Cross-reactive antibodies elicited to conserved epitopes on SARS-CoV-2 spike protein after infection and vaccination

SARS-CoV-2 is a novel betacoronavirus that caused coronavirus disease 2019 and has resulted in millions of deaths worldwide. Novel coronavirus infections in humans have steadily become more common. Understanding antibody responses to SARS-CoV-2, and identifying conserved, cross-reactive epitopes among coronavirus strains could inform the design of vaccines and therapeutics with broad application. Here, we determined that individuals with previous SARS-CoV-2 infection or vaccinated with the Pfizer-BioNTech BNT162b2 vaccine produced antibody responses that cross-reacted with related betacoronaviruses. Moreover, we designed a peptide-conjugate vaccine with a conserved SARS-CoV-2 S2 spike epitope, immunized mice and determined cross-reactive antibody binding to SARS-CoV-2 and other related coronaviruses. This conserved spike epitope also shared sequence homology to proteins in commensal gut microbiota and could prime immune responses in humans. Thus, SARS-CoV-2 conserved epitopes elicit cross-reactive immune responses to both related coronaviruses and host bacteria that could serve as future targets for broad coronavirus therapeutics and vaccines.

Multiple studies have identified that individuals without previous SARS-CoV-2 infection had varying levels of pre-existing antibodies that could cross-react with SARS-CoV-2 and may influence SARS-CoV-2 immunity and disease severity [40][41][42][43][44] . Furthermore, other groups have identified regions on the SARS-CoV-2 spike that are highly conserved and could serve as targets to generate more broad cross-reactive immunity to many types of coronaviruses 45,46 . These naturally occurring or elicited poly-reactive antibodies could recognize phylogenetically closely related viruses and could potentially provide immunity against emerging viral variants or related strains. With an increase in novel coronaviruses infecting humans in recent history, pan-coronavirus vaccines targeting a multitude of related viruses would be an effective, possibly necessary, tool against future viral threats. Previous studies with other viral targets have exemplified that immunization to a virus or antigen could provide a broad antibody response and cross-reactivity against similarly structured antigens regardless of viral origin 47,48 . It will be critical to identify and characterize epitopes on SARS-CoV-2 that are the source of cross-reactive immune responses.
In humans, there is considerable variability in the immune responses to both infections and vaccinations within the population. Several genetic and environmental factors have been identified that could help explain the mechanisms of this variation. One potential source of cross-reactive immunity, apart from viruses with similar protein sequence or structure, could be exposure to commensal bacteria that contain proteins with sequence homology to infecting viruses. Components of the host gut microbiome have been shown to prime or alter the immune system response to vaccination prior to exposure to antigens 49,50 . Recently, epitopes within the SARS-CoV-2 receptor binding domain and S2 subunits of the spike protein have been shown to share epitopes with commensal bacteria 51,52 . These shared epitopes may have primed immune responses to individuals naïve to the viral components. However, the origins of the SARS-CoV-2 cross-reactive antibodies and their potential impacts on vaccines and infection have not been fully elucidated.
In this study, we determined cross-reactive antibody responses elicited to related alpha-and betacoronaviruses after SARS-CoV-2 infection or COVID-19 BNT162b2 vaccination. Moreover, we identified immunodominant peptide epitopes within the SARS-CoV-2 spike protein with high sequence conservation amongst betacoronaviruses that could serve as platforms for future vaccine designs. Additionally, we identified that antibody responses to SARS-CoV-2 spike S2 subunit epitopes could cross-react with commensal gut bacteria proteins that may prime SARS-CoV-2 antibody responses 51,52 .

Results
Comparative analysis of closely related coronavirus spike proteins. We performed a Multiple Sequence Alignment (MSA) to determine the amino acid sequence similarities between the viral spike protein sequences of closely related coronaviruses to SARS-CoV-2 and generated a phylogenetic tree from the underlying alignment. Specifically, we compared the coronavirus spike protein sequences of SARS-CoV-1 and SARS-CoV-2 with four SARS-like coronaviruses that infect animals and are not yet zoonotic, as well as four human seasonal coronavirus spike protein sequences and MERS-CoV ( Fig. 1A & Supplemental Fig. 1A). Newly emergent Bat CoV RaTG13 and Pangolin CoV QIQ54048.1 were closely related to SARS-CoV-2 while Bat CoV WIV1 and Bat CoV RsSHC014 were more closely related to SARS-CoV-1 (Fig. 1A). MERS-CoV was more distantly related to the SARS coronaviruses we compared (Fig. 1A). It is important to note that although MERS-CoV, SARS-CoV-1 and SARS-CoV-2 all belong to the betacoronavirus genus, the phylogenetic origin of MERS-CoV is classified as clade II while SARS-CoV-1 and SARS-CoV-2 are clade I, clusters IIb and IIa respectively 53,54 . With regards to the four seasonal human coronaviruses that generally cause moderate or mild respiratory symptoms, Human CoV 229E and Human CoV NL63 clustered together (both alphacoronaviruses) and Human CoV OC43 clustered with Human CoV HKU1 (both betacoronaviruses) 55,56 (Fig. 1A). Thus, the spike proteins of SARS-CoV-1, SARS-CoV-2, and several emergent coronaviruses have high sequence similarity within the spike protein that is critical for host cell attachment and infectivity.
Next, we collected plasma from 24 individuals of diverse age, sex, and race/ethnicity who had laboratory confirmed SARS-CoV-2 infection 12-42 days prior to collection (Supplementary Table 1). Using an enzyme-linked immunosorbent assay (ELISA) that measured immunoglobulin (Ig) G antibody responses to the SARS-CoV-2 spike protein, we found that 20 of 24 individuals (83%) had detectable antibodies to SARS-CoV-2 (Fig. 1B). We then investigated whether the previously infected SARS-CoV-2 individuals (seropositive) had cross-reactive antibody responses to the related coronaviruses using ELISA to the spike proteins for SARS-CoV-1, the emergent Bat CoV RaTG13, and the more distantly related MERS-CoV. We found that 23 of 24 samples (96%) demonstrated detectable antibody response to SARS-CoV-1 and Bat CoV RaTG13 (Fig. 1C, D). Surprisingly, 21 of 24 samples (88%) also demonstrated an antibody response to the more distantly related MERS-CoV spike protein (Fig. 1C, D). These results suggested that infection with SARS-CoV-2 elicited antibody responses in recovered individuals that cross-reacted with other coronaviruses, even those belonging to different betacoronavirus clades.
The Pfizer-BioNTech BNT162b2 vaccine elicited cross-reactive antibodies to related coronaviruses. We next investigated whether vaccination with SARS-CoV-2 spike could elicit cross-reactive antibod-  Table 2). We found that most individuals did not have detectable antibody levels, as measured by ELISA for SARS-CoV-2, SARS-CoV-1, Bat CoV RaTG13, or MERS-CoV spike proteins at week 0 ( Fig. 2A-D). We did detect low levels of reactivity at week 0 in one individual against SARS-CoV-2 and three individuals against MERS-CoV, but only at the lowest plasma dilution ( Fig. 2A, D). After both immunizations with the COVID-19 vaccine at Week 7, all individuals (16 of 16) had high levels of antibodies against SARS-CoV-2 spike protein ( Fig. 2A). Moreover, all individuals also had detectable levels of cross-reactive antibodies against SARS-CoV-1 spike protein and to the emergent Bat CoV RaTG13 spike protein (Fig. 2B, C). Albeit detected at lower levels than SARS-CoV-1 and Bat CoV RaTG13, 15 of the 16 individuals had increased antibodies against the more distantly related MERS-CoV spike protein after vaccination with SARS-CoV-2 mRNA vaccine (Fig. 2D). These results suggested that 2 doses of the Pfizer-BioN- Pre-existing antibody levels targeting seasonal coronaviruses were largely not significantly affected by the Pfizer-BioNTech BNT162b2 vaccine. Seasonal coronaviruses are more divergent in viral sequence to SARS-CoV-2 and are reasonably common, first infecting humans in childhood, and contribute to up to 30% of illnesses that are categorized as the common cold 57 . There are currently four major seasonal coronaviruses that infect humans, Human CoV 229E, Human CoV OC43, Human CoV HKU1 and Human CoV NL63. The amino acid sequence homology of spike proteins of the endemic coronaviruses to SARS-CoV-2 are less than 30% (Supplemental Fig. 1A). We measured antibody levels against the spike proteins from all four seasonal coronaviruses in the individuals with no history of SARS-CoV-2 infection at the two COVID-19 mRNA vaccine time points (week 0 and week 7 described above) from our hospital workers based in the United States. We found that all individuals (16 of 16) had detectable antibodies by ELISA to the spike proteins from all four seasonal coronaviruses prior to immunization at baseline (week 0), reflecting the high prevalence of pre-existing immunity to these four seasonal coronaviruses ( Fig. 3A-D). After immunization with the COVID-19 mRNA vaccine (week 7), there was a modest but statistically significant increase in antibody magnitude in individuals against the Human CoV OC43 spike protein (Paired Wilcoxon p < 0.0001; Fig. 3A). However, there was no significant increase from baseline in antibody levels to the other three seasonal coronavirus spike proteins after SARS-CoV-2 mRNA vaccination ( Fig. 3B-D). We did observe that Human CoV NL63 had a statistically significant decrease in antibody response (Paired Wilcoxon p < 0.0001; Fig. 3D), but the mechanism of this decrease is unclear. These data demonstrated that individuals had high levels of preexisting antibody levels against the four seasonal coronaviruses prior to COVID-19 vaccination and that the Pfizer-BioNTech BNT162b2 vaccine did not significantly boost the antibody responses to these coronavirus spike proteins, apart from the modest increase observed for Human CoV OC43.

Identification of immunodominant regions with high sequence conservation on the SARS-CoV-2 spike protein.
Since we observed the elicitation of cross-reactive antibody responses against related coronaviruses after exposure to the SARS-CoV-2 spike protein, we next determined the precise antibody epitopes targeted after SARS-CoV-2 infection using a peptide microarray. Specifically, we performed a SARS-CoV-2 peptide array that included overlapping peptides across the S1 and S2 subunits of the SARS-CoV-2 spike protein using convalescent blood samples from 14 individuals previously infected with SARS-CoV-2 ( Fig. 4A, B). We calculated z-scores across the group of individuals for each specific peptide and identified peptide regions in both the spike S1 and S2 subunits that were immunodominant among the 14 individuals. We considered a z-score greater than one as "immunodominant" as the levels would be one standard deviation higher than binding to the other peptides on the spike protein. These immunodominant regions were present in the RBD of the spike protein that is critical for recognizing the host cell receptor ACE2, but also outside of the RBD, in the N-terminal domain (NTD) of the S1, near the fusion peptide and the C-terminal of the S2 subunit (Fig. 4A, B). In order to measure sequence conservation among related coronaviruses for each peptide, we generated a Bitscore for each peptide from a BLAST sequence alignment using SARS-CoV-2, SARS-CoV-1, Bat CoV RaTG13, and MERS-CoV spike protein amino acid sequences (Fig. 4A, B). To identify immunodominant peptides that also had high sequence conservation among related coronaviruses, z-scores and Bitscores were compared in both the S1 and S2 regions of the SARS-CoV-2 spike protein (Fig. 4C, D). We identified four peptides in the S1 region (S1-76, S1-94, S1-105, S1-111) and three peptides in the S2 region (S2-78, S2-97, S2-96) that were all immunodominant (z-score ≥ 1) and had high sequence conservation with Bitscores one standard deviation above the mean (Fig. 4C, D). These results demonstrated that there are regions within the S1 and S2 subunit that are conserved at the amino acid level between multiple coronaviruses and are also frequently targeted by antibodies in individuals after infection with SARS-CoV-2.
Using ELISA, we found that 15 of the 24 seropositive individuals (63%), had high levels of antibody binding to the S2-78 peptide (Fig. 5B). None of the seronegative individuals that were vaccinated at week 0 demonstrated any antibody binding to the S2-78 peptide at baseline (Fig. 5B). However, after COVID-19 vaccination (week 7) all but a single individual (96%) had high levels of antibody binding to the S2-78 peptide (Fig. 5B). This confirmed the frequent targeting of this epitope after SARS-CoV-2 infection or vaccination.
We next sought to determine if immunization with the S2-78 peptide could elicit a cross-reactive antibody response to related coronaviruses. Since peptide antigens are often poorly immunogenic alone, we synthesized the S2-78 peptide and conjugated it with a diphtheria toxin carrier protein CRM197 to boost immunogenicity as a peptide-conjugate vaccine (Fig. 5C). We immunized five BALB/c mice with the S2-78 conjugate vaccine three times on days 0, 21 and 35 (Fig. 5D). We collected blood samples at baseline and seven days after each immunization at days 0, 28, and 42 (Fig. 5D). After the second (day 28) or third (day 42) immunizations we detected antibody binding to the S2-78 peptide for all five animals (Fig. 5E). Most mice also had detectable antibodies against the full SARS-CoV-2 spike protein (day 28, 5 of 5; day 42, 4 of 5) as well as the SARS-CoV-1 (day 28, 4 of   (Fig. 5E). We did not detect any antibodies in the mice targeting the more divergent MERS-CoV spike protein after immunization with the S2-78 peptide conjugate vaccine. We tested a single concentration of serum from each of the mice (1:40) in a SARS-CoV-2 and SARS-CoV-1 pseudovirus neutralization assay and did not detect a significant reduction in infection or neutralization (Supplemental Fig. 2). This may indicate that higher levels are required to determine neutralization or the other conserved epitopes with stronger neutralization activity could be selected for future vaccine designs. Thus, immunization with a peptide conjugate vaccine in mice using a peptide that is conserved elicited cross-reactive antibodies against multiple coronaviruses.

Cross-reactivity with commensal gut bacteria may prime antibody responses to the S2 subunit of SARS-CoV-2.
A recent report showed that a region in the SARS-CoV-2 S2 subunit that contained the S2-78 peptide sequence elicited pre-existing antibodies that cross-reacted with commensal gut bacteria 51 .
Using the SARS-CoV-2 S2-78 sequence as reference, we performed a sequence alignment against the entire bacterial proteome to identify sequence homology between the viral and commensal bacterial proteins that could be targeted by the humoral immune response. We identified numerous bacterial proteins with similar sequences that existed across a broad range of bacteria, including proteins from many genera of bacteria that exist within the human microbiome. Specifically, the top 10 bacterial proteins had sequence homology with SARS-CoV-2 S2-78 between 66.6 and 83.3%, which is more conserved than the 58.3% homology of distantly related MERS-CoV (Fig. 6A). Serum from the mice immunized with the S2-78 peptide conjugate vaccine at baseline (day 0) and after three immunizations (day 42) was used to determine antibody cross-reactivity to commensal gut bacteria that was collected from human feces. We found that 4 of the 5 mice had increased antibody reactivity compared to baseline against the human fecal protein lysate after immunization at day 42 at either 1:5 or 1:10 serum dilutions (Fig. 6B). This data suggested that proteins contained in commensal gut bacteria shared protein sequence homology with the coronavirus S2-78 peptide and could provide preemptive exposure to cross-reactive epitopes without exposure to the coronavirus spike protein.
Interestingly, we observed heterogeneity in the magnitude of pre-existing cross-reactive antibodies targeting the S2 subunit in human individuals (n = 140) at baseline before COVID-19 vaccination. We hypothesized that this difference in pre-existing S2 antibody levels could be due to differences in priming by commensal bacteria or other microbes and resulted in differences in antibody responses after vaccination. We used the baseline S2-targeting antibody levels to stratify individuals into three groups, high (top 25% of binding, MFI > 1557), medium (middle 50% of binding, MFI 526-1557), and low (bottom 25% of binding, MFI < 526) antibody response (Fig. 6C). These grouped samples were then measured again for S2 subunit binding 3 weeks after the first immunization to the Pfizer-BioNTech's BNT162b2 vaccine. Samples with low S2 binding before vaccination at week 0 had significantly lower levels of antibodies binding the S2 subunit after the first immunization at week 3 compared to individuals with medium or high pre-existing S2 antibody levels (Fig. 6D). We did not observe any significant differences between the groups for antibody levels to the S1 or RBD subunits of the spike protein (Supplemental Fig. 3). We also did not observe any significant correlation of S2 binding levels with the antibody levels against the four seasonal coronaviruses measured in Fig. 3. This suggested that there may be preexisting antibodies elicited by reactivity with commensal gut bacterial that cross-reacted with the S2 subunit providing higher antibody response to the S2 subunit of SARS-CoV-2 without prior infection.

Discussion
In this study, we investigated the antibody responses of individuals with recent SARS-CoV-2 infection and determined if these antibodies could cross-react with other related strains of coronavirus. We identified that natural infection with SARS-CoV-2 produced cross-reactive antibody responses that could target other known and emerging coronaviruses. Moreover, we found that vaccination with two doses of the Pfizer-BioNTech BNT162b2 mRNA COVID-19 vaccine also elicited antibodies that could react with other SARS-like and MERS betacoronaviruses.
Consistent with previous studies, we observed that the SARS-CoV-2 spike protein shares high amino acid sequence similarity to other closely related infectious coronavirus spike proteins 1,14,54,55 . These spike protein Figure 6. Immunodominant S2-78 peptide was cross-reactive with commensal gut bacteria. (A) Amino acid sequence alignments of the S2-78 peptide across 12 example bacterial sequences identified using sequence alignment of the bacteria proteome (Fig. 5A). Blue shading indicates amino acid conservation, dashes indicate gaps in sequence. Percentage of sequence homology shown on the right of the alignments. (B) Bar and line graph of antibody responses obtained by ELISA at different serum dilutions of mouse serum immunized with the S2-78 peptide against the commensal gut bacteria cell lysate (n = 5 mice per each time point). Endpoint titers are designated by the most dilute plasma concentration detected above the minimum threshold of the background OD450 multiplied by 3. Dots are representative of individual samples. Bars are representative of the mean of all samples. (C, D) Multiplex bead-based antibody binding assay that measured the IgG antibody response to SARS-CoV-2 spike subunit 2 (S2) in distinct healthy individuals (C) at baseline (n = 140) prior to receiving COVID-19 vaccine or (D) at week 3 after receiving the first COVID-19 vaccine immunization. Individual values of Median Fluorescent Intensity (MFI) are calculated; background subtraction has been used to remove nonspecific signal. Line indicates group median. Individuals at baseline with the top 75% MFI were grouped as high (green) and individuals with the lowest 25% MFI were grouped as low (blue) and individuals with 25-75% MFI were grouped as medium (red). P values were determined with a Wilcoxon-Mann-Whitney test. www.nature.com/scientificreports/ phylogenies include several emergent bat and pangolin originating coronaviruses (Bat CoV WIV1, Bat CoV RsSHC014, Bat CoV RaTG13, and Pangolin CoV identifier QIQ54048.1), that have yet to become pandemic in nature, but are being closely monitored due to their potential high-risk 61 . Due to the similarity of these aforementioned spike proteins to SARS-CoV-2 spike protein, we speculated there would be cross-reactivity with antibody responses to these other coronaviruses for individuals exposed to SARS-CoV-2, especially for SARS-CoV-1 and the bat and pangolin variants as they are all known to be classified as clade 1 betacoronaviruses 53,54 . The fact that we found cross-reactive antibodies to these related coronaviruses after infection with SARS-CoV-2 raised the hypothesis that broadly neutralizing antibodies to multiple coronaviruses could be induced. Indeed, cross-reactive antibodies have been isolated to multiple coronavirus epitopes with diverse effector functions 62 . Further characterization of the most potent neutralizing antibodies with the broadest coverage of coronaviruses will identify the optimal epitopes to target in future vaccines and therapies. The determination that both seropositive and vaccinated individuals had high antibody binding, not only to the targeted SARS-CoV-2, but to other coronaviruses with phylogenetically similar spike protein sequences (SARS-CoV-1, MERS-CoV, Bat CoV RaTG13) could provide a template to create more broad coronavirus therapeutics or a vaccine. These novel vaccines could elicit broadly neutralizing antibodies to protect against current and future viral threats. 21 of 24 SARS-CoV-2 seropositive samples and 3 seronegative samples had antibody responses to the distantly related MERS-CoV spike protein, indicating that previous exposure to similar coronavirus epitopes could elicit cross-reactive responses to the MERS spike protein, even though it is one of the least homologous of the coronaviruses we examined. Interestingly, we found that while the whole spike protein of Human CoV OC43 had low sequence homology to SARS-CoV-2 spike protein, the S2-78 peptide region of SARS-CoV-2 had 75% identify to the same region in Human CoV OC43. This indicated that specific epitopes in even more distantly related coronaviruses could impact the antibody response. This is supported with evidence of increased antibody responses to both seasonal coronavirus Human CoV OC43 (Fig. 3A) and MERS-CoV (Fig. 2D) after vaccination with the SARS-CoV-2 spike protein. Although still detectable, we did detect a significant decrease in cross-reactive antibodies to the seasonal coronavirus NL63 spike after vaccination. It is unclear what the reason for this shift and kinetics of these changes are over the course of COVID-19 vaccination. Previous studies have demonstrated that cross-reactive immune responses with chimeric spike protein mRNA vaccines elicited protection against a multitude of related coronaviruses and SARS-CoV-2 variants by inclusion of bivalent and trivalent epitopes from multiple locations of the spike protein sequence 45,46,62,63 . Future work characterizing the pan-coronavirus cross-reactivity of these antibodies, isolation and identification of the most immunodominant epitopes through binding, structural, functional, and sequence conservation analysis would be needed to provide strong vaccine target candidacy.
We used a SARS-CoV-2 spike peptide array and sequence homology analysis to identify epitopes on the spike protein that are both immunodominant and highly conserved among related coronaviruses. Using this approach, we identified one peptide in the S2 subunit and performed a proof-of-concept study of designing a conjugate vaccine and immunizing mice. This resulted in the induction of broadly cross-reactive antibodies in the immunized mice and clearly demonstrates that peptide conjugate vaccines could induce cross-reactive coronavirus antibody responses. Although S2-78 represented the strongest candidate through binding and conservation of sequence, the antibodies produced from immunizing the mice did not result in neutralization of SAR-CoV-2 or related SARS-CoV-1 from binding to the ACE2 receptor [58][59][60][64][65][66] . Additionally, the importance of the sequence conservation can be exemplified by the lack of antibody binding to the least conserved MERS-CoV spike protein by all S2-78 peptide vaccinated mice. However, prior studies identified that antibodies targeting this epitope had neutralizing activity and correlated with disease outcomes after SARS-CoV-2 infection 58-60 . We did observe higher antibody titers to the peptide alone compared to the native spike proteins after immunization with our peptide vaccine. This suggested that the peptide vaccine could elicit a population of antibodies that did not recognize the spike protein as it is presented on the virion and would not neutralize the virus. Alternatively, serum neutralization after immunization may not have been detected because the neutralizing antibody levels were at too low at the dilution of mouse serum that was utilized. Future studies determining the specific properties of the antibodies elicited by this peptide vaccination, including other Fc-mediated functions beyond neutralization, and studies selecting other epitopes in the spike protein that may be more optimal for vaccine protection, such as those in the RBD, will be required.
There are two preliminary reports that found that antibodies elicited to SARS-CoV-2 could cross-react with proteins in the commensal gut microbiome 51,52 . One of these epitopes overlapped with the S2-78 peptide sequence we used for our immunization study. We also found high sequence similarity between S2-78 peptide and proteins found in bacteria. Moreover, we found that individuals with higher levels of pre-existing S2 antibodies before COVID-19 vaccination had significantly higher levels after immunization. Prior studies with influenza vaccines have shown that treatment with antibiotics and disruption of the microbiome may result in impairment of immunoglobulin neutralization 49 . These observations raised the hypothesis that the composition of the microbiome could influence antibody levels. Our findings identified high sequence conservation of SARS-like coronavirus spike proteins in viruses and bacteria and ascertained the possibility of the Pfizer-BioNTech BNT162b2 vaccine to boost immunity against more than just SARS-CoV-2. Future studies of components of both commensal gut bacteria and the virome should be carried out to determine how exposure to other bacteria and viruses impact the SARS-CoV-2 antibody response. A limitation of these findings is whether a neutralizing response occurs from gut bacteria exposure or if these antibodies recognize intact microbiome or if bacterial lysis must occur. Lastly, in the pursuit of generating cross-reactive antibody responses that could react with the microbiome, determinization of deleterious effects towards the microbiome or host proteins should be measured.
In summary, our data demonstrated that SARS-CoV-2 infection or vaccination elicited cross-reactive antibodies that target not only SARS-CoV-2, but related coronaviruses. Moreover, we showed that immunization with conserved peptide regions of SARS-CoV-2 induce cross-reactive antibodies in mice. Lastly, we provided www.nature.com/scientificreports/ evidence that suggested that the makeup of the gut microbiome could influence SARS-CoV-2 antibody levels. These data lay the groundwork for developing a pan-coronavirus vaccine that could elicit cross-reactive immunity to a broad range of coronavirus species. Further research will be needed to optimize this process in order to identify epitopes and vaccine platforms that induce protective immunity.

Materials and methods
Phylogenetic construction. Spike protein amino acid sequences, for the respective virus, were collected from the NCBI protein database (Supplemental Table 3). Multiple Sequence alignment (MSA) was done using MUSCLE (v3.8.1551) 67 . RAXML (v2.0) was used for generating maximum likelihood phylogenetic tree with Blosum62 substitution model and 100 replicates 68 . Final visualization of phylogenetic tree was done using Environment for Tree Exploration (ETE) toolkit 69 .
Human subjects. Informed consent was obtained from all subjects and/or their legal guardian(s). This study was reviewed and approved by the Children's Mercy IRB (#00001670 and #00001317). Participants selfenrolled after they had reviewed a study information letter and were given opportunity to ask questions. All methods were carried out in accordance with relevant guidelines and regulations as reviewed and approved by the IRB. Healthcare workers from our children's hospital were enrolled prior to the administration of the Pfizer BNT162b2 SARS-CoV-2 vaccine. Plasma from peripheral blood was collected before vaccination as a baseline (week 0), after primary immunization (week 3), and after second immunization (week 7) from individuals with no known history of infection (n = 140). Sample population consisted of mostly adult middle aged, white, females who did not identify as Hispanic or Latino (Supplementary Table 1 Manufactured ELISA. Detection and quantification of S1 IgG class antibodies was performed using the High-Sensitivity SARS-CoV-2 S1 IgG ELISA kit (41A232, Biovendor, Asheville, NC, USA) following standard protocol with serum or plasma diluted of 1:100. SARS-CoV-2 spike peptide array. Plasma samples were diluted 1:200 and used to probe a single SARS-CoV-2 protein and peptide microarray (CDI Labs, Mayaguez, Puerto Rico). After probing arrays with serum antibodies, the arrays were washed, labeled with an Alexa 647 anti-human IgG Fc secondary antibody and scanned using a GenePix 4000B scanner (Molecular Devices, San Jose, CA, USA). Array data was collected using the MAGPIX software (Innopsys, Chicago, IL, USA). Each protein or peptide was represented in triplicate on the microarray. There were positive control proteins (human IgG, anti-human IgG, and ACE2_Fc) and blank wells served as negative controls. The signal intensity was measured in the detection channel 635 nm (F635). The average of the F635 for each peptide was calculated and log2 transformed for graphical presentation.
Spike protein sequence from the seven different strains (Bat CoV RaTG13, Pangolin coronavirus, SARS-CoV-2, SARS-CoV-1, MERS-CoV, Bat CoV RsSHC014, BAT CoV WIV1) of coronavirus were used as reference database. Amino acid sequences for Peptide ID (S1-1 to S2-97) were then blasted against this reference database. Blastp-task blastp-short (blast-v2.2.29+) was used to find sequence homology between viral sequences and peptide of interest 70  Mouse immunization with SARS-CoV-2 peptide conjugate vaccine. The following methods reported are in accordance with the ARRIVE guidelines (https:// arriv eguid elines. org) for the reporting of animal experiments. The mouse study was performed at Hooke Laboratories, Lawrence, MA, USA. All methods were reviewed and approved by the Hooke and Children's Mercy IACUC and performed in according to the relevant international laboratory animal welfare guidance and regulations. Mice were sourced from Taconic Biosciences, Rensselaer, NY, USA. 5 BALB/c female mice between 7 and 10 weeks of age were immunized subcutaneously with 0.1 ml/mouse at 37.5 ug/mL of the FKEELDKYFKNHC-CRM197 peptide (Creative Biolabs), mixed 1:1 with Addavax (VAC-ADX-10, Invivogen, San Diego, CA, USA) on days 0, 21, and 35. Serum was collected on day 0, 28, and 42. Baseline sample (day 0) was used as the negative control for each mouse. No randomization or blinding was utilized in this study. Antibody level outcomes were performed across the timepoints during immunization and tested for statistical significance using nonparametric tests due to the exploratory sample size.

SARS-CoV-2 viral antigen multiplexed binding assay.
To measure antibody levels to SARS-CoV-2 spike subunit proteins, spike subunit 1 (S1), spike subunit 2 (S2), and receptor-binding domain (RBD), were used on a bead-based multiplex assay based on the Luminex xMAP technology. Reagent kits with secondary antibodies specific for immunoglobulin G (IgG) were used (HC19SERG1-85K, MilliporeSigma, Burlington, MA, USA) following manufacture protocol. The kit provided a set of SARS-CoV-2 antigen conjugated beads (S1, S2, RBD) along with 3 positive control beads and a negative control bead set. The positive control beads were beads coated with different concentrations of IgG. The negative control beads did not have antigen conjugated to determine nonspecific binding. The 3 antigen-conjugated beads, 3 positive control beads, and 1 negative control beads were mixed and incubated with each plasma sample at a dilution of 1:100 with assay buffer. Samples were run in duplicate. Each plate contained at least two wells with only buffer and no plasma to determine background activity. PE-anti-human IgG conjugate detection antibody was utilized to determine antibody response to each SARS-CoV-2 antigen. Using the positive control beads, inter-assay (plate-to-plate) co-efficient of variation (CV) was determined to be 5.16% for IgG. We utilized the Luminex analyzer (MAGPIX) and Luminex xPONENT acquisition software to acquire and analyze data. After acquisition net MFI was calculated by subtracting background MFI (no plasma).
After net MFI was determined for all samples, individuals before vaccination at week 0 were grouped based on their binding to S1, S2, and RBD and categorized into low (bottom 25% of binding), medium (middle 50% of binding), and high (top 25% of binding) for each antigen. These groupings were then tracked on their response at week 3 after vaccination to determine if initial binding to antigen determined future antibody response to each antigen.
Collection of commensal gut bacterial protein from human fecal sample. A human fecal sample (D6323, Zymo Research, Irvine, CA, USA) was centrifuged at 200 g for 5 min twice to remove debris. Supernatant was collected and centrifuged at 9000 g for 10 min to collect bacterial pellet. Pellet was resuspended in PBS with 1X Halt protease and phosphatase inhibitor (78442, ThermoFisher, Waltham, MA, USA) and sonicated for 30 s with a probe sonicator 3 times. Protein concentrations were measured using the Pierce BSA protein assay kit (23225, ThermoFisher).
Commensal gut bacteria ELISA. ELISA was performed using bacteria lysate collected from human fecal samples diluted to 1 ug/mL in 0.1 M sodium bicarbonate and incubated on high-binding plates (3369, Corning) overnight at 4 degrees. Mouse serum was diluted to 1:5 in superblock buffer with sodium azide followed by subsequent dilutions at 1:10, 1:30, 1:90, and 1:270. Secondary antibodies were purchased from Jackson Immu-noResearch: Goat anti-mouse IgG (Cat# 115-035-003, Lot# 153294). Secondary antibody dilutions were done in superblock buffer without sodium azide within range of manufacturer's recommendations at 1:50,000 dilution. SureBlue Reserve Microwell Substrate (95059-294, VWR) was added and incubated in the dark for 15 min. Absorbance was measured at 450 nm immediately after 0.33 N HCl Acid Stop solution was added to the plate.
Statistical analysis. The statistical analysis was performed using Graphpad Prism 9.1. For multiple comparison, the statistical significance was determined with a Wilcoxon-Mann-Whitney test with two-tailed p values.

Data availability
All relevant data are within the manuscript and supplementary information files. Any additional data is available from the corresponding author at reasonable request.