Effect of Double Mutation (L452R and E484Q) on the Binding Affinity of Monoclonal Antibodies (mAbs) against the RBD—A Target for Vaccine Development

The COVID-19 pandemic, caused by SARS-CoV-2, emerges as a global health problem, as the viral genome is evolving rapidly to form several variants. Advancement and progress in the development of effective vaccines and neutralizing monoclonal antibodies are promising to combat viral infections. In the current scenario, several lineages containing “co-mutations” in the receptor-binding domain (RBD) region of the spike (S) protein are imposing new challenges. Co-occurrence of some co-mutations includes delta (L452R/T478K), kappa (L452R/E484Q), and a common mutation in both beta and gamma variants (E484K/N501Y). The effect of co-mutants (L452R/E484Q) on human angiotensin-converting enzyme 2 (hACE2) binding has already been elucidated. Here, for the first time, we investigated the role of these RBD co-mutations (L452R/E484Q) on the binding affinity of mAbs by adopting molecular dynamics (MD) simulation and free-energy binding estimation. The results obtained from our study suggest that the structural and dynamic changes introduced by these co-mutations reduce the binding affinity of the viral S protein to monoclonal antibodies (mAbs). The structural changes imposed by L452R create a charged patch near the interfacial surface that alters the affinity towards mAbs. In E484Q mutation, polar negatively charged E484 helps in the formation of electrostatic interaction, while the neutrally charged Q residue affects the interaction by forming repulsive forces. MD simulations along with molecular mechanics-generalized Born surface area (MMGBSA) studies revealed that the REGN 10933, BD-368-2, and S2M11 complexes have reduced binding affinity towards the double-mutant RBD. This indicates that their mutant (MT) structures have a stronger ability to escape from most antibodies than the wild type (WT). However, EY6A Ab showed higher affinity towards the double MT-RBD complex as compared to the WT. However, no significant effect of the per-residue contribution of double-mutated residues was observed, as this mAb does not interact with the region harboring L452 and E484 residues.


Introduction
The ongoing coronavirus pandemic  has affected lives across the globe. Vaccination and neutralizing antibodies (NAbs) have emerged as the most promising strategies to end the pandemic [1]. The emergence of more virulent variants of SARS-CoV-2 has triggered intensive genomic surveillance [2]. S glycoprotein (S) of the viral genome mediates the fusion of viral proteins with the human angiotensin-converting enzyme 2 (hACE2) receptor [3]. Therefore, S glycoprotein is considered the primary target for most vaccine and therapeutic antibodies [3,4]. The rapidly spreading lineage B.1.617.1 was called the kappa variant of interest (VOI) by the World Health Organization (WHO) [5]. It was first identified in India in December 2020 and has spread across the globe to at least with the RBD of the S protein of SARS-CoV-2 are available in the PDB. A representative Ab complex for each class was considered in this study. A representative Ab complex for each class-REGN10933 for Class I [25], BD-388-2 for Class II [26], S2M11 for Class III [27], EY6A for Class IV [28], and JMB2002 [21] for Class V-was selected ( Figure 1). MD simulations were performed for the Ab complex of each class to study the effect of the double mutation on antibody binding. This work contributes to understanding the molecular interaction between the viral RBD and different monoclonal antibodies (mAbs). This can aid in developing new drugs and vaccine strategies in the future, as well as in ascertaining the efficacy of existing drug therapy.
Vaccines 2023, 11, x FOR PEER REVIEW 4 of 22 pressure at 1 atm and temperature at 300 K. The cut-off distance for short-range interaction was set at 10.0 Å, while the long-range columbic interactions were assessed using the smooth particle mesh Ewald method (PME). A time-reversible reference system propagator algorithm (RESPA) integrator was used to accelerate the run, with an inner time step of 2.0 fs. The MT-RBD complexed structure with mAbs was extracted at 100 ns of MD simulations and was superimposed with their respective WT-RBD structures to identify the conformation changes using PYMOL.

Preparation of Wild-Type and Mutated Ag-Ab Complex
The three-dimensional coordinates of structures of the wild-type (WT) RBD bound to monoclonal antibodies (mAbs) were retrieved from the Protein Data Bank (Table 1). The structures were prepared using the Protein Preparation Wizard by assigning the correct bond order and ionization states; adding missing atoms/side chains; assigning partial charges; and eliminating unwanted sugar molecules, cofactors, and ligand molecules [29]. In all the four complexed structures, the missing loop regions of the RBD were built using Schrödinger Prime [30]. Hydrogen atoms were added, and a standard protonation state at pH 7.0 was used [31]. The co-mutant complexes in Classes I to V were generated by mutating the amino acids at Position 452 from leucine (L) to arginine and replacing glutamic acid (E) with glutamine (Q) at Position 484. These mutant complexes (MT) were similarly processed using the Protein Preparation Wizard. For the Class V mAb, the interactions of the Omicron-mutated RBD with the mAb were obtained using the Pymol and PDBsum server.
MD simulations were implemented using the Desmond module. The OPLS force field parameter was employed to run all calculations [32]. Both wild-type (WT) and mutant-type (MT) RBD structures, along with the mAb complex, were positioned in an orthorhombic box that was solvated with the TIP3P water system [33]. This step was performed using the System Builder menu of Desmond. Sodium and chloride ions were added to neutralize the system. Both WT and MT-RBD complexes, along with their respective mAbs, were subjected to Desmond's default eight-stage relaxation protocols before the commencement of the production run. All simulations were carried out in an isothermal-isobaric ensemble (NPT) system. For the simulations, the Nose-Hoover thermostat and the isotropic Martyna-Tobias-Klein barostat were used to maintain the pressure at 1 atm and temperature at 300 K. The cut-off distance for short-range interaction was set at 10.0 Å, while the long-range columbic interactions were assessed using the smooth particle mesh Ewald method (PME). A time-reversible reference system propagator algorithm (RESPA) integrator was used to accelerate the run, with an inner time step of 2.0 fs. The MT-RBD complexed structure with mAbs was extracted at 100 ns of MD simulations and was superimposed with their respective WT-RBD structures to identify the conformation changes using PYMOL.

Trajectory Analysis after Molecular Dynamics (MD) Simulation
The root mean square deviation (RSMD) represents the deviation from the initial minimized crystal structure for each of the system studies, while root mean square fluctuation (RMSF) represents the overall trend of flexibility of each residue in the protein. The change in RMSD values indicates the local conformational alternations occurring in the backbone of the protein residues due to these mutations. A higher RMSF value indicates increased flexibility, which points towards their potential to interact with other molecules, while a lower RMSF indicates lower fluctuation, implying less flexibility and diminished binding potential. RMSD and RMSF values for the protein backbone were taken using the simulation interaction diagram tool implemented in Desmond, and a graph was generated for the same values. The molecular interactions of the S protein RBD with mAbs were determined by submitting the obtained PDB at the PDBsum server [34].

MMGBSA Analysis
Binding free energies of the RBD complex with different mAbs were calculated using Schrödinger Prime employing the VSGB 2.0 solvation model [35]. For each RBD-mAb complex, energy was computed by taking 200 frames from the last 20 ns of simulation time [36]. The MMGBSA analysis calculates the binding energy and its constituent individual energy terms, including columbic, covalent, van der Waals (vdW), lipophilic (lipo), generalized Born electrostatic solvation (Solv GB), and hydrogen bonding (H bond) terms [37]. The obtained MMGBSA energy values were then averaged, and the standard deviation was calculated. Herein, the net binding free energy was computed using the thermal_mmgbsa.py script.
Per-residue decomposition was obtained, to identify the energy contribution of the crucial RBD residue that participated in effective binding with mAbs. The per-residue interaction energy was calculated using breakdown_MMGBSA_by_residue.py.

Comparisons of Biomolecular Interaction between the mAb and RBD Complexes
The molecular interactions between the SARS-CoV-2 RBD and mAb were studied after 100 ns MD simulation.

Interaction of REGN10933 with WT and MT-RBD
The cryo-electron microscopy (cryo-EM) [38] structure of the Fab region of REGN10933 reveals that it binds to the RBD from the topmost site and overlaps with the hACE2 binding site [38]. We studied the effect of the RBD double mutation on mAb binding by comparing the molecular interactions between the wild-type (WT) and mutated-type (MT) RBD complexes. The interfacial residues between the RBD and REGN10933 were analyzed in the post-MD complexes (Figure 2A,B). Several hydrogen bonds present between the heavy-chain (HC), complementarity-determining regions (CDRs) and the RBD, involving the HC residues D31, Y33, and T52 and the RBD residues K417, Y453, E484, F486, C488, and Y489 ( Figure 2A). The complementarity-determining regions (CDRs) are the hypervariable domains of the Ab that determine the specific antibody interaction with the antigen molecule. The light-chain (LC) residues of the mAb are responsible for stabilizing the HC residues to facilitate the interaction with the RBD of the S protein. Here, we focused specifically on the co-mutant residues of the RBD, i.e., L452 and E484, and their induced local perturbations on the protein's overall structure. The L452 (RBD) residue made key intramolecular hydrophobic interactions with L492 and F490 that constitute the hydrophobic patch on the surface of the RBD. The E484 residue forms intermolecular interactions with the HC of REGN10933 through the S56. L452 and E484, and their induced local perturbations on the protein's overall structure. The L452 (RBD) residue made key intramolecular hydrophobic interactions with L492 and F490 that constitute the hydrophobic patch on the surface of the RBD. The E484 residue forms intermolecular interactions with the HC of REGN10933 through the S56.  Here, only intermolecular interactions are shown. RBD represents receptor-binding domain, while HC represents heavy chain of monoclonal antibody (mAb). The interacting residues are shown in sticks. RBD is shown in magenta cartoon, and HC is shown in green color as a residue of mAbs ( Figure 2B). In the MT-RBD-REGN1093 complex, the replacement of the hydrophobic Leu (L) residue with positively charged polar Arg (R) residue abolishes the intramolecular hydrophobic contact with L490 ( Figure S1A,B), while the substitution of Glu (E484), which has a polar acidic side chain with Gln (Q), results in the formation of a new hydrogen bond interaction with T52 and Y33 residues of HC. The key salt bridge interaction of K417 in RBD with D31 (HC region) was found to be lost in the MT-RBD complex ( Figure 2C,D). The WT-RBD complex with REGN10933 displayed additional stabilizing interactions as compared to the MT complex.

Interaction of BD-368-2 with WT and MT-RBD
BD-368-2 is one of the most potent neutralizing monoclonal Abs against SARS-CoV-2 and gave an IC50 of 1.2 and 15 ng/mL against both pseudotyped and authentic virus, respectively [39]. This mAb also exhibited high therapeutic and prophylactic efficacy in human ACE2 (hACE2) transgenic mice infected by SARS-CoV-2 [39]. BD-368-2 fully blocks the hACE2 recognition and targets both the "up" and "down" RBD conformations. The result showcased that the WT-RBD complex with the mAb complex is stabilized by multiple hydrogen bonds, salt bridge interactions, and non-bonded contacts as compared to its mutant type ( Figure 3A,B). Three Fab regions of heavy-chain BD-368-2 (CDRH1, CDRH3, and DE loop in the VH domain) are involved in interaction with the RBD [26]. However, in CDRH1, the residues G26, F27, and A28 form a prominent non-covalent interaction with the residue Y449 (RBD). The residue N450 of the RBD forms hydrogen bond interactions with N74 and N77 of the HC. The amino acid residue L452 is involved in intermolecular contacts and interacted through non-covalent contacts with T31 of the HC (mAb) ( Figure 3A). E484 (RBD) was observed to be the key residue that facilitated both salt bridge and hydrogen bond interactions with the R100 (HC), R102 (HC), and D106 (HC) residues ( Figure 3A,B). The exchange of the non-polar leucine residue at Position 452 with a hydrophilic, polar arginine residue (L452R mutation) sterically restricted the interaction of L452 with the T31 residue of the HC. This interaction also influenced the hydrogen bond interactions formed by N450 (RBD). The substitution E to Q in the mutant resulted in a slight movement of N450 (3.5 Å) away from the HC of the mAb, leading to the disruption of this interaction ( Figure 3C). The salt bridge that formed between E484 (RBD) and R102 (HC) at the interfacial binding region was also not observed due to altered polarity and different spatial conformation imposed by the uncharged Gln residue ( Figure 3D).

Interaction of S2M11 with WT and MT-RBD
S2M11 belongs to Class III mAbs that bind away from the hACE2 binding region. These mAbs possess the ability to bind the RBD in both up and down conformation. The S2M11 mAb associates with the RBD to form a bridge between the neighboring RBDs present on the S trimer surface [40]. This mAb has the ability to fully block hACE2 binding by interacting with the two adjacent or all three RBDs simultaneously. S2M11 is known to exhibit strong therapeutic and prophylactic effects against both WT and pseudotyped (PT) virus [41]. This mAb targets the two neighboring RBDs of the same trimer consisting of the receptor-binding motif (RBM) residues from one S monomeric unit and the RBM from an adjacent S monomeric unit. Most of the WT-RBD interactions are mediated through heavy-chain CDRs (HCCDRs) of S2M11. The L452 residue does not interact with the mAb molecule; however, it mediates the formation of the intramolecular hydrophobic patch with F490 and L492. The substitution of hydrophobic L452 with the positively charged R residues abolishes the hydrophobic interaction with L492. The side chains of residues Y449 and E484 of the RBD are hydrogen-bonded to the F29 backbone amide and the N52/S55 side chains of S2M11 heavy chain, respectively ( Figure 4A,B). Q493 makes hydrogen-bonded interactions with both T30 and Y103 of the heavy-chain CDR of S2M11 ( Figure 4A). The replacement of E484 (RBD) with the negatively charged Q residue retained the original interaction with N52 and S55 of the HC residues and resulted in the formation of an additional hydrogen bond with Y33 (HC) ( Figure 4C,D). In the MT-RBD complex, an extra hydrogen-bonded interaction was observed in comparison to the WT complex. However, the WT complex displayed the presence of a higher number of non-bonded contacts. CDR of S2M11 ( Figure 4A). The replacement of E484 (RBD) with the negatively c Q residue retained the original interaction with N52 and S55 of the HC residues sulted in the formation of an additional hydrogen bond with Y33 (HC) (Figure 4 the MT-RBD complex, an extra hydrogen-bonded interaction was observed in c son to the WT complex. However, the WT complex displayed the presence of a number of non-bonded contacts.

Interaction of EY6A with WT and MT-RBD
The Fab region of EY6A binds at the RBD site, which is completely distinct from the hACE2 binding site. Previous experiments with surface plasmon resonance (SPR) revealed that the Fab fragment of EY6A binds to immobilized SARS-CoV-2 RBD with a KD value of 2 nM [42]. EY6A binds to the RBD at the α2-helix (Residues 365-371) and the α3-helix (Residues 384-388) away from the loop region where the residues L452 and E484 are located. A total of 12 residues from the HC and 13 residues from the RBD participate to form the binding interphase. The residue T385 of the RBD interacts with G101 and Y106 residues of the HC through hydrogen bond interaction. The residues K386 (RBD) and D99 (HC) are involved in a salt bridge interaction and further stabilize the complex. Y106 residue of the HC also forms hydrogen bond contacts with T385 of the RBD. However, six hydrogen bond interactions along with other non-bonded hydrophobic interactions further increase the binding affinity of EY6A with the RBD (Figure 5A,B). Since the residues L452 and E484 do not directly make any contact when EY6A binds to the RBD, the effect of these mutations on EY6A binding was not observed (Figure 5C,D).

Interaction of JMB2002 with WT and MT Omicron-RBD
JMB2002 represents a new class of SARS-CoV-2 neutralizing antibody that has a different mechanism of binding from other reported Nabs, and hence has been classified as a Class V Nab. The reported electron microscopy structure revealed that the Fab region of JMB2002 binds only to the down conformation of the RBD and inhibits hACE2 binding. The residues R346 and K444 of the RBD dually interact with E107 and D108 of mAbs HC, respectively, to form both a salt bridge and a hydrogen bond ( Figure 6A,B). The complex is further stabilized through a hydrogen bond interaction with the S446 (RBD)/R50 (HC) and Y449 (RBD)/T57 (HC) residues. The L452 residue situated in the middle of the binding epitope forms a hydrophobic interaction with Y102 and S103 of the HC of Nab. The E484A mutation in Omicron helps to improve the interaction between the RBD and hACE2 and displays strong affinity towards hACE2. The residue A484 in Omicron is observed to lie away from the binding epitope of JMB2002 ( Figure 6A). In the L452R MT Omicron-JMB2002 complex, as compared to the WT, an additional salt bridge interaction is observed between K444 (RBD) and E107 (HC) that is responsible for the enhanced affinity of the RBD against this Nab. R452 residue in the mutated RBD complex forms hydrophobic contacts with the F55, Y102, and S103 residues of the HC. The predominance of non-polar residues can result in hydrophobic clash of this Nab with the RBD (Figure 6C,D).

RMSF Calculations
The RBD consists of both β sheets and a few loop regions, while the HC and LC regions of mAbs mostly contain random coil structures. The higher fluctuations of both WT and MT complexes are mainly due to the presence of the loop regions. MT-RBDs of REGN10933, BD-368-2, S2M11, EY6A, and JMB-2002 display relatively more fluctuation as compared to their WT complexes (Figure 8). The values for the WT and MT-L452R/E484Q residue have been provided in Table 2. Analysis of RMSF values of L452R and E484Q revealed a greater degree of fluctuation in the mutant structure as compared to WT.

Per-Residue Energy Contribution
Per-residue decomposition analysis was performed to estimate the energy contribution of the single residues of the RBD by summing its interactions over all HC and LC residues in the complex. In the WT-RBD-REGN10933 complex, the ∆Gbind contribution of L452 (−1.36 kcal/mol) and E484 (−4.93 kcal/mol) residues was higher as compared to the residues in the MT complex (L452R (−0.36 kcal/mol) and E484Q (0.51 kcal/mol) ( Table 4). This is due to the more negative columbic, van der Waals, and hydrogen bonding interactions observed as compared to the MT complex. The L452R mutation constitutes the replacement of non-polar Leu residue by the polar Arg, which contributes to the formation of interactions with solvent molecules as well as to the overall stabilization of the complex. The ∆Gbind contribution clearly points towards the formation of more hydrogen bond and non-bonded contact in the WT complex over the MT complex, leading to a stronger binding ability to the Ab (Table 4). In WT-RBD-BD-6322, the energy contribution of L452 (−1.17 kcal/mol) and E484 (−5.37 kcal/mol) residues was found to be comparatively higher than its respective MT complex ((L452R (−0.87 kcal/mol) and E484Q (−0.90 kcal/mol)) ( Table 4). The L452 residue of the RBD forms intermolecular van der Waals contacts with T31 (HC) of BD-6322, contributing to its more negative van der Waals energy (∆G Vdw −1.54 kcal/mol) over the MT-RBD-BD-6322 complex (∆G Vdw −0.54 kcal/mol). The columbic and hydrogen bond energy contribution of E484 residue was found to be higher for the WT-RBD-BD-6322 complex as compared to MT-RBD-BD-6322 (Table 4). This can be accounted for by the strong salt bridge interaction and more hydrogen bond interaction in the WT complex as compared to the MT complex ( Figure 4). The energy contribution for the residue L452 (−2.57 kcal/mol) in WT-RBD-S2M11 was comparatively more negative than its respective mutant complex (−0.86 kcal/mol). The increased energy is associated with favorable intramolecular van der Waals energy in WT-RBD over MT-RBD. Q484 has more ∆G H bond (−0.33 kcal/mol) as compared to E484 (−0.21 kcal/mol), which is further complemented with an additional hydrogen bond observed in WT ( Table 4). The EY6A mAb does not present any interaction with L452 and E484 residues of the RBD. No significant change in per-residue binding energy contribution of co-mutant L452R and E484Q was observed, as this mAb binds at a region situated away from the hACE2 binding site (Table 4). Since it does not bind at the hACE2 binding site, EY6A mAb can be used in combination with other neutralizing nanobodies to provide a blueprint for designing antibody cocktails and therapeutics. In Omicron-RBD, the ∆Gbind contribution of L452 (−2.19 kcal/mol) residues was higher as compared to its L452R complex (L452R (−0.76 kcal/mol) ( Table 4). The more negative ∆Gbind accounts for its stable interaction with Omicron-RBD as compared to its mutant.

Conclusions
The RBD (Residues 331-524) of SARS-CoV-2 S glycoprotein constitutes an essential drug, vaccine, and mAb target due to its interaction with the human receptor ACE2, as well as its response in the host immune response [43]. Many mutations in the emerging viral variants are located in the RBD, which are targeted by a majority of human-neutralizing mAbs from COVID-19 patients and other mAb therapies under investigation [44]. The SARS-CoV-2 variants in the RBD interfere with the host immune system and impair the antibody-mediated neutralization of the virus. A majority of mAbs exhibit neutralizing activity against SARS-CoV-2, generated by blocking the attachment of the RBD to the hACE2 receptor [45]. However, due to the emergence of several new variants in the RBD during the pandemic, these mAbs showcase weak or reduced neutralization activities. Our MD simulation combined with MMGBSA study elucidates the effect of the RBD double mutation (L452R and E484Q) on the binding affinity of different mAbs that interact with the S protein. These mutations were found to alter the interfacial interactions between the spike RBD and mAb. REGN10933, BD368-2 (Class II), S2M11 (Class III), and JMB2002 (Class V) were found to display better binding affinity with WT-RBD as compared to MT-RBD. This can be attributed to the comparatively more stable hydrophobic, electrostatic, and hydrogen bond interactions in the WT complex as compared to MT-RBD. The decrease in binding affinity of mutated complexes is majorly due to the occurrence of unfavorable interactions between the RBD and mAb. The observed double mutation can be associated with weakened neutralizing activity and immune evasion. EY6A mAb was observed to have greater binding affinity with MT-RBD than WT, indicating that these were also consistent in their action against a mutated virus. This study may provide useful information in the ongoing development of effective vaccines and therapeutic antibodies. Author Contributions: D.G. and M.K. designed research with inputs from P.K. and M.K. performed unbiased molecular dynamics simulations and analyzed results. A.P., T.M. and R.K. designed data tabulation and figures. P.S., D.G. and P.K. wrote and revised the paper. All authors have read and agreed to the published version of the manuscript.