Computational Models Warn Ebola Antibody Drugs May Lose Power Against Bundibugyo Virus

As a new outbreak of Bundibugyo virus disease unfolds across the Democratic Republic of the Congo and Uganda in 2026, an urgent question has moved from the laboratory bench to the treatment centre: will the monoclonal antibody therapies developed and validated against Ebola virus actually work against this distinct ebolavirus species? A team of researchers from Uganda, the Democratic Republic of the Congo, Belgium, France, the United States and beyond has now provided the first systematic computational assessment of that question, and their findings are a mixture of reassurance and caution. Writing in Nature Communications, the group reports that while one leading antibody cocktail appears to retain its targeting of the Bundibugyo virus glycoprotein, another widely deployed therapeutic may be compromised by mutations at its binding site, underscoring how quickly genomic surveillance must be coupled to therapeutic decision-making during an active outbreak.

The study, led by Kevin Cissy Nabukeera, Gradi Luakanda-Ndelemo and Syrus Semawule, with joint supervision from Daudi Jjingo, Tony Wawina-Bokalanga, Placide Mbala-Kingebeni and Isaac Ssewanyana, drew on 44 Bundibugyo virus genome sequences, including 12 generated from the ongoing 2026 outbreak. The team’s central analytical target was the viral glycoprotein, the trimeric surface spike that mediates attachment and entry into host cells and that serves as the sole target of the current generation of approved and investigational ebolavirus antibody therapeutics. Because these antibodies were raised and clinically evaluated against Ebola virus, the Zaire ebolavirus species, their cross-reactivity with Bundibugyo virus cannot be assumed, and the researchers set out to map precisely where the two viruses’ glycoproteins agree and diverge at the amino acid level.

The analytical strategy combined genomic comparison with structural modelling. By aligning the Bundibugyo virus glycoprotein sequences against the Ebola virus reference and projecting the resulting substitutions onto three-dimensional models of the glycoprotein trimer, the team could identify which residues fall within the epitopes, the antibody-binding footprints, of the major therapeutic candidates. Where substitutions occurred within or adjacent to these footprints, the researchers used structural prediction tools, including resources from Google DeepMind’s AlphaFold suite, to model how the altered residues would affect the geometry, charge complementarity and binding affinity of the antibody-antigen interface. This approach allowed the team to generate predictions of therapeutic efficacy rapidly, without waiting for the slow process of isolating and testing live virus in high-containment laboratories.

The most concerning finding concerns mAb114, marketed as Ebanga, a monoclonal antibody that received regulatory approval for the treatment of Ebola virus disease following its demonstration of survival benefit in the PALM trial conducted during the 2018-2020 outbreak in the Democratic Republic of the Congo. The computational analysis revealed that two mutations in the Bundibugyo virus glycoprotein, E112D and P116A, fall within the mAb114 epitope. According to the structural models, these substitutions do not merely weaken binding; they cause the antibody to engage in off-target binding and substantially reduce its affinity for the viral surface. In practical terms, the models predict that mAb114’s efficacy against Bundibugyo virus may be compromised, a result that, if confirmed experimentally, would have immediate implications for treatment protocols in the affected region.

The picture for the second major therapeutic, the three-antibody cocktail known as Inmazeb, proved more complex and, in some respects, more encouraging. Inmazeb combines atoltivimab, maftivimab and odesivimab, three antibodies that bind distinct sites on the Ebola virus glycoprotein and were also validated in the PALM trial. The structural modelling showed that odesivimab maintains its epitope binding against the Bundibugyo virus glycoprotein, but that atoltivimab and maftivimab individually fail to bind their designated sites on the Bundibugyo virus spike. Crucially, however, when the complete trimeric cocktail was modelled against the Bundibugyo virus glycoprotein, the analysis demonstrated synergistic binding, suggesting that the cooperative action of the three antibodies together may preserve therapeutic activity even where individual components fall short.

This combinatorial effect carries a broader lesson for antiviral therapeutics. Antibody cocktails were originally designed with viral escape in mind: targeting multiple non-overlapping epitopes makes it far harder for a virus to mutate its way around the treatment. The Bundibugyo virus analysis adds a second rationale, showing that cooperation between antibodies can compensate for the loss of individual binding interactions when a therapeutic is deployed against a related but antigenically divergent virus species. The authors emphasize that the complex binding behaviour of the Inmazeb components points to the importance of antibody combinatorial effects for treatment efficacy, a consideration that should inform both the design of future pan-ebolavirus therapeutics and the interpretation of cross-species efficacy predictions.

By contrast, the analysis of the MBP134 cocktail offers the clearest cause for optimism. MBP134, a two-antibody combination developed to provide broad protection across ebolavirus species, was found to have epitopes that remain conserved in the Bundibugyo virus glycoprotein. The computational models predict that MBP134 retains its conserved targeting and holds promise as a broadly protective therapeutic against the 2026 outbreak strain. Because MBP134 was explicitly engineered to bind conserved regions of the glycoprotein across the Orthoebolavirus genus, its resilience in this analysis is consistent with its design intent, and the finding positions it as a leading candidate for experimental deployment should clinical evaluation proceed.

The authors are careful to frame their conclusions as predictions rather than proven clinical facts. Computational structural modelling, however sophisticated, cannot fully capture the dynamics of antibody-antigen interaction in biological systems, and neutralization assays performed with authentic Bundibugyo virus isolates remain the definitive test of therapeutic efficacy. The team explicitly calls for urgent experimental validation through neutralization assays, work that would require high-containment laboratory capacity and access to outbreak-derived viral stocks. The gap between computational prediction and wet-laboratory confirmation is a recurring challenge in outbreak response, and this study illustrates both the value and the limitations of moving fast with in silico evidence while definitive experiments are organized.

The collaborative structure of the research reflects the realities of modern outbreak science. The analysis brought together the African Center of Excellence in Bioinformatics and Data-Intensive Sciences at Makerere University, the Institut National de Recherche Biomédicale in Kinshasa, Uganda’s National Health Laboratory and Diagnostic Services, the Uganda Virus Research Institute, and partners including the Institute of Tropical Medicine in Antwerp, Université de Montpellier, the Pandemic Sciences Institute at Oxford, and several public health ministries and agencies across both affected countries. The work was supported in part by Google DeepMind’s AlphaFold tools and resources, a Google.org AIMaR grant, the US National Institutes of Health, the Gates Foundation and the Lacuna Fund, alongside the She Data Science programme supporting Ugandan women in health data science. This blend of African-led genomic capacity, international structural biology expertise and philanthropic and governmental funding represents a template for how future cross-border outbreak analyses can be assembled quickly.

As the 2026 Bundibugyo virus disease outbreak continues, the study’s practical message is twofold. First, treatment decisions in the field should not assume that Ebola virus therapeutics transfer automatically to Bundibugyo virus; the E112D and P116A substitutions that undermine mAb114 binding demonstrate that even single-residue changes at an epitope can reshape the therapeutic landscape. Second, genomic surveillance must be treated as a live input to therapeutic strategy, with continuous monitoring of glycoprotein sequences from new cases feeding into updated structural models and, ultimately, into neutralization testing. The rapid genomic and structural assessment published this week provides exactly that kind of early insight, and its predictions, once validated experimentally, could help determine which antibody therapies offer the best chance of saving lives in the current outbreak and in the Bundibugyo virus eruptions that will inevitably follow.

Subject of Research: Predicted efficacy of Ebola monoclonal antibody therapeutics against Bundibugyo virus in the 2026 outbreak

Article Title: Early insights into predicted efficacy of Ebola monoclonal antibodies for the 2026 Bundibugyo virus disease outbreak

Article References: Nabukeera, K. C., Luakanda-Ndelemo, G., Semawule, S., Ayitewala, A., Adroba-Tandele, P., Walakira, A., Amuri-Aziza, A., Jansen, D., Galiwango, R., Kinganda-Lusamaki, E., Tebba, A., Ngandu, C., Ssemaganda, A., Tshiani-Mbaya, O., Kanyerezi, S., Kyokushaba, J., Akil-Bandali, P., Makoha, C., Oundo, H. R., … Ssewanyana, I. (2026). Early insights into predicted efficacy of Ebola monoclonal antibodies for the 2026 Bundibugyo virus disease outbreak. Nature Communications. https://doi.org/10.1038/s41467-026-78077-9

Image Credits: AI Generated

DOI: 10.1038/s41467-026-78077-9

Keywords: Bundibugyo virus, Ebola, monoclonal antibodies, glycoprotein, MBP134, Inmazeb, mAb114, Ebanga, structural modeling, genomic surveillance, outbreak response, neutralization assays