Natural history of Sudan ebolavirus infection in rhesus and cynomolgus macaques

ABSTRACT Due to its high mortality rate and continued re-emergence, Ebolavirus disease (EVD) continues to pose a serious threat to global health. A group of viruses within the genus Ebolavirus causes this severe hemorrhagic disease in humans: Ebola virus (EBOV; species Zaire ebolavirus), Sudan virus (SUDV; species Sudan ebolavirus), Bundibugyo virus, and Taï Forest virus. EBOV and SUDV are associated with the highest case fatality rates. While the host response to EBOV has been comprehensively examined, limited data exists for SUDV infection. For medical countermeasure testing, well-characterized SUDV nonhuman primate (NHP) models are thus needed. Here, we describe a natural history study in which rhesus (N = 11) and cynomolgus macaques (N = 14) were intramuscularly exposed to a 1000 plaque-forming unit dose of SUDV (Gulu variant). Time-course analyses of various hematological, pathological, serological, coagulation, and transcriptomic findings are reported. SUDV infection was uniformly lethal in cynomolgus macaques (100% mortality), whereas a single rhesus macaque subject (91% mortality) survived to the study endpoint (median time-to-death of ∼8.0 and ∼8.5 days in cynomolgus and rhesus macaques, respectively). Infected macaques exhibited hallmark features of human EVD. The early stage was typified by viremia, granulocytosis, lymphopenia, albuminemia, thrombocytopenia, and decreased expression of HLA-class transcripts. At mid-to-late disease, animals developed fever and petechial rashes, and expressed high levels of pro-inflammatory mediators, pro-thrombotic factors, and markers indicative of liver and kidney injury. End-stage disease was characterized by shock and multi-organ failure. In summary, macaques recapitulate human SUDV disease, supporting these models for use in the development of vaccines and therapeutics.


Introduction
The Ebolavirus genus (family Filoviridae) is comprised of six recognized species: Reston ebolavirus, Bombali ebolavirus, Zaire ebolavirus, Sudan ebolavirus, Bundibugyo ebolavirus, and Taï Forest ebolavirus [1]. African fruit bats are the purported natural reservoirs of these filamentous, enveloped, non-segmented, negative-sense RNA viruses [2]. Reston virus (RESTV; species Reston ebolavirus) causes disease in nonhuman primates and pigs, but no cases have been reported in people [3][4][5]. In 2018, Bombali virus (BOMV; species Bombali ebolavirus) was discovered in bats in Sierra Leone [2], but it is still unclear whether the virus is pathogenic in humans or animals [6]. The other filovirus species comprise a group of viruses that are important human pathogens with case-fatality rates ranging from 50% to 90% for Ebola virus (EBOV; species Zaire ebolavirus), ∼ 55% for Sudan virus (SUDV; species Sudan ebolavirus), and ∼25-50% for Bundibugyo virus (BDBV; species Bundibugyo ebolavirus) [1,[7][8][9]. Taï Forest virus (TAFV; species Taï Forest ebolavirus) was the culprit of a single nonfatal case of Ebola virus disease (EVD) [1,10]. SUDV and EBOV have been responsible for most cases of EVD and were the first ebolaviruses discovered. These viruses appeared to emerge simultaneously in the summer of 1976 in the Sudanese towns of Nzara and Maridi and Yambuku, Democratic Republic of Congo (formerly Zaire), respectively [7], and ostensibly spread independently to people in each of the affected areas. Emerging filovirus species causing outbreaks in Central and West Africa are hypothesized to coincide with the proportion of seropositivity to each filovirus species in fruit bats or other potential reservoirs. For example, filoviruses causing outbreaks in West and Central Africa between 2005 and 2012 shifted from Zaire ebolaviruses to Sudan and Bundibugyo ebolaviruses, concurrent with a change in the serologically dominant virus species in bats [11]. Further investigation of SUDV in animal and human hosts is clearly needed given its historic serodominance in bats, high lethality, and high potential for future spillover into human populations.
In humans and preclinical animal models, the host response to EBOV is well-characterized, whereas limited data is available for SUDV infection. In rodents (mice, guinea pigs, and hamsters), adaptation of filoviruses is necessary to cause lethal disease in immunocompetent hosts [12,13]. Typically, rodent models do not exhibit the disordered coagulopathy seen in human EVD cases, e.g. pro-thrombotic changes or extensive fibrin deposition. Ferrets are susceptible to wild-type SUDV and exhibit some of the coagulopathies associated with human infection [14,15], but few immunological reagents are available to study host responses. Nonhuman primates (NHPs), particularly macaque species, are considered the "gold standard" animal model for filovirus infection as they recapitulate most human manifestations of EVD including the development of disseminated intravascular coagulation and a maculopapular rash [16]. Macaques are highly susceptible to SUDV without the need for virus adaptation [17], and ample immunological resources are available to study the immune response.
To date, only an aerosol model of SUDV infection has been characterized in an NHP model [17]. Here, we describe a detailed natural history study of cynomolgus macaques (CM) and rhesus macaques (RM) exposed to 1000 plaque-forming units of SUDV via the intramuscular (i.m.) route. The i.m. model is important as it represents the most lethal route of infection (e.g. needlestick) and is the standard for evaluating medical countermeasures as well as serving as a comparison for other filovirus models (i.e. there are limited mucosal challenge data available). Virology and various hematological, pathological, serological, coagulation, and transcriptomic findings are reported. Our results shed light on the pathophysiology of SUDV infection and highlight the utility of NHPs for testing medical countermeasures against this deadly pathogen.

Ethics statement
Animal studies were conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animal experimentation. Details are provided in the Supplemental Methods.

Challenge virus
The SUDV seed stock (Gulu variant) originates from the serum of a fatal patient (

Animal challenge
Fourteen cynomolgus (Macaca fascicularis) [18][19][20] and eleven rhesus (Macaca mulatta) [21][22][23][24] macaques of Chinese origin (PreLabs, Worldwide Primates) that served as virus positive controls from 17 studies at the Galveston National Laboratory (GNL) were employed for this project. Results from the remaining seven cynomolgus and three rhesus macaques have not been published. Details are provided in the Supplemental Methods.

Blood collection
Blood was collected by venipuncture into EDTA and serum tubes throughout the course of the study. Details are provided in the Supplemental Methods.

Hematology and clinical chemistry
Blood samples were analyzed using a laser-based hematologic analyzer and serum samples were tested using a Piccolo point-of-care analyzer and Biochemistry Panel Plus analyzer discs (Abaxis). Details are provided in the Supplemental Methods.

Viral load determination
SUDV viral loads were determined using RT-qPCR and standard plaque assays. Details are provided in the Supplemental Methods.

Transcriptomics
Targeted transcriptomics was performed on macaque whole blood as previously described [25]. Immune cell profiling was accomplished via CIBERSORT webbased deconvolution software [26] using the LM22 signature matrix file. The 6-way Venn diagram was created with InteractiVenn [27]. Details are provided in the Supplemental Methods.

Histopathology and immunohistochemistry
Tissue samples of all major organs were harvested in 10% neutral buffered formalin for histopathologic and immunohistochemical (IHC) examination. Slides were reviewed by a board-certified veterinary pathologist. Details are provided in the Supplemental Methods.

Statistical analysis
Statistical comparisons were carried out in GraphPad Prism. Details are provided in the Supplemental Methods.

Results
Eleven rhesus macaques (RM) and fourteen cynomolgus macaques (CM) were i.m. challenged with a 1000 PFU target dose of SUDV-Gulu. The median time-todeath (TTD) for CM and RM was ∼8.0 (mean TTD 7.64 ± 1.72) and ∼8.5 days (mean TTD 8.3 ± 1.27), respectively. Disease was uniformly lethal in CM. A single RM subject survived to the 28-day post infection (DPI) study endpoint, corresponding to a 91% mortality rate within this group (Figure 1(A)). No statistically significant difference in survival was noted between the two species (log-rank test). Viral loads were assessed by performing RT-qPCR amplification and conventional plaque assays on whole blood and plasma samples, respectively. Viral RNA (vRNA) (Figure 1(B)) and infectious SUDV particles ( Figure  1(C)) were first detected in both species at ∼ 4 DPI. Peak vRNA titres ranged from 7.2 to 12.4 log 10 copies/ml, whereas infectious titres ranged from 5.2 to 8.6 log 10 PFU/ml. CM compared to RM tended to have higher titres at 6-8 DPI, although no statistically significant difference was noted between the two and rhesus macaques (black; N = 11) i.m. exposed to 1000 PFU of SUDV-Gulu. Macaque icons were created with BioRender (https://biorender.com/). (B) Viral loads were measured by RT-qPCR in whole blood and reported as log10 copies/ml at the denoted time points. The limit of detection for this assay is 1000 copies/ml (indicated by a dotted horizontal line). (C) Plasma viremia was measured by standard plaque assay at the denoted time points and reported as log10 PFU/ml. The limit of detection for this assay is 25 PFU/ml (indicated by a dotted horizontal line). For (B) and (C), each bar represents the average titre ± SEM for each cohort. Individual subjects are represented by circles. Abbreviations: SUDV, Sudan virus; i.m., intramuscular; DPI, days post infection; PFU, plaque-forming units.
cohorts for this parameter at any time point (Mann-Whitney t-tests). The single rhesus survivor had a peak titre of 5.1 log 10 PFU/ml at 5 DPI, but no infectious virus was detected in this subject by 11 DPI. SUDV-exposed macaques displayed hallmark features of human EVD (Figure 2, Tables S1 and S2). Early common clinical signs (3-4 DPI) included decreased food consumption, fever, lymphopenia, monocytosis, and granulocytosis. At mid-to-late disease (5-6 DPI), RM and CM exhibited anorexia, thrombocytopenia, hypoalbuminemia, hypoamylasemia, and a petechial rash. Liver enzymes (ALT, AST, ALP, GGT), markers indicative of kidney injury (BUN, CRE), and C-reactive protein (CRP; general inflammation marker) were elevated in all subjects at this stage. Preagonal disease (7-10 DPI) was characterized by weakness, lateral recumbency, and decreased body core temperature.
Postmortem gross examination of animals at necropsy revealed one or more lesions consistent with EVD including petechial to ecchymotic rash, necrotizing hepatitis (characterized as hepatic pallor with reticulation), splenomegaly, lymphadenitis, hemorrhagic interstitial pneumonia, and gastrointestinal ulceration ( Figure 3 Microscopically, tissue sections were consistent with EVD induced by SUDV infection. Predominant microscopic findings consisted of mixed inflammatory infiltrates largely composed of mononuclear cells that were IHC positive for SUDV antigen and widespread throughout multiple organs (Figure 3 (B, G, J, K)). Inflammation was often accompanied with hemorrhage fibrin deposition and necrosis. Prominent fibrin deposition was noted particularly in highly vascularized organs such as the spleen and liver (Figure 3(D, I, M)). Conversely, the survivor lacked inflammatory findings, fibrin deposition, and immunolabeling for SUDV antigen (Figure 3 (P, Q)).
Functional enrichment of transcripts was executed to determine canonical signalling pathways and upstream regulators associated with SUDV infection. This analysis supported activation of pathways involved in cytokine/chemokine signalling, viral sensing, and antiviral immunity ( Figure 5(C)). To capture shifts in circulating cell populations, we performed digital cell quantitation via transcriptional profiling ( Figure 5(D)). SUDV infection of RM and CM was associated with predicted recruitment of monocyte, M2 macrophages, and various granulocyte populations (mast cells, eosinophils, and neutrophils), in accordance with our hematology results. Disease and function terms predominantly mapped to terms involved in myeloid cell phagocytosis ( Figure S1B). The topmost upstream regulators included LPS, poly IC-RNA, TNF, IFNG, and IL1B, factors that have been previously identified for other ebolaviruses ( Figure S1C) [25]. A pharmacological inhibitor of p38 mitogen-activated protein kinase (MAPK), SB203580 [33], was conversely predicted to significantly decrease. Congruent with our serum biochemistry results, predicted tox functions included terms associated with liver and kidney injury, e.g. "apoptosis of kidney cells," "increased levels of ALT," "increased levels of BUN," and "acute renal failure" (Figure S1D). These transcriptional changes were evident by early disease.
Lastly, we measured protein levels of plasmaderived inflammatory mediator and thrombosis analytes. Like EBOV, SUDV disease was associated with elevated IL-6, IP-10, and MCP-1 concentrations ( Figure 6, Figure S2) [31,32]. Of the 15 analytes that had a statistically significant association, 4 were cytokines or chemokines, and 3 of those -IL-10, IP-10, and RANTESwere associated with an age-dependent survival outcome. tPA levels were also increased, whereas plasminogen and Factor XIII were decreased, possibly due to consumptive coagulopathy.
Overall temporal changes in SUDV-infected NHPs are summarized in Figure 7. Specifically, clinical signs, viral load, pathology, transcriptomic findings, and plasma cytokine/chemokine and thrombosis analyte profiles at each stage of disease are highlighted. Figure 7. Summary of the most salient features of disease in SUDV-exposed cynomolgus and rhesus macaques. Infected macaques and humans share many disease features including perturbations in liver enzymes (ALT, AST) and kidney function markers (BUN, CRE) along with hemorrhagic manifestations such as a petechial rash and hematochezia. Tissue factor, tPA, and PAI-1 were elevated in infected macaques, which may play a role in disseminated intravascular coagulation, fibrin deposition, and an overactive endothelium, respectively. Transcriptional biomarkers were detectable at the early disease stage followed by shifts in plasma inflammatory and thrombosis mediators at the mid disease stage. The rhesus macaque survivor exhibited only transient shifts in liver and kidney markers, and this subject had a reduced viral load and delayed disease onset in the absence of hemorrhagic manifestations. Survival was also associated with increasing levels of anti-SUDV IgM by the late disease stage followed by increasing levels of anti-SUDV IgG in the convalescent phase.

Discussion
The continued re-emergence of ebolaviruses emphasizes the need for animal models that accurately recapitulate human EVD, particularly for less characterized members of the genus such as SUDV, BDBV, TAFV, RESTV, and BOMV. While research focus has been placed on the development of vaccines and treatments against SUDV, there remains a paucity of clinical and preclinical studies to describe the natural progression and pathophysiology of infection with this virus. Here we assessed the pathogenic potential of SUDV infection in CM and RM by i.m. exposure.
Infection with a ∼ 1000 PFU dose of SUDV was 100% lethal in CM and near uniformly lethal in RM (91% mortality rate), with a mean TTD of 7.64 ± 1.72 and 8.3 ± 1.27, respectively. In comparison, the mean TTD for the largest outbreak of SUDV in the Gulu district of Uganda was 8 days after onset of symptoms with a case fatality rate of 53% [34]. The high challenge dose used and i.m. route of infection in this study likely hastened disease progression in NHPs as these factors are known to accelerate the course of EVD [16]. Inoculation via the i.m. route is intended to mimic a worstcase scenario involving an accidental needle stick of either a healthcare worker performing procedures on an infected patient or a laboratory staff member performing procedures on an infected animal. Understanding the disease kinetics of this route of infection is important for identifying the prophylactic window for vaccines and treatments. The clinical pathology and progression of disease were analogous between the two macaque species. These results suggest both CM and RM serve as suitable animal models for SUDV infection. Although we observed earlier detection of fever, rash, and hunched posture in some RM subjects, the disease course appeared slightly faster and more lethal in CM than RM, suggesting the latter model may be more appropriate for treatment studies to evaluate therapeutic effect more sensitively [16].
Immune dysregulation is a hallmark of EVD [1]. Some key features include hypercytokinemia, hyperchemokinemia, and a failed or delayed adaptive response [25]. Infection of NHPs with SUDV led to early and dramatic upregulation of transcripts encoding IP-10, IL-6, MCP-1, pro-inflammatory mediators that were previously reported to correlate with lethality in humans [32,35]. By mid-disease, we detected plasma secretion of these inflammatory proteins. IP-10 and MCP-1 are powerful chemoattractants for monocytes/macrophages and dendritic cells, whereas IL-6 is a molecular control involved in differentiation of monocytes to macrophages [36]. As monocytes, macrophages, and dendritic cells are early and preferred sites of filovirus replication [37], secretion of these cytokines and chemokines likely serves to recruit more cells to the site of infection, thereby promoting dissemination of the virus. This reasoning is in agreement with our hematology and histopathology results.
To identify corelates of natural protection against SUDV infection, we conducted a targeted assessment of the whole blood transcriptome in the single RM survivor. Our results demonstrated the survivor expressed lower expression of granulocyte markers (CEACAM8, CEACAM1, CEACAM3) and higher levels of major histocompatibility complex class II transcripts (e.g. HLA-DRA, HLA-DMB, HLA-DMA, CD74). Thus, antigen presentation and formation of an adaptive response is implicated in resistance to SUDV infection, whereas prolonged innate signalling correlates with lethality. This hypothesis is supported by the induction of SUDV-specific IgM and IgG and transcriptional evidence of B-and T-cell activation (CD79A, CD79B, TBX21) in the RM survivor but not fatal subjects. We and others have shown host MHC class II proteins are strikingly downregulated on monocytes following filovirus exposure [38][39][40] independent of direct infection of these cells [41]. The cytokine milieu, release of immature cells as a consequence of emergency myelopoiesis, or recruitment of myeloid-derived suppressor cells might explain this phenomenon, although the mechanism of MHC class II downregulation is still undetermined [25,41]. Reduced monocyte antigen presentation may contribute to the lack of an adaptive response and ultimate inability to clear the virus.
Our analyses indicate macaques and humans infected with SUDV share many disease features. Coagulopathy and acute liver and kidney injury are significantly associated with severe cases of EVD [1,42,43]. Accordingly, perturbations in liver enzymes (ALT, AST, ALP, GGT) and kidney function markers (BUN, CRE) were prominent findings in fatal macaques along with hemorrhagic manifestations such as petechial rash, epistaxis, and hematochezia. Tissue factor and PAI-1 were also elevated in infected macaques, which may play a role in disseminated intravascular coagulation and an overactive endothelium, respectively [31,44]. Importantly, the single RM survivor exhibited only transient shifts in liver and kidney markers, and this subject had a reduced viral load and delayed disease onset in the absence of hemorrhagic manifestations. While few postmortem examinations in humans with SUDV disease have been carried out, notable findings in two patients resembled those of an acute viral infection with some, but not exclusively, defining EVD features [7]. Some mutual histological findings between SUDV-infected humans and NHPs included focal eosinophilic necrosis of the liver, renal tubular necrosis, massive deposition of fibrin in tissues, and extensive lymphocyte depletion.
In conclusion, this natural history study has expanded our understanding of the pathogenesis of SUDV infection and has enhanced our understanding of immunological factors that mediate natural protection. Our results revealed many similarities between NHP and human SUDV-induced EVD, suggesting the reliability of CM and RM models for medical countermeasure development to fight endemic disease and for biodefense purposes.