Recombinant Bombali ebolavirus in cynomolgus macaques as a survival model of Ebola virus disease Authors: Jacquelyn Turcinovic, Karla A Fenton, Krystle N Agans, Viktoriya Borisevich, Daniel J Deer, Joan B Geisbert, Rachel O’Toole, Dafna Abelson, Natalie S Dobias, Abhishek N Prasad, Courtney Woolsey, John H Connor, César G Albariño, Thomas W Geisbert, Robert W Cross Published: 2026 ABSTRACT AbstractEbolaviruses are known to cause severe hemorrhagic fever in humans. A new ebolavirus species, Bombali orthoebolavirus (BOMV), was recently detected in bats from Sierra Leone, Kenya, Guinea, Mozambique, and Tanzania. No infectious BOMV has been isolated from field samples, but an infectious recombinant BOMV clone (rBOMV) was recently established. To estimate disease potential in humans, we challenged cynomolgus macaques with rBOMV. Macaques exhibit hallmark features of Ebola virus disease, including vascular leak, lymphopenia, neutrophilia, and elevated inflammation markers. Consistent with these hallmarks, we detected rBOMV in multiple target tissues at peak viremia, including the liver, lymph nodes, and adrenal gland. All animals survive challenge and develop robust adaptive immune responses, including strong BOMV-specific IgG and neutralizing antibody titers. These data suggest that BOMV may have an attenuated disease phenotype compared to more pathogenic ebolaviruses such as Ebola, Bundibugyo, Sudan, or Taï Forest and may serve as an important model of resolving Ebola Virus Disease.Subject terms: Ebola virus, Viral pathogenesis, Model vertebrates Introduction Ebolaviruses are negative-sense, nonsegmented RNA viruses. Four species cause severe disease in humans and nonhuman primates (NHPs): Ebola virus (EBOV), Sudan virus (SUDV), Taï Forest virus (TAFV), and Bundibugyo virus (BDBV)1. A fifth species, Reston virus (RESTV), causes a lethal disease in macaques, but no human illness has been documented to date (reviewed in refs. 1,2). EBOV, SUDV, and BDBV have caused repeated outbreaks of severe hemorrhagic fever in humans with high mortality, which have become larger and more frequent since EBOV and SUDV were first identified in 19761,3. Most human ebolavirus outbreaks occur in central Africa, particularly in the Democratic Republic of Congo, with two notable exceptions in western Africa: the largest ebolavirus outbreak originating in Guinea4 and a single documented human TAFV infection that occurred in Côte d’Ivoire in 19945,6. Bombali virus (BOMV) is the third ebolavirus discovered in western Africa7. The viral genome was first identified through metagenomic analysis of NGS samples taken from insectivorous bats roosting in human dwellings in Bombali District, Sierra Leone. Subsequent studies detected BOMV RNA in bats from Kenya, Guinea, Mozambique, and Tanzania8–11. The two bat species that tested positive for BOMV (Chaerephon pumilus and Mops condylurus) commonly roost in buildings with habitats ranging from western to central and southern Africa12,13, suggesting human contact with BOMV-infected bats is likely. No symptomatic human BOMV infection has been documented, nor have BOMV-specific serosurveillance studies been reported. BOMV’s disease incidence and outbreak potential in humans is unknown, though studies using a replication-competent vesicular stomatitis virus expressing the BOMV glycoprotein (rVSV-ΔG-GPBOMV) demonstrated that BOMV GP could facilitate entry into human cells7 using the ebolavirus receptor NPC114,15. Attempts to recover infectious BOMV from ecological samples have been unsuccessful, likely due to challenges in maintaining proper cold chain in field environments or low infectious titers present in sampled tissues. Nonetheless, generating infectious filovirus clones from genome sequences using reverse genetics techniques has afforded new opportunities to study viruses that are challenging to recover16. Recently, two groups generated recombinant infectious clones of Bombali ebolavirus (rBOMV)17,18. Initial in vitro characterization showed that rBOMV replicates more slowly than EBOV in primate kidney and human lung cell lines18. Infection of interferon-alpha/beta receptor knockout (IFNAR−/−) and humanized mice has shown limited to no signs of disease18,19. Unfortunately, mouse models of ebolavirus infection have limited utility in modeling Ebola virus disease (EVD) since they do not develop the most salient features of EVD, including coagulopathy, vascular leak, and lymphopenia seen in human patients20,21. To date, nonhuman primates (NHPs) are considered the gold standard animal species for modeling ebolavirus infection that fully recapitulates human EVD20,21. Lethal models of EBOV, SUDV, and BDBV have been established in cynomolgus and rhesus macaques, and their disease course has been extensively characterized22–30. Hallmark features of lethal EVD in NHPs and humans involve fever, a hyperinflammatory innate immune response, coagulopathy with skin rashes and internal organ hemorrhage, lymphopenia, damage and necrosis of viral replication target organs, and failure to develop an adaptive immune response. Late-stage disease is characterized by neutrophilia and multi-system organ failure. BOMV infection in NHPs has not been described, and whether BOMV requires adaptation to cause disease in primates is unknown. Here, we present virological, clinical, immunological, and pathological findings describing Ebola-like disease observed in cynomolgus macaques following intramuscular (i.m.) challenge with rBOMV using a dose known to be more than 100 times the EBOV LD50 in this species31–33. All macaques survive rBOMV challenge despite displaying a combination of EVD hallmarks, including petechial rash, viremia, lymphopenia, thrombocytopenia, neutrophilia, and infection of tissue-resident mononuclear cells. rBOMV-challenged macaques mirror known human EVD survivor correlates: they develop lower peak viremia and lower peak clinical chemistry markers of organ damage (e.g., AST and BUN) than fatally infected EBOV-challenged macaques, and their development of a robust adaptive immune response correlates with viral clearance. Finally, genomic analysis of circulating rBOMV returns no evidence of mutation or adaptation in vivo, indicating a low genetic barrier to primate spillover. Results BOMV challenge causes nonlethal ebolavirus-like disease in cynomolgus macaques Four adult cynomolgus macaques (NHPs A-D) were challenged i.m. with 1000 plaque-forming units (PFU) of recombinantly recovered BOMV (rBOMV). All survived to the predetermined study endpoint of 28 days postinfection (DPI) (Fig. 1a). Historical controls challenged with EBOV Kikwit (n = 44 NHPs) succumbed to disease by 9 DPI. Three of four rBOMV-challenged NHPs exhibited clinical signs consistent with EVD starting at 8 DPI (Fig. 1b and Supplementary Table 1), including fever, decreased appetite and anorexia, diarrhea, mild depression, hunched posture, and petechial rash. We then challenged four additional cynomolgus macaques with rBOMV using the same route and dose, but terminated the study at 7 DPI to better understand pathogenesis. These four cynomolgus macaques (NHPs E-H) did not develop clinical signs before the scheduled 7 DPI euthanasia (Supplementary Table 1). Infectious rBOMV (8/8 NHPs, peak titer 3.16e5 PFU/mL) and rBOMV RNA (2/8 NHPs, peak titer 1.37e9 GEq/mL) were detected in circulating blood by 5 and 7 DPI, respectively. This peak titer was significantly lower than in historical EBOV Kikwit-challenged controls (n = 12 NHPs, peak titers 1.10e14 GEq/mL and 2.75e7 PFU/mL) (Fig. 1c, d). NHPs cleared rBOMV viremia by 15 DPI. Tissues from two of the four NHPs euthanized at 7 DPI contained detectable replicating virus (Supplementary Table 4): spleen (2/4 NHPs, 500 PFU/g), tonsil (2/4 NHPs, 1250–1750 PFU/g), mesenteric lymph node (2/4 NHPs, 500–2750 PFU/g), uterus (1/4 NHPs, 250 PFU/g), and inguinal lymph node (1/4 NHPs, 500 PFU/g). Viral RNA not detectable in tissues via qRT-PCR. Survival, clinical signs, and circulating viral load in BOMV-challenged cynomolgus macaques compared to historical EBOV Kikwit-challenged controls. a Kaplan–Meier survival curves of BOMV natural history cynomolgus macaques (n = 4 NHPs; NHPs A-D) compared to historical EBOV Kikwit controls (n = 44 NHPs). b Presence of EVD-associated clinical signs in each BOMV-challenged NHP (black = present). Viral load as detected in circulating blood by qRT-PCR (c) and plaque assay (d) in BOMV-challenged NHPs (n = 8) compared to EBOV Kikwit controls (n = 12 NHPs). Viral loads at 3, 5, and 7 DPI were compared via two-sided Mann–Whitney U tests. **: p < 0.01. *: p < 0.05. p-values for c: 3 DPI = 0.003, 5 DPI = 0.015, and 7 DPI = 0.033. p-values for d: 3 DPI = 0.007, 5 DPI = 0.016, and 7 DPI = 0.010. Mean and standard deviation plotted for EBOV-challenged NHPs (red). Source data are provided as a Source data file. Measurements of hematology and clinical chemistry values in the rBOMV-challenged macaques (n = 8 NHPs) were made and compared to EBOV Kikwit-challenged historical controls (n = 8 NHPs) (Fig. 2 and Supplementary Figs. 2, 3). rBOMV-infected NHPs experienced thrombocytopenia and lymphopenia to a similar degree to EBOV-infected controls, with minimum lymphocyte concentrations of 0.84e9 ± 2.30e9 cells/L and 1.12e9 ± 1.19e9 cells/L, respectively; and minimum platelet concentrations of 1.13e11 ± 1.01 cells/L and 0.44e11 ± 1.23e11 cells/L, respectively (median and standard deviation reported for each; lymphocyte p = 0.19, platelet p = 0.13; two-sided Mann–Whitney U test with FDR-adjusted p-values). Despite similar depletion levels, the onset of thrombocytopenia and lymphopenia was delayed in rBOMV-challenged NHPs (~7 DPI) relative to historical EBOV-challenged controls (~3 and ~5 DPI, respectively) (Fig. 2a, b). Platelet levels in rBOMV-infected NHPs recovered to greater than or equal to baseline by 15 DPI. Likewise, C-reactive protein (Fig. 2e), a general inflammation marker, spiked at 7 and 10 DPI (4/8 and 3/4 NHPs, respectively) to levels comparable with EBOV-challenged NHPs at 5–8 DPI (rBOMV: 102–148 mg/L; EBOV: 113–185 mg/L), which returned to baseline concentrations by 15 DPI. Selected hematology and clinical chemistry markers of BOMV-challenged NHPs (n = 8) compared to EBOV Kikwit-challenged control NHPs (n = 8). BOMV-challenged NHPs developed thrombocytopenia (a) and lymphopenia (b) comparable to EBOV control NHPs. Two of four natural history BOMV NHPs experienced elevated AST at 10 DPI (c) but not at the same level as EBOV NHPs. BOMV NHPs did not develop elevated BUN (d). All BOMV-challenged NHPs experienced elevated CRP (e). Other measured hematology and clinical chemistry analytes can be found in Supplementary Figs. 2a–p and 3a–m. Mean and standard deviation plotted for EBOV-challenged NHPs (red). Source data are provided as a Source data file. rBOMV NHPs differed from EBOV controls in classical liver and kidney dysfunction markers. For example, elevated circulating aspartate aminotransferase (AST), often used to indicate liver dysfunction or damage, was elevated in only two of four BOMV NHPs (Fig. 2c). In these two NHPs, AST peaked later (10 DPI vs. 6–8 DPI) and to lower levels (~300–500 U/L vs. ~800–7000 U/L) than seen in EBOV-infected NHPs before returning to baseline levels. Circulating blood urea nitrogen (BUN), a marker of kidney dysfunction, did not deviate from baseline levels in BOMV NHPs through the study duration (Fig. 2d); in contrast, BUN spiked to nearly double its baseline concentration (from 21–30 mg/dL to 53–128 mg/dL) by 6 DPI in EBOV controls and remained high until the animals succumbed to infection. These findings suggest BOMV-challenged animals did not experience liver or kidney damage comparable to EBOV-challenged animals. BOMV infection activates host response pathways involved in EVD To address whether BOMV infection induced canonical EVD host response pathways, we profiled serum proteins involved in the coagulation cascade and inflammation (natural history only, n = 4 NHPs, NHPs A-D; Supplementary Figs. 4 and 5). D-dimer spiked to ≥3x pre-infection baseline levels (log2 fold change 1.5–5) in all BOMV NHPs at 7–10 DPI before decreasing to less-than-baseline levels by 15 DPI (Fig. 3a). Fibrinolysis regulators tPA and PAI-1 also increased from baseline at 7–10 DPI (Fig. 3b, c). Pro- and anti-inflammation proteins IFN gamma, IL-1 receptor agonist, and IL-6 similarly increased at 7–10 DPI (Fig. 3d–f) to ≥8x baseline (log2 fold change ≥3). The change in these proteins’ circulating concentrations during the study shows BOMV-induced but controlled inflammation and coagulation networks in the challenged macaques. Selected coagulation and inflammation markers detected in BOMV-challenged NHP circulating blood. All log2 fold changes were calculated relative to pre-infection baseline (0 DPI). Coagulation markers D-dimer (a), tPA (b), and PAI-1 (c) peaked at 7–10 DPI. NHP A had substantially increased D-dimer levels relative to other NHPs; NHP C experienced higher increases in tPA and PAI-1 than the others. d, e Inflammation markers IFNg (d), IL-1ra (e), and IL-6 (f) peak at 7-DPI. IL-1ra and IL-6 returned to baseline by 21 DPI; IFNg remained elevated through the study endpoint in NHPs A and C. Source data are provided as a Source data file. Transcriptomics reveals a modular host immune response to infection To examine the host response at a systems level, we analyzed bulk RNA-seq transcriptomic data from serially sampled whole blood by comparing each time point postinfection to the pre-infection baseline. Globally, total differentially expressed genes peaked twice at 7 DPI and 15 DPI (Fig. 4a). Naïve K-means clustering on genes’ fold change profiles supported this finding by grouping upregulated genes into three distinct phenotypic transcriptional response modules (Fig. 4b): early, late, and prolonged. Gene enrichment analysis identified functional and signaling pathways characteristic of each module (Fig. 4c) and upstream regulators for each postinfection time point (Fig. 4d). Bulk transcriptomic analyses of circulating blood in BOMV-challenged NHPs. Gene expression at each time point post-infection was compared to baseline (−8 DPI). Significantly differentially expressed (DE) genes had log2 fold change >2 or <−2 and FDR-adjusted p-value < 0.05. a Total DE genes per postinfection timepoint with peaks at 7 and 15 DPI. b K-means clustering defined three distinct upregulated DE gene modules. The gray ribbon shows the median log2 fold change ± standard deviation. The colored lines show log2 fold change for selected representative genes. c Selected enriched functional and canonical signaling pathways from IPA. d Selected enriched upstream regulators from IPA. e Selected cell type abundance scores from CIBERSORTx. Early upregulated genes typically increased starting at 5 DPI, peaking at 7 DPI, and returned to baseline by 10 DPI. This module includes innate immune interferon-stimulated genes (ISGs), including ISG15, IFIT2, and OAS3. Correspondingly, interferon signaling (α, β, and γ) and antiviral response pathways were enriched at 5–7 DPI. Upstream regulator analysis identified interferon α, β, and other ISG regulators (e.g., IRF3 and IRF7) as early as 3 DPI. Late upregulated genes peaked at 15 DPI after BOMV was no longer detectable in circulating blood. Many of these are associated with neutrophils, including MMP8, ELANE, and LCN2. Neutrophil degranulation and activation pathways (Fig. 4c) and upstream differentiation regulator CEBPA (Fig. 4d) were also enriched at 15 DPI. Genes in the prolonged upregulation module increased expression early (3–5 DPI) and remained upregulated through 15 DPI. They include genes associated with the complement cascade (e.g., C1QA and SERPING1), ISGs (e.g., OAS1), cytokines and their regulators (e.g., CXCL10 and SOCS1), and acute phase response markers (e.g., S100A8 and HERC5). Finally, several chemokines involved in recruiting various types of immune cells, including neutrophils and monocytes, had either no change in expression (MCP-1, also known as CCL2) or were downregulated in many postinfection time points (IL-8 and MCP-3, also known as CXCL8 and CCL7, respectively). Digital cell quantitation via CIBERSORTx predicted a spike in monocyte and neutrophil populations at 7 DPI (Fig. 4e). By 15 DPI, monocyte and neutrophil populations returned to baseline, and memory CD4+ T-cell populations had increased. NHPs A and B had transiently increased plasma B-cell scores at 28 DPI (Supplementary Fig. 1f), but no other consistent B-cell involvement was predicted. Adaptive immune response Viral replication and clearance correlated with a BOMV-specific antibody response starting at 7 DPI as measured by IgM and IgG anti-rBOMV GP ELISA, with titers of 400–1600 and 1600–25,600 by the study endpoint for IgM and IgG, respectively (Fig. 5a, b). Plasma from rBOMV-challenged NHPs increasingly neutralized rBOMV over time (Fig. 5c), and PRNT50 increased from below the detected limit at 0 DPI to 40 PRNT50 by the study endpoint (Supplementary Fig. 6a–d). Humoral responses in BOMV-challenged NHPs. Anti-BOMV GP IgM (a) was detectable by 5 DPI in the sera of 2 of 4 NHPs and all 4 NHPs by 7 DPI. Anti-BOMV GP IgG (b) was detectable in sera by 7 DPI in all NHPs. c rBOMV neutralization by rBOMV-challenged NHP sera increased significantly over time. The box shows the first quartile, median, and third quartile; whiskers extend to the largest and smallest values no further than 1.5 × IQR from the first and third quartiles, respectively. Time points were compared via two-sided Mann–Whitney U tests. *: p < 0.05; p = 0.029 for all comparisons. d NHP sera cross-reacted with multiple species of ebolaviruses after challenge and clearance of rBOMV via IgG ELISA. Bar height indicates the mean; error bars indicate standard deviation. Samples were run in technical duplicates for ELISA and neutralization assays, with the mean value of technical duplicates reported. Source data are provided as a Source data file. Finally, we addressed the hypothesis that surviving an rBOMV infection could generate a humoral immune response that cross-reacts with other ebolaviruses. IgG ELISA showed that rBOMV-challenged NHP sera from 28 DPI cross-reacted with other ebolaviruses (Fig. 5d); however, BOMV-neutralizing antibodies did not cross-neutralize EBOV Makona or TAFV in plaque neutralization assays (Supplementary Fig. 6e–l). Only one of four NHPs developed a mild, albeit quantifiable, cross-neutralizing response against EBOV Makona and TAFV. BOMV adaptation to primate infection We analyzed BOMV-aligning RNA-seq reads to address whether BOMV accumulates potentially adaptive mutations during primate infection. Viral genome coverage from three of four BOMV-challenged NHPs at 7 DPI was sufficient for single-nucleotide variant (SNV) analysis (median aligned depth >10). No consensus-level (>50%) mutations were detectable in any NHP. We identified 1–3 sub-consensus, nonsynonymous SNVs in the 7 DPI samples analyzed (Supplementary Table 2). All were found in the RNA-dependent RNA polymerase (L) at <25% frequency. The nonsynonymous substitution L A1517T was found in all three analyzed NHPs. The L frameshift insertion at nucleotide 16,949 (C to CT) was found in two of three analyzed NHPs. Gross pathology findings Examination of the NHPs that survived to 28 DPI (NHPs A-D) identified minimal gastrointestinal mucosal reddening at the duodenal junction in two of four macaques. This finding is a disease feature common in NHP acute-phase ebolavirus infections. No other lesions were found in these animals. Mild hepatic reticulation (3/4 NHPs), splenomegaly (4/4 NHPs), and congestion (4/4 NHPs) were noted in NHPs euthanized at 7 DPI (NHPs E-H; Fig. 6b). No other appreciable lesions typically associated with fulminant EBOV infection were identified (Fig. 6a, b, s)24,28–30,34,35. Gross pathology, histopathology, and immunohistochemistry findings for rBOMV-challenged NHPs relative to an EBOV Kikwit-challenged historical control NHP. Scale bars in the upper right corner denote 10 μm for 40× images and 100 μm for 20× and 10× images. a Liver with no appreciable lesions (NHP A). b Liver with mild reticulation (hepatitis) (NHP H). c Liver, H&E, 40×, no appreciable lesions (NHP A). d Liver, IHC, 40×, no appreciable rBOMV NP (red) (NHP A). e Liver, H&E, 40×, mild granulomatous hepatitis (NHP H). f Liver, IHC, 40×, IHC dual positive for CD68 (brown) and rBOMV NP (red) in hepatic sinusoids (arrow). g Spleen, H&E, 10×, no appreciable lesions (NHP A). h Spleen, IHC, 40×, no appreciable rBOMV NP (red) (NHP A). i Spleen, H&E, 10×, disruption of normal architecture with lymphocyte loss (NHP H). j Spleen, IHC, 40×, with rare mononuclear cells within the red and white pulp that are IHC positive for rBOMV NP (red) (NHP H). k Tonsil, IHC, 20×, no appreciable rBOMV NP (red) (NHP A). l Adrenal gland, IHC, 20×, no appreciable rBOMV NP (red) (NHP A). m Tonsil, IHC, 20x, with rare mononuclear cells within the lymphoid tissues of the tonsil that are IHC positive for rBOMV NP (red) (NHP H). n Adrenal gland, IHC, 20×, with a rare mononuclear cell within the cortex that is IHC positive for rBOMV NP (red) (NHP H) (arrow). o Ovary, IHC, 20×, no appreciable rBOMV NP (red) (NHP A). p Cecum, IHC, 40×, no appreciable rBOMV NP (red) (NHP A). q Ovary interstitium, IHC, 20×, with rare mononuclear cells that are IHC positive for rBOMV NP (brown) (NHP H). r Cecum, IHC, 20×, with rare mononuclear cells within the lymphoid tissues that are IHC positive for rBOMV NP (red) (NHP H). s Liver with rounded edges, and diffuse pallor (hepatitis). t Liver, IHC, 40×, IHC dual positive for CD68 (brown) and EBOV VP40 (red) in hepatic sinusoids. u Spleen, IHC, 40×, IHC dual positive for CD68 (brown) and EBOV VP40 (red) with markedly disrupted white pulp. v Adrenal gland, IHC, 40×, IHC positive for EBOV VP40 (brown) in cortical cells. Histopathology and immunohistochemistry findings Histopathologic evaluation of the four NHPs examined at 28 DPI (NHPs A-D) lacked any significant lesions indicative of EVD, supporting the gross findings (Supplementary Table 3 and Fig. 6c, g). Minimal eosinophilic and lymphohistiocytic inflammation was noted in at least one of the following tissues: trachea, stomach, duodenum, ileum, cecum, colon, and nasal mucosa. Incidental reddening of the gastrointestinal tract at necropsy corroborated these inflammatory findings. No appreciable positive immunohistochemistry for anti-BOMV antigen was noted in any of the examined tissue sections (Fig. 6d, h, k, l, o, p). A similar evaluation of the four NHPs challenged with rBOMV and examined at 7 DPI (NHPs E-H) showed minimal to mild lesions within EVD target organs, including the liver, lymphoid tissues, adrenal gland, gonad, and urinary bladder. Minimal sinusoidal leukocytosis and occasional microgranulomas were noted in the liver (Supplementary Table 3 and Fig. 6e). CD68-positive mononuclear cells within these granulomas and sinusoids were occasionally co-labeled with anti-EBOV antigen in all 7 DPI-examined NHPs (Fig. 6f). The splenic white pulp architecture was mildly disrupted with lymphocyte loss (Fig. 6i). Individual to small clusters of mononuclear cells within the white pulp were IHC-positive for anti-EBOV antigen (Fig. 6j). Similar histiocytic infiltrates and IHC positivity were noted in at least one of the following lymphoid tissues: axillary lymph node, inguinal lymph node, submandibular lymph node, tonsil (Fig. 6m), mesenteric lymph node, and/or the gastrointestinal lymphoid tissues (Fig. 6r). Minimal histiocytic inflammatory infiltrates were noted in the urinary bladder of one animal with associated IHC-positive cells. No other inflammatory lesions were noted histologically; however, IHC-positive mononuclear cells within the adrenal cortex and the interstitium of the ovary were present (Fig. 6n, q). These histologic findings were corroborated by minimal hepatitis and splenomegaly noted at necropsy. No appreciable positive immunohistochemistry for anti-BOMV antigen was present in the remaining examined tissue sections. Discussion The study and analyses presented above revolve around four major questions. First, does the nonlethal infection observed in rBOMV-challenged macaques resemble human infection and other NHP models of EVD? Second, does nonlethal rBOMV infection in macaques parallel the illness seen in human EVD survivors? Third, why do macaques survive rBOMV challenge while other filoviruses are partially or uniformly lethal? And finally, how can this study inform us about barriers to spillover and cross-protection? These themes are addressed sequentially below. rBOMV infection in cynomolgus macaques recapitulated several hallmarks of EVD found in human patients and other partially to uniformly lethal NHP models of EVD. rBOMV-challenged macaques developed classic NHP clinical signs, including decreased appetite, hunched posture, fever, and petechial rash24. Viremia, lymphopenia, neutrophilia, and acute inflammation were present, as well as multiple signs of a consumptive coagulopathy reminiscent of DIC; these are well-established components of human and NHP EVD29,36–40. Tissue-resident mononuclear cells in the liver, spleen, lymph nodes, and adrenal gland contained rBOMV antigen in the macaques euthanized at predicted peak viremia (7 DPI). Such mononuclear cells, particularly Kupffer cells, have been well-established as early targets of filovirus infection, hotspots of viral replication, and are thought to be drivers of the host’s aberrant and uncontrolled inflammatory response1,24,30,40. While all these EVD hallmarks were detected in rBOMV-challenged macaques, they were less consistent and were observed at later times than in NHP models of EBOV or SUDV infection23,24. The most obvious way rBOMV diverges from other filovirus NHP challenge models is the difference in lethality. Studies of human EVD and partially lethal NHP models have identified multiple differential variables in outcome that remain predictive across different outbreaks and filovirus species, including viral load41–44, clinical chemistry markers of liver and kidney function (AST, ALT, BUN, and CRE)36,41,45, and the rapid development of a robust adaptive immune response45–49. Our comparison of rBOMV-challenged NHPs versus EBOV Kikwit-infected NHPs aligns with these predictors. rBOMV-survivor NHPs had lower circulating viral load than fatally-infected EBOV Kikwit NHPs as measured by both viral genome copies and infectious virus; clinical chemistry parameters did not spike to the levels seen in historical controls; and rBOMV-survivor NHPs rapidly developed a quantifiable and specific adaptive immune response as measured by virus neutralization, IgM and IgG ELISAs, and a predicted increase in memory CD4+ T-cells. Finally, no infectious virus, viral genomes, or viral antigen were detected in circulating blood or sampled tissues by 28 DPI, with only minimal residual inflammation, which is suggestive of systemic clearance by an effective and controlled host immune response. These lines of evidence show rBOMV infection recapitulates human EVD survival. Based on these lines of evidence and previous work, we can build a working hypothesis as to why NHPs survive rBOMV infection but do not uniformly survive an equivalent challenge with other filoviruses. Importantly, rBOMV replicates more slowly in primates relative to EBOV. This has been shown previously in vitro18 and is shown here in vivo, where viremia is first detectable at 3 DPI and 5 DPI in rBOMV- and EBOV Kikwit-challenged NHPs, respectively. A delayed viral replication timeline would decrease the time that the host spends in an innate immune response state before transitioning to an adaptive immune-driven state. This would result in lower peak viremia, inflammation, and resultant tissue damage during the innate immune stage— all of which we observed in rBOMV-challenged NHPs—making it more probable that the host would survive long enough to enter the adaptive immune stage, clear the infection, control the inflammation, and survive. Thus, while rBOMV survival is likely due to a combination of factors, we hypothesize that the key principle is time, where slower replication affords opportunity for effective host response to infection. A separate but parallel line of inquiry in this study revolves around BOMV as a potential human pathogen. While BOMV has only been detected in bats, prior work has shown that BOMV can enter and replicate in human cells7,17; however, it remained unknown whether BOMV accumulated adaptive mutations when replicating in primates, whether in vitro or in vivo. Our analysis of circulating rBOMV genomes in vivo did not show evidence of adaptive mutation accumulation or strong selective pressure on the input viral population, which supports the hypothesis that the genetic barrier to BOMV spillover into primates is low to nonexistent. Furthermore, while rBOMV survivor sera reacted against other ebolaviruses via IgG ELISA, rBOMV did not consistently neutralize infectious EBOV Makona or Taï Forest virus, both of which have been found in western Africa, regionally near where BOMV was first detected4,5,7. While preliminary, this evidence supports the notion that human exposure to BOMV would not completely protect against other regional ebolaviruses. Finally, the presence of mild disease in NHPs does not preclude the ability of BOMV to cause severe disease in humans, but it suggests a milder or asymptomatic infection is possible. The conventional NHP model of EVD employs high i.m. challenge doses of ebolaviruses to achieve uniform or near-uniform lethality24,31–33. These lethal models have been essential for testing and characterizing medical countermeasures, including the rVSV-EBOV vaccine (Ervebo) and multiple postexposure therapeutics50–56, and have been accepted across several BSL-4 laboratories. NHP models to capture moderate-to-severe, nonlethal EVD remain an area of active research since even low i.m. challenge doses result in lethal outcomes31–33. Established survivor models include EBOV challenge via intranasal27, oral57,58, or conjunctival57,59 routes; BDBV i.m. challenge22; and TAFV i.m. challenge60–62. These models either skew toward fatal outcomes and cannot reliably produce sufficient survivors for robust statistical comparison without using prohibitively high numbers of NHPs per study, or survivors in these models do not routinely develop signs of disease. Given that hallmark EVD features were observed with the standard 1000 PFU i.m. dose of rBOMV, the possibility of more severe disease manifestations, including lethality resulting from higher challenge dosing, cannot be ruled out. The rBOMV model presented here was uniformly nonlethal but produced mild to severe clinical disease, which makes it a promising candidate for studying host responses to ebolavirus infection associated with survival. While outside the scope of this study, future work with the BOMV model could include investigation of post-EVD sequelae63. While data presented here suggest robust immunity resulting in uniform viral clearance, we cannot rule out the potential for this infection model to afford persistent viral replication in immune-privileged sites64–66. The work presented here supports the hypothesis that the successful host response to rBOMV infection is due, at least in part, to three factors: (1) delayed, reduced, or controlled viral replication in target organs, (2) the lack of or successful control of the host’s inflammatory innate immune response, and (3) development of an effective adaptive immune response before irreversible organ damage occurs. All three of these factors are correlated with EVD recovery in human patients1,24,30,41–44,46–49, supporting the rBOMV challenge of cynomolgus macaques as a viable model of nonlethal EVD. Methods Virus rBOMV was rescued as previously described17 and used for i.m. challenge. Importantly, the original field sample (MF319186.1 [https://www.ncbi.nlm.nih.gov/nuccore/MF319186.1]) was re-sequenced, and four low-coverage nucleotides were corrected (OK633884.1 [https://www.ncbi.nlm.nih.gov/nuccore/OK633884.1]) before rBOMV rescue. Analysis of the rescued rBOMV consensus sequence found no mutations relative to the corrected BOMV input sequence. Nonhuman primate challenge All animal work conducted in the study was ethically approved by the UTMB Institutional Animal Care and Use Committee (IACUC). UTMB facilities used in this work are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International and adhere to principles specified in the eighth edition of the Guide for the Care and Use of Laboratory Animals, National Research Council. In an initial study, four (two male and two female) healthy adult captive-bred cynomolgus macaques (Macaca fascicularis) (Worldwide Primates, Miami, FL) of Asian origin, 5–7 years of age and weighing 2.5–4.3 kg, were challenged in the left quadricep with a 1000 PFU target dose (actual dose 1175 PFU) of rBOMV. Serial blood draws were performed on days −8, 0, 1, 5, 7, 15, 21, and at the predetermined study endpoint (28 days postinfection). In a second study, four healthy adult captive-bred cynomolgus macaques (two male and two female, Worldwide Primates) of Asian origin, 4–6 years of age and weighing 3.5–6.5 kg, were challenged i.m. in the left quadricep with a 1000 PFU target dose (actual dose 1425 PFU) of rBOMV. All four animals in the second study were euthanized 7 days after the rBOMV challenge. All eight macaques in both studies were monitored daily and scored for disease progression following the study protocol. The scoring changes measured from baseline included posture and activity level, attitude and behavior, food intake, respiration, and disease manifestations, such as visible rash, hemorrhage, ecchymosis, or flushed skin. A score of ≥9 indicated that an animal met the criteria for euthanasia. Survival, viral replication, clinical chemistry, and hematology parameters were compared to EBOV Kikwit-challenged historical controls (n = 2 NHPs from67; n = 1 NHP from68; n = 41 NHPs unpublished; n = 44 total) to quantify similarities and differences between BOMV infection and a known uniformly lethal filovirus infection. The historical control animals were challenged with EBOV Kikwit comparably to the rBOMV-challenged animals (1000 PFU, i.m.) and represent similar age and sex demographics. Subsets of these controls have been used where appropriate to maintain experimental and statistical integrity; numbers are stated in the corresponding figure captions and methods subsections. Hematology and serum biochemistry Total white blood cell counts, white blood cell differentials, red blood cell counts, platelet counts, hematocrit values, total hemoglobin concentrations, mean cell volumes, mean corpuscular volumes, and mean corpuscular hemoglobin concentrations were analyzed from blood (Supplementary Fig. 2a–p) collected in tubes containing EDTA using a VetScan HM5 laser-based hematologic analyzer (Zoetis). Serum samples were tested for concentrations of albumin, amylase, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), blood urea nitrogen (BUN), calcium, creatinine (CRE), C-reactive protein (CRP), gamma-glutamyltransferase (GGT), glucose, total protein, and uric acid (Supplementary Fig. 3a–m) by using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer discs (Abaxis). Only eight of the 44 EBOV Kikwit-challenge historical controls were assayed using these machines; this subset was used for downstream figures and statistical analyses. Detection of viremia RNA was isolated from whole blood with the Viral RNA mini kit (Qiagen), using 100 µL of blood added to 600 µL of viral lysis buffer. Tissue RNA was extracted with the RNeasy Mini Kit (Qiagen) using 0.07 g tissue homogenized in 700 μL of lysis buffer. Primers and probes targeting the VP24 gene of BOMV were used for real-time quantitative PCR (RT-qPCR) with the following probes: 6FAM-5′-ACC-TGA-CAA-GTC-GGC-CCT-TGA-3′-TAMRA (TAMRA) (Life Technologies). Viral RNA was detected using the CFX96 detection system (Bio-Rad Laboratories, Hercules, CA) in one-step probe RT-qPCR kits (Qiagen) with the following cycle conditions: 50 °C for 10 min, 95 °C for 5 min, and 40 cycles of 95 °C for 10 s and 56 °C for 30 s. Samples were plated in technical duplicate. Threshold cycle (CT) values representing viral genomes were analyzed with CFX Manager software, and the data are shown as genome equivalents (GEq) per milliliter. To create the GEq standard, RNA from viral stocks was extracted, and the number of strain-specific genomes was calculated using Avogadro’s number and the molecular weight of each viral genome. Virus titration was performed for rBOMV by plaque assay with Vero E6 cells (ATCC, Manassas, Virginia, #CRL-1586) from all plasma and tissue samples, as previously described for filoviruses55,56. Increasing 10-fold dilutions of the samples were adsorbed to Vero E6 monolayers in technical duplicate wells (200 µL per well) for one hour before applying a 0.8% agarose overlay. Cells were stained with neutral red at 12 DPI, and plaques were read at 13 DPI. The detection limit for this assay is 25 PFU/mL for plasma and 250 PFU/g for tissues. Only 12 of the 44 EBOV Kikwit-challenge historical controls were assayed at comparable timepoints; this subset was used for downstream figures and statistical analyses. Histopathology and immunohistochemistry Necropsy was performed on all subjects. Tissue samples of all major organs were collected for histopathologic and IHC examination, immersion-fixed in 10% neutral buffered formalin, and processed for histopathology. An American College of Veterinary Pathologists board-certified veterinary pathologist assigned relative severity scores for histological lesions and immunoreactivity. Representative photomicrographs were qualitatively considered to display lesions that were nominally or ordinally measured by masking the veterinary pathologist after examination and ranking lesions to satisfy the study objectives, as previously established. Tissue sections were deparaffinized and rehydrated through xylene and graded ethanols. Antigen retrieval was performed in a steamer at 95 °C for 20 min in Sigma Citrate Buffer, pH 6.0, 10× (Sigma Aldrich, St. Louis, MO). To block endogenous peroxidase activity, slides were treated with 3% hydrogen peroxide and rinsed in distilled water. The tissue sections were processed for IHC using the Autostainer 360 (ThermoFisher, Kalamazoo, MI). Specific anti-EBOV and anti-rBOMV immunoreactivity was detected using the methods described below. Anti-nucleoprotein (NP) antibody staining with ImmPact red Specific anti-EBOV NP was detected using an anti-EBOV NP primary antibody at a 1:4000 dilution for 60 min (Absolute Antibody, United Kingdom, #Ab00692-23.0). The secondary antibody used was biotinylated goat anti-rabbit IgG (Vector Laboratories, Burlingame, CA, #BA-1000) at a 1:200 dilution for 30 min, followed by Vector streptavidin alkaline phosphatase at a 1:200 dilution for 15 min (Vector Laboratories, #SA-5100). Slides were developed with the ImmPact Red Substrate Kit (Vector Laboratories, #SK-5105) for 20 min and counterstained with hematoxylin for 30 s. Anti-EBOV glycoprotein (GP) and anti-EBOV VP40 staining with DAB Sequential 15-min incubations with avidin D and biotin solutions (Vector Laboratories, Burlingame, CA #SP-2001) were performed to block endogenous biotin reactivity. Specific anti-EBOV GP immunoreactivity was detected using an anti-EBOV GP primary antibody (Integrated Bio Therapeutics, Rockville, MD #0301-015) at a 1:500 dilution for 60 min. Specific anti-EBOV VP40 immunoreactivity was detected using an anti-EBOV VP40 primary antibody (Integrated Bio Therapeutics #0301-010) at a 1:4000 dilution for 60 min. The secondary antibody used was biotinylated goat anti-rabbit IgG (Vector Laboratories, Burlingame, CA #BA-1000) at a 1:200 dilution for 30 min, followed by Vector Horseradish Peroxidase Streptavidin, R.T.U. (Vector Laboratories #SA-5704) for 30 min. Slides were developed with Dako DAB chromogen (Dako, Carpinteria, CA #K3468) for 5 min and counterstained with hematoxylin for 30 s. RNA sequencing and analysis Total RNA was extracted from whole blood in TRIzol LS according to the manufacturer’s protocol. RNA cleanup was performed via QIAGEN RNeasy cleanup columns, and DNA was digested on-column during cleanup. Globin and ribosomal RNA were removed via the NEBNext Globin and rRNA Depletion Kit, and sequencing libraries were prepared with the NEBNext Ultra II RNA Library Prep Kit. Libraries were sequenced via Illumina NextSeq 500 in a paired-end, 75-bp format with a target depth of 50 million reads per sample. Transcriptomic analysis Raw FASTQ files were aligned to the combined M. fascicularis (Ensembl 102 v5.0) and BOMV (OK633884.1 [https://www.ncbi.nlm.nih.gov/nuccore/OK633884.1]) genomes using STAR aligner v2.7.11b69, and transcript-aligning reads were quantified with Rsubread featureCounts v2.20.070. The sequencing batch effect was regressed using SVA ComBat-seq v3.54.071 and verified using principal component analysis (Supplementary Fig. 1a-d). Postexposure samples were binned by day (DPI) and compared to baseline (−8 DPI) using DESeq2 v1.46.072. Significantly differentially expressed (DE) genes for each comparison had an FDR-adjusted p-value < 0.05 and a shrunken log2 fold change >2 or <−2. Upregulated genes were binned into three modules as assigned by k-means, with k = 3 determined using the within-cluster sum of squares. Ingenuity Pathway Analysis (IPA)73 was used to identify enriched canonical pathways and functions. CIBERSORTx74 was employed for digital cell quantitation using the LM22 matrix, B-mode batch correction, and was run in absolute mode. BOMV genomic analysis The rBOMV genome was analyzed for potential primate-adaptive SNVs by aligning raw FASTQ files to the corrected BOMV reference sequence (OK633884.1 [https://www.ncbi.nlm.nih.gov/nuccore/OK633884.1]) using Bowtie2 v2.3.4.375. Viral genome coverage was assessed via SAMtools v1.2076, and SNVs were called via LoFreq v.2.1.3.177. SNVs were annotated as nonsynonymous, synonymous, or noncoding in R via Biostrings v.2.74.078. Only the 7 DPI samples from NHPs A, C, and D had adequate viral genome coverage for SNV analysis (median aligned depth >10). Immunology Bead-based multiplex assays Plasma inflammatory mediators, thrombosis markers, and fibrinolysis analytes were measured by flow cytometry using BioLegend LegendPlex™ assays and a FACS Canto-II cytometer (Becton Dickinson). Samples were processed in technical duplicate for each Nonhuman Primate Inflammation 13-plex (1:4 dilution), Human Thrombosis (1:100 dilution), and Human Fibrinolysis (1:40,000 dilution) panel according to the manufacturer’s instructions. Standards were mixed in batches and aliquoted across all plates to ensure consistency. Optional wash steps were incorporated to reduce the background. ELISA: purified rBOMV GP To generate BOMV glycoprotein for anti-sera detection, a gene fragment encoding residues 1–642 of Bombali GP (ASJ82195.1) was codon optimized for Drosophila cell expression and cloned into pMT puro (Addgene) with a C-terminal fibritin domain and a double Strep-tag II. Schneider 2 (S2) insect cells were transfected using Effectene (Qiagen) following the manufacturer’s instructions, and a stable line was established by selection with 6 µg/mL puromycin. The cells were scaled up, and Bombali GP was purified from clarified supernatant via streptavidin affinity chromatography, followed by size-exclusion chromatography into phosphate-buffered saline, pH 7.4. The resulting antigen was adsorbed onto uncoated 96-well plates for ELISA at 0.2 μL/mL in ELISA coating buffer (pH 9.4). EDTA plasma collected at the indicated time points was tested for total anti-rBOMV IgG by ELISA using the plates mentioned above and MaxiSorp uncoated plates (ThermoFisher, Rochester, NY). Plasma was initially diluted 1:100 and then twofold through 1:25,600 in ELISA diluent (2% BSA in 1× PBS with 0.2% Tween 20). Samples were plated in technical duplicate. After a one-hour incubation, cells were washed four times with wash buffer (1× PBS with 0.2% Tween 20) and incubated for an hour with a dilution of horseradish peroxidase (HRP)-conjugated anti-monkey IgG (1:8000) and IgM (1:5000) (Fitzgerald Industries International, Acton, MA). SigmaFast O-phenylenediamine (OPD) substrate (Sigma; P9187) was added to the wells after four additional washes to develop the colorimetric reaction. The reaction was stopped with 3 M sulfuric acid approximately 5 min after the OPD addition, and absorbance values were measured at a wavelength of 492 nm on a spectrophotometer (BioTek Cytation 5). Absorbance values were normalized by subtracting uncoated from antigen-coated wells at the corresponding serum dilution. Endpoint titers were defined as the reciprocal of the last adjusted serum dilution with a value ≥0.15 for IgG and ≥0.25 for IgM. ELISA: cell lysate Diluted Vero E6 cell lysates were adsorbed onto uncoated 96-well plates in ELISA coating buffer (pH 9.4). Lysate from cells infected with EBOV Makona C07 was diluted 1:5000 before adsorption; SUDV Boniface, 1:1000; BDBV 200706291 Uganda, 1:1000; TAFV, 1:1000; uninfected control Vero E6 cells, 1:5000. After an overnight incubation at 4 °C, excess coating buffer was removed, and the plates were washed twice with wash buffer. ELISA blocking buffer (4% BSA in PBS) was added to all coated and uncoated plates, and plates were incubated at room temperature for 2 h, followed by two washes. EDTA plasma was plated in technical duplicate and diluted twofold from 1:100 through 1:25,600 on plates coated with each infected cell lysate (EBOV, SUDV, BDBV, and TAFV), uninfected cell lysate, and uncoated plates. Following a 1-h incubation at room temperature, the plates were washed four times and then incubated for an hour at room temperature with a dilution of HRP-conjugated anti-monkey IgG (1:8000). The plates were washed an additional four times before adding the OPD substrate and incubating for 10 min. The reaction was stopped with 3 M sulfuric acid, and absorbance values were measured at 492 nm. Absorbance values were normalized by subtracting the uncoated and uninfected cell lysate-coated values from the corresponding wells of the plates coated with infected cell lysate. Endpoint titers were defined as the reciprocal of the last normalized serum dilution with a value ≥0.15. Plaque reduction neutralization test Neutralization titers were calculated by determining the dilution of serum that reduced plaques by 50% or more (PRNT50). We incubated a standard 100 PFU amount of rBOMV, EBOV, or TAFV with two-fold serial dilutions of serum samples in Modified Eagle Medium (MEM). The virus-serum mixture was then used to inoculate Vero E6 cells for 60 min. Samples were plated in technical duplicates. Cells were overlaid with 0.7% agarose in 2X MEM and incubated for 7, 11, and 12 days for EBOV, BOMV, and TAFV, respectively. Plates were stained with 5% neutral red in PBS with 5% FBS, and plaques were counted 24 h after staining. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review File Source data Source data Acknowledgements The authors thank the UTMB Animal Resource Center for laboratory animal husbandry support and Dr. Kevin Melody, Dr. Chad Mire, and Mr. Mack B. Harrison for their assistance with animal studies. This study was supported in part by the Department of Health and Human Services, National Institutes of Health grant number U19AI142785 to T.W.G., in part by the UTMB Department of Microbiology and Immunology funds to T.W.G., and UC7AI094660 for BSL-4 operations support of the Galveston National Laboratory. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the University of Texas Medical Branch. Author contributions T.W.G., R.W.C. conceived and designed animal challenge experiments. C.G.A. provided the recombinant virus and advised on its use. D.J.D., J.B.G. performed animal challenges. J.B.G., A.N.P., D.J.D., T.W.G., and R.W.C. performed procedures and conducted clinical observations. K.N.A., V.B. performed clinical pathology assays. K.N.A. performed the PCR assays. V.B. performed the rBOMV plaque assays and neutralization assays. R.O. performed the ELISAs. C.W. performed the LegendPlex assays. N.S.D. performed the IHC assays. K.A.F. performed gross pathologic, histologic, and immunohistochemical analyses of the data. D.A. developed and provided purified rBOMV glycoprotein. All authors analyzed the data. J.T. wrote the paper with additions from K.A.F., T.W.G., and R.W.C. J.T., K.A.F. prepared the figures. T.W.G., J.T., and K.A.F. prepared the tables. J.H.C., T.W.G., and R.W.C. edited the manuscript. All authors had access to the data and approved the final manuscript. Peer review Peer review information Nature Communications thanks Miles Carroll and the other anonymous reviewer for their contribution to the peer review of this work. A peer review file is available. Data availability The RNA sequencing data generated in this study have been deposited in the NCBI GEO database under accession code GSE290671. The remaining data generated in this study are provided in the Supplementary Information/Source data file. Source data are provided with this paper. Code availability Bioinformatic code is hosted on GitHub (https://github.com/neidl-connor-lab/bomv-nhps)79. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Thomas W. Geisbert, Email: twgeisbe@utmb.edu Robert W. Cross, Email: rwcross@utmb.edu Supplementary information The online version contains Supplementary material available at 10.1038/s41467-026-73405-5. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review File Source data Data Availability Statement The RNA sequencing data generated in this study have been deposited in the NCBI GEO database under accession code GSE290671. The remaining data generated in this study are provided in the Supplementary Information/Source data file. Source data are provided with this paper. Bioinformatic code is hosted on GitHub (https://github.com/neidl-connor-lab/bomv-nhps)79.