key: cord-1031694-cksr4x39
authors: Baral, Prabin; Bhattarai, Nisha; Hossen, Md Lokman; Stebliankin, Vitalii; Gerstman, Bernard S.; Narasimhan, Giri; Chapagain, Prem P.
title: Mutation-induced changes in the receptor-binding interface of the SARS-CoV-2 Delta variant B.1.617.2 and implications for immune evasion
date: 2021-08-15
journal: Biochem Biophys Res Commun
DOI: 10.1016/j.bbrc.2021.08.036
sha: 191c006e0c49d2ccc24ab8b4a9b570ddc79d1081
doc_id: 1031694
cord_uid: cksr4x39

Following the initial surges of the Alpha (B.1.1.7) and the Beta (B.1.351) variants, a more infectious Delta variant (B.1.617.2) is now surging, further deepening the health crises caused by the pandemic. The sharp rise in cases attributed to the Delta variant has made it especially disturbing and is a variant of concern. Fortunately, current vaccines offer protection against known variants of concern, including the Delta variant. However, the Delta variant has exhibited some ability to dodge the immune system as it is found that neutralizing antibodies from prior infections or vaccines are less receptive to binding with the Delta spike protein. Here, we investigated the structural changes caused by the mutations in the Delta variant's receptor-binding interface and explored the effects on binding with the ACE2 receptor as well as with neutralizing antibodies. We find that the receptor-binding β-loop-β motif adopts an altered but stable conformation causing separation in some of the antibody binding epitopes. Our study shows reduced binding of neutralizing antibodies and provides a possible mechanism for the immune evasion exhibited by the Delta variant.

While the vaccination efforts against SARS-CoV-2 infections are ongoing worldwide, new genetic variants of the virus are emerging and spreading. Notable variants of concern include the Alpha variant B. is fast becoming the most dominant variant in many countries 3 . Fortunately, the current vaccines appear to be effective against many variants of concern, including the Delta variant 4, 5 . However, lack of vaccination coverage worldwide has allowed the virus to spread and continue to evolve, decreasing the chances of quickly ending the pandemic. The significant and rapid rise in the number of cases due to infections by the Delta variant in areas that appeared to overcome the earlier surges (e.g. India) or even in areas with high coverage of effective vaccines (e.g. Israel) indicates that the antibodies elicited from vaccines or prior infections by other strains of SARS-CoV-2, or other pathogens through molecular mimicry are not as effective for the Delta variant. Indeed, recent studies 7, 8 have shown that compared to the previous variants, the Delta variant is not only able to evade immunity conferred by previous infections but is also less sensitive to neutralizing Abs from recovered patients. Some anti-RBD Abs have been shown to have reduced RBD binding, resulting in a 4-fold decrease in the potency of the sera from prior infections and a 3-5 fold decrease in vaccine-generated Abs against the Delta variant 8 . Similarly, it has been shown recently that the mutations in the B.1.427/B.1.429 variants cause reduced or complete loss of sensitivity to RBD-binding antibodies (Abs) 9 . This suggests that, in addition to possibly a high affinity of the RBD to ACE2 or its ability to present itself in an up conformation in the spike trimer 10, 11 , antibody evasion due to RBD mutations may also be J o u r n a l P r e -p r o o f contributing to the increased transmissibility of the Delta variant. Therefore, it is important to understand both the RBD-ACE2 binding as well as the RBD-Ab binding.

In this study, we investigate the effects of the mutations in the Delta variant on the structure of the receptor-binding interface of the RBD as well as the RBD-ACE2 interactions and RBDneutralizing Abs interactions. We examine the SARS-CoV-2 Ab-RBD complexes available in the protein data bank (PDB) and compare the differences in the RBD-Ab interactions due to the mutations in the Delta variant. Our results suggest that the Delta variant features a stable but slightly reorganized receptor-binding interface that can lead to weakened interactions with some neutralizing Abs resulting in immune evasion.

The Delta variant RBD was prepared by introducing the L452R/T478K mutations to the WT structure taken from the CHARMM-GUI 12, 13 COVID19 repository (spike protein-ACE2 complex, PDB ID 6VSB, 6VW1) [14] [15] [16] . The RBD-only system for the Delta variant was set up and simulated the same way as in our earlier work in Bhattarai et al. 17 The RBD-ACE2 complex for the Delta variant was modeled by superimposing the Delta RBD (the last frame of the 600 ns simulation) onto the WT RBD-ACE2 complex, with the receptor-binding interface of the RBD selected for the structural alignment. The structures of the RBD-Ab complexes were retrieved from the RCSB Protein Data Bank (PDB) 15 . Four representative RBD-Ab systems were prepared for simulation.

The same MD procedure is used as in our previous work 17 . Briefly, all-atom molecular dynamics (MD) simulations were performed with NAMD 2.14 18 using the Charmm36m force field 19, 20 . The structures were equilibrated for 2 ns with a timestep of 2 fs after a short minimization. The production runs were performed under constant pressure of 1 atm and constant temperature (303 K). The RBD-only systems were run for 600 ns, whereas the RBD-ACE2 and RBD-Ab systems were run for 100 ns. The simulations and systems in this work are summarized in Table S1 . Visualization and analysis of the trajectories were done with Visual Molecular Dynamics (VMD) 21 .

J o u r n a l P r e -p r o o f

To evaluate the Ab-binding to the RBD, we used a pre-trained MaSIF-search geometric deep learning model 22 . The model evaluates the protein-protein binding potential, given two surface regions (patches) from distinct proteins. The deep learning model converts patches into 80-dimensional feature space, such that distance between embedded vectors from native binders is minimized. This distance is referred to as the binding cost, i.e., a lower output score represents a better interaction. The binding cost of the Ab-RBD complex was computed as the average of the output scores of three best patch pairs.

The RBD of the spike protein has mutations N501Y in B. 

To investigate the RBD dynamics and the structural changes due to mutations, we performed MD simulations of the RBD of the Delta variant B.1.617.2 and compared the results with the WT, B.1.1.7, and B.1.351 variants. Both of the mutations, L452R and T478K, in the Delta RBD are in the receptor-binding interface comprised of a motif spanning residues 438 to 508. The same interface is a target for many neutralizing antibodies. Therefore, any changes in the receptorbinding interface can affect both the receptor binding to the host ACE2 as well as Ab-binding. To assess the structural changes in this interface, we analyzed different regions of the interface given in Fig. S1 . Figure S2 shows that the amino acid residues in the -loop- motif (Region 3, residues 472-490) have the largest flexibility for all variants. Interestingly, the analysis of Fig. 1c shows that a slight reorientation of residue G496 in the Delta variant results in much stronger hydrogen bonding between the β-strands. Most notably for the Delta variant, a significantly enhanced salt-bridge interaction between the R454 side chain and D467 side chain is observed. This change is possibly due to the mutation L452R which gives a slightly enhanced -structural propensity 24, 25 in 5. It has recently been shown that the L452R mutation in another variant of concern, B.1.427/B.1.429, caused reduction in nearly half of the tested monoclonal Abs. 9 , highlighting the dangerous consequences of this mutation. We analyzed 300 ns re-runs for each of these variants and compared in Fig. S3 . Although the R454-D467 backbone hydrogen bond has not switched to a side chain interaction by 300 ns in the re-run of the 

A recent study illustrated the mechanism of immune evasion by a variant of concern B.1.427/B.1.429 9 . Specifically, the same mutation found in the Delta variant, L452R, was responsible for reduced neutralizing activities in many of the monoclonal Abs tested, whereas regrouping of a disulfide bond in a different RBD site caused the loss of activities for all Abs tested.

To assess the impact on the Ab binding due to the changes in the receptor-binding interface caused by the mutations in the Delta RBD, we first examined the interfacial interactions in the Ab-RBD complexes available in the protein data bank (PDB) in the WT. Of the 118 RBD-Ab complexes with Ab bound in the receptor-binding interface retrieved from the Protein Data Bank, 47 nonrepeating complexes were considered for further analysis. The Ab-RBD complexes (pdb IDs) are listed in Table S2 and visualized in Movie S2. We identified the RBD residues involved in ionic or hydrogen bond interactions in each complex and plotted in Fig. 3 the frequency of occurrences of the important residues in all complexes. While this distribution may be inherently biased due to the available pdb structures of the complexes, it provides a general idea of the preferred interfacial RBD binding epitope sites for a sample of Abs. From Fig. 3 , we see that the majority of the Abs have interactions with the -loop- residues. 

Inspection of the Ab-RBD complexes for the WT shows that many of the Abs anchor at multiple sites. For example, in many Ab-RBD complexes, including 6xe1, 7b3o, 7cdi, and 7cjf

Abs bind at A475/G485 at one site (site A in Fig. 3 ) and R457/K458 (site F in Fig. 3 ) at another site, as shown in the figure. We grouped different sites and color-coded as shown in Fig. 3 . To examine how the changes in these sites may affect the Ab binding, we plotted the C distance between the residues K458 and A475 belonging to two Ab-binding sites in Fig. 4 for both the WT and the Delta variant. As shown in Fig. 4b , the distance between these residues (458-475) mostly remains at ~9 Å for the WT. However, the same (458-475) distance for the Delta variant in Fig.   4c increases to ~14 Å by 150 ns and remains stable at that distance. This 4-5 Å increase in the Abbinding sites suggests that the Ab-binding will be severely affected, and the Ab becomes insensitive to Delta RBD binding. While ACE2 binds at the site of 475/487 (site A in Fig. 3) , it does not bind at the site of 457/458 (site F) and therefore the increase in the K458-A475 distance does not affect the ACE2 binding. Instead, ACE2 binds at 475/487 and Q493 (in the middle of 6). Therefore, we also plotted the distance between the residues N487 and Q493. Interestingly, despite the structural changes, this distance in both the WT and the Delta variant remains nearly the same (16-17 Å) as seen in Fig 4b and 4c . This suggests that the structural changes have not affected the ACE2 binding sites but significantly affected the Ab-binding sites, suggesting a possible immune evasion by the Ab while maintaining the ability of receptor binding. 

With the observation of the increase in the distance between the two sites in the Delta RBM, we investigated how the changes affect ACE2 and Ab binding. If the ACE2 binding is maintained or enhanced but the Ab binding is weakened, at least for a set of neutralizing Abs, that would mean that the virus is less sensitive to the Abs thereby making it more effective at infecting and spreading. To explore this, we performed simulations of the RBD-ACE2 complex and Ab- (Table S3) . With some of the major interactions, including the K417-D30 salt-bridge, still present in the complex, ACE2 binding seems tolerate the structural changes in the Delta RBD.

We next compared the Ab binding in the WT and the Delta RBD. We considered two examples of the Ab-Delta RBD complexes modeled from the WT RBD complexed with the neutralizing Ab CV30 Fab (pdb ID 6xe1) 27 and complexed with the Ab BD-236 Fab (pdb ID 7chb) 28 . The 100 ns simulation of the CV30-Delta RBD model shows a less stable complex with significantly reduced interactions. As shown in Fig. S4 , Ab in the WT has interactions with residues in three clusters that are intact during the simulation. However, in the Delta RBD, the Ab is only able to bind at site A or F but not both. This is consistent with the argument made on the basis of Fig. 4 . The major hydrogen bonds, including those with Y473, Y421, L455, A475, R457, R403, K417, and Y505 in WT are broken or weakened in the Delta variant. The Ab-RBD complex modeled from 7chb (complex with BD-236 Fab) also shows reduction in the number and strength of the hydrogen bonding ( Fig. S4 and Table S4 ). Binding analyses using MaSIF-search shows that both Ab-RBD complexes in the Delta variant have significantly reduced binding (Fig. S5) . The violin plot in Fig. S5 shows the distribution of the binding costs extracted from the Ab-Protein complexes in the SAbDab structural antibody database 29 . The scores for the WT complexes lie within the distribution but the complexes for the Delta variant show significantly higher binding costs. While Abs that bind at other sites may not be affected, Abs that bind at A-F can become insensitive, albeit differentially, to the Delta RBD binding.

The Delta variant (B.1.617.2) of the SARS-CoV-2 has become one of the most worrisome variants so far during the pandemic and is rapidly spreading worldwide, making it responsible for the recent surges in infections and deaths. While current vaccines are still shown to be protective against this variant, it is also becoming clearer that it can escape the immune system by making neutralizing Abs from prior infections or elicited by vaccines less sensitive to binding with the spike protein. In this work, we performed molecular dynamics simulations of the Delta variant RBD with the mutations L452R/T478K and investigated the resulting structural changes in the receptor-and Ab-binding interfaces. We find that the Delta variant presents a noticeably different J o u r n a l P r e -p r o o f receptor-binding interface compared to the WT, B.1.1.7, and B.1.351. Specifically, the receptorbinding -loop- motif adopts an altered conformation which appears to cause shifts in the Abbinding epitope regions that can reduce the binding affinities for some neutralizing Abs. We investigated this by performing simulations of two Ab-RBD complexes and found that one of the complexes shows significantly reduced interactions between the Ab and the RBD, suggesting a possible mechanism of the immune escape by the Delta variant. Even though the Ab-resistant conformations obtained in these simulations may represent only a subset of the conformational ensemble, they can still contribute considerably to the reduced sensitivity of the Abs. Future work with a full mapping of the conformational space of the receptor-binding interface may shed further light on the nature of the interactions with the common anti-RBD Abs, providing useful information on vaccine efficacies. Understanding how the structural changes alter the RBD's ability to present itself in the up conformation in the spike trimer or its ACE2 binding affinity can also inform us on the variant's transmissibility.

Additional figures showing the hydrogen-bonding residues in the complexes and tables listing the Ab-RBD complexes and the interacting residues. Movie S1: showing the 600 ns trajectory of the Delta variant RBD. Movie S2: showing a sample of Ab-RBD complexes from the protein data bank.

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

Delta to dominate world

Covid-19: Single vaccine dose is 33% effective against variant from India, data show

Prevention, SARS-CoV-2 variant classifications and definitions

SARS-CoV-2 B. 1.617. 2 Delta variant emergence and vaccine breakthrough

Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization

SARS-CoV-2 immune evasion by the B. 1.427/B. 1.429 variant of concern

Structural basis for enhanced infectivity and immune evasion of SARS-CoV-2 variants

Evolutionary and structural analysis elucidates mutations on SARS-CoV2 spike protein with altered human ACE2 binding affinity. Biochemical and biophysical research communications 2021

Spike Protein Model in a Viral Membrane

CHARMM-GUI: a web-based graphical user interface for CHARMM

Structural basis of receptor recognition by SARS-CoV-2

The protein data bank

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

Structural and Dynamical Differences in the Spike Protein RBD in the SARS-CoV-2 Variants B. 1.1. 7 and B. 1.351

Scalable molecular dynamics with NAMD

CHARMM36 all-atom additive protein force field: Validation based on comparison to NMR data

Force field development and simulations of intrinsically disordered proteins

VMD: visual molecular dynamics

Deciphering interaction fingerprints from protein molecular surfaces using geometric deep learning

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

Structural propensities and entropy effects in peptide helix-coil transitions

The intrinsic conformational propensities of the 20 naturally occurring amino acids and reflection of these propensities in proteins

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

Structural basis for potent neutralization of SARS-CoV-2 and role of antibody affinity maturation

Structurally resolved SARS-CoV-2 antibody shows high efficacy in severely infected hamsters and provides a potent cocktail pairing strategy

SAbDab: the structural antibody database

This work was supported by the National Science Foundation under Grant No. 2037374. NB acknowledges the Dissertation Year Fellowship support from the University Graduate School at Florida International University. The authors thank the COVID-Informatics research team at Florida International University for helpful discussions.