The E484K Substitution in a SARS-CoV-2 Spike Protein Subunit Vaccine Resulted in Limited Cross-Reactive Neutralizing Antibody Responses in Mice

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), especially emerging variants, poses an increased threat to global public health. The significant reduction in neutralization activity against the variants such as B.1.351 in the serum of convalescent patients and vaccinated people calls for the design of new potent vaccines targeting the emerging variant. However, since most vaccines approved and in clinical trials are based on the sequence of the original SARS-CoV-2 strain, the immunogenicity and protective efficacy of vaccines based on the B.1.351 variant remain largely unknown. In this study, we evaluated the immunogenicity, induced neutralization activity, and protective efficacy of wild-type spike protein nanoparticle (S-2P) and mutant spike protein nanoparticle (S-4M-2P) carrying characteristic mutations of B.1.351 variant in mice. Although there was no significant difference in the induction of spike-specific IgG responses in S-2P- and S-4M-2P-immunized mice, neutralizing antibodies elicited by S-4M-2P exhibited noteworthy, narrower breadth of reactivity with SARS-CoV-2 variants compared with neutralizing antibodies elicited by S-2P. Furthermore, the decrease of induced neutralizing antibody breadth at least partly resulted from the amino acid substitution at position 484. Moreover, S-4M-2P vaccination conferred insufficient protection against live SARS-CoV-2 virus infection, while S-2P vaccination gave definite protection against SARS-CoV-2 challenge in mice. Together, our study provides direct evidence that the E484K substitution in a SARS-CoV-2 subunit protein vaccine limited the cross-reactive neutralizing antibody breadth in mice and, more importantly, draws attention to the unfavorable impact of this mutation in spike protein of SARS-CoV-2 variants on the induction of potent neutralizing antibody responses.


Introduction
As of February 2022, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of COVID-19, had infected more than 400 million people and led to more than 5 million deaths worldwide (https://covid19.who.int, accessed on 15 February 2022). SARS-CoV-2 is an enveloped, positive-sense, single-stranded RNA virus that belongs immunization, resulting at least partly from immunogenicity or conformational changes caused by mutations at site 484. To our knowledge, there are very few reports comparing the immunogenicity and protection of prototype S protein nanoparticle vaccines with mutant S protein carrying characteristic mutations of the B.1.351 variant. Our results suggest that the prototype S protein nanoparticle vaccine elicits a much broader neutralization response than mutant S carrying the E484K mutation, further highlighting that the effects of site 484 mutations should be considered more carefully when designing new vaccines based on SARS-CoV-2 variants carrying E484K mutation.

Cloning, Protein Expression and Purification of SARS-CoV-2 Spike Protein
The codon-optimized SARS-CoV-2 spike gene encoding residues 1-1252 of the spike protein of the Wuhan-Hu-1 strain (GenBank: MN908947) was synthesized and inserted into the pOET1 vector to generate the WT-S construct. The S-2P construct was made through two prolines introduced at residues 986 and 987 (986-KV-987 to 986-PP-987) to stabilize the prefusion conformation and "QQAQ" substituted at the furin cleavage site (682-RRAR-685 to 682-QQAQ-685) to confer protease resistance. Four key mutations in the B.1.351 variant, namely K417N, E484K, N501Y, and D614G, were simultaneously introduced into S-2P to obtain the S-4M-2P construct. Recombinant baculoviruses expressing the S-2P or S-4M-2P proteins were generated using the flashBAC ULTRA system (Oxford Expression Technologies, Oxford, UK) following the manufacturer's instructions. Sf-βX cells were cultured in serum-free medium (Vigor-S100S, BIOENGINE) and infected with recombinant baculovirus at an MOI = 0.5 in a ReadyToProcess WAVE 25 Rockor (GE Healthcare Life Sciences, Marlborough, MA, USA) at 27 • C. Cells were harvested by centrifugation at 72 h post infection and then suspended in 25 mM Tris HCl (pH 8.0), 50 mM NaCl, and 0.5-1.0% (v/v) TERGITOL NP-9. The supernatant containing recombinant protein S-2P or S-4M-2P was clarified by centrifugation at 10,000× g for 30 min, followed by subsequent purification with TMAE anion exchange and lentil lectin affinity chromatography. Purified spike protein was formulated in 25 mM sodium phosphate (pH 7.2), 300 mM NaCl, and 0.02% (v/v) polysorbate 80 (PS 80) by a tangential flow ultrafiltration system and then subjected to SDS-PAGE and Western blot analyses.

SARS-CoV-2 Pseudovirus Titration
The titration of the SARS-CoV-2 pseudovirus was performed as described previously [24], with some modifications. Briefly, 293T-hACE2 cells (3.5 × 10 5 cells/mL) were seeded in 96-well plates one day before virus infection and then infected with a series of 3-fold dilutions of SARS-CoV-2 pseudovirus (6 wells per dilution factor). Cells without the addition of pseudovirus were used as controls. After 48 h of incubation, the culture supernatant was aspirated gently, and the cells were lysed for luminescence detection using a Bright-Glo Luciferase Assay (Promega, Dane County, WI, USA), according to the manufacturer's instructions, in a microplate luminometer (GloMax, Promega). A signal with ten-fold relative luminescence unit (RLU) values higher than those of the control was determined to be positive. The 50% tissue culture infectious dose (TCID 50 ) was calculated using the Reed-Muench method.

SARS-CoV-2 Pseudovirus-Based Neutralization Assay
The pseudovirus-based neutralization assays were performed as reported [25] with minor modifications. In brief, the heat-inactivated serum samples were initially diluted at 1:50 (v/v) and then mixed with 250 TCID 50 SARS-CoV-2 pseudovirus at 1:1 (v/v). After incubation for 30 min at 37 • C, the mixture was added to 293T-hACE2 cells in a 5% CO 2 37 • C incubator. Forty-eight hours later, cell lysates were collected for luminescence detection using a Bright-Glo Luciferase Assay (Promega) following the manufacturer's protocol in a microplate luminometer (GloMax, Promega). Neutralization curves were made with GraphPad Prism (GraphPad Software, Inc., La Jolla, CA, USA), and the ID 50 (50% inhibitory dilution neutralization titers) was calculated using a nonlinear regression (log(inhibitor) versus normalized response − variable slope) algorithm.

Mouse Immunization
Mouse immunization experiments were performed in Foshan Huamiao Biotechnology Company Ltd. (Foshan, China) Wild-type BALB/c mice were injected intramuscularly with 0.2 or 1 µg SARS-CoV-2 spike protein twice on day 0 (prime) and 14 (boost) adjuvanted with aluminum hydroxide (n = 8 for each dosage). A placebo group (n = 8) with adjuvant alone served as a nonimmunized control. Serum was collected for analysis before the initial immunization and at the indicated time points after the final immunization.

Mouse Challenge Experiments
hACE2 transgenic mice (6 weeks old, C57BL/6-Tgtn, CAG-human ACE2-IRES-Luciferase-WPRE-polyA) were provided by Shanghai Model Organisms (Shanghai, China). Mouse challenge experiments were performed in the Biosafety Level 3 (BSL3) Laboratories of Guangzhou Customs District Technology Center. hACE-2 mice were vaccinated through intramuscular injection with 1 or 5 µg SARS-CoV-2 spike protein twice at day 0 and day 14 adjuvanted with aluminum hydroxide (n = 10 for each dosage). A placebo group vaccinated with only adjuvant served as a nonvaccinated control (n = 10). Boost sera were collected 2 weeks after the final immunization. Vaccinated and control animals were intranasally challenged with 1 × 10 5 pfu of live SARS-CoV-2 virus (GenBank: MT123290) 16 days after the final immunizations. At 3 days post infection, mice were sacrificed. Lung tissues were collected to examine viral RNA levels or fixed with 10% formalin for histological examination using hematoxylin and eosin (H&E) staining. SARS-CoV-2 spike protein-specific IgG was analyzed by ELISA. Briefly, 96-well microtiter plates were coated with 1.0 µg /mL of target protein overnight at 4 • C. Plates were washed with phosphate-buffered saline with 0.1% Tween 20 (PBST) and blocked for 2 h using 200 µL of 5% nonfat milk. Mouse serum samples were serially diluted twofold and added to the blocked plates, followed by a 2 h incubation at 37 • C. Plates were washed with PBST and incubated with HRP-conjugated goat anti-mouse IgG (Sigma Aldrich, MI, USA) at a 1:10,000 dilution. After incubation for another 1 h at room temperature, the plates were washed with PBST and developed with TMB/E substrate (Merck Millipore, Burlington, MA, USA). Reactions were stopped with 1 M H 2 SO 4 , and the optical density at 450 nm (OD 450 ) values were read with an Epoch microplate spectrophotometer (BioTek Instruments Inc., Winooski, VT, USA). All raw OD 450 values had blank values subtracted before analysis. A subtracted OD 450 value above 0.1 and two-fold greater than that of the no-serum control was considered positive, and the maximum dilution with a positive result was used as the ELISA titer.

Electron Microscopy (EM)
Electron microscopy was performed using a Tecnai G2 Spirit transmission electron microscope operated at 120 kV. SARS-CoV-2 S protein samples were applied to nitrocellulosesupported 400-mesh copper grids and stained with uranyl formate. High-magnification images were acquired with an FEI Eagle 4 k × 4 k CCD camera.

Construction and Characterization of SARS-CoV-2 Spike Protein Nanoparticle Vaccines
The neutralization activity of sera from convalescents or individuals immunized with many approved vaccines against the B.1.351 variant decreased significantly [3]. Thus, it is extremely urgent to enhance protection against B.1.351 variant infection. To date, most current authorized or approved vaccines have been developed based on the wild-type SARS-CoV-2 (Wuhan-Hu-1, GenBank: MN908947.3) sequence, whose immunogenicity and protective effect of vaccines based on SARS-CoV-2 variants sequences remain unknown. Therefore, we designed two vaccines displaying the wild-type S protein or mutant S protein carrying the featured mutations of the B.1.351 variant. The coding sequence of the original SARS-CoV-2 strain (Wuhan-Hu-1, GenBank: MN908947.3) S protein (WT spike) was codonoptimized for expression in insect cells. S-2P was generated by introducing two proline substitutions at residues 986 and 987 as well as QQAQ substations between residues 682 and 685 of the WT spike. Substitutions of 986-KV-987 to 986-PP-987 (2P) were introduced to stabilize the prefusion conformation of spike protein [26][27][28][29][30], and modifications at the furin cleavage site (682-RRAR-685 to 682-QQAQ-685) were introduced to protect S-2P from protease digestion [31,32]. S-4M-2P was constructed by further introducing four featured mutations (K417N, N501Y, E484K, and D614G) of the B.1.351 variant into S-2P simultaneously ( Figure 1A). Both S-2P and S-4M-2P were cloned into recombinant baculovirus, expressed in Sf-βX cells, and verified by SDS-PAGE and Western blotting after further purification. The purified S-2P and S-4M-2P proteins had a molecular weight of 170 kDa ( Figure 1B,C). Transmission electron microscopy (TEM) showed that in the presence of polysorbate 80, both S-2P and S-4M-2P formed nanoparticles with uniform size and shape ( Figure 1D). that in the presence of polysorbate 80, both S-2P and S-4M-2P formed nanoparticles with uniform size and shape ( Figure 1D).

S-2P Induced a Broader Spectrum of Neutralizing Responses Than S-4M-2P
Then, the immunogenicity of S-2P and S-4M-2P was assessed in mice. BALB/c mice were immunized with low doses (0.2 μg) or high doses (1 μg) of S-2P and S-4M-2P with aluminum hydroxide adjuvant using a prime/boost regimen spaced 14 days apart ( Figure  2A). Mice immunized with S-2P or S-4M-2P had high S-specific IgG titers following the prime immunization and significantly elevated S-specific IgG titers after the boost immunization ( Figure 2B,C). The neutralization titers of these vaccinated sera against WT and variant SAR-CoV-2 were determined using in vitro pseudovirus-based neutralization assays. Animals immunized with S-2P or S-4M-2P displayed comparable neutralization titers against B.1.351 pseudoviruses after the boost immunization ( Figure 2D,E). However, the neutralization titers against wild-type SARS-CoV-2 pseudoviruses were significantly lower in S-4M-2P-immunized animals than in S-2P-immunized animals ( Figure 2F,G).

S-2P Induced a Broader Spectrum of Neutralizing Responses Than S-4M-2P
Then, the immunogenicity of S-2P and S-4M-2P was assessed in mice. BALB/c mice were immunized with low doses (0.2 µg) or high doses (1 µg) of S-2P and S-4M-2P with aluminum hydroxide adjuvant using a prime/boost regimen spaced 14 days apart ( Figure 2A). Mice immunized with S-2P or S-4M-2P had high S-specific IgG titers following the prime immunization and significantly elevated S-specific IgG titers after the boost immunization ( Figure 2B,C). The neutralization titers of these vaccinated sera against WT and variant SAR-CoV-2 were determined using in vitro pseudovirus-based neutralization assays. Animals immunized with S-2P or S-4M-2P displayed comparable neutralization titers against B.1.351 pseudoviruses after the boost immunization ( Figure 2D,E). However, the neutralization titers against wild-type SARS-CoV-2 pseudoviruses were significantly lower in S-4M-2P-immunized animals than in S-2P-immunized animals ( Figure 2F,G).  from S-2P-or S-4M-2P-immunized animals ( Figure 3A-I). The results showed that S-4M-2P induced significantly lower neutralization activity than S-2P against pseudoviruses without the E484K mutation ( Figure 3J) and, by contrast, obviously higher neutralization activity against pseudoviruses carrying E484K ( Figure 3K). Our results indicated that S-4M-2P-induced neutralizing antibodies mostly neutralized pseudotyped viruses carrying the E484K mutation, suggesting that the E484K mutation in S-4M-2P was the prime determinant for the induction and recognition of neutralizing antibodies.

NAbs Induced by S-4M-2P Mostly Recognized Pseudoviruses Carrying Lys at Site 484 in Spike Protein
To investigate the impact of mutations in the spike protein on neutralizing responses, we constructed a panel of eight SARS-CoV-2 pseudoviruses, including four single mutations and four combined mutations, based on four key mutations of B.1.351 spike protein (K417N, D501Y, E484K, and D614G) in S-4M-2P. Then, we performed neutralization tests using sera from S-2P-or S-4M-2P-immunized animals ( Figure 3A-I). The results showed that S-4M-2P induced significantly lower neutralization activity than S-2P against pseudoviruses without the E484K mutation ( Figure 3J) and, by contrast, obviously higher neutralization activity against pseudoviruses carrying E484K ( Figure 3K). Our results indicated that S-4M-2P-induced neutralizing antibodies mostly neutralized pseudotyped viruses carrying the E484K mutation, suggesting that the E484K mutation in S-4M-2P was the prime determinant for the induction and recognition of neutralizing antibodies.

S-2P Elicited Stronger Protective Immune Responses against SARS-CoV-2 Infection Than S-4M-2P in Mice
To further evaluate the potential efficacy of nanoparticle vaccines, hACE2 transgenic mice were challenged with live SARS-CoV-2 (Wuhan-Hu-1) intranasally 16 days after S-2P or S-4M-2P boost immunization ( Figure 5A). Consistent with previous results, S-2P elicited comparable neutralizing antibody titers against wild-type SARS-CoV-2 and B.1.351 variant pseudotyped virus in mice immunized with a high dose, whereas S-4M- Previous studies have shown that the titers of neutralizing antibody in sera from convalescent and vaccinated individuals gradually decline over time [33][34][35]. We wondered whether the titers of neutralizing antibodies in S-2P-immunized animals have a similar tendency to decline over time and hence measured the S-specific IgG and neutralizing antibody titers at different time points after boost immunization. The results showed no evident change in the level of S-specific IgG from 14 to 125 days after boost immunization ( Figure 4D). Although the neutralizing antibody titers against wild-type SARS-CoV-2 pseudoviruses slowly declined over time, they remained comparatively high even at 110 days after S-2P boost immunization ( Figure 4E,F). These results suggested that S-2P immunization could confer long-lasting protection.

S-2P Elicited Stronger Protective Immune Responses against SARS-CoV-2 Infection Than S-4M-2P in Mice
To further evaluate the potential efficacy of nanoparticle vaccines, hACE2 transgenic mice were challenged with live SARS-CoV-2 (Wuhan-Hu-1) intranasally 16 days after S-2P or S-4M-2P boost immunization ( Figure 5A). Consistent with previous results, S-2P elicited comparable neutralizing antibody titers against wild-type SARS-CoV-2 and B.1.351 variant pseudotyped virus in mice immunized with a high dose, whereas S-4M-2P induced neutralizing antibodies against only B.1.351 pseudovirus and hardly elicited neutralizing antibodies against wild-type SARS-CoV-2 pseudovirus even in mice immunized with a high dose ( Figure 5B,C). We subsequently measured genomic RNA loads in lungs of SARS-CoV-2-infected mice by RT-qPCR. For mice receiving either a low dose or high dose of S-2P vaccine, we observed noteworthy decreased levels of viral RNA in the lung in comparison with levels seen in the placebo control mice ( Figure 5D). For S-4M-2P, however, obviously declined levels of viral RNA were only observed in the lungs of mice immunized with high-dose but not low-dose S-4M-2P vaccine ( Figure 5D). 2P induced neutralizing antibodies against only B.1.351 pseudovirus and hardly elicited neutralizing antibodies against wild-type SARS-CoV-2 pseudovirus even in mice immunized with a high dose ( Figure 5B,C). We subsequently measured genomic RNA loads in lungs of SARS-CoV-2-infected mice by RT-qPCR. For mice receiving either a low dose or high dose of S-2P vaccine, we observed noteworthy decreased levels of viral RNA in the lung in comparison with levels seen in the placebo control mice ( Figure 5D). For S-4M-2P, however, obviously declined levels of viral RNA were only observed in the lungs of mice immunized with high-dose but not low-dose S-4M-2P vaccine ( Figure 5D). Lung tissues were also collected at 3 days post challenge, and sections were stained with hematoxylin and eosin for pathologic analysis. In sharp contrast to the lungs of placebo control animals ( Figure 5F), which were injured and characterized by infiltration of inflammatory cells, protein exudation, and thickened alveolar septa, the lungs of mice immunized with high-dose S-2P were relatively healthy, with little inflammation ( Figure  5G,H). However, the lungs of S-4M-2P-immunized mice displayed injury features similar to those of placebo control mice ( Figure 5I,J). Collectively, these data suggest that immunization with S-2P is able to generate a protective immune response, alleviate SARS-CoV-2-induced lung damage, and protect against SARS-CoV-2 infection in mice. Lung tissues were also collected at 3 days post challenge, and sections were stained with hematoxylin and eosin for pathologic analysis. In sharp contrast to the lungs of placebo control animals ( Figure 5F), which were injured and characterized by infiltration of inflammatory cells, protein exudation, and thickened alveolar septa, the lungs of mice immunized with high-dose S-2P were relatively healthy, with little inflammation ( Figure 5G,H). However, the lungs of S-4M-2P-immunized mice displayed injury features similar to those of placebo control mice ( Figure 5I,J). Collectively, these data suggest that immunization with S-2P is able to generate a protective immune response, alleviate SARS-CoV-2-induced lung damage, and protect against SARS-CoV-2 infection in mice.

Discussion
The most cost-effective and potentially long-term solution to the fight against a viral disease is to develop highly efficacious vaccines. Recent SARS-CoV-2 vaccine candidates are classified into five types: DNA vaccines, mRNA vaccines, virus vectored vaccines, inactivated vaccines, and protein subunit vaccines [36]. Protein subunit vaccines have a proven safety record in humans but are usually not sufficiently immunogenic. Generally, there are two major approaches to improve subunit vaccine immunogenicity for the production of high levels of neutralizing antibodies. One approach is to optimize vaccine design, such as the structure-based antigen design used in this study. Structure-based vaccine design seeks to create more efficient immunogens that will elicit protective antibody responses against the target pathogen based on a progressive understanding of antigen structure and function [37]. Due to the successful application of structure-based designed immunogens to induce the desired immune responses against RSV [38,39], HIV [40,41], PIV [42], and HCV [43,44], we designed a SARS-CoV-2 spike subunit vaccine based on the current understanding of the structures and fusion mechanisms of class I viral fusion proteins, such as RSV F [45], coronavirus S [46], and influenza HA protein [47]. Specifically, we introduced two proline substitutes to stabilize the prefusion conformation of the spike protein: a mutant furin cleavage site to protect the spike protein from protease digestion and a truncated cytoplasmic tail to increase spike protein expression on the cell surface [48]. Potent and broad-spectrum NAbs elicited by S-2P nanoparticle immunization and protective effects against live original SARS-CoV-2 virus infection suggested the great potential of the vaccine we designed. Regrettably, the protective effect of S-2P or S-4M-2P immunization against live B.1.351 variant infection has not been tested because of the inaccessibility of live B.1.351 variant virus. However, S-2P-induced broad-spectrum NAbs against SARV-CoV-2 variants pseudovirus at least suggest that this vaccine has the potential to protect infections caused by multiple variants.
Another approach to improve subunit vaccine immunogenicity is the addition of effective adjuvants. We noticed that despite the very low viral RNA load, viral RNA could still be detected in the lungs of mice inoculated with the S-2P vaccine ( Figure 5D), highlighting that there is still room for improvement in the protective effect. We speculate that the inability to completely prevent the infection was at least partly due to the insufficient efficiency of the aluminum adjuvant used in this study. Although aluminum adjuvant is the most frequently used adjuvant in human vaccines, it usually results in polarization toward a Th2 response, which has been regarded as unfavorable in the case of coronavirus and other viral infections and vaccinations [17,49]. Recently, several vaccine adjuvants have shown great potential in coronavirus vaccine strategies but are still in development, including MF59, AS03, Addavax, CpG ODN, and Matrix-M [17]. The effect of the S-2P vaccine in this study combined with these new adjuvants is promising.
Amino acid position 484 is located in the interaction interface with ACE2 of the receptor binding domain (RBD) of the spike protein and is involved in binding with ACE2 [50]. The E484K mutation first appeared in March 2020 and then rapidly spread and became one of the shared defining mutations of many SARS-CoV-2 variants, including the B.1.351 (Beta), P.1 (Gamma), P.2 (Zeta), P.3 (Theta), B.1.621 (Mu), B.1.526 (Lota), and B.1.525 (Eta) variants (https://covariants.org, accessed on 15 February 2022). Recent studies have suggested that the E484K mutation increases ACE2 affinity [51] and accounts for immune escape of B.1.351 and P.1 due to the reduction in neutralizing antibody recognition caused by conformational changes [7,[52][53][54][55]. However, the effect of the E484K mutation on immunogenicity and neutralization breadth remains unknown. Our results indicated that more potent and cross-reactive neutralizing responses were induced by the wild-type spike protein than by the spike protein carrying the E484K mutation. The NAbs elicited by spike protein carrying the E484K mutation exclusively recognized SARS-CoV-2 pseudoviruses with Lys at 484 site of spike protein but not recognized SARS-CoV-2 pseudoviruses with Glu, Ala, or Gln at 484 site of spike protein ( Figure 4B,C). The limited neutralization breadth and potency resulting from the E484K mutation raised concern about the protection of designed vaccines based on the SARS-CoV-2 variants carrying E484K mutation in spike protein against infection with the emerging SARS-CoV-2 lineages without E484K mutation. As expected, S-4M-2P vaccination offered relatively poor protection in hACE2 mice challenged with the original SARS-CoV-2. Of note, our findings are consistent with recent studies that have suggested that plasma from B.1.351 variant-infected patients displayed remarkably lower neutralization of the original variant than of the B.1.351 variant [56,57]. Substitution of negatively charged residue Glu484 by Lys with positive charge changed the electrostatic energies and in turn contributed to the increased binding affinity with hACE2 as well as reduced interactions with neutralizing antibodies [58,59]. We speculate that the change of electrostatic energies caused by E484K mutation is at least partly responsible for the narrow breadth of the neutralizing antibody. The underlying mechanisms, however, need further investigation.
To counteract the impact of circulating variants, one option is to develop new vaccines that more closely reflect the variants. However, it is not clear how beneficial such vaccines designed specifically to target new variants will be. Although the underlying mechanism is not entirely clear, our findings provide the first preliminary evidence that the E484K mutation results in poor cross-reactivity of NAbs except for resulting in immune escape by affecting the recognition of neutralizing antibodies. Consequently, careful consideration should be given when designing new vaccines based on SARS-CoV-2 variants, especially variants with the E484K mutation in spike protein, to potentiate vaccine efficacy.