Immunogenicity of differentially glycosylated Marburg virus glycoproteins expressed in mammalian and insect cells

Abstract

Marburg virus (MARV) infection can cause severe disease, and there is no available vaccine or therapeutic method. Research into potential vaccine design is focused on the glycoprotein (GP), which mediates the adherence and invasion process of the virus. However, it is unclear whether the degree of GP glycosylation is associated with vaccine efficacy. Here we constructed two versions of the GP expressed using insect and mammalian cell systems, respectively, either containing the mucin-like domain (MARV GPΔTM including residues 1-637) or deleting residues 264-425 to remove the part of mucin-like domain (MARV GPΔTM ΔMuc). Physicochemical properties, antigenicity, and immunogenicity were compared for soluble GPs produced in different cell expression systems. The GPΔTM ΔMuc produced in mammalian cells was more immunogenic, as evidenced by the induction of higher titers of binding antibodies and more antibodies targeting the protective epitope. Our results may offer a better understanding of glycosylation for the development of vaccines.

Introduction

Marburg virus (MARV) is a group of highly virulent pathogens that can cause viral hemorrhagic fevers in humans and nonhuman primates [1, 2]. Since Marburg virus (MARV) was first identified in 1967 in Germany and Yugoslavia [3], Marburg virus (MARV) has sporadically emerged several times, mainly throughout equatorial Africa [4]. Most are local outbreaks, but in 1998 and 2004, two major outbreaks occurred with nearly 90% lethality [1, 2]. In 2025, Tanzania experienced a Marburg virus disease (MVD), and it is the second occurrence of the virus in the country after the initial outbreak in 2023. Rwanda also reported the first Marburg outbreaks, with 66 confirmed cases having recovered as of 8 November 2024. Due to its high lethality rates, MARV was declared as a pathogen with the potential to generate a public health emergency of international concern by the World Health Organization (WHO) in 2018 and identified as a priority pathogen that requires urgent vaccine development [5]. However, there is no licensed treatment or vaccine for MARV infection [6, 7]. Therefore, an effective preventative vaccine is urgently needed.

MARV is a non-segmented, negative-strand RNA virus [1, 2, 4, 8]. The genus Marburgvirus contains Marburg virus (MARV) and Ravn virus (RAVV) [9, 10]. Based on the phylogenetic analysis of genomic sequence data, MARV has four different lineages: Angola, Ci67, Musoke, and poPP, and the most pathogenic is Angola [11]. MARV glycoprotein (GP) is the surface glycoprotein responsible for viral entry into the host cells [12], which serves as one of the major targets for the immune response [13] and is the major focus of vaccine development [14]. The World Health Organization (WHO) has published a list of four priority vaccine candidates, all of which are viral-vectored vaccines encoding the surface glycoprotein (GP) of the Marburg virus. These vaccines are currently in various stages of clinical and preclinical studies. Among them, only the ChAd3 vector vaccine, a chimpanzee adenovirus vector vaccine, has completed Phase I clinical trials. Clinical trials have demonstrated that the vaccine is safe and immunogenic, and the future Phase II clinical trials are planned to be conducted in Ghana, Kenya, and Uganda [15]. GP is cleaved by furin in the producer cells to yield disulfide-linked subunits GP1 and GP2, present as a homotrimer in the viral envelope [16]. Due to the shift of the furin cleavage site (Arg-435 in MARV and Arg-501 in Ebola virus), the amino acids (aa) 436-501 remain attached to the GP2 after proteolytic separation of GP precursor into subunits [17]. This 66-amino-acid N-terminal extension of GP2 is termed the “GP2 wing” due to its outward projection and flexibility [18], which is the only epitope other than the receptor-binding site (RBS) shown to elicit protective antibodies against MARV [10].

The glycans linked to the surface protein play extensive roles in the virus-host recognition and immune evasion [19], which can affect the recognition epitopes of monoclonal antibodies (mAbs) in the immune system, such as Human immunodeficiency virus (HIV) and influenza virus [20]. Different expression systems, including mammalian cells [21] and insect cells [22, 23], could serve as a strategy to modulate glycosylation and produce distinct glycan patterns [24]. Mammalian cells predominantly confer complex- and high-mannose-type glycans, while insect cells mainly impart simple paucimannose-type glycans [25]. Moreover, MARV GP exhibits a high degree of glycosylation [26], which contains 19 potential N-linked glycosylation sites along with multiple O-glycosylation sites [16, 27, 28]. Various research has shown that antigens with different degrees of glycosylation have different immunogenicity [29, 30]. However, there is no literature on the function and immunogenicity of MARV GP that differ in glycosylation degree. Here, in order to compare the difference, the mucin-containing GP and the mucin-deleted GP were constructed, then expressed in Expi293F cells and Drosophila S2 cells, and then the differences were compared, including antigenicity and immunogenicity. Our research hopes to give an insight into whether the glycosylation of the antigen could affect the exposure of epitopes to the humoral immune system.

Method

Expression plasmids and transient transfections

To investigate the influence of glycosylation degree on the antigenicity and immunogenicity of the Marburg virus GPs, Angola-GP (KU978782) strain MARV GP ectodomains (including residues 1-637), containing the mucin-like domain (MARV GPΔTM) or deleting residues 264-425 to remove the mucin-like domain (MARV GPΔTM ΔMuc), were expressed in different cells. For Drosophila S2 cell expression, DNA encoding the glycoproteins was synthesized and codon-optimized for expression in insect cells at General Biol. The gene was cloned into the vector pMT/BiP/V5-His B(PMT-GP). For Expi293F cell expression, the gene was codon-optimized for expression in mammalian cells, and the gene of MARV GPΔTM was cloned into the vector pCAGGS (PCAGGS-GP); the other gene of MARV GPΔTM ΔMuc was cloned into the vector pcDNA3.4(pcDNA3.4-GP). All constructs contained a C-terminal Twin-Strep-tag for purification.

To generate GP glycoprotein from insect cells, S2 cells cultured in Schneider’s Drosophila Medium (Gibco, 21720024) containing 10% heat-inactivated FBS (Gibco, 10099141-C), 0.1% Pluronic F-68 (Gibco, 24040032) and 50 U/mL Penicillin-Streptomycin (Gibco, 15140122) were co-transfected with pMT-GP using calcium phosphate transfection (YEASEN, 40803ES70). Prepare cultured cells for transfection by seeding 1 × 10⁶ S2 cells in a 6-well plate in 2 mL complete Schneider’s Drosophila Medium. After growing for 16 h at 27 °C until cells reach a density of 2 × 10⁶-4 × 10⁶ cells/mL, 6 µg pMT-GP by calcium phosphate transfection was co-transfected for each well, and the medium was changed after 24 h of incubation. Copper sulfate was added to the medium to a final concentration of 500 µM 2 days after transfection. Six days after transfection, the culture medium was collected and then clarified by centrifugation at 800 × g for 15 min, 4000 × g for 15 min, and filtered using a 0.45 μm filter (Millipore, SLHVR33RB). For Expi293F, PCAGGS-GP (MARV GPΔTM) or pcDNA3.4-GP (MARV GPΔTM ΔMuc) constructs were transiently expressed in Expi293F cells with the ExpiFectamine™ 293 transfection kit (Gibco, A14524) at 37 °C at 100 rpm with 5% CO₂ according to the manufacturer’s protocol. The supernatant was collected at 120 h post-transfection.

The supernatant was purified through a StrepTactin affinity purification strepTrap(Cytiva,29401322) purification column according to the manufacturer’s instructions. Then, the elution was exchanged with 0.01 M phosphate-buffered saline (PBS, pH 7.2) and concentrated using a 30 kD Centrifugal Filter Unit (Millipore, UFC903096). Proteins were further purified by SEC on a Superdex 200 Increase 10/300 GL column (Cytiva, 28990944). Then the fractions were harvested and analyzed by SDS-PAGE. The resulting purified Marburg GPs glycoprotein was used to ELISA, western blotting, and immunization experiments.

Antibody production and purification

The antibody sequences of MR78, MR191, MR228 were used as reported [22, 23] and codon optimized for mammalian cells. The heavy and light chains were cloned into the pcDNA3.4 expression vector, and then the constructs were transfected into Expi293F cells using the ExpiFectamine™ 293 Transfection Kit (Gibco, A14524) according to the manufacturer’s specifications. Antibodies were purified from culture supernatants using Protein A column (Cytiva, 17040201) followed by running buffer (PBS). Bound proteins were eluted using elution buffer (0.1 M glycine, pH 2.7). This fraction was concentrated by 50 kD  Amicon Ultra concentrators (Millipore, UFC905096) and was identified using SDS-PAGE. The purified protein was aliquoted and stored at -80 °C until further use.

SDS-PAGE, Native-PAGE and Western blotting

For SDS-PAGE, 5 µg of reduced or non-reduced protein was loaded onto a SurePAGE™ Plus, Bis-Tris, 4-12% gel (GenScript, M00653). After electrophoresis, the gel was stained with Coomassie brilliant blue G-250. For western blotting, gels were transferred onto nitrocellulose membranes (Cytiva, 10600001) using the trans-Blot Turbo transfer system (GenScript). Membranes were then blocked with 5% skim milk for 1 h. After washing three times with wash buffer containing PBS and 0.2% Tween 20 (PBST), the membranes were incubated with MR78 (1 µg/mL) or anti-strep-tag II antibody (1:2000 dilution; Abcam, ab307676) for 1 h. The membranes were washed again and then incubated with HRP-conjugated Goat Anti-Human IgG Fc (1:5000 dilution; Abcam, ab97225) or Goat Anti-Rabbit IgG Secondary Antibody (HRP) (0.01 µg/mL; Sino Biological, SSA004) for 1 h. Membranes were washed again and then incubated with Chemiluminescent Substrate Kit (Millipore, WBKLS0100). Images were acquired using an iBright FL1500 imaging system (Invitrogen).

For native-PAGE, purified protein samples were mixed with 3× loading buffer (Thermo Scientific, BN2008) and then loaded onto 4-15% Native-Page Tris-Gly gradient gels (Beyotime, P0465S) as described by the manufacturer. Proteins were electrophoresed at 150 V for 60 min. After electrophoresis, the protein was also stained with Coomassie brilliant blue G-250.

High-performance liquid chromatography (HPLC)

All high-purity GPs were analyzed on a HPLC system (Waters) using a TSK Gel G5000PWXL 7.8 × 300 mm column (TOSOH, 0008023), which was pre-equilibrated with PBS, pH 7.4 (Gibco, C10010500BT). Samples were loaded at a flow rate of 0.4 mL/min, and eluted proteins were detected at 280 nm.

Deglycosylation and glycan staining

Glycans were removed from purified GPs using peptide-N-glycosidase F (PNGase F) (New England BioLabs, P0704S) and Endoglycosidase (Endo H) (New England BioLabs, P0702S). In brief, 20 µg of GP protein was denatured at 100 °C for 10 min in 1× glycoprotein denaturing buffer, then digested with 1 µL PNGase F or endo-N-acetylglucosaminidase H (Endo H) at 37 °C for 1 h according to the manufacturer’s instructions. The deglycosylated protein was analyzed by Western blot with anti-Strep antibody as the detection antibody. For MARV GPΔTM ΔMuc, the detection method is as described above. For MARV GPΔTM, the membranes were incubated with StrepMAB-Classic HRP (1:3000 dilution; iba, 2-1509-001). The glycan staining was performed using periodic acid schiff (PAS) staining (Thermo Scientific, 24562) for glycoproteins as described by the manufacturer.

Enzyme-linked immunosorbent assay (ELISA)

To determine the antigenicity of Marburg virus GP, Costar™ 96-well assay plates (Costar, 9018) were coated overnight at 4 °C with serially diluted Marburg GP. Then washed three times with wash buffer containing PBS and 0.2% Tween 20 (PBST). Each well was then coated with 100 µl of blocking buffer containing PBST with 2% BSA for 1 h at 37 °C and then washed three times with wash buffer. Marburg mAbs MR78 and MR191 were then separately added at a concentration of 10 µg/mL, MR228 was added at a concentration of 1 µg/mL, followed by incubation for 1 h at 37 °C. Then, after washing three times with wash buffer, Goat Anti-Human IgG Fc (HRP) (1:10000 dilution; Abcam, ab97225) was added and incubated for 1 h. Finally, after washing three times with wash buffer, the wells were developed with 100 µL of TMB (Solarbio, PR1200) for 3 min at 37 °C, then the reaction was stopped with 50 µL of ELISA Stop Solution (Solarbio, C1058). The absorbance was measured at 450 nm (reference: 630 nm). Data was analyzed by four-parameter nonlinear regression using GraphPad Prism 8.4.2 software.

Mice immunization

Specific pathogen-free (SPF) female BALB/c mice (6-8 weeks) were randomly divided into 6 groups with 8 mice per group, used for intramuscular immunization. Each mouse was injected with 100 µL of vaccine sample including 5 µg purified Marburg GP glycoprotein antigens, 200 µg aluminum adjuvant (InvivoGen, vac-alu-50), or PBS solution. The five purified Marburg GP glycoprotein antigens included Expi293F-derived GPΔTM, S2-derived GPΔTM, Expi293F-derived GPΔTM ΔMuc, S2-derived GPΔTM ΔMuc peak1, and S2-derived GPΔTM ΔMuc peak2. The control mice were vaccinated with PBS. The mice were immunized at weeks 0 and 3, and the blood samples were collected from the tail vein and processed by centrifugation on days 0, 14, 21 and 35. Then the processed serum was heated and inactivated at 56 °C for 30 min, aliquoted, and then stored at -80 °C until analysis was performed.

Antibody measurement

To measure GP-specific antibody responses in serum samples by ELISA, purified GP was coated into the wells of 96-well microtiter plates (Costar, 9018) at 10 µg/mL (100 µL) and incubated at 4 °C overnight. After blocking, Sera at 1:100 were three-fold serially diluted and added to the wells (100 µL) as primary antibodies, followed by HRP-conjugated anti-mouse IgG antibody (1:10000 dilution; Abcam, ab97265). After color development, the absorbance was read at 450 nm and 630 nm on a microplate reader (TECAN). For the given serum sample, the cutoff value is defined as 2.1 times the reading of the blank control (without serum added). The endpoint titers were defined as the dilution of the cutoff value.

Cell-surface binding using fluorescence activated cell sorting (FACS)

For the cell surface-displayed GP, 4 µg of plasmid encoding full-length GP was transfected into HEK293T cells cultured in T75 flasks using VirusGEN transfection reagent (Mirus, MIR 6700). After 24 h, cells were detached with 2% (v/v) FBS in PBS and transferred at a total of 5 × 10⁵ cells to each flow tube, then stained with polyclonal sera (1:100) obtained from mice vaccinated with different forms of GP. Cells were subsequently stained with anti-mouse IgG antibody conjugated APC (BioLegend, 405308) followed by flow cytometry analysis on a FACSCanto II flow cytometer (BD Biosciences). Data were analyzed with FlowJo software, using the following gating strategy: size & granularity > single cells > GP+ (Ab positive). All results were expressed as mean fluorescence intensity (MFI) or cell percentage.

Pseudovirus-based neutralization assays

For pseudovirus packaging, the gene encoding the full-length GP of Angola was human codon-optimized and inserted into the pcDNA3.1 vector to construct GP protein-expressing plasmids. HEK293T cells were inoculated into T75 flasks (Corning, 430641) and cultured at 37 °C and 5% CO₂ to 70-90% confluence for transfection. 4 µg of pcDNA3.1-Marburg GP were co-transfected with 32 µg of HIV backbone vector pNL4-3.Luc.R-E- into HEK293T cells using the VirusGEN transfection reagent (Mirus, MIR 6700). Supernatants containing pseudovirus particles were collected at 48 h post-transfection by centrifugation at 800 × g for 5 min. The harvested pseudovirus solution was stored in aliquots at − 80 °C after filtering through a 0.45 μm filter.

The titers of neutralizing antibodies in these sera were detected by HIV-based pseudovirus-type neutralization assay. For neutralization testing, serum samples were diluted at 1:20 using Dulbecco’s Modified Eagle’s Medium (Gibco, C11995500BT) with 10% fetal bovine serum at 50 uL per well. The serum samples were mixed with HIV-based pseudovirus-type virus with an equal volume (50 µL) and then incubated at 37 °C for 1 h before adding to HEK293T cells. After 48 h at 37 °C, the luciferase activity was measured by a microplate reader (TECAN) using the Bright-Lite Luciferase Assay System (Vazyme, DD1204). The formula for calculating the percentage of neutralization is “100% - (sample signals - blank control signals) / (virus control signals - blank control signals) × 100%”. The neutralization percent of each sample was performed using the GraphPad Prism 8.4.2 software.

Competitive ELISA

To compare the antibody epitope-specific responses, competitive binding enzyme-linked immunosorbent assay (ELISA) was performed. For the competitive ELISA, the steps were the same as the ELISA method described previously. First, Costar™ 96-well assay plates (Costar, 9018) were coated overnight at 4 °C with 1 µg/mL of Marburg GP. Wash buffer was used to wash the 96-well plate three times; mouse antisera, which was serially diluted from 1:10, was added to the wells and incubated at 37 °C for 1 h. MR228 antibodies were conjugated with HRP by using the EZ-Link Plus Activated Peroxidase kit (Solarbio, EX7000) according to the instructions. The plates were washed 3 times prior to the addition of HRP-conjugated MR228 (0.03 µg/mL) to the wells and incubated at 37 °C for 1 h. Finally, the cleaning step was repeated, and the wells were developed with 100 µL of TMB (Solarbio, PR1200) for 3 min at 37 °C, then the reaction was stopped with 50 µL of termination buffer (Solarbio, C1058). After color development, colorimetric analysis was performed at 450 nm and 630 nm in a microplate reader (TECAN). The formula for calculating the inhibition rate is “(blank control signals - sample signals) / blank control signals × 100%.”

Statistical analysis

Where applicable, results are expressed as the mean ± SD of representative results. Comparisons of GP-specific antibody titers (Fig. 1B) and cell-surface binding results (Fig. 1C) were performed using Kruskal-Wallis one-way ANOVA. Comparisons of neutralization assays (Fig. 1D) and inhibition rates (Fig. 1F) between groups were performed using one-way ANOVA. Differences were considered statistically significant at a P value of 0.05 (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). All statistical analyses were performed with GraphPad Prism 8.4.2 software.

Fig. 1.

Immunogenicity of GPs derived from Expi293F and S2 cells. (A) Mice immunization schedules. Mice (n = 8) were received 5 µg GP formulated with aluminum adjuvant at weeks 0 and 3. (B) GP-specific antibody titers induced by five GPs at week 5. (C) Binding of cell-surface Marburg GP proteins to mouse sera at week 5 measured by flow cytometry. P values for graphs B and C were determined using Kruskal-Wallis ANOVA, and significant P values between immunized groups are indicated (*P < 0.05). (D) Neutralization assay. Mouse antisera collected at week 5 were diluted 1:20 and then tested for neutralization of HIV-based pseudovirus-type virus. Each symbol represents one animal. (E) Competitive ELISA. Mouse antisera were serially diluted and tested for inhibition of MR228 binding of GP. Means ± SD of OD450-630 nm readings for all animals in each group are shown. (F) Inhibition ratio. The inhibition rate was calculated at a dilution of 1:10 in competitive ELISA. The higher value represented the more MR228-like antibodies contained in the serum. The asterisks in Figures D and F represented significant differences (one-way ANOVA) between groups: **P < 0.01, ****P < 0.0001. All significant differences from the PBS group are not labelled.

Results

Expression of soluble Marburg GP glycoprotein in Expi293F and Drosophila S2 cells

The highly glycosylated GP demonstrates conservation at both the N- and C-terminal regions while exhibiting more variability in its middle region [31]. Removal part of mucin-like immunogens can affect the epitope exposure [32] and was used to direct the immune response away from the highly variable mucin-like domains [18]. In order to investigate the contribution of glycosylation to the antigenicity and immunogenicity, GP was produced in mammalian (Expi293F) and insect (S2) cell lines, which yield altered glycan structures [33]. Two constructs of the Marburg GP were constructed: the ectodomain lacking the hydrophobic C-terminal transmembrane anchor which was called GPΔTM (including amino acids 1-637), then removed a portion of the mucin-like domain (residues 264-425) which was called GPΔTM ΔMuc (Fig. 2A).

Fig. 2.

Expression of soluble Marburg GPs in Expi293F and Drosophila S2 cells. (A) Schematic representation of GP expression cassettes used in mammalian (Expi293F-GP) and insect cell (S2-GP) vectors. Dashed lines represent deleted regions. MLD, mucin-like domain; IFL, internal fusion loop; TM, transmembrane domain. A red triangle indicates the furin cleavage site. The GP2-wing region, which is unique to MARV, is colored orange. (B and C) Superdex 200 size-exclusion chromatography profile of GPs following StrepTrap column. Graphs from left to right respectively represent Expi293F-derived GPΔTM, S2-derived GPΔTM, Expi293F-derived GPΔTM ΔMuc, S2-derived GPΔTM ΔMuc peak1 and S2-derived GPΔTM ΔMuc peak2. Two peaks were marked by arrow. mAU, milliabsorbance units.

To produce the GP, we used the Expi293F and Drosophila S2 cell expression system. GPs were mostly soluble expressed and secreted into the culture medium. To define the oligomeric state of our cell-produced GPs, we further purified GP using Superdex 200 columns (GE Healthcare). Interestingly, two major peaks were identified in the SEC chromatogram for S2-derived GPΔTM ΔMuc peak2, whereas the other proteins all showed one peak (Fig. 2B).

Physiochemical properties of GPs from the different expression systems

To analyze the expressed GP protein, SDS-PAGE and Western blot were employed under reducing and non-reducing conditions. GP protein expressed from mammalian cells (Expi293F) showed as high-molecular-weight proteins, which is probably due to the high levels of glycosylation or the presence of aggregates. For S2-derived GPΔTM, there was a main band that migrated between the 100 and 130 kDa molecular weight markers (Fig. 3A, C and E). And GPΔTM ΔMuc showed a single band at ~ 80 kDa that was predominantly detected in peak2, whereas peak1 contained large-molecular-mass proteins probably representing GP aggregates (Fig. 3B, D and F). The size of Expi293F-derived GPs was slightly larger than S2-derived GPs. One possible explanation is that the extent of glycosylation was higher in the Expi293F cells.

Fig. 3.

Physiochemical properties of GPs from the different expression systems. (A and B) SDS-PAGE analysis of purified Expi293F- and S2-derived GPΔTM or GPΔTM ΔMuc under nonreducing and reducing conditions. GP, GP1 and GP2 are indicated by the arrows. (C-E) Western blot analysis of purified Expi293F- and S2-derived GPΔTM or GPΔTM ΔMuc. (C and D) The membranes were subjected to Western blotting with MR78. (E and F) Anti-Strep antibody was used as the detection antibody. The + represents the reduced condition, and the - represents the non-reduced condition. (G) High-pressure liquid chromatography (HPLC) profiles of the purified GPs in Expi293F and S2 cells. (H) Native-PAGE of purified GPs. The results are a representative result of the experiment.

The presence of additional bands above the GP protein on the SDS-PAGE and Western blot may indicate the presence of aggregates. In order to assess the potential nature of the bands, we next investigated the physiochemical properties of the GPs from mammalian cells and insect cells. First, HPLC was carried out to measure the oligomerization of the GP proteins. All GPs showed a single major peak in the HPLC profiles, and the peak of S2-derived GPΔTM ΔMuc peak2 was narrower (Fig. 3G), indicating a higher homogeneity. Furthermore, to get an insight into the high-molecular-weight constructs, native PAGE was further performed. The results indicated the formation of aggregation of higher molecular weight. Specifically, the S2-derived GPΔTM ΔMuc peak2 ran between 300 and 440 kDa, likely representing a trimer, which was consistent with the native oligomeric state of the protein (Fig. 3H).

GPs derived from Expi293F and S2 cells display different glycosylation and affect the exposure of antigenic epitopes

The molecular weight of purified GPs was much higher than the predicted mass (72 kDa and 56 kDa) based on its amino acid sequence using the Geneious 4.8.3 software, suggesting possible glycosylation. To investigate the extent of glycosylation, GPs were stained by the periodic acid Schiff (PAS) technique for glycoproteins, using a Pierce glycoprotein staining kit (Thermo Scientific). The results revealed clear glycoprotein-containing bands that were consistent with the WB results, showing that the GPs produced by Expi293F cells were indeed more glycosylated (Fig. 4A). To further examine the extent and pattern of glycosylation, GPs were digested with the endoglycosidases H (Endo H) and peptide-N-glycosidase F (PNGase F). Endo H could remove the N-linked high-mannose from the glycoprotein but leave behind one N-acetyl Glucosamine (GlcNAc) on the N-linked glycoprotein [34]. And PNGase F is an amide hydrolase, which could hydrolyze almost all N-linked sugars in glycoproteins. As shown in Fig. 4B, PNGase F treatment led to a more discernible reduction in Expi293F-derived GP, which indicated the extent of glycosylation expressed from Expi293F was higher as compared with that derived from the S2 cells. In contrast, Endo H digestion resulted in a band only slightly below that of untreated GP, suggesting that deglycosylated GP may leave a GlcNAc on the N-linked glycoprotein or contain endo H-resistant glycan types [17], such as paucimannose N-glycans, as previously reported for other S2 cell-produced glycoproteins [33].

Fig. 4.

Analysis of glycosylation extent of GPs and their effect on antibody recognition abilities. (A) Periodic acid schiff staining to visualize glycoproteins under reducing and nonreducing conditions. The + represents the reduced condition, and the - represents the non-reduced condition. (B) Analysis of GP by PNGase F or Endo H digestion. Glycosidase-treated and untreated samples were subjected to Western blotting with anti-Strep antibody as the detection antibody. (C) Reactivities of different GPs with antibodies tested by ELISA.

Antigens with different degrees of glycosylation may influence activities. To investigate the differences of binding activity with antibodies, GPs were assessed by using two well-characterized, GP-specific antibodies like MR78 and MR191, which target RBS [22, 23]. MR228, which anchors to the Marburgvirus-specific “wing” region, has also been chosen. ELISA analysis revealed that GPs efficiently reacted with all tested mAbs in a dose-dependent manner (Fig. 4C). These data demonstrated that GP acquired a conformation critical for binding antibodies. Moreover, mammalian GPs bound mAbs less efficiently than insect GPs did, suggesting that epitope exposure of GP may be affected by glycosylation.

Immunization of mice with GPs induces Marburg GP-specific antibodies

To investigate the impact of the GP glycosylation and oligomeric state on specific antibody induction, five groups of BALB/c mice were immunized with GP antigens, respectively (Fig. 1A). Firstly, GP-specific IgG was measured by end-point ELISA. The GP-specific antibodies were induced in all GP-immunized groups. Notably, the GP-specific IgG titers in the Expi293F-derived GPΔTM ΔMuc group reached 4-5 log, which was higher than those observed in the other groups (3-4-log) (Fig. 1B). Moreover, to study the reactivity of mouse serum that bound to the cell surface-displayed GP, full-length GP proteins were expressed on 293T cells, and the reactivity was assessed by flow cytometry (Fig. 1C). The results showed that the sera of group Expi293F-derived GPΔTM ΔMuc reacted well to the cell-surface-expressed antigens, whereas the polyclonal sera elicited by other forms of GP reacted weakly, demonstrating a consistent trend with the response of binding antibodies measured by ELISA.

Next, the neutralizing ability of the sera was measured by a pseudovirus-based neutralization assay. These results demonstrated that mouse serum did not neutralize the MARV pseudovirus (Fig. 1D), as previously reported that most MARV vaccines under development generate low to no neutralizing antibody response in animal models [35]. Some animal experiments have shown that certain non-neutralizing antibodies can also provide complete protection in mice. Such as MR228 [18], a non-neutralizing mAb, anchors to the Marburg virus glycoprotein-specific “wing” region, conferring in vivo protection through the induction of strong Fc domain-mediated effector [10]. To determine whether the induced sera contain mAbs targeting the protective epitopes like MR228, a competitive ELISA in which HRP-conjugated MR228 was used as a detecting antibody was conducted. The results showed that the binding of GPs to MR228 was blocked in a dose-dependent manner by the preincubation with the sera from the five GP groups (Fig. 1E). Among these groups, the sera from Expi293F-derived GPΔTM ΔMuc was observed to have superior inhibitory effects (Fig. 1F).

Discussion

The spread of the Marburg epidemic to Rwanda in 2024 indicates further expansion, which highlighted the urgency of more research on this highly lethal infectious disease pathogen to find available effective vaccines and therapeutics. For MARV, the glycoprotein GP was particularly valuable as an antigen due to it being the only antigen exposed on the surface of the virus and its role in virus-host interactions, making GP a main candidate antigen for the development of vaccines and antibodies. Numerous studies have shown that viral proteins expressed by distinct expression systems exhibited divergent immunogenicity due to their unique glycosylation patterns. For example, the Hepatitis C virus (HCV) E2 produced in Drosophila insect (S2) cells was found to be more immunogenic than the corresponding protein produced in HEK293T cells [29]. In contrast, the fully glycosylated 293 F S-2P induced higher IgG and neutralizing antibody titers than Bac S-2P in Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) [30]. These findings suggested that distinct glycosylation patterns should be taken into consideration in the development of a recombinant vaccine. We hypothesized that the immunogenicity of Marburg virus GP protein may also be influenced by its expression in different cell lines. To further test the hypothesis that differential glycosylation may influence the immunogenicity of GP, we performed antigenic and immunogenic comparisons of GP produced in Expi293F cells and S2 cells, which confer different glycans [25].

In the present study, we demonstrated that the glycosylation pattern associated with Expi293F cell-derived GP may be more favorable for eliciting bNAbs than S2 cell-produced GP. It may be due to the principle that immunogens with high molecular weights may perform stronger immunogenicity. Another possibility is that the insect cells are phylogenetically distant from human cells and bear a similar yet simpler glycosylation pattern as compared with mammalian cells [36]. The mammalian-derived proteins demonstrated high percentages of complex N-linked glycans, but the insect-derived proteins were found to have the majority of simple N-linked glycans and high mannose structures. Furthermore, the mammalian system also produced sialylated N-glycans, but the level of sialylation in insect cell expression systems was minimal [37]. Sialylation has been shown to play an important role in the function of some glycoproteins, which could impart significant functional differences [38], such as immunoregulation [39]. Moreover, mammalian cells produce more extensive O-linked glycosylation than insect cells, which may also affect the protein function [40]. The mucin-like domain may mask the GP from immune system surveillance, with it containing a dense clustering of N- and O-linked glycans [41, 42], whereas its removal may leave some viral epitopes of the internal fusion loop, base, or stalk more exposed to the humoral immune system [35, 43].

Notably, in our study, production of GP in S2 cells resulted in higher antigenicity than the equivalent material produced in mammalian cells. We hypothesized that this phenomenon was primarily attributable to the increased exposure of antigenic epitopes as a result of its reduced glycosylation. A number of studies have highlighted the role of N-linked glycosylation in masking or shielding epitopes from recognition by bNAbs [20, 44, 45], further validating our suspicions. Additionally, the SDS-PAGE bands except peak2 appeared more complex under reducing conditions, which we attributed to the possibility of disulfide bond formation between GP and other intra- or extracellular (from media) proteins, which was also similar in the proteins of Crimean-Congo hemorrhagic fever virus glycoprotein Gc [46]. Ebola virus also identified an additional contamination band of approximately 80 kDa by mass spectrometry as bovine serum transferrin, a component of fetal bovine serum in the culture medium [40].

Meanwhile, there are limitations in this study. Pseudovirus neutralization assays were evaluated only under BSL-2 conditions owing to the limitation of the BSL-4, and the protective efficacy in vivo needs to be assessed to prove whether the Expi293F-derived GPΔTM ΔMuc deserves further development. At present, the biophysical characterization of Expi293F-derived GPΔTM ΔMuc remains relatively limited, and further experiments (such as thermal stability analysis and negative-stain electron microscopy) are required to comprehensively evaluate its potential as a vaccine. Additionally, mass spectrometry analysis is needed to further confirm the glycan compositions of GPs expressed in different cells. As glycosylation modification could modulate the immunogenicity and neutralizing-epitope recognition [34], future efforts could focus on complete deletion of specific N-glycans proximal to neutralizing epitopes. Instead of relying on expression system-dependent modification of all glycans, the enzymatic truncation of glycoforms for targeted deletion of specific glycans may offer a more precise strategy to further enhance immunogenicity. For example, influenza virus vaccine design was improved by removing the outer part of glycans from the virus surface of the HA protein [47, 48], and HIV improved the ability of immunogenicity by site-selective deglycosylation [49]. Future studies could also consider structure-based design. Building upon the continuous advancements in structural biology techniques, the three-dimensional structures of numerous antigens have been resolved, unveiling promising key antigenic epitopes. Leveraging this structural information has become instrumental for the rational design of vaccines. Rational vaccine design strategies, such as helix breaker insertion and cavity filling [50], have demonstrated remarkable utility in developing vaccines against various pathogens [51–53]. Studies have demonstrated that Ebola and Marburg virus GP trimer structures designed with stabilizing mutations enhance trimer expression levels [54]. Therefore, structure-based rational design targeting Expi293F-derived GPΔTM ΔMuc may also serve as a promising strategy to further improve their immunogenicity.

Conclusions

In summary, the alteration of Marburg GP glycosylation indicates that Expi293F-derived GPs contain more glycosylation, and the result of immunogenicity indicates that Expi293F-derived GPΔTM ΔMuc is a more potent immunogen. Research efforts not only will lead to a better understanding of the importance of glycosylation for vaccine development but also could provide a basis for the design of more effective MARV immunogens.

Abbreviations

MARV

Marburg virus

GP

Glycoprotein

MVD

Marburg virus disease

RAVV

Ravn virus

RBS

Receptor-binding site

mAbs

Monoclonal antibodies

HIV

Human immunodeficiency virus

HPLC

High-performance liquid chromatography

PNGase F

peptide-N-glycosidase F

Endo H

Endoglycosidase H

FACS

Fluorescence activated Cell Sorting

HCV

Hepatitis C virus

SARS-CoV-2

Severe acute respiratory syndrome coronavirus 2

IFL

Internal fusion loop

Author contributions

CMY and GYZ conceived and designed the study; JL, SYW, YC, LYS, ZWS, PH, YXW and LB performed the experiments; JML, XYC, TF, YZD, RHL and PFF guided the experiments; JL and GYZ wrote and revised the manuscript. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

BALB/c mice studies were approved by the Institutional Animal Care and Use Committee of the Laboratory Animal Centre of the Sino Animal (Beijing) Science and Technology Development Company (approval no. IACUC-20240301YZA-3R) in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals.

Consent for publication

The manuscript has been given publication permission by all authors.

Competing interests

The authors declare no competing interests.