Profiling of SARS‐CoV‐2 neutralizing antibody‐associated antigenic peptides signature using proteome microarray

Abstract The profile of antibodies against antigenic epitopes of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) during neutralizing antibody (NAb) decay has not been clarified. Using a SARS‐CoV‐2 proteome microarray that contained viral antigenic peptides, we analyzed the characteristics of the humoral response in patients with coronavirus disease 19 (COVID‐19) in a longitudinal study. A total of 89 patients were recruited, and 226 plasma samples were serially collected in 2020. In the antigenic peptide microarray, the level of immunoglobulin G (IgG) antibodies against peptides within the S2 subunit (S‐82) and a conserved gene region in variants of interest, open reading frame protein 10 (ORF10‐3), were closely associated with the presence of SARS‐CoV‐2 NAbs. In an independent evaluation cohort of 232 plasma samples collected from 116 COVID‐19 cases in 2020, S82‐IgG titers were higher in NAbs‐positive samples (p = 0.002) than in NAbs‐negative samples using enzyme‐linked immunosorbent assay. We further collected 66 plasma samples from another cohort infected by Omicron BA.1 virus in 2022. Compared with the samples with lower S82‐IgG titers, NAb titers were significantly higher in the samples with higher S82‐IgG titers (p = 0.04). Our findings provide insights into the understanding of the decay‐associated signatures of SARS‐CoV‐2 NAbs.


INTRODUCTION
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has emerged and has been spreading worldwide for more than 3 years. 1SARS-CoV-2 contains a single-strand RNA genome that encodes four structural proteins (the spike [S], nucleocapsid [N], membrane [M], and envelope [E] proteins), 16 nonstructural proteins (NSP1−16), and nine accessory proteins (ORF3a-b, ORF6, ORF7a-b, ORF8, ORF9b-c, and ORF10). 2 Gene site mutations and recombination have generated many SARS-CoV-2 variants, and some of the variants have shown antigenic shifts, causing changes in transmissibility and immune escape. 3,4he level of neutralizing antibodies (NAbs) that is induced by SARS-CoV-2 is critical in evaluating the effects of humoral immunity against viral infections. 5The major epitopes that induce NAbs are within the receptor-binding domain (RBD) and N-terminal domain of the S protein. 6,71][12] The titers of NAbs and IgG antibodies against the SARS-CoV-2-S, N, and RBD proteins decreased during convalescence post-symptom onset (PSO) and remained detectable at 1 year PSO. 13 However, the dynamic signature of antibodies against the SARS-CoV-2 antigenic epitopes and the correlations with NAbs decay remain unclear.
Proteome microarray is an efficient diagnostic tool for the detection of specific antibodies in infectious and autoimmune diseases. 14,15This immunoassay has also been used to analyze B-cell responses, 16 antibody dynamics, 17 and variant-specific epitopes 18 in coronavirus disease 2019 (COVID-19) patients.
In this study, we applied a high-throughput proteome microarray to characterize the repertoire and kinetics of IgG antibodies against SARS-CoV-2 antigenic peptides.The specific peptides related to NAbs seropositivity were also identified and verified.The peptide screening in our study provides candidates for serological assays that can estimate neutralizing capacity post-SARS-CoV-2 infection.

The construction of the proteome microarray against SARS-CoV-2 peptides
To profile the antibodies against SARS-CoV-2 antigenic peptides, we constructed a proteome microarray with peptides designed according to the original SARS-CoV-2 strain (GenBank MN908947.3),as previously reported. 19he peptides used to construct the microarray were 15 amino acids (aa) long with five aa overlap (Figure 1A).The peptides were labeled with a C-terminal biotin group and printed in duplicate onto a three-dimensional modified microscope slide using biotin-streptavidin chemistry.Phosphate-buffered saline (PBS), bovine serum albumin (BSA), and hemagglutinin (HA) peptides were used as negative controls.Biotinylated BSA, human IgG and immunoglobulin M (IgM), and polio peptides were used as positive controls.The lowest limit of detection was 94 pg/mL.The antibody detection efficacy of the proteome microarray was evaluated by comparing it with N-IgG titers in plasma samples from 222 patients with COVID-19 that were measured using the ELISA method.A strong positive correlation was observed between the results derived by using the two methods (Spearman r = 0.73; p < 0.0001), indicating the validity and reproducibility of the proteome microarray (Figure 1B).

Profile of SARS-CoV-2 antibodies in plasma samples using proteome microarray
To profile the antibodies in COVID-19 patients PSO using the proteome microarray, we recruited 89 recovered patients with COVID-19, and a total of 226 plasma samples were separately collected in 2020 in April (baseline), June (follow-up 1 or F1), and October-December (follow-up 2 or F2).The median age of the patients was 47 years (range: 5−80; interquartile range [IQR]: 36−58; Table S1).Three plasma samples were serially collected from 49 patients (Figure S1).Furthermore, 26 samples were collected from 26 healthy individuals (H group) and 16 samples were collected from 16 influenza virus-infected individuals (IFV group, Figure 1C).
The NAbs titer in each patient was tested using the microneutralization method with cultured virus in our previous report. 20The NAbs-positive rate was 57.3% (51 out of 89) in all patients at baseline, decreased to 52.8% (47 out of 89) at F1, and remained at 49.4% (44 out of 89) at F2.The patients were divided into four groups based on NAbs seropositivity at baseline and F2: NAbsnegative (N group; n = 33), NAbs-positive (P group, n = 39), NAbs-positive conversion during follow-up (NP group; n = 5), and NAbs-negative conversion during follow-up (PN group; n = 12; Figures 1C-D).In the PN group, eight individuals were NAbs-negative at F1, and four were negative at F2 (Figure 1C).

Antibody signatures related to the presence of NAbs in patients who had recovered from COVID-19
To gain insight into the antibody repertoire related to the presence of NAbs, the IgG levels in NAbs-positive samples were assessed.We observed that N-IgG, S-IgG, S1-IgG, S2-IgG, and RBD-IgG showed higher Z scores in NAbspositive samples than in negative samples (FDR < 0.0001; Figure 2A).As expected, these IgG antibodies were also dominant (Z score > 1.96) in most NAbs-positive samples but not in NAbs-negative samples (Table S2).
Two peptide-IgG antibodies against S-82 (FDR < 0.0001), and ORF10-3 (FDR < 0.0001) exhibited higher Z scores in NAbs-positive plasma samples than in NAbs-negative samples (Figure 2A).Notably, we noticed the positive correlations of S82-IgG (Spearman r = 0.866, p = 0.003) and ORF10-3-IgG (Spearman r = 0.714, p = 0.03) with NAbs during follow-up in the PN group (Figure 2B).These findings supported the close association of the levels of these two peptides with NAb titers.S82-IgG exhibited the highest mean Z score among all of the S-derived peptides in all of the samples (Figure 2C) and the IgG against this peptide was dominant in 77.9% (102 out of 131) of NAbs-positive samples (Table S2).Based on logistic regression analysis, the Z scores of S82-IgG (β = 0.13, p < 0.0001) and ORF10-3-IgG (β = 4.38, p = 0.002) also supported their significant association with the probability of NAbs seropositivity.These findings suggested that S-82 and ORF10-3 may be antigenic peptide signatures that indicate SARS-CoV-2 NAbs seropositivity following viral infection.
Based on the proteome data before F2, we observed that the levels of S82-IgG (Figure 3C; p = 0.002) and ORF10-3-IgG (Figure 3D; p = 0.03) were lower in patients who were NAbs-negative at F2 than in NAbs-positive patients during follow-ups, suggesting that the detection of the two IgGs were a potential early warning signal for NAbsnegative conversion.Therefore, S82-IgG and ORF10-3-IgG levels may contribute to identifying the presence of NAbs following viral infection.

Kinetics of SARS-CoV-2 antigenic peptides associated with NAbs
A total of 49 patients were sampled three times in our study, allowing us to perform longitudinal observation of the IgG antibodies against SARS-CoV-2 peptides and structural proteins.Linear mixed-effect models showed that the levels of IgG antibodies against 349 peptides declined significantly in the N group, including S-IgG (FDR = 0.01) and RBD-IgG (FDR = 0.02) (Figure 4A).The IgGs against structural proteins (S, N, S1, S2, and RBD) showed nonsignificant changes in the P group (Table S3).Given that the kinetics of IgGs against S-82 and ORF10-3 may impact NAbs seropositivity during convalescence, we focused on the dynamics of these IgGs in patients.ORF10-3-IgG declined significantly in the N group (FDR = 0.003) but not in the P group (FDR = 0.17) (Figure 4A).The decline of S82-IgG was non-significant in the P group (FDR = 0.65) and N group (FDR = 0.34) during the follow-ups (Table S3).However, by using fuzzy c-means clustering, we found that the kinetics of S82-IgG in the patients belonging to the N group and PN group were mainly observed in clusters 2, 3, and 4, all of which exhibited significantly declining trends in IgG levels in serially collected plasma samples (Figure 4B).Our findings suggested that the kinetics of S82-IgG and ORF10-3-IgG were stable with the persistence of NAbs post-viral infection.

Correlation of SARS-CoV-2 S82-IgG levels with NAb titers
To verify the efficacy of S82-IgG on NAb titers, we evaluated S82-IgG levels in two independent evaluation cohorts using ELISA.One cohort included 232 plasma samples collected from 116 recovered COVID-19 patients in June-September 2020 (F1) and December 2020-January 2021 (F2), and 66 plasma samples collected from healthy individuals.NAbs were positive in 98 patients (84.5%) at F1, and 90 patients remained positive (P group), while eight patients were negative at F2 (PN group).Eighteen patients (15.5%) were NAbs-negative (N group; Figure 5A).As expected, the S82-IgG titers were higher in NAbs-positive plasma samples than in NAbs-negative plasma samples (p = 0.002; Figure 5B).
Another cohort included 66 individuals recruited in February 2022 who were vaccinated with inactivated SARS-CoV-2 vaccine and then infected with Omicron BA.1 virus (Figure S3).A total of 62 (93.9%) plasma samples were NAbs-positive against Omicron BA.1.The presence of S82-IgG was tested using ELISA; among the 26 positive plasma samples, 25 (96.2%)samples were NAbs-positive.The NAb titers were significantly higher in the samples where the S82-IgG titers were over 0.80 than those with titers between 0.32 and 0.80 (p = 0.04; Figure 5C).Furthermore, NAb titers were higher than or equal to 100 in 55% (11 out of 20) of the plasma samples in which the S82-IgG titer was between 0.32 and 0.80, and this ratio increased to 100% (six out of six) in the samples with S82-IgG titers over 0.80 (Figure 5D).
In all of the 298 plasma samples from the two evaluation cohorts, the chi-square test indicated that NAbs seropositivity was significantly higher in the S82-IgG positive samples (χ 2 = 10.45,p = 0.001; Figure 5E).The sensitivity of S82-IgG for NAbs seropositivity evaluation was 50.4% and the specificity was 75% (Figure 5E).Moreover, logistic regression analysis showed a positive correlation between the S82-IgG titer and the probability of NAbs seropositivity (β = 1.63, p = 0.008; Figure 5F).These results suggested the correlation of SARS-CoV-2 S82-IgG levels with NAb titers, a finding consistent with that of the proteome assay in our study.In summary, our findings confirmed that S82-IgG levels can reflect the NAbs-associated immune signature.

DISCUSSION
In this study, we profiled antibody signatures post-SARS-CoV-2 infection at the amino acid resolution level by using a proteome microarray containing 966 SARS-CoV-2 antigenic peptides. 19We observed that the levels of IgG antibodies against structural proteins (S, N, S1, S2, and the RBD) in NAbs-positive plasma increased significantly compared with those in NAbs-negative plasma, suggesting higher immunogenicities of these antigens in NAbs-positive patients.Furthermore, two novel IgG The proteome microarray has been used to depict the epitope landscape of the SARS-CoV-2 spike protein in COVID-19 patients. 21Significant decreases in the IgG response against several epitopes (S1-93, S1-97, and S2-78) in non-survivor patients observed, indicating the protective roles of the corresponding antibodies. 22In another study, longitudinal and proteome-wide analyses were used to conclude that the levels of S82-IgG differed significantly between mild and severe COVID-19 patients. 23In our study, we applied the SARS-CoV-2 proteome microarray to profile the specific epitopes related to NAbs decay during convalescence.As expected, strong humoral responses against SARS-CoV-2 structural proteins were associated with the presence of NAbs.Notably, we also identified two antigenic peptides (S-82 and ORF10-3) that can be used to evaluate NAbs seropositivity and decay.This finding revealed an important role of the S-82 peptide in reflecting the NAb response post-SARS-CoV-2 infection.NAb titers have been shown to increase over time in parallel with the rise in IgG antibody levels. 9However, the effective signatures of humoral responses that could allow the evaluation of SARS-CoV-2 NAbs seropos-itivity post-infection have remained unclear.The levels of IgG against structural proteins (S1, N, and the RBD) have shown different predictive effects on NAb titers postinfection and vaccination. 8,24Our study revealed that the combination of S82-IgG and ORF10-3-IgG effectively estimated NAbs decay during follow-ups, providing insight into the relationship between these antigenic peptides and neutralizing capacity and durability against SARS-CoV-2.Moreover, longitudinal observation revealed a stable relationship between S82-IgG and ORF10-3-IgG with the persistence of NAbs.These findings suggested the stable role of the peptide-IgGs in evaluating NAbs seropositivity.
The IgM antibody against S-82 has been reported to be enriched in patients with COVID-19. 25This 15-mer peptide is located within the fusion peptide (FP) domain, which is partially exposed on the surface of the spike protein, including in SARS-CoV-2 variants of concern. 26he conserved residue was observed to be engaged in electrostatic interactions with several FP-specific monoclonal antibodies (VN01H1 and C77G12) that exhibited heterogeneous neutralizing activities and reduced viral burden in vivo. 27Moreover, two broad-spectrum NAbs (COV44-62 and COV44-79) that could neutralize alpha and beta coronaviruses including the Omicron BA.1 and BA.2 subvariants were mapped to the S2' cleavage site using a surface plasmon resonance-based high-throughput peptide array. 28A previous report suggested that mice immunized with a peptide that included the S2'site and the FP region (aa 816−826) did not exhibit significant neutralization activity. 22The positive correlation between the level of S82-IgG and the NAbs titer observed in our study indicated the potential neutralizing capacity induced by this peptide.Moreover, the decline of IgG levels against this antigenic peptide may reflect the NAbs decay observed during follow-up visits.
Our study had some limitations.First, the sample size was limited; therefore, the effectiveness of S82-IgG and ORF10-3-IgG in the evaluation of NAbs seropositivity should be confirmed with a large and long cohort study of SARS-CoV-2 variants of concern.Second, an ELISA for ORF10-3-IgG was not performed because of ORF10-3-IgG's high hydrophobicity; therefore, the efficacy of ELISA in the analysis of NAbs must be investigated further.Third, the sensitivity and specificity of the proteome microarray should be confirmed given that the microarray contained only linear epitopes; further studies are needed to investigate the neutralization potential of SARS-CoV-2 conformational epitopes.

CONCLUSION
In this study, we used a proteome microarray to characterize IgG antibody signatures against antigenic epitopes post-SARS-CoV-2 infection.We observed robust humoral responses against SARS-CoV-2 structural proteins and antigenic domains after viral infection.Furthermore, IgG kinetics after infection were correlated with epitope sites and NAb titers.We identified two novel peptide-IgG antibodies against the SARS-CoV-2 spike S2 subunit-S-82 (aa 811−825) and ORF10-3 (aa 21−38)-as effective NAbs-associated immune signatures.Our study provides insights into the humoral immune response against SARS-CoV-2 infection and proposes novel candidates for the evaluation of SARS-CoV-2 NAbs.

Participants and samples
The plasma samples were collected from Wuhan, China.The sampling procedure, during which SARS-CoV-2 infections were diagnosed in accordance with the Chinese clinical guidance for COVID-19 pneumonia diagnosis and treatment, 29 has been described in our previous report on viral seroprevalence. 20Demographic information for all of the participants is shown in Table S1.The patients were diagnosed with symptomatic or asymptomatic infection according to whether they had a self-reported fever or respiratory symptoms (including but not limited to cough, anhelation, stuffy nose, rhinorrhea, sore throat, pneumonia, or both) during clinical care.The plasma samples from all patients were collected on April 14−15 (baseline), June 11−13 (follow-up 1 or F1), and October 9-December 5, 2020 (follow-up 2 or F2), when participants were recruited and venous blood samples were collected.Sixteen plasma samples collected from 16 healthy volunteers and 26 plasma samples collected from 26 individuals who recovered from influenza virus infection were used as controls (Table S4).Two validation cohorts were involved.As the test cohort for S82-IgG ELISAs, we recruited 116 individuals who had recovered from confirmed COVID-19 in Wuhan, China.The median age in the test cohort of patients with COVID-19 was 57 years (range: 28−95; IQR: 49−65; Table S5).Two paired plasma samples (total = 232) were collected from each patient in June-September 2020 (F1) and December 2020-January 2021 (F2).The number of SARS-CoV-2 NAbs-positive samples was 188, and the number of negative samples was 44.We also collected 66 plasma samples from 66 additional healthy individuals in Wuhan (Table S5).
Another validation cohort included 66 individuals who had been vaccinated with inactivated SARS-CoV-2 vaccine and then infected with Omicron BA.1 virus in Tianjin, China.The median age of these cases was 40 years (range: 18−69; IQR: 33−56; Table S6).In total, 66 plasma samples were collected from patients on February 11−23, 2022.
All plasma samples were inactivated at 56 • C for 30 min before use.Recombinant N protein was used to evaluate the antibodies against SARS-CoV-2.NAb titers were assessed at each visit using in-house microneutralization assays for the original SARS-CoV-2 strain or the Omicron BA.1 strain. 20

Detection of viral antibodies using a SARS-CoV-2 proteome microarray
The peptide microarrays were assembled in an incubation tray and blocked with 5% (weight-bulk ratio) milk/(PBS

1
Immunoglobulin G (IgG) antibodies against SARS-CoV-2 in recovered patients with COVID-19 were analyzed using a proteome microarray.(A) Schematic diagram of the proteome microarray used for the detection of SARS-CoV-2 peptide-IgGs.(B) Spearman's correlation for the ELISA and proteome microarray.The X-axis indicates the rank of N-IgG fluorescence intensity for each sample detected using proteome microarray.The Y-axis indicates the rank of N-IgG absorbance for each sample detected by ELISA.(C) Research design and participants' information.The plasma samples were collected from patients who had recovered from COVID-19 during three visits.(D) Neutralizing antibody (NAb) titers against SARS-CoV-2 in patients with COVID-19 during the three visits.The box outlines represent the 25th-75th percentiles and the middle lines indicate the median values.The whiskers indicate 1.5 times the interquartile range (values greater than or lower than the extremes were regarded as outliers).F1, Follow-up 1; F2, Follow-up 2; N, NAbs-negative; P, NAbs-positive; PN, NAbs-negative conversion during follow-up; NP, NAbs-positive conversion during follow-up.

2
Humoral immune features in patients who had recovered from COVID-19.(A) The IgG differences between neutralizing antibodies (NAbs)-positive and NAbs-negative samples.The middle points indicate the mean Z score of each IgG, and the upper and lower lines indicate the 95% confidence interval (CI).(B) The Spearman coefficients of IgG changes and NAbs titer decay in the patients with NAbs-negative conversion (PN group).(C) The mean Z scores of S peptide-IgG antibodies in all of the samples.S82-IgG exhibited the highest Z score across all S peptide-IgGs.The amino acid sequence and location of S-82 are shown in the upper right.

3
The potency of S-82 and ORF10-3 immunoglobulin G (IgG) for SARS-CoV-2 neutralizing antibody (NAb) seropositivity evaluation.(A) Receiver operating characteristic (ROC) curve of logistic regression to evaluate NAbs seropositivity in the testing set using S82-IgG and ORF10-3-IgG.The X-axis indicates the false positive rate, and the Y-axis indicates the true positive rate.ROC curves (blue) were generated after 100 runs of computational cross-validation and the mean ROC curve (red) was generated by averaging all of the ROC curves.The mean and 95% confidence interval (CI) of the areas under the curves (AUCs) after 100 runs of computational cross-validation were also calculated.(B) The mean accuracy of different IgG combinations for SARS-CoV-2 NAbs seropositivity evaluation in support vector machine classifiers with 10-fold cross-validation.The two IgGs combined in the classifier were referred to Antibody 1 and Antibody 2. (C and D) The level of IgGs against S-82 (C) and ORF10-3 (D) at the last visit before F2 across groups.The dots indicate the IgG absorbance in the proteome microarray.The box outlines represent the 25th-75th percentiles and the middle lines indicate the median values.The whiskers indicate 1.5 times the interquartile range (values greater than or lower than the extremes were regarded as outliers).The p values in the Kolmogorov-Smirnov test are shown.OD 532 = fluorescent signal intensity at 532 nm.

4
Immunoglobulin G (IgG) kinetics in patients who had recovered from COVID-19.(A) The decline rates of IgG antibodies in patients with COVID-19.The X-axis indicates the gradient of the linear mixed-effects model and represents the fluorescence intensity decline per day.The Y-axis indicates the false discovery rate (−log10).(B) Fuzzy c-means clustering of S82-IgG kinetics in patients with COVID-19.Clusters 1−6 represent the different S82-IgG kinetics patterns after clustering.The colors represent the membership values of the lines in each cluster, which gradually change from grey to red.The X-axis indicates the median day at each visit, and the median day at baseline refers to Day 0. The Y-axis indicates normalized intensity at each visit.antibodies against antigenic peptides in the S (S-82, 811−825 aa) and ORF10 (ORF10-3, 21−38 aa) proteins also exhibited strong associations with NAb titers and are potential early warning signals of SARS-CoV-2 NAbs decay.

F I G U R E 5
The association of S82-IgG and neutralizing antibodies (NAbs) in ELISA validation.(A) The NAb titers from an independent evaluation cohort involving 232 plasma samples collected from 116 COVID-19 cases in 2020.N group, NAbs-negative patients; P group, NAbs-positive patients; PN group, patients with NAbs-negative conversion.(B) S82-IgG titers in NAbs-positive samples versus NAbs-negative samples.The dots in the violin plot indicate the S82-IgG titers in ELISA and the middle line represents the median value.The p value in the Mann-Whitney U test is shown.(C) The titers of NAbs against Omicron BA.1 subvariant in the groups with differing S82-IgG titers.The dots indicate the titers of NAbs against Omicron BA.1 subvariant.The box outlines represent the 25th-75th percentiles and the middle lines indicate the median values.The whiskers indicate 1.5 times the interquartile range (values greater than or lower than the extremes were regarded as outliers).The p values in the Mann-Whitney U tests are shown.(D) The ratio of individuals with different titers of NAbs against Omicron BA.1 subvariant in the groups with differing S82-IgG titers.(E) The number of NAbs-positive and NAbs-negative plasma samples (N = 298) across the groups based on S82-IgG seropositivity in two independent evaluation cohorts using ELISA.(F) The logistic regression for S82-IgG titer and the probability of NAbs seropositivity in plasma samples (N = 298) from two independent evaluation cohorts.The X-axis indicates the S82-IgG titer tested using ELISA.The Y-axis indicates the probability of NAbs seropositivity.