3.1. IgM and IgG responses to RBD and NP proteins
Serum levels of IgM and IgG against RBD and NP proteins were determined in patients’ sera (n=97) by ELISA (Fig. 1). Sera from 23 healthy individuals were used as controls. Our data demonstrated that based on assigned cut-off OD values (cut-off =mean ±2SD of normal individuals) anti-RBD and anti-NP IgG were positive in 94% and 92%. and anti-RBD and anti-NP IgM were positive in 90% and 80% of patients’ sera, respectively. Subsequently, 66 serum samples from patients with IgG titer higher than the cut-off level (0.33 for anti-RBD IgG and 0.43 for anti-NP IgG) were selected for further Pepscan experiments.
Identification of linear epitopes on the SARS-CoV-2 RBD and NP proteins
We next assessed the linear dominant antigenic determinants within the RBD domain of the spike protein (319-541 aa), NP protein (1-419 aa), and three different fragments within the S2 domain of the spike protein, namely FE-18 (802-819 aa), FI-22 (888-909 aa) and CK-16 (1254-1269 aa), by Pepscan analysis. A total of 66 serum samples from COVID-19 patients were selected for epitope mapping which had been verified to be reactive to the target proteins by ELISA. Serum samples from 23 healthy subjects were also included as negative controls. A set of 20-aa overlapping peptides spanning the entire RBD domain and NP protein in pools of 3 adjacent peptides were used as coating antigens in ELISA (Table. 1), and serum samples were tested against each of the peptides. To test whether the reactivity is proportional to linear epitopes or conformational ones, reduced as well as native proteins were also used as coating antigens, and all sera were assessed with these antigens as well.
  1. Reactivity against native and reduced SARS-CoV-2 RBD proteins Native and reduced SARS-CoV-2 RBD proteins were employed to assess the reactivity of RBD-specific IgG antibodies with cysteine bonds dependent and independent epitopes. Significantly higher reactivity was observed with native non-reduced protein as compared to the reduced preparation (p=0.0002 ) (Fig. 2).
  2. Reactivity against RBD peptide pools While all of the 66 serum samples from COVID-19 patients were reactive against native RBD protein, less than 40% of these samples recognized peptide pools A to D, and pool E alone reacted with about 40% of patients’ sera. On the other hand, reactivity with all RBD peptide pools was significantly higher in patients’ sera than healthy controls (Fig. 3A).
  3. Identification of the immunodominant peptide in pool E Since patients’ sera showed better reactivity with pool ”E” of RBD peptides, we evaluated the reactivity of 25 pool ”E” positive serum samples with individual peptides of this peptide pool. Most samples reacted either with peptide P204-223 or 181-200 of RBD (Fig. 3B).
  4. Depletion assay confirmed specific reactivity with P204-223 on SARS-CoV-2 RBD protein Further assessment was performed to verify specific reactivity of RBD-specific antibodies from 10 patients to P204-223 peptide from RBD protein (Fig. 4). Patients’ sera were adsorbed with P204-223 as the most reactive peptide and P196-215 as the non-reactive peptide, and evaluated by ELISA against native RBD (Fig. 4A). Two of these samples were also tested by Western blotting (Fig. 4B). The reactivity of non-adsorbed sera against RBD was slightly higher in comparison with that of P204-223 adsorbed sera (p=0.274 ). It confirms that the antibody response against RBD is largely related to the conformational epitopes.
  1. Identification of linear epitopes on the SARS-CoV-2 NP protein
  2. Reactivity against native and reduced SARS-CoV-2 NP proteinsNative and reduced SARS-CoV-2 NP proteins were employed to assess the reactivity of NP-specific antibodies with cysteine bonds dependent and independent epitopes. The reactivity to native NP was similar to the reduced NP (p=0.206 ) (Fig. 5), suggesting that denaturation does not significantly alter recognition of the NP protein by patients’ antibodies.
  3. Reactivity against NP peptide pools Pepscan analysis of peptide pools covering the entire NP protein against sera obtained from COVID-19 and healthy controls also revealed significantly higher reactivity of patients’ sera than healthy controls in each block of peptides (Fig. 6A). Most pools of peptides were recognized by less than 50% of patients’ sera, while one distinct antigenic site corresponding to aa 136 to 185 in the N-terminal domain of NP protein (pool I) reacted very strongly with more than 75% of the serum samples from COVID-19 patients. This suggests that the identified region, namely pool I, contains at least one of the major linear immunodominant epitopes that induces the antibody response in COVID-19 patients. The results also indicate that the reactivity against pool I was collectively weaker than reactivity to native NP protein (p<0.001) (Fig. 6A), which may reflect the contribution of other potential conformational epitopes in native NP.
  4. Identification of the immunodominant peptide in pool I Next, we further assessed individual peptides within pool I from NP protein which includes aa 136-155, 151-170, and 166-185, to precisely determine the exact 20-mer epitope which attributes to the highest reactivity of COVID-19 patients’ sera (Fig. 6B). Sera obtained from 25 patients were used in this experiment. The data revealed that peptide P151-170 is the dominant hit from pool ”I”.
  5. Depletion assay confirmed specific reactivity with P151-170 on SARS-CoV-2 NP protein Further assessment was performed to verify the specific reactivity of NP-specific antibodies from 10 patients to the dominant P151-170 peptide from NP protein (Fig. 7). Patients’ sera were individually adsorbed with P151-170, as well as P136-155 as the non-reactive peptide, and evaluated by ELISA against native NP (Fig. 7A). Two of these samples were also tested by Western blotting (Fig. 7B). The data revealed that peptide P151-170 adsorbed sera have decreased reactivity (p=0.0156 ) with native NP in comparison with non-adsorbed sera.
  6. Identification of linear epitopes on the SARS-CoV-2 S2 proteinThree peptides of S2 domain, including FE-18 (aa 802-819), FI-22 (aa 888-909), and CK-16 (aa 1254-1269) were selected to assess serum reactivity of 66 patients and 23 normal subjects. Almost 50% of patients’ sera reacted with these three peptides (Fig. 8), among which CK-16 displayed the highest reactivity.
Discussion
Identification of major antigenic determinants of SARS-CoV-2 proteins which provoke remarkable antibody response in COVID-19 patients may provide valuable information for understanding the virus-neutralizing antibody response and developing efficient vaccines and serological assays.
Here, we have mapped the linear immunodominant epitopes on SARS-CoV-2 NP and RBD proteins by Pepscan analysis using 20-aa overlapping peptides spanning the whole sequence of both proteins. We showed that Iranian patients with COVID-19 develop significant anti-RBD and anti-NP IgG and IgM antibody responses. However, while the antibody response against RBD seems to be largely raised against the S-S bond-dependent conformational determinants, the antibody response against NP protein is mostly directed linear epitopes.
Despite strong serological reactivity of convalescent patients against RBD reported in studies using full-length RBD antigen of SARS-CoV-2 (16, 17), reports on serological reactivity to ”linear” immunodominant sites on SARS-CoV-2 RBD are very limited in the literature. Although Zhang et al. reported four linear immunodominant sites on RBD detected by sera from COVD-19 patients (18), we and others observed low reactivity to peptides designed from the RBD of SARS-CoV-2. Weak serological reactivity to peptides within the RBD was also observed in a recent study using a highly multiplexed peptide assay (PepSeq) platform by Ladner et al. (19). Meng Poh and colleagues using a pepscan analysis, failed to find a linear immunodominant epitope exactly localized to RBD region with high reactivity to patients’ sera (20). This weak/ lack of reactivity in peptide-based approaches implies that antibodies reactive to RBD region are largely directed against conformational epitopes and/or epitopes generated by post-translational modifications. Pepscan analysis in our assay and peptide-based antibody assays exclusively detects linear epitopes created by the primary structure of proteins and are blind to epitopes generated in the secondary or tertiary structure of proteins. Interestingly, the reduction of S-S bonds within the native RBD molecule by a reducing agent resulted in a significant reduction of antibody reactivity in patients’ sera (Fig. 2).
To assess the linear epitopes which provoke anti-RBD antibody response, we performed Pepscan analysis using 20-aa long overlapping peptide covering the entire RBD domain. Although antibody reactivity against 5 different peptide pools, each consisting of 3 peptides, was significantly higher in sera from patients compared to healthy subjects, it was significantly lower when compared with reactivity to the native RBD protein (Fig. 3A). Of the 5 peptide pools, pool ”E” which covers the C-terminal residues of RBD (aa 181-223) displayed the highest reactivity. Thus, we investigated the antibody response to the three peptides of this pool in 25 selected patients’ sera to identify the most immunodominant one. Both peptides P181-200 and particularly P204-223, but not peptide P196-215, displayed modest reactivity with some of the samples (Fig. 3B). Preincubation of serum samples from 10 of these patients with peptide P204-223 resulted in slightly lower reactivity of these samples to native RBD by ELISA and Western blot (Fig. 4). All these findings suggest that the antibody response to RBD is dominated by S-S bond-dependent conformational epitopes. The fact that the immunogenicity of RBD mainly relies on the conformational structures and/or post-translational modifications (i.e., glycosylation) is proved by studying the binding footprint of neutralizing monoclonal antibodies that inhibit RBD binding to ACE2. Studies revealed residues that are distal in the linear sequence of RBD and their presence and glycosylation state contributed to antibody binding (21, 22).
We adopted a similar approach to investigate the antibody response against the NP protein. SARS-CoV-2 NP is a 419 aa phosphoprotein that associates with the viral RNA genome as well as other proteins to form the ribonucleoprotein core (23). Like SARS-CoV, NP protein of SARS-CoV-2 consists of three distinct domains: an N-terminal RNA-binding domain (NTD) which associates with the RNA genome, an intrinsically disordered central Ser/Arg (SR)-rich linker and a C-terminal domain (CTD) which allows dimerization of NP proteins (24, 25).
Our results revealed that in contrast to RBD, NP-specific antibody response is mainly directed against linear epitopes. No significant difference was observed between serum levels of antibody against reduced and non-reduced NP protein (Fig. 5). This was also supported by the Pepscan data which showed substantially higher reactivity of the anti-NP antibodies from patients’ sera with all peptide pools, particularly peptide pool ”I”, which covers aa 136-185 (Fig. 6A). When the three peptides within this pool were dissected and tested in a number of patients’ sera, the antibody response was almost entirely directed against one of these three peptides which encompasses aa 151-170 (Fig. 6B).
Amino acids 151-170 are located on RNA-binding terminal domain of NP protein which has been previously reported by Amruna et al. as an immunodominant epitope in SARS-CoV-2 and as a disease severity correlate in COVID-19 patients (26). Sequence homology analysis demonstrated that the epitope we identified in SARS-CoV-2 NP protein is highly conserved among human coronavirus strains, including SARS-CoV, MERS-CoV, HKU1-CoV, and OC43-CoV. Although most studies only address the diagnostic value of nucleocapsid protein and its epitopes for detection of seroconversion in COVID-19 patients, the highly conserved nature of this epitope in NP protein would be an advantage in developing epitope-based vaccines against SARS-CoV-2 which can develop cross-protection against other human coronaviruses. The majority of vaccines which are under clinical investigation against SARS-CoV-2 are developed based on viral spike protein and mostly rely on the ensuing neutralizing antibody response which blocks viral entry rather than killing infected cells. This could be largely attributed to the localization of NP protein. While the three other structural proteins of the virus, S, M, and E reside at the interface of the virus to the external environment, the NP protein is located inside the viral particles and is not accessible to the antibody during the course of infection. Therefore, although elevated anti-SARS-CoV-2 NP IgG and IgM antibody titers were observed in our study and have been reported by other investigators (27, 28), they do not seem to have neutralization potential.
Since most of our current knowledge about SARS-CoV-2 NP protein comes from previous studies on SARS-CoV, better evidence in this regard could be obtained from SARS-CoV literature. Several studies during the SARS outbreak clearly showed that SARS-CoV NP protein could not induce strong neutralizing antibody responses neither in human nor in animal (29-31); however, significant cytotoxic T lymphocyte (CTL) response against NP has been reported using vector-based vaccines containing NP protein (29, 32, 33) . Kim and colleagues introduced an effective DNA vaccine using SARS-CoV NP protein as a target antigen fused to calreticulin (CRT) to enhance MHC class I presentation of linked antigen to CD8(+) T cells. Their results showed that the NP-based vaccine generated strong NP-specific humoral and T-cell immune responses in mice (34). Recently, it has been suggested that SARS-CoV-2 NP could be considered as an advantageous vaccine target owing to its conserved nature and strong immunogenicity (35). A recently published data by Ahlén et al. (36), also showed a DNA vaccine based on a codon-optimized SARS-CoV-2 NP gene induced high titers of anti-NP antibodies in immunized rabbits, but most interestingly, they showed that immunization of mice with a DNA vaccine expressing the SARS-CoV-2 NP protein induced the strongest T cell response against a peptide region spanning our P151-170 peptide.
In this study, we also observed a substantial IgG response against CK-16 peptide fragment from the S2 domain of spike protein. The S2 subunit, composed of a fusion peptide, HR1, HR2, transmembrane and cytoplasmic domains, is responsible for viral fusion and host cell entry and thus plays a crucial role in the virus pathogenicity (8). The CK-16 peptide (aa 1254-1269) is located at the very C-terminal end of the spike protein (in the cytoplasmic domain of S2). It is surprising to see the high immunogenicity of CK-16 because it is located in the cytoplasmic domain of S2 and after viral entry, is not accessible to B cells. Whether the antibodies that recognize this epitope are neutralizing or not necessitates further study. However, it is interesting that CK-16 peptide shares a high sequence identity with C-terminus end of spike protein from all other coronaviruses, and the CK-16 peptide region is identical among SARS-CoV-2, SARS-CoV-1, SARS-CoV S Urbani, SARSr-CoV ZXC21and SARSr CoV-RaTG13 (8, 37). It implies that this epitope may serve a functional role for the virus. It has been shown that C-terminus of S2 subunit contributes to the stabilization of the trimeric structure of the spike protein (38), and destabilization of this area in the spike protein of SARS-CoV interferes with spike trimerization and reduces fusion and infectivity of SARS-CoV (39). A similar mechanism could be postulated for antibodies directed against CK-16 in SARS-CoV-2 that may destabilize the trimeric structure of spike protein and reduces viral pathogenicity. Yang Li et al. have also reported IgG antibodies against this epitope in the sera from some convalescent COVID-19 patients, however, the neutralization assay was not reported for this epitope (40).
On the other hand, due to high homology with other coronaviruses, it is possible that some patients have preexisting antibodies to this region, and the corresponding B cell clones against this region are expanded after SARS-CoV-2 infection. This epitope has also been identified in 2 other publications studying antibody response in convalescent sera from SARS patients (41, 42); hence the functional importance of this epitope also remains to be elucidated. Most recently, in a comprehensive linear epitope landscape using sera from 1,051 COVID-19 patients, Yang Li and colleagues revealed an immunodominant region at the C-terminal domain of S2 subunit rich in linear epitopes which covers our CK-16 (43). Although the clinical and functional significance of antibodies against this region were not proved in their study, the reactivity of a large number of COVID19 patients’ sera against this region reinforces its immunogenicity, making it more deserving for further investigation.
In summary, we identified the dominant B cell linear epitopes on RBD and NP proteins of SARS-CoV-2 in COVID-19 patients. These epitopes may serve as a vaccine candidate after being confirmed for their neutralization function using appropriate assays.