Expanding SARS-CoV-2 antigen targets beyond spike using a baculovirus expression system

TL;DR

This study explores using a baculovirus system to produce multiple SARS-CoV-2 proteins for better vaccines and diagnostics.

Contribution

The study introduces a baculovirus system to co-express multiple SARS-CoV-2 proteins for improved immune response profiling.

Findings

Nucleocapsid and membrane proteins elicited stronger antibody responses than spike protein.

BEVS effectively presented multiple antigens for profiling immune responses in convalescent sera.

Non-spike antigens showed significant diagnostic and immunological value.

Abstract

The ongoing antigenic evolution of SARS-CoV-2, particularly within the spike glycoprotein, threatens the long-term efficacy of spike-focused vaccines and serodiagnostics. While most authorized COVID-19 vaccines exclusively target the spike protein, growing evidence underscores multivalent strategies incorporating conserved viral antigens. Here, we employed a baculovirus expression vector system (BEVS) to co-express all four structural proteins, including spike, nucleocapsid, membrane, and envelope, in Sf9 insect cells. Surface-displayed antigens were used in a cell-based ELISA to profile IgG responses in convalescent sera. All antigens elicited detectable antibody binding, with nucleocapsid and membrane proteins provoking significantly stronger responses than spike (p < 0.001), and envelope protein showing intermediate reactivity. Statistical analyses revealed distinct patterns of antigenic reactivity. These findings validate the utility of BEVS for multiplex antigen presentation and highlight the immunological and diagnostic value of conserved non-spike antigens. Combined, this study advocates multivalent vaccine strategies and refines serodiagnostics by leveraging broader antigenic targets to counter immune escape.

The online version contains supplementary material available at 10.1038/s41598-026-38282-4.

Figures

Workflow of baculovirus expression vector system (BEVS) for SARS-CoV-2 antigen production. Target genes encoding the structural proteins of SARS-CoV-2, including S, N, M, and E proteins, were cloned into a pFastBac plasmid. (A) Recombinant plasmids were transformed into *E. coli* DH10Bac cells containing the bacmid and helper plasmid. (B) Tn7-mediated transposition integrated the target gene into the bacmid genome. (C) Colonies with successful transposition were identified through blue-white screening. (D) Recombinant bacmid DNAs were extracted from the white colonies and (E) transfected into

Light microscopy of Sf9 cell morphology at 96 h post-transfection with recombinant bacmid DNA. (A) Schematic overview of the transfection assay conducted in a six-well plate. (B) Untreated Sf9 cells serving as a negative control. (C–F) Sf9 cells transfected with S-bacmid, N-bacmid, M-bacmid, and E-bacmid, respectively. Cells infected with recombinant baculovirus exhibited characteristic cytopathic effects, including nuclear enlargement and a granular or vesicular appearance, compared to the untreated control. Images were captured using an inverted phase-contrast microscope at 400× magnificatio

Optimization of infection parameters in Sf9 cells. Recombinant baculoviruses encoding SARS-CoV-2 antigens were used to infect Sf9 cells at MOIs of 1 and 2 pfu/cell. Transcript levels of target genes were measured by qPCR at (A) 24 h and (B) 48 h post-infection. No statistically significant differences in transcript expression were observed between the two MOIs at either time point. In contrast, panel (C) shows a significant increase in transcript levels between 24 and 48 h post-infection at an MOI of 2 pfu/cell (*p* < 0.05), revealing that extending the infection period can improve gene expres

Serum IgG responses to SARS-CoV-2 structural antigens. IgG binding levels to S, N, M, and E antigens were measured using a spectrophotometric in-cell ELISA. Serum samples from 54 COVID-19 patients were analyzed in duplicate, and results are presented as adjusted mean absorbance values at 450 nm optical density (OD). Statistical comparisons were performed using one-way ANOVA followed by Tukey’s HSD post-hoc test. Letters above each box plot represent the results of Tukey’s HSD pairwise comparisons; groups that share the same letter are not significantly different.

Full text

Introduction

The emergence of coronavirus disease 2019 (COVID-19) sparked an extraordinary global effort, leading to the rapid development of vaccines at a pace unmatched in medical history^1^. Unlike traditional vaccines, which typically require a decade of development, the first COVID-19 vaccines received emergency authorization within 11 months of the viral genome’s release in 2020^2^. This remarkable acceleration was driven by several factors, including extensive global collaboration, substantial public-private investment, overlapping clinical phases, and expedited regulatory reviews through Emergency Use Authorizations (EUAs)^3^. While this achievement represents a milestone in virology, the continuing evolution of SARS-CoV-2 highlights the urgent need for next-generation vaccines with broader and more durable protection. Rational antigen design that incorporates spike protein mutations from multiple Variants of Concern (VOCs), such as Alpha, Beta, Gamma, and Delta, into recombinant constructs, as well as the exploration of non-spike antigens, may help sustain vaccine efficacy against viral evolution^4^.

Most approved COVID-19 vaccines, including mRNA-based, non-replicating viral vector, and protein subunit platforms, rely exclusively on the spike (S) protein as the immunogen^5^. However, repeated global surges have been driven by emerging VOCs, many exhibiting enhanced transmissibility, and virulence^6^. The spike protein, central to viral entry and immune recognition with 1,273 amino acids, has accumulated numerous mutations across all documented variants relative to the original Wuhan strain. Notably, Omicron subvariants BA.2.86 and JN.1 harbor over 30 spike mutations, representing roughly 2.4% of the total amino acid sequence, undermining neutralizing antibody efficacy and necessitating repeated boosters^7^. This antigenic drift emphasizes the limitations of S-only vaccines and the need for broader antigen coverage. Importantly, other structural proteins, including nucleocapsid (N), membrane (M), and envelope (E) proteins, are considerably more conserved, with lengths of 419, 222, and 75 amino acids, and sequence conservation of approximately 90%, > 95%, and > 95%, respectively^8^. Incorporating these targets could enhance T cell-mediated responses, improve protection against severe disease, and reduce immune escape^4^.

Beyond vaccination, accurate diagnostics remain vital for public health. Serological assays, which detect host antibody responses, are particularly important for assessing exposure and population immunity^9^. The N protein, the most abundant structural protein, is widely used in immunodiagnostic assays, largely due to previously reported high antigenicity and its substantial sequence conservation^10^. Unlike the variable S protein, N-based assays exhibit strong sensitivity across variants. However, N protein shares homology with endemic human coronaviruses (e.g. OC43 and HKU1), raising specificity concerns. This N antigen may be most effective when incorporated alongside other antigens in multiplex assays. Similarly, the M and E proteins, while highly conserved and abundantly expressed, may contribute diagnostic value when integrated alongside S and N^11^. Such multiplex approaches could improve diagnostic resolution between natural infection and vaccine-induced immunity. Yet, despite these advantages, the diagnostic potential of M and E remains underexplored.

Here, we address these gaps by employing the baculovirus expression vector system (BEVS) to heterologously express all four SARS-CoV-2 structural proteins (S, N, M, and E) in Spodoptera frugiperda (Sf9) insect cells for systematic evaluation of their antigenic reactivity and potential diagnostic utility (Fig. 1). Unlike prior BEVS-based studies that primarily focused on spike protein expression or virus-like particles (VLP) production, this work applies an established expression system to enable comparative serological profiling of all four SARS-CoV-2 structural proteins using human convalescent sera. This strategy highlights value of underutilized non-spike antigens and provides a framework for multivalent vaccine design and precision serodiagnostics capable of withstanding viral evolution and immune escape.

Fig. 1. Workflow of baculovirus expression vector system (BEVS) for SARS-CoV-2 antigen production. Target genes encoding the structural proteins of SARS-CoV-2, including S, N, M, and E proteins, were cloned into a pFastBac plasmid. (A) Recombinant plasmids were transformed into E. coli DH10Bac cells containing the bacmid and helper plasmid. (B) Tn7-mediated transposition integrated the target gene into the bacmid genome. (C) Colonies with successful transposition were identified through blue-white screening. (D) Recombinant bacmid DNAs were extracted from the white colonies and (E) transfected into Sf9 insect cells to generate recombinant baculoviruses expressing the target genes. (F) Primary viral stock (P1) was harvested and sequentially amplified to obtain high-titer viral stocks (P2 & P3). (G) Viral titers were quantified using plaque assay to standardize infection conditions. (H) Sf9 cells were infected at an optimized MOI, and expression of viral antigens was assessed using cell-based ELISA to evaluate antibody responses in COVID-19 patient sera.

Materials and methods

Recombinant pFastBac plasmid construction

To construct recombinant pFastBac vectors encoding the S, N, M, and E genes of SARS-CoV-2, codon-optimized sequences were employed to ensure efficient expression in human and insect cells (Addgene plasmids #149329, #149330, #141274 and #141273)^12^. Target genes were amplified by PCR using a mixture of standard Taq and high-fidelity DNA polymerase (Ampliqon, Denmark) with primers listed in Supplementary Table S1. Amplified products were purified (GeneAll PCR purification kit), and digested with RsrII and SalI (Thermo Fisher Scientific, USA), and ligated into corresponding sites of the pFastBac vector (Addgene plasmid #48295) using T4 DNA Ligase (Thermo Fisher Scientific, USA). Ligation products were transformed into chemically competent E. coli XL1-Blue cells, and transformants were selected on LB agar plates supplemented with ampicillin. Recombinant clones were identified by colony PCR and confirmed by Sanger sequencing (Supplementary Table S1).

Recombinant bacmid DNA generation

Chemically competent E. coli DH10Bac cells were transformed with purified pFastBac constructs containing the genes of interest to facilitate transposition into bacmid DNA. Transformants were plated onto LB agar supplemented with kanamycin (50 µg/mL), gentamicin (10 µg/mL), tetracycline (10 µg/mL), IPTG (40 µg/mL), and X-gal (100 µg/mL) for blue-white colony screening. White colonies were re-streaked under the same conditions and incubated at 37 °C for 48 h. Successful transposition was verified by colony PCR (Supplementary Table S1).

To isolate recombinant bacmid DNA, a verified colony was cultured in LB medium supplemented with mentioned antibiotics, and incubated at 37 °C. Bacmid DNA was extracted using an alkaline lysis procedure^13^. Briefly, cells were resuspended in GTE buffer [50 mM glucose, 25 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 100 µg/mL RNase A], lysed with 1% SDS and 0.2 N NaOH, and neutralized with 5 M potassium acetate (pH 4.8). After centrifugation, bacmid DNA was precipitated with ethanol, resuspended, and analyzed by agarose gel electrophoresis (0.5% agarose gel, 23 V, 12 h). Samples were stored at 4 °C for downstream applications.

Insect cell culture and transfection

Spodoptera frugiperda (Sf9; ATCC: CRL-1711) cells were employed as the host for bacmid replication and recombinant baculovirus production. Cells were cultured adherently in IPL-41 insect medium (UGA Biopharma, Germany), supplemented with sodium bicarbonate, sodium chloride, and lactalbumin hydrolysate (Bio Basic, Canada). Medium pH was adjusted to 6.2 prior to use and further supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific, USA) and 100 U/mL penicillin-streptomycin. Cells were maintained at 27 °C in a non-humidified, CO_2_-free incubator.

For transfection, Sf9 cells with ~ 90% confluency were gently detached by sloughing method and seeded into six-well plates at a density of 1.2 × 10^6^ cells/well in insect medium supplemented with 1.5% FBS and no antibiotics. Plates were incubated at 27 °C for 10 h to allow the cell attachment. For each transfection, 3 µg of recombinant bacmid DNA was diluted in 100 µL of IPL-41 medium, combined with 7 µL Cellfectin II (Thermo Fisher Scientific, USA) diluted in 100 µL of IPL-41 medium, and incubated at room temperature for 30 min to form complexes. Mixtures were added to cells, with controls lacking bacmid DNA or transfection reagent. After 5 h, complexes were replaced with 2 mL of fresh medium supplemented with 7% FBS and no antibiotics. The cells were further incubated at 27 °C for 5 days to allow for recombinant baculovirus production. Following incubation, supernatants were carefully collected, clarified by low-speed centrifugation, passed through 0.2 μm filters, and stored at 4 °C, protected from light.

Recombinant baculovirus stock amplification

To amplify recombinant baculoviruses, Sf9 cells were seeded into six-well plates at a density of 2 × 10^6^ cells/well in complete IPL-41 medium and incubated at 27 °C for 8 h. At infection, medium was replaced with 1.2 mL of IPL-41 medium supplemented with 0.1 mL of the corresponding P1 virus stock. The following formula was used to calculate the required volume of virus stock. To achieve a multiplicity of infection (MOI) of 0.1 plaque-forming units (pfu)/cell using a P1 virus stock with an assumed titer of 2 × 10^6^ pfu/mL, the required virus volume was calculated using the following formula:

Inoculum volume (mL) = [MOI (pfu/cell) × Number of cells] / [Virus titer (pfu/mL)]

The virus inoculum was added to the cells and allowed to adsorb for 1.5 h with gentle rocking at regular intervals. Following the adsorption, 1.3 mL of IPL-41 medium supplemented with 7% FBS was added to each well, bringing the final volume to 2.5 mL. Infected cells were incubated at 27 °C for 72 h in a humidified, CO_2_-free incubator. Supernatants were harvested, clarified by low-speed centrifugation, and stored. For secondary amplification, Sf9 cells were infected with 0.01 mL of the P2 virus stocks following the same protocol. At 24 h post-infection, P3 virus-containing supernatants were harvested and clarified. All virus stocks were stored at 4 °C protected from light, for subsequent use.

Recombinant baculovirus stock quantification

A plaque assay was conducted to determine the infectious titer of P3 recombinant baculovirus stocks. Sf9 cells were seeded into six-well plates at a density of 1 × 10^6^ cells/well in IPL-41 medium supplemented with 4% FBS and incubated overnight to ~ 50% confluency. An 8-log serial dilution (10^− 1^ to 10^− 8^) of each baculovirus was prepared in IPL-41 medium. Immediately before use, the plaquing overlay medium was prepared by mixing 75 mL of IPL-41 medium supplemented with 10 mL of FBS and 25 mL of melted 4% low-melting-point agarose (Sigma-Aldrich, Merck, Germany).

For each baculovirus, 1 mL of the viral suspension ranging from 10^− 4^ to 10^− 8^ was added in duplicate to the designated wells following the removal of the existing medium. Two wells received virus-free medium as negative controls. Cells were incubated at room temperature for 1 h with gentle agitation every 15 min. Following incubation, the virus inoculum was replaced with 2 mL of 37 °C plaquing medium. The overlay was allowed to harden for 1 h and plates were then incubated at 27 °C for 10 days in a humidified, CO_2_-free incubator. At 10 days post-infection, 1 mg/mL neutral red solution was used to stain plaques. The virus titer was determined by multiplying the average number of plaques per dilution by the dilution factor, and dividing by the inoculum volume per well.

Optimization of conditions for recombinant protein expression

Recombinant protein expression was optimized by evaluating two multiplicities of infection (MOI: 1 and 2 pfu/cell) and two post-infection incubation periods at 24 and 48 h. Sf9 cells were seeded into six-well plates at a density of 2 × 10^6^ cells/well in IPL-41 medium supplemented with 3% FBS, and incubated for 8 h. After rinsing with medium, cells were infected with virus stocks at the designated MOIs in IPL-41 medium supplemented with 0.5% FBS. The virus was allowed to adsorb for 30 min with gentle rocking at regular intervals. At 24 and 48 h post-infection, the medium was aspirated, and cell monolayers were rinsed with medium. Total RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific, USA) and quantified using a Nanodrop spectrophotometer.

Subsequently, RNA was reverse transcribed to cDNA using the RevertAid reverse transcriptase (Thermo Fisher Scientific, USA). Quantitative real-time PCR (RT-qPCR) was performed using RealQ Plus Master Mix Green (Ampliqon, Denmark) in duplicate using gene-specific primers (Supplementary Table S1). The thermal cycling conditions were as follows: initial denaturation at 95 °C for 15 min, followed by 40 cycles of 95 °C for 15 s and 63 °C for 30 s. All primers were analyzed using the Primer-BLAST against the Spodoptera frugiperda and Escherichia coli to eliminate potential off-target amplification.

Cell-based ELISA for detection of antibody binding

Sf9 cells were seeded into 96-well plates at a density of 6 × 10^4^ cells/well in 120 µL of IPL-41 medium supplemented with 3% FBS, and incubated at 27 °C for 8 h. Following cell attachment, the medium was replaced with 100 µL of recombinant baculovirus at an MOI of 2 pfu/cell, diluted in IPL-41 medium supplemented with 0.5% FBS. Each serum sample was tested in duplicate against cells infected with S-BV, N-BV, M-BV, and E-BV. In parallel, ten wells per baculovirus served as negative controls. Viral adsorption was performed for 30 min with gentle rocking at regular interval, followed a 72-hour incubation.

Cells were fixed with 4% paraformaldehyde (pH 7.4) at 25 °C for 20 min and permeabilized using PBS supplemented with Tween-20, incubating at 25 °C for 30 min. Subsequently, wells were blocked with blocking solution (PBS supplemented with 3% BSA and Tween-20) at 25 °C for 2 h, followed by two washes with PBS. Serum samples were diluted 1:100 in serum dilution buffer (PBS containing 1% BSA and Tween-20) and added to each well (100 µL/well), then incubated at 25 °C for 2 h. Wells designated as negative control received only serum dilution buffer to account for background signal. After three washes with PBS containing Tween-20, HRP-conjugated goat anti-human IgG (1:5000 dilution in blocking solution, Millipore Sigma) was added to each well and incubated at 25 °C for 2 h. Following an additional three washes, TMB substrate (Sigma-Aldrich, Merck, Germany) was added and incubated in the dark for 20 min. Then 1 M sulfuric acid was added and absorbance was measured at 450 nm using a microplate reader (Hiperion MPR4++).

Serum sample collection

Human serum samples were collected from 54 individuals with RT-PCR-confirmed SARS-CoV-2 infection, between 3 and 6 weeks after the onset of COVID-19 symptoms. All samples were stored at − 80 °C until further analysis.

Statistical analyses

For expression optimization assays, statistical significance between two groups was assessed using an unpaired two-tailed Student’s t-test. A p-value less than 0.05 was considered statistically significant. In the cell-based ELISA assay, the mean optical density (OD) of the negative control wells for each recombinant baculovirus type was subtracted from the mean OD of corresponding virus-infected wells on the same plate to determine the final ELISA value for each serum sample. One-way analysis of variance (ANOVA) was used to evaluate differences among groups, with p < 0.05 considered statistically significant. Pairwise comparisons were performed using Tukey’s Honestly Significant Difference (HSD) post-hoc test. Statistical analyses were conducted using R software (version 4.5.1).

Results

Construction and verification of recombinant bacmid DNAs carrying S, N, M, andE Genes

To express SARS-CoV-2 structural proteins in insect cells, we produced recombinant bacmid DNAs encoding the S, N, M, and E proteins using the baculovirus expression vector system. In this process, we successfully amplified the full-length, codon-optimized gene sequences by PCR (Supplementary Figure S1). The amplified products were digested and ligated into the pFastBac vector, yielding constructs designated as S-pFastBac, N-pFastBac, M-pFastBac, and E-pFastBac (Supplementary Figure S2 & S3). We transformed the recombinant constructs into E. coli DH10Bac competent cells, which harbor a bacmid containing a mini-attTn7 attachment site. Site-specific transposition of the recombinant expression cassette into the bacmid genome was facilitated by Tn7 transposition, disrupting the LacZ alpha-complementation system and enabling selection of recombinant clones by blue-white screening. White colonies, indicating successful recombination, were re-streaked for clonal isolation and subjected to colony PCR using pUC/M13 primers flanking the transposition site. For the M and E constructs, PCR produced distinct amplicons of 3,093 bp and 2,652 bp, respectively, while the non-recombinant control band (342 bp) was not detected (Supplementary Figure S4). Due to the large length of the S and N inserts, combinations of flanking and gene-specific primers were employed for verification (Supplementary Tables 1 and Supplementary Figures S5 & S6), producing the expected amplicon patterns in each case.

To confirm sequence integrity and rule out potential mutations introduced during amplification or cloning, we performed bidirectional Sanger sequencing on all constructs. For complete coverage of the large S and N genes, overlapping primer pairs were designed as listed in Supplementary Table S1. The sequencing results confirmed the accuracy and consistency of all genes, with no mutations or frame-shifts detected (Supplementary File S1). The validated recombinant bacmid DNAs were designated as S-bacmid, N-bacmid, M-bacmid, and E-bacmid, and were subsequently used for transfection into insect cells for virus production and protein expression (Supplementary Figure S7).

Production and amplification of recombinant baculoviruses

To generate infectious recombinant baculoviruses encoding SARS-CoV-2 structural genes, we transfected Sf9 cells with validated recombinant bacmid DNAs. Two wells mock-transfected with vehicle alone served as negative controls. We monitored the cells daily by phase-contrast microscopy to track cytopathic changes. By ~ 48 h post-transfection (hpt), hallmark features of productive infection, including cellular hypertrophy, granularity, and detachment, began to appear and become increasingly prominent by 72–96 hpt (Fig. 2; Supplementary Figures S8 & S9). In contrast, mock-transfected cells retained a normal, adherent morphology. At approximately five days post-transfection, once widespread cytopathic effects were evident, we harvested supernatants containing the first-generation recombinant baculovirus (P1 stocks), designated as S-BV, N-BV, M-BV, and E-BV.

Fig. 2. Light microscopy of Sf9 cell morphology at 96 h post-transfection with recombinant bacmid DNA. (A) Schematic overview of the transfection assay conducted in a six-well plate. (B) Untreated Sf9 cells serving as a negative control. (C–F) Sf9 cells transfected with S-bacmid, N-bacmid, M-bacmid, and E-bacmid, respectively. Cells infected with recombinant baculovirus exhibited characteristic cytopathic effects, including nuclear enlargement and a granular or vesicular appearance, compared to the untreated control. Images were captured using an inverted phase-contrast microscope at 400× magnification.

To enrich for fully infectious viral particles and minimize defective background, we performed two subsequent amplification rounds. Infected cells were incubated until late-stage cytopathic effect observed, and clarified culture supernatants were harvested to generate high-titer P3 viral stocks (supplementary Figure S10). We subsequently quantified viral titers using plaque assay to prepare for downstream protein expression and immunoassay applications.

Optimization of conditions for efficient recombinant protein expression

To ensure consistency in infection efficiency and uniform antigen presentation across all experimental conditions, we determined the titers of the amplified S-BV, N-BV, M-BV, and E-BV baculoviral stocks using plaque assay and normalized them accordingly. The resulting titers were 0.4 × 10^7^, 0.375 × 10^7^, 0.5 × 10^7^, and 0.563 × 10^7^ pfu/mL, respectively (Supplementary Figure S11 & S12), enabling adjusted MOI-based infection.

To identify optimal conditions for maximal recombinant protein expression, we employed a quantitative RT-PCR-based approach to assess transgene transcript levels following infection^14^. Sf9 monolayer cells were infected with each P3 viral stocks at MOIs of 1 and 2 pfu/cell, and total RNA was extracted at 24 and 48 h post-infection. While no significant differences in transcript abundance were observed between the two MOIs at either time point, transcript levels at 48 h post-infection were significantly higher than those at 24 h (p < 0.05; Fig. 3), suggesting continued transcriptional activity beyond the first day of infection. Based on these findings, an MOI of 2 pfu/cell was selected for subsequent experiments to ensure reproducible infection. Furthermore, to accommodate the inherent delay between mRNA transcription and detectable protein accumulation, a time interval of 72-hour post-infection was chosen for optimal antigen presentation.

Fig. 3. Optimization of infection parameters in Sf9 cells. Recombinant baculoviruses encoding SARS-CoV-2 antigens were used to infect Sf9 cells at MOIs of 1 and 2 pfu/cell. Transcript levels of target genes were measured by qPCR at (A) 24 h and (B) 48 h post-infection. No statistically significant differences in transcript expression were observed between the two MOIs at either time point. In contrast, panel (C) shows a significant increase in transcript levels between 24 and 48 h post-infection at an MOI of 2 pfu/cell (p < 0.05), revealing that extending the infection period can improve gene expression efficiency in insect cells. In all panels, expression levels of S-BV, N-BV, M-BV, and E-BV baculoviral stocks are depicted in blue, red, orange, and yellow, respectively, with two technical replicates per assay. Each panel includes negative controls without viral infection.

In-Cell ELISA using surface display of SARS-CoV-2 antigens

To assess serum IgG responses against SARS-CoV-2 structural proteins, we developed an in-cell ELISA assay using Sf9 insect cells expressing the S, N, M, and E antigens. Unlike conventional serological assays that typically target the receptor-binding domain (RBD) or truncated spike variants, our platform employed the entire spike protein, comprising both S1 and S2 subunits, to retain conformational epitopes and native glycosylation patterns with full-length antigenic structure^15^. To establish the ELISA platform, Sf9 cells were infected with recombinant baculoviruses encoding each antigen. Serum samples from 54 individuals with confirmed SARS-CoV-2 infection were applied to each well. IgG reactivity was detected using horseradish peroxidase (HRP)-conjugated anti-human IgG antibodies. To minimize technical variability stemming from factors such as plate handling, reagent heterogeneity, and particularly wash efficiency as previously described^16^, signals were corrected by subtracting the mean optical density (OD) of five negative control wells on each plate.

All serum samples exhibited IgG reactivity against the four antigens, while negative control wells produced only background signal (Fig. 4). Quantitative analysis revealed the nucleocapsid antigen elicited the highest mean IgG response (mean OD = 1.56), followed closely by the membrane (mean OD = 1.53) and envelope (mean OD = 1.46) antigens. In contrast, the spike antigen displayed the lowest mean IgG reactivity (mean OD = 1.00), highlighting higher antigenicity of internal and membrane-associated viral proteins relative to the surface-expressed spike under our assay conditions. A one-way ANOVA confirmed a highly significant difference in IgG binding across the four antigens (p < 0.001). Tukey’s HSD post-hoc analysis further revealed significantly greater IgG recognition for N, M, and E antigens compared to S (p < 0.01 for all comparisons), while N antigen also surpassing E in immunoreactivity (p < 0.05). No significant differences were observed between the N and M antigens or between the M and E antigens (Table 1). These findings underscore the immunodominance of internal and membrane-associated structural proteins, particularly the N, M, and E antigens, in eliciting antibody responses during natural infection.

Fig. 4. Serum IgG responses to SARS-CoV-2 structural antigens. IgG binding levels to S, N, M, and E antigens were measured using a spectrophotometric in-cell ELISA. Serum samples from 54 COVID-19 patients were analyzed in duplicate, and results are presented as adjusted mean absorbance values at 450 nm optical density (OD). Statistical comparisons were performed using one-way ANOVA followed by Tukey’s HSD post-hoc test. Letters above each box plot represent the results of Tukey’s HSD pairwise comparisons; groups that share the same letter are not significantly different.

Table 1. Tukey’s HSD post-hoc analysis of pairwise comparisons among SARS-CoV-2 antigenic groups.Pairwise comparisonsTukey HSDQ statisticTukey HSDp-valueTukey HSDinferenceS vs. N20.3711.0779e-10SignificantS vs. M19.1321.0779e-10SignificantS vs. E16.5341.078e-10SignificantN vs. M1.2390.81711Not significantN vs. E3.8370.036041SignificantM vs. E2.5970.25911Not significant

Discussion

In the wake of the COVID-19 pandemic, rapid vaccine development primarily targeting the spike protein has been pivotal in mitigating severe disease and mortality worldwide. However, the continual emergence of SARS-CoV-2 variants with multiple mutations in the spike region has challenged vaccine efficacy and highlighted the need for next-generation vaccines and diagnostics that go beyond the spike antigen^17,18^. Our study contributes to this endeavor by leveraging the BEVS to express four major SARS-CoV-2 structural proteins, including S, N, M, and E, in Sf9 insect cells, providing a platform to comparatively evaluate their serological reactivity and diagnostic potential. Our findings reveal a robust antibody response not only to the S protein but also, and more importantly, to the N and M proteins, with N showing the highest IgG reactivity in COVID-19 patient sera. This aligns with prior studies emphasizing the immunodominance of the N protein, which is abundantly expressed during infection and exhibits greater sequence conservation relative to the S protein^19,20^. The relative stability of the N protein across variants suggests its utility as a reliable serological marker and underscores its potential relevance for antigen screening in the context of multivalent vaccine design, rather than direct evidence of protective efficacy^21,22^. However, the comparatively lower IgG reactivity observed for spike may also reflect technical factors related to heterologous expression in Sf9 cells, including differences in protein folding, trimer stability, or insect cell-specific glycosylation, rather than solely reflecting biological immunodominance.

Surprisingly, the M protein, often overlooked in antigen-focused studies, induced antibody responses comparable to N. Given its critical role in virion assembly and its relatively conserved sequence, the M protein may contribute to cross-reactive T cell immunity against diverse SARS-CoV-2 variants^23^. This is particularly relevant as ongoing viral evolution continues to erode spike-specific immunity. Because only a single reference spike sequence was evaluated, the observed spike reactivity may also reflect variant mismatch with circulating strains at the time of serum collection. This limitation further underscores the vulnerability of spike-centric serodiagnostics to ongoing viral evolution. Moreover, because variant-level infection data were unavailable for the analysed sera, we were unable to assess whether antibody recognition of non-spike antigens is presented across infections with distinct SARS-CoV-2 variants. Future studies incorporating variant-confirmed cohorts will be essential to directly test the variant-agnostic diagnostic potential of N, M, and E antigens. Collectively, our data support the inclusion of the M antigen into multivalent antigen screening strategies aimed at broadening immune coverage, rather than implying direct protective efficacy. Envelope protein-specific antibodies, although slightly less prevalent, still elicited a significant response. The E protein’s multifunctional role in virus assembly and pathogenesis, combined with its conserved nature, suggests that it may complement other antigens in multiplex serological assays^24^. However, given its small size and lower abundance on the virion surface, its direct neutralizing potential may be limited, warranting further functional studies to clarify its protective role.

Our use of BEVS to express full-length antigens in Sf9 cells enabled retention of key structural features, including proper folding and insect-cell-derived glycosylation, which is particularly important for maintaining conformational epitopes of the spike protein^25^. This system also facilitated standardized antigen production for cell-based ELISA. The consistent detection of IgG antibodies across all antigens supports the feasibility of the BEVS platform as a proof-of-concept system for exploratory serological profiling and antigen screening. From a diagnostic perspective, the strong and consistent antibody responses to N and M proteins highlight their potential utility as complementary antigens in future serological assay development. Unlike the S protein, whose antigenic variability requires frequent assay adjustments, conserved nature of N and M reduces false negatives due to antigenic drift, enhancing test sensitivity across variants^26,27^. Nevertheless, cross-reactivity with endemic human coronaviruses remains a concern, particularly for N and M proteins, which share homology with common cold coronaviruses like OC43 and HKU1^28,29^. While our findings suggest clear IgG reactivity of these antigens in COVID-19 patient sera, a key limitation of our study is the absence of protein-level validation to confirm translation, as well as the lack of cross-reactivity assessments using hyperimmune sera, monoclonal antibodies, or convalescent sera from individuals previously infected with other coronaviruses. As a result, we were unable to determine the extent to which the observed antibody responses are specific to SARS-CoV-2 versus potentially cross-reactive with related coronaviruses. Future studies should address this limitation by incorporating pre-pandemic sera as negative controls and employing competitive binding assays such as homologous competition through serum pre-incubation with soluble SARS-CoV-2 antigens and heterologous competition with endemic HCoV antigens to rigorously evaluate whether antibody binding is directed specifically to SARS-CoV-2 epitopes rather than conserved regions shared across coronaviruses. Another limitation of our study is absence of comparison with conventional plate‑based ELISA, which prevented determination of the limit of detection (LOD) and benchmarking of assay performance. Altogether, multiplex serological assays combining S, N, M, and E antigens may improve diagnostic specificity in principle; however, rigorous benchmarking against conventional ELISA platforms using defined negative controls will be required to substantiate this potential. Although our data do not evaluate neutralizing or cellular immune responses, the observed differential antigenic reactivity provides valuable insights for antigen screening and may help inform the rational design of next-generation multivalent formulations.

Our statistical analyses further underscored significant differences in antibody binding, with N and M outperforming S in IgG reactivity, and E protein eliciting a slightly weaker yet meaningful response. These distinctions highlight the necessity of a multi-antigen approach to capture the full spectrum of host immune responses, particularly when aiming to reconstruct natural immunity more faithfully than spike-only vaccines^30^. While our study highlights the promise of multi-antigen approaches, several limitations must be acknowledged. We focused on IgG responses without evaluating neutralizing activity or cellular immune responses, which are integral to long-term protection^31,32^. Future work should include T cell assays such as ELISpot and intracellular cytokine staining, as well as neutralization tests, to comprehensively characterize immune responses elicited by multi-antigen vaccine candidates. Additionally, expanding sample diversity to include vaccinated individuals and those infected with newer variants would clarify the breadth and durability of cross-reactive immunity. Finally, while BEVS offers advantages in protein expression and scalability, glycosylation patterns in insect cells differ from mammalian cells, which may influence antigenicity, especially for heavily glycosylated proteins like spike^33–35^. Complementary expression in mammalian systems or glycoengineering strategies could enhance translational relevance.

In summary, our work underscores the immunological relevance of SARS-CoV-2 structural proteins beyond the S antigen, particularly nucleocapsid and membrane proteins, in eliciting strong and broadly reactive antibody responses. Overall, our findings support the exploration of multi-antigen strategies as a framework for antigen prioritization and immune profiling, rather than as a standalone diagnostic solution. Further validation incorporating benchmarked sensitivity and specificity measurements will be essential before translational diagnostic application. Continued research integrating humoral and cellular immunity, variant-specific responses, and functional neutralization will be critical to realize the full potential of these findings.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2