endogeneous furin (Figure 6B ) during protein expression the S1 and S2 subunits remained together in solution ( Figure 6A) . Consistent with the enhanced cleavage observed for the S-HRV3C/D614G spike relative to the S-HRV3C spike, in the furin-site restored spikes we observed a higher proportion of cleaved spike in S-RRAR/D614G relative to S-RRAR, suggesting that the D614G mutation makes the spike more susceptible to furin cleavage. NSEM of the purified S-RRAR ( Figure 6C ) and S-RRAR/D614G ( Figure 6D) confirmed that both of these furin site-restored spikes formed well-folded spike ectodomains. We next digested the SEC isolated fractions of the S-RRAR and S-RRAR/D614G ectodomains ( Figure 6A-D) in vitro by adding furin ( Figure 6E ). As observed for the S-HRV3C constructs, the D614 version of the spike was less susceptible to cleavage than the G614 mutant for the same incubation time with the enzyme. SEC purification of the fully digested S-RRAR/D614G ectodomain revealed a peak corresponding to the ectodomain ( Figure 6F) . On SDS-PAGE, this peak migrated as two distinct bands corresponding to the S1 and S2 domains thus confirming isolation of only the cleaved portion of the protein ( Figure 6G ). NSEM showed fully folded ectodomains for the furin digested and SEC purified S-RRAR/D614G protein ( Figure 6H ).

In summary, these results show that acquisition of the D614G mutation the S protein SD2 domain resulted in increased furin cleavage of the S ectodomain.

To visualize the structure of the furin-cleaved S ectodomain at atomic level resolution, we obtained a cryo-EM dataset, and resolved two populations of the furin-cleaved S ectodomain -a 1-RBD-up and a 3-RBD-down population ( Figure 7A, Figure S9 and S10 and Table S1 ).

We observed an increased proportion of the 3-RBD-down S compared to the uncleaved D614G S ectodomain, thus reporting a change in the RBD conformational dynamics upon furin cleavage.

Consistent with this result, we observed reduced binding to ligands such as ACE-2 and CR3022 that require the RBD to be in the up conformation for binding ( Figure 7B ). As expected, decrease in binding was also observed with antibody 712199, isolated from a convalescent COVID-19 donor, with an epitope overlapping with the ACE2 binding site. Antibody 2G12 that binds a quaternary glycan epitope in the S2 subunit showed a small decrease in binding with the furin-cleaved S ectodomain, whereas another COVID-19-derived S2 antibody 511584 showed increase in binding with the furin-cleaved S ectodomain.

We compared the different protomers in the two structures by overlaying three protomers in the asymmetric 1-RBD-up structure and one protomer from the symmetric 3-RBD-down structure using residues 908-1035 (comprising the CH and HR1 regions) for superposition ( Figure 7C ). Similar to observations made with the S-GSAS/D614G S ectodomain structure, the RBD up/down motion in the furin-cleaved G614 S ectodomain was associated with a movement in the SD1 domain and in the region of the RBD-to-NTD linker that joined the SD1 b sheet ( Figure 7C, S8B) . As observed for S-GSAS/D614G, the SD2 domain showed little conformational change and formed a stable motif anchoring the mobile NTD and RBD domains.

These observations reinforce the divergent roles of the SD1 and SD2 domain in RBD motion.

We next examined the region of the SD2 domain proximal to the NTD, and asked whether we could detect any structural changes in this region and if yes, could these be related to NTD motion. In the symmetric 3-RBD-down S ectodomain, all NTDs are identical, each stacking against the down RBD of the adjacent promoter. In the asymmetric 1-RBD-up structure, each NTD were distinct. To distinguish between these, we named the NTDs as follows: the NTD that was part of the protomer with the RBD in the up conformation was named NTD1. NTD1 stacked against a down RBD that contacted the up-RBD at one end and the second down-RBD at the other. The NTD2 stacked against a down-RBD that contacted a down-RBD at one end, and the NTD3 had the least amount of RBD contact by virtue of contacting the up-RBD ( Figure 7A ).

Observing the NTD-proximal region on the SD2 domain (marked by a dotted square on Figure   7C ) that also contacted the RBD-to-NTD linker, we noted shifts in the T602-606 loop between the different protomers. While the shifts were modest (with a maximal displacement of 2.2 Å), interestingly, identical trends were observed in the 1-RBD-up structures of the S-GSAS, S-GSAS/D614G and furin-cleaved S-GSAS/D614G S ectodomains, suggesting that this region of the SD2 domain responds to NTD motion and adopts a different conformation depending on the NTD environment ( Figure 7D ).

Thus, these data provide further evidence for allostery in the S protein, with changes in the SD2 domain impacting the RBD conformational dynamics. While the SD2 domain remains almost structurally invariant, we observe small but reproducible changes in SD2 loops in response to RBD/NTD movement suggesting that small changes in the SD2 region may translate to large motions in the RBD/NTD region.

Stabilized ectodomain constructs have proven to be useful tools to understand conformational dynamics of CoV S proteins. In particular, these have enabled high-resolution structural determination and atomic level understanding of the S ectodomain. They also are key components in vaccine development pipelines. The structural similarities in the S proteins of diverse CoVs have often enabled quick translation of structural rules and ideas from one CoV S ectodomain to another. Indeed, after the onset of the recent and ongoing COVID-19 pandemic, the SARS-CoV-2 S ectodomain could be rapidly stabilized and structurally characterized by exploiting its similarities with other CoVs and following strategies that had proved successful previously Pallesen et al., 2017; Wrapp et al., 2020) . Some of these stabilization strategies, such as introduction of proline residues in the fusion subunit to prevent transition from pre-to post-fusion, have been successful in stabilizing the pre-fusion conformation of diverse class I fusion proteins including RSV F (Krarup et al., 2015) , HIV-1 Env (Sanders et al., 2002) , Ebola and Marburg GP (Rutten et al., 2020) , influenza HA (Qiao et al., 1998) and Lassa GPC (Hastie et al., 2017) . While the underlying hypothesis for the stabilization of the S ectodomain was that introduction of 2 proline residues at the junction of the CH and HR1 helices would arrest conformational transition to the post-fusion form, we found that even without the PP mutations, the SARS-CoV-2 S ectodomain retained its pre-fusion form.

Not only so, even following furin cleavage the S ectodomain retained its pre-fusion conformation. These differences between the observed behavior of the SARS-CoV-2 S relative to other CoVs suggests that even though they retain similar overall topology and structural folds, there are differences between these CoVs that profoundly affect their structural and biological properties. Studying and accounting for these will be essential not only to understand SARS-CoV-2 but also to appreciate the nature and origin of these differences between CoVs for anticipating, preparing for and rapidly combating future CoV pandemics. Viral surface proteins that are involved in receptor binding mediated cellular entry typically consist of flexible and moving parts that exhibit large conformational changes. While this conformational flexibility is necessary for function, structural checkpoints are required to prevent premature activation and destabilization or unfolding of the protein structure.

Conformationally-silent structural islands provide the necessary stabilizing anchors for adjacent regions undergoing large motions. In this study we have identified the SD2 domain in the SARS-CoV-2 S protein as such a conformational anchor that is spatially interspersed between the highly mobile NTD and RBD regions, while itself remaining relatively invariant in its conformation. This conformational invariability of the SD2 subdomain is reminiscent of the beta sandwich structure in the HIV-1 envelope glycoprotein that connects and anchors a mobile layered architecture of the gp120 inner domain (Pancera et al., 2010) . The conformationally invariant SD2 also serves to contain the movements of the RBD and NTD to the S1 subunit, such that the S2 subunit was unchanged between the various RBD "up" and "down" protomers ( Figure 4 and Figure S8 ). This suggests a role for the SD2 domain in preventing premature triggering due to the stochastic up/down RBD motions in the SARS-CoV-2 S protein, as well as the importance of downstream events such as ACE2 receptor engagement and TMPRSS2 protease cleavage (Bestle et al., 2020; Hoffmann et al., 2020b; Matsuyama et al., 2020) in orchestrating the full extent of pre-to post-fusion transformation. In this study, we also assigned a key role to the N2R linker that connects the NTD to the RBD within a protomer. Rather than just being a connector, this 28-residue linker is also a modulator of conformational changes that are critical for receptor engagement. The linker contributes a beta strand to each of the SD1 and SD2 subdomains thus connecting all the structural domains in the S1 subunit.

In addition to the much discussed D614G mutation, the SD2 subdomain also houses the multibasic furin cleavage site that demarcates the S1 and S2 subunits. Furin cleavage is an essential processing step for the S protein and is necessary for viral infection and transmission (Hoffmann et al., 2020a; Shang et al., 2020) . We provide evidence in this study that the D614G mutation enhances susceptibility of the SARS-CoV-2 S ectodomain to furin cleavage, thus raising the possibility that this is a contributor to increased fitness and transmissibility of D614G

isolates.

In this paper, we study the effect of the D614G mutation on RBD dynamics and susceptibility to furin cleavage. We find that the D614G mutation results in increased furin cleavage susceptibility, which could be responsible for the increased transmissibility of the SARS-CoV-2 with the D614G mutation. It is important to consider though that these results are 

Further information and requests for resources and reagents should be directed to Priyamvada Acharya (priyamvada.acharya@duke.edu) .

Data and Code Availability Cryo-EM reconstructions and atomic models generated during this study are available at wwPDB and EMBD (https://www.rcsb.org; http://emsearch.rutgers.edu) under the accession codes PDB IDs 7KDG, 7KDH, 7KDK, 7KDL, 7KDI, 7KDJ, 7KE4, 7KE6, 7KE7, 7KE8, 7KE9, 7KEA, 7KEB, 7KEC and EMDB IDs EMDB-22821, EMD-22822, EMD-22825, EMD-22826, EMD-22823, EMD-22824, EMD-22831, EMD-22832, EMD-22833, EMD-22834, EMD-22835, EMD-22835, EMD-22837, EMD-22838 .

Gibco FreeStyle 293-F cells (embryonal, human kidney) were incubated at 37°C and 9% CO2 in a humidified atmosphere. Cells were incubated in FreeStyle 293 Expression Medium (Gibco) with agitation at 120 rpm. Plasmids were transiently transfected into cells using Turbo293 (SpeedBiosystems) and incubated at 37 °C, 9% CO2, 120 rpm for 6 days. On the day following transfection, HyClone CDM4HEK293 media (Cytiva, MA) was added to the cells. Antibodies were produced in Expi293 cells (embryonal, human kidney). Cells were incubated in Expi293 Expression Medium at 37°C, 120 rpm and 8% CO2 in a humidified atmosphere. Plasmids were transiently transfected into cells using the ExpiFectamine 293 Transfection Kit and protocol (Gibco).

All genes in this study were synthesized and sequenced by GeneImmune Biotechnology (Rockville, MD). The SARS-CoV-2 spike protein ectodomain constructs used comprised the protein residues 1−1208 (GenBank: MN908947) with or without the D614G mutation, with or without the furin cleavage site RRAR (residue 682-685) mutated to GSAS or LEVLFQGP (HRV3C protease site), a C-terminal T4 fibritin trimerization motif, a C-terminal HRV3C protease cleavage site (except for the constructs where the furin site was mutated to an HRV3C site), a TwinStrepTag and an 8XHisTag. All spike ectodomain constructs were cloned into the mammalian expression vector pαH (Wrapp et al., 2020) . For the ACE-2 construct, the C-terminus was fused a human Fc region.

Protein purification Spike ectodomains were harvested from filtered and concentrated supernatant using StrepTactin resin (IBA) and further purified by SEC using a Superose 6 10/300 GL Increase column preequilibrated in 2mM Tris, pH 8.0, 200 mM NaCl, 0.02% sodium azide. All protein purification steps were performed at room temperature in a single day. The purified proteins were flash frozen and stored at -80 °C in single-use aliquots. Each aliquot were thawed by incubation (~20 min) at 37 °C before use. Antibodies were produced in Expi293F cells and purified by Protein A affinity. ACE-2 with human Fc tag was purified by Protein A affinity chromatography.

Negative-stain electron microscopy Samples were diluted to 100 µg/ml in 20 mM HEPES pH 7.4, 150 mM NaCl, 5% glycerol, 7.5 mM glutaraldehyde and incubated for 5 minutes before quenching the glutaraldehyde by the addition of 1 M Tris (to a final concentration of 75 mM) and 5 minutes incubation. A 5-µl drop of sample was then applied to a glow-discharged carbon-coated grid for 10-15 seconds, blotted, stained with 2% uranyl formate, blotted and air-dried. Images were obtained using a Philips EM420 electron microscope at 120 kV, 82,000× magnification, and a 4.02 Å pixel size. The RELION (Scheres, 2012) program was used for particle picking, 2D and 3D class averaging.

Differential scanning fluorimetry DSF assay was performed using Tycho NT. 6 (NanoTemper Technologies). Spike ectodomains were diluted to approximatively 0.15 mg/ml. Intrinsic fluorescence was measured at 330 nm and 350 nm while the sample was heated from 35 to 95 °C at a rate of 30°C/min. The ratio of fluorescence (350/330 nm) and inflection temperatures (Ti) were calculated by the Tycho NT. 6 apparatus.

ELISA assays Spike samples were pre-incubated at different temperatures then tested for antibody-or ACE-2binding in ELISA assays as previously described . Assays were run in two formats. In the first format antibodies or ACE2 protein were coated on 384-well plates at 2 µg/ml overnight at 4°C, washed, blocked and followed by two-fold serially diluted spike protein starting at 25 µg/mL. Binding was detected with polyclonal anti-SARS-CoV-2 spike rabbit serum (developed in our lab), followed by goat anti-rabbit-HRP and TMB substrate. Absorbance was read at 450 nm. In the second format, serially diluted spike protein was bound in individual wells of 384-well plates, which were previously coated with streptavidin at 2 µg/mL and blocked. Proteins were incubated at room temperature for 1 hour, washed, then human mAbs were added at 10 µg/ml. Antibodies were incubated at room temperature for 1 hour, washed and binding detected with goat anti-human-HRP and TMB substrate.

Cryo-EM Purified SARS-CoV-2 spike preparations were diluted to a concentration of ~1.5 mg/mL in 2 mM Tris pH 8.0, 200 mM NaCl and 0.02% NaN3. A 2.5-µL drop of protein was deposited on a Quantifoil-1.2/1.3 grid that had been glow discharged for 10 seconds in a PELCO easiGlow™ Glow Discharge Cleaning System. After a 30 seconds incubation in >95% humidity, excess protein was blotted away for 2.5 seconds before being plunge frozen into liquid ethane using a Leica EM GP2 plunge freezer (Leica Microsystems). Frozen grids were imaged in a Titan Krios (Thermo Fisher) equipped with a K3 detector (Gatan).

No statistical analysis were performed in this study. Table  REAGENT or RESOURCE  SOURCE  IDENTIFIER  Antibodies ACE2 N/A CR3022 N/A 2G12 N/A Ab712199 N/A Ab511584 N/A Goat anti-rabbit-HRP Abcam ab97080 Goat anti-human-HRP Jackson ImmunoResearch Laboratories First derivative (ratio) Figure 3C with the S1 subunit colored by domain and the S2 subunit colored grey. RBD is colored red, NTD green, SD1 dark blue, SD2 orange and the linker between the NTD and RBD colored cyan. B. Overlay of the individual protomers in the 1-RBD-up structure and a protomer in the C3 symmetric 3-RBD-down structure shown in Figure 3C . The structures were superimposed using S2 subunit residues 908-1035 (spanning the HR1 and CH regions). The domain colors of the up-RBD chain are as described in panel A. The down-RBDs are colored salmon, the SD1 domains from the down RBD chains are colored light blue. The linker between the NTD and RBD in the down RBD chains are colored deep teal. C. Zoomed-in view showing the association of the linker connecting the NTD and RBD with the SD1 and SD2 domains. D. Zoomed-in views of individual domains marked in panel B. The N2R linker spanning residues 306-334 connects the NTD and the RBD. Residues 324-328 of the N2R linker contribute a b-strand to the SD1 subdomain together forming the SD1' "super" subdomain. Residues 311-319 of the N2R linker contribute a b-strand to the SD2 subdomain together forming the SD2' "super" subdomain. E. Difference distance matrices (DDM) showing structural changes between different protomers for the structures shown in Figure 3C . The blue to white to red coloring scheme is illustrated at the bottom. Ab712199 (RBD-directed neutralizing antibody) and Ab511584 (S2-directed non-neutralizing antibody) to S-GSAS/D614G (in blue) and the furin-cleaved S-RRAR/D614G ectodomain (in green) measured by ELISA. The assay format was the same as in Figure 2D . C. Overlay of the individual protomers in the 1-RBD-up structure and a protomer in the C3 symmetric 3-down-RBD structure shown in panel A. RBD-up chain with the S1 subunit colored by domain and the S2 subunit colored grey. RBD is colored red, NTD colored green, SD1 dark blue, SD2 orange and the linker between the NTD and RBD colored cyan. The down RBDs are colored salmon, the SD1 domains from the down RBD chains are colored light blue. The linker between the NTD and RBD in the down RBD chains are colored deep teal. Insets show zoomed-in views of individual domains similar to the depiction in Figure 4D . D. (left) The protomers of the 1-RBD-up structure of the furin-cleaved S-RRAR/D614G ectodomain superimposed using residues 908-1035 and colored by the color of their NTD as depicted in panel A. Zoomed-in views show region of the SD2 domain proximal to the NTD.

A glycan cluster on the SARS-CoV-2 spike ectodomain is recognized by Fab-dimerized glycan-reactive antibodies. bioRxiv

Real-space refinement inPHENIXfor cryo-EM and crystallography

TMPRSS2 and furin are both essential for proteolytic activation of SARS-CoV-2 in human airway cells

SARS-CoV-2 mRNA vaccine design enabled by prototype pathogen preparedness

SARS and MERS: recent insights into emerging coronaviruses

An interactive web-based dashboard to track COVID-19 in real time

Cold sensitivity of the SARS-CoV-2 spike ectodomain

Data, disease and diplomacy: GISAID's innovative contribution to global health

Features and development ofCoot

UCSF ChimeraX: Meeting modern challenges in visualization and analysis

The Bio3D packages for structural bioinformatics

Cryo-electron microscopy structures of the SARS-CoV spike glycoprotein reveal a prerequisite conformational state for receptor binding

Structural basis for antibody-mediated neutralization of Lassa virus

Controlling the SARS-CoV-2 spike glycoprotein conformation

A Multibasic Cleavage Site in the Spike Protein of SARS-CoV-2 Is Essential for Infection of Human Lung Cells

SARS-CoV-2 Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Structures and distributions of SARS-CoV-2 spike proteins on intact virions

Tracking Changes in SARS-CoV-2 Spike: Evidence that D614G Increases Infectivity of the COVID-19 Virus

A highly stable prefusion RSV F vaccine derived from structural analysis of the fusion mechanism

The Impact of Mutations in SARS-CoV-2 Spike on Viral Infectivity and Antigenicity

Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix

Enhanced isolation of SARS-CoV-2 by TMPRSS2-expressing cells

Immunogenicity and structures of a rationally designed prefusion MERS-CoV spike antigen

Structure of HIV-1 gp120 with gp41-interactive region reveals layered envelope architecture and basis of conformational mobility

UCSF Chimera?A visualization system for exploratory research and analysis

cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination

Specific single or double proline substitutions in the "spring-loaded" coiled-coil region of the influenza hemagglutinin impair or abolish membrane fusion activity

Identification of structural motifs from protein coordinate data: secondary structure and first-level supersecondary structure

Structure-Based Design of Prefusion-Stabilized Filovirus Glycoprotein Trimers

Stabilization of the soluble, cleaved, trimeric form of the envelope glycoprotein complex of human immunodeficiency virus type 1

A Bayesian View on Cryo-EM Structure Determination

Processing of Structurally Heterogeneous Cryo-EM Data in RELION

NIH Image to ImageJ: 25 years of image analysis

Pandemic Preparedness: Developing Vaccines and Therapeutic Antibodies For COVID-19

Cell entry mechanisms of SARS-CoV-2

situ structural analysis of SARS-CoV-2 spike reveals flexibility mediated by three hinges. Science

Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

D614G Spike Mutation Increases SARS CoV-2 Susceptibility to Neutralization. MedRxiv

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Cryo-EM structures of MERS-CoV and SARS-CoV spike glycoproteins reveal the dynamic receptor binding domains

Structural and Functional Analysis of the D614G SARS-CoV-2 Spike Protein Variant

Coronaviruses -drug discovery and therapeutic options

RBD "up" (1-RBD-up S) vs RBD "down" protomer #2 (1-RBD-up S)

RBD "up" (1-RBD-up S) vs RBD "down" protomer (3-RBD-down S)

RBD "up" (1-RBD-up S) vs RBD "down" protomer #1 (1-RBD-up S)

RBD "down" protomer #1(1-RBD-up S) vs RBD "down" protomer #2 (1-RBD-up S)

RBD "down" protomer #1(1-RBD-up S) vs RBD "down" protomer (3-RBD-down S)

Cryo-EM data were collected at the National Center for Cryo-EM Access and Training (NCCAT) and the Simons Electron Microscopy Center located at the New York Structural Biology Center, supported by the NIH Common Fund Transformative High Resolution Cryo-Electron Microscopy program (U24 GM129539) and by grants from the Simons Foundation