SARS-CoV-2 Delta Variant Displays Moderate Resistance to Neutralizing Antibodies and Spike Protein Properties of Higher Soluble ACE2 Sensitivity, Enhanced Cleavage and Fusogenic Activity

The SARS-CoV-2 B.1.617 lineage variants, Kappa (B.1.617.1) and Delta (B.1.617.2, AY) emerged during the second wave of infections in India, but the Delta variants have become dominant worldwide and continue to evolve. Here, we compared B.1.617 variants for neutralization resistance by convalescent sera, mRNA vaccine-elicited sera, and therapeutic neutralizing antibodies using a pseudovirus neutralization assay. B.1.617.1, B.1.617.2, and AY.1 pseudoviruses showed a modest 1.5- to 4.4-fold reduction in neutralization by convalescent sera and vaccine-elicited sera. In comparison, similar modest reductions were also observed for C.37, P.1, R.1, and B.1.526 pseudoviruses, but 7- and 16-fold reductions for vaccine-elicited and convalescent sera, respectively, were seen for B.1.351 pseudoviruses. Among twenty-three therapeutic antibodies tested, four antibodies showed either complete or partial loss of neutralization against B.1.617.2 pseudoviruses and six antibodies showed either complete or partial loss of neutralization against B.1.617.1 and AY.1 pseudoviruses. Our results indicate that the current mRNA-based vaccines will likely remain effective in protecting against B.1.617 variants. Finally, the P681R substitution confers efficient cleavage of B.1.617 variants’ spike proteins and the spike of Delta variants exhibited greater sensitivity to soluble ACE2 neutralization, as well as fusogenic activity, which may contribute to enhanced spread of Delta variants.


Introduction
Since its origin in December 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread globally to cause a coronavirus disease 2019 (COVID- 19) pandemic that recorded more than 263 million infections and has claimed 5. The global dominance and ongoing evolution of B.1.617 lineage variants require continuing assessment of the neutralization potency of convalescent sera, vaccine-elicited sera, and therapeutic neutralizing antibodies against emerging B.1.617 variants. Here, we measured the neutralization potency of convalescent sera, vaccine-elicited sera, and therapeutic neutralizing antibodies against two independent variants each in the Kappa (B.1.617.1) and Delta (B.1.617.2, AY.1) lineages and assessed the contribution of the RBD substitutions in conferring resistance. We found that resistance to convalescent and vaccine-elicited sera was predominantly conferred by RBD substitutions E484Q, T478K, and L452R. Furthermore, out of 23 therapeutic neutralizing antibodies tested, Kappa and Delta pseudoviruses displayed complete resistance to five neutralizing antibodies and partial resistance to one antibody. Finally, we showed that the P681R substitution confers enhanced furin processing in spike protein of B.1.617 lineage variants that corresponded to enhanced cell-cell fusion activity. However, only Delta spike protein exhibited greater sensitivity to soluble ACE2 inhibition, implying enhanced ACE2 affinity. This feature along with enhanced cell-cell fusion activity may contribute to the dominance of the B.1.617.2 variant.

Ethics Statement
Use of de-identified sera in this study was approved by the U.S. Food and Drug Administration Research in Human Subjects Committee. Vaccine-elicited sera were collected at the U.S. Food and Drug Administration with written consent under an approved Institutional Review Board (IRB) protocol (FDA IRB Study # 2021-CBER-045).

Plasmids and Cell Lines
Codon-optimized full-length open reading frames of the S genes of SARS-CoV-2 variants were cloned into pcDNA3.1(+) or pVRC8400 by GenScript (Piscataway, NJ, USA). The codon optimization for gene expression in human cell was performed by GenScript's OptimumGene algorithm system to optimize the following parameters: codon usage bias, GC content, CpG dinucleotides content, mRNA secondary structure, internal chi sites and ribosomal binding sites, negative CpG islands, cryptic splicing sites, premature PolyA sites, PolyT sites, RNA instability motif (ARE), and repeat sequences (direct repeat, reverse repeat). The codon-optimized sequence of Wuhan-Hu-1 S gene is shown in Supplementary Figure S1. The spikes of variants used in this study are listed in Table 1. The HIV gag/pol (pCMV∆R8.2), and luciferase reporter (pHR'CMV-Luc) plasmids were obtained from the Vaccine Research Center (National Institutes of Health, Bethesda, MD, USA) [31,32]. 293T-ACE2.TMPRSS2 cells stably expressing human angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease 2 (TMPRSS2) were previously described (BEI Resources, Manassas, VA, USA; Cat no: NR-55293) [33]. The 293T and 293T-ACE2.TMPRSS2 cells were maintained at 37 • C in Dulbecco's modified eagle medium (DMEM) supplemented with high glucose, L-glutamine, minimal essential media (MEM) non-essential amino acids, penicillin/streptomycin, and 10% fetal bovine serum (FBS).

SARS-CoV-2 Pseudovirus Production and Neutralization Assay
HIV-based lentiviral pseudoviruses with spike proteins were generated as previously described [33]. The B.1 spike containing D614G was used as wild type (WT(D614G)). Briefly, pseudoviruses bearing the spike glycoprotein and packaging a firefly luciferase (FLuc) reporter gene were produced in 293T cells by co-transfection of 5 µg of pCMV∆R8.2, 5 µg of pHR'CMVLuc and 4 µg of pcDNA3.1(+) or 0.5 µg of pVRC8400 encoding a codon-optimized spike gene with the desired substitutions. Pseudovirus supernatants were collected approximately 48 h post transfection, filtered through a 0.45 µm low protein binding filter, and stored at −80 • C. Neutralization assays were performed using 293T-ACE2.TMPRSS2 cells in 96-well plates as previously described [33]. Briefly, pseudoviruses with titers of approximately 1 × 10 6 relative luminescence units (RLU)/mL of luciferase activity were incubated with serially diluted sera or antibodies for two h at 37 • C. Pseudovirus and serum or antibody mixtures (100 µL) were then inoculated onto the plates pre-seeded one day earlier with 3.0 × 10 4 cells/well. Pseudovirus infectivity was scored 48 h later for luciferase activity. The antibody concentration or inverse of the sera dilutions causing a 50% reduction of RLU compared to control was reported as the neutralization titer. Titers were calculated using a nonlinear regression curve fit (GraphPad Prism Software Inc., La Jolla, CA, USA). The mean titer from at least two independent experiments each with intra-assay duplicates was reported as the final titer. WT(D614G) pseudovirus was run as a control for every assay.
For ACE2 neutralization assay, serially diluted recombinant human soluble ACE2 was incubated with indicated pseudovirus (~1 × 10 6 RLU/mL) for one hour at 37 • C and 100 µL of pseudovirus and soluble ACE2 mixture was added to 293T-ACE2.TMPRSS2 cells. Luciferase activity was measured 48 h later. The soluble ACE2 concentration causing a 50% reduction of RLU compared to control was reported as the 50% inhibitory concentration or IC 50 .

Western Blotting
In total, 1.25 mL of pseudoviruses was pelleted at 4 • C for 2 h at 15,000 rpm using Tomy TX-160 ultracentrifuge. Pseudovirus pellet was resuspended in 1× Laemmli loading buffer and heated at 70 • C for 10 min. Samples were resolved by 4-20% SDS PAGE and transferred onto nitrocellulose membranes. Membranes were probed for SARS-CoV-2 S1 using rabbit polyclonal antibodies against the RBD domain (Sino Biological, Wayne, PA, USA).

Cell-Cell Fusion Assay
For measuring spike-protein-mediated cell-cell fusion, β-gal complementation assay was performed as described previously [35]. Briefly, β-gal ω subunit-expressing 293T cells were transfected with 1 µg spike plasmids, whereas 293T-ACE2.TMPRSS2 cells were transfected with β-gal α subunit using the Fugene 6 reagent. At 24 h post transfection, the transfected cells were detached using a nonenzymatic cell dissociation solution (Sigma, St. Louis, MO, USA) and washed with DMEM. Spike-transfected/β-gal ω subunit-expressing 293T cells were mixed with β-gal α subunit-transfected/293T-ACE2.TMPRSS2 cells at 1:1 ratio to a total of 6 × 10 4 cells per well on a 96-well plate. The cells were co-cultivated for 24 h at 37 • C. The culture supernatants were then removed, and cell-cell fusion was scored by determination of the β-gal activity in co-cultured cell lysates using a Galacto-Star kit (Applied Biosystems, Waltham, MA, USA) according to the manufacturer's instructions.
To ensure equivalent amount of spike cell surface expression levels among treatments, spike-transfected β-gal ω subunit-expressing 293T cells were quantified for cell surface Viruses 2021, 13, 2485 6 of 20 spike levels by flow cytometry. Spike-transfected 293T cells employed in cell-cell fusion assay were concurrently stained with SARS-CoV-2 positive human polyclonal sera at 1:20 dilution, washed twice, and then incubated with FITC-conjugated goat anti-human (KPL Inc., Gaithersburg, MD, USA). The cells were washed twice and then fixed with 2% paraformaldehyde. The results were acquired using BD LSRFortessa™ X-20 Cell Analyzer (BD Biosciences, San Jose, CA, USA). The mean fluorescence intensities of spike positive cells were recorded.

Antigenic Cartography
We created a geometric interpretation of neutralization titers against the tested SARS-CoV-2 pseudoviruses using Racmacs antigenic cartography software (Sam Wilks (2021), Racmacs: R Antigenic Cartography Macros. R package version 1.1.16. https://github.com/ acorg/Racmacs, accessed on: 15 October 2021) [36,37]. The map is presented on a grid in which each square indicates one antigenic unit, corresponding to a two-fold dilution of the antibody in the neutralization assay. Antigenic distance is measured in any direction on the map.

Furin Prediction Score Calculations
The prediction of furin-specific cleavage site in spike proteins was computed using the ProP 1.0 Server hosted at http://www.cbs.dtu.dk/services/ProP/ (accessed on: 15 July 2021) and the PiTou V3 software hosted at http://www.nuolan.net/reference.html (accessed on: 15 July 2021).

Statistics Analysis
One-way analysis of variance (ANOVA) with Dunnett's multiple comparisons tests (variants compared to WT(D614G)), Mann-Whitney test for the comparison of two groups with unmatched pairs (Pfizer BNT162b2 compared to Moderna) and geometric mean titers (GMT) with 95% confidence intervals were performed using GraphPad Prism software. The p values of less than 0.05 were considered statistically significant. All neutralization titers were log 2 transformed for analyses.
Because the convalescent sera came from individuals who were previously infected by different variants (Table 2), we also explored differences in the neutralization titers between those infected by D614G variants lacking L452R and those infected by variants containing L452R (all have L452R and D614G, except one lacking D614G). The small number of samples in each group precludes conclusions, but we noticed modest differences against some variants. In the group infected by D614G variants that lacked L452R, a 5.2-fold, 1.6-fold, and 2.5-fold reduction in cross-neutralization potency was seen against B.  [41]. The generally lower titers and small numbers of samples in the L452R group in our study may impact the fold changes. We also note that in our study the titers of the sera from WT(D614G)-infected individuals are generally higher than the neutralization titers of the sera from L452R-infected individuals, but the small number of samples prevents a meaningful comparison. Differences in titers may be due to the timing of collection of sera since infection, severity of infection, or other factors.

Neutralization of B.1.617 Variant Pseudoviruses by Vaccine-Elicited Sera
We next assessed the neutralization potency of mRNA vaccine-elicited sera against WT(D614G) and B.1.617 variant pseudoviruses. Sera from fourteen individuals who received two doses of Moderna mRNA-1273 vaccine and fifteen individuals who received two doses of Pfizer/BioNtech BNT162b2 vaccine were collected approximately two weeks after the second immunization. Each vaccine-elicited serum had high neutralization titers against WT(D614G) pseudoviruses, ranging between 578 and 3935 for Pfizer/BioNtech A prior study showed that convalescent sera and vaccine-elicited antibody neutralization titers against pseudoviruses bearing spikes containing L452R-E484Q-P681R substitutions displayed 2-5-fold reduction, compared to the neutralization titers against WT (D614G) pseudoviruses [25]. In this study, depending upon the infecting variant, convalescent sera displayed a modest 2-4-fold reduction in neutralization titers Overall, the trends in neutralization titers for the vaccine-elicited sera against a large panel of variant pseudoviruses were similar to those for convalescent sera, though the GMTs were approximately 3-to 5-fold higher for vaccine-elicited sera.
A prior study showed that convalescent sera and vaccine-elicited antibody neutralization titers against pseudoviruses bearing spikes containing L452R-E484Q-P681R substitutions displayed 2-5-fold reduction, compared to the neutralization titers against WT(D614G) pseudoviruses [25]. In this study, depending upon the infecting variant, convalescent sera displayed a modest 2-4-fold reduction in neutralization titers for B.1.617.1 compared to WT(D614G), while the vaccine-elicited sera displayed only 2.1-fold reduction. Our findings using B.1.617.1 pseudoviruses with the full complement of spike substitutions rather than just the L452R-E484Q-P681R spike substitutions confirm and extend the prior findings [25]. Finally, the 1.8-3.4-fold reduction in neutralization in the vaccination group against B.1.617.2 pseudoviruses seen in our study and others suggests that the two-dose mRNA vaccines could significantly contribute to protection against both Kappa and Delta variants [17,18,38,41,43].

Antigenic Cartography
We used antigenic cartography to compare the relative difference in neutralization of distinct variants, exploiting heterogeneity in individual antibody responses to identify antigenic differences among strains. Antigenic map dimensionality was tested in 1-5 dimensions using cross-validation. The 3D maps had only slightly greater predictive power compared to the 2D maps; both the 2D and 3D maps performed much better than 1D (Supplementary Table S1). The 2D antigenic maps are presented for convalescent infection sera ( Figure 3A) and vaccine sera ( Figure 3B). To evaluate reproducibility in positioning of antigens on the map, we created bootstrap confidence intervals in which n = 10,000 antigenic maps were made by resampling sera with replacement (bottom panels in Figure 3A,B). Antigenic maps are also shown in 3D (Supplementary Figure S3). Maps made with convalescent sera only from infection by WT(D614G) or only from infection by strains containing L452R mutation were not sufficiently robust for antigen positioning.
On the convalescent sera antigenic map, most of the pseudoviruses clustered close to the WT(D614G) pseudovirus and the WT(D614G) sera and L452R sera, including K417N, L452R, T478K, L452Q + F490S, and B.1.429 (0.25 to 0.72 antigenic units (AU) from WT (Table 3). B.1.617.2, C.37, and AY.1 clustered below WT(D614G) and slightly further away (1.14-1.60 AU), while L452R + T478K was also further away, but in the opposite direction (1.20 AU). However, in the bootstrap confidence interval 2D map and 3D map, L452R + T478K and AY. antigenic differences among strains. Antigenic map dimensionality was tested in 1-5 dimensions using cross-validation. The 3D maps had only slightly greater predictive power compared to the 2D maps; both the 2D and 3D maps performed much better than 1D (Supplementary Table S1). The 2D antigenic maps are presented for convalescent infection sera ( Figure 3A) and vaccine sera ( Figure 3B). To evaluate reproducibility in positioning of antigens on the map, we created bootstrap confidence intervals in which n = 10,000 antigenic maps were made by resampling sera with replacement (bottom panels in Figure 3A,B). Antigenic maps are also shown in 3D (Supplementary Figure S3). Maps made with convalescent sera only from infection by WT(D614G) or only from infection by strains containing L452R mutation were not sufficiently robust for antigen positioning.   The antigenic map of the same SARS-CoV-2 variant pseudoviruses for vaccine-elicited sera showed similar patterns to the convalescent sera map but with some notable differ-ences. Again, pseudoviruses K417N, L452R, T478K, L452Q + F490S, and B.1.429 were close to WT(D614G) (0.05-0.51 AU). However, E484Q and C.37 were both closer to WT(D614G) but were poorly coordinated and extended in elongated shapes around WT(D614G). B.1.617.1 was also slightly closer to WT(D614G) at 1.55 AU and was positioned adjacent to the other variants. Unexpectedly, AY.1 was between B.1.617.2 and B.1.351 and was further away from WT(D614G) (1.89 AU) than in the convalescent sera map (1.60 AU), being positioned only slightly closer to WT(D614G) than B.1.351 was from WT(D614G) (3.02 AU). These antigenic maps reinforce what was observed in terms of neutralizing antibody titers. Compared to the convalescent sera map, there was a larger antigenic difference between AY.1 and WT(D614G) on the vaccine-elicited sera map.
Notably, given that the vaccine-elicited sera had much higher titers than convalescent sera across variants, the titers against AY.1 were still higher in vaccinated individuals, meaning the antigenic difference may not translate into loss of vaccine protection. Furthermore, vaccine-elicited sera saw a smaller difference between B.1.351 and WT(D614G) than the convalescent sera, which may have aligned AY.1 and B.1.351 closer together. Overall, while these antigenic maps provide meaningful information on the relative positions of antigens, they are limited by the sera being so tightly clustered. Future antigenic maps with sera against distinct variants would enable more accurate evaluation of antigenic variation among the variants.

Spike RBD Substitutions in B.1.617 Variants Affect Sensitivity to Therapeutic Antibodies
We next evaluated 23 clinical-stage therapeutic neutralizing antibodies for potency against the B.1.617 variants. These antibodies were evaluated as part of the U.S. Government COVID-19 response effort to inform the clinical testing and use of these antibodies [19]. Due to an agreement with the manufacturers who provided the antibodies, only blind codes are used to identify the antibodies. Amongst the therapeutic antibodies tested were thirteen nAbs, six cnAbs, and four pAbs.

B.1.617 Pseudovirus Infectivity and Spike Protein Processing
Furin cleavage score analysis of CoV S1/S2 cleavage sites. CoV spike and influenza hemagglutinin (HA) sequences were analyzed using the ProP 1.0 and PiTou 3.0 furin prediction algorithms, generating a score with bold numbers indicating predicted furin cleavage. "|" denotes the position of the furin cleavage site. Sequences corresponding to the S1/S2 region of NL63 (APF29063.  Figure 5B). Our findings extend previous studies indicating that the P681R substitution increases spike proteolytic cleavage and facilitates cell-cell fusion [41,50,66,67]. Our findings also suggest that fusogenic potential of spike proteins may be influenced by both ACE2 binding affinity as well as proteolytic cleavage of spike.

Conclusions
Here we show that pseudoviruses bearing B.1.617.1 spike with L452R and E484Q substitutions, and B.1.617.2 spike with K417N, L452R and T478K substitutions, have modestly reduced susceptibility to neutralization by Pfizer/BioNtech BNT162b2 or Moderna mRNA-1273 vaccine-elicited sera and convalescent sera compared to pseudoviruses bearing WT(D614G) spike. The individual L452R, T478K, E484Q, and dual L452R + T478K substitutions accounted for most but not all of the reduction in neutralization potency of the sera, suggesting contributions from substitutions in the NTD/CTD. Neutralization titers, as well as antigenic maps, indicated that the full set of RBD substitutions in combination with substitutions outside the RBD contributes to antigenic differences of B.1.617.1, B.1.617.2, and C.37 variants. Antigenic distances between the variants also tended to be more spaced apart in the map generated by the convalescent sera compared to the vaccine-elicited sera. Limitations in our study include the small number of sera samples in the convalescent and vaccine cohorts. Potential differences in COVID-19 severity in the convalescent sera cohort and time of sera collection could also affect neutralization titers. Nonetheless, most sera from convalesced and vaccinated individuals neutralized the B.1.617.1, B.1.617.2, and AY.1 variants. Furthermore, 17 of 23 therapeutic neutralizing antibodies retained complete neutralization against B.1.617 variants. Resistance to the remaining therapeutic neutralizing antibodies is due to RBD substitutions, K417N, L452R, and E484Q, but not T478K. These findings suggest that the two-dose immunization with current mRNA vaccines will likely induce protective immunity against the tested B.1.617 variants. However, as B.1.617.2 variants continue to evolve, it will be important to continue to monitor how new substitutions in spike impact their resistance to therapeutic neutralizing antibodies and vaccine efficacy.