Comprehensive deep mutational scanning reveals the pH induced stability and binding differences between SARS-CoV-2 spike RBD and human ACE2

Abstract The SARS-CoV-2 spike (S) glycoprotein with its mobile receptor-binding domain (RBD), binds to the human ACE2 receptor and thus facilitates virus entry through low-pH-endosomal pathways. The high degree of SARS-CoV-2 mutability has raised concern among scientists and medical professionals because it created doubt about the effectiveness of drugs and vaccinations designed specifically for COVID-19. In this study, we used computational saturation mutagenesis approach, including structure-based free energy calculations to analyse the effects of the missense mutations on the SARS-CoV-2 S-RBD stability and the S-RBD binding affinity with ACE2 at three different pH (pH 4.5, pH 6.5, and pH 7.4). A total of 3705 mutations in the S-RBD protein were analyzed, and we discovered that most of these mutations destabilize the RBD protein. Specifically, residues G404, G431, G447, A475, and G526 were important for RBD protein stability. In addition, RBD residues Y449, Y489, Y495, Q498, and N487 were critical for the RBD-ACE2 interaction. Next, we found that the distribution of the mean stability changes and mean binding energy changes of RBD due to mutations at both serological and endosomal pH correlated well, indicating the similar effects of mutations. Overall, this computational analysis is useful for understanding the effects of missense mutations in SARS-CoV-2 pathogenesis at different pH. Communicated by Ramaswamy H. Sarma


Introduction
Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is a non-segmented, positive-sense single-stranded RNA virus of �30 kb genome size.The spike (S) glycoprotein is essential for its pathogenicity.This homotrimeric S protein facilitates the release and entry of the viral genome into host cells and mediates the attachment and binding of the virus to host cell receptors, ACE2 (Hoffmann et al., 2020;Lan et al., 2020;Walls et al., 2020;Q. Wang et al., 2020).The S protein is composed of two subunits: Subunit 1 (S1), which contains the ACE2 receptor-binding domain (RBD), and Subunit 2 (S2), which contributes to the fusion process (Hoffmann et al., 2020;Shang et al., 2020).The S protein from SARS-CoV-2 has a greater affinity for ACE2 compared with those of SARS-CoV (Walls et al., 2020) and bat coronavirus S (Tai et al., 2020).The binding interface between RBD and ACE2 and the atomic features of the SARS-Cov-2 S proteins were both uncovered by structural analysis.The two common prefusion conformations for uncleaved and furin-cleaved SARS-CoV-2 spikes are: a single-up conformation and an all-down conformation of RBD in the S1 subunit (Walls et al., 2020;Wrapp et al., 2020;Wrobel et al., 2020).The 'up' orientation of RBD is linked to the epitope accessibility of RBD-directed antibodies and is required for interaction with the ACE2 receptor.Also, a postfusion structure of the SARS-CoV-2 spike further revealed even more structural differences between prefusion and post-fusion conformations (Cai et al., 2020).
With respect to the reference Wuhan genome, SARS-CoV-2 has experienced more than 10,000 mutations (Wang et al., 2020).The majority of the mutations associated with the different SARS-CoV-2 variants are known to have increased affinity for the ACE2 receptor and are crucial for the virus spread (Chan et al., 2020;Starr et al., 2020).These variants are also demonstrated to increase transmissibility (Davies et al., 2021), alter infectivity (Li et al., 2020;L. Zhang et al., 2020), ineffective towards neutralizing antibodies and escape from neutralization (Dejnirattisai et al., 2021;Mannar et al., 2022;Wibmer et al., 2021;Xie et al., 2021;Xu et al., 2022;J. Zhang et al., 2021).The underlying reason for the increased binding affinity in these variants has not been linked to a sizable alteration in either the spike or ACE2 receptor's three-dimensional structures.Therefore, it is important to study possible RBD mutations and their potential effects on COVID-19 development and vaccination interactions.Accordingly, RBD mutations that can improve binding affinity with ACE2 and/or affect antibody neutralization have been extensively mapped through high-throughput mutational studies (C.Chen et al., 2021;J. Chen et al., 2020;Greaney et al., 2021;Starr et al., 2021).These investigations emphasise the significance of monitoring single and multiple RBD mutations and their increased potential for ACE2 binding and/or immune evasion.The various computational techniques have been used in SARS-CoV-2 research and are easily able to estimate the consequences of harmful mutations on protein function and structure (Ahamad et al., 2022;Alam et al., 2021;Ali et al., 2022;C. Chen et al., 2021;Gan et al., 2021;Laurini et al., 2021;Teng et al., 2009;Teng et al., 2021).
The interaction of RBD and ACE2, which can happen through serological pH as well as low-pH-endosomal routes, is what allows SARS-CoV-2 to infect receptive cells.Therefore, it is crucial to look into the RBD stability and RBD-ACE2 binding affinity.According to this viewpoint, we apply the computational saturation mutagenesis to study the impacts of S-RBD mutations on the stability and binding free energy (BFE) of the RBD protein and the ACE2 at three different pH, namely 5.5, 6.5, and 7.4.Because SARS-CoV-2 variants exhibit different mutation patterns, it is critical to comprehend their mutation consequences.To this purpose, we have used computational saturation mutagenesis to examine 3705 RBD (residue 333-526 on S protein) mutations in experimentally established protein structures at three different pH levels (Q.Wang et al., 2020;Zhou et al., 2020).This is, to the best of our knowledge, the first study to look at the mutation landscape of the SARS-CoV-2 RBD region at different pH levels.

Structure preparation
All the three-dimensional structures are collected from Protein Data Bank (PDB) [16].The cryo-EM structures of SARS-CoV-2 S in complex with single ACE2 at serological pH 7.4 (PDB ID: 7KNB) and at endosomal pH 5.5 (PDB ID: 7KNE) (Zhou et al., 2020), along with the crystal structure of RBD-ACE2 complex at pH 6.5 (PDB ID: 6LZG) (Q.Wang et al., 2020) was used for stability and interaction analysis.

Free energy and binding affinity calculations of S-RBD mutations
To determine the effect of mutations on protein's stability and binding, we here applied saturation mutagenesis to mutate each residue in the complex structure to the other 19 amino acids.The mutation-induced change in the stability and binding affinity of S-RBD was investigated using the FoldX tool (Schymkowitz et al., 2005).Foldx calculates free energy, DG, by incorporating the contributions of hydrophobic and polar groups to the solvation energy, Van der Waals, hydrogen bonding, and electrostatic interactions.These energy parameters were determined experimentally (Schymkowitz et al., 2005).FoldX tool is extensively utilized for computational saturation mutagenesis studies (Cheng et al., 2012;Teng et al., 2021;Vedithi et al., 2020;Xue et al., 2022).Mutations in the protein complex structure were introduced using the BuildModel module of FoldX (Guerois et al., 2002).The folding energy change (DDG) between the mutant structure (mut) and wild-type (wt) structure was calculated using: DDG ¼ DG ðmutÞ À DG ðwtÞ A negative DDG value indicates that the mutation leads to stabilization of the protein structure whereas, a positive value indicates that mutation leads to destabilization of the protein.
Next, the BFE change was computed by the 'AnalyseComplex' command and is mathematically represented as: DDG bind ¼ DG bindðmutÞ -DG bindðwtÞ A negative DDG bind value indicates that the mutation increased the binding energy and a positive value suggests that the mutation decreased the binding energy.

Heatmaps of mutational stability and binding energy
The changes in stability and binding energy for single mutations were used to generate heatmaps in Excel.The y-axis represented the mutation residues, while the x-axis represented the amino acid types of single mutations.The stability or binding energy values of all mutations were colored where the blue color indicated the destabilization of the protein, and the red color indicated the stabilization of the protein.The values of stability or binding energies were represented as a gradient between these two color ranges.

Ligplot analyses
Further, the residual interactions in the three-dimensional complex structures were analysed through PDBSUM server (http:// www.ebi.ac.uk/pdbsum).The molecular interactions in each complex structure of the S-RBD protein with ACE2 were analyzed using DIMPLOT module of ligplotþ (Laskowski & Swindells, 2011).

Mutations induced changes in the stability of SARS-CoV-2 S-RBD protein at different pH
To analyse the effects of the systematic mutations on SARS-CoV-2 S-RBD stability, we generated 3705 mutations by mutating all 195 residues of RBD to all other 19 canonical amino acids and computed the folding energy changes (DDG) introduced by these mutations in monomeric S-RBD structures (Supplementary Tables 1-3).The structures used in the present study are (i) the crystal structure of the RBD-ACE2 complex at pH 6.5 (PDB ID: 6LZG) (Q.Wang et al., 2020), (ii) the cryo-EM structure of SARS-CoV-2 spike in complex with single ACE2 at pH 7.4 (PDB ID: 7KNB) (Zhou et al., 2020), and (iii) cryo-EM structure of SARS-CoV-2 spike in complex with single ACE2 at pH 5.5 (PDB ID: 7KNE) (Zhou et al., 2020).

Stability changes of S-RBD protein at pH 6.5 (6LZG)
The mutations induced free energy changes at pH 6.5 calculated by FoldX are reported in Figure 1.Lower energy values indicate favorable mutations that stabilize the RBD and can improve the binding with the ACE2.Whereas, higher energy values indicate mutations that destabilize the RBD and possibly affect the binding dynamics with the ACE2. Figure 1(A) shows that out of 3705 missense mutations, 2434 (66%) mutations increased the free energy of the RBD by at least 0.5 kcal/mol whereas, 288 (8%) mutations decreased the free energy of the RBD by at most À 0.5 kcal/mol, and 983 (26%) exhibited neutral effect on the stability of the RBD.The Foldx suite has a standard error of energy computation of �0.5 kcal/mol (Schymkowitz et al., 2005).As a result, the folding energy changes within the range (-0.5 < DDG <0.5) are insignificant or classified as neutral.Specifically, 1311 (�35%) mutations highly destabilize the RBD at pH 6.5 (DDG > 2.5 kcal/mol), and 1123 (�30%) mutations moderately destabilize the RBD (0.5 < DDG � 2.5 kcal/mol).Only 12 (�0.32%)mutations had a highly stabilizing effect (DDG <-2.5 kcal/mol), and 276 (�7%) moderately stabilize the RBD (-2.5<DDG � À 0.5 kcal/mol).A total of 983 (�27%) mutations had a neutral effect (-0.5 < DDG � 0.5 kcal/mol) on RBD stability.
The line chart depicts the mutation distribution along the whole length of the RBD (Figure 1B).The positive mean values are represented by the upward lines, while the negative mean values are represented by the downward lines.As illustrated in Figure 1(B), the mean value of DDG ranges from þ25.68 kcal/mol in G431 to À 1.71 kcal/mol in S514, which lies in the S2 subunit of the S protein.Based on the mean value of DDG, mutations at V401, G431, P507, and G526 showed the greatest destabilizing effects (Figure 1B and Table 1).The three most destabilizing missense mutations are V401W (45.63 kcal/mol), G431W (45.34 kcal/mol), P507W (42.45 kcal/mol), and G526H (29.69 kcal/mol).Also, mutations with the highest stabilizing effects are found in residues S375, T385, G504, and S514.The most stabilizing missense mutations are S399L (-2.68 kcal/mol), A397L (-3.90 kcal/mol), and S514F (-3.87 kcal/mol) (Figure 1B and Table 1).

Comparison of effects of mutation on stability of RBD in RBD-ACE2 complex structures at different pH
We compared the mutational effects on the stability change of RBD in the RBD-ACE2 complex at three different pH.Five glycine residues (G404, G416, G526, G431, and G447) showed the highest mean destabilizing effects along with N422.
Whereas, the residues G339, G504, S514, T500, and T523 showed the highest mean stabilizing effects.Interestingly, the distribution of the effect of all missense mutations in the RBD mean stability changes at pH 6.5 and pH 7.4 were highly correlated (R 2 ¼ 0.6193) (Supplementary Figure 1A).Whereas, the distribution of the effect of missense mutations in the RBD mean stability at pH 6.5 and pH 5.5 did not significantly correlated (R 2 ¼ 0.4366) (Supplementary Figure 1B).Moreover, the distribution of the mean stability changes due to mutations at pH 5.5 and pH 7.5 corelated well with R 2 ¼ 0.6884 (Supplementary Figure 1C).
Next, the RBD interfacial mutants induced stability changes were shown as a heatmap (Figure 4), where the y-axis represents the interfacial residues and the x-axis represent the mutations.The computed stability values were coloured with a gradient ranging from blue (stabilized binding) to red (destabilized binding).Figure 4 depicts the different results of single-point mutations and highlights those that are unstable.For some residues, the presence of a mutation has no effect on the stability.Most mutations at other locations will decrease protein stability.Interestingly, mutations to Trp, Tyr, Phe or His for some of the RBD residues generally improve structural stability.The results revealed that mutations at K417, N487, Q493, S494, Q498, and T500 were mostly destabilizing in all the three structures (Figure 4A-C).Moreover, the observation of the mutations induced stability changes indicated analytic similarity in stability analysis in the three complex structures (Supplementary Figure S3).

Mutations induced changes in binding affinity of SARS-CoV-2 S-RBD protein to ACE2 at different pH
Missense mutations in the RBD region could alter the key interaction site and influence RBD-ACE2 binding affinity.The binding free energy (BFE) change due to mutation (DDG bind ) is represented as, DDG bind ¼ DG mut À DG wt , where DG wt is the wild-type BFE and DG mut is the mutant BFE.A positive BFE change suggests that the mutation decreases the free energy of the binding, and negative BFE suggests an increase in the binding affinity of the complex, making the virus more infective.We calculated the BFE changes (DDG bind ) of total 3705 RBD mutations in the RBD-ACE2 complex and classified them into one of five categories according to their DDG bind (Supplementary Tables 4-6).

Binding affinity changes of S-RBD protein in RBD-
ACE2 complex structures at endosomal pH 5.

Comparison of effects of mutations on binding affinities of RBD to ACE2 in RBD-ACE2 structures at different pH
We compared the mutational effects of RBD binding on RBD-ACE2 complexes structure at three different pH.Three residues (L455, A475, and G502) showed the highest mean destabilizing binding affinities, while six residues (N487, Q493, Y495, G496, Q498, and N501) showed the largest mean stabilizing binding effects.Moreover, the distribution of the effect of mutations in the BFE changes at pH 6.5 and pH 7.4 did not show any correlation (R 2 ¼ 0.08) (Supplementary Figure 4A).The distribution of the effect of mutations in the RBD binding affinity at pH 6.5 and pH 5.5 correlated with R 2 ¼ 0.26 (Supplementary Figure 4B) whereas BFE changes due to mutations at pH 5.5 and pH 7.5 corelated with R 2 ¼ 0.32 (Supplementary Figure 4C).

Binding changes of S-RBD interfacial variants at different pH
Figure 6 maps the changes in binding energy upon mutations and exposes unstable ones.Observation of heat maps of binding energy of individual mutations revealed similarity in binding analysis in all the three complex structures.The results revealed that mutations at G476, N487, Y489, G496, and Q498 mostly increased the binding to ACE2 in all the three complex structures (Figure 6A-C and Table 2).In addition, the mutations in the residues Y449, T500, and N501 mostly favor the binding to ACE2 at serological and endosomal pH (Figure 6B and 6C).The mutations at residues K417, G476, and G502 were frequently shown to disrupt the ACE2 binding.Interestingly, the amino acids changes to Asn, Gln, Cys, Gly, or Ala of the RBD interfacial residues generally reduces the binding to ACE2.Moreover, the observation of the mutation induced stability changes indicated analytic similarity in stability analysis in the three complex structures (Supplementary Figure S5).

Discussion
The SARS-CoV-2 spike protein is one of the primary targets for vaccine development and neutralising antibodies, and many have acquired immune evasion mechanisms, some of which match characteristics of the recently revealed endosomal pH-dependent conformational masking (Zhou et al., 2020).The stability of the S protein is decisive for the speedy transmissions of infection (Moreira et al., 2020) and is critical in producing therapeutic drugs and vaccines (Kyriakidis et al., 2021).
Charles et al. (Charles et al., 2023) recently used ITC to compare the binding affinities of these variants to hACE2.For each variant, the K D values ranged from 2 nM for Alpha-RBD to 6.0 nM for the Delta variant.In comparison to the original Wuhan strain (WHCV), all of the variants demonstrated tighter binding to the hACE2 receptor.The trend of binding affinity was as follows: This study along with many others have indicated that all of the variants had a stronger affinity for hACE2 than WHCV (Ali et al., 2022;Tian et al., 2021;X. Xue et al., 2021).This suggests that mutations that promote tighter binding could be a contributing factor in the observed difference in transmissibility.
To understand how mutations affect the stability of S-RBD protein at different pH, we predicted the stability changes of RBD protein at three different pH.Our prediction showed that at both serological (pH 7.4) and endosomal pH (pH 5.5), half of the RBD mutations (�54-56%) destabilize the RBD.Whereas, at pH 6.5, more than half of the S-RBD mutations (�66%) are destabilizing.Majority of these destabilizing mutations included the substitution of Gly, Val and Ala residues, which are hydrophobic amino acids with longer hydrophobic side chains.Prediction of the effects of mutations on the stability of RBD at different pH revealed nearly identical results.The top five amino acid residues with the highest average destabilizing effect were G404, G431, G447, A475, and G526 (Table 1).This study also showed a significant correlation in the effects of mutations on the RBD stability at both serological and endosomal pH, suggesting that a mutation in the RBD will have a similar effect on the stability and folding of the RBD regardless of the different pH values.
The binding of the S-RBD to human ACE2 allows SARS-CoV-2 to enter the human cells (Hoffmann et al., 2020).Our results showed that most of the RBD mutations have small BFE changes, while some of them showed large BFE changes.Almost 20% of the RBD mutations decreased the binding affinity of the RBD protein to human ACE2.Whereas, 30%-35% mutations increased the binding affinity, leading to more infectious SARS-CoV-2.It is worthy to note that residues Y449, Y489.Y495, Q498, and N487 are potential hot spots and will increase the S-RBD-ACE2 binding (Table 2).Interestingly, at pH 5.5 and 7.5, the effects of mutations on S-RBD binding affinity to hACE2 are somewhat corelated.

Conclusion
The infectivity of SARS-CoV-2 is a critical factor for preventive measurements against COVID-19.However, determining the viral infectivity of all SARS-CoV-2 variants experimentally is extremely difficult.The continuous evolution of SARS-CoV-2 worsens these issues.We investigated the potential hotspots of SARS CoV-2 RBD that can destabilise and decrease ACE2 binding at both endosomal and serological pH in this work.Computational analysis revealed that the interfacial RBD residues K417, N487, Q493, S494, Q498, and T500 are hotspots because their variations can significantly disrupt the RBD protein.Notably, different studies also supported the K417, N487, Q493, and S494 as hotspots.The mutations at residues K417, G476, and G502 were frequently seen to affect ACE2 binding.
In conclusion, our findings shed light on the putative effects of potential hotspots on interactions with ACE2 at different pH levels, which can aid in the development of new therapeutic drugs against potential SARS-CoV-2 variants.

Figure 1 .
Figure 1.Effects of missense mutations on RBD stability at pH 6.5.(A) Pie charts summarize the contribution of mutations on stability changes in RBD at pH 6.5.(B) Line chart showing mean DDG values of destabilizing mutations (positive values depicted as upward lines) and stabilizing mutations (negative values depicted as downward lines).

Figure 2 .
Figure 2. Effects of missense mutations on RBD stability at serological pH 7.4.(A) Pie charts summarize the contribution of mutations on stability changes in RBD at pH 7.4.(B) Line chart summarizing the stability changes for DDG mean of residues.The destabilizing mutations are depicted as positive values and stabilizing mutations are depicted as negative values.

Figure 3 .
Figure 3. Effects of missense mutations on RBD stability at endosomal pH 4.5.(A) Pie charts summarize the contribution of mutations on stability changes in RBD at pH 4.5.(B) Line chart summarizing the stability changes for DDG mean of residues.The destabilizing mutations are depicted as positive values and stabilizing mutations are depicted as negative values.

Figure 4 .
Figure 4.The effect of SARS-CoV-2 S-RBD interfacial mutations on protein stability.Heatmap shows the stability changes, DDG of all interfacial mutations of RBD at pH (A) 6.5, (B) 7.4, and (C) 4.5.The boxes of each mutation were colored with the gradient of a range between blue (destabilized) and red (stabilized) along with their DDG values.
Note: The residues or mutations with the maximum or minimum binding affinity changes, DDG bind mean and DDG bind values at three different pH are shown as bold fonts.

Figure 5 .
Figure 5.The effect of RBD mutations on ACE2 binding affinity.Pie charts summarize the contribution of mutations on binding affinity changes in RBD at (A) pH 6.5, (B) pH 7.4, and (C) pH 4.5.

Figure 6 .
Figure 6.Effects of RBD interfacial mutations on ACE2 binding interaction.Heatmap shows the binding free energy changes, DDG bind of all interfacial mutations of RBD at pH (A) 6.5, (B) 7.4, and (C) 4.5.The mutations stabilizing (red) and destabilizing (blue) the binding interactions are shown in the heatmaps along with their DDG bind values.

Table 1 .
Important residue and mutations based on stability changes of SARS-CoV-2 S-RBD.
Note: The residues or mutations with the maximum or minimum stability changes, DDG mean and DDG values at three different pH are shown as bold fonts.