Abstract
Marburg virus (MARV) and Ravn virus (RAVV) are highly pathogenic filoviruses that cause Marburg virus disease (MVD), for which no approved treatment currently exists. In this study, we describe a panel of 10 novel mouse-derived monoclonal antibodies (mAbs), all exhibiting neutralizing activity against MARV. Notably, five of these mAbs also cross-neutralize RAVV. Based on amino acid sequences, neutralization profiles, and escape mutations in the MARV glycoprotein (GP), the mAbs were classified into three groups. mAbs targeting the receptor-binding domain (RBD) and mucin-like domain exhibited MARV-specific neutralization, whereas those targeting epitopes spanning the RBD and internal fusion loop exhibited cross-neutralization. Post-exposure administration of cross-neutralizing mAbs (AGP-R7, AGP-R12) provided significant protection in MARV and RAVV infection mouse models. These findings identify a conserved neutralizing epitope with therapeutic potential and highlight AGP-R7 and AGP-R12 as promising candidates for MVD treatment. This study offers new insights into cross-reactive antibody strategies against multiple filoviruses.
Similar content being viewed by others
Introduction
Marburg virus (MARV) and Ravn virus (RAVV) are members of the family Filoviridae, genus Orthomarburgvirus, and species Orthomarburgvirus marburgense. These filoviruses cause Marburg virus disease (MVD), a severe hemorrhagic fever that is often fatal in humans and nonhuman primates. The case fatality rate for MVD varies between 24% and 88%, depending on the MARV variant involved and the quality of case management1,2. MARV infection was first identified in 1967, and since then, sporadic outbreaks have been reported in several African countries, including the Democratic Republic of the Congo, Angola, Uganda, Guinea, Ghana, Equatorial Guinea, Tanzania, and Rwanda3,4. Notably, recent outbreaks of MVD in Guinea, Ghana, Equatorial Guinea, Tanzania, and Rwanda—regions with no prior reported cases before 2020—have underscored the urgent need to develop effective prophylactic and therapeutic interventions against this deadly disease.
Among the seven structural proteins of filoviruses, the surface glycoprotein (GP) is the primary mediator of viral entry, facilitating both binding to host cell receptors and fusion of the viral envelope with the host cell membrane. Notably, GP is the sole target for neutralizing antibodies5. In the case of Ebola virus disease, caused by Ebola virus (EBOV)—a member of the Orthoebolavirus genus within the same family—two FDA-approved GP-specific antibody therapies are available: the monoclonal antibody cocktail REGN-EB3 and the single monoclonal antibody (mAb) 1146. However, these therapeutic antibodies are ineffective against other human-pathogenic filoviruses, including other orthoebolaviruses (e.g., Sudan virus and Bundibugyo virus), MARV, and RAVV, due to significant antigenic differences. Although several broadly cross-reactive mAbs targeting orthoebolavirus GPs have been identified7,8,9, none have effectively neutralized MARV or RAVV10,11,12. Consequently, cross-reactive and protective mAbs and target B-cell epitopes for MVD treatments remain poorly characterized. Several MARV-specific neutralizing antibodies, such as MR78 and MR191, have been reported to confer protection in experimental animals challenged with MARV or RAVV10,11,12. However, their neutralizing potency and breadth are limited, and their incomplete neutralization raises concerns about the potential for viral escape10. Thus, there is a pressing need to develop novel cross-neutralizing antibodies with strong neutralizing activity against both MARV and RAVV.
In this study, we generated mouse-derived mAbs with potent neutralizing activity against MARV. The 10 mAbs produced were classified into three groups based on predicted epitope specificity and the amino acid sequences of their heavy and light chain variable regions. Five of these mAbs (AGP-R1, AGP-R7, AGP-R12, AGP-R15, AGP-R17) exhibited cross-reactivity with both MARV and RAVV, likely targeting an epitope that spans the GP fusion loop and receptor-binding domain (RBD). Conversely, AGP-R4, AGP-R5, AGP-R11, AGP-R20, and AGP-R22 specifically neutralized MARV but not RAVV and were thought to bind epitopes located within the GP head region. The epitopes targeted by AGP-R4 and AGP-R5 appear to overlap with, but differ slightly from, those targeted by AGP-R11, AGP-R20, and AGP-R22. The therapeutic efficacy of these mAbs was validated in mouse models of MARV and RAVV infection. Notably, AGP-R7 and AGP-R12, which exhibited strong cross-neutralizing activity, provided complete protection against lethal challenges with both viruses, highlighting their potential as broadly protective therapeutic candidates for MVD. These mAbs exhibited neutralizing and protective effects comparable to, or surpassing, those of the previously reported antibody MR7810,11. Therefore, we provide a detailed characterization of the biological properties of the mAbs generated in this study.
Results
Generation of AGP-R mAbs against MARV GP
To generate mAbs targeting the MARV GP, we produced hybridomas from spleen cells of mice infected with a recombinant, replication-competent vesicular stomatitis Indiana virus (VSIV) expressing the MARV GP gene (rVSV/MARV-Angola) in place of the native VSIV G gene13,14. This was followed by booster immunization using purified soluble recombinant MARV GP7,15. Hybridoma supernatants were screened for neutralizing activity against replication-incompetent VSIV pseudotyped with MARV and RAVV GPs (VSVΔG*MARV-Angola, VSVΔG*MARV-Musoke, VSVΔG*RAVV)7,16,17. From this screening, we identified 10 mAb clones (designated AGP-R mAbs: AGP-R1, R4, R5, R7, R11, R12, R15, R17, R20, and R22) exhibiting divergent amino acid sequences in the variable regions of their heavy and/or light chains, which differed from all previously characterized mAbs with available sequence data18,19,20,21,22. These mAbs possessed kappa light chains and IgG2a heavy chains, except for AGP-R7, which had an IgG3 heavy chain.
In vitro properties of AGP-R mAbs
The 10 AGP-R mAbs were classified into three groups—R1 (AGP-R1, -R7, -R12, -R15, and -R17), R4 (AGP-R4 and R5), and R11 (AGP-R11, -R20, and -R22)-based on the amino acid sequences of their complementarity-determining regions (CDRs) (Fig. 1) and their neutralization profiles against pseudotyped viruses (VSVΔG*MARV-Angola, VSVΔG*MARV-Musoke, and VSVΔG*RAVV) (Fig. 2a and Supplementary Table 1). The R1-group mAbs efficiently neutralized all three pseudotyped viruses. Among them, AGP-R7 exhibited the highest neutralizing activity in the R1 group with 50% inhibitory concentrations (IC50) values of 0.39, 0.41, and 0.61 μg/mL against VSVΔG*MARV-Angola, VSVΔG*MARV-Musoke, and VSVΔG*RAVV, respectively. mAbs in the R4 and R11 groups showed high neutralizing activity against VSVΔG*MARV-Angola and VSVΔG*MARV-Musoke but failed to neutralize VSVΔG*RAVV. Consistent with these neutralization profiles, R1-group mAbs bound to both MARV and RAVV GPs, with a slight preference for MARV GPs, whereas R4- and R11-group mAbs bound exclusively to MARV GP (Supplementary Fig. 1).
Amino acid sequences of the VH (a) and VL (b) regions of R1-group mAbs. Amino acid sequences of the VH (c) and VL (d) regions of R4- and R11-group mAbs.
a Neutralizing activities of mAbs were assessed using VSVΔG*MARV-Angola (blue), VSVΔG*MARV-Musoke (orange), VSVΔG*RAVV (gray), and VSVΔG*EBOV (green). Relative infectivity was calculated by setting the infectivity value in the absence of mAbs as 100%. Data represent the mean ± standard errors of three independent experiments. Logistic curves were fitted using nonlinear four-parameter regression and are shown as solid lines. b Competitive binding among mAbs—AGP-R7, -R12, -R17, -R4, -R5, -R11, -R20, MR78, and MR191—was examined using ELISA. AGP-R7, -R12, -R17, -R4, -R5, -R11, -R20, and MR78 were labeled with HRP, and the unlabeled mAbs shown on the right were used as competitors.
R1-group mAbs did not compete for epitope binding with mAbs from the R4 or R11 groups but did compete among themselves within the R1 group. Conversely, mAbs in the R4 and R11 groups exhibited competitive binding with each other (Fig. 2b). None of the mAbs generated in this study exhibited epitope competition with previously characterized antibodies MR78 and MR19110,11. Subsequently, the neutralizing activity of the mAbs against authentic MARV and RAVV was assessed using plaque reduction neutralization (Fig. 3, Supplementary Table 2). R1-group mAbs exhibited over 80% neutralization of both MARV-Angola and RAVV at a concentration of 100 μg/mL, comparable to or exceeding the activity of protective mAbs reported in previous studies10,11,21,23,24,25,26. Notably, AGP-R7 achieved more than 80% neutralization against both MARV and RAVV at 25 μg/mL, with IC50 values of 3.36 and 5.19 μg/mL, respectively. In contrast, MR78, AGP-R4, and AGP-R11 only partially neutralized RAVV. Similarly, although MR78 and the R4- and R11-group mAbs showed high neutralizing activity against pseudotyped viruses expressing MARV GPs, they only partially neutralized authentic MARV.
Neutralizing activities of mAbs against MARV (Angola strain) (a) and RAVV (b) were evaluated using plaque reduction neutralization tests. Relative infectivity was calculated by setting the infectivity value of the control in the absence of antibodies as 100%. Data represent the mean ± standard error of three independent experiments. Logistic curves were generated using nonlinear four-parameter regression analysis and are shown as solid lines; dashed lines represent 95% confidence intervals. The coefficient of determination (R2) values, indicating the goodness of fit, are shown in the graph.
Identification of putative epitopes of AGP-R mAbs
To identify the putative epitopes targeted by the mAbs, GP escape mutants were generated using rVSV/MARV-Angola. rVSV/MARV-Angola were propagated in the presence of each mAb, and escape mutants were subsequently isolated through plaque cloning7. Full-length GP genes from these mutants were sequenced and compared with the parental virus to identify amino acid substitutions (Fig. 4). Escape mutants selected by R1-group mAbs contained amino acid substitutions at residues K58, K90, K120, D459, N508, E509, N510, A514, and N564, which are located within the RBD or internal fusion loop (IFL) of the GP. Structural modeling revealed that these amino acid positions were clustered within the lateral surface of the MARV GP trimer11. Conversely, escape mutants selected by R4-group mAbs exhibited amino acid substitutions at residues within the mucin-like domain (MLD), including F447, L448, L451, L452, P455, I456, F458, and D459. Due to gaps in the available crystal structure (residues 1–32, 181–468, 499–504, 630–681), these MLD-associated substitutions could not be mapped to the 3D structure. Escape mutants selected by R11-group mAbs exhibited amino acid substitutions both within the RBD (Q126, H131, Y197, T199) and the MLD (F447, L451, I452, P455, F458, D459), partially overlapping with the mutations observed in the R1- and R4-group mutants. While the substitutions in MLD could not be structurally localized due to the aforementioned gaps in the structural data, some unique positions in RBD, specifically Q126 and H131, were mapped near the glycan cap at the head region of the GP.
a Amino acid substitutions in MARV GP identified in escape mutants selected by each mAb are shown. b A schematic of MARV GP domains: the GP1 subunit comprises the receptor binding domain (RBD), glycan cap, and mucin-like domain (MLD); the GP2 subunit includes the internal fusion loop (IFL), heptad repeats 1 and 2 (HR1 and HR2), the transmembrane domain (TM), and the cytoplasmic tail. Amino acid substitutions identified in escape mutants of R1-, R4-, and R11-group mAbs are indicated in red, green, and blue, respectively. c Amino acid residues associated with putative mAb epitopes are mapped on the GP trimer structure (PDB ID: 6BP2). Structural data are unavailable for amino acid positions 1–32, 181–468, 499–504, and 630–681.
Epitopes for cross-neutralization by R1-group mAbs
To gain further insight into the putative epitopes recognized by cross-neutralizing R1-group mAbs, we conducted molecular modeling and docking analyses using the available crystal structure of the RAVV GP and the AlphaFold-predicted structure of AGP-R7 (Fig. 5a). The resulting docking model indicated that the surface region of RAVV GP predicted to interact with AGP-R7 included amino acid residues K58, K90, K120, N510, E508, E509, and A514—all of which were also identified in escape mutants selected by R1-group mAbs. To assess the breadth of cross-neutralization conferred by R1-group mAbs, we analyzed the phylogenetic relationships among MARV and RAVV GP sequences and compared the amino acid sequences within the putative epitope regions. Phylogenetic analysis based on full-length GP sequences revealed four distinct viral lineages of MARVs and RAVVs (Supplementary Fig. 2). Lineage A included variants from Angola, Ghana, and Equatorial Guinea, representing recent West African outbreaks. Lineage B encompassed classic East African strains such as Musoke and Ci67, while lineage C consisted of Central African strains from the Democratic Republic of the Congo and Uganda. RAVV strains from Kenya and Uganda formed a separate cluster designated as lineage D. A comparison of the putative epitope regions across multiple MARV and RAVV variants revealed that the amino acid residues involved in the epitope of R1-group mAbs are highly conserved among all lineages (Fig. 5b). Although the amino acid residues implicated in the putative epitopes of R4- and R11-group mAbs also appeared largely conserved, this finding was inconsistent with their observed neutralization and binding profiles. However, many of these amino acid residues are located within the MLD (Fig. 5b), a region known to vary among MARV and RAVV strains, suggesting that additional amino acid residues outside the aligned region may contribute to epitope formation for these MARV-specific mAbs. The high level of amino acid sequence conservation across these lineages in key functional domains of GP—particularly the RBD, IFL, and heptad repeat 1 (HR1) (Fig. 5b, Supplementary Fig. 3)—likely underlies the broad cross-neutralizing activity observed with R1-group mAbs.
a Docking model of the RAVV GP trimer in complex with the AGP-R7 Fv fragments. The GP protomers are depicted in black, gray, and white, with the AGP-R7 variable heavy (VH) and light (VL) chains shown in light pink and light blue, respectively, in both surface and ribbon representations. Predicted contact residues of RAVV GP with AGP-R7 are highlighted in pale green, while amino acid residues identified in escape mutants are marked in red. b Amino acid conservation of putative epitopes across MARV and RAVV GPs. Partial multiple sequence alignments of GP regions are shown for the receptor-binding domain (RBD; 55–60, 86–95, 116–135, 191–200), mucin-like domain (MLD; 441–460), internal fusion loop (IFL; 501–515), and heptad-repeat 1 (HR1; 561–570). Sequences include MARV variants Angola, Ghana/2022, Equatorial Guinea 2023, Musoke, Ci67, 07DRC99, 01Uga07, 05DRC99, and RAVV variants Kenya 1987, Uganda 20070648, and Uganda 02Uga07. Asterisks denote amino acids identical to those in MARV-Angola. Putative epitope residues targeted by AGP-R1, -R4, and -R11 groups are shown in bold and indicated by red, blue, and green bars, respectively.
Neutralization mechanism of AGP-R mAbs
To explore the mechanisms underlying the neutralizing activity of AGP-R mAbs, we tracked the dynamics of fluorescently labeled virus-like particles (VLPs) composed of MARV matrix protein, nucleoprotein, and GP with or without mAbs using confocal microscopy (Fig. 6). This approach enabled us to assess the effect of mAbs on MARV entry into host cells7. At 0 h post-inoculation, AGP-R1, -R7, -R12, and -R4 significantly reduced the number of VLPs attached to the cell surface compared with control IgG, indicating effective inhibition of initial viral attachment (Fig. 6a). Conversely, AGP-R11 and MR78 exhibited minimal impacts on VLP attachment. At 5 h post-inoculation, the number of internalized VLPs treated with AGP-R1, -R7, -R12, and -R4 was also significantly reduced, consistent with the observed reduction in the number of VLPs attached to the cell surface (Fig. 6b). However, the size and fluorescence intensity of the internalized VLPs did not differ significantly between mAb-treated and untreated samples (Fig. 6c, d), suggesting that these mAbs do not inhibit viral membrane fusion once the VLPs have entered the cells.
DiI-labeled MARV VLPs (DiI-VLPs) were incubated with either untreated, control IgG, AGP-R mAbs, or MR78 and inoculated into confluent Vero E6 cells for 30 min at room temperature. After adsorption, cells were either fixed immediately (0 h) to assess DiI-VLP attachment to the cell surface (a) or incubated at 37 °C for 5 h to evaluate DiI-VLP internalization (b) and membrane fusion (c, d). The number of DiI-VLP dots per cell at the cell surface (a) and in the cytoplasm (b), along with the size and fluorescence intensity of DiI-VLP dots per field of view (c, d), were quantified using confocal laser scanning microscopy. Data are presented as means ± standard errors from three independent experiments. Statistical analysis was performed using a one-way repeated-measures ANOVA, followed by Dunnett’s multiple comparison test. Asterisks indicate significant differences compared with untreated controls (*p < 0.05, **p < 0.01, ****p < 0.0001).
Therapeutic efficacy of AGP-R mAb treatments in mouse models
The therapeutic efficacy of AGP-R mAbs was evaluated in lethal mouse models infected with MARV and RAVV. BALB/c mice were infected with a lethal dose of mouse-adapted MARV-Angola and treated with 100 µg of each mAb at 1 day post-infection (dpi) (Fig. 7a). Most mice treated with AGP-R mAbs exhibited no clinical symptoms, whereas all mice receiving negative control polyclonal mouse IgGs succumbed to infection within 7 days (Fig. 7a). In the AGP-R12-treated group, one mouse developed a scruffy coat at 6 dpi but recovered shortly thereafter. Similarly, in the AGP-R17 group, one mouse exhibited a scruffy coat at 6 dpi, and another showed a scruffy coat, hunched posture, and over 15% body weight loss at 7 dpi; both mice recovered by 9 dpi. One mouse in the AGP-R5 group died at 5 dpi. The previously reported protective mAb (MR78) conferred 60% protection against lethal outcomes. To further assess the cross-protective efficacy of AGP-R mAbs, mice infected with mouse-adapted RAVV were treated with either cross-neutralizing mAbs (AGP-R1, AGP-R7, AGP-R12, AGP-R17), a MARV-specific neutralizing mAb (AGP-R11), MR78, or control IgG (Fig. 7b). Neither AGP-R11 nor the control IgG conferred protection against lethal infections. Conversely, AGP-R7 and AGP-R12 conferred complete protection (100%) without any observable symptoms. MR78 also conferred 100% protection, while AGP-R1 and AGP-R17 achieved 80% protection against RAVV infection.
BALB/c mice (n = 5) infected with mouse-adapted MARV-Angola (a) or RAVV (b). At 1 day post-infection, mice were administered 100 µg of AGP-R mAbs, mouse MR78, or a mouse IgG control through intraperitoneal injection. Mice were monitored over 21 days for changes in body weight (top), survival (middle), and clinical score of disease (bottom). Each line represents an individual mouse.
Discussion
Recent outbreaks of MVD in sub-Saharan African countries, including Equatorial Guinea, Tanzania, and Rwanda3,4, have raised concerns about the potential for broader transmission of MARV and RAVV, underscoring the urgent need to develop effective medical countermeasures against these highly lethal pathogens. Despite their high case fatality rates and capacity for human-to-human transmission, no vaccines or therapeutics are currently approved for clinical use against MARV or RAVV. This contrasts with Ebola virus disease, for which mAb therapies such as REGN-EB3 (Inmazeb) and mAb114 (Ansuvimab) have shown strong clinical efficacy and received regulatory approval6. Given the potential of MARV and RAVV to cause future outbreaks and their genetic diversity, developing broadly cross-reactive neutralizing mAbs with potent therapeutic efficacy remains a significant global health priority.
In this study, we developed and characterized 10 mouse-derived mAbs targeting MARV GP. Based on their putative epitopes, the antibodies were categorized into R1, R4, and R11 groups. Among these, the R1 group (AGP-R1, R7, R12, R15, R17) exhibited strong neutralizing activity against both MARV and RAVV. Notably, although R4- and R11-group mAbs showed high neutralizing activity against VSVΔG*MARV-Angola in vitro, their ability to neutralize authentic MARV was limited. Similarly, the R1-group mAbs showed significantly higher neutralizing activities against pseudotyped viruses than against live MARV or RAVV. This discrepancy may be attributed to differences in the morphology and size of viral particles. Given that MARV and RAVV virions are filamentous and incorporate a greater number of GP molecules than the smaller VSIV particles, a higher quantity of mAb molecules may be required to neutralize each virion. Similar observations have been reported for other anti-MARV mAbs10, emphasizing the importance of validating mAb efficacy against authentic filoviruses. AGP-R7 showed exceptional potency, outperforming benchmark mAbs such as MR78. At concentrations of 25–100 µg/mL, AGP-R7 achieved near-complete neutralization, with IC50 values of 3.355 and 5.186 μg/mL against MARV and RAVV, respectively. Notably, among the R1-group mAbs, AGP-R7 contains several unique amino acids in its CDRs (one in HCDR1, two in HCDR2, one in HCDR3, and one in LCDR3), which may contribute to its enhanced neutralizing capacity.
To gain insight into the epitopes recognized by our cross-neutralizing mAbs, we analyzed the amino acid substitutions found in the escape mutants. Although neutralization assays using each escape mutant are needed to definitively determine which amino acids are crucial for antibody binding, our data provide important information on the putative epitopes of these mAbs. Multiple sequence alignments of MARV and RAVV GPs revealed high conservation of the amino acid residues involved in the putative epitopes of R1-group mAbs across representative viruses from each lineage. These residues localize to functionally constrained regions of GP, including RBD (K58, K90, K120) and IFL (N508, E509, N510, A514), which mediate receptor binding and membrane fusion, respectively5,27,28,29. This pattern of epitope targeting is conceptually similar to that observed for pan-orthoebolavirus mAbs, such as ADI-15878, ADI-15742, CA45, and 6D6, which bind to conserved conformational epitopes located at the base of GP, encompassing the GP1–GP2 interface, the IFL, or adjacent regions7,30,31,32,33. Their mechanisms of action include locking the GP trimer in its prefusion conformation, disrupting its interaction with the Niemann–Pick C1 (NPC1) protein (the only known endosomal fusion receptor for all mammalian filoviruses), or inhibiting subsequent membrane fusion. By analogy, R1-group mAbs are expected to exhibit similar cross-reactive properties. Conversely, R4- and R11-group mAbs likely target more variable regions, primarily within the MLD, which shows lineage-dependent amino acid differences, particularly between MARV and RAVV. Consistent with this, R4- and R11-group mAbs failed to neutralize RAVV in vitro, and AGP-R11 did not protect mice from lethal RAVV infection. Considering the molecular flexibility of the MLD, it should also be noted that MLD-recognizing antibodies may exhibit weaker neutralizing activity than those targeting more stable regions. The high level of amino acid conservation within the R1-group mAb epitope offers a plausible mechanistic explanation for its broad cross-neutralizing activity. These antibodies are thus considered promising therapeutic candidates against newly emerging MARV variants. For instance, recent strains from Ghana and Equatorial Guinea, both belonging to lineage A, retained all the key amino acid residues in RBD and IFL identified in this study.
Several neutralizing mAbs targeting MARV have been reported to inhibit viral entry into host cells10,11,21,34. For example, MR191 and MR78 bind to overlapping epitopes at the apex of GP1, adjacent to the RBD corresponding to the NPC1-C binding site5,10,11. MR78 primarily contacts surface residues P63, S67, W70, F72, I95, and I125, whereas MR191 engages the same structural region but extends more deeply into the hydrophobic trough, interacting with W70, F72, M154, and Q1285,10,11. By occupying this site, both antibodies prevent the interaction between GP and NPC1, thereby inhibiting subsequent membrane fusion10,11,35,36. Several escape mutations identified for the R11-group mAbs, such as Q126R and H131P, are located near the residues targeted by MR78 and MR191, suggesting that R11-group mAbs likely recognize a similar epitope on GP1. An escape mutation in the R4-group mAbs at P455 within the MLD corresponds to a previously reported MR78 escape site, indicating that the R4-group mAbs may interact with an adjacent region extending toward the MLD10. Conversely, the R1-group antibodies appear to recognize a distinct conformational epitope spanning the GP1–GP2 interface rather than the apex of GP1, implying a neutralization mechanism independent of MR78 and MR191. Other mAbs, such as AF-03, exhibit partial neutralizing activity through similar interference with the GP–NPC1 interaction21. Furthermore, some previously reported anti-EBOV-neutralizing mAbs, such as KZ52 and 6D6, which target the IFL, primarily function during the post-internalization process of viral entry, particularly membrane fusion10. Conversely, our fluorescence-based imaging of VLPs suggests that AGP-R mAbs primarily exert their neutralizing effects during early stages of viral entry, including initial viral attachment to the host cell surface. This inhibition may be mediated by steric hindrance, preventing interactions between GP and glycan-dependent attachment factors—such as C-type lectins—as well as phosphatidylserine-binding receptors such as TIM-129. Given that R1-group mAbs likely recognize epitopes spanning both RBD and IFL, the possibility that they also interfere with membrane fusion cannot be excluded, although our VLP assay did not visualize this process. Further functional and structural studies are necessary to elucidate the precise mechanism by which R1-group mAbs inhibit viral entry.
Furthermore, we showed that AGP-R7 and AGP-R12 conferred 100% therapeutic protection in mouse models of MARV and RAVV infection. Notably, the majority of AGP-R mAbs belong to the IgG2a subclass, suggesting that Fc-mediated immune functions, including antibody-dependent cellular cytotoxicity and antibody-dependent cellular phagocytosis, may contribute to their protective effects25,37. Even in cases where neutralizing activity is limited (e.g., R4- and R11-group mAbs), such Fc-mediated mechanisms could play a critical role in mediating protection in vivo25,37. Interestingly, AGP-R7, which exhibited the highest antiviral activity both in vitro and in vivo, is an IgG3 antibody that is generally believed to have low binding affinity for Fc receptors37,38. This observation suggests that the protective effect of AGP-R7 is likely mediated by its neutralizing activity, with a limited contribution from Fc-mediated functions. The therapeutic efficacy of MR191 and MR78 has also been shown in guinea pig models of MARV and RAVV infection12. Importantly, MR191 conferred 80% and 100% protection from lethal MARV and RAVV infection, respectively, in rhesus macaques12. In 2024, clinical trials of MBP091—a plant-produced version of MR191—were initiated in Rwanda during an outbreak of MVD, showing promising antiviral potential39. However, due to the limited number of clinical cases and the potential for viral escape, the development of new therapeutic mAbs remains necessary39. AGP-R7 and AGP-R12 target nonoverlapping epitopes compared to MR191, positioning them as promising candidates for inclusion in therapeutic antibody cocktails that complement existing antibodies. As shown by REGN-EB3, an FDA-approved antibody therapy for Ebola virus disease, cocktails comprising multiple mAbs targeting nonredundant GP epitopes can enhance therapeutic efficacy and minimize the risk of viral escape40. REGN-EB3 comprises three mAbs: REGN3479, a potent neutralizing antibody that targets the IFL, and REGN3470/REGN3471, which contribute to partial neutralization and promote Fc-mediated clearance of infected cells. Based on this REGN-EB3 model, a combination therapy including MR191, the broadly cross-reactive AGP-R7, and an Fc-active mAb such as MR228 or MR23525 may offer more comprehensive protection against MARV and RAVV variants. Evaluating such combinations in nonhuman primate models and clinical trials will be essential for establishing their clinical potential as frontline therapeutics for MVD.
In conclusion, our study identified a novel group of antibodies against MARV and RAVV that target conserved functional regions on GP. Among these, AGP-R7 and AGP-R12 emerged as strong candidates, exhibiting potent in vitro neutralization and complete protection in lethal animal models. These findings lay a solid foundation for advancing these antibodies toward clinical application of these mAbs and offer a strategic framework for designing broadly protective immunotherapies and vaccines against MVD.
Methods
Cells
African green monkey kidney Vero E6 cells and human embryonic kidney HEK293T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Corning) supplemented with 10% fetal bovine serum (FBS; Gibco), 100 U/mL penicillin, and 0.1 mg/mL streptomycin (Gibco). Mouse P3U1 myeloma cells were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco) supplemented with 10% FBS, 4 mM L-glutamine (Gibco), 100 U/mL penicillin, and 0.1 mg/mL streptomycin. These cell lines were cultured at 37 °C in a humidified atmosphere with 5% CO2. Expi293F cells were maintained in Expi293F Expression Medium (Thermo Fisher Scientific) at 37 °C with 8% CO2, according to the manufacturer’s instructions.
Viruses
Replication-incompetent pseudotyped VSIV encoding the green fluorescent protein (GFP) gene in place of the VSV-G gene and bearing GPs from MARV variants Angola and Musoke, RAVV, or EBOV variant Mayinga (designated VSVΔG*MARV-Angola, VSVΔG*MARV-Musoke, VSVΔG*RAVV, and VSVΔG*EBOV, respectively) were generated as previously described7,16,17. Briefly, HEK293T cells were transfected with pCAGGS expression plasmids encoding the GP genes. After 24 h, the cells were infected with VSVΔG*-VSV-G at a multiplicity of infection of 1.0. Supernatants were collected after 16 h of incubation and centrifuged to remove cell debris. Infectious units (IU) of the pseudotyped viruses were determined as previously described7,16,17. Briefly, serial 10-fold dilutions of pseudotyped viruses were inoculated onto confluent Vero E6 cell monolayers in 96-well plates. After 24 h, GFP-positive cells were counted using an In Cell Analyzer 2500 system (GE Healthcare). To minimize background infectivity from residual parental VSVΔG*-VSV-G, all pseudotyped virus stocks were treated with a neutralizing mAb specific to the G protein (VSV-G[N]1-9)41.
Replication-competent rVSV/MARV-Angola was generated as previously described13,14. The virus was propagated in Vero E6 cells and stored at −80 °C until further use. Virus titers were determined using a plaque assay14. Briefly, 10-fold serial dilutions of virus stocks were inoculated onto confluent Vero E6 cells in 12-well plates and incubated for 1 h at 37 °C with 5% CO2. Following removal of the inoculum, cells were washed with DMEM and overlaid with Eagle’s minimum essential medium (EMEM) supplemented with 0.8% Bacto Agar (Becton Dickinson), 0.3% bovine serum albumin, 100 U/mL penicillin, and 0.1 mg/mL streptomycin. After 48 h of incubation, cells were fixed with 10% formalin and stained with crystal violet. Viral titers were expressed as plaque-forming units (PFU/mL).
Infectious MARV Angola variant (Angola 200501379)42 and RAVV (Kenya 1987)43 were initially obtained from the Special Pathogens Branch, U.S. Centers for Disease Control (CDC), and maintained at the World Reference Center for Emerging Viruses and Arboviruses at the University of Texas Medical Branch (UTMB). Both viruses were propagated in Vero E6 cells and stored at −80 °C until use. All experiments involving infectious MARV and RAVV were conducted in Biosafety Level 4 (BSL-4) facilities at the Galveston National Laboratory (GNL), UTMB, following institutional guidelines. Virus titers were determined by plaque assay as previously described, with slight modifications10. Confluent Vero E6 cells in 12-well plates were inoculated with 10-fold serial dilutions of virus stocks and incubated for 30 min at 37 °C with 5% CO2. Following removal of the inoculum, cells were washed with DMEM and overlaid with EMEM supplemented with 5% FBS, 1% penicillin–streptomycin, and 0.6% tragacanth (Sigma). After 9 days of incubation, cells were fixed with 10% formalin and stained with 0.25% crystal violet. Viral titers were expressed as PFU.
Expression and purification of His-tagged MARV, RAVV, and EBOV GPs
Expi293F cells were transfected with expression plasmids (pCAGGS) encoding histidine-tagged soluble forms of the GPs from MARV (Angola), RAVV, and EBOV15, using the Expi293 Expression System (Thermo Fisher Scientific) according to the manufacturer’s instructions. After 96 h post-transfection, the culture supernatant was harvested and centrifuged at 3000 rpm for 15 min at 4 °C to remove cellular debris. Recombinant His-MARV, -RAVV, and -EBOV GPs were purified from the supernatant using the Ni-nitrilotriacetic acid affinity purification system (Invitrogen) following the manufacturer’s instructions. The purified His-MARV GP was used for booster immunization in mice and as a coating antigen in ELISA. Purified His-RAVV and His-EBOV GPs were also used as antigens in ELISA.
Generation of anti-MARV neutralizing antibodies
All animal experiments for hybridoma generation were conducted strictly in accordance with the Guidelines for Proper Conduct of Animal Experiments of the Science Council of Japan. The study protocol was approved by the Hokkaido University Animal Care and Use Committee (Approval No. 18-0026). Six-week-old female BALB/c mice were obtained from Sankyo Lab Service and housed in animal biosafety level 2 (BSL-2) facilities at the International Institute for Zoonosis Control, Hokkaido University. Animals were maintained for a minimum of one week to allow acclimation. Isoflurane was used as the anesthetic. Mice were intraperitoneally (i.p.) inoculated with 106 PFU of rVSV/MARV-Angola. At 4 weeks post-immunization, they received an intravenous booster injection of 100 µg of purified His-MARV GP. Three days after the booster, mice were euthanized, and splenocytes were harvested and fused with P3U1 myeloma cells using a standard hybridoma generation protocol7,44. The resulting hybridomas were maintained and screened for secretion of neutralizing mAbs using VSVΔG*MARV-Angola. The hybridoma clones producing neutralizing mAbs were then subjected to two rounds of cloning by limiting cell dilution to ensure monoclonality. The resulting neutralizing mAbs were purified from mouse ascites. The isotypes and subclasses of the mAbs were determined using a Mouse Isotyping Kit (Biorad). Previously characterized mouse mAbs MR78 and MR191 were obtained from Absolute Antibody (Drydock Avenue, Boston, MA, USA) and ProteoGenix (Schiltigheim, France), respectively, and were used in neutralization assays and ELISA-based binding studies.
5’RACE for determining variable regions of mAbs
Total RNA was extracted from hybridoma cells producing mAbs using TRIzol™ Reagent (Thermo Fisher Scientific) and reverse-transcribed with the SMARTer RACE 5’/3’ Kit (Clontech) using reverse transcription primers45. Subsequently, the variable gene segments encoding the heavy chain (VH) and light chain (VL) were amplified by polymerase chain reaction (PCR) using VH- and VL-specific primer sets45 and the Universal Primer A Mix (Clontech), in combination with KOD One DNA polymerase (TOYOBO), following the manufacturer’s instructions. The resulting PCR products for the VH and VL genes were cloned into linearized pRACE vectors (Clontech) using the In-Fusion HD Cloning Kit (Clontech) and sequenced. Nucleotide sequences were determined using Sanger sequencing (GENEWIZ, South Plainfield, NJ, USA). The framework regions (FRs) and complementarity-determining regions (CDRs) were determined by the IMGT information system (https://www.imgt.org/IMGT_vquest/input)46,47.
Neutralization tests
VSVΔG*MARV-Angola, VSVΔG*MARV-Musoke, VSVΔG*RAVV, and VSVΔG*EBOV were appropriately diluted to yield 500–3000 IU/well and incubated with serial dilutions of mAbs for 1 h at room temperature. The mixtures were then inoculated into confluent Vero E6 cells, which were seeded in 96-well plates. At 20 h post-inoculation, GFP-positive cells were counted. Relative percentage of infectivity was calculated by setting the number of infected cells in the absence of mAbs as 100%. Neutralizing activity against authentic MARV (Angola variant) and RAVV was determined using a plaque reduction neutralizing test (PRNT). Viruses were diluted in DMEM supplemented with 2% FBS to obtain 100–150 PFU/well and incubated with serially diluted mAbs for 30 min at 37 °C. The virus–antibody mixtures were then inoculated onto monolayers of Vero E6 cells in six-well plates and incubated for an additional 30 min at 37 °C in a CO2 incubator. The inoculum was then removed, and cells were overlaid with EMEM supplemented with 5% FBS, 1% penicillin–streptomycin, and 0.6% tragacanth. After 7 days of incubation, cells were fixed with 10% formalin, and plaques were visualized by staining with crystal violet. The relative percentage of infectivity was calculated by setting the number of plaques in the absence of mAbs as 100%.
Enzyme-linked immunosorbent assay
A filovirus GP-based ELISA was conducted as previously described7,15. Briefly, ELISA plates (Thermo Fisher Scientific) were coated with purified His-tagged MARV, RAVV, or EBOV GPs (50 ng/50 µL/well) diluted in phosphate-buffered saline (PBS) and incubated overnight at 4 °C. The plates were then washed with PBS supplemented with 0.05% Tween 20 (PBST) and blocked with 3% skim milk (180 μL/well) for 1 h at room temperature. After washing with PBST, serial dilutions of purified mAbs prepared in PBST supplemented with 1% skim milk were added and incubated overnight at 4 °C. After three washes with PBST, bound antibodies were visualized using horseradish peroxidase (HRP)-conjugated goat antimouse IgG (H + L) (Invitrogen) and 3,3’,5,5’-tetramethylbenzidine (TMB) solution (Sigma). The enzymatic reaction was stopped by adding 1 N phosphoric acid, and the optical density was measured at 450 nm using a Biotek ELx808 microplate reader (Agilent).
Competitive ELISA
Each mAb was purified and labeled with HRP using the Peroxidase Labeling Kit-SH (Dojindo Molecular Technologies, Inc.) according to the manufacturer’s instructions. Briefly, 100 µg of each purified antibody (1.0 mg/mL) was subjected to the labeling reaction, followed by washing with the provided spin column, and the antibodies were recovered in 200 µL of the provided Storage Buffer. ELISA plates were coated with purified His-tagged MARV GP (200 ng/50 µL/well) in PBS for 2 h at room temperature and subsequently blocked with 3% skim milk for 1.5 h at room temperature. After washing with PBST, serially diluted unlabeled competing mAbs in PBST supplemented with 1% skim milk were added and incubated for 1 h at room temperature. Following additional washes with PBST, HRP-conjugated mAbs, for which the optimal dilutions had been determined prior to use, were added and incubated at room temperature for 1 h. Bound HRP-conjugated mAbs were visualized using TMB solution (Sigma). The reaction was stopped by adding 1 N phosphoric acid, and absorbance at 450 nm was measured using a SPECTRAMAX 190 microplate reader (Molecular Devices).
Selection of rVSV/MARV-Angola escape mutants and identification of putative epitopes
The selection of escape mutants and identification of putative epitopes on the MARV GP molecule were performed as previously described7. Briefly, 10-fold serial dilutions (104–106 PFU) of rVSV/MARV-Angola were incubated with 10 μg/mL of each mAb for 1 h at room temperature and then inoculated onto confluent Vero E6 cell monolayers grown in six-well plates. After 1 h of adsorption, the inoculum was removed, and cells were overlaid with EMEM supplemented with 0.8% Bacto Agar, 0.3% bovine serum albumin, 100 U/mL penicillin, 0.1 mg/mL streptomycin, and 10 μg/mL of the respective mAb. The cultures were then incubated at 37 °C for 2 days. Mutant viruses that grew in the presence of mAbs were isolated from individual plaques and propagated in Vero E6 cells. Viral RNAs were extracted from the supernatants and purified using the Viral RNA Mini Kit (QIAGEN), followed by PCR amplification. The GP gene sequences from the parent virus and the escape mutants were determined using the BigDye Terminator v3.1 Cycle Sequencing Kit and analyzed on an Applied Biosystems 3130xl Genetic Analyzer. Deduced amino acid sequences were aligned and compared among the viral isolates. The reference amino acid sequence of MARV-Angola was obtained from GenBank (accession no. KY047763.1). Identified amino acid substitutions were mapped onto a trimeric GP structure modeled using the Schrödinger Suite (version 2022-4; Schrödinger, LLC, New York, NY, USA), based on the crystal structure of RAVV GP11 (Protein Data Bank, PDB ID: 6BP2). We also propagated rVSV/MARV-Angola in the absence of mAbs and confirmed by Sanger sequencing that no consistent non-synonymous substitutions were found in the GP gene.
Imaging of attachment, internalization, and membrane fusion of DiI-labeled VLPs in live cells
The purification and fluorescent labeling of MARV-Angola VLPs were performed as previously described7. Vero E6 cells were cultured in 35 mm glass-bottom culture dishes (MatTek Corporation). MARV VLPs labeled with 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanine perchlorate (DiI; Thermo Fisher Scientific) were incubated with 100 µg/mL of AGP-R1, -R7, -R12, -R4, -R11, -MR78, or a control IgG (S139/1) for 1 h at 37 °C. Cells were then washed with 1.0 mL of phenol red-free DMEM (Invitrogen) and incubated with either mAb-treated or untreated VLPs in the same medium at room temperature for 30 min. The cells were washed with the same medium to remove unbound VLPs and incubated with 2 ml of phenol red-free DMEM supplemented with 2% FCS and 4% bovine serum albumin at 37 °C for 0 and 5 h to assess viral attachment, internalization, and membrane fusion, respectively. In this assay, Dil fluorescence signal increases when the VLP envelope fuses with the endosomal membrane. To quantify Dil-labeled VLPs, cells were fixed with 4% paraformaldehyde for 10 min at room temperature. Nuclei were stained with 1 μg/mL of Hoechst 42232 (Cell Signaling Technology, Danvers, MA, USA) for 10 min at room temperature. Imaging was performed using a Fluoview FV3000 confocal laser scanning microscope (Evident Scientific, Tokyo, Japan) equipped with a ×60 oil immersion objective lens (×60 magnification) and controlled using FV31S-SW software (Evident Scientific). To quantify DiI-labeled VLPs, z-stack images comprising 4–20 optical sections were acquired at 0.5–1 μm intervals. DiI signals were quantified in ~100 individual cells (1–20 dots/cell), and the average number of fluorescent dots per cell was calculated under each condition. The number, size, and fluorescence intensity of the DiI-labeled VLPs were analyzed using the ImarisCell module (OXFORD Instruments, Oxfordshire, UK).
Therapeutic effects of mAb AGP-Rs in mouse models of MARV and RAVV infection
Five- to 7-week-old female BALB/c mice were obtained from Charles River and housed in ABSL-2 and ABSL-4 facilities at the GNL, UTMB. All animal procedures were reviewed and approved by the Institutional Animal Care and Use Committee of UTMB (IACUC2306037) and were conducted according to the National Institutes of Health guidelines. Animals were randomly allocated to control and treatment groups. Each mouse was considered an independent experimental unit. Sample sizes were determined according to those commonly used in published mouse models of filovirus infection. Animals were maintained for a minimum of one week to allow acclimation. Body weights were measured using an Ohaus CX1201 portable scale. Isoflurane was used as the anesthetic. Mice were i.p. inoculated with 50 PFU (~50 × 50% lethal dose [LD50]) of mouse-adapted MARV-Angola48 or 1 PFU (LD50 was not determined while significant mortality [two out of four mice succumbed, and the remaining two mice showed clinical signs, including ruffled fur and significant body weight loss] in mice infected with this dose was confirmed in our preliminary experiment) of mouse-adapted RAVV49 in a total volume of 100 µL of PBS. At 1 dpi, mice infected with mouse-adapted MARV Angola were administered 100 µg of mAbs AGP-R1, -R7, -R12, -R17, -R4, -R5, -R11, -R20, MR78, or a control mouse polyclonal IgG (BioXCell) through i.p. injection. Mice infected with mouse-adapted RAVV received 100 µg of mAbs AGP-R1, -R7, -R12, -R17, -R11, MR78, or a control mouse polyclonal IgG under the same conditions. Mice were monitored daily for 21 dpi and were humanely euthanized if they exhibited more than 20% body weight loss or were unable to access their food or water. Clinical scoring was performed using the following criteria: 0 = normal; 1 = one clinical sign of the disease; 2 = two or more clinical signs of the disease; 3 = a score of two plus one additional clinical sign and/or >15% weight loss; 4 = >20% weight loss, labored breathing, partial paralysis, paralysis, or bleeding. Animals that met euthanasia criteria or died naturally were both assigned a final score of 4. Inclusion and exclusion criteria were provided, and no animals or data points were excluded.
Molecular modeling
The crystal structure of RAVV GP in complex with neutralizing antibodies was downloaded from the PDB (PDB ID: 6BP2). The apo-GP structure was obtained by removing the antibody components from the complex. To model the variable fragment (Fv) region of AGP-R7, five structures were generated using AlphaFold2, as implemented via ColabFold v1.5.550. Molecular docking between the apo-RAVV GP and the AGP-R7 Fv structures was performed using the “antibody mode” in the ClusPro 2.0 server, which performs rigid-body docking based on a fast Fourier transform (FFT) correlation search51. Among the resulting models, a structure in which the AGP-R7 mAb was predicted to bind near the putative epitope on RAVV GP was selected for further analysis. Residue contacts were identified using the Protein Contacts module in Molecular Operating Environment software (version 2022; Chemical Computing Group, Montreal, QC, Canada). We also attempted direct complex predictions using AlphaFold2-Multimer and AlphaFold3. However, none of the predicted interfaces simultaneously accounted for all experimentally identified escape mutations. Accordingly, the ClusPro-derived pose, which best matched the experimental constraints, was selected for further analysis.
Phylogenetic analysis and multiple alignment of MARV and RAVV GP amino acid sequences
Amino acid sequences of MARV and RAVV GPs available in GenBank were analyzed. The MARV Equatorial Guinea 2023 sequence was retrieved from https://virological.org/t/first-emergence-of-marburg-virus-in-equatorial-guinea-2023/924. A phylogenetic tree was constructed using the maximum-likelihood method with the Jones–Taylor–Thornton matrix-based model implemented in MEGA11. The following GP amino acid sequences were included in the analysis: MARV-Angola (Angola 20051379, AMZ00481.1), Ghana/2022 (MV-22-114C/Ghana/2022, WKR37932.1), Equatorial Guinea 2023, Musoke (CAA78117.1), Ci67 (ABS17558.1), 07DRC99 (ABE27078.1), 01Uga07 (ACT79229.1), 05DRC99 (ABE27085.1), RAVV (ABE27071.1), Uganda 20070648 (AMZ00488.1), and 02Uga07 (ACT79201.1). Multiple sequence alignment was performed using the MUSCLE algorithm in MEGA11.
Statistical analysis
All data were analyzed using GraphPad Prism software (version 9.1.2). Logistic curves representing neutralizing activity were generated using nonlinear four-parameter regression analysis. The fitting curves of neutralization against MARV and RAVV are presented along with their corresponding R2 values. The IC50 and 95% confidence intervals for neutralization were calculated based on these curves, defined as the antibody concentration required to achieve a 50% reduction in infectivity of pseudotyped virus or plaque counts of MARV or RAVV. For VLP attachment, internalization, and membrane fusion assays, one-way repeated-measures analysis of variance (ANOVA) was performed, followed by Dunnett’s multiple comparison test to evaluate differences between AGP-R mAbs and untreated cells. p-values < 0.05 were considered statistically significant.
Data availability
The data used in the current study are provided within the manuscript and supplementary information files or are available from the corresponding author upon reasonable request.
Code availability
There is no custom code associated with the submission.
References
Marburg Virus Disease (accessed June 2025), https://www.who.int/news-room/fact-sheets/detail/marburg-virus-disease.
Asad, A. et al. Past and current advances in the treatment of Marburg virus disease: a review. Infez. Med. 28, 332–345 (2020).
CDC. History of Marburg Outbreaks. Marburg Virus Disease(CDC, accessed June 2025), https://www.cdc.gov/marburg/outbreaks/index.html.
Srivastava, S. et al. Emergence of Marburg virus: a global perspective on fatal outbreaks and clinical challenges. Front. Microbiol. 14, 1239079 (2023).
Hashiguchi, T. et al. Structural basis for Marburg virus neutralization by a cross-reactive human antibody. Cell 160, 904–912 (2015).
Gao, Y. et al. Effects of therapies for Ebola virus disease: a systematic review and network meta-analysis. Lancet Microbe 3, e683–e692 (2022).
Furuyama, W. et al. Discovery of an antibody for pan-ebolavirus therapy. Sci. Rep. 6, 20514 (2016).
Flyak, A. I. et al. Cross-reactive and potent neutralizing antibody responses in human survivors of natural ebolavirus infection. Cell 164, 392–405 (2016).
King, L. B., Milligan, J. C., West, B. R., Schendel, S. L. & Ollmann Saphire, E. Achieving cross-reactivity with pan-ebolavirus antibodies. Curr. Opin. Virol. 34, 140–148 (2019).
Flyak, A. I. et al. Mechanism of human antibody-mediated neutralization of Marburg virus. Cell 160, 893–903 (2015).
King, L. B. et al. The Marburgvirus-neutralizing human monoclonal antibody MR191 targets a conserved site to block virus receptor binding. Cell Host Microbe 23, 101–109.e4 (2018).
Mire, C. E. et al. Treatment of Marburg and Ravn virus infection in nonhuman primates with a human monoclonal antibody. Sci. Transl. Med. 9, 8711 (2017).
Takada, A. et al. Identification of protective epitopes on Ebola virus glycoprotein at the single amino acid level by using recombinant vesicular stomatitis viruses. J. Virol. 77, 1069–1074 (2003).
Saito, T. et al. A surrogate animal model for screening of Ebola and Marburg glycoprotein-targeting drugs using pseudotyped vesicular stomatitis viruses. Viruses 12, 923 (2020).
Nakayama, E. et al. Enzyme-linked immunosorbent assay for detection of filovirus species-specific antibodies. Clin. Vaccine Immunol. 17, 1723–1728 (2010).
Maruyama, J. et al. Characterization of the envelope glycoprotein of a novel filovirus, lloviu virus. J. Virol. 88, 99–109 (2014).
Takada, A. et al. A system for functional analysis of Ebola virus glycoprotein. Proc. Natl. Acad. Sci. U.S.A. 94, 14764–14769 (1997).
Aman, M. J. et al. Broadly neutralizing binding molecules against Marburgviruses. WO Patent WO2022020327A1 (2022).
Crowe, J. E., Flyak, A. I., Bukreyev, A. & Ilinykh, P. Antibody-mediated neutralization of Marburg virus. CA Patent CA2977499A1 (2016).
Berry, J. & Saphire, E. Marburg monoclonal antibodies. CA Patent No. CA2939572A1 (2015).
Zhang, Y. et al. A novel MARV glycoprotein-specific antibody with broad-spectrum neutralization to filovirus. eLife 12, RP91181 (2024).
Addetia, A. et al. Potent neutralization of the Marburg virus by a vaccine-elicited monoclonal antibody. https://doi.org/10.1038/s41586-025-09868-1 (2025). (In press).
Bozhanova, N. G. et al. Discovery of Marburg virus neutralizing antibodies from virus-naïve human antibody repertoires using large-scale structural predictions. Proc. Natl. Acad. Sci. USA. 117, 31142–31148 (2020).
Kajihara, M. et al. Inhibition of Marburg virus budding by nonneutralizing antibodies to the envelope glycoprotein. J. Virol. 86, 13467–13474 (2012).
Ilinykh, P. A. et al. Non-neutralizing antibodies from a Marburg infection survivor mediate protection by Fc-effector functions and by enhancing efficacy of other antibodies. Cell Host Microbe 27, 976–991.e11 (2020).
Marceau, C. D. et al. Novel neutralizing monoclonal antibodies protect rodents against lethal filovirus challenges. Trials Vaccinol. 3, 89–94 (2014).
Igarashi, M., Hirokawa, T. & Takada, A. Structural and energetic basis for differential binding of Ebola and Marburg virus glycoproteins to a bat-derived Niemann–Pick C1 protein. J. Infect. Dis. 228, S479–S487 (2023).
Liu, N., Girvin, M. E., Brenowitz, M. & Lai, J. R. Conformational and lipid bilayer-perturbing properties of Marburg virus GP2 segments containing the fusion loop and membrane-proximal external region/transmembrane domain. Heliyon 5, e03018 (2019).
Hunt, C. L., Lennemann, N. J. & Maury, W. Filovirus entry: a novelty in the viral fusion world. Viruses 4, 258–275 (2012).
Wec, A. Z. et al. Development of a human antibody cocktail that deploys multiple functions to confer pan-ebolavirus protection. Cell Host Microbe 25, 39–48 (2019).
Zhao, X. et al. Immunization-elicited broadly protective antibody reveals ebolavirus fusion loop as a site of vulnerability. Cell 169, 891–904 (2017).
Janus, B. M. & Ofek, G. Structural basis for broad neutralization of ebolaviruses by an antibody targeting the glycoprotein fusion loop. Nat. Commun 9, 3934 (2018).
West, B. R. et al. Structural basis of pan-ebolavirus neutralization by a human antibody against a conserved, yet cryptic epitope. mBio 9, e01674–18 (2018).
Janus, B. M. et al. Macaque antibodies targeting Marburg virus glycoprotein induced by multivalent immunization. J. Virol. 98, e00155–24 (2024).
Carette, J. E. et al. Ebola virus entry requires the cholesterol transporter Niemann–Pick C1. Nature 477, 340–343 (2011).
Côté, M. et al. Small molecule inhibitors reveal Niemann–Pick C1 is essential for Ebola virus infection. Nature 477, 344–348 (2011).
Bruhns, P. & Jönsson, F. Mouse and human FcR effector functions. Immunol. Rev. 268, 25–51 (2015).
Dekkers, G. et al. Affinity of human IgG subclasses to mouse Fc gamma receptors. MAbs 9, 767–773 (2017).
Absolomon, G. et al. Marburg virus disease in Rwanda: an observational study of the first 10 days of outbreak response, clinical interventions, and outcomes. BMC Med. 23, 292 (2025).
Rayaprolu, V. et al. Structure of the Inmazeb cocktail and resistance to Ebola virus escape. Cell Host Microbe 31, 260–272.e7 (2023).
Nakayama, E. et al. Antibody-dependent enhancement of Marburg virus infection. J. Infect. Dis. 204, S978–S985 (2011).
Towner, J. S. et al. Marburgvirus genomics and association with a large hemorrhagic fever outbreak in Angola. J. Virol. 80, 6497–6516 (2006).
Johnson, E. D. et al. Characterization of a new Marburg virus isolated from a 1987 fatal case in Kenya. Arch. Virol. Suppl. 11, 101–114 (1996).
Shahhosseini, S. et al. Production and characterization of monoclonal antibodies against different epitopes of Ebola virus antigens. J. Virol. Methods 143, 29–37 (2007).
Meyer, L. et al. A simplified workflow for monoclonal antibody sequencing. PLoS ONE 14, e0218717 (2019).
Brochet, X., Lefranc, M.-P. & Giudicelli, V. IMGT/V-QUEST: the highly customized and integrated system for IG and TR standardized V-J and V-D-J sequence analysis. Nucleic Acids Res. 36, W503–W508 (2008).
Giudicelli, V., Brochet, X. & Lefranc, M.-P. IMGT/V-QUEST: IMGT standardized analysis of the immunoglobulin (IG) and T cell receptor (TR) nucleotide sequences. Cold Spring Harb. Protoc. 2011, 695–715 (2011).
Qiu, X. et al. Establishment and characterization of a lethal mouse model for the Angola strain of Marburg virus. J. Virol. 88, 12703–12714 (2014).
Warfield, K. L. et al. Development and characterization of a mouse model for Marburg hemorrhagic fever. J. Virol. 83, 6404–6415 (2009).
Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).
Desta, I. T., Porter, K. A., Xia, B., Kozakov, D. & Vajda, S. Performance and its limits in rigid body protein–protein docking. Structure 28, 1071–1081 (2020).
Acknowledgements
We thank the Animal Resources Center (ARC) at the University of Texas Medical Branch (UTMB) for their support of the animal experiments. This work was supported by the Japan Agency for Medical Research and Development (AMED) under grant numbers JP24fk0108624, JP24jf0126002, and JP24fm0208101; the Cooperative Research Project Program of the National Research Center for the Control and Prevention of Infectious Diseases and Tropical Disease at the Institute of Tropical Medicine (NEKKEN), Nagasaki University; and the Chemo-Sero-Therapeutic Research Institute (grant number 2023a018). Additional support was provided by KAKENHI grants from the Japan Society for the Promotion of Science (21H02734, 24K02283, 23K27064), as well as by the Institute for Human Infections and Immunity at UTMB and the John S. Dunn Foundation. T.S. received funding from the Uehara Memorial Foundation, the McLaughlin Fellowship Fund, and the Overseas Research Fellowship of the Japanese Society for the Promotion of Science.
Author information
Authors and Affiliations
Contributions
T.S., J.M., and A.T. designed the experiments. T.S., H.M., A.N., R.H., R.P., G.S., K.L., A.M., C.C., W.F., J.M., and A.T. conducted the experiments. T.S., H.M., A.N., W.F., and M.I. performed data analysis. T.S. and A.T. drafted the manuscript. All authors contributed to the manuscript review and editing and approved the final version.
Corresponding authors
Ethics declarations
Competing interests
T.S., M.I., J.M., and A.T. are inventors on a patent application filed jointly by the University of Texas Medical Branch and Hokkaido University that covers monoclonal antibodies described in this study (Patent application no. 63/575,085). All other authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Saito, T., Miyamoto, H., Igarashi, M. et al. Discovery of potent cross-neutralizing antibodies against Marburg and Ravn viruses with therapeutic potential. npj Viruses 3, 84 (2025). https://doi.org/10.1038/s44298-025-00168-z
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s44298-025-00168-z
