Comparative pathogenesis of three Mayaro virus genotypes in the cynomolgus macaque

Abstract

Mayaro virus (MAYV), a mosquito-borne alphavirus, is considered an emerging threat to public health with epidemic potential. Phylogenetic studies show the existence of three MAYV genotypes. In this study, we provide a preliminary analysis of the pathogenesis of all three MAYV genotypes in cynomolgus macaques (Macaca facicularis, Mauritian origin). Significant MAYV-specific RNAemia and viremia were detected during acute infection in animals challenged intravenously with the three MAYV genotypes, and strong neutralizing antibody responses were observed. MAYV RNA was detected at high levels in lymphoid tissues, joint muscle and synovia over 1 month after infection, suggesting that this model could serve as a promising tool in studying MAYV-induced chronic arthralgia, which can persist for years. Significant leucopenia was observed across all MAYV genotypes, peaking with RNAemia. Notable differences in the severity of acute RNAemia and composition of cytokine responses were observed among the three MAYV genotypes. Our model showed no outward signs of clinical disease, but several major endpoints for future MAYV pathology and intervention studies are described. Disruptions to normal blood cell counts and cytokine responses were markedly distinct from those observed in macaque models of CHIKV infection, underlining the importance of developing non-human primate models specific to MAYV infection.

Data summary

The authors confirm all supporting data, code and protocols have been provided within the article or through supplementary data files.

Introduction

Mayaro virus (MAYV) is an emerging mosquito-borne arboviral alphavirus endemic to South and Central America that can cause debilitating febrile and rheumatic disease in humans [1,4]. Since the initial isolation of MAYV in 1954 in Trinidad and Tobago [5], MAYV outbreaks have been reported in multiple countries within Latin America and the Caribbean [6,10]. Genetic characterization of MAYV isolates suggests the existence of three genotypes (genotypes D: dispersed, L: limited, and N: new) that vary primarily in their geographic distribution [10,12]. MAYV is naturally transmitted in a rural/sylvatic cycle between canopy-dwelling mosquito vectors of the Haemagogus genus and non-human primate (NHP) hosts [13,14]. However, MAYV has also been detected in several urban-adapted mosquito species, including those within the Culex and Aedes genera [15,18], both of which are widely distributed throughout the Americas. Vector competence studies among some of these species, as well as several North American Anopheles species, report that these mosquitoes are capable of transmitting MAYV [19,24]. Notably, the large outbreaks of similar arboviruses like Chikungunya virus (CHIKV) and Zika virus (ZIKV) in recent decades have been ascribed, at least in part, to viral adaptations to urbanized mosquito species [25,28]. The availability of a variety of potential MAYV vector species combined with the expansion of their geographic ranges due to trade and climate change [29,31] has caused increasing concern that MAYV could become a global threat to public health.

MAYV is unique among the New World alphaviruses as it presents with a hallmark characteristic symptom of Old World alphaviruses: arthralgia [30]. This arthralgia can be significantly debilitating and may persist, or even reoccur, for months to years following infection [31]. There is evidence that the arthralgic persistence is a byproduct of the continuous expression of pro-inflammatory cytokines that initiates with MAYV infection [3]. Notably, these pro-inflammatory cytokines differ from those observed in CHIKV models [32], suggesting that CHIKV models may serve as less useful surrogates for MAYV models than previously assumed.

Until recently, MAYV pathogenesis models were limited to murine strains including BALB/c, IFNαR^-/-, IFNα/βR^-/- (A129), IFNα/β/γR^-/- (AG129), and C57BL/6. These models have been used in prophylactic, therapeutic, and transmission studies with adequate reliability and reproducibility [1,33,46]. However, as countermeasures advance to potential licensure, it is important to evaluate safety and efficacy in more clinically relevant animal models. Non-human primate (NHP) models are preferable to murine models given the greater physiological and immunological homology between NHPs and humans, as well as the often-variable pathogenesis observed between murine and NHP models. NHP models for CHIKV were established as early as the mid-twentieth century when it was reported that rhesus macaques infected with viremic human sera developed neutralizing antibodies responses against CHIKV [47]. More recently, CHIKV infections have been demonstrated in pregnant and/or various aged rhesus macaques [48] and in cynomolgus macaques [45]. CHIKV pathogenesis studies in NHPs have focused on both acute stages of infection (i.e. 1–6 days post-infection) and convalescent stages (i.e. 7–34 days post-infection) with major emphasis on virus detection, innate response, cytokine/chemokine production, and virus-specific immunity [34]. While these studies offer a relevant reference for the development of appropriate MAYV pathogenesis models in NHPs, it is imperative to develop NHP models that are specific to MAYV to facilitate quantitative evaluations of therapeutic and/or prophylactic potency against MAYV-induced disease. Marmosets (Callithrix argentata and Callithrix humeralifer) experimentally infected with MAYV have been shown to develop transient and moderate viremia [49]; however, practical concerns (i.e. reagent availability) have precluded the use of these models in Mayaro virus studies to date [47]. Within the last year, a rhesus macaque model of MAYV infection was shown to recapitulate many aspects of human infection [48]. The small size of this study, however, speaks to the current pressure on rhesus macaque stocks and underlines the importance of establishing multiple NHP models of MAYV infection that are accessible to a variety of researchers [50]. Additionally, understanding the differences in pathogenesis among MAYV genotypes is important for selecting appropriate challenge viruses used in efficacy studies, and for developing vaccines and therapeutics that will have broad clinical applicability. Here we characterize the pathophysiology of cynomolgus macaques (Macaca fascicularis, Mauritian origin) infected with representative viruses of the three MAYV genotypes (N, L, and D) to establish a novel experimental MAYV-specific NHP model. We show that cynomolgus macaques are susceptible to MAYV infection, which generates significant RNAemia peaking 2 days post-infection (dpi). We demonstrate virus amplification in a variety of organs and tissues, with high levels of MAYV RNA in lymphoid and joint tissues up to 1 month after infection. While MAYV-infected cynomolgus macaques did not present with overt clinical signs of illness (e.g. fever), a strong neutralizing antibody (nAb) response was observed. Significant leucopenia in MAYV-infected macaques as a result of the depletion of circulating neutrophil, lymphocyte, and eosinophil populations was observed concomitant with RNAemia. All MAYV-infected animals mounted a robust cytokine response in the first 4 days of infection; notably, these differed among the three MAYV genotypes assessed here, and from published data characterizing CHIKV cytokine responses in the same animal model [32].

Altogether, our studies establish cynomolgus macaques as a suitable NHP model for studying MAYV pathogenesis and assessing MAYV countermeasures.

Methods

Animals

The 12 adult (5–8 years old) female cynomolgus macaques (Macaca fascicularis, Mauritian origin) used in this study were housed in ABSL-2 facilities at BIOQUAL, Inc. (BIOQUAL) in Rockville, MD. BIOQUAL facilities are accredited by the Association for Accreditation and Assessment of Laboratory Animal Care (AALAC) International, and all study procedures involving animals were conducted in accordance with the National Research Council (NRC) Guide for the Care and Use of Laboratory Animals, 8th Edition and the Animal Welfare Act. The study was reviewed and approved by the BIOQUAL Institutional Animal Care and Use Committee (IACUC) under Protocol #18–016P. Animals were recycled from BIOQUAL immunogenicity studies evaluating influenza or Lyme disease interventions. Serum samples from each animal were shipped to VRL Laboratories (Dan Antonio, TX) for ELISA prescreening for the following viral infections: herpes B virus; SRV-1,–2, -3,–4, and −5; SIV, STLV-1; all animals tested seronegative. Additionally, all animals tested negative for active ZIKV infection via PCR. Animals were housed singly in open metal caging units that allowed visual inspection and prevented contact with other animals in the room. At the conclusion of the study, all animals were euthanized according to the recommendations of the American Veterinary Medical Association (AVMA) 2013 Panel on Euthanasia.

Viruses

The MAYV strains used in this study were obtained from the World Reference Centre for Emerging Viruses and Arboviruses (WRCEVA) at the University of Texas Medical Branch (Galveston, TX). Contemporaneous virus strains representing the N, L, and D genotypes of MAYV with minimal passage histories (hereafter referred to as MAYV N, MAYV L, and MAYV D, respectively) were preferentially selected for inclusion in the study. The MAYV N strain FMD 3213 (GenBank Accession # KP842812) is a human isolate from Madre de Dios, Peru in 2010 [10]. The MAYV L strain BeAr505411 (GenBank Accession # DQ487382.1) is a mosquito isolate from Pará, Brazil in 1991 [10]. The MAYV D strain INHRR 12 a (GenBank Accession # KM400594) is a human isolate from an outbreak in Portuguesa, Venezuela in 2010 [10].

MAYV strains were propagated in Vero-76 cells (African green monkey kidney cells; CRL-1587 ATCC, Manassas, VA). Briefly, cells were infected with a multiplicity of infection (MOI) of 0.1 and virus was harvested when >75 % of cells demonstrated cytopathic effects (i.e. between 36–48 h post-infection). The cellular supernatant was collected and clarified by centrifugation at 4 °C before storage. Virus stocks were titrated by Vero cell plaque assay. Virus titres for the N, L, and D strains were 8.0×10⁷ plaque forming units (PFU) ml⁻¹, 1.0×10⁸ PFU ml⁻¹, and 1.0×10⁷ PFU ml⁻¹, respectively.

Study design and virus challenge

Twelve adult female cynomolgus macaques (Macaca fascicularis) were randomized into four groups (three animals per group) for challenge with one of the three MAYV genotypes (N, L, or D) or a phosphate-buffered saline (PBS) control. Before any observations or procedures requiring handling, animals were sedated with an intramuscular injection of ketamine HCl (10–25 mg kg⁻¹). A veterinarian conducted a physical examination of each animal prior to the initiation of the study to ensure the animals were in good health. Animals were challenged with 10⁶ PFU virus via the saphenous vein. Clinical observations were recorded twice daily throughout the study. Rectal body temperature and weights were recorded during each physical exam. Body temperatures were also monitored continuously via transponders implanted in the quadriceps muscle. Blood samples were collected 1–4, 7, 11, and 14 days post-infection (dpi) from sedated animals into serum separator (SST) tubes (for post-study analyses of sera) and tubes containing ethylenediaminetetraacetic acid (EDTA) (for post-study analyses of whole blood and plasma). Approximately 1 month post-infection (31–33 dpi), all animals were euthanized in accordance with AVMA guidelines and necropsied for the following tissues: axial, inguinal, mesenteric, and mandibular lymph nodes (LNs), spleen, small intestine, knee joint muscle and synovial fluid (synovia), lung, liver, kidney, vagina, uterus, ovary, brain, and spinal cord.

Viral RNA quantification via RT-qPCR

Total RNA was extracted from plasma using a QIAamp MinElute Virus Kit (Qiagen) and from homogenized tissues using a standard phenol/chloroform extraction. A TaqMan RNA real-time quantitative PCR (RT-qPCR) assay (Applied Biosystems, Foster City, CA) was performed with a TaqMan probe (5′ FAM CATAGACATCCTGATAGACTGCCACC-BHQ1 3′) and primers (forward: 5′ CCTTCACACAGATCAGAC 3′ and reverse: 5′ GCCTGGAAGTACAAAGAA 3′) targeting a region of the MAYV genome that encodes the nonstructural protein nsP1. The RT-qPCR was carried out using a 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) and One-Step RT-PCR Master Mix (Applied Biosystems, Foster City, CA) with the following cycling conditions: 48 °C for 30 min (reverse transcription), 95 °C for 10 min (polymerase activation), and 40 cycles of 95 °C for 15 s followed by 60°C for 1 min (amplification). The signal was compared to a standard curve of known concentrations ranging from 10⁷ to one copy (i.e. the linear range of concentration/signal relation spans 8 Logs). All samples were quantified in triplicate for consistency and accuracy. MAYV challenge virus controls of known PFU ml⁻¹ and viral RNA samples were extracted and tested in parallel and under the same conditions.

Biopsied tissues were evaluated for MAYV RNA via a high-sensitivity semiquantitative RT-PCR assay normalized to the expression of the housekeeping gene GAPDH. RNA was extracted from 6 ml of tissue homogenate using a standard phenol/chloroform extraction with the addition of tRNA prior to precipitation to improve RNA recovery. Relative quantitative RT-PCR was performed using MAYV and GAPDH primers and probes simultaneously, as previously described [48]. The inter-assay variability of this method was 23.4 %; intra-assay variability was 0.3Log, yielding a sensitivity threshold of two viral RNA copies ml⁻¹.

Infectious viral titer determination via plaque assay

Viremic titres in sera were determined by plaque assay on Vero cells. Briefly, confluent monolayers of Vero-76 cells seeded in 6-well plates were infected with seven tenfold serial dilutions of serum samples. Cells were incubated for 1 h while rocking every 10 min to ensure sufficient viral adsorption. A 0.4 % agarose overlay supplemented with growth medium containing 2 % fetal bovine serum (FBS) and 1 % penicillin/streptomycin was added. Infected cells were incubated for 2 days at 37 °C with 5 % CO₂. Following incubation, cells were fixed with 10 % formaldehyde solution, the agarose overlay was removed, and cells were stained with 2 % crystal violet dissolved in deionized water. The number of plaques was counted and expressed as plaque forming units per millilitre (PFU ml⁻¹). All samples without a detectable titre, or a titre below the lower limit of detection (LLOD), were substituted with a constant value representative of half the LLOD (i.e. 1 Log) for statistical comparisons.

Evaluation of neutralizing antibodies by PRNT₈₀

Neutralizing antibody (nAb) activity in serum was assayed using a plaque reduction neutralization test (PRNT) performed using standard methods [51]. For this assay, nAb titres were calculated as the highest dilution of serum that inhibited 80 % of plaque formation (PRNT₈₀); samples with reciprocal PRNT₈₀ titres≥20 were considered positive. The lower and upper limits of detection were 20 and 2560, respectively. For statistical comparisons, all data points that resulted in values below the LLOD were substituted with a value representative of half the limit of detection (i.e. 10).

Hematology and serum biochemistry analyses

Whole blood samples collected in EDTA tubes and serum samples isolated from blood collected in SST tubes were sent to IDEXX BioResearch (Westbrook, ME) for complete blood count (CBC) and serum biochemistry analyses. Whole blood was assayed for total white blood cell (WBC) counts, total red blood cell (RBC) counts, total haemoglobin (HGB), hematocrit (HCT), mean corpuscular volume (MCV), mean corpuscular haemoglobin (MCH), mean corpuscular haemoglobin concentration (MCHC), along with platelet counts (PLT), and WBC differentials (e.g. neutrophils, lymphocytes, monocytes, eosinophils, and basophils). Serum was assayed for concentrations of alkaline phosphatase (ALP), alanine transaminase (ALT), aspartate transaminase (AST), creatine kinase, albumin (ALB), total protein (TP), globulin (GLOB), total bilirubin (TBIL), conjugated and unconjugated bilirubin, blood urea nitrogen (BUN), creatinine (CRE), total cholesterol (TCHO), glucose (GLU), calcium (Ca), inorganic phosphorus (IP), bicarbonate (HCO₃), chloride (Cl), potassium (K), and sodium (Na). ALB/GLOB, BUN/creatinine, and Na/K ratios were also determined from serum in the same IDEXX panel.

Tissue collection, processing, and histopathology

Biopsies of the quadriceps muscle were performed using a biopsy punch at 3 and 11 dpi. In-life cerebrospinal fluid samples were collected 1 day prior to virus challenge (to establish baseline measurements) and at 3, 7, 11, 14, and 31–33 dpi. Biopsies and cerebrospinal fluid samples were flash-frozen on dry ice for viral load analyses. All 12 cynomolgus macaques were euthanized between 31 and 33 dpi. The following tissues were sampled during complete necropsy immediately following euthanasia: axial, inguinal, mesenteric, and mandibular lymph nodes (LNs), spleen, knee joint muscle and synovia, small intestine, lung, liver, kidney, vagina, uterus, ovary, brain, and spinal cord. Necropsy samples for viral load analyses were flash frozen on dry ice, and samples for histopathologic assessment were fixed in formalin and shipped to IDEXX BioResearch (Westbrook, ME). Submitted tissues were trimmed and embedded in paraffin, sectioned at 5 µm, and stained with hematoxylin and eosin (H&E) for histological analysis.

Cytokine profiling

Circulating cytokine, chemokine, and growth factor levels were measured using a Milliplex Non-Human Primate Cytokine Magnetic Bead Panel kit (EMD Millipore) for 23 soluble analytes: G-CSF, GM-CSF, IFN-γ, IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12p40, IL-13, IL-15, IL-17a, MCP-1, MIP-1β, MIP-1α, sCD40L, TGF-α, TNF-α, VEGF, and IL-18. This 23-plex immunoassay, based on Luminex xMAP technology, was performed on serum samples according to the manufacturer’s instructions, which included appropriate standards and quality controls. Samples were run in duplicate, and data was acquired on a Luminex 200 instrument system.

Statistics

With the exception of hierarchical clustering performed on cytokine data, all statistical analyses were carried out in GraphPad Prism 10.1.0 (GraphPad Software, San Diego, CA). Data were normalized by Log10 transformation when necessary and data distributions were evaluated for normality. Data were analysed using one-way and two-way ANOVAs where appropriate and represented in figures as the mean±standard deviation (SD) where relevant. In cases where missing data points precluded the use of two-way ANOVA, mixed-effects models (REML) were applied. In cases where the data could not be meaningfully normalized (i.e. absolute cell counts in CBC datasets), nonparametric Friedman’s tests with post-hoc Dunn’s multiple comparisons were used. For some statistical comparisons of assay readouts, values registering as below the LLOD were substituted with a value equal to half the LLOD prior to log transformation. Hierarchical clustering and creation of heatmaps for Luminex data were performed in R version 4.4.0 (https://www.r-project.org) on log2-transformed MFI fold changes from the PBS control group.

Results

MAYV challenge of cynomolgus macaques results in RNAemia, viremia, and tissue viral loads that suggest genotype-specific infection kinetics

To characterize the kinetics of MAYV infection in the cynomolgus macaque, blood plasma was collected 6 h post-infection (shown in Fig. 1a as 0 dpi) and at 1–4, 7 and 11 dpi for viral load analysis by RT-qPCR. Plasma RNAemia in all three MAYV-challenged groups peaked at 2 dpi, with MAYV N and D strains achieving mean viral RNA titres of 7.3Log₁₀ and 7.5Log₁₀ RNA copies ml⁻¹, respectively (Fig. 1a). RNAemia waned in these two groups to ~2.6Log₁₀ RNA copies ml⁻¹ by 7 dpi before eventually diminishing to near undetectable levels by 11 dpi. Animals infected with MAYV L presented with comparatively lower peak mean RNAemia titres of 5.7Log₁₀ RNA copies ml⁻¹ and exhibited lower overall RNAemia compared to N and D genotypes in a two-way ANOVA of all longitudinal plasma RNA (P=0.0013). RNAemia in MAYV L- and N-challenged groups also appeared to decrease from peak more rapidly compared to the MAYV D group, though this trend was only statistically significant in a comparison between MAYV D and MAYV L (4 dpi; P=0.0234). Of note, viral RNA was detectable in plasma of the MAYV D-challenged group at 0 dpi, just 6 h post-infection; MAYV RNA was not detected in other groups at this early time point. In summary, the MAYV D genotype appears to elicit the earliest onset, highest peak, and slowest kinetic decline of RNAemia compared to other genotypes.

Fig. 1. Characterization of acute MAYV infection and early immune response in cynomolgus macaques. (a) Kinetics of MAYV RNAemia as determined by RT-qPCR of blood plasma sampled 0–4, 7, and 11 dpi; the 0 dpi time point was collected 6 h post-infection. The lower limit of detection (LLOD) for this assay was 50 RNA copies ml⁻¹ (1.7Log₁₀). Viral RNA was not detected in the PBS control group at any time point. (b) Viremia, as measured by plaque assay, in terms of plaque forming units per millilitre (PFU ml⁻¹), in serum samples collected at 1–4 dpi. Plaque assay was performed with raw sera on Vero cells without isolating and/or passaging the virus. Viremia was not detectable by 3 dpi (not shown). (c) Log₁₀ MAYV RNA in muscle biopsies collected at 3 and 11 dpi as determined by a high-sensitivity semiquantitative RT-PCR assay normalized to expression of the housekeeping gene GAPDH. The LLOD for this assay was 2 RNA copies per copy of endogenous GAPDH (0.3Log₁₀ RNA:GAPDH). (b–c) Statistical analyses were performed using mixed-effects model analyses (REML) with Tukey’s post-hoc multiple comparisons tests; *P<0.05, **P<0.01, ***P<0.001. (d) Neutralizing antibody (nAb) titres, as measured by 80% plaque reduction neutralization test (PRNT₈₀), in sera collected at 14 dpi. Statistical analyses were performed using ordinary one-way ANOVA with Tukey’s post-hoc multiple comparisons tests; *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. (a–d) Error bars indicate ±standard deviation (SD) from the mean. Dotted black lines indicate LLODs for each assay.

To evaluate viremia (i.e. infectious viral titres in the blood), serum samples were collected 1–4 dpi and directly evaluated via plaque assay on Vero-76 cells. All MAYV-challenged animals displayed transient but detectable viremia in the first two days following infection (Fig. 1b). Viremia waned rapidly and could not be detected by plaque assay at 3–4 dpi (data not shown). The highest viremic titres at 1 dpi were observed in animals infected with MAYV D (~3.3Log₁₀ PFU ml⁻¹), consistent with the early development of RNAemia seen in this group. By 2 dpi, the highest viremia was observed in animals infected with MAYV N (~3.2 Log₁₀ PFU ml⁻¹), which was significantly higher compared to animals infected with MAYV L (P=0.0113).

Acute MAYV infection in humans often presents with myalgia [51,53] and MAYV has been shown in murine models to preferentially replicate in skeletal muscle [34]. To assess this phenomenon in the cynomolgus macaque, quadriceps biopsies were performed at 3 and 11 dpi and assayed for viral RNA via a high-sensitivity semiquantitative RT-PCR assay normalized to the expression of the housekeeping gene GAPDH. Muscle biopsies of MAYV-challenged animals collected during acute infection (3 dpi) exhibited mean viral loads between 4.7Log₁₀ and 6.1Log₁₀ RNA copies:GAPDH (Fig. 1c), comparable to those observed in plasma at this time point (Fig. 1a). By 11 dpi, in the convalescent stage of MAYV infection, mean muscle viral loads had waned only somewhat, with titres between 1.9Log₁₀ and 3.6Log₁₀ RNA copies:GAPDH (Fig. 1c). At this same time point, MAYV RNA was barely detectable in plasma, suggesting productive viral replication in muscle. MAYV D and MAYV N viral loads in muscle biopsies were significantly higher than MAYV L viral loads in a mixed-effect model analysis of all genotypes (P=0.0027), as observed in plasma. Furthermore, the same analysis demonstrated that while MAYV L and MAYV D muscle viral loads decreased significantly from 3 dpi to 11 dpi (P=0.0117 and P=0.0293, respectively), MAYV N did not, corroborating the observation that the kinetics of MAYV infection may differ between genotypes.

MAYV infection elicits a strong neutralizing antibody response in cynomolgus macaques

A plaque reduction neutralization test using an 80 % threshold (PRNT₈₀) was conducted to evaluate neutralizing antibody (nAb) titres in sera collected at 14 dpi. All MAYV-infected macaques displayed significant nAb titres at this time point (Fig. 1d). Animals challenged with MAYV N and L displayed higher levels of nAbs (PRNT₈₀ titres between 320 and 2560) than animals challenged with MAYV D (PRNT₈₀ titre of 160) (Fig. 1d), despite the fact that MAYV D challenge generally produced higher RNAemia, viremia, and tissue viral loads during acute infection. This was significant in a one-way ANOVA of all MAYV-challenged groups (P=0.0422).

MAYV RNA persists at high levels in lymphoid and skeletomuscular tissues one month after infection

To explore the possibility of persistent infection in this model, all MAYV-infected macaques were necropsied for a variety of tissues and organs at 31–33 dpi (at the transition between the convalescent and persistent stages of infection) for viral load determination via RT-qPCR. The highest MAYV RNA titres at necropsy were observed in lymphoid tissues (Fig. 2a), including spleen (~6.9Log₁₀ RNA copies g⁻¹) and all lymph node (LN) types collected, including axial (~5.8Log₁₀ RNA copies g⁻¹), inguinal (~6.5Log₁₀ RNA copies g⁻¹), mesenteric (~4.4Log₁₀ RNA copies g⁻¹), and mandibular (~6.0Log₁₀ RNA copies g⁻¹) lymph nodes. Moderately high titres were detected in the knee joint muscle (~4.1Log₁₀ RNA copies g⁻¹) and synovia (~3.6Log10 RNA copies g⁻¹) (Fig. 2b), as well as the kidney (~4.0Log₁₀ RNA copies g⁻¹) and liver (~3.2Log₁₀ copies g⁻¹) (Fig. 2c), and some reproductive organs (Fig. 2e). MAYV RNA was detectable at the lowest levels (<3.0Log₁₀ RNA copies g⁻¹) in other non-lymphoid organs (Fig. 2e) and central nervous system tissues (Fig. 2d). Consistent with plasma and muscle viral load data at earlier time points, infection with MAYV L produced significantly lower tissue and organ viral loads 1 month post-infection compared to infection with MAYV N (P=0.0004) or MAYV D (P<0.0001). Notably, MAYV D infection generally produced the highest viral titres in the lymphoid organs (Fig. 2a), while MAYV N produced the highest titres in joint tissues associated with myalgia and arthralgia (Fig. 2b). Cerebrospinal fluid was also collected at 3, 7, 11, 14 and 31–33 dpi, but MAYV RNA was largely undetectable (data not shown).

Fig. 2. Persistence of viral RNA in cynomolgus macaque tissues 1 month following MAYV infection. Log₁₀ MAYV viral RNA as determined by RT-qPCR of (a) lymphoid tissues, (b) joint tissues, (c) non-lymphoid organs, (d) reproductive tissues, and (e) central nervous system (CNS) tissues. (a–e) All tissues were collected at necropsy between 31–33 dpi. Statistical analyses were performed via two-way ANOVA with post-hoc Tukey’s multiple comparisons test; *P<0.05, **P<0.01. Error bars indicate ±standard deviation (SD) from the mean. Dotted black lines indicate the lower limit of detection (LOD) for these assays (200 RNA copies g⁻¹).

MAYV challenge results in significant leucopenia concomitant with RNAemia

Complete blood count (CBC) analyses were performed on whole blood collected at 1, 2, 4, 7, and 14 dpi; baseline (BL) values for each animal were established 1 day prior to infection (−1 dpi). MAYV-challenged animals across all groups exhibited a significant (P=0.0125) decrease in absolute white blood cell (WBC) counts (i.e. leucopenia) from baseline to 2 dpi (Fig. 3a), concomitant with peak RNAemia. From 7 dpi onwards, WBC counts in MAYV-challenged animals began to recover to levels comparable to baseline. There were no significant differences in absolute WBC counts observed between MAYV genotypes or compared to the small (n=3) PBS control group. Small group size and high inter-individual variation in this data set (20.7 % of variation due to subject as determined by two-way ANOVA; P=0.0003) likely accounts for the latter.

Fig. 3. White blood cell total and subtype counts in MAYV-infected cynomolgus macaque whole blood. Absolute counts of (a) total white blood cells (WBCs), (b) neutrophils, (c) lymphocytes, (d) eosinophils, (e) basophils, and (f) monocytes in whole blood from baseline (−1 dpi) to 14 dpi. (a–f) Data for MAYV-infected animals are shown for individuals identified by a four-digit ID; MAYV N animals are shown in red, MAYV L in blue, and MAYV D in green. Mean values for the PBS control group are shown as a dashed black line for reference; these were not included in the longitudinal statistical analysis. Grey areas indicate values outside of the normal clinical range for adult female cynomolgus macaques. Statistical analyses comparing baseline values of MAYV-infected animals to post-infection time points were performed using Friedman’s test with Dunn’s post-hoc multiple comparisons test; arrows with P values indicate a significant difference from the baseline (−1 dpi).

Nevertheless, the leucopenia observed in MAYV-infected animals appears largely due to decreases in neutrophil counts (Fig. 3b), lymphocyte counts (Fig. 3c), and eosinophil counts (Fig. 3d) during acute infection. Compared to baseline, neutrophil counts in MAYV-infected animals were significantly lower at 2 dpi (P=0.0188) and 4 dpi (P=0.0082) (Fig. 3b). Similarly, lymphocyte counts were significantly decreased from baseline at 2 dpi (P=0.0008) and 4 dpi (P=0.0408) (Fig. 3c). Eosinophils, while comprising a smaller percent of total WBC counts, were also significantly decreased from baseline at 2 dpi (P=0.0013) and 4 dpi (P<0.0001) (Fig. 3d). Basophil counts in MAYV-infected animals (Fig. 3e) appear to be decreased from baseline at the same time points, but these differences were not statistically significant. By 7 dpi, however, basophil counts (Fig. 3e) and monocyte counts (Fig. 3f) in some MAYV-infected animals appeared elevated compared to baseline; animal 6639 (infected with MAYV L), animal 6630 (infected with MAYV N), and animal 6667 (infected with MAYV D) account for the highest values observed in both data sets at 7 dpi. No significant differences in absolute cell counts or differentials as a percentage of WBC were observed between MAYV genotypes; other CBC parameters (e.g. RBC, HCT, MCV, PLT) appeared unaffected by MAYV infection (Fig. S1, available in the online version of this article).

MAYV-infected cynomolgus macaques present with limited clinical symptoms

MAYV-challenged animals did not present with any visible clinical signs (e.g. poor grooming, hunched posture, low appetite, lethargy, diarrhoea) during the study (data not shown). None of the animals challenged in this study succumbed to MAYV infection, and no euthanasia was performed prior to the scheduled sacrifice date 1 month post-infection. MAYV-challenged animals did not present with fever above 39.5 °C (Fig. 4a) or statistically significant body weight loss (Fig. 4b). It is worth noting, however, that MAYV N-challenged animals displayed a mean body weight loss of ~5 % at 3 dpi (Fig. 4a), shortly after peak RNAemia and viremia. Cachexia of this degree may be significant in larger studies. Likewise, all three MAYV-challenged groups exhibited a small (~1 °C) spike in in mean body temperature from 2 to 3 dpi (Fig. 4b), and telemetry data from transponders implanted into the quadriceps (Fig. S2) show temperature spikes approaching fever (39.5 °C) at 2 dpi in animal 6630 (MAYV N), animal 6639 (MAYV L), animal 6743 (MAYV D) and animal 6747 (MAYV D).

Fig. 4. Clinical parameters in MAYV-infected cynomolgus macaques. (a) Macaque mean body temperatures from 0 to 11 dpi. (b) Mean percent (%) changes in macaque body weights from 0 to 11 dpi. (c) AST levels in macaque serum from baseline (−1 dpi) to 14 dpi, shown in units per litre (U l^-1). (d) ALT levels in macaque serum from baseline (−1 dpi) to 14 dpi, shown in units per litre (U l^-1). Grey areas indicate values outside of normal clinical range.

While the majority (>80 %) of reported MAYV clinical cases do not present with abnormal serum biochemistry, elevated levels of hepatic function markers indicating liver injury (i.e. AST and ALT) have been documented in a number of human Mayaro fever patients [52]. In our study, clinical biochemistry analyses were performed on MAYV-challenged macaque serum samples collected 1–4, 7, and 14 dpi; sera collected at −1 dpi were used to determine group baselines. AST appeared elevated from baseline in MAYV D-infected animals at 4 dpi, but this difference was not significant compared the PBS-treated controls (Fig. 4c). Similarly, ALT showed moderate increases from baseline in all MAYV groups, but ALT levels in PBS controls were consistently higher (Fig. 4d). Other values for serum chemistry analytes (ALP, creatine kinase, ALB, TP, GLOB, TBIL, conjugated and unconjugated BIL, BUN, CRE, TCHO, GLU, Ca, P, HCO₃, Cl, K, Na, ALB/GLOB ratio, BUN/CRE ratio, Na/K ratio) were largely within clinical range and/or did not demonstrate a significant group effect in two-way ANOVAs (data not shown).

MAYV infection results in moderate histological abnormalities in the lung, liver, and lymph node

Infection of cynomolgus macaques with MAYV resulted in histological abnormalities detected primarily in the lung, liver, and inguinal lymph node (Table 1). Moderate to severe congestion and/or oedema was observed in the lung (n=2/2). In the liver, mild congestion or multifocal fatty change was observed (n=2/2). All inguinal and mesenteric lymph nodes assessed were enlarged and/or reactive (n=3/3 and n=2/2, respectively), though no microscopic changes were observed. At the 10⁶ PFU challenge dose used in this study, MAYV infection did not appear to cause any histological abnormalities in the central nervous system tissues, kidneys, spleen, pancreas, or gallbladder. No microscopic changes were observed in the patellar muscles (n=0/2), and no joint synovia were assessed.

Table 1. Summary of histopathologic assessment of select cynomolgus macaque tissues collected 1 month after MAYV infection.

Animal ID	6639	6647	6794	6667	6743	6747
Lung	Congestion and oedema, moderate to severe	-	-	-	Congestion, moderate	-
Liver	Fatty change; multifocal, minimal	-	-	-	Congestion, mild	-
Gallbladder	Normal	-	-	-	Normal	-
Mand. LN	Normal	-	-	-	-	Reactive
Sal. LN	Normal	-	-	-	-	-
Mes. LN	Mildly reactive	-	-	-	-	Reactive
Ing. LN	Reactive	-	-	Enlarged, reactive	Enlarged, reactive	-
Axe. LN	Normal	-	-	-	-	Mildly reactive
Vagina	Mild lymphoid infiltrate; mucosa	-	-	-	-	Keratinized
Uterus	Normal	-	-	-	-	Normal
Ovary	Multiple corpora lutea	-	-	-	Large follicular cyst	-
Patellar muscle	-	Normal	Normal	-	-	-
Forebrain	-	Normal	-	-	Normal	-
Midbrain	-	Normal	-	-	Normal	-
Cerebellum	-	Normal	-	-	Normal	-
Hindbrain	-	Normal	-	-	Normal	-
Spinal Cord	-	Normal	-	-	-	Normal
Kidney	-	Normal	-	Normal	-	-
Spleen	-	Normal	-	-	-	Normal
Pancreas	-	Normal	-	-	-	-

Dashes in this table indicate that no sample was assessed.

MAYV infection results in genotype-specific cytokine, chemokine, and growth factor responses in macaques

Cytokine profiles have been shown to vary significantly between humans infected with CHIKV and humans infected with MAYV [31]. To explore this phenomenon, as well as any MAYV genotype-specific differences in immune response to MAYV, we analysed sera from all groups collected 1–4 dpi against a panel of 23 cytokines, chemokines, and growth factors (Fig. 5a). The responses of individual analytes appear to fall into several categories of interest: 1) analytes that display a sharp, transient spike at 2 dpi across all MAYV genotypes, as with IL-1ra (Fig. 5b), 2) analytes that are exclusively or primarily elevated in MAYV N- and/or MAYV L-infected animals from 1 to 2 dpi onwards, as with IL-13 (Fig. 5c), and 3) analytes that are elevated in all three MAYV-infected groups from 1 to 2 dpi onwards, as with IL-2 (Fig. 5d). In two-way ANOVAs of untransformed analyte concentrations, four analytes exhibited a significant group effect over time: IL-2 (P=0.0165), IL-15 (P=0.0225), IL-13 (P=0.0211), and VEGF (P=0.0323). IL-2 and IL-15 were broadly elevated in all three MAYV-infected groups compared to the PBS control, while IL-13 and VEGF were elevated exclusively in MAYV N-infected animals (Figs 5a and S3).

Fig. 5. Acute cytokine, chemokine, and growth factor responses in MAYV-infected macaques. Serum was collected from MAYV-infected and PBS-treated control macaques at 1–4 dpi and evaluated for cytokine, chemokine, and growth factor concentrations via 23-plex Luminex bioassay. (a) Hierarchical clustering heat map of all 23 Luminex analytes showing the log₂ fold change of mean fluorescence intensity (MFI) for each group calculated based on the PBS control group MFI at each time point. The x-axis labels indicate the MAYV challenge genotype and time point, i.e. “N1” indicates MAYV N sera sampled at 1 dpi, etc. Rows (analytes) were clustered by Euclidean distance. (b) IL-1ra serum concentrations shown in pg ml⁻¹. (c) IL-13 serum concentrations shown in pg ml⁻¹. (d) IL-2 serum concentrations shown in pg ml⁻¹. (b–d) Statistical analyses were performed using two-way ANOVA with Tukey’s post-hoc multiple comparisons test; *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Error bars represent ±standard deviation (SD) from the mean. Black dotted lines indicate the lower limit of quantification (LLOQ) for each analyte.

Discussion

As MAYV increases its geographic spread and causes new outbreaks, it presents a significant threat to global public health. Yet, as with many other arboviruses, there are currently no FDA-approved vaccines or therapeutics available to contain MAYV outbreaks. Developing and evaluating these interventions will require MAYV-specific animal models with immediate preclinical relevance. Our aim was to provide a preliminary characterization of MAYV infection in non-human primates, specifically cynomolgus macaques (Macaca fascicularis), and evaluate potential differences in pathogenesis and immune response among the three known MAYV genotypes: N (New), L (Limited), and D (Dispersed). Due to sparse genotyping of clinical cases and a dearth of direct comparisons between MAYV genotypes in available animal models, little is known about potential genotype-specific differences regarding the epidemiology and pathology of MAYV. All historical MAYV outbreaks have been primarily of genotype D; MAYV L has only been reported outside of Brazil three times (though once in Haiti, a known hub of viral emergence from South America), and MAYV N has not been detected outside of Peru [4,10]. Importantly, however, L and D strains have been shown to vary significantly in their vector competence, with MAYV L exhibiting greater ability to infect and disseminate in urbanized mosquito species like Ae. aegypti [24]. Considering that CHIKV and ZIKV are believed to have escaped their respective enzootic cycles – at least in part – by acquiring mutations that improved their vector competence in various anthropophilic Aedes species, it would be remiss to ignore differences in MAYV strains in the pursuit of preventing future outbreaks [53,55].

In this first study of Mayaro virus infection in the cynomolgus macaque, three animals each were challenged with 10⁶ PFU of a representative strain (or PBS as a control) via intravenous route. All twelve animals were evaluated for virologic, immunologic, and histologic readouts for a study period of 1 month post-infection, the end of which marks the transition from the convalescent to the persistent stage of many alphavirus infections [36]. Significant MAYV-specific RNAemia and viremia were detected in cynomolgus macaques challenged with all three MAYV genotypes. RNAemia peaked at 2 dpi, with titres reaching 7.5Log₁₀ MAYV RNA ml⁻¹ of plasma, and then waned rapidly, becoming nearly undetectable by RT-qPCR by 11 dpi. These kinetics closely recapitulate those observed in a recent MAYV challenge study of three rhesus macaques, in which animals were infected with 10⁵ PFU virus [56] and compares with the proposed time course of human MAYV infection based on collated serological data from clinical cases [3]. Intravenous inoculation of cynomolgus macaques with CHIKV, considered a close evolutionary cousin of MAYV, also yields a similar curve in longitudinal RT-qPCR experiments [32]. Notably, infection of cynomolgus macaques with MAYV D elicited the earliest onset, highest peak, and slowest kinetic decline of RNAemia compared to other genotypes, while MAYV L viral RNAemia was generally less severe and was cleared more quickly.

Human MAYV infections are characterized by a short viremic period, lasting 5 days at most [3]. In our study, viremia as determined by plaque assay was transient, only present for 2 days immediately following infection with each genotype. Overall, viremic titres were low (~3Log₁₀ PFU ml⁻¹), especially considering the high viral RNA loads in plasma and muscle as determined by RT-qPCR. It should be considered, however, that the plaque assay methodology employed here neither isolated nor passaged virus from the serum samples assayed, as has been done in other studies evaluating viremia in MAYV challenge experiments [48]. Additionally, the viremic titres observed here are comparable to previous NHP CHIKV infection studies, which showed that rhesus macaques infected with either CHIKV-African E-103 or CHIKV-Asian BAH-306 developed viremia that lasted 4–5 days with peak viremia levels reaching 10⁴–10⁶ PFU ml⁻¹ [57]. Cynomolgus macaques infected with CHIKV La Reunion strain show that disease severity is dependent on dose and that a moderate dose of 10⁶ PFU caused only fever, rash and RNAemias (5×10⁹ CHIKV RNA ml⁻¹ plasma) comparable to those observed in our study [45]. While cynomolgus macaques are generally more susceptible to development of high titre viremia for CHIKV infection, it must be noted that MAYV is a New World alphavirus and thus may not be particularly well adapted for significant viremia in this model. Further studies involving various inoculation routes and higher doses are needed to explore the range of viremia presented in cynomolgus macaques. Nonetheless, the presence of viremia and RNAemia in this study confirms this is a measurable outcome that can be used for pathogenesis and efficacy studies.

The replication of arthralgic alphaviruses in muscle tissues has been well documented and has been shown to be important for pathogenesis and playing a role in persistent infections [45,46]. Along with the development of viremia and RNAemia, viral RNA was detected in skeletal muscle of MAYV-infected cynomolgus macaques at 3 and 11 dpi. Despite variability among and within groups, there was consensus of high MAYV viral loads in the muscle biopsies in all three genotype infections. Additionally, MAYV L and MAYV D muscle viral loads decreased significantly from 3 dpi to 11 dpi while MAYV N did not, further supporting our observation that the kinetics of MAYV infection may differ between genotypes.

All MAYV-challenged macaques had developed MAYV-specific neutralizing antibodies by 14 dpi. Interestingly, animals infected with MAYV N and L produced significantly higher levels of neutralizing antibodies than animals infected with MAYV D. While this may be a coincidental observation, it is possible that the induction of lower neutralizing antibody titres could be related to the more robust replication in the MAYV D infection group as seen via RT-qPCR [49]. However, further studies are needed to truly determine if there are differences in neutralizing antibody responses after infection with the various MAYV genotypes. This will be critically important for an increased understanding of herd immunity effects in natural infection, and if such a difference does exist, this will be important for antigen selection and design for MAYV vaccines.

Importantly, MAYV RNA was detected in most organs and tissues collected 1 month after infection, with exceptionally high loads detected in lymphoid and joint tissue (i.e. knee). Viral RNAs were also detected in the liver, a known site of alphavirus replication. At this time point bridging the convalescent and persistent phases of alphavirus infection, we observed that infection with the MAYV D genotype resulted in higher viral loads in the lymphoid tissues, while infection with the MAYV N genotype resulted in higher viral loads in tissues associated with arthralgia (knee joint muscle and synovia). MAYV N was also detected at higher levels in the quadriceps muscles at 11 dpi, during the convalescent stage of infection. Considering that persistent and debilitating arthralgia is arguably the most concerning symptom of human MAYV infections, it will be essential to include genotype N strains in future investigations of MAYV pathogenesis and intervention development.

We observed significant leucopenia, primarily due to decreases in lymphocyte and neutrophil counts, in the blood of all three MAYV-challenged groups during and shortly after peak RNAemia. This is consistent with serological data from many human cases [58], but notably in opposition to serological data from mouse models of acute MAYV infection [59]. MAYV infection elicited negligible effects on other CBC analytes, like those associated with anaemia. While thrombocytopenia is present in some clinical cases [60] and has been documented in a mouse model [39], we did not detect significant changes in platelet counts in any MAYV-challenged group. The broad depletion of circulating WBC subtypes, with the exception of monocytes, at 2–4 dpi suggests potential heavy recruitment of immune cells to infected tissues and/or significant replication of virus within these cell types, both of which have been documented in alphavirus infection models [61]. Flow cytometry immunophenotyping of putative early sites of viral replication (e.g. spleen, lymph nodes), along with viral load determination in isolated immune cell populations sampled from a variety of tissues and peripheral blood mononuclear cells (PBMCs), should be considered for further verifying the phenotypic distribution of the immune cells to broaden the current understanding of the model. Regardless, monitoring white blood cell counts during acute infection will prove to be a major endpoint for future MAYV pathogenesis and efficacy studies in the cynomolgus macaque.

Cynomolgus macaques presented with no visibly apparent clinical disease after infection with MAYV. There were also no significant changes in macaque internal temperatures measured during physical exams, though spikes in body temperature approaching clinical fever were observed in some animals via telemetry data. Small decreases in mean body weight were observed in MAYV-challenged groups shortly after peak viremia and RNAemia, but these differences were not significant compared to unchallenged controls. However, animals challenged with MAYV N exhibited mean body weight loss of ~5 % at 3 dpi, which may prove statistically significant in other experimental designs. Changes to markers of hepatic function (ALT and AST), which have been documented in clinical cases and murine models of MAYV infection [52], were equivocal. However, elevated values in the unchallenged control animals may have obfuscated any infection-related increases in these analytes. Overall, however, infection of cynomolgus macaques at a dose of 10⁶ PFU did not produce severe disease. This should be noted when considering this model for assessing the efficacy of countermeasures.

A cursory histopathology assessment of select tissues from MAYV L- and D-challenged animals 1 month post-infection revealed signs of inflammation and damage to known sites of alphavirus replication, including the liver and lymph nodes [36,52]. However, no signs of microscopic abnormalities were observed in muscles attached to joints, as would be expected from an arthritogenic infection. Due to technical challenges, no tissues from MAYV N-infected animals were available for evaluation, which is unfortunate considering that these animals demonstrated the highest viral titres in arthritogenic tissues and displayed the most robust and generalized inflammatory cytokine response. Appreciating that this preliminary tissue pathology testing was conducted with a limited number of animals and samples, we recommend a comprehensive histopathologic assessment in future studies of MAYV pathogenesis in the cynomolgus macaque, and investigation of other MAYV strains, especially those belonging to genotype N. Additionally, specific scoring for joint inflammation during acute infection should determine if this model will serve in studies evaluating MAYV-induced arthralgia.

Finally, we observed robust cytokine and chemokine responses during acute MAYV infection that were both genotype-specific and distinct from those reported in CHIKV infection. Of the 23 analytes assessed, nine were upregulated exclusively or primarily in N- or L-infected groups, with MAYV N eliciting the most severe and generalized cytokine profile. While there is not yet evidence that human Mayaro disease outcomes differ among MAYV genotypes, the fact that the three analytes associated most with persistent arthralgia – IL-13, IL-17, and VEGF – were upregulated only in MAYV N infections is compelling and merits further investigation. It is also interesting that genotype D infections, which elicited the weakest overall cytokine response, also resulted in the lowest neutralizing antibody titers. As observed in human analyses, MAYV infection in this model produced cytokine responses distinct from those observed in CHIKV infection of cynomolgus macaques: the upregulation of IL-10, IL-1β, IL-12p40, and IL-17 observed here is notably absent in similar infection studies of CHIKV [44], further underscoring the importance of developing experimental models specific to MAYV. The inflammatory cytokines TNF-α and IL-6 were undetectable except in MAYV N-infected animals, and only at very low levels. Considering that these cytokines are detected at high levels in human cases of Mayaro disease [41], and that MAYV-infected mice also display elevated TNF-α and IL-6 [62], further exploration of these analytes is warranted.

In conclusion, we have shown that intravenous infection of cynomolgus macaques (Macaca fascicularis) with all three MAYV genotypes leads to the induction of viremia and RNAemia, high viral loads in lymphoid tissues, leucopenia, unique cytokine response profiles, and development of neutralizing antibodies. Our data suggest significant differences in infection kinetics, replication site preference, and immune response among genotypes, which warrants further investigation and reinforces the importances of assessing multiple viral strains in pathogenesis studies. Despite limited histological data, we present a preliminary non-human primate model that can facilitate future research on MAYV pathogenesis and efficacy studies for novel therapeutics and prophylactics.

Abbreviations

ANOVA

analysis of variance

CBC

complete blood count

CHIKV

Chikungunya virus

dpi

days post-infection

EDTA

ethylenediaminetetraacetic acid

ELISA

enzyme-linked immunosorbent assay

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

LLOD

lower limit of detection

LLOQ

lower limit of quantification

LN

lymph node

MAYV

Mayaro virus

MAYV D

Mayaro virus genotype D

MAYV L

Mayaro virus genotype L

MAYV N

Mayaro virus genotype N

MOI

multiplicity of infection

nAb

neutralizing antibody

NHP

non-human primate

PBS

phosphate-buffered saline

PFU

plaque-forming unit

PRNT

plaque reduction neutralization test

REML

restricted maximum likelihood

RT-qPCR

real-time quantitative polymerase chain reaction

SST

serum separator tube

WBC

white blood cell

ZIKV

Zika virus