key: cord-0708479-gmy3ldpn
authors: Patrick, Casey; Upadhyay, Vaibhav; Lucas, Alexandra; Mallela, Krishna M.G.
title: Biophysical fitness landscape of the SARS-CoV-2 Delta variant receptor binding domain
date: 2022-05-06
journal: J Mol Biol
DOI: 10.1016/j.jmb.2022.167622
sha: 4831f25a4e5189c82472e236912b690d33d91890
doc_id: 708479
cord_uid: gmy3ldpn

Among the five known SARS-CoV-2 variants of concern, Delta is the most virulent leading to severe symptoms and increased mortality among infected people. Our study seeks to examine how the biophysical parameters of the Delta variant correlate to the clinical observations. Receptor binding domain (RBD) is the first point of contact with the human host cells and is the immunodominant form of the spike protein. Delta variant RBD contains two novel mutations L452R and T478K. We examined the effect of single as well as the double mutations on RBD expression in human Expi293 cells, RBD stability using urea and thermal denaturation, and RBD binding to angiotensin converting enzyme 2 (ACE2) receptor and to neutralizing antibodies using isothermal titration calorimetry. Delta variant RBD showed significantly higher expression compared to the wild-type RBD, and the increased expression is due to L452R mutation. Despite their non-conservative nature, none of the mutations significantly affected RBD structure and stability. All mutants showed similar binding affinity to ACE2 and to Class 1 antibodies (CC12.1 and LY-CoV016) as that of the wild-type. Delta double mutant L452R/T478K showed no binding to Class 2 antibodies (P2B-2F6 and LY-CoV555) and a hundred-fold weaker binding to a Class 3 antibody (REGN10987), and the decreased antibody binding is determined by the L452R mutation. These results indicate that the immune escape from neutralizing antibodies, rather than receptor binding, is the main biophysical parameter that determined the fitness landscape of the Delta variant RBD.

In late 2019, a novel coronavirus (2019-nCoV), later renamed as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was discovered in Wuhan, China and quickly became the center of the ongoing pandemic coronavirus disease 19 . SARS-CoV-2 enters human host cells with its spike protein interacting with the angiotensin converting enzyme 2 (ACE2) located on the cell surface. [1] [2] [3] [4] [5] A specific structural region within the spike protein, known as the receptor binding domain (RBD), binds to the ACE2 receptor. SARS-CoV-2 has been shown to continuously mutate in multiple regions of the spike protein leading to new variants of interest (VOI) and more severe variants of concern (VOC). VOCs in general have been shown to have increased infectivity, 6-9 enhanced ACE2 binding, 10-12 escape from the human immune system, 4,10, 13 and evade FDA-approved monoclonal antibody therapies. 4, 6, 9 To date, there have been five known VOCs, which include Alpha, Beta, Gamma, Delta, and the newly discovered Omicron.

Out of all the five VOCs, Omicron is the most transmissible variant, whereas Delta is the most virulent leading to more severe symptoms and an increased mortality among infected patients. Delta variant arose from the B.1.617 lineage, and is specifically labeled as variant B.1.617.2. It was first identified in India and has since been accounted for the majority of COVID-19 deaths worldwide. 14, 15 Delta variant has been shown to have higher viral titers in COVID-19 patients compared to previous variants. [15] [16] [17] Prior to the emergence of Omicron variant, increased breakthrough infections of COVID-19 in vaccinated patients have been attributed to the Delta variant. [16] [17] [18] [19] A single dose of vaccination was found to be only 33% effective in protecting against the Delta variant as opposed to 48.7% against the Alpha variant, and two vaccination doses were only 88% effective for the Delta variant compared to 93.7% for the Alpha variant. 20 Delta variant introduces several mutations in the N-terminal domain (NTD), RBD, and the furin cleavage site of the spike protein that makes it distinct from the unmutated, wild-type (WT) virus. 21 Unlike previous variants, which have had mutations that have been predicted by in vitro evolution through biophysical parameters such as ACE2 binding, the mutations in the Delta variant, particularly in the RBD, have not previously been predicted to lead to a more dangerous variant. 12, 22 Delta RBD contains two mutations which change the characteristic nature of the amino acid: a hydrophobic amino acid leucine mutated to a positively charged amino acid arginine at position 452 (L452R) and an uncharged amino acid threonine mutated to a positively charged lysine at position 478 (T478K) in the primary structure of the protein. 6, 9, 14, 23, 24 None of these two mutations are part of the RBDs of previously discovered VOCs that include Alpha (N501Y), Beta (K417N/E484K/N501Y), or Gamma (K417T/E484K/N501Y). The newly discovered Omicron VOC RBD contains only the T478K mutation and not the L452R mutation. 25 Analyzing the biophysical parameters that determine the fitness landscape of viruses is of considerable interest in recent years, particularly in the case of HIV, influenza, dengue, hepatitis C, and other retroviruses. [26] [27] [28] [29] [30] [31] In the case of SARS-CoV-2 RBD, we along with others have recently shown that increased receptor binding, escape from neutralizing antibodies, and maintaining protein structure, stability, and expression despite the non-conservative nature of amino acid mutations are important parameters that direct the natural selection of mutations and determine the biophysical fitness landscape of emerging variants. 10,32-37 These biophysical analyses were done on Alpha, Beta, and Gamma VOCs before the Delta variant has emerged.

Whether the two novel mutations of the Delta RBD, which were not part of the previous VOCs, follow similar natural selection principles is not clear. This study examined the effect of the two single amino acid mutations L452R, T478K and the double mutant L452R/T478K on the biophysical properties (structure, stability, receptor binding, and binding to neutralizing antibodies) of the RBD. Our results indicate that the Delta RBD does not show increased ACE2 receptor binding unlike the previous VOCs (Alpha, Beta, and Gamma), but shows increased protein expression, consistent with increased spike protein expression and viral titers in Delta patients, and escapes multiple neutralizing antibodies. Increased expression and antibody escape is determined by the L452R mutation.

VOCs including Delta are known to have decreased levels of neutralization titers in both vaccinated and unvaccinated individuals. [15] [16] [17] 38 In addition, Delta variant COVID-19 patients have viral titers ten times higher than that of the other variants. 15 A plausible explanation is that the mutations in the Delta variant might have a selective advantage in terms of increased expression of viral proteins over the wild-type virus. Higher quantities of the viral proteins could allow for more virus particles to be created. 39 In order to test if this clinical observation could be correlated with the increased expression of RBD, which is a major part of the spike protein, expression of the wild-type (WT) RBD and its Delta mutants (L452R, T478K, and L452R/T478K) was tested in human embryonic kidney cells (Expi293 HEK). HEK cells can be used to test the mutation effects on protein expression levels, since protein expression profiles and the quality control mechanisms in different human cells are similar. In addition, SARS-CoV-2 has been shown to infect the kidney cells, 40-42 and HEK cells naturally express the ACE2 receptor. After 48 hours of post-transfection, secreted proteins in the supernatants were analyzed using SDS-PAGE ( Figure 1A ) and the expression levels were quantified as a ratio of the mutant over WT RBD (Figure 1B) . Both the L452R single mutant and the Delta double mutant L452R/T478K showed ~70% higher expression compared to WT RBD ( Figure 1B) . No significant differences in expression were observed for the T478K mutant compared to the WT RBD. These results indicate that the L452R mutation is responsible for the increased expression of Delta variant RBD and possibly the spike protein expression.

Both mutations L452R and T478K are non-conservative mutations where one type of amino acid is mutated to another type of amino acid with differing physical properties. Such mutations tend to destabilize proteins if the amino acid prior to mutation is involved in stabilizing the protein structure. To test the effect of mutations on RBD structure, we used far-UV circular dichroism and fluorescence spectroscopy. Figure 2A shows SDS-PAGE of purified proteins, and the single bands on the gel show the high purity of protein samples used for biophysical analyses reported in this manuscript. Figure 2B shows the far-UV circular dichroism (CD) and Figure 2C shows the intrinsic protein fluorescence spectra of the WT RBD and its Delta single mutants and the double mutant. Spectra of the WT match those reported in the literature. 10, [43] [44] [45] More importantly, none of the mutations caused significant changes in the spectra, implying that the Delta mutations do not affect the global protein structure.

Protein stability could provide valuable insight into both the viability and flexibility of proteins and has been shown to play a big role in the fitness of viruses. 27, 46 To evaluate how the Delta mutations alter the RBD stability, both thermal and urea denaturation melts were utilized.

Change in protein structure with increase in temperature ( Figure S1 ) was fit to a two-state unfolding model (Equation 1 in Materials and Methods) to obtain the midpoint melting temperature (T m ) of the proteins. Since the thermal melts are not reversible, T m values can only be used as a qualitative measure of protein stability. 10,47 Table S2 lists the mean fit parameters obtained from three independent batches of protein expression. Compared to WT RBD, which showed a T m of 56.1 ± 0.7°C, L452R displayed a similar stability of 56.5 ± 0.2°C, while T478K displayed a slightly decreased T m of 54.1 ± 0.1°C. However, the Delta double mutant L452R/T478K exhibited a T m of 56.6 ± 0.2°C, similar to that of the WT ( Figure S1 & Table   S1 ).

Urea denaturation melts of RBD variants are completely reversible ( Figure S2 ). The native signal showed a large change with denaturant concentration, which might indicate partial unfolding and non-2-state unfolding behavior that needs to be further probed. However, since we Table S2) . These equilibrium stability values obtained from urea denaturant melts agree quite well with the trends observed with thermal denaturation melts ( Table S1 ) and indicate that none of the Delta mutants significantly affect RBD stability.

Since SARS-CoV-2 enters host cells with its RBD binding to ACE2, the relative binding affinity of the RBD can play a key role in how variants are evolving. An increase in the affinity of the Delta variant to ACE2 could allude to a potential mechanism where the VOC allows more viral entry into host cells. Previous VOCs (Alpha, Beta, and Gamma) have shown enhanced ACE2 binding compared to the WT RBD. 10 Location of the two amino acid mutations L452R and T478K in the Delta variant RBD with respect to ACE2 binding interface is shown in Figure   3A . None of the two mutations are part of the ACE2 interface. Figure 4 shows isothermal titration calorimetry (ITC) thermograms for the WT RBD and Delta mutants, and Table 1 lists the average fit parameters from three independent batches of protein expression. WT RBD binding to ACE2 shows a typical exothermic reaction with a K d of 10.0 ± 3.1 nM and a ΔH value of -11.8 ± 0.2 kcal/mol. None of the Delta mutations significantly altered the binding affinity of RBD for ACE2. Both L452R and T478K mutants and the Delta double mutant L452R/T478K displayed similar K d values of 6.2 ± 3.7 nM, 17.1 ± 4.8 nM, and 10.8 ± 3.3 nM, respectively, similar to that of the WT RBD ( Table 1) .

Clinical observations associated with Delta variant could be related to SARS-CoV-2 escaping the human immune system. Neutralizing antibodies against SARS-CoV-2 RBD have been found to belong to four major classes depending on the mechanism of action and the location of their epitopes on the RBD. 48 ,49 SARS-CoV-2 spike protein is a trimer in its native state and exists in multiple conformations, mainly RBD in "up" position that is accessible for binding to ACE2 or in "down" position in which RBD is buried and not accessible for ACE2 binding. [50] [51] [52] Class 1 antibodies bind to RBD in the up conformation and compete with ACE2 binding. Class 2 antibodies bind to RBD both in the up or down conformations, and their epitope partially overlaps with the ACE2 binding site and hence compete against ACE2 binding. Class 3 antibodies bind to RBD in both up or down positions with their epitope on RBD far away from ACE2 binding site, and hence can neutralize the RBD through an uncompetitive mechanism.

Class 4 antibodies are relatively rare as their epitope is close to the hinge region connecting RBD to the rest of the spike protein, which is relatively buried compared to other epitopes, and none of the VOCs contain mutations in this region. Since the first step in neutralization is binding of neutralizing antibodies to RBD, we examined how the Delta mutations affect RBD binding to the three major classes of antibodies.

One of the first Class 1 antibodies that was identified from patients recovered from WT SARS-CoV-2 infection was CC12.1. 53 Location of the two Delta mutants L452 and T478 in RBD with respect to the CC12.1 binding interface is shown in Figure 3B . Binding of WT RBD and its Delta mutants to CC12.1 single chain variable fragment (ScFv) was measured using ITC ( Figure 5) , and the average thermodynamic parameters obtained from fitting the data from three independent batches of protein expression was included in Table 2 . All proteins showed similar exothermic binding profiles. WT RBD binds to CC12.1 with a K d of 23.9 ± 5.7 nM. Both Delta single mutants L452R and T478K as well as the double mutant L452R/T478K either bind with a similar affinity (similar K d value for T478K) or with a higher affinity (lower K d values for L452R and L452R/T478K), implying that the Delta variant RBD does not escape from Class 1 antibodies. ITC binding data is also consistent with the location of these two residues L452 and T478 with respect to the epitope of CC12.1 on RBD ( Figure 3B ). The two mutations in the Delta variant RBD are located far from the CC12.1 epitope, and do not affect the interactions between RBD and CC12.1. 54, 55 As of January 2022, FDA approved four therapeutic antibodies for emergency use authorization (EUA): Eli Lilly's LY-CoV016 (Etesevimab) and LY-CoV555 (Bamlanivimab), and Regeneron's REGN10933 (Casirivimab) and REGN10987 (Imdevimab). LY-CoV016 and REGN10933 are Class 1 antibodies whose epitopes are very similar on RBD. 56,57 LY-CoV555 is a Class 2 antibody, whereas REGN10987 is a Class 3 antibody. To determine whether Delta variant escapes FDA-approved Class 1 antibodies, we tested WT RBD and its Delta mutants binding to LY-CoV016 in ScFv format. Figure 3C shows the location of the two residues L452 and T478 with respect to the RBD interface with LY-CoV016. Figure 6 shows the ITC binding data for the single and double mutants. All interactions displayed exothermic binding profiles. Table 3 lists the average parameters obtained from fitting the ITC data from three independent batches of protein expression to a one-site binding model. WT RBD binds to LY-CoV016 with a K d of 49.3 ± 10.1 nM. Both single mutants L452R and T478K and the double mutant L452R/T478K bind to LY-CoV016 with a stronger affinity, implying that none of the Delta mutations escape LY-CoV016. This is consistent with the location of the two residues L452 and T478 with respect to the RBD binding interface with LY-CoV016 ( Figure 3C) , and also consistent with binding of Delta mutants to another Class 1 antibody CC12.1 described above ( Figure 5 and Table 2 ).

While the Delta variant does not evade Class 1 antibodies, a large amount of clinical data suggests that neutralizing antibodies discovered against WT SARS-CoV-2 are not effective in neutralizing emerging VOCs. 58 Similar to CC12.1, P2B-2F6 is one of the first neutralizing antibodies discovered in recovered COVID-19 patients. 59,60 P2B-2F6 is a Class 2 antibody. 48, 49 We examined whether Delta variant escapes from Class 2 antibodies by determining the binding of WT RBD and its Delta mutants to P2B-2F6 ScFv. Location of the two residues L452 and T478 in RBD with respect to its binding interface with P2B-2F6 is shown in Figure 3D . Figure   7 shows the ITC binding curves and Table 4 lists the thermodynamic parameters obtained from fitting ITC data from three independent batches of protein expression to a one-site binding model. Both WT RBD and T478K show similar K d values of 81.4 ± 6.2 nM and 80.5 ± 6.3 nM, respectively. However, L452R resulted in a complete loss of binding (Figure 7) . Similar results were observed for the Delta double mutant L452R/T478K. These results indicate that the Delta variant escapes Class 2 antibodies, and the escape is determined by the L452R mutation. Our results on the effect of L452R mutation on RBD binding to P2B-2F6 are also consistent with a deep mutational scanning analysis that indicates L452R as one of the possible immune-escaping hotspots. 61 We further examined whether Delta variant escapes FDA-approved Class 2 antibody LY-CoV555 62,63 . Location of the two residues L452 and T478 in RBD with respect to its interface with LY-CoV555 is shown in Figure 3E . Figure 8 shows the ITC binding curves for WT RBD and its Delta mutants, and Table 5 lists the mean thermodynamic parameters obtained from fitting ITC data from three independent batches of protein expression to one-site binding model.

WT RBD binds to LY-CoV555 ScFv with a K d of 3.8 ± 1.9 nM. T478K mutant shows a similar binding affinity to LY-CoV555 with a K d of 9.0 ± 4.6 nM. This was however not the case for the L452R mutation and the Delta double mutant L452R/T478K. Both showed no binding to LY-CoV555 (Figure 8) . These results indicate that Delta variant escapes Class 2 antibodies and is determined by the L452R mutation, which is consistent with escape from another Class 2 antibody P2B-2F6 described above (Figure 7 and Table 4 ).

We also examined whether Delta variant escapes Class 3 antibodies. FDA-approved antibody therapeutics contain a Class 3 antibody REGN10987. Location of the two residues L452 and T478 in RBD with respect to its interface with REGN10987 is shown in Figure 3F . Figure 9 shows the ITC binding curves of WT RBD and its Delta mutants, and Table 6 lists the mean thermodynamic parameters obtained from fitting ITC data from three independent batches of protein expression to a one-site binding model. Both WT RBD and T478K mutant showed similar binding affinity with K d values of 34.3 ± 8.1 nM and 15.9 ± 1.9 nM, respectively. This was not the case for both the L452R mutant and the Delta double mutant L452R/T478K. Both proteins showed a ~100 fold weaker binding affinity with K d values of 1,340 ± 100 nM for the L452R mutant and 1,150 ± 100 nM for the Delta double mutant ( Table 6 ). These results indicate that the Delta variant escapes Class 3 antibodies, and the escape is determined by the L452R mutation.

SARS-CoV-2 Delta variant has adapted unlike any other previous VOCs (Alpha, Beta, and Gamma). It is the VOC which is responsible for more severe symptoms and the maximum mortality among infected patients compared to other VOCs including Omicron. Deep mutational scanning has been able to predict that the previous VOCs had a high probability of becoming dominant strains, yet these studies were unable to predict the two novel mutations in the Delta variant with high probability. None of the two mutations L452R and T478K were part of Alpha, Beta, and Gamma VOCs, and only T478K (and not L452R) is present in the recently discovered Omicron VOC. In this manuscript, we examined the effect of these two novel mutations on the biophysical fitness landscape of Delta variant RBD.

Delta VOC displays unique biophysical characteristics unlike the previous VOCs Alpha, Beta, and Gamma. Biophysical data on Omicron VOC is not yet available, and hence we will be comparing the results on Delta VOC with other VOCs discovered prior to Delta. Table 7 lists the summary of biophysical parameters we have examined. Delta mutations do not significantly alter the binding affinity of RBD towards the ACE2 receptor (Figure 4 & Table 1 ). While a common belief is that VOCs should result in increased binding to ACE2, which would correlate with increased viral entry, data on Delta variant shows that this cannot be a ubiquitous thought. VOC having no effect on ACE2 binding affinity is unique to the Delta variant, as all previously VOCs showed increased affinity to ACE2. 10,34 This is primarily because of the presence of N501Y mutation in previous VOCs that is responsible for increased ACE2 binding. 35 Figure 3A) . ACE2 binding has been one of the most common factors when attempting to predict emerging VOCs, 12 which would explain why the two novel RBD mutations that resulted in the Delta variant have not been predicted by earlier studies. Thus, it is important to consider a more robust system for predicting variants that is not so heavily weighted towards ACE2 binding, as our results show that immune escape rather than receptor binding determines the fitness of Delta VOC. Since the full-length spike protein exists in multiple conformations with RBDs in up or down positions, 13,66 measured binding affinity for isolated RBD towards ACE2

represents the upper value of the binding affinity. Any conformation with RBD in down position in equilibrium will only decrease the relative population of RBDs in up conformation, and hence will result in decreased affinity of the complete spike protein towards ACE2.

Compared to other VOCs, patients contracted with Delta have increased viral titers. 15, 67 Results show that the two Delta mutations did not affect either the secondary (far-UV CD; Figure 2B ) or tertiary structure (intrinsic protein fluorescence of aromatic sidechains; Figure   2C ) of the RBD. This is consistent with the three-dimensional structural alignment of the two RBDs using the Multiprot program. 68 Delta RBD (PDB ID: 7v8b) has similar structure as that of WT RBD (PDB ID: 6m0j) with an RMSD of 0.82 Å. Additionally, the two mutations L452R and T478K did not alter the stability of the RBD when measured by both thermal denaturation ( Figure S1 & Table S1 ) or urea denaturation ( Figure S2 & Table S2 ) experiments. However, Delta RBD showed ~70% higher expression in human Expi293 (modified HEK293) cells compared to the WT RBD (Figure 1) . Similar high expression of RBD was not seen in the case of Alpha, Beta, and Gamma VOCs. 10 Increased expression of RBD is determined by the L452R mutation (Figure 1 and Table 7) , and this mutation is not present in other VOCs. One single mutation increasing the protein expression by 70% is very rare in protein literature in general.

Increased expression of viral proteins can lead to increased viral titers, although whether L452R results in increased expression of the complete spike protein needs to be examined. If that is the case, relative expression levels of viral protein mutants need to be considered as a criterion in predicting the emergence of future VOCs.

Another crucial factor to consider when evaluating viral fitness is the ability of mutations to escape neutralizing antibodies generated by human immune system in response to WT SARS-CoV-2 infection, vaccination, or those authorized by FDA for treating infected patients.

Determining whether the variants escape FDA-approved antibodies, which are derived either from patients recovered from WT SARS-CoV-2 infection (Eli Lilly) or from humanized mice models (Regeneron), will tell us whether the current therapies work against the emerging variants or new antibody therapies have to be developed. Neutralizing antibodies have been broadly classified into different classes depending on the location of their epitopes on RBD. 48, 49 Locations of mutations in the structures of RBD-antibody complexes (Figure 3 ) and analyzing the stabilizing interactions in which they participate can sometimes predict which antibodies the VOCs can escape. However, experiments have to confirm such predictions based on protein structure because long-range mutation effects and the dynamics of various protein regions can also play a critical role in protein function. The two mutations L452R and T478K are far away from the binding interface of two Class 1 antibodies we examined (CC12.1 ( Figure 3B ) & FDAapproved Eli Lilly Class 1 antibody LY-CoV016 (Figure 3C) ), and both mutations are not part of the RBD interface with Class 1 antibodies. Consistently, Delta mutants did not escape Class 1 antibodies (Figure 5, Figure 6 , Table 2, Table 3 & Table 7 ). In contrast, L452 stabilizes the RBD interactions with Class 2 antibodies by forming a hydrophobic cluster with I103 and V105 of the variable heavy chain of P2B-2F6 and I54 and L55 residues of the variable heavy chain of LY-CoV555. Replacing the hydrophobic residue leucine with a positively charged arginine in the middle of these hydrophobic clusters is expected to destabilize the RBD interactions with Class 2 antibodies. Accordingly, after L452R mutation, RBD does not bind to Class 2 antibodies (Figure 7, Figure 8 , Table 4, Table 5 & Table 7 ). In the case of RBD binding to Class 3 antibodies (Figure 3F) , L452 does not form any direct contacts with the antibody REGN10987, but neighbors the epitope composed of residue 450 in RBD. 69, 70 Change of a hydrophobic amino acid leucine with a positively charged residue arginine next to the epitope is expected to change the electrostatic nature of RBD interface with REGN10987, and thus might explain the decrease in binding affinity by ~100-fold upon L452R mutation (Figure 9 , Table 6 & Table 7 ). T478 residue is far away from any of the binding interfaces and do not participate in any stabilizing inter-molecular interactions, and hence do not contribute to the immune escape potential of the Delta variant (Table 7) .

Our results indicate that L452R mutation determines the antigenic drift of the Delta VOC against Class 2 and Class 3 antibodies. Since L452R also occurs in other variants such as Epsilon, Iota, and Kappa, 71 we expect these variants also escape Class 2 and Class 3 antibodies.

Despite the number of infected cases with T478K mutation are increasing, 72, 73 its role in the viral fitness is quite elusive. Our results indicate that it does not offer any fitness advantage to SARS-CoV-2 RBD in terms of receptor binding or immune escape from neutralizing antibodies compared to the WT. Some studies have suggested that T478K mutation might have evolved from previously infected Beta VOC patients, since it escapes an antibody specific to the Beta VOC. 74 In addition, the newest Omicron VOC contains S477N mutation in addition to T478K mutation, and whether T478K mutation enhances the antibody escape determined by S477N needs to be examined. Further, our study has examined the role of T478K mutation on the biophysical properties of RBD, and its specific role needs to be examined in the context of the full-length spike protein, particularly in controlling the up/down conformations of the spike protein and long-range allosteric effects on other important structural regions of the spike protein.

Results presented here indicate that the Delta variant has evolved towards escape from Class 2 and Class 3 antibodies, rather than enhancing the receptor binding or escape from Class 1 antibodies. Class 1 antibodies bind to RBD only in up conformation where RBD is accessible to ACE2 binding, whereas Class 2 and Class 3 antibodies bind to RBD irrespective of whether it is in up conformation (accessible to ACE2) or down conformation (inaccessible to ACE2). 49 Escape from Class 2 antibodies mainly contributes to escape from polyclonal plasma, 56 (Table 7) . However, Eli Lilly's Class 2 antibody LY-CoV555 will be completely ineffective in neutralizing the Delta variant as L452R completely abolished RBD binding to Class 2 antibodies, whereas Regeneron's Class 3 antibody REGN10987 will be much less effective and requires much higher concentration as the RBD binding affinity is reduced by ~100 fold upon Delta mutations ( Table 7) .

The immune escape and high expression capabilities of the SARS-CoV-2 Delta variant requires a necessity for robust therapeutic options. As the virus adapts, every VOC has shown increased immune escape potential. Thus, it is necessary for the vaccination rates to continue to rise in order to combat the emergence of future VOCs. Simultaneously, it is necessary for improved monoclonal antibody therapeutics to be developed against future variants. Delta VOC is clearly distinct from other VOCs. Our previous work has shown that other VOCs can escape Class 1 antibodies, while expression is not significantly different from that of the unmutated WT. 10 Delta does not escape Class 1 antibodies and shows higher protein expression. In addition, all previously studied variants showed enhanced ACE2 binding which is not the case for the Delta variant. These results also point to the fact that the virus is still under continuous evolution, as none of the VOCs is still able to escape all classes of neutralizing antibodies. Any combination of mutations that confer immune escape potential to SARS-CoV-2 against all classes of neutralizing antibodies will be of a major concern. In terms of applicability of our results to the new Omicron variant, Omicron might still show higher ACE2 binding because of the presence of N501Y mutation. Since Omicron lacks L452R mutation, it might not escape Class 2 and Class 3 antibodies, unless a separate set of amino acid mutations confer the ability to escape these neutralizing antibodies. Since Omicron is the most transmissible and Delta is the most virulent out of the five known VOCs, if Omicron picks up the critical Delta mutations resulting in a Deltacron variant, it could turn out to be a more dangerous variant. Note added during Revision: As predicted above, recent results show that Omicron VOC has higher ACE2 binding. 76 It escapes Class 2 and Class 3 neutralizing antibodies, 74, 77 and the escape has to be due to a different set of mutations since Omicron lacks L452R mutation.

Sequences for RBD and human ACE2 were obtained from Uniprot (RBD ID: P0DTC2 hACE2 ID: Q9BYF1). Sequences for neutralizing antibodies were obtained from Research 

An Applied Photophysics Chirascan Plus spectrometer was used to record the CD spectra and chemical denaturation melts for all variants. CD Spectra were obtained for each RBD variant from 190 nm to 260 nm. Protein spectra were recorded with a 0.5 mm cuvette at 5 µM protein concentration in a buffer consisting of 10 mM sodium phosphate, 4 mM NaCl, pH 7.0. Spectra were recorded every 1 nm wavelength and averaged over 2 seconds. Runs were repeated 5 times and averaged.

Thermal denaturation melts were performed using an Applied Photophysics Chirascan Plus spectrometer. All experiments were performed in a 0.5 mm cuvette at a protein concentration of 20 µM in buffer containing 50 mM sodium phosphate, 20 mM NaCl, pH 7.0.

Temperature scan rate was 1°C/min, sample was equilibrated for 30 sec at each temperature increment, and the CD signal at 222 nm was averaged over 2 seconds. Data was plotted and analyzed using the equation 47 (Eq. 1) =

where S T is the measured signal as a function of temperature T, S N and S U are the signals corresponding to the native and unfolded baselines, m N and m U are the slopes of linear dependence of S N and S U with temperature, T m is the midpoint melting temperature, ΔH m is the enthalpy change at the T m , R is the universal gas constant, and T is the absolute temperature in Kelvin, respectively.

A CCD detector was used with the Applied Photophysics Chirascan Plus spectrometer to record the chemical denaturation spectra. An excitation wavelength of 280 nm was used, and fluorescence emission was recorded in a 1 cm cuvette at 2 µM protein concentration in a buffer containing 50 mM sodium phosphate, 20 mM NaCl, pH 7.0. Urea was utilized as the denaturant for equilibrium protein unfolding measurements, and was dissolved in 50 mM sodium phosphate, refractive index measurements. 78 

In order to determine the significance of differences between the biophysical properties of WT RBD and mutant proteins, an unpaired t-test was used. Statistical analyses were ScFv. Top panels represent the raw differential power vs. time thermographs, while bottom panels represent the integrated heat plots. 

 Delta variant RBD shows higher protein expression compared to the wild-type.

 Delta variant RBD does not affect ACE2 receptor binding affinity compared to the wild-type.

 Delta variant RBD does not escape Class 1 antibodies.

 Delta variant RBD escapes Class 2 and Class 3 antibodies.

 Fitness landscape of the Delta variant RBD is determined by the L452R mutation.

☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: WT 81.4 ± 6.2 0.9 ± 0.0 -8.0 ± 0.5 -9.5 ± 0.1 -1.5 ± 0.5

No binding ----T478K 80.5 ± 6.3 0.9 ± 0.0 -8.7 ± 0.4 -9.5 ± 0.0 -0.8 ± 0.4

No binding ---- Table 5 . Thermodynamic parameters of WT RBD and its Delta mutants binding to FDAapproved Class 2 therapeutic antibody LY-CoV555 ScFv. Statistical analyses of mutant properties with respect to the WT were performed using unpaired ttest, and P-values less than 0.05 were considered significant. Superscripts indicate *P < 0.05, **P < 0.01, ***P < 0.001, and no superscript indicates not a significant difference between the mutant and the WT.

(+68%) 0.2 3.3 *** ***

Engineering human ACE2 to optimize binding to the spike protein of SARS coronavirus 2

Effect of mutation on structure, function and dynamics of receptor binding domain of human SARS-CoV-2 with host cell receptor ACE2: a molecular dynamics simulations study

Structural variations in human ACE2 may influence its binding with SARS-CoV-2 spike protein

Structural and energetic profiling of SARS-CoV-2 receptor binding 11

Spike protein recognition of mammalian ACE2 predicts the host range and an optimized ACE2 for SARS-CoV-2 infection

SARS-CoV-2 variant prediction and antiviral drug design are enabled by RBD in vitro evolution

Molecular basis of immune evasion by the Delta and Kappa SARS-CoV-2 variants

SARS-CoV-2 Spike Mutations, L452R, T478K, E484Q and P681R

SARS-CoV-2 B.1.617.2 Delta variant replication and immune evasion

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

Predominance of delta variant among the COVID-19 vaccinated and unvaccinated individuals

Transmission event of SARS-CoV-2 delta variant reveals multiple vaccine breakthrough infections

Differences in the Concentration of Anti-SARS-CoV-2 IgG Antibodies

Recovery or Post-Vaccination

Effectiveness of Covid-19 Vaccines against the B.1.617.2 (Delta) Variant

Quantifying the Evolutionary Constraints and Potential of Hepatitis C Virus NS5A Protein. mSystems

Mutational fitness landscapes reveal genetic and structural improvement pathways for a vaccine-elicited HIV-1 broadly neutralizing antibody

Structural and functional ramifications of antigenic drift in recent SARS-CoV-2 variants

Structural analysis of receptor binding domain mutations in SARS-CoV-2 variants of concern that modulate ACE2 and antibody binding

Effects of common mutations in the SARS-CoV-2 Spike RBD and its ligand, the human ACE2 receptor on binding affinity and kinetics

N501Y mutation of spike protein in SARS-CoV-2 strengthens its binding to receptor ACE2

SARS-CoV-2 infects the human kidney and drives fibrosis in kidney organoids

Cross-validation of SARS-CoV-2 responses in kidney organoids and clinical populations

The Nuts and Bolts of SARS-CoV-2 Spike Receptor-Binding Domain Heterologous Expression

Stabilization of the SARS-CoV-2 Spike Receptor-Binding Domain Using Deep Mutational Scanning and Structure-Based Design

Structural and functional comparison of SARS-CoV-2-spike receptor binding domain produced in Pichia pastoris and mammalian cells

Characterizations of SARS-CoV-2 mutational profile, spike protein stability and viral transmission

Using circular dichroism collected as a function of temperature to determine the thermodynamics of protein unfolding and binding interactions

Neutralizing monoclonal antibodies for treatment of COVID-19

SARS-CoV-2 neutralizing antibody structures inform therapeutic strategies

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

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

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

Isolation of potent SARS-CoV-2 neutralizing antibodies and protection from disease in a small animal model

Specific epitopes form extensive hydrogen-bonding networks to ensure efficient antibody binding of SARS-CoV-2: Implications for advanced antibody design

Structural basis of a shared antibody response to SARS-CoV-2

Mapping mutations to the SARS-CoV-2 RBD that escape binding by different classes of antibodies

A broadly cross-reactive antibody neutralizes and protects against sarbecovirus challenge in mice

Structural Basis of SARS-CoV-2 and SARS-CoV Antibody Interactions

Human neutralizing antibodies elicited by SARS-CoV-2 infection

Comprehensive Deep Mutational Scanning Reveals the Immune-Escaping Hotspots of SARS-CoV-2 Receptor-Binding Domain Targeting Neutralizing Antibodies

Complete map of SARS-CoV-2 RBD mutations that escape the monoclonal antibody LY-CoV555 and its cocktail with LY-CoV016

The new SARS-CoV-2 strain shows a stronger binding affinity to ACE2 due to N501Y mutant

Higher infectivity of the SARS-CoV-2 new variants is associated with K417N/T, E484K, and N501Y mutants: An insight from structural data

Effect of natural mutations of SARS-CoV-2 on spike structure, conformation, and antigenicity

Viral infection and transmission in a large

BeStSel: a web server for accurate protein secondary structure prediction and fold recognition from the circular dichroism spectra

Prospective mapping of viral mutations that escape antibodies used to treat COVID-19

Studies in humanized mice and convalescent humans yield a SARS-CoV-2 antibody cocktail

Acquisition of the L452R Mutation in the ACE2-Binding Interface of Spike Protein Triggers Recent Massive Expansion of SARS-CoV-2 Variants

Preliminary report on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Spike mutation T478K

SARS-CoV-2 Delta (B.1.617.2) Variant: A Unique T478K Mutation in Receptor Binding Motif

SARS-CoV-2 Omicron-B.1.1.529 leads to widespread escape from neutralizing antibody responses

An Alternative Binding Mode of IGHV3-53 Antibodies to the SARS-CoV

SARS-CoV-2 Omicron variant: Antibody evasion and cryo-EM structure of spike protein-ACE2 complex

An infectious SARS-CoV-2 B.1.1.529 Omicron virus escapes neutralization by therapeutic monoclonal antibodies

Determination and analysis of urea and guanidine hydrochloride denaturation curves

On the Refractive Indices of Aqueous Solutions of Urea

Unfolding free energy changes determined by the linear extrapolation method. 1. Unfolding of phenylmethanesulfonyl. alpha.-chymotrypsin using different denaturants

Data reduction and error analysis

Conceptualization, Methodology, Validation, Formal analysis, Investigation, Data Curation, Writing -Original Draft, Writing -Review & Editing, Visualization Vaibhav Upadhyay: Conceptualization, Methodology, Resources, Writing -Review & Editing, Visualization Alexandra Lucas: Conceptualization, Methodology, Writing -Review & Editing

Writing -Review & Editing, Supervision, Project administration

We thank David Bain and Carlos Catalano for their critical reading of the manuscript and helpful comments. This work was supported by an Associate Dean of Research seed grant program from the University of Colorado Skaggs School of Pharmacy and Pharmaceutical Sciences.