Browse the corpus

Oral obeldesivir provides postexposure protection against Marburg virus in nonhuman primates. The recent outbreak of Marburg virus (MARV) in Rwanda underscores the need for effective countermeasures against this highly fatal pathogen, with case fatality rates reaching 90%. Currently, no vaccines or approved treatments exist for MARV infection, distinguishing it from related viruses such as Ebola. Our study demonstrates that the oral drug obeldesivir (ODV), a nucleoside analog prodrug, shows promising antiviral activity against filoviruses in vitro and offers significant protection in animal models. Here with cynomolgus macaques (n = 6), a 10 day regimen of once-daily ODV, initiated 24 h after exposure, provided 80% protection against a thousandfold lethal MARV challenge, delaying viral replication and disease onset. Transcriptome analysis revealed that early adaptive responses correlated with successful outcomes. Compared with intravenous options, oral antivirals such as ODV offer logistical advantages in outbreak settings, enabling easier administration and broader contact coverage. Our findings support the potential of ODV as a broad-spectrum, oral postexposure prophylaxis for filoviruses.

Marburg virus (MARV), belonging to the Filoviridae family and Orthomarburgvirus genus, is a hemorrhagic fever virus associated with high lethality in humans, with mortality rates in some outbreaks approaching 90% (refs. 1,2). Marburg outbreaks have occurred sporadically since 1967, concentrated in Central Africa. Outbreaks have largely been limited in size and duration but have periodically caused large and uncontrolled disease outbreaks sometimes taking years to fully control3,4. On 27 September 2024, the Rwandan Ministry of Health declared the first MARV outbreak in the country. As of 17 December 2024, 66 cases of MARV disease (MVD) and 15 deaths have been reported, with most cases occurring among healthcare workers5.

d uncontrolled disease outbreaks sometimes taking years to fully control3,4. On 27 September 2024, the Rwandan Ministry of Health declared the first MARV outbreak in the country. As of 17 December 2024, 66 cases of MARV disease (MVD) and 15 deaths have been reported, with most cases occurring among healthcare workers5. The 2013–2016 epidemic of Orthoebolavirus zairense (Ebola virus (EBOV)) that resulted in 28,600 cases and 11,325 deaths6 alerted the public health community and government agencies about the pandemic potential of filoviruses. As a result of this epidemic, a strong focus was placed on developing medical countermeasures against EBOV. These increased efforts played a large role in the licensure of two preventive vaccines and two monoclonal antibody (mAb)-based treatments against EBOV7–10. However, these interventions are specific to EBOV and there are currently no licensed vaccines or treatments for other orthoebolaviruses or orthomarburgviruses. Several MARV vaccine candidates have been evaluated in preclinical models, including non-human primates (NHP), showing efficacy against lethal MARV challenge11,12. The most promising candidates include the chimpanzee adenovirus (cAd3)- and recombinant vesicular stomatitis virus-based vaccines, as these candidates have been shown to induce protective immunity in NHPs following a single injection, making them the most practical vaccines to deploy in outbreaks13–15. Despite strong preclinical efficacy of the cAd3-based vaccine, which is being used in Rwanda16, there is a delay of 7–10 days between vaccination and production of protective immunity, during which the virus may continue to replicate and spread.

tion, making them the most practical vaccines to deploy in outbreaks13–15. Despite strong preclinical efficacy of the cAd3-based vaccine, which is being used in Rwanda16, there is a delay of 7–10 days between vaccination and production of protective immunity, during which the virus may continue to replicate and spread. mAbs (for example, MBP091) and remdesivir are candidate MARV treatments that have shown preclinical efficacy in late-stage lethal infection in NHPs, both alone and in combination17. These compounds have been included as investigational therapies for MARV treatment in the World Health Organization platform clinical trial protocol in Rwanda16,18. Despite substantial progress toward the development of MARV treatments, both MBP091 and remdesivir require intravenous delivery, limiting access to treatment centers with trained staff and specialized equipment. Thus, there remains a critical unmet medical need for an oral antiviral that may be more rapidly and widely deployed as a postexposure prophylaxis (PEP) in a MARV outbreak setting. Obeldesivir (ODV, GS-5245) is an oral prodrug of nucleoside analog GS-441524 that is metabolized in tissues to the same active triphosphate as remdesivir19. We recently reported that ODV has potent in vitro antiviral activity against several filoviruses, including EBOV, Sudan virus (SUDV) and MARV, and completely protected cynomolgus macaques against a lethal SUDV challenge when administered orally for 10 days beginning 24 h postinfection20. Here we report the in vivo postexposure efficacy of ODV against lethal MARV infection in NHPs.

To assess ODV efficacy as PEP against MARV, we challenged a cohort of healthy adult cynomolgus macaques (n = 6) with 1,000 plaque forming units (PFU) of MARV (Angola variant) by intramuscular injection. This dose has been used as an industry standard for filovirus research for decades by almost all filovirus medical countermeasure developers as it has been found to be uniformly lethal without exception. Beginning 1 day postinfection (DPI), the experimental cohort (n = 5) received oral ODV (100 mg kg−1) once daily for a total of 10 consecutive days. One animal served as an in-study placebo control and was treated in parallel with vehicle. The in-study placebo control (C-1) developed clinical signs of disease, including fever, at 5 DPI which progressed to severe MVD, and the animal succumbed at 7 DPI (Fig. 1a,b and Table 1). Clinical signs of illness in this animal were consistent with those of historical controls21, and this animal showed marked changes in hematological and serum chemistry parameters compared with baseline (day of challenge) values, including lymphocytopenia, thrombocytopenia and elevated markers of hepatic injury (for example, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), gamma-glutamyltransferase (GGT) and acute systemic inflammation (C-reactive protein (CRP)) (Table 1). One of the five ODV-treated macaques (Tx-4) showed clinical evidence of illness, including fever on 6 DPI (Table 1). The disease course in this ODV-treated animal was protracted compared with that in the in-study positive control (C-1) and historical controls. The animal succumbed to severe MVD on 13 DPI (Fig. 1a,b and Table 1). The remaining four ODV-treated animals survived the infection with mild and transient MVD, initiated 8–9 DPI and resolved by 17 DPI, and showed no evidence of clinical disease at the 35 DPI study endpoint (Fig. 1a,b and Table 1).Fig. 1Survival analysis, clinical scoring and virus replication kinetics in cynomolgus macaques challenged with MARV and treated with ODV.a, Kaplan–Meier survival curves for MARV-challenged cynomolgus macaques.

ed by 17 DPI, and showed no evidence of clinical disease at the 35 DPI study endpoint (Fig. 1a,b and Table 1).Fig. 1Survival analysis, clinical scoring and virus replication kinetics in cynomolgus macaques challenged with MARV and treated with ODV.a, Kaplan–Meier survival curves for MARV-challenged cynomolgus macaques. The in-study control is plotted separately; however, for statistical comparison, the in-study control was grouped with historical controls (HC) from previous studies. Differences in curves were tested by the Mantel–Cox log–rank test. b, Clinical scores were assigned based on daily cage-side observations of behavior and apparent physical health. The horizontal dashed line indicates the minimum clinical score by which euthanasia criteria were met. c,d, Viral load was determined by RT-qPCR of RNA from whole blood (c) or plaque titration of plasma (d) collected at predetermined sampling points or at euthanasia. e, Viral load in selected tissues collected at necropsy as determined by RT-qPCR detection of vRNA. For c–e, individual data points represent the mean of two technical replicates. For HC, the geometric mean ± geometric s.d. for the cohort is plotted. To fit on a log-scale axis, zero values (below LOQ) are plotted as ‘1’ (100). Dashed horizontal lines indicate the LLOQ for the assay (1,000 GEq ml−1 of blood or GEq g−1 of tissue for RT-qPCR; 25 PFU ml−1 for plaque titration). To fit on a log-scale axis, zero values (below LLOQ) are plotted as ‘1’ (100).

t is plotted. To fit on a log-scale axis, zero values (below LOQ) are plotted as ‘1’ (100). Dashed horizontal lines indicate the LLOQ for the assay (1,000 GEq ml−1 of blood or GEq g−1 of tissue for RT-qPCR; 25 PFU ml−1 for plaque titration). To fit on a log-scale axis, zero values (below LLOQ) are plotted as ‘1’ (100). ALN, axillary lymph node; ILN, inguinal lymph node; Liv, liver; Spl, spleen; Kid, kidney; Adr, adrenal gland; BrFr, brain frontal cortex; BrSt, brain stem; CSC, cervical spinal cord; Pan, pancreas; Uri, urinary bladder; Gon, gonad; Ut/Pro, uterus or prostate; NaMu, nasal mucosa; Conj, conjunctiva.Source dataTable 1Clinical description and outcome of cynomolgus macaques following MARV challenge with ODV treatmentNHPSexTreatmentClinical illnessClinical pathologyTx-1FODVFever (d6) and anorexia (d8–9); subject survived to study endpoint (d35)Leukocytosis (d7,14), lymphopenia (d7), monocytopenia (d1, 7, 10, 28, 35), monocytosis (d4, 14), neutrophilia (d1, 7, 14), neutropenia (d4), eosinophilia (d1, 7, 14, 21, 28, 35), basophilia (d14), CRP ↑ (d10)Tx-2MODVFever (d6–9), decreased appetite (d14–16), anorexia (d8–13, 17), mild depression (d11–14), hunched posture (d10–14), petechial rash (d9–13); subject survived to study endpoint (d35)Leukopenia (7, 10), leukocytosis (d14), lymphopenia (d4, 7, 10), lymphocytosis (d21), thrombocytopenia (d7, 10, 14), monocytosis (d10, 14, 21, 28), neutropenia (d10), neutrophilia (d14, 21, 28), eosinopenia (d4, 7, 10), basopenia (d7, 10), basophilia (d14, 21), ALT ↑ (d7, 21, 28), ALT ↑↑ (d14), ALT ↑↑↑↑ (d10), AST↑ (d7, 14, 21), AST ↑↑↑↑ (d10), ALP↑ (d10, 14, 21), GGT↑ (d10), CRP ↑↑↑ (d7, 10)Tx-3FODVFever (d7); subject survived to study endpoint (d35)Leukopenia (d4, 7), lymphopenia (d7), thrombocytopenia (d7, 35), monocytopenia (d1, 35), monocytosis (d10, 14), neutropenia (d4, 7, 10, 14), eosinophilia (d14, 21, 35), basopenia (d7), ALT ↑ (d4, 10), AST ↑ (d4), CRP ↑ (d10)Tx-4MODVFever (d6–10), hypothermia (d13), anorexia (d8, 10–13), mild depression (d11–12), depression (d13), weakness (d13), hunched posture (d10–12), recumbency (d13), petechial rash (d10–13); subject succumbed on d13Leukopenia (d7, 10, 13), lymphopenia (d7, 10), thrombocytopenia (d7, 10, 13), monocytopenia (d4, 7), neutropenia (d10, 13), neutrophilia (d1), eosinopenia (d7, 10, 13), basopenia (d7, 10, 13), anemia (d13), hypoalbuminemia (d13), hypoproteinemia (d13), hypoglycemia (d13), hypocalcemia (d13), BUN ↑↑ (d13), CRE ↑ (d10, 13), ALT ↑↑ (d10), ALT ↑↑↑↑ (d13), AST ↑ (d7

0), thrombocytopenia (d7, 10, 13), monocytopenia (d4, 7), neutropenia (d10, 13), neutrophilia (d1), eosinopenia (d7, 10, 13), basopenia (d7, 10, 13), anemia (d13), hypoalbuminemia (d13), hypoproteinemia (d13), hypoglycemia (d13), hypocalcemia (d13), BUN ↑↑ (d13), CRE ↑ (d10, 13), ALT ↑↑ (d10), ALT ↑↑↑↑ (d13), AST ↑ (d7 ), AST ↑↑↑↑ (d10, 13), ALP ↑ (d13), GGT ↑ (d10, 13), CRP ↑↑↑ (d7), CRP ↑↑ (d10, 13)Tx-5MODVAnorexia (d9); subject survived to study endpoint (d35)Leukopenia (d4, 35), lymphopenia (d7), monocytopenia (d35), neutropenia (d4, 7, 10, 21, 35), eosinopenia (d7, 35), eosinopenia (d4, 7, 35)C-1MControl (vehicle)Fever (d5–7), decreased appetite (d6), hunched posture (d7), depression (d7), weakness (d7), unresponsiveness (d7), dyspnea (d7), petechial rash (d6–7); subject succumbed on d7Leukopenia (d4, 7), lymphopenia (d4, 7), thrombocytopenia (d7), monocytopenia (d7), neutropenia (d7), basopenia (d4, 7), hypoglycemia (d7), hypoamylasemia (d7), ALT ↑ (d4), ALT ↑↑↑↑ (d7), AST ↑↑↑↑ (d7), ALP ↑ (d7), GGT ↑ (d7), CRP ↑ (d7)Days after MARV challenge are in parentheses.

l rash (d6–7); subject succumbed on d7Leukopenia (d4, 7), lymphopenia (d4, 7), thrombocytopenia (d7), monocytopenia (d7), neutropenia (d7), basopenia (d4, 7), hypoglycemia (d7), hypoamylasemia (d7), ALT ↑ (d4), ALT ↑↑↑↑ (d7), AST ↑↑↑↑ (d7), ALP ↑ (d7), GGT ↑ (d7), CRP ↑ (d7)Days after MARV challenge are in parentheses. All reported findings are in comparison to baseline (day of challenge (d0)) values. Decreased appetite is defined as ≤65% of food consumed from the previous day. Anorexia is defined as no food consumed from the previous day. Fever is defined as a temperature more than 2.5 °F over baseline, or at least 1.5 °F over baseline and ≥103.5 °F. Hypothermia is defined as a temperature ≤3.5 °F below baseline. Lymphocytopenia, monocytopenia, erythrocytopenia, thrombocytopenia, neutropenia, eosinopenia and basopenia are defined by a ≥35% drop in numbers of lymphocytes, monocytes, erythrocytes, platelets, neutrophils, eosinophils or basophils, respectively. Lymphocytosis, monocytosis, neutrophilia, eosinophila and basophilia are defined by a 100% or greater increase in numbers of lymphocytes, monocytes, neutrophils, eosinophils and basophils, respectively. Hyperglycemia is defined as a 100% or greater increase in levels of glucose. Hypoglycemia is defined by a ≥25% decrease in levels of glucose. Anemia is defined as a concurrent ≥25% decrease in erythrocyte count, Hct and Hgb. Hypoalbuminemia is defined by a ≥25% decrease in levels of albumin. Hypoproteinemia is defined by a ≥25% decrease in levels of total protein. Hypoamylasemia is defined by a ≥25% decrease in levels of serum amylase. Hypocalcemia is defined by a ≥25% decrease in levels of serum calcium. Increases in ALT, AST, ALP, CRE, CRP, Hct and Hgb were graded on the following scale: ↑ = 1–5-fold, ↑↑ = >5–10-fold, ↑↑↑ = >10–20-fold, ↑↑↑↑ = >20-fold, ↓ = ≥50% decrease.

ein. Hypoamylasemia is defined by a ≥25% decrease in levels of serum amylase. Hypocalcemia is defined by a ≥25% decrease in levels of serum calcium. Increases in ALT, AST, ALP, CRE, CRP, Hct and Hgb were graded on the following scale: ↑ = 1–5-fold, ↑↑ = >5–10-fold, ↑↑↑ = >10–20-fold, ↑↑↑↑ = >20-fold, ↓ = ≥50% decrease. a, Kaplan–Meier survival curves for MARV-challenged cynomolgus macaques. The in-study control is plotted separately; however, for statistical comparison, the in-study control was grouped with historical controls (HC) from previous studies. Differences in curves were tested by the Mantel–Cox log–rank test. b, Clinical scores were assigned based on daily cage-side observations of behavior and apparent physical health. The horizontal dashed line indicates the minimum clinical score by which euthanasia criteria were met. c,d, Viral load was determined by RT-qPCR of RNA from whole blood (c) or plaque titration of plasma (d) collected at predetermined sampling points or at euthanasia. e, Viral load in selected tissues collected at necropsy as determined by RT-qPCR detection of vRNA. For c–e, individual data points represent the mean of two technical replicates. For HC, the geometric mean ± geometric s.d. for the cohort is plotted. To fit on a log-scale axis, zero values (below LOQ) are plotted as ‘1’ (100). Dashed horizontal lines indicate the LLOQ for the assay (1,000 GEq ml−1 of blood or GEq g−1 of tissue for RT-qPCR; 25 PFU ml−1 for plaque titration). To fit on a log-scale axis, zero values (below LLOQ) are plotted as ‘1’ (100). ALN, axillary lymph node; ILN, inguinal lymph node; Liv, liver; Spl, spleen; Kid, kidney; Adr, adrenal gland; BrFr, brain frontal cortex; BrSt, brain stem; CSC, cervical spinal cord; Pan, pancreas; Uri, urinary bladder; Gon, gonad; Ut/Pro, uterus or prostate; NaMu, nasal mucosa; Conj, conjunctiva.

plotted as ‘1’ (100). ALN, axillary lymph node; ILN, inguinal lymph node; Liv, liver; Spl, spleen; Kid, kidney; Adr, adrenal gland; BrFr, brain frontal cortex; BrSt, brain stem; CSC, cervical spinal cord; Pan, pancreas; Uri, urinary bladder; Gon, gonad; Ut/Pro, uterus or prostate; NaMu, nasal mucosa; Conj, conjunctiva. Source data Clinical description and outcome of cynomolgus macaques following MARV challenge with ODV treatment

plotted as ‘1’ (100). ALN, axillary lymph node; ILN, inguinal lymph node; Liv, liver; Spl, spleen; Kid, kidney; Adr, adrenal gland; BrFr, brain frontal cortex; BrSt, brain stem; CSC, cervical spinal cord; Pan, pancreas; Uri, urinary bladder; Gon, gonad; Ut/Pro, uterus or prostate; NaMu, nasal mucosa; Conj, conjunctiva. Source data Clinical description and outcome of cynomolgus macaques following MARV challenge with ODV treatment Days after MARV challenge are in parentheses. All reported findings are in comparison to baseline (day of challenge (d0)) values. Decreased appetite is defined as ≤65% of food consumed from the previous day. Anorexia is defined as no food consumed from the previous day. Fever is defined as a temperature more than 2.5 °F over baseline, or at least 1.5 °F over baseline and ≥103.5 °F. Hypothermia is defined as a temperature ≤3.5 °F below baseline. Lymphocytopenia, monocytopenia, erythrocytopenia, thrombocytopenia, neutropenia, eosinopenia and basopenia are defined by a ≥35% drop in numbers of lymphocytes, monocytes, erythrocytes, platelets, neutrophils, eosinophils or basophils, respectively. Lymphocytosis, monocytosis, neutrophilia, eosinophila and basophilia are defined by a 100% or greater increase in numbers of lymphocytes, monocytes, neutrophils, eosinophils and basophils, respectively. Hyperglycemia is defined as a 100% or greater increase in levels of glucose. Hypoglycemia is defined by a ≥25% decrease in levels of glucose. Anemia is defined as a concurrent ≥25% decrease in erythrocyte count, Hct and Hgb. Hypoalbuminemia is defined by a ≥25% decrease in levels of albumin. Hypoproteinemia is defined by a ≥25% decrease in levels of total protein. Hypoamylasemia is defined by a ≥25% decrease in levels of serum amylase. Hypocalcemia is defined by a ≥25% decrease in levels of serum calcium. Increases in ALT, AST, ALP, CRE, CRP, Hct and Hgb were graded on the following scale: ↑ = 1–5-fold, ↑↑ = >5–10-fold, ↑↑↑ = >10–20-fold, ↑↑↑↑ = >20-fold, ↓ = ≥50% decrease.

ein. Hypoamylasemia is defined by a ≥25% decrease in levels of serum amylase. Hypocalcemia is defined by a ≥25% decrease in levels of serum calcium. Increases in ALT, AST, ALP, CRE, CRP, Hct and Hgb were graded on the following scale: ↑ = 1–5-fold, ↑↑ = >5–10-fold, ↑↑↑ = >10–20-fold, ↑↑↑↑ = >20-fold, ↓ = ≥50% decrease. The ODV-treated animal that succumbed to MARV infection had severe changes in hematological and serum chemistry parameters consistent with MVD, including lymphocytopenia, thrombocytopenia and elevated markers of hepatic and kidney injury (for example, ALT, AST, ALP, GGT, blood urea nitrogen (BUN), creatinine (CRE) and acute systemic inflammation (CRP) (Table 1). The four ODV-treated animals that survived infection also had various transient changes in hematological and serum chemistry values during the peak of viremia on 7–10 DPI and returned to baseline by study endpoint. For statistical comparison, the in-study control animal was grouped with 44 historical control cynomolgus macaques (n = 45, mean time to death (MTD) = 8.2 ± 0.7 DPI) challenged with the same MARV stock, dose and inoculation route. There was a significant difference in both the survival curves (P < 0.0001; Mantel–Cox log–rank test) and the proportional survival (P < 0.0001; Fisher’s exact test) between the ODV-treated and control groups (Fig. 1a).

ime to death (MTD) = 8.2 ± 0.7 DPI) challenged with the same MARV stock, dose and inoculation route. There was a significant difference in both the survival curves (P < 0.0001; Mantel–Cox log–rank test) and the proportional survival (P < 0.0001; Fisher’s exact test) between the ODV-treated and control groups (Fig. 1a). The magnitude and kinetics of filovirus replication and disease onset can vary based on the NHP species, which may impact the efficacy of candidate therapies in these models. To determine whether a difference in the progression of MARV infection exists between NHP species, we compared survival data from the 45 cynomolgus macaque control animals with data from 25 rhesus macaque positive control animals (MTD = 8.1 ± 0.7 DPI). There was no significant difference in either the survival curves (P = 0.494; Mantel–Cox log–rank test) or the MTD (P = 0.451; Welch’s t-test) between species (Fig. 2a,b). To determine whether there was a difference in the replication kinetics of MARV between the two NHP species, we compared the abundance of circulating MARV genomic RNA (vRNA) and circulating infectious virus at the earliest post-challenge sampling timepoint (3 DPI; cynomolgus, n = 35; rhesus, n = 11). There was no significant difference in the viral load, as assessed by quantitative reverse transcription polymerase chain reaction (RT-qPCR), at the earliest post-challenge sampling point (3 DPI) between the two species (P = 0.772; Mann–Whitney U-test; Fig. 2c); however, there was a significant difference in the amount of circulating infectious MARV (P = 0.046; Mann–Whitney U-test; Fig. 2d) indicating that some difference in the kinetics of infection between the two species may exist. Therefore, it is possible that greater protection may be achieved in rhesus macaques given the slightly faster development of infectious MARV in cynomolgus macaques.Fig. 2Survival analysis and comparisons of circulating MARV vRNA and infectious virus in cynomolgus and rhesus macaques.a, Kaplan–Meier survival curves for MARV-challenged HC cynomolgus and rhesus macaques. Survival data from the in-study control cynomolgus macaque (C-1) is included. Differences in curves were tested by the Mantel–Cox log–rank test. b, Comparison of the MTD of unprotected control cynomolgus macaques and that of rhesus macaques following infection with MARV. Statistical significance was tested by Welch’s t-test. c, Comparison of circulating MARV vRNA in cynomolgus and rhesus macaques at 3 DPI as measured by RT-qPCR.

by the Mantel–Cox log–rank test. b, Comparison of the MTD of unprotected control cynomolgus macaques and that of rhesus macaques following infection with MARV. Statistical significance was tested by Welch’s t-test. c, Comparison of circulating MARV vRNA in cynomolgus and rhesus macaques at 3 DPI as measured by RT-qPCR. d, Comparison of circulating infectious MARV in cynomolgus and rhesus macaques at 3 DPI as measured by plaque titration. For b, statistical comparison was made using Welch’s t-test. For c and d, the horizontal dashed lines indicate the LLOQ for the assay (1,000 GEq ml−1 for RT-qPCR; 25 PFU ml−1 for plaque titration). Individual data points represent the mean of two technical replicates. The bars indicate the geometric mean ± s.d. For statistical comparisons, undetectable values are plotted as values below the LLOQ (999 GEq ml−1 for RT-qPCR; 24 PFU ml−1 for plaque titration). Statistical comparisons were made using the Mann–Whitney U-test. All reported P values are two tailed. NS, not significant.

tes. The bars indicate the geometric mean ± s.d. For statistical comparisons, undetectable values are plotted as values below the LLOQ (999 GEq ml−1 for RT-qPCR; 24 PFU ml−1 for plaque titration). Statistical comparisons were made using the Mann–Whitney U-test. All reported P values are two tailed. NS, not significant. a, Kaplan–Meier survival curves for MARV-challenged HC cynomolgus and rhesus macaques. Survival data from the in-study control cynomolgus macaque (C-1) is included. Differences in curves were tested by the Mantel–Cox log–rank test. b, Comparison of the MTD of unprotected control cynomolgus macaques and that of rhesus macaques following infection with MARV. Statistical significance was tested by Welch’s t-test. c, Comparison of circulating MARV vRNA in cynomolgus and rhesus macaques at 3 DPI as measured by RT-qPCR. d, Comparison of circulating infectious MARV in cynomolgus and rhesus macaques at 3 DPI as measured by plaque titration. For b, statistical comparison was made using Welch’s t-test. For c and d, the horizontal dashed lines indicate the LLOQ for the assay (1,000 GEq ml−1 for RT-qPCR; 25 PFU ml−1 for plaque titration). Individual data points represent the mean of two technical replicates. The bars indicate the geometric mean ± s.d. For statistical comparisons, undetectable values are plotted as values below the LLOQ (999 GEq ml−1 for RT-qPCR; 24 PFU ml−1 for plaque titration). Statistical comparisons were made using the Mann–Whitney U-test. All reported P values are two tailed. NS, not significant.

tes. The bars indicate the geometric mean ± s.d. For statistical comparisons, undetectable values are plotted as values below the LLOQ (999 GEq ml−1 for RT-qPCR; 24 PFU ml−1 for plaque titration). Statistical comparisons were made using the Mann–Whitney U-test. All reported P values are two tailed. NS, not significant. As treatment was initiated 1 DPI, all animals, including the in-study vehicle control, had undetectable levels of vRNA or infectious MARV at the time ODV was first administered, as measured by RT-qPCR of whole blood or plaque titration of plasma, respectively (Fig. 1c,d). At 4 DPI, 3 ODV-treated animals (Tx-3, Tx-4 and Tx-5) had moderate levels of circulating vRNA (6.25–7.61 log10(genome equivalents (GEq) ml−1)) (Fig. 1c) and 4 animals (Tx-1, Tx-2, Tx-3 and Tx-4) had low levels of circulating infectious MARV (1.40–3.41 log10(PFU ml−1)) (Fig. 1d). At 4 DPI, 2 ODV-treated animals (Tx-1 and Tx-2) remained free of detectable quantities of circulating vRNA and 1 ODV-treated animal (Tx-5) remained free of infectious MARV (Fig. 1c,d). This is in contrast to the vehicle-treated control animal (C-1) that had a relatively high level of circulating vRNA (9.81 log10 (GEq ml−1)) and infectious MARV (5.07 log10(PFU ml−1)). The geometric mean circulating vRNA abundance was 5.47 log10(GEq ml−1) for ODV-treated macaques, compared with 6.99 log10(GEq ml−1) for the pooled in-study and historical controls (n = 9) at 4 DPI (values below the lower limit of quantitation (LLOQ) in both cohorts were assigned as 999 GEq ml−1), indicating that ODV treatment substantially reduced circulating viral loads at 4 DPI.

q ml−1) for ODV-treated macaques, compared with 6.99 log10(GEq ml−1) for the pooled in-study and historical controls (n = 9) at 4 DPI (values below the lower limit of quantitation (LLOQ) in both cohorts were assigned as 999 GEq ml−1), indicating that ODV treatment substantially reduced circulating viral loads at 4 DPI. By 7 and 10 DPI, 3 of the ODV-treated macaques (Tx-1, Tx-3 and Tx-5) had moderate levels of circulating vRNA (6.47–8.02 log10(GEq ml−1)) that cleared by 14 DPI (Fig. 1c). These three surviving ODV-treated animals also had low levels of circulating infectious MARV (1.70–2.35 log10(PFU ml−1)) at 7 DPI that cleared by 10 DPI (Fig. 1d). The fourth surviving ODV-treated macaque (Tx-2) had high levels of circulating vRNA at 7 and 10 DPI (10.53 log10(GEq ml−1 average)) that declined on 14 DPI (7.90 log10(GEq ml−1)) and 21 DPI (6.42 log10(GEq ml−1)) and cleared by 28 DPI (Fig. 1c); however, the levels of circulating infectious MARV in this animal at 7 and 10 DPI (4.13 and 3.03 log10(PFU mL−1), respectively) were relatively low and cleared by 14 DPI (Fig. 1d). By contrast, the ODV-treated animal that succumbed on 13 DPI (Tx-4) had high levels of circulating vRNA at 7 and 10 DPI (10.06 and 10.67 log10(GEq ml−1), respectively) and peaked at 13 DPI (12.40 log10(GEq ml−1)) (Fig. 1c); this animal also had a relatively high level of infectious MARV at 13 DPI (6.35 log10(PFU ml−1)) (Fig. 1d). Consistent with historical control cynomolgus macaques, the in-study vehicle control animal (C-1) had high levels of circulating vRNA (13.86 log10(GEq ml)) and infectious MARV (7.66 log10(PFU ml−1)) when it succumbed on 7 DPI (Fig. 1c,d).

atively high level of infectious MARV at 13 DPI (6.35 log10(PFU ml−1)) (Fig. 1d). Consistent with historical control cynomolgus macaques, the in-study vehicle control animal (C-1) had high levels of circulating vRNA (13.86 log10(GEq ml)) and infectious MARV (7.66 log10(PFU ml−1)) when it succumbed on 7 DPI (Fig. 1c,d). Tissues were collected at necropsy from all animals following euthanasia due to advanced disease or at the predetermined study endpoint. Most ODV-treated animals that survived the challenge had moderate levels of vRNA (5.79–9.29 log10(GEq g−1 tissue)) in some tissues assayed, except subject Tx-5, which had a high level of vRNA (11.42 log10(GEq g−1 tissue)) in the axillary lymph nodes (Fig. 1e). However, we were unable to detect any infectious MARV by plaque titration in the axillary lymph nodes of this animal or in select tissues with higher vRNA levels in other surviving animals (Tx-2, spleen and testes; Tx-3, kidney). Evidence of residual vRNA in the absence of infectious virus is not unexpected in recovered animals and is likely due to the presence of ongoing immune clearance mechanisms (that is, neutralizing antibodies, cellular immunity and so on) and has been documented by us in filovirus20 and arenavirus22–24 vaccine and treatment studies and by others for a number of other virus infection models25. By contrast, MARV vRNA was detected in moderate to high abundance (8.15–13.74 log10(GEq g−1 tissue)) in all tissues sampled from both the ODV-treated animal that succumbed at 13 DPI (Tx-4) and the untreated control animal (C-1) (Fig. 1e). MARV vRNA was not detectable in CNS tissue (frontal cortex, brain stem, cervical spinal cord) from any of the four surviving ODV-treated animals but was detected at moderate to high levels (8.76–10.00 log10(GEq g−1 tissue)) in CNS tissue from subject Tx-4. CNS tissue was not assayed for the vehicle-treated control animal (Fig. 1e).

not detectable in CNS tissue (frontal cortex, brain stem, cervical spinal cord) from any of the four surviving ODV-treated animals but was detected at moderate to high levels (8.76–10.00 log10(GEq g−1 tissue)) in CNS tissue from subject Tx-4. CNS tissue was not assayed for the vehicle-treated control animal (Fig. 1e). The presence of anti-MARV neutralizing antibodies was assessed by plaque reduction neutralization test (PRNT) using serum collected before the challenge (0 DPI), at 14 DPI, and before euthanasia due to terminal disease or at the study endpoint (35 DPI) (Fig. 3a,b). As expected, pre-challenge serum uniformly failed to achieve 50% reduction in MARV titers compared with the virus control plate (Fig. 3a). Likewise, neutralization capacity was below 50% for all surviving ODV-treated macaques at 14 DPI (Tx-4 and C-1 succumbed at 13 DPI and 7 DPI, respectively). Only two animals (Tx-3 and Tx-5) had weak neutralizing antibody titers (expressed as the reciprocal of the last dilution at which at least 50% neutralization was observed) of 10 and 20, respectively, at the study endpoint (Fig. 3b).Fig. 3Serum MARV-neutralizing antibody titers in cynomolgus macaques exposed to MARV and treated with ODV.Neutralizing antibody titers were determined by PRNT with a lower threshold of 50% (PRNT50). a, MARV-neutralizing antibody titers on the day of challenge (0 DPI) and at 14 DPI. The asterisk indicates that data from animals Tx-4 and C-1, which succumbed at 13 and 7 DPI, respectively, are plotted with 14 DPI data. The bars indicate the mean percentage neutralization ± s.d. at each timepoint after the challenge. b, Neutralization curves from surviving ODV-treated macaques at the study endpoint (35 DPI). The horizontal dashed line indicates 50% neutralization.Source data

at 13 and 7 DPI, respectively, are plotted with 14 DPI data. The bars indicate the mean percentage neutralization ± s.d. at each timepoint after the challenge. b, Neutralization curves from surviving ODV-treated macaques at the study endpoint (35 DPI). The horizontal dashed line indicates 50% neutralization.Source data Neutralizing antibody titers were determined by PRNT with a lower threshold of 50% (PRNT50). a, MARV-neutralizing antibody titers on the day of challenge (0 DPI) and at 14 DPI. The asterisk indicates that data from animals Tx-4 and C-1, which succumbed at 13 and 7 DPI, respectively, are plotted with 14 DPI data. The bars indicate the mean percentage neutralization ± s.d. at each timepoint after the challenge. b, Neutralization curves from surviving ODV-treated macaques at the study endpoint (35 DPI). The horizontal dashed line indicates 50% neutralization. Source data

Neutralizing antibody titers were determined by PRNT with a lower threshold of 50% (PRNT50). a, MARV-neutralizing antibody titers on the day of challenge (0 DPI) and at 14 DPI. The asterisk indicates that data from animals Tx-4 and C-1, which succumbed at 13 and 7 DPI, respectively, are plotted with 14 DPI data. The bars indicate the mean percentage neutralization ± s.d. at each timepoint after the challenge. b, Neutralization curves from surviving ODV-treated macaques at the study endpoint (35 DPI). The horizontal dashed line indicates 50% neutralization. Source data In the representative lymphoid tissue sections (Fig. 4a–o), lesions ranged from sinus histiocytosis to necrosis of germinal centers in the in-study positive control (C-1) and the ODV-treated animal that succumbed to disease (Tx-4). Positive immunohistochemistry (IHC) for the MARV antigen was noted in mononuclear cells throughout the lymphoid tissues (Fig. 4b,e). All four ODV-treated survivors lacked lesions and IHC positivity associated with MARV (Fig. 4h). In the representative gastrointestinal tissue sections, lesions ranged from lymphohistiocytic inflammatory infiltrates to multifocal hepatic necrosis and/or necrotizing gastroenteritis in the positive control (C-1) and the ODV-treated animal that succumbed to disease (Tx-4). Positive IHC was noted in hepatic sinusoidal lining cells of these nonsurviving animals (Fig. 4a,d), mononuclear cells within the lamina propria and rarely the overlying epithelial cells. Surviving macaques lacked lesions and IHC positivity associated with MARV (Fig. 4g). In the representative urogenital tissue sections, lesions were mild and ranged from lymphohistiocytic interstitial inflammatory infiltrates to lymphohistiocytic arteritis of the kidneys (C-1, Tx-1 and Tx-2). Positive IHC was noted in mononuclear cells within the interstitium of the kidney, urinary bladder and epididymis of the animals that succumbed to disease (C-1 and Tx-4). Two of the four ODV-treated survivors (Tx-1 and Tx-2) showed chronic inflammatory lesions (Fig. 4m,n) but lacked associated IHC positivity for the MARV antigen. Other organs evaluated included the adrenal glands, lungs, eyes and brain, including the pituitary gland and trigeminal ganglion. Mild lymphohistiocytic infiltrates were the prominent lesion across these organs. IHC positivity of the alveolar septum and few alveolar macrophages were noted for the animals that succumbed to disease (Fig. 4c,f). Mild lymphocytic perivascular infiltrates with no IHC immunolabeling were noted in one ODV-treated survivor (TX-1). The remaining survivors lacked pulmonary lesions and IHC positivity associated with MARV (Fig. 4i).

ar septum and few alveolar macrophages were noted for the animals that succumbed to disease (Fig. 4c,f). Mild lymphocytic perivascular infiltrates with no IHC immunolabeling were noted in one ODV-treated survivor (TX-1). The remaining survivors lacked pulmonary lesions and IHC positivity associated with MARV (Fig. 4i). Scattered mononuclear cells and clusters of cells within the adrenal cortex were IHC positive for MARV for the two animals that succumbed to disease (C-1 and Tx-4). The brain was not collected for the in-study control animal (C-1). Mild lymphohistiocytic meningitis, choroid plexitis and ganglionitis of the trigeminal ganglion were noted as a cluster of lesions for the ODV-treated animal that succumbed to disease (Tx-4). IHC positivity colocalized with lesions in satellite cells, glial cells and axons of the trigeminal ganglion (Fig. 4j), ependymal cells and mononuclear cells of the choroid plexus (Fig. 4k) and mononuclear cells clustered within the meninges (Fig. 4l). One surviving ODV-treated animal had mild lymphohistiocytic choroid plexitis but lacked IHC positivity (Fig. 4o). No lesions or IHC positivity associated with MARV was noted in the examined sections of eye across all macaques enrolled in the study.Fig. 4Representative hematoxylin and eosin-stained and IHC images of anti-MARV NP antibody in NHPs.a–o, Images taken from the in-study positive control, C-1 (a–c); ODV-treated nonsurvivor at 13 DPI, Tx-4 (d–f,j–l); and ODV-treated survivors at 35 DPI, Tx-5 (g–i), TX-2 (m) and Tx-1 (n,o). The images were captured with ×20 (a–k,n,o), ×40 (l) and ×10 (m) objectives (×200, ×400 and ×100 total magnification, respectively). Liver with widespread, diffuse immunolabeling (red) of hepatic sinusoidal lining cells (a), multifocal to coalescing immunolabeling of hepatic sinusoidal lining cells (d) and no appreciable immunolabeling in the liver (g). Spleen with immunolabeling of individual to clusters of mononuclear cells within the white and red pulp (b), scattered individual immunolabeling of mononuclear cells within the red and white pulp (arrows) (e) and no appreciable immunolabeling in the spleen (h). Lung with locally extensive immunolabeling of the alveolar septum and rare alveolar macrophages (c), immunolabeling of rare clusters of mononuclear cells within the alveolar septum (arrow) (f) and no appreciable immunolabeling in the lung (i).

white pulp (arrows) (e) and no appreciable immunolabeling in the spleen (h). Lung with locally extensive immunolabeling of the alveolar septum and rare alveolar macrophages (c), immunolabeling of rare clusters of mononuclear cells within the alveolar septum (arrow) (f) and no appreciable immunolabeling in the lung (i). Positive immunolabeling was also noted in satellite cells, glial cells and axons of the trigeminal ganglion (j), ependymal cells and mononuclear cells of the choroid plexus (arrow) (k) and mononuclear cells clustered within the meninges (arrow) (l). Chronic inflammatory infiltrates were noted in various organs in surviving NHPs. Lymphohistiocytic and eosinophilic epididymitis (m), lymphohistiocytic arteritis of the kidney (arrows) (n) and lymphohistiocytic choroid plexitis (o). For l, scale bar = 10 µm. For all other panels, scale bars = 100 µm.

inges (arrow) (l). Chronic inflammatory infiltrates were noted in various organs in surviving NHPs. Lymphohistiocytic and eosinophilic epididymitis (m), lymphohistiocytic arteritis of the kidney (arrows) (n) and lymphohistiocytic choroid plexitis (o). For l, scale bar = 10 µm. For all other panels, scale bars = 100 µm. a–o, Images taken from the in-study positive control, C-1 (a–c); ODV-treated nonsurvivor at 13 DPI, Tx-4 (d–f,j–l); and ODV-treated survivors at 35 DPI, Tx-5 (g–i), TX-2 (m) and Tx-1 (n,o). The images were captured with ×20 (a–k,n,o), ×40 (l) and ×10 (m) objectives (×200, ×400 and ×100 total magnification, respectively). Liver with widespread, diffuse immunolabeling (red) of hepatic sinusoidal lining cells (a), multifocal to coalescing immunolabeling of hepatic sinusoidal lining cells (d) and no appreciable immunolabeling in the liver (g). Spleen with immunolabeling of individual to clusters of mononuclear cells within the white and red pulp (b), scattered individual immunolabeling of mononuclear cells within the red and white pulp (arrows) (e) and no appreciable immunolabeling in the spleen (h). Lung with locally extensive immunolabeling of the alveolar septum and rare alveolar macrophages (c), immunolabeling of rare clusters of mononuclear cells within the alveolar septum (arrow) (f) and no appreciable immunolabeling in the lung (i). Positive immunolabeling was also noted in satellite cells, glial cells and axons of the trigeminal ganglion (j), ependymal cells and mononuclear cells of the choroid plexus (arrow) (k) and mononuclear cells clustered within the meninges (arrow) (l). Chronic inflammatory infiltrates were noted in various organs in surviving NHPs. Lymphohistiocytic and eosinophilic epididymitis (m), lymphohistiocytic arteritis of the kidney (arrows) (n) and lymphohistiocytic choroid plexitis (o). For l, scale bar = 10 µm. For all other panels, scale bars = 100 µm.

inges (arrow) (l). Chronic inflammatory infiltrates were noted in various organs in surviving NHPs. Lymphohistiocytic and eosinophilic epididymitis (m), lymphohistiocytic arteritis of the kidney (arrows) (n) and lymphohistiocytic choroid plexitis (o). For l, scale bar = 10 µm. For all other panels, scale bars = 100 µm. To identify immune correlates linked to ODV-mediated protection, we conducted a targeted transcriptomic analysis of whole-blood RNA from MARV-exposed macaques (n = 5 ODV treated, n = 1 in-study control, n = 4 historical controls). Samples were compared across several stages: after the challenge (1–2 DPI) and early (4–5 DPI), mid (7 DPI) and late disease (10 DPI or terminal for controls) (Source Data Fig. 5). Most transcriptional shifts occurred at the mid and late stages (Extended Data Fig. 1), with a larger number of downregulated transcripts than upregulated ones. Functional enrichment of differentially expressed transcripts revealed early activation of ‘interferon gamma signaling’ and T cell pathways (for example, ‘costimulation by the CD28 family’ and ‘T-cell receptor signaling’). Among treated survivors, the ‘peroxisome proliferator-activated receptor (PPAR) signaling’ pathway was notably upregulated in both early and late disease stages, which was shown previously following ODV treatment against other filoviruses20 (Source Data Extended Data Fig. 1). PPAR-specific ligands are known to suppress pro-inflammatory cytokines such as interleukin-6 (IL-6), tumor necrosis factor (TNF), interleukin-1β (IL-1β) and nitric oxide across different cell types26, indicating a possible mechanism by which early ODV treatment modulates excessive inflammation. Additional upregulated pathways in treated survivors involved antigen presentation, including ‘phosphatase and tensin homolog deleted on chromosome 10 signaling’ and ‘class I major histocompatibility complex-mediated antigen processing and presentation’, along with immunoregulatory processes such as ‘IL-10 signaling’.

Additional upregulated pathways in treated survivors involved antigen presentation, including ‘phosphatase and tensin homolog deleted on chromosome 10 signaling’ and ‘class I major histocompatibility complex-mediated antigen processing and presentation’, along with immunoregulatory processes such as ‘IL-10 signaling’. We also performed a pathway analysis comparison of treated survivors and untreated controls, revealing a significant downregulation of pathways associated with MVD—such as ‘pathogen-induced cytokine storm signaling’, ‘NF-κB signaling’ and ‘IL-6 signaling’—at various timepoints in the treated survivors. By contrast, these pathways remained predictably upregulated in the control subjects (Fig. 5a). We further compared the treated survivors with the treated fatal subject. Among treated survivors, the most significantly upregulated transcripts encoding proteins associated with B cell activation (CD79A, CD79B, MS4A1, CD19, CD22), antigen presentation (CD1C, CIITA, FN1, CD40, GPR183, HLA-DOB, HLA-DPB1) and immunoregulation (BTLA, CD22) (Fig. 5b). Notably, BTLA and CD22 are genes whose protein products are key regulators of B cell activity27,28, helping to prevent excessive immune responses and autoimmunity. Many of these transcripts, such as CD79B, CD40 and CIITA, showed early upregulation postinfection (1–2 DPI) in treated survivors, but were downregulated in the treated fatal subject (Fig. 5c). Conversely, the transcripts most significantly downregulated in treated survivors compared with the fatal subject included ATG7, CARD9 and TRADD (Fig. 5c). These encoded molecules play diverse roles in immunity: ATG7 is essential for autophagy29, CARD9 activates NF-κB signaling to promote cytokine production in response to pathogens30 and TRADD mediates TNF receptor signaling, triggering either apoptosis or inflammation through NF-κB31.Fig. 5Transcriptional profiling of ODV-treated macaques exposed to MARV.Blood RNA samples from all subjects (n = 4 ODV-treated survivors; n = 1 ODV-treated fatal; n = 5 control subjects). a, Pathway analysis of differentially expressed transcripts (Benjamini–Hochberg-adjusted P < 0.01, two tailed) comparing transcriptomic changes in ODV-treated survivors versus controls across each disease progression stage (1–2 DPI, 4–5 DPI, 7 DPI and 10 DPI). Pathways are sorted by the topmost positive (top) or negative (bottom) z-scores for the ODV survivor group at 10 DPI.

s (Benjamini–Hochberg-adjusted P < 0.01, two tailed) comparing transcriptomic changes in ODV-treated survivors versus controls across each disease progression stage (1–2 DPI, 4–5 DPI, 7 DPI and 10 DPI). Pathways are sorted by the topmost positive (top) or negative (bottom) z-scores for the ODV survivor group at 10 DPI. Red indicates increased expression; blue indicates decreased expression; white indicates no change in expression. b, Volcano plot depicting −log10(P value, two-tailed) and log2(fold change) values of differentially expressed transcripts between ODV-treated survivors (n = 4) and the treated fatal case (n = 1) across all timepoints. Key to horizontal lines: solid, P < 0.01; dashed, P < 0.05; dotted, P < 0.10; dotted-dashed, P < 0.50. c, Heat map showing top differentially expressed genes (Benjamini–Hochberg-adjusted P < 0.05, two tailed) in ODV-treated survivors compared with the treated fatal at each timepoint (sorted by the ODV survivor group at 1–2 DPI). Any differentially expressed transcripts with a Benjamini–Hochberg false discovery rate-corrected P value less than 0.05 were deemed significant. d, Immune cell transcriptional profiling of controls (n = 5), the treated fatal (n = 1) and ODV-treated survivors across all timepoints. Higher cell-type scores indicate a higher abundance of transcripts mapping to the specific cell subset.Source data

overy rate-corrected P value less than 0.05 were deemed significant. d, Immune cell transcriptional profiling of controls (n = 5), the treated fatal (n = 1) and ODV-treated survivors across all timepoints. Higher cell-type scores indicate a higher abundance of transcripts mapping to the specific cell subset.Source data Blood RNA samples from all subjects (n = 4 ODV-treated survivors; n = 1 ODV-treated fatal; n = 5 control subjects). a, Pathway analysis of differentially expressed transcripts (Benjamini–Hochberg-adjusted P < 0.01, two tailed) comparing transcriptomic changes in ODV-treated survivors versus controls across each disease progression stage (1–2 DPI, 4–5 DPI, 7 DPI and 10 DPI). Pathways are sorted by the topmost positive (top) or negative (bottom) z-scores for the ODV survivor group at 10 DPI. Red indicates increased expression; blue indicates decreased expression; white indicates no change in expression. b, Volcano plot depicting −log10(P value, two-tailed) and log2(fold change) values of differentially expressed transcripts between ODV-treated survivors (n = 4) and the treated fatal case (n = 1) across all timepoints. Key to horizontal lines: solid, P < 0.01; dashed, P < 0.05; dotted, P < 0.10; dotted-dashed, P < 0.50. c, Heat map showing top differentially expressed genes (Benjamini–Hochberg-adjusted P < 0.05, two tailed) in ODV-treated survivors compared with the treated fatal at each timepoint (sorted by the ODV survivor group at 1–2 DPI). Any differentially expressed transcripts with a Benjamini–Hochberg false discovery rate-corrected P value less than 0.05 were deemed significant. d, Immune cell transcriptional profiling of controls (n = 5), the treated fatal (n = 1) and ODV-treated survivors across all timepoints. Higher cell-type scores indicate a higher abundance of transcripts mapping to the specific cell subset.

false discovery rate-corrected P value less than 0.05 were deemed significant. d, Immune cell transcriptional profiling of controls (n = 5), the treated fatal (n = 1) and ODV-treated survivors across all timepoints. Higher cell-type scores indicate a higher abundance of transcripts mapping to the specific cell subset. Source data To further explore immune cell shifts, we performed digital cell quantitation using transcriptional profiling (Fig. 5d). This analysis aligned with our differential expression findings, indicating that successful ODV treatment was associated with a predicted increase in B cell-related transcripts in macaques that survived MARV infection. By contrast, untreated controls and the treated fatal subject showed marked neutrophilia and the upregulation of neutrophil activation-related transcripts, a common finding in MVD. Collectively, these findings suggest that ODV-treated survivors showed enhanced markers of adaptive immunity and immunoregulation, while also showing a reduction in markers linked to inflammation.

As treatment was initiated 1 DPI, all animals, including the in-study vehicle control, had undetectable levels of vRNA or infectious MARV at the time ODV was first administered, as measured by RT-qPCR of whole blood or plaque titration of plasma, respectively (Fig. 1c,d). At 4 DPI, 3 ODV-treated animals (Tx-3, Tx-4 and Tx-5) had moderate levels of circulating vRNA (6.25–7.61 log10(genome equivalents (GEq) ml−1)) (Fig. 1c) and 4 animals (Tx-1, Tx-2, Tx-3 and Tx-4) had low levels of circulating infectious MARV (1.40–3.41 log10(PFU ml−1)) (Fig. 1d). At 4 DPI, 2 ODV-treated animals (Tx-1 and Tx-2) remained free of detectable quantities of circulating vRNA and 1 ODV-treated animal (Tx-5) remained free of infectious MARV (Fig. 1c,d). This is in contrast to the vehicle-treated control animal (C-1) that had a relatively high level of circulating vRNA (9.81 log10 (GEq ml−1)) and infectious MARV (5.07 log10(PFU ml−1)). The geometric mean circulating vRNA abundance was 5.47 log10(GEq ml−1) for ODV-treated macaques, compared with 6.99 log10(GEq ml−1) for the pooled in-study and historical controls (n = 9) at 4 DPI (values below the lower limit of quantitation (LLOQ) in both cohorts were assigned as 999 GEq ml−1), indicating that ODV treatment substantially reduced circulating viral loads at 4 DPI.

The presence of anti-MARV neutralizing antibodies was assessed by plaque reduction neutralization test (PRNT) using serum collected before the challenge (0 DPI), at 14 DPI, and before euthanasia due to terminal disease or at the study endpoint (35 DPI) (Fig. 3a,b). As expected, pre-challenge serum uniformly failed to achieve 50% reduction in MARV titers compared with the virus control plate (Fig. 3a). Likewise, neutralization capacity was below 50% for all surviving ODV-treated macaques at 14 DPI (Tx-4 and C-1 succumbed at 13 DPI and 7 DPI, respectively). Only two animals (Tx-3 and Tx-5) had weak neutralizing antibody titers (expressed as the reciprocal of the last dilution at which at least 50% neutralization was observed) of 10 and 20, respectively, at the study endpoint (Fig. 3b).Fig. 3Serum MARV-neutralizing antibody titers in cynomolgus macaques exposed to MARV and treated with ODV.Neutralizing antibody titers were determined by PRNT with a lower threshold of 50% (PRNT50). a, MARV-neutralizing antibody titers on the day of challenge (0 DPI) and at 14 DPI. The asterisk indicates that data from animals Tx-4 and C-1, which succumbed at 13 and 7 DPI, respectively, are plotted with 14 DPI data. The bars indicate the mean percentage neutralization ± s.d. at each timepoint after the challenge. b, Neutralization curves from surviving ODV-treated macaques at the study endpoint (35 DPI). The horizontal dashed line indicates 50% neutralization.Source data

In the representative lymphoid tissue sections (Fig. 4a–o), lesions ranged from sinus histiocytosis to necrosis of germinal centers in the in-study positive control (C-1) and the ODV-treated animal that succumbed to disease (Tx-4). Positive immunohistochemistry (IHC) for the MARV antigen was noted in mononuclear cells throughout the lymphoid tissues (Fig. 4b,e). All four ODV-treated survivors lacked lesions and IHC positivity associated with MARV (Fig. 4h). In the representative gastrointestinal tissue sections, lesions ranged from lymphohistiocytic inflammatory infiltrates to multifocal hepatic necrosis and/or necrotizing gastroenteritis in the positive control (C-1) and the ODV-treated animal that succumbed to disease (Tx-4). Positive IHC was noted in hepatic sinusoidal lining cells of these nonsurviving animals (Fig. 4a,d), mononuclear cells within the lamina propria and rarely the overlying epithelial cells. Surviving macaques lacked lesions and IHC positivity associated with MARV (Fig. 4g). In the representative urogenital tissue sections, lesions were mild and ranged from lymphohistiocytic interstitial inflammatory infiltrates to lymphohistiocytic arteritis of the kidneys (C-1, Tx-1 and Tx-2). Positive IHC was noted in mononuclear cells within the interstitium of the kidney, urinary bladder and epididymis of the animals that succumbed to disease (C-1 and Tx-4). Two of the four ODV-treated survivors (Tx-1 and Tx-2) showed chronic inflammatory lesions (Fig. 4m,n) but lacked associated IHC positivity for the MARV antigen. Other organs evaluated included the adrenal glands, lungs, eyes and brain, including the pituitary gland and trigeminal ganglion. Mild lymphohistiocytic infiltrates were the prominent lesion across these organs. IHC positivity of the alveolar septum and few alveolar macrophages were noted for the animals that succumbed to disease (Fig. 4c,f). Mild lymphocytic perivascular infiltrates with no IHC immunolabeling were noted in one ODV-treated survivor (TX-1). The remaining survivors lacked pulmonary lesions and IHC positivity associated with MARV (Fig. 4i).

To identify immune correlates linked to ODV-mediated protection, we conducted a targeted transcriptomic analysis of whole-blood RNA from MARV-exposed macaques (n = 5 ODV treated, n = 1 in-study control, n = 4 historical controls). Samples were compared across several stages: after the challenge (1–2 DPI) and early (4–5 DPI), mid (7 DPI) and late disease (10 DPI or terminal for controls) (Source Data Fig. 5). Most transcriptional shifts occurred at the mid and late stages (Extended Data Fig. 1), with a larger number of downregulated transcripts than upregulated ones. Functional enrichment of differentially expressed transcripts revealed early activation of ‘interferon gamma signaling’ and T cell pathways (for example, ‘costimulation by the CD28 family’ and ‘T-cell receptor signaling’). Among treated survivors, the ‘peroxisome proliferator-activated receptor (PPAR) signaling’ pathway was notably upregulated in both early and late disease stages, which was shown previously following ODV treatment against other filoviruses20 (Source Data Extended Data Fig. 1). PPAR-specific ligands are known to suppress pro-inflammatory cytokines such as interleukin-6 (IL-6), tumor necrosis factor (TNF), interleukin-1β (IL-1β) and nitric oxide across different cell types26, indicating a possible mechanism by which early ODV treatment modulates excessive inflammation. Additional upregulated pathways in treated survivors involved antigen presentation, including ‘phosphatase and tensin homolog deleted on chromosome 10 signaling’ and ‘class I major histocompatibility complex-mediated antigen processing and presentation’, along with immunoregulatory processes such as ‘IL-10 signaling’.

The recent MARV outbreak in Rwanda is a reminder of the continuous threat that filoviruses pose to regions across Africa. The 2013–2016 EBOV epidemic highlights the potential for global filovirus spread and warrants the advancement of promising oral PEP agents as critical components of future outbreak preparedness. Moreover, antivirals with broad coverage across genetically diverse viruses are needed in the event of resistance mutations to mAbs or vaccines. Remdesivir has shown promising success in protecting NHPs when given after virus challenge against orthoebolaviruses32,33 and orthomarburgviruses17,34, thereby showing the value of both remdesivir and ODV in serving as broad-spectrum filovirus therapeutics. While remdesivir remains a critical countermeasure for treatment of symptomatic filovirus disease, ODV is poised for filovirus PEP as part of early outbreak response owing to its oral administration route. There may also be benefit to the use of oral ODV in combination with mAbs for the treatment of diagnosed filovirus infection, should they become available for use. This is based on the increased efficacy observed in NHP models of MARV and SUDV virus challenge when remdesivir was combined with monoclonal antibodies17,33. Our analysis of MARV-exposed macaques treated with ODV revealed key immune markers associated with survival. B cell activation and development of neutralizing antibodies, antigen presentation and immunoregulation were enhanced in survivors. By contrast, inflammatory pathways commonly associated with MVD, such as NF-κB and IL-6 signaling, were downregulated in ODV-treated survivors but remained elevated in the controls and the fatal subject. Digital cell profiling showed increased B cells in survivors, while controls and the fatal case showed neutrophilia, underscoring the role of ODV in promoting adaptive immunity over excessive inflammation. The correlates of immunity derived from treatment with ODV are more likely akin to correlates of immunity associated with natural survival, where both cellular and humoral immunity are offered the opportunity to mature in the context of the full complement of viral antigens.

aptive immunity over excessive inflammation. The correlates of immunity derived from treatment with ODV are more likely akin to correlates of immunity associated with natural survival, where both cellular and humoral immunity are offered the opportunity to mature in the context of the full complement of viral antigens. This may result in a complementary balance of both cellular and antibody-mediated protection from subsequent infections; however, as the goal of the study was to assess the protective potential of the therapy, the focused transcriptomic work offered presents merely a snapshot of the development of this type of immunity and more work is needed to fully elucidate this outcome.

ar and antibody-mediated protection from subsequent infections; however, as the goal of the study was to assess the protective potential of the therapy, the focused transcriptomic work offered presents merely a snapshot of the development of this type of immunity and more work is needed to fully elucidate this outcome. The objective of our study was to determine whether the efficacy of ODV as PEP recently reported for SUDV20 could be translated to MARV. Our results showed that ODV PEP in cynomolgus macaques conferred 80% survival and an accompanying reduction in plasma viral loads in an otherwise uniformly lethal infection model using one of the most lethal MARV strains35 at a dose 1,000 times greater than previously documented as uniformly lethal36. The efficacy of ODV administered later than 24 h after MARV exposure warrants additional studies, but the proof-of-concept data reported herein support the potential utility of ODV for PEP of subjects with known exposures to MARV, including accidental exposures in laboratories or clinics, and high-risk contacts, particularly among healthcare workers who are facing the highest risk of exposure. This has specific relevance to the recent MARV outbreak in Rwanda, which disproportionately affected the healthcare personnel. Deployment of oral ODV in coordination with other experimental vaccines and therapies may ease the burden on treatment centers, help to break the chain of transmission and more effectively contain the outbreak. The pathways to regulatory approval through efficacy trials for filovirus countermeasures have long been challenged by the sporadic nature of outbreaks. Nonetheless, efforts to introduce deployable compassionate use or randomized clinical trial protocols are underway by groups such as the World Health Organization-led MARV vaccine consortium and others to ensure that countermeasures such as ODV can be rapidly evaluated in the context of new outbreaks11. Indeed, a clinical trial has been initiated to evaluate this drug in the context of a MARV outbreak37. Another option may be to use the US Food and Drug Administration Animal Rule, representing an alternative regulatory pathway through which well-controlled animal studies provide evidence of efficacy when clinical trials are not practical or ethical owing to the nature of the disease threat38.

text of a MARV outbreak37. Another option may be to use the US Food and Drug Administration Animal Rule, representing an alternative regulatory pathway through which well-controlled animal studies provide evidence of efficacy when clinical trials are not practical or ethical owing to the nature of the disease threat38. Ideally, both the implementation of the deployable clinical trial protocols and the Animal Rule regulatory path should be pursued in parallel for further testing and development of ODV and other promising filovirus medical countermeasures.

The MARV Angola p2 seed stock originates from the 2004–2005 Uige, Angola, outbreak (DQ: 447653.1). The virus was isolated from the serum of a patient who died from MVD in 2005. The study challenge material was created by passaging the original isolate in Vero E6 cells (ATCC, CRL-1586). No mycoplasma or endotoxin could be detected (˂0.5 endotoxin units per ml). The ODV drug substance used for the NHP study was prepared using a synthetic route previously described and purified by crystallization as Form III19. All study protocols described were approved by the University of Texas Medical Branch (UTMB) Institutional Animal Care and Use Committee, which were compliant with UTMB Institutional Biosafety Committee guidelines under biosafety level 4 containment. UTMB animal 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.

biosafety level 4 containment. UTMB animal 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. Before the study was conducted, a power analysis was performed to determine the minimum cohort size for the study. For the MARV ‘Angola isolate’, assuming a one-tailed alpha of 0.05, sample sizes of 5 per group will provide >80% power to detect a difference in the proportion of surviving animals between the treatment group (100% survival rate) and the control group (0% survival rate), using a Fisher’s exact test. One control animal per challenge period is used to confirm the lethality of the challenge material used in the treatment studies wherein clinical parameters are compared with those of historical control animals derived from other experiments using the identical challenge material (that is, same virus passage). This allows an ethical reduction in animal number use for data with highly predictable, lethal outcomes. Accordingly, six healthy captive-bred cynomolgus macaques (Macaca fascicularis) (Worldwide Primates) of Indonesian origin ~3 years of age and weighing ~2.1–3.5 kg were challenged by intramuscular injection with a target dose of 1,000 PFU of MARV Angola variant (actual dose, 988 PFU). Assignment to the treatment group or control group was determined before the challenge by randomization using Excel. Five animals were treated by oral gavage with 100 mg kg−1 ODV (as a suspension in a 0.5% methylcellulose in water vehicle) beginning 24 h after MARV exposure. These five animals received daily doses of ODV for 10 days (1–10 DPI). The MARV-positive control animal was treated in parallel with 0.5% methylcellulose in water. The duration of the NHP study was 35 days. All six macaques were monitored daily and scored for disease progression with an internal MARV humane endpoint scoring sheet approved by the UTMB Institutional Animal Care and Use Committee. 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.

Animal Care and Use Committee. 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. Total white blood cell counts, white blood cell differentials, red blood cell counts, platelet counts, hematocrit (Hct) values, total hemoglobin (Hgb) concentrations, mean cell volumes, mean corpuscular volumes and mean corpuscular Hgb concentrations were analyzed from blood collected in tubes containing EDTA using a Vetscan HM5 laser-based hematologic analyzer (Zoetis). Serum samples were tested for concentrations of albumin, amylase, ALT, AST, ALP, BUN, calcium, CRE, CRP, GGT, glucose, total protein and uric acid using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer disks (Abaxis).

es containing EDTA using a Vetscan HM5 laser-based hematologic analyzer (Zoetis). Serum samples were tested for concentrations of albumin, amylase, ALT, AST, ALP, BUN, calcium, CRE, CRP, GGT, glucose, total protein and uric acid using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer disks (Abaxis). RNA was isolated from blood and tissues of MARV-infected macaques as previously described17. One-Step Probe RT-qPCR kits (Qiagen) and CFX96 system/software (BioRad) were used to determine viral copies in samples. To detect MARV RNA, we targeted the MARV nucleoprotein (NP) gene with primer pairs and a 6-carboxyfluorescein (6FAM)–5′-CCCATAAGGTCACCCTCTT-3′–6 carboxytetramethylrhodamine (TAMRA) probe as previously described17. The limit of detection for this assay is 1,000 copies per ml. Data were analyzed using CFX Maestro 1.1 (Qiagen). Virus titration was performed by plaque assay using Vero E6 cells (ATCC CRL-1586) from all plasma samples as previously described17. The limit of detection for this assay was 25 PFU ml−1 for plasma and 250 PFU g−1 for tissues. Neutralization titers were calculated by determining the dilution of serum that reduced 50% of plaques (PRNT50) as previously described13.

RNA was isolated from blood and tissues of MARV-infected macaques as previously described17. One-Step Probe RT-qPCR kits (Qiagen) and CFX96 system/software (BioRad) were used to determine viral copies in samples. To detect MARV RNA, we targeted the MARV nucleoprotein (NP) gene with primer pairs and a 6-carboxyfluorescein (6FAM)–5′-CCCATAAGGTCACCCTCTT-3′–6 carboxytetramethylrhodamine (TAMRA) probe as previously described17. The limit of detection for this assay is 1,000 copies per ml. Data were analyzed using CFX Maestro 1.1 (Qiagen). Virus titration was performed by plaque assay using Vero E6 cells (ATCC CRL-1586) from all plasma samples as previously described17. The limit of detection for this assay was 25 PFU ml−1 for plasma and 250 PFU g−1 for tissues. Neutralization titers were calculated by determining the dilution of serum that reduced 50% of plaques (PRNT50) as previously described13. Necropsy was performed on all subjects. Tissue samples from major organs were collected from all subjects for histopathological and IHC examination, immersion fixed in 10% neutral buffered formalin and processed for histopathologic analysis as previously described17. The primary antibody (mouse anti-MARV NP) was used at 1:4,000 dilution. The secondary antibody (goat anti-mouse IgG) was used at 1:200 dilution. Micrographs from surviving animals depicted in Fig. 4 are representative of what was observed for the given tissue in all surviving animals, unless otherwise noted.

scribed17. The primary antibody (mouse anti-MARV NP) was used at 1:4,000 dilution. The secondary antibody (goat anti-mouse IgG) was used at 1:200 dilution. Micrographs from surviving animals depicted in Fig. 4 are representative of what was observed for the given tissue in all surviving animals, unless otherwise noted. Targeted transcriptomic analysis was performed on blood samples from macaques as previously described39. The NHPV2_Immunology reporter and capture probe sets (NanoString Technologies) were hybridized with 3 µl of each RNA sample for approximately 24 h at 65 °C. After hybridization, RNA complexes were loaded onto an nCounter microfluidics cartridge and analyzed using the NanoString nCounter SPRINT Profiler. Samples with binding densities above 2.0 or below 0.20 were reprocessed with either 1 µl or 5 µl of RNA to meet quality control (QC) standards. Three samples (C-5 0 DPI, TX-2 4 DPI and C-5 7 DPI) failed the normalization QC and were excluded from analysis.

dge and analyzed using the NanoString nCounter SPRINT Profiler. Samples with binding densities above 2.0 or below 0.20 were reprocessed with either 1 µl or 5 µl of RNA to meet quality control (QC) standards. Three samples (C-5 0 DPI, TX-2 4 DPI and C-5 7 DPI) failed the normalization QC and were excluded from analysis. nCounter.RCC files were imported into NanoString nSolver 4.0 for analysis. To adjust for RNA input variability and reaction efficiency, raw read counts were normalized using a panel of ten housekeeping genes along with positive and negative controls, following established methods40,41. nSolver’s Advanced Analysis module automatically selected optimal housekeeping genes by minimizing pairwise variation, using the geNorm algorithm in the Bioconductor package NormqPCR42. Differential expression analysis, heat maps and cell-type trend plots were generated with NanoString nSolver Advanced Analysis 2.0. Immune cell profiling was enhanced by adding human annotations to the respective mRNA in nSolver. Normalized data (fold-change- and Benjamini–Hochberg-adjusted P values) were exported as .xlsx files and further analyzed in GraphPad Prism 9.3.1 to produce transcript heat maps, available in supplementary material (Source Data Fig. 5 and Source Data Extended Data Fig. 1). For the heat maps, ODV-treated macaque samples were compared with controls at each timepoint (1–2, 4–5, 7 and 10 DPI or terminal timepoint for the controls), with the most upregulated and downregulated transcripts (Benjamini–Hochberg-adjusted P < 0.05, two tailed) sorted based on the 1–2 DPI rhesus ODV-treated survivor group. Cell-type trend plots were generated by comparing treated versus control samples across all timepoints. Differentially expressed transcripts (Benjamini–Hochberg-adjusted P < 0.10, two tailed) were further analyzed for pathway enrichment using the Canonical Signaling Pathways module within Ingenuity Pathway Analysis (Qiagen), highlighting top upregulated and downregulated pathways, sorted by the 10 DPI rhesus ODV-treated survivor group.

expressed transcripts (Benjamini–Hochberg-adjusted P < 0.10, two tailed) were further analyzed for pathway enrichment using the Canonical Signaling Pathways module within Ingenuity Pathway Analysis (Qiagen), highlighting top upregulated and downregulated pathways, sorted by the 10 DPI rhesus ODV-treated survivor group. Data collection and analysis were not performed blind to the conditions of the experiments. No animals or data points were excluded from analysis in this work, other than three samples for Nanostring targeted transcriptome profiling (C-5 0 DPI, TX-2 4 DPI and C-5 7 DPI), which failed the normalization QC and were excluded from analysis. Due to the small numbers used in this study, determination of the effects of sex on treatment success was not possible; however, we did attempt to ensure that a close-to-even distribution of each sex was represented in the experimental groups. The a priori power analysis used to determine group size is described in the ‘Nonhuman Primate Challenge and Treatment’ section of Methods. With regard to histopathological analysis of photomicrographs, representative photomicrographs were qualitatively considered to show lesions that were nominally or ordinally measured by masking of the pathologist after examination and ranking lesions to satiate the study objectives. In addition, thorough examinations of multiple slides of the target tissues multiple times (at least two times per tissues) were performed in a timely manner to maintain interpretation consistency, which comports with established criteria43. Additional data analysis and plotting were performed in Microsoft Excel (current Office 360 version) and/or GraphPad Prism v. 9.3.1 and/or GraphPad Prism v. 10.3.1.

imes (at least two times per tissues) were performed in a timely manner to maintain interpretation consistency, which comports with established criteria43. Additional data analysis and plotting were performed in Microsoft Excel (current Office 360 version) and/or GraphPad Prism v. 9.3.1 and/or GraphPad Prism v. 10.3.1. Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

The MARV Angola p2 seed stock originates from the 2004–2005 Uige, Angola, outbreak (DQ: 447653.1). The virus was isolated from the serum of a patient who died from MVD in 2005. The study challenge material was created by passaging the original isolate in Vero E6 cells (ATCC, CRL-1586). No mycoplasma or endotoxin could be detected (˂0.5 endotoxin units per ml). The ODV drug substance used for the NHP study was prepared using a synthetic route previously described and purified by crystallization as Form III19.

All study protocols described were approved by the University of Texas Medical Branch (UTMB) Institutional Animal Care and Use Committee, which were compliant with UTMB Institutional Biosafety Committee guidelines under biosafety level 4 containment. UTMB animal 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.

Before the study was conducted, a power analysis was performed to determine the minimum cohort size for the study. For the MARV ‘Angola isolate’, assuming a one-tailed alpha of 0.05, sample sizes of 5 per group will provide >80% power to detect a difference in the proportion of surviving animals between the treatment group (100% survival rate) and the control group (0% survival rate), using a Fisher’s exact test. One control animal per challenge period is used to confirm the lethality of the challenge material used in the treatment studies wherein clinical parameters are compared with those of historical control animals derived from other experiments using the identical challenge material (that is, same virus passage). This allows an ethical reduction in animal number use for data with highly predictable, lethal outcomes. Accordingly, six healthy captive-bred cynomolgus macaques (Macaca fascicularis) (Worldwide Primates) of Indonesian origin ~3 years of age and weighing ~2.1–3.5 kg were challenged by intramuscular injection with a target dose of 1,000 PFU of MARV Angola variant (actual dose, 988 PFU). Assignment to the treatment group or control group was determined before the challenge by randomization using Excel. Five animals were treated by oral gavage with 100 mg kg−1 ODV (as a suspension in a 0.5% methylcellulose in water vehicle) beginning 24 h after MARV exposure. These five animals received daily doses of ODV for 10 days (1–10 DPI). The MARV-positive control animal was treated in parallel with 0.5% methylcellulose in water. The duration of the NHP study was 35 days. All six macaques were monitored daily and scored for disease progression with an internal MARV humane endpoint scoring sheet approved by the UTMB Institutional Animal Care and Use Committee. 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.

Total white blood cell counts, white blood cell differentials, red blood cell counts, platelet counts, hematocrit (Hct) values, total hemoglobin (Hgb) concentrations, mean cell volumes, mean corpuscular volumes and mean corpuscular Hgb concentrations were analyzed from blood collected in tubes containing EDTA using a Vetscan HM5 laser-based hematologic analyzer (Zoetis). Serum samples were tested for concentrations of albumin, amylase, ALT, AST, ALP, BUN, calcium, CRE, CRP, GGT, glucose, total protein and uric acid using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer disks (Abaxis).

RNA was isolated from blood and tissues of MARV-infected macaques as previously described17. One-Step Probe RT-qPCR kits (Qiagen) and CFX96 system/software (BioRad) were used to determine viral copies in samples. To detect MARV RNA, we targeted the MARV nucleoprotein (NP) gene with primer pairs and a 6-carboxyfluorescein (6FAM)–5′-CCCATAAGGTCACCCTCTT-3′–6 carboxytetramethylrhodamine (TAMRA) probe as previously described17. The limit of detection for this assay is 1,000 copies per ml. Data were analyzed using CFX Maestro 1.1 (Qiagen). Virus titration was performed by plaque assay using Vero E6 cells (ATCC CRL-1586) from all plasma samples as previously described17. The limit of detection for this assay was 25 PFU ml−1 for plasma and 250 PFU g−1 for tissues.

Necropsy was performed on all subjects. Tissue samples from major organs were collected from all subjects for histopathological and IHC examination, immersion fixed in 10% neutral buffered formalin and processed for histopathologic analysis as previously described17. The primary antibody (mouse anti-MARV NP) was used at 1:4,000 dilution. The secondary antibody (goat anti-mouse IgG) was used at 1:200 dilution. Micrographs from surviving animals depicted in Fig. 4 are representative of what was observed for the given tissue in all surviving animals, unless otherwise noted.

Targeted transcriptomic analysis was performed on blood samples from macaques as previously described39. The NHPV2_Immunology reporter and capture probe sets (NanoString Technologies) were hybridized with 3 µl of each RNA sample for approximately 24 h at 65 °C. After hybridization, RNA complexes were loaded onto an nCounter microfluidics cartridge and analyzed using the NanoString nCounter SPRINT Profiler. Samples with binding densities above 2.0 or below 0.20 were reprocessed with either 1 µl or 5 µl of RNA to meet quality control (QC) standards. Three samples (C-5 0 DPI, TX-2 4 DPI and C-5 7 DPI) failed the normalization QC and were excluded from analysis.

nCounter.RCC files were imported into NanoString nSolver 4.0 for analysis. To adjust for RNA input variability and reaction efficiency, raw read counts were normalized using a panel of ten housekeeping genes along with positive and negative controls, following established methods40,41. nSolver’s Advanced Analysis module automatically selected optimal housekeeping genes by minimizing pairwise variation, using the geNorm algorithm in the Bioconductor package NormqPCR42. Differential expression analysis, heat maps and cell-type trend plots were generated with NanoString nSolver Advanced Analysis 2.0. Immune cell profiling was enhanced by adding human annotations to the respective mRNA in nSolver. Normalized data (fold-change- and Benjamini–Hochberg-adjusted P values) were exported as .xlsx files and further analyzed in GraphPad Prism 9.3.1 to produce transcript heat maps, available in supplementary material (Source Data Fig. 5 and Source Data Extended Data Fig. 1). For the heat maps, ODV-treated macaque samples were compared with controls at each timepoint (1–2, 4–5, 7 and 10 DPI or terminal timepoint for the controls), with the most upregulated and downregulated transcripts (Benjamini–Hochberg-adjusted P < 0.05, two tailed) sorted based on the 1–2 DPI rhesus ODV-treated survivor group. Cell-type trend plots were generated by comparing treated versus control samples across all timepoints. Differentially expressed transcripts (Benjamini–Hochberg-adjusted P < 0.10, two tailed) were further analyzed for pathway enrichment using the Canonical Signaling Pathways module within Ingenuity Pathway Analysis (Qiagen), highlighting top upregulated and downregulated pathways, sorted by the 10 DPI rhesus ODV-treated survivor group.

Data collection and analysis were not performed blind to the conditions of the experiments. No animals or data points were excluded from analysis in this work, other than three samples for Nanostring targeted transcriptome profiling (C-5 0 DPI, TX-2 4 DPI and C-5 7 DPI), which failed the normalization QC and were excluded from analysis. Due to the small numbers used in this study, determination of the effects of sex on treatment success was not possible; however, we did attempt to ensure that a close-to-even distribution of each sex was represented in the experimental groups. The a priori power analysis used to determine group size is described in the ‘Nonhuman Primate Challenge and Treatment’ section of Methods. With regard to histopathological analysis of photomicrographs, representative photomicrographs were qualitatively considered to show lesions that were nominally or ordinally measured by masking of the pathologist after examination and ranking lesions to satiate the study objectives. In addition, thorough examinations of multiple slides of the target tissues multiple times (at least two times per tissues) were performed in a timely manner to maintain interpretation consistency, which comports with established criteria43. Additional data analysis and plotting were performed in Microsoft Excel (current Office 360 version) and/or GraphPad Prism v. 9.3.1 and/or GraphPad Prism v. 10.3.1.

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41591-025-03496-y.

Source Data Fig. 1Plaque assay and RT-qPCR data for quantification of viral load. Source Data Fig. 3PRNT data. Source Data Fig. 5The −log10(P values) and log2(fold change) data from Nanostring targeted transcriptome profiling. Source Data Extended Data Fig. 1The z-scores derived from P values for pathway enrichment analysis from Nanostring targeted transcriptome profiling data. Plaque assay and RT-qPCR data for quantification of viral load. PRNT data. The −log10(P values) and log2(fold change) data from Nanostring targeted transcriptome profiling. The z-scores derived from P values for pathway enrichment analysis from Nanostring targeted transcriptome profiling data.