Humoral immune responses during SARS-CoV-2 mRNA vaccine administration in seropositive and seronegative individuals

We enrolled healthcare workers from our children’s hospital prior to the administration of the Pfizer BNT162b2 COVID-19 vaccine. Peripheral blood was collected before vaccination at baseline (week 0), after primary immunization (week 3), and secondary booster immunization (week 7) from healthcare workers who had no known history of infection (N = 152) or previous laboratory-confirmed SARS-CoV-2 infection 30–60 days (N = 42) prior to the administration of the BNT162b2 vaccine (administered at week 0 and week 3). This population consisted of mostly adult middle aged, white, females who did not identify as Hispanic or Latino (Supplementary Table 1). The SARS-CoV-2 vaccine specimens were collected at Children’s Mercy Kansas City and were reviewed and approved by the Children’s Mercy IRB. Serum or plasma was utilized to perform the immunoassays that were isolated from venous whole blood collection and stored in frozen in ultra-low temperature freezers until use.

To measure antibody levels to SARS-CoV-2 spike subunit proteins (spike subunit 1 (S1), spike subunit 2 (S2), receptor-binding domain (RBD)) and nucleocapsid protein (NP) antigens, we utilized a bead-based multiplex assay based on the Luminex xMAP technology using reagent kits that had secondary antibodies that were specific for immunoglobulin isotypes (IgG, IgM, IgA). We used the following kits: IgG (Millipore, #HC19SERG1-85 K), IgM (Millipore, #HC19SERM1-85 K), and IgA (Millipore, #HC19SERA1-85 K) following standard manufacture protocols. Each kit provided the same sets of SARS-CoV-2 antigen conjugated beads (S1, S2, RBD, NP) along with 4 positive control beads and a negative control bead set. The positive control beads were beads coated with different concentrations of IgG, IgM, or IgA (depending on the isotype kit utilized). The negative control beads did not have antigen conjugated to determine nonspecific binding. The 4 antigen-conjugated beads, 4 positive control beads, and 1 negative control beads were mixed and incubated with each plasma sample that was diluted 1:100 with assay buffer. With each assay plate, at least two sample wells with only buffer and no plasma were included to determine assay background. Finally, PE-anti-human IgG, IgM, or IgA conjugate detection antibodies were utilized to determine antibody isotype responses to each of the SARS-CoV-2 antigens. Using the positive control beads, we determined the inter-assay (plate-to-plate) coefficient of variation (CV) for each assay. We determined that the CVs were 5.16%, 7.42%, and 11.45% for the IgG, IgM, and IgA assays, respectively. In order to acquire and analyze data, we utilized the Luminex analyzer (MAGPIX) and Luminex xPONENT acquisition software. Samples were run in technical duplicate and after acquisition Net MFI was utilized which is MFI with background well (no plasma) MFI subtracted. Positive control beads were utilized to ensure positive detection of the well and to identify any inter- and/or intra-assay technical variation. We next determined the level of nonspecific binding by using the negative control samples MFI (beads without antigen mixed with plasma) for each isotype (IgG, IgM, IgA). The IgG had the highest background MFI, so we used the average MFI plus the standard deviation of the IgG samples to set the detection threshold for IgG, IgM, and IgA isotype assays (442 MFI).

To detect viral neutralizing antibodies the SARS-CoV-2 Surrogate Virus Neutralization Test kit was utilized (Genscript, #L00847) according to the standard protocol. Samples were run in duplicate with blocking values averaged. This kit detects antibodies that can block the interaction between the receptor-binding domain of the viral spike glycoprotein with the angiotensin-converting enzyme 2 (ACE2) cell surface receptor and has been approved by the FDA for emergency use. Plasma samples along with positive (anti-RBD antibody) and negative (buffer only) were incubated with a Horseradish peroxidase (HRP) conjugated recombinant SARS-CoV-2 RBD fragment. The mixture was then added to a capture plate that was coated with the human ACE2 protein. The unbound HRP-RBD will bind to the plate. After washing, 3,3′,5,5′-tetramethylbenzidine (TMB) solution was added to develop the HRP signal and was read at 450 nm in a microtiter plate reader. The absorbance of the sample is inversely dependent on the titer of the anti-SARS-CoV-2 neutralizing antibodies. Inhibition was calculated by (1 − OD value of sample/OD value of negative control) × 100 which gives percent inhibition. A cutoff of ≥ 30% is considered positive for SARS-CoV-2 neutralizing antibody. Plasma samples were diluted 1:10 for all samples and 1:100 for a more dilute assay of seropositive samples at week3 and week 7 timepoints.

Plasma samples were diluted 1:200 and used to probe a single SARS-CoV-2 protein and peptide microarray (CDI Labs). After probing arrays with serum antibodies, the arrays were washed, labeled with an Alexa647-anti-human IgG Fc secondary antibody and scanned using a GenePix 4000B scanner. Array data was collected using the MAGPIX software (Innopsys). 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 graphing. Z-scores were calculated across all peptides for each individual prior to t test analysis with Bonferroni adjustment for multiple comparisons.

Antigens: SARS-CoV-2 spike protein RBD (Genscript # Z03483), SARS-CoV2 nucleocapsid protein (Genscript #Z03488), SARS-CoV-2 S1 subunit spike protein (Genscript #Z03501), SARS-CoV-2 S2 subunit spike protein (R&D Systems #10594-CV) were all diluted to 1 μg/mL in 0.1 M sodium bicarbonate and incubated on high-binding plates (corning #3369) overnight at 4 degrees. Serum was diluted to 1:30 in superblock buffer with sodium azide followed by subsequent 1:3 dilutions until a final dilution of 1:21870 was reached. Secondary antibodies were purchased from Southern Biotech: Mouse Anti-Human IgG1 Fc-HRP (#9054-05), Mouse Anti-Human IgG2 Fc-HRP (#9060-05), Mouse Anti-Human IgG3 Hinge-HRP (#9210-05), Mouse Anti-Human IgG4 Fc-HRP (#9200-05). Secondary antibody dilutions were done in superblock buffer without sodium azide within range of manufacturers recommendations: IgG1 1:6000, IgG2 1:5000, IgG3 1:7000, IgG4 1:6000. SureBlue Reserve Microwell Substrate (VWR #95059-294) was added to each well 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.

All statistical analyses were performed using Prism 8.0 (GraphPad Software Inc.) software and R. For group comparisons, an unpaired nonparametric Wilcoxon-Mann-Whitney test was used with a two-tailed P value reported. We used a P value significance threshold of P < 0.05. Both the Wilcoxon-Mann-Whitney and descriptive statistical analysis such as mean, median, and range were calculated using non-transformed data. For the peptide microarray data, the log2 fluorescence intensity was used for the heatmap and these values were transformed into Z-score based on row (sample) for the Z-score plotting of the group average. We selected Z-score values > 1 (one standard deviation above mean) to identify immunodominant peptides within the vaccine group.

Peripheral blood was collected before vaccination at baseline (week 0), after primary immunization (week 3), and secondary booster immunization (week 7) from individuals who had no known history of infection (seronegative; N = 152) or laboratory-confirmed SARS-CoV-2 infection 30–60 days (seropositive; N = 42) prior to the administration of the BNT162b2 vaccine (administered at week 0 and week 3). This population consisted of mostly adult middle aged, white, females who did not identify as Hispanic or Latino (Additional file 1: Table S1).

We first measured immunoglobulin G (IgG) antibody levels at the three timepoints (week 0, week 3, and week 7) to SARS-CoV-2 spike subunit proteins (spike subunit 1 (S1), spike subunit 2 (S2), receptor-binding domain (RBD)) and nucleocapsid protein (NP) using a multiplexed bead-based binding assay. As expected at baseline, the seronegative group had low median fluorescence intensity (MFI) for all four antigens (S1, 16.8; S2, 886.7; RBD, 218.2; NP, 482.9) whereas the seropositive group had significantly higher median antibody binding levels (S1, 10793.3; S2, 12923; RBD, 16409.7; NP, 20766). Cross-reactivity of pre-existing antibody immunity has been reported in the absence of SARS-CoV-2 infection [20‐24]. In parallel, we analyzed a group of individuals that had peripheral blood collected prior to the emergence of SARS-CoV-2 and the COVID-19 pandemic in the year 2019 (Additional file 1: Figure S1A). We found that the median MFI of the groups (S1, 20.5; S2, 732; RBD, 331.3; NP, 759.8) in the pre-pandemic individuals were similar to the baseline levels of the seronegative group prior to vaccination. Interestingly, the MFI for RBD was consistently higher than S1 in the prepandemic samples and further study into the protein background levels or conformation-dependent nature of the antibody response should be performed. Thus, MFI levels significantly above the seronegative individuals at baseline or the individuals collected prior to the pandemic would represent elevation in antibody levels due to vaccination or more recent SARS-CoV-2 infection. The lowest median MFI in the seronegative group was to the S1 protein, with a median MFI less than 20.5 for the seronegative vaccine group or group sampled prior to the COVID-19 pandemic (Additional file 1: Figure S1). We identified six individuals at baseline who had S1 antibody levels MFI ≥ 1000, similar to the levels observed at baseline in the seropositive group. These individuals also had higher levels of antibody binding to the other SARS-CoV-2 antigens tested and may have had undiagnosed or asymptomatic infection prior to vaccination (Additional file 1: Figure S1B). After primary immunization (week 3), both seronegative and seropositive individuals had significantly increased IgG antibody titers to all three spike subunit antigens when compared to baseline (S1, S2, RBD; P < 0.0001; Fig. 1A). Furthermore, seropositive individuals had significantly higher titers to all spike antigens when compared to the seronegative group at week 3 after primary immunization (P < 0.0001; Fig. 1A). The undiagnosed individuals resembled seropositive participants in week 3 serological assays and therefore we added them to the seropositive group for subsequent analysis at week 7 (green points). While the NP is contained in the whole virus, only the SARS-CoV-2 spike protein is a vaccine antigen. As expected, we found higher NP antibody titers only in the group with prior natural infection (seropositive), and there were no significant changes in NP antibody levels after vaccination in either group (seropositive or seronegative) (Fig. 1A). Four weeks after the second vaccine dose (week 7), seronegative individuals had significantly increased antibody titers to all 3 spike subunit proteins when compared to week 3 (P < 0.0001), whereas seropositive individuals had significant increases to S1 and S2 but not the RBD, with smaller magnitudes of increases compared to week 3 levels. Antibody titers in seropositive individuals at week 3 were comparable to week 7 titers in seronegative individuals for S1 and RBD; however, seropositive individuals had significantly higher S2 antibody levels at both week 3 and week 7 when compared to seronegative individuals (P < 0.0001; Fig. 1A).

We next used an assay that allows indirect detection of SARS-CoV-2 neutralizing antibodies through determination of antibody blocking of SARS-CoV-2 RBD binding to the human host viral receptor angiotensin-converting enzyme 2 (ACE2). This assay has been demonstrated to have a significant correlation with levels of neutralizing antibodies determined by primary virus or pseudovirus neutralization assays [25]. Levels of blocking antibodies were significantly increased after primary vaccination (week 3) for both seropositive and seronegative groups, and the seropositive group had significantly higher blocking antibodies when compared to the seronegative group (P < 0.0001; Fig. 1B). The second vaccine dose only significantly increased blocking for the seronegative group at week 7 (P < 0.0001; Fig. 1B). There was no significant difference in blocking antibodies in the seropositive group after the first vaccine dose (week 3) when compared to the seronegative group after the second vaccine dose (week 7). The median percent blocking for the seropositive samples was 96.3% at week 3 and 97.9% at week 7. We further performed an additional 10-fold dilution of the plasma samples in order to determine if there were any increases in the magnitude of neutralizing antibodies in seropositive individuals after the second vaccine dose. Using this plasma dilution, we found that there was an increase in the seropositive group from week 3 (median 89.4%) to week 7 (median 93.4%) that was modest in magnitude but statistically significant (P = 0.02; Additional file 1: Figure S2). The threshold for positive detection of SARS-CoV-2 neutralizing antibodies has been determined to be ≥ 30% for this assay. Using this threshold as a marker of seropositivity after vaccination, there were 86.1% and 97.5% of individuals who had antibody levels above the threshold after the first immunization in the seronegative and seropositive/undiagnosed groups, respectively (Fig. 1B). After the second dose, nearly all participants in both groups had positive neutralizing antibody titers (seronegative: 99.2%; seropositive: 100%; Fig. 1B). These results demonstrated that individuals with prior SARS-CoV-2 infection have higher magnitudes of binding and neutralizing antibodies after a single dose that is equivalent to levels observed in seronegative individuals after two vaccine doses. Moreover, seropositive individuals only receive a minor increase in antibody binding magnitude after the second dose.

Next, we determined the contribution of sex and age to vaccine neutralizing antibody responses by stratifying the week 3 blocking assay results by sex and age. After the first vaccine dose (week 3), no significant differences in blocking antibody levels based on sex were detected (Fig. 2). However, we did observe that older individuals (≥ 50 years of age) in the seronegative group had significantly lower blocking antibodies compared to younger individuals (< 50 years of age) after primary immunization, but no significant difference in age was observed for the seropositive group (P = 0.0003; Fig. 2).

While IgG isotype antibodies are the major surrogates of vaccine and infection-mediated immunity, the kinetics and contributions of IgM and IgA isotypes are less clear. We found that IgM antibody responses to S1, S2, and RBD were only significantly boosted after immunization in the seronegative individuals, whereas the antibody levels in seropositive individuals were not boosted and even waned by week 7 (Fig. 3A). However, IgM levels at baseline in the seropositive individuals remained significantly higher than levels elicited by two doses of vaccine (week 7) in seronegative individuals (Fig. 3A). Similarly, IgA antibody levels were only increased by vaccination in seronegative individuals, with no significant change during vaccination in seropositive individuals (Fig. 3B). Only IgM S1 antibody levels had a significant increase in seronegative individuals after the second dose, indicating a limited contribution of secondary immunization to IgM and IgA antibody responses (Fig. 3A, B). There was no change in seronegative or seropositive individuals IgM or IgA titers against NP over the course of vaccination as expected (Fig. 3A, B). These results indicated that immunization with SARS-CoV-2 vaccines could elicit IgM and IgA isotype antibodies in seronegative individuals, but not for individuals with a previous history of infection.

There are four IgG subclasses (IgG1, IgG2, IgG3, IgG4) that differ in efficiencies in recruiting immune effector cells [26, 27]. In a recent study, it was determined that SARS-CoV-2-infected individuals make IgG1 and IgG3 subclass antibody responses to the spike RBD with little contribution from IgG2 or IgG4 subclasses [28]. Using IgG subclass-specific enzyme-linked immunosorbent assays (ELISAs), we found that in 24 SARS-CoV-2-infected individuals at baseline before immunization, IgG1 and IgG3 were the dominant subclasses for spike protein S1, S2, and RBD subunits and the NP (Fig. 4A; Additional file 1: Figure S3). Only two individuals made low titer IgG2 responses, and no individuals had detectable IgG4 antibody responses (Fig. 4A; Additional file 1: Figure S3). Similarly, after primary immunization with SARS-CoV-2 vaccine, seronegative individuals displayed the same pattern of subclass response to the spike RBD when compared to seropositive individuals at baseline, with IgG1 and IgG3 levels being the highest. These results indicated that similar class-switch antibody responses to SARS-CoV-2 spike protein occurred after both vaccination and natural infection (Fig. 4B; Additional file 1: Figure S3).

Finally, we selected 14 seronegative individuals after primary immunization with the SARS-CoV-2 vaccine (week 3) and determined linear epitopes recognized by humoral immune responses using a SARS-CoV-2 spike peptide microarray that contained 196 soluble 12-mer overlapping peptides (Additional file 2). Heatmaps using the Log2 MFI values for each peptide were displayed for each spike subunit (Fig. 5; Additional file 3). Using a threshold of mean Z-score greater than 1 as “immunodominant,” we found that a peptide in the RBD (S1-61) was the most immunodominant in the S1 region, targeted by antibodies from most individuals tested (Fig. 5A; Additional file 4). There were also three other peptides in the RBD that had high antibody binding levels (S1–66, S1–72, S1–76), with only one occurring in the region identified to make critical contacts with host cell receptor ACE2 (S1-76; Fig. 5A). Outside of the RBD, there were also peptides highly recognized by antibodies within the N-terminal domain (NTD; S1–24, S1–34, S1–45) and near the S1/S2 cleavage site (S1–111, S1–105, S1–97; Fig. 5A). In the S2, the antibody response was mainly targeted to four peptides in the C-terminal region near the heptad repeat 2 (HR2), transmembrane, and cytoplasmic domains (S2–78, 21–81, S2–83, S2–94), but there was a single peptide in the heptad repeat 1 (HR1) domain that was also immunodominant (S2–47; Fig. 5B, Additional file 4). This data identified regions in the spike protein that are commonly recognized by antibodies from many SARS-CoV-2 vaccinated individuals and may reveal epitopes critical for vaccine-mediated protection.