Multimodal assessments of Zika virus immune pathophysiological responses in marmosets.

Animal models that recapitulate the human pathophysiology have been developed as useful research tools. Although laboratory mice are widely used, they are phylogenetically "distant" to humans. New world monkeys, such as the common marmoset (Callithrix jacchus) have steadily gained prominence. In this report, marmosets are explored as an alternate in vivo model to investigate infection and immunity of Zika virus (ZIKV). Multimodal platforms, including ultrasound and magnetic resonance imaging (MRI), flow cytometry, and multiplex microbead immunoassays were established to comprehensively decipher immune responses and pathophysiological outcomes. While ZIKV-infected marmosets had detectable ZIKV RNA load in various body fluids, animals did not develop any observable lesions in their testes and brains as shown by ultrasound and MRI. Immune-phenotyping detected differences in the numbers of B cells, CD8+ T cells and HLADR+ NK cells during the first two weeks of infection. Neutralizing ZIKV-specific antibodies were elicited to high levels and targeted epitopes in the E protein. This study presents a one-stop-shop platform to study infection and pathophysiology in marmosets. While marmoset-specific research tools are being refined, the research values of these animals present them as a good model for immune-based therapies.


Virus infection of marmosets and detection of viral RNA in various body fluids.
Six adult marmosets (three males and three females) aged between 101-125 weeks old were selected for this study (Fig. 1A). An initial "baseline" 3-week period was carried out to optimize experimental procedures (Fig. 1B). Virus infection was performed with inoculation of 1E5 PFU of ZIKV virions via the saphenous vein. ZIKV-infected marmosets were observed at stated time-points over 4 months (Fig. 1B). Physiological and immune changes were assessed at selected time-points. Marmosets were also assessed for the presence of ZIKV RNA load in the various body fluids (Fig. 1B).
Corroborating the findings in ZIKV-infected patients 40 , ZIKV RNA load was detected in marmosets' whole blood, plasma, saliva and urine specimens. Virus detection was highest in the plasma, followed by whole blood samples, with the first traces of ZIKV RNA detected as early as 2 days post-infection (dpi). Detectable levels started to decrease from 6 dpi to trace levels of ~100 copies/ul at 15 dpi and went below detection limit by 30 dpi (Fig. 1C). In the urine and saliva specimens, low viral RNA load was observed at 2 dpi and peaked at 6 dpi ( Fig. 1C). Although no gender-bias was observed in viral load, variation across the ZIKV-infected marmosets in terms of absolute quantities, especially in the urine and saliva samples was observed (Fig. 1C).

Clinical and pathophysiological changes in ZIKV-infected marmosets. ZIKV-infected marmosets
showed mild clinical signs during the first week of infection and exhibited no further disease complications for the remaining study period. However, all five ZIKV-infected marmosets were observed to have some soft dark stools from day 1-7 dpi, which was resolved by the second week. Although animals appeared slightly lethargic during the first week of infection, they were alert and responsive. None of the animals showed any signs of joint pain, limping or conjunctivitis. These symptoms were consistent with a mild viral infection 41,42 . These animals also did not experience any changes in temperature and weight (Fig. S1). Similarly, non-infected marmosets (3 females and 2 males) (Fig. S1A), also did not experience any changes in body weight and temperature ( Fig. S1B-E). It was reported that ZIKV could infect testicular and neural cells 43,44 , thus both testes and the brains of the marmosets were evaluated in this study. Fortunately, there were no pathological sonographic findings on grey scale and Doppler evaluation of both ZIKV-infected and non-infected groups (Fig. 2). Images indicated an absence of inflammation in the testes during both acute (6 dpi; Fig. 2A and C) or post-acute (120 dpi; Fig. 2B and D) disease phase. Testicular volume derived manually with a vernier caliper ( Fig. S2A-C), and with ultrasound imaging (Fig. S2D-F) showed no statistical significance between the two groups of animals (P = 0.09, 0.18 respectively) and between the two groups with the passage of time (P = 0.11, 0.47 respectively) ( Fig. S2C and S2F).
Given the small size of the marmosets, optimization for the MRI was needed (see Materials and methods), and MRI on the brains of ZIKV-infected and non-infected animals did not reveal any focal abnormality or significant finding during acute (6 dpi; Fig. 3A and C) or post-acute (120 dpi; Fig. 3B and D) disease phase. The calculated brain volumes ( Fig. 3E and Fig. S2G-I) also did not show any significant differences between the two groups of animals (P = 0.28) and between the two groups with the passage of time (P = 0.83) (Fig. 2I).

Changes in profile of immune cells during post ZIKV infection.
As ZIKV RNA load was cleared within 30 days (Fig. 1C), specific peripheral immune cell changes within this period were studied ( Fig. S3; see supplementary discussion online). Notably, of the 15 immune subsets identified (Fig. S4), only the B cells, CD8+ T cells, HLADR+ NK cells and CD16+ monocytes exhibited significant differences between the ZIKV-infected and non-infected groups (Fig. 4A). ZIKV infection resulted in lower B-cell and CD8+ T-cell numbers at 2 dpi, higher HLADR+ NK cells at 6 dpi and reduced CD16+ monocytes at 15dpi (Fig. 4A). However, it would be crucial to note that in the non-infected marmosets, some fluctuations were observed in their peripheral immune cell numbers, which may confound findings presented in Fig. 4A. Nonetheless, to further define any phenotypic differences within these immune subsets, data was interpreted with tSNE, a method that conserves the high dimensionality of the data and allows for easy visualization of the data on two dimensions 45 . Despite the differences in cell numbers (Fig. 4A), no further phenotypic differences were observed between the non-infected and ZIKV-infected marmosets (Fig. 4B).

ZIKV infection elicits neutralizing anti-ZIKV antibodies.
Utilizing a ZIKV virion-based ELISA 46 , anti-ZIKV IgM was detected as early as 2 dpi (Fig. 5A). Consistently, titers of ZIKV-specific IgM peaked at 15 dpi in all six ZIKV-infected marmosets, before declining sharply thereafter (Fig. 5A). The peak of ZIKV-specific IgM response at 15 dpi coincided with the first traces of detectable ZIKV-specific IgG antibodies that continued to increase up to 4 months post-infection (mpi) (Fig. 5A). Plasma obtained from non-infected marmosets was devoid of any ZIKV-specific antibodies (Fig. 6A).
Functionally, efficient in vitro neutralization was observed in ZIKV-specific antibodies taken from 15 dpi onwards till 1-2 mpi. (Fig. 5B). The strong inhibition concurs with the production of ZIKV-specific IgG (Fig. 5A), while plasma samples from non-infected animals were non-neutralizing (Fig. 6B). Interestingly, plasma samples from ZIKV-infected marmosets did not cross-react with all 4 serotypes of the DENV virus (Fig. 6C).
ZIKV-specific antibodies recognizes algorithm-predicted linear epitopes on ZIKV E protein. It was recently reported the ZIKV E protein contains numerous surface-exposed regions that could be targeted by ZIKV antibodies 47 . To determine if ZIKV-specific antibodies induced in the marmosets recognize algorithm-predicted immunogenic linear epitopes within the E protein 47 , five key peptides (labeled as E1, E2, E3 E4 and E5) were selected for the screening (Fig. 5C). Specifically, ZIKV-specific antibodies from > 4mpi were only able to recognize the E1 peptide making it the immune-dominant peptide (Fig. 5D). These five screened peptides were next mapped on to the ZIKV E protein structure shown in either its side view (

Discussion
While larger Old World monkeys continue to remain as an attractive choice to study pathophysiology of diseases 7 , the common marmoset of the New World monkeys could be another interesting consideration 48 . One of the most important features that makes marmosets appealing is their small frame and weight 48,49 that makes handling easier. In addition, due to the reduced body size, husbandry costs have been estimated to be reduced by approximately 10-15 folds 49 . These animals are also known for their early maturity, great fecundity, short gestation period, which will allows for a quick setup of an in-house breeding colony 49 .
In this report, ZIKV-infected marmosets were free from any visible discomfort or clinical illness despite detection of ZIKV RNA in various body fluids, corroborating earlier studies 50 . These findings also closely reflects ZIKV patients and ZIKV-infected macaques where the majority displayed little or no symptoms 24 plasma, urine and saliva specimens were also consistent with findings in patients 52 . These results further complement findings from earlier reports where ZIKV RNA load could be detected in serum, saliva and urine specimens for up to 2-3 weeks post-infection 50,53 , albeit with some variations in the levels and persistence of the detected ZIKV RNA. These differences could be due to differences in dosage of the virus inoculum, inoculation route, the strain of ZIKV being used 50,53 , and the physiological states of the animals, as one study was done on pregnant marmosets 53 . Importantly, the detection of ZIKV RNA load in urine and saliva specimens accentuate their potential as non-invasive samples for diagnostic purposes 52 , especially when it is reported that ZIKV RNA load can be detected in urine samples for up to 60 days after symptoms onset in humans 54 . MRI has been shown to be effective in detecting fetal brain damage in pregnant macaques 18 , while ultrasound is a sensitive technique often used to evaluate for testicular lesion and inflammation. Therefore, non-invasive brain MRI and testes ultrasound procedures were established for longitudinal follow-up. Interestingly, ZIKV infection in marmosets did not reveal any detectable lesions, inflammation or gross structural changes. Inspite of this, it is worthwhile to note that replicating ZIKV has been found in the testes of immunocompromised mice leading to the shrinkage of the organs as well as overall sperm count and motility 55 . ZIKV RNA has also been found in the seminal fluid of ZIKV-infected adult macaques 28 , marmosets 50 and men 56,57 . ZIKV infection dynamics in the immune-privileged testes of these larger animals should be studied to better understand any long-term impairments of the organ. MRI did not reveal any significant brain abnormalities in ZIKV-infected marmosets as these animals were immunocompetent and like humans, had an intact blood brain barrier. Nevertheless, these techniques are useful as non-invasive follow-up in large animal models. In terms of immunological responses, ZIKV infection did not lead to significant changes in numbers of the majority of peripheral immune cells. The only notable differences were in the B cells, CD8+ T cells, HLADR+ NK cells and CD16+ monocytes. However, it should also be noted that in non-infected marmosets, changes in peripheral immune cell numbers were similarly recorded, making it hard to interpret if the changes observed were a true consequence of ZIKV infection. Nonetheless, changes in NK cells and B cells have been reported in another marmoset study of ZIKV infection 50 . In ZIKV-infected macaques, induction of activated CD4+ T cells was reported 21,23,29 , which was not observed in the marmosets 50 . In another study, ZIKV-infected rhesus macaques displayed altered levels of peripheral monocytes and dendritic cells 31 . Reasons for this disparate immune responses between marmosets and macaques could be due to the differences in inoculation routes 58 , virus titer or ZIKV isolates used. In ZIKV-infected humans, a transient decrease in numbers of most peripheral immune subsets was detected during the acute phase of the disease (<7 days post-illness onset), before recovering 40 . In the quantification of soluble immune mediators present in the blood, a lack of significant differences were observed after ZIKV infection (Supplementary Table 2). Further optimization would be required as the NHP microbead kit used in this study may not be the most specific for marmosets, thus resulting in the lack of quantifiable differences during the post infection phases. It would be interesting to point out that similar observation was earlier reported in another cohort of ZIKV-infected marmoset 50 , where majority of the immune mediators assayed were not affected by ZIKV infection. This contrasts findings reported in human patients, where levels of soluble immune mediators present in their plasma were significantly modulated by the infection 40,59 . This further accentuates the urgent need to develop a more sensitive and marmoset-specific microbead assay. Elicitation of robust neutralizing ZIKV-specific antibodies indicates active infection of the marmosets, with high titers of ZIKV-specific IgG being detected up to at least 4 mpi. While it was not performed in this report, the elicited ZIKV-specific IgG should be able to cross-protect against a re-infection by another ZIKV strain 50 . In fact, heterologous ZIKV infection in marmosets further boosted the humoral response 50 . Importantly, these antibodies exhibited minimal cross-reactivity to DENV. Despite inefficient neutralization, ZIKV-specific IgM could be a good diagnostic measure of acute ZIKV infection because these IgM do not cross-react with DENV. The ZIKV E protein is known to be responsible for virus entry 60 and typically covers the entire viral surface 61 . The specificity of marmoset ZIKV antibodies in targeting the solvent-exposed E1 peptide of the E protein, further emphasizes its importance as a therapeutic or diagnostic target (28). Notably, numerous eptiopes are concealed in the ZIKV E protein, rendering it a major target of ZIKV-specific antibodies from patients as well as murine monoclonal antibodies [61][62][63][64] . Nevertheless, screening of other epitopes could also be expanded to include other ZIKV proteins, and such screens should also include ZIKV antibodies derived from other organisms, such as mice and other NHPs. These concerted efforts will allow the identification of universal immunodominant epitopes that could be formulated into peptide-based vaccines 65 .
In vivo primate models should not be restricted to macaques only. The idea of using marmosets as the next ideal primate for studying infectious diseases should be further explored as marmoset-specific tools are being developed. This study demonstrated that marmosets could be utilized to understand infection and immunity and present themselves as a feasible platform for evaluating the efficacy of developed immune-based therapies.

Methods
Study approval. All animals were housed at the animal facility Biological Resource Centre of the Agency for Science, Technology and Research, Singapore (A*STAR), which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International (AAALAC). Common marmosets aged between 101 to 132 weeks were used in this study. All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC No. 161137) of A*STAR, in accordance with the guidelines of the Agri-Food and Veterinary Authority and the National Advisory Committee for Laboratory Animal Research of Singapore. The study was conducted under Animal Biosafety Level 2 (ABSL-2) conditions. All marmosets were screened negative for ZIKV and dengue virus (DENV) by both PCR and serology assays 46,66 prior to the commencement of the study. Housing and husbandry. Animals were pair-housed whenever possible and singly-housed marmosets were given visual, auditory and olfactory access to other marmosets. Marmosets were housed in stainless steel cages of 775 mm × 775 mm × 1750 mm and were always provided with enrichment materials. They received a balanced diet containing commercial pellets (Altromin) and a variety of fresh fruits, eggs and dairy products. The holding room was illuminated with fluorescent lights on a 12 hour light-dark cycle with temperature and humidity maintained at 24-28 °C and 40-70% respectively.

Virus stock. ZIKV strain (accession KJ776791) used in this study was originally isolated from the French
Polynesia outbreak in 2013 67 . The virus was propagated and quantified in Vero-E6 cells (ATCC; CRL-1587) as previously described 68 .
Virus inoculation. Animals were first sedated with ketamine (10-15 mg/kg) and Atropine (0.025-0.05 mg/ kg) via intramuscular injection and maintained on inhalation of anaesthesia using isoflurane provided via a facemask, if sedation was inadequate. 1E5 plaque-forming units (PFU) of ZIKV was prepared in a final volume of 300 μL with PBS and inoculated into the sedated animals via an intravenous catheter placed in the saphenous vein. Immediately following the virus inoculation, 500 μL of saline was infused into the catheter to ensure entry of the entire virus preparation into the circulatory system. Non-infected animals received an equal volume of PBS and saline without virus.

Physiological assessments. Clinical observation. Marmosets were observed for clinical signs and behav-
iour changes such as reduced appetite, lethargy, weakness, signs of joint pain, limping, conjunctivitis and changes in stools. This clinical assessment was conducted by facility veterinarians daily for the first 14 days of infection, twice weekly up to day 30 then weekly for the remaining period of study.
Sample Collection (blood, urine, saliva). Samples were collected on 2, 6, 15, 30, 60, 90, and 120 dpi. Urine samples were collected from restrained marmosets by gentle manual massage of the lower abdomen into Eppendorf tubes and kept on ice prior to analysis. Blood and saliva collection were performed under light sedation with ketamine (10-15 mg/kg) and Atropine (0.025-0.05 mg/kg) via intramuscular injection. Blood samples were collected into EDTA tubes via venipuncture from femoral veins. Volume collected did not exceed 10% total blood volume weekly. Blood was subsequently spun down at 1200 rpm for 10 minutes for plasma isolation. Saliva samples were collected by swabbing the oral cavity with a sterile swab and placing into 300 μL of RNAlater ® solution (ThermoFisher Scientific).
Magnetic Resonance Imaging of the Brain. MRI brain was performed on the Siemens Skyra 3T scanner.
A total of 11 marmosets (n = 5 non-infected, n = 6 ZIKV-infected) were imaged. Imaging schedule coincided with sample collection schedule. Marmosets were scanned under anaesthesia. The imaging protocol consisted of axial T2-weighted fast spin-echo (FSE), coronal T2-weighted FSE, fluid-attenuated inversion recovery (FLAIR) with fat saturation, readout-segmented diffusion-weighted imaging (DWI) 70 , and pre-and post-contrast T1-weighted MPRAGE. Between 1 to 4 months after the last time point, additional pre-contrast MPRAGE and MP2RAGE images 71 were also obtained to aid automated volume quantitation (see Supplementary Table 1 for details of scan parameters). The body volume coil was used for transmission and a 4 cm single-channel surface coil for reception, which was positioned over the vertex of the marmoset head and secured by a custom rig. Apparent Diffusion Coefficient (ADC) maps were generated from DWI images. T1 maps were generated from the MP2RAGE data, to which each MPRAGE image was registered for segmentation of gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), and whole brain volume at every time point. Gadolinium contrast was administered intravenously (Magnevist 0.2 mmol/kg). Images were evaluated by a radiologist (co-author K.C.L) for any pathophysiological changes. MRI brain volume analysis. Automated image analysis was performed on MRI to yield brain volumes.
As conventional brain extraction methods (FSL Brain Extraction Tool 72 ) were ineffective due to inhomogeneity arising from sensitivity variations in the surface coil, an atlas-based brain extraction method robust to such inhomogeneity was optimized. This was performed in 3 phases: pre-processing, extraction of brain and segmentation of the grey matter (GM), white matter and cerebrospinal fluid (CSF). During the pre-processing, bias field inhomogeneity correction using the Computational Morphometry Toolkit (CMTK) 73 (http://nitrc.org/projects/cmtk) was performed. In the second phase, a marmoset brain atlas 74 was used to extract the brain using linear registration (atlasBREX, FSL 5.0) to yield total brain volume. Subsequently, segmentation into GM, WM and CSF was obtained by deriving a probability map from the calculated T1 map. The masks of GM, WM and CSF obtained from the T1 map were transformed to MPRAGE at each timepoint using linear and nonlinear registration (  45 . Amount of specific immune subset present in per microliter of whole blood is back-calculated by multiplying its percentage of total CD45+ cells by the absolute leucocytes count determined via CBC. Subsequently, fold change differences in calculated cell numbers were expressed relative to the baseline levels. Virion-and peptide-based ELISA. Presence and titers of ZIKV-specific antibodies in the plasma of infected marmosets (n = 5 non-infected, n = 6 ZIKV-infected) were determined with a ZIKV virion-based ELISA as previously described 76 . Briefly, purified ZIKV were coated on 96-well microtiter plates (Maxisorp, NUNC). Wells with coated virions were then blocked with 5% skim milk (Nacalai Tesque) in 0.05% Tween-20 (Sigma-Aldrich) PBS for 1.5 hours at 37 °C. Diluted (1:50) heat inactivated marmoset plasma samples were then added and incubated for 1 hour at 37 °C. Secondary antibody, horse-radish peroxidase (HRP)-conjugated goat anti-human IgM (diluted 1:2000) (Invitrogen) or anti-human marmoset IgG (diluted 1:5000) (Invitrogen) was subsequently incubated for 30 minutes at 37 °C before luminescence reactions were developed with 3,3′, 5,5′-Tetramethylbenzidine (TMB) substrate (Sigma-Aldrich). Absorbance was then measured at 450 nm with the Tecan Infinite M200 plate reader (Tecan). Samples were assayed in duplicates. Individual samples were assayed in duplicates. Data obtained (after subtracting signals from non-specific binding and secondary antibody) is presented as fold change in antibody titers relative to baseline levels. DENV virion based Elisa is performed in the same approach. For the peptide-based ELISA, linear epitopes of marmoset ZIKV-specific antibodies were screened as previously described 76 , with synthesized biontinylated peptides (EMC) consisting of 18-mer overlapping peptides generated from the ZIKV Polynesian isolate (KJ776791). Peptides selected for the screening are found on the ZIKV E protein and these region have previously been reported to be targets of ZIKV-specific antibodies of ZIKV-infected patients 47 . Briefly, streptavidin-coated plates (Pierce) were blocked overnight with 0.1% Tween-20, 1% Bovine Serum Albumin (BSA) and 1% Casein (all from Sigma-Aldrich) in PBS. Diluted peptides (1:1000 in 0.1% Tween-20 in PBS (PBST)) were then incubated for 1 hour at room temperature, before heat inactivated pooled marmoset plasma samples (n = 5 non-infected, n = 6 ZIKV-infected) were added at dilutions of 1:50, 1:100, 1:200 and 1:400 and incubated for 1 hour at room temperature. Secondary antibody, horse-radish peroxidase (HRP)-conjugated goat anti-marmoset IgG (diluted 1:2500) (Invitrogen) was then added and incubated for another 1 h at room temperature before TMB substrate (Sigma-Aldrich) was added to develop the luminescence reactions. Thereafter, absorbance was measured at 450 nm with the Tecan Infinite M200 plate reader (Tecan). Pooled plasma sample were assayed in triplicates. Epitope modeling was done on the ZIKV E protein structural data obtained from the Protein Data Bank (PDB) (ID: 5IZ7) and visualized using the PyMOL Molecular Graphics System (http://www.pymol.org). Epitopes mapped on the ZIKV E protein are illustrated in either the side or or top (solvent-exposed) view. with the secondary goat anti-rabbit secondary antibody conjugated to fluorescein isothiocyanate (FITC) (Life Technologies) for 1 hour at 37 °C. Stained cells were resuspended in a final volume of 60 μL of PBS and acquisition was performed with the MACSquant ® Flow Cytometer (Miltenyi Biotec). Neutralization capacity is calculated as the percentage of infection in HEK293T cells relative to virus-only (positive) control.

Multiplex microbead immunoassay for cytokine quantification. Cytokine and chemokine levels
in in the plasma of marmosets (n = 5 non-infected, n = 6 ZIKV-infected) were measured simultaneously using a multiplex microbead-based immunoassay, ProcartaPlex NHP Cytokine/Chemokine/Growth Factor 37-plex (Thermo Scientific). Additionally, 4 replicates of the plasma of control marmosets (n = 5) were ran on all plates to allow plate normalization. Preparation of reagents as well as immunoassay procedures were performed according to manufacturers' instructions. Luminex concentrations obtained via the Bio-plex Manager software (using 5 parameter logistic curve fitting) were normalized using median centering to remove potential plate effects. The final adjusted concentrations were then logarithmically transformed to assume normality prior to further data analysis and visualization.
Statistics. Data analysis for multiplex microbead immunoassay was conducted using one-way ANOVA with post hoc Dunnet's test to detect for differences between the two groups of animals across the various collections. Results were corrected for multiple testing using the method of Benjamini and Hochberg. Statistical analysis on MRI brain volume, testicular volume derived from vernier caliper and ultrasound was performed using a general linear mixed model with repeated measurements. Marmosets were modeled as random effects, while group (infected, control), follow-up imaging (2, 6, 15, 30, 60, 90,120 dpi) and group by day interaction were modeled as fixed effects after controlling for differences in body weight. The objectives were to determine, (1) if there was an overall statistically significant difference between groups, and (2) if the difference between groups changed significantly with the passage of time (Group x Day interaction). Statistical analysis between ZIKV-infected and non-infected marmosets for changes in peripheral blood cell numbers was determined using non-parametric, Mann Whitney test, two tailed. P value < 0.05 is considered significant.