key: cord-0951402-jiiwtib7
authors: Socher, Eileen; Conrad, Marcus; Heger, Lukas; Paulsen, Friedrich; Sticht, Heinrich; Zunke, Friederike; Arnold, Philipp
title: Mutations in the B.1.1.7 SARS-CoV-2 spike protein reduce receptor-binding affinity and induce a flexible link to the fusion peptide
date: 2021-04-06
journal: bioRxiv
DOI: 10.1101/2021.04.06.438584
sha: 60a3b7c62ce46ab40552f44e3f2f1dd9b00a72d5
doc_id: 951402
cord_uid: jiiwtib7

The B.1.1.7 variant of the SARS-CoV-2 virus shows enhanced infectiousness over the wild type virus, leading to increasing patient numbers in affected areas. A number of single amino acid exchanges and deletions within the trimeric viral spike protein characterize this new SARS-CoV-2 variant. Crucial for viral entry into the host cell is the interaction of the spike protein with the cell surface receptor angiotensin-converting enzyme 2 (ACE2) as well as integration of the viral fusion peptide into the host membrane. Respective amino acid exchanges within the SARS-CoV-2 variant B.1.1.7 affect inter-monomeric contact sites within the spike protein (A570D and D614G) as well as the ACE2-receptor interface region (N501Y), which comprises the receptor-binding domain (RBD) of the viral spike protein. However, the molecular consequences of mutations within B.1.1.7 on spike protein dynamics and stability, the fusion peptide, and ACE2 binding are largely unknown. Here, molecular dynamics simulations comparing SARS-CoV-2 wild type with the B.1.1.7 variant revealed inter-trimeric contact rearrangements, altering the structural flexibility within the spike protein trimer. In addition to reduced flexibility in the N-terminal domain of the spike protein, we found increased flexibility in direct spatial proximity of the fusion peptide. This increase in flexibility is due to salt bridge rearrangements induced by the D614G mutation in B.1.1.7 found in pre- and post-cleavage state at the S2’ site. Our results also imply a reduced binding affinity for B.1.1.7 with ACE2, as the N501Y mutation restructures the RBD-ACE2 interface, significantly decreasing the linear interaction energy between the RBD and ACE2. Our results demonstrate how mutations found within B.1.1.7 enlarge the flexibility around the fusion peptide and change the RBD-ACE2 interface, which, in combination, might explain the higher infectivity of B.1.1.7. We anticipate our findings to be starting points for in depth biochemical and cell biological analyses of B.1.1.7, but also other highly contagious SARS-CoV-2 variants, as many of them likewise exhibit a combination of the D614G and N501Y mutation.

The outbreak of the severe acute respiratory syndrome (SARS) caused by the SARS-like coronavirus SARS-CoV-2 has emerged to a global pandemic with daily increasing numbers of infections and deaths exceeding 2.5 million world-wide 1 On top of the rapid, global spread of SARS-CoV-2, new and more contagious virus variants comprise an additional threat 2-5 . The novel SARS-CoV-2 variant B.1.1.7 first emerged in southeast England in November 2020 and is estimated to be 56% more transmissible 6 . This variant is characterized by a number of amino acid deletions and exchanges, with most of the protein-coding mutations found within the surface-anchored spike (S) protein of the virus: del69-70HV, del144Y, N501Y, A570D, D614G, P681H, T761I, S982A, D1118H 7 (sFig1 a). The homotrimeric S protein facilitates viral entry into host cells by interaction of its receptor-binding domain (RBD) with the cell surface receptor angiotensin-converting enzyme 2 (ACE2) [7] [8] [9] . The S protein consist of a S1 subunit, harboring the receptor-binding domain, and a S2 subunit, that is essential for viral entry into the host cell after dissociation from S1 (Fig1 a, sFig1b). Separation of S1 and S2 subunit is mediated by host cell surface proteases at the S1/S2 and S2' cleavage sites of the S protein 10, 11 . Previous results suggest that dramatic structural changes of the S2 unit lead to the exposure of a fusion peptide that facilitates cell membrane fusion and virus entry 12 . For S protein priming, human serin protease TMPRSS2 cleavage within a multi-basic (furin) cleavage site between AAs685/686 (S1/S2 cleavage site) or AAs815/816 (S2' cleavage site) has been proposed 9, 13 and a recent study even suggests blocking SARS-CoV-2 infection by utilizing a TMPRSS2 inhibitor 9 . Cellular uptake as well as protein structure of the S protein of the SARS-CoV-2 virus have been studied extensively. Hence, cryo-electron microscopy and crystallographic analyses of the S protein trimer 14, 15 , ACE-2-bound S protein 16 as well as ACE-2-bound receptor-binding domain 17 have shed light onto the structural mechanisms of viral entry. Until now this structural information is missing for globally emerging SARS-CoV-2 variants, exhibiting several mutations within the S protein and higher infectivity. We only have a vague understanding of the molecular reasons leading to enhanced contagion and cellular uptake of SARS-CoV-2 variants like the B.1.1.7 (British), but also the B.1.351 (South African) and P.1 (Brazilian/Japanese) variants. Interestingly, these three Variants of Concern (VOC) have the N501Y and D614G mutation in common 2, 18 In the present study, we focus on the B.1.1.7 variant and utilize molecular dynamics (MD) simulations of S protein trimer to assess its dynamic behavior in terms of conformational stability as well as interaction of isolated viral RBD in complex with human ACE2 to calculate linear interaction energies for wild type and mutated (N501Y) RBD-ACE2 complexes. Our data suggest increased flexibility around the fusion peptide induced by the D614G mutation and reduced binding affinity between RBD and ACE2 due to conformational reorganization of the RBD-ACE2 interface mediated by N501Y. Upon cleavage at the S2' cleavage site (Arg815/Ser816), a conformational switch alters partial salt bridge formation between arginine 815 and negatively charged aspartate and glutamate residues in the vicinity to keep the new C-terminus in place via non-covalent bindings.

To compare flexibility between the SARS-CoV-2 wild type (wt) and B.1.1.7 variant, root-meansquare fluctuation (RMSF) values were calculated for the backbone atoms of all individual residues over 200 ns simulation time, averaged for all six calculated monomers (two runs with three subunits per trimer) and visualized as spheres of different diameter and color according to their average RMSF values (Fig1 b, sFig2 a,b). Highest RMSF values were calculated for the RBD with flexibility of up to 10 Å especially at the interface positions where interaction with ACE2 would occur upon receptor binding (sFig2 a,b,c). In the N-terminal domain (NTD) deletions of amino acids 69, 70 and 144 induce a reduced flexibility for amino acid residues arginine 78, leucine 249 and threonine 250 in the B.1.1.7 variant (sFig2 d, Tab. 1). These residues are all located at the surface of the spike protein and are most exposed to interact with other spike trimers at the viral surface (sFig2 b,e). The loop region C-terminal of the S2' cleavage site and the fusion peptide showed increased flexibility in the B.1.1.7 variant when compared to wt (Fig1 c, sFig2 c). Flexibility increased from a maximum of 4 Å in wt to a maximum of about 7 Å in B.1.1.7. Residues 835-843 showed markedly increased RMSF values (Tab. 2). To understand this change in flexibility we analyzed the salt bridges formed by charged residues of this amino acid stretch.

We found lysine 835 and lysine 854 as positively charged amino acid residues in the region of this increased flexibility. Analyzing the wild type structure during MD simulation revealed interchain salt bridge formation between lysine 835 from one chain and aspartate 568 from the neighboring chain and interchain salt bridge formation between lysine 854 and aspartate 614 (Fig2 a). We also identified interchain salt bridges between arginine 646 and glutamate 868 and aspartate 867 (Fig2 a) although to a lesser extent than the other two described charged residue pairs. To analyze the consistency of these identified salt bridges we measured the interatomic distance over all three trimer subunits for both simulation runs. Therefore, we defined distances between the carbon atoms of the carboxyl groups of aspartate or glutamate and the nitrogen of the NH3 + group in the side chain of lysine or the carbon atom in ζ-position (most distal carbon in side chain) in arginine under 4 Å (preferred distance for salt bridge formation), between 4-5 Å (important for arginine residues due to the usage of the carbon atom in ζ-position) and above 5 Å (for no salt bridge formation). The interchain salt bridge, which can only be present in wild type, formed by aspartate 614 and lysine 854, is in the preferred distance of under 4 Å in about 60% of the simulation time. This indicates a certain importance for this salt bridge for local stability. For the salt bridge formed by lysine 835 from one subunit and aspartate 568 from an adjacent subunit we found distances smaller 4 Å in over 40% of the time for the wt S protein (Fig2 a). Additionally, there was no interaction found between lysine 854 and aspartate 586, arguing for a stable conformation of this fusion peptide adjacent loop region in wild type as it is supported by two salt bridges. Analyzing the same salt bridges in the B.1.1.7 variant reveals a different salt bridge pattern. As aspartate 614 is missing (D614G), lysine 854 is unmatched in this variant. We found that both lysine residues in this region, lysine 854 and lysine 835 form partial salt bridges with aspartate 568. However, even combined, both residues are only in the preferred distance of under 4 Å in about 20% of the time. To analyze the direct effect of the D614G mutation in this matter, we also calculated MD simulations of a wt variant with an inserted glycine for aspartate at position 614 (wt+D614G). Calculating the RMSF values also revealed an increase in flexibility in the region of lysine 854 (sFig3 a). Analyzing the distances between lysine residues 854 and 835 with aspartate 568 also revealed a partial interaction between lysine 854 and aspartate 568 as found in the B.1.1.7 variant (sFig3 b). However, the interaction between lysine 835 and aspartate 568 is as strong as in wt, indicating an additional destabilizing mechanism in the B. We also re-analyzed the stabilizing salt bridges between aspartate 614 and lysine 854 and aspartate 568 and lysine 835 in wt (Fig2 g) and found almost no change as compared to pre-cleavage conditions (Fig2 a). For B.1.1.7 we found, that the small interaction between aspartate 568 and lysine 835 is completely lost in the post-cleavage state and only the small interaction between aspartate 568 and lysine 854 remains in B.1.1.7 (Fig2 g). Of note, we also identified that interactions between arginine 646 and aspartate 867 and glutamate 868 are different in wt and B.1.1.7. While there is a more even distribution of contact formation between arginine 646 and the two negatively charged residues in wt, there is a clear preference towards glutamate 868 in B.1.1.7 (Fig2 g). Taken together, we identified how the D614 mutation (in combination with the A570D mutation) induces a loss of conformational stability that increases flexibility in a pre-and post-cleavage state. We also identified changes in salt bridge engagement of arginine 815 between a pre-cleavage (continuous protein chain) and post-cleavage state (discontinuous protein chain), that then stabilizes the new C-terminus via salt bridge rearrangement. To analyze whether this increase in flexibility in B.1.1.7 is also propagated towards the RBD over the entire S protein or is a local phenomenon, we analyzed the RMSD values over the simulation time for the RBDs (sFig3 ce). The RMSD values were low and very similar for the variants analyzed (wt, wt+D614G and B.1.1.7). Thus we hypothesized that changes in RBD behavior are a direct effect of the N501Y mutation and further analyzed it.

In order to analyze binding of the S protein to the ACE2 receptor we used the isolated receptor-binding domain (RBD) in complex with ACE2 (PDB ID code: 7KMB 16 ) as starting structure for molecular dynamics simulations. The only amino acid exchange localized into the RBD is the N501Y exchange that resides at the RBD-ACE2 interface (sFig1 a). Simulation for 500 ns revealed no marked differences in structural flexibility between wt and B.1.1.7, as it can be deduced from the RMSF values (sFig4 a,b). Compared to the MD simulation of the trimeric S protein without ACE2 (unbound state), the overall flexibility was very similar, but the bound RBD (RBD-ACE2 complex) showed markedly lower RMSF values in the region between RBD residue 475 and residue 520, where most of the RBD residues are located which closely interact with ACE2 (sFig4 c). To identify changes between wt and B.1.1.7 we analyzed contacts below 5 Å over time. For the B.1.1.7 variant we identified that glutamine 498 loses almost all contacts to ACE2 residues aspartate 38 and lysine 353, while the number of contacts is strongly reduced for tyrosine 41 (sFig5 a, Tab. 3). We also identified a markedly increased number of contacts for the newly inserted tyrosine at position 501 to lysine 353 in B.1.1.7 when compared to the wt asparagine residue (sFig5 a, individual runs shown in sFig6 and sFig7). Structural analysis revealed the expulsion of glutamine 498 from the RBD-ACE2 interface due to the bulkier tyrosine side chain in the B.1.1.7 variant (sFig5 b,c). We also noticed a change in side chain conformation for lysine 353 explaining the altered contact numbers (sFig5 b,c). We were now interested in individual binding energies and decomposed the RBD-ACE2 interface.

To decompose the electrostatic linear interaction energy all residues within 4 Å of ACE2 were analyzed. We identified the main electrostatic interaction in the central part of the RBD-ACE2 interface for wt and B.1.1.7. The electrostatic interaction is comprised by interaction of glutamine 493, lysine 417 and glutamine 498 with ACE2 in the wt RBD (Fig3 a,b) . Regarding the B.1.1.7 variant glutamine 493 and lysine 417 also contribute majorly to electrostatic interaction between RBD and ACE2. However, glutamine 498, which was displaced from the binding interface, does not contribute to electrostatic interaction as in wt (Fig3 a,b) as markedly reduced electrostatic interaction was calculated (Fig3 c, Tab. 4). Decomposition of electrostatic interaction on the individual residue level for ACE2 reveals that lysine 353 loses its interaction with the RBD in the B.1.1.7 variant (sFig8 a,b, Tab. 5). This fits to our data on contact formation as lysine 353 from ACE2 and glutamine 498 from the RBD lose all their contacts in the B.1.1.7 variant (sFig5 a). In total, we calculated an electrostatic interaction energy of -200 kcal/mol for wt and -165 kcal/mol in B.1.1.7. Computationally more demanding calculations of the binding energy using Molecular Mechanics Generalized Born Surface Area (MM/GBSA) also imply the previously observed reduced electrostatic binding affinity (data not shown). As tyrosine is a bulkier residue than asparagine at position 501, we were now interested in van der Waals contacts also at the level of the individual residues. 1.7 (Fig4 a,b) . A major difference was calculated for mutated residue 501 accounting for an asparagine in wt and tyrosine in B.1.1.7 (Fig4 c, Tab. 6). Here we found an increased van der Waals interaction energy in B.1.1.7. Van der Waals interaction energies did not differ in total and were calculated to be around -90 kcal/mol for both wt and B.1.1.7. The van der Waals linear interaction energy was lowered for tyrosine 41 and glutamine 42 from ACE2 (sFig9 a,b, Tab. 7).

We used molecular dynamics simulations on (i) S protein trimers and (ii) S protein RBD-ACE2 complexes to compare wt and B.1.1.7 variants. Understanding dynamic stability of the S protein is of vital importance to comprehend molecular mechanisms underlying viral entry into the host cell. In particular, this could help to assess superior infectivity of newly emerging viral variants. Although MD simulation data on wt S protein RBD-ACE2 interaction has been reported 19, 20 , data on dynamic changes induced by the N501Y variant found in B.1.1.7 (British), B.1.351 (South African) and P.1 (Brazilian, Japanese) is missing 7 . These three variants also harbor the D614G mutation 7 , which has previously been associated with higher infectivity 5, 21-23 and viral replication 21, 24 . For variants carrying this D614G mutation it was also suggested that more functional S protein is incorporated into the virion membrane 5 . However, molecular mechanisms explaining the higher infectivity of the D614G mutation are still elusive. Our data derived from simulations of the entire S trimer and comparing wt with B.1.1.7, reveal a rearrangement of salt bridges caused by the loss of interaction between position 614 due to the D614G mutation and lysine 854 in B.1.1.7. This induced flexibility at residues C-terminal of the fusion peptide is enhanced in post-cleavage conformations as also salt bridges formed by aspartate 568 are majorly lost in B.1.1.7. Release of the fusion peptide might allow a faster insertion of the latter into the host cell membrane due to the increased flexibility of the adjacent linker. This might also happen after cleavage at the S2' position by TMPRSS2 9, 25 or allow faster entry into host cells after priming the S2' position by furin during protein secretion 9 , especially as enhanced proteolytic cleavage was proposed for D614G S proteins 26 . Simulation also reveals structural stability of the post-cleavage state induced by salt bridge formation involving arginine 815 that would allow such priming during protein secretion without destabilizing the overall structural integrity. The newly formed salt bridge of aspartate 570 with lysine 964 does not compensate for the loss of aspartate 614 in terms of protein stability, but might rather weaken salt bridges formed by aspartate 568. Additionally, as aspartate 614 comes from the S1 region and lysine 854 from the S2 region of the S protein, dissociation of the S1 and S2 domain of the S protein, important for membrane fusion after entry of the fusion peptide into the host cell membrane, might be enhanced 27 . Furthermore, we show that deletions in the NTD reduce flexibility in the outer region of the S trimer, potentially allowing more spike proteins at the virion host cell interface of B.1.1.7 and thus increasing the probability of viral uptake 5 . Binding to ACE2 via the receptor-binding domain of the viral S protein is the first step of cellular uptake and infection. Previous reports already highlighted the importance of residues lysine 417, tyrosine 449, glutamine 493 and glutamine 498 on the RBD of the S protein for interface formation with ACE2 7, 20 . At ACE2, residues aspartate 30, lysine 31 and lysine 353 were also identified as important interactors with the RBD of the S protein 7, 20, 28 . We could confirm the importance of the above-mentioned residues for interface formation between RBD and ACE2 here. In addition, our data suggests that the insertion of the bulkier tyrosine at position 501, as found in B.1.1.7, B.1.351 and P.1 excludes glutamine 498 from the RBD-ACE2 interface and alters the molecular architecture of the RBD-ACE2 interface of the S protein. We measured a reduced electrostatic affinity of glutamine 498 from the S protein to residues on ACE2. For tyrosine 501 from the S protein RBD we found an increased van der Waals interaction with lysine 353 (ACE2) in the N501Y variant. Similar RMSD values of the RBD for wt and B.1.1.7 suggest that amino acid deletions/exchanges in the NTD and fusion peptide adjacent loop region cause local changes and do not change the overall flexibility of the RBD. In conclusion, our data suggest a higher mobility of the fusion peptide after release and reduced electrostatic affinity of the S protein RBD to ACE2 in B.1.1.7. This combination could allow faster membrane fusion of the virion with the host cell membrane and could be explained by (i) faster insertion of the fusion peptide and (ii) faster dissociation of the S protein RBD-ACE2 complex due to reduced electrostatic affinity. Dissociation of the S protein components is important for efficient membrane fusion after insertion of the fusion peptide. Our data might serve as a basis for advanced biochemical and cell biological experiments to better understand the importance of individual amino acid exchanges for viral pathology.

To generate starting structures for wild type, wild type with the D614G mutation (wt+D614G) and B.1.1.7 SARS-CoV-2 we used the protein sequence annotated in UniProt (www.uniprot.org) with the identifier P0DTC2 (SARS-CoV-2; wild-type) and changed the sequence in accordance with the reported deletions (del69-70, del144) for the SARS-CoV-2 B.1.1.7 variant. Using Swiss-Model Expasy (www.swissmodel.expasy.org) we generated a model for the SARS-CoV-2 wild type, wild type with the D614G mutation and SARS-CoV-2 B.1.1.7 variant on the basis of the Cryo-EM solved structure of the triple ACE2 bound SARS-CoV-2 spike protein trimer (PDB ID: 7KMS) 16 and used chain A as a template. After modeling, structures were transferred into UCSF Chimera 29 and single amino acid exchanges were inserted into the B.1.1.7 structure using the swapaa command. These were N501Y, A570D, D614G, P681H, T716I, S982A and D1118H. Post-translational modifications (here sugars) were removed. Monomeric spike proteins of wt, wt+D614G and B.1.1.7 were then C3 symmetrized to generate trimeric spike protein complexes and used as starting structures for molecular dynamics (MD) simulations. Cleavage of wild type and the B.1.1.7 variant was structurally implemented in the PDB files used as starting structures by adding TER records (indicates end of a chain) between arginine 815 and serine 816 in all three chains of the trimer.

To investigate the interface between the RBD of the spike protein and ACE2, the respective wild type starting structure was obtained from the PDB database (PDB ID code: 7KMB) 16 . In order to generate also the starting structure for the MD simulations of the B.1.1.7 variant the N501Y amino acid exchange was introduced with Swiss-PdbViewer 4.1.0.

Molecular dynamics simulations were performed using version 20 of the Amber Molecular Dynamics software package (ambermd.org) 30 and the ff14SB force field 31 . With the Amber Tool LEaP, all systems were electrically neutralized with Na + ions and solvated with TIP3P 32 water molecules. The trimeric SARS CoV-2 spike protein was solvated in a cuboid water box with at least 15 Å distance from the borders to the solute, whereas the receptor-binding domain (RBD) complexed with ACE2 was solvated in a water box with the shape of a truncated octahedron and at least 25 Å distance from the borders to the solute.

The simulations followed a previously applied protocol 33 . At first, a minimization was carried out in three subsequent steps to optimize the geometry of the starting structures. In the first step of the minimization, the water molecules were minimized, while all remaining atoms were restrained with a constant force of 10 kcal·mol -1 ·Å -2 to the initial positions. In the second step, additional relaxation of the sodium ions and the hydrogen atoms of the protein were allowed, while the remaining protein was restrained with 10 kcal·mol -1 ·Å -2 . In the last step, no restraints were used, so that the whole protein, the ions, and the water molecules were minimized. All three minimization parts started with 2500 steps using the steepest descent algorithm, followed by 2500 steps of a conjugate gradient minimization. After the minimization, the systems were equilibrated in two successive steps. In the first step, the temperature was raised from 10 to 310 K within 0.1 ns and the protein was restrained with a constant force of 5 kcal·mol -1 ·Å -2 . In the second step (0.4 ns length), only the Cα atoms of the protein were restrained with a constant force of 5 kcal·mol -1 ·Å -2 . In both equilibration steps, the time step was 2 fs. Minimization and equilibration were carried out on CPUs, whereas the subsequent production runs were performed using pmemd.CUDA on Nvidia A100 GPUs [34] [35] [36] . In the following, 200 ns (trimeric spike protein) or 500 ns (RBD complexed with ACE2) long production runs were conducted without any restraints and at 310 K (regulated by a Berendsen thermostat 37 ). Furthermore, the constant pressure periodic boundary conditions were used with an average pressure of 1 bar and isotropic position scaling. For bonds involving hydrogen, the SHAKE algorithm 38 was applied in the equilibration and production phase. In order to accelerate the production phase of the molecular dynamics (MD) simulations, hydrogen mass repartitioning (HMR) 39 was used in combination with a time step of 4 fs. The MD simulations were performed two times for all forms of the trimeric spike protein and four times for both forms of the RBD in complex with ACE2. Trajectory analysis (analysis of root-mean-square deviation of atomic positions (RMSD), rootmean-square fluctuations (RMSF), analysis of native contacts, measurement of interatomic distances, calculation of linear interaction energy (electrostatic and van der Waals interactions, used default cutoff of 12.0 Å) was carried out using the Amber tool cpptraj 40 . Contacts were evaluated with an in-house Perl script parsing the trajectory using the prior named Amber tools and assigning contacts based on a distance criterion of ≤5 Å between any pair of atoms, as it was done previously 41 .

Statistical analyses were generated with GraphPad Prism (version 8.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com) and statistical tests were applied as indicated below the figure. Plots were generated in GraphPad and Gnuplot (version 4.6). All structure images were made with UCSF Chimera 1.15 29 . Figure 1: Structural flexibility in the B.1.1.7 SARS-CoV-2 S protein. a) Schematic primary structure of the SARS-CoV-2 S protein indicating the site of amino acid deletions (red arrowheads) or single amino acid exchanges (blue arrowheads; NTD, N-terminal domain; RBD, receptor-binding domain; S1/S2, furin cleavage site at positions 685/686; S2', furin/TMPRSS2 cleavage site at positions 815/816; FP, fusion peptide; TM, transmembrane domain and Cterminal end). b) S protein trimer as it would reside on the cell surface with one subunit colored for structural flexibility as calculated during simulation (root-mean-square fluctuations (RMSF), n = 6). NTD denotes the N-terminal domain, RBD the receptor-binding domain and the grey arrowhead the loop region between amino acids 835-843. c) Line plot of RMSF values for amino acid residues 800-900 reveals increased flexibility for residues 835 and 843 in B.1.1.7 (orange) when compared to wt (blue). The arrowhead denotes the same region as in b) and the asterisk indicates statistical differences for these amino acids (n = 6; two-way ANOVA; statistical significance assumed for *p<0.05; full statistic can be found in Tab. 2). RBD residues with atoms within a maximum distance of 4 Å from ACE2 are shown as spheres with radii and colors according to their van der Waals linear interaction energy to ACE2. b) View on the interacting interface of the RBD with ACE2. Residue color and atom size represent the van der Waals linear interaction energy to ACE2. c) Quantification of van der Waals linear interaction energy for all RBD residues within 4 Å distance of ACE2 (n = 4; two-way ANOVA; statistical significance assumed for *p<0.05; full list of results in Tab. 6). sFig1: a) Structural representation of the SARS-CoV-2 S protein on the viral surface with amino acid residues shown as spheres that are deleted (red) or mutated (blue) in the B.1.1.7 variant. b) Structural representation of the SARS-CoV-2 S protein on the viral surface with domains highlighted in color as in figure 1a (N-terminal domain, NTD in pink; receptor-binding domain, RBD in green; S1 region in khaki; S1/S2 and S2' cleavage sites in yellow with amino acids as spheres; fusion peptide in cyan; S2 domain in red). sFig3: a) Line plot of root-mean-square fluctuation (RMSF) values for amino acid residues 800 to 900. Averages were calculated for all three trimer subunits and both simulation runs of the wild type (wt, blue) and wt+D614G variant (green). b) Percentage of salt bridge formation over time for four different residue pairs. All residue pairs represent interchain interactions with residues from two different, but directly neighboring chains within the trimeric spike protein. c-e) Conformational stability was measured as root-mean-square deviation (RMSD) values over simulation time for the receptor-binding domain (RBD) of c) the wild type d) the B.1.1.7 variant and e) the wild type with the D614G mutation. One representative plot is shown for wild type spike protein and its variants. sFig6: Individual intermolecular contact plots for all four molecular dynamics (MD) simulation runs of the wild type RBD-ACE2 complex. The number of contacts, calculated as the number of interresidue atom pairs that are within a maximum distance of 5 Å from each other, was plotted color-coded for intermolecular residue pairs over the simulation time. For instance, the residue pair Lys353-Asn501, with Lys353 expressed on ACE2 and Asn501 expressed on the RBD, has in all four MD simulations a calculated number of contacts between 40 and 60 over the whole simulation time. Residue pairs with asparagine 501 are highlighted in blue. sFig7: Individual intermolecular contact plots for all four molecular dynamics (MD) simulation runs of the B.1.1.7 RBD-ACE2 complex. The number of contacts, calculated as the number of interresidue atom pairs that are within a maximum distance of 5 Å from each other, was plotted color-coded for intermolecular residue pairs over the simulation time. For instance, the residue pair Lys353-Tyr501, with Lys353 expressed on ACE2 and Tyr501 expressed on the mutated RBD, has in all four MD simulations a calculated number of contacts around 80 over the whole simulation time. Residue pairs with mutated tyrosine 501 are highlighted in orange. sFig8: a) Electrostatic affinity shown on the RBD-ACE2 interface for ACE2 residues. Residues in closer contact than 4 Å are shown as spheres of different size and color according to their electrostatic affinity with the RBD. b) Quantification of the electrostatic linear interaction energy for all residues within a radius of 4 Å around the receptor-binding domain (n =4; twoway ANOVA; statistical significance was assumed for *p<0.05; full list of results in Tab. 5). sFig9: a) Van der Waals linear interaction energy shown on the RBD-ACE2 interface for ACE2 residues. Residues in closer contact than 4 Å are shown as spheres of different size and color according to their van der Waals linear interaction energy with the RBD. b) Quantification of the van der Waals linear interaction energy for all residues within a radius of 4 Å around the receptor-binding domain (n =4; two-way ANOVA; statistical significance was assumed for *p<0.05; full list of results in Tab. 7).

A. contributed to the design of the study as well as discussion of the data

A. wrote the initial draft; all authors reviewed the manuscript prior to submission

Emergence of SARS-CoV-2 B.1.1.7 Lineage -United States

Early transmissibility assessment of the N501Y mutant strains of SARS-CoV-2 in the United Kingdom

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

SARS-CoV-2 spike-protein D614G mutation increases virion spike density and infectivity

Estimated transmissibility and severity of novel SARS-CoV-2 Variant of Concern 202012/01 in England

Structure of SARS coronavirus spike receptor-binding domain complexed with receptor

A pneumonia outbreak associated with a new coronavirus of probable bat origin

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

Cell entry mechanisms of SARS-CoV-2

Activation of the SARS coronavirus spike protein via sequential proteolytic cleavage at two distinct sites

Identification of the Fusion Peptide-Containing Region in Betacoronavirus Spike Glycoproteins

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

Cryo-EM Structure of the 2019-nCoV Spike in the Prefusion Conformation

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

Cryo-EM Structures of SARS-CoV-2 Spike without and with ACE2 Reveal a pH-Dependent Switch to Mediate Endosomal Positioning of Receptor-Binding Domains

Structure of the SARS-CoV-2 spike receptor-binding domain bound to the ACE2 receptor

Dynamics of the ACE2-SARS-CoV-2/SARS-CoV spike protein interface reveal unique mechanisms

SARS-CoV-2 D614G Variant Exhibits Enhanced Replication ex vivo and Earlier Transmission in vivo

Empirical transmission advantage of the D614G mutant strain of SARS-CoV-2. medRxiv 2020.09.22

Evaluating the Effects of SARS-CoV-2 Spike Mutation D614G on Transmissibility and Pathogenicity

Efficient activation of the severe acute respiratory syndrome coronavirus spike protein by the transmembrane protease TMPRSS2

D614G Mutation Alters SARS-CoV-2 Spike Conformation and Enhances Protease Cleavage at the S1/S2 Junction

Receptor binding and priming of the spike protein of SARS-CoV-2 for membrane fusion

Deep Mutational Scanning of SARS-CoV-2 Receptor Binding Domain Reveals Constraints on Folding and ACE2 Binding

UCSF Chimera--a visualization system for exploratory research and analysis

Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB

Comparison of Simple Potential Functions for Simulating Liquid Water

The conformational stability of nonfibrillar amyloid-beta peptide oligomers critically depends on the C-terminal peptide length

Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 2. Explicit Solvent Particle Mesh Ewald

Routine Microsecond Molecular Dynamics Simulations with AMBER on GPUs. 1. Generalized Born

SPFP: Speed without compromise-A mixed precision model for GPU accelerated molecular dynamics simulations

Molecular-Dynamics with Coupling to an External Bath

Numerical-Integration of Cartesian Equations of Motion of a System with Constraints -Molecular-Dynamics of N-Alkanes

Long-Time-Step Molecular Dynamics through Hydrogen Mass Repartitioning

Software for Processing and Analysis of Molecular Dynamics Trajectory Data

5374. sFig4: a) Structural flexibility of the receptor-binding domain-ACE2 (RBD-ACE2) complex. Residues are colored and shown as spheres of different size according to their individual RMSF values. RBD denotes the receptor-binding domain and ACE2 the angiotensin-converting enzyme 2. b) Line plot of RMSF values averaged over four molecular dynamics (MD) simulation runs for the RBD-ACE2 complex with wild type (wt)

Line plot of RMSF values from the RBD in the unbound state (from the S protein trimer MD simulations) and in the bound state to ACE2 (from the RBD-ACE2 complex MD simulations)

Wild type (wt) is shown in blue and B.1.1.7 in orange (n = 4, two-way ANOVA; statistical significance was assumed for *p<0.05; full list of results in Tab. 3). b) Structure of the RBD-ACE2 interface with residues asparagine 501 (Asn501; wild type in blue) or tyrosine 501 (Tyr501orange), glutamine 498 (Gln498) (all from RBD) and lysine 353 (green) from ACE2 shown in ball-andstick

The authors gratefully acknowledge the compute resources and support provided by the Erlangen Regional Computing Center (RRZE) and by NHR@FAU.