Neurocognitive impact of Zika virus infection in adult rhesus macaques

Animals were housed at the AAALAC International-accredited, Armed Forces Research Institute of Medical Sciences (AFRIMS) in Bangkok, Thailand. The protocol was approved by the AFRIMS Institutional Animal Care and Use Committee. Research was conducted in compliance with Thai laws, the Animal Welfare Act and other U.S. federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, 2011 edition [36].

Eight adult, Indian-origin rhesus macaques (Macaca mulatta, 4 males, 4 females), median age 6 (range 5–7) years, median weight 8.2 (range 5.8–10.3) kg were included in the study. Prior to entry into the study, animals were screened to confirm normal physical condition, complete blood count, blood chemistry and sero-negativity to mosquito-borne Flaviviruses endemic to Thailand, including DENV (serotypes 1, 2, 3 and 4), JEV and ZIKV. Animals were trained to undergo neurocognitive assessment for 24 weeks prior to ZIKV inoculation at day 0. Animals were then divided into two groups and were humanely euthanized early (day 8 PI, n = 4) or late (day 28 PI, n = 4). In both groups, plasma ZIKV RNA was monitored daily for 6 days PI. Animals in the late group also underwent CSF collection at days 4, 14 and during necropsy and lymph node collection at day 4 and during necropsy (Fig. 1A). Characteristics of the animals including age, sex, weight and group assignment are listed in Fig. 1B.

The Monkey CANTAB Intellistation with Pellet Reward (Lafayette Instrument, Lafayette, IN, USA) was used to perform neurocognitive testing. In this study, rhesus macaques were assessed using the Motor Screening Task (MOT) and Self-Ordered Spatial Search (SOSS). The MOT assesses sensorimotor deficits and reaction speed. The SOSS assesses frontal lobe working memory functions. Both of these tests have been widely used in humans and monkeys, across a range of ages [35, 37‐40] and pathological conditions, including neurodegenerative [41, 42], neuropsychiatric [43], infectious [44, 45] and the effects of psychotropic agents [46‐49].

The animals were trained to undergo CANTAB assessments 2–3 times a week, in 5–25 min afternoon sessions, for 24 weeks (Additional file 1: Fig. S1). Step 1: To familiarize the animal, CANTAB Intellistation was brought in front of the animal’s cage and the animal was allowed to explore it and take rewards from the dispenser in response to an auditory stimulus from a manual clicker. Step 2: A visual stimulus, filling the whole screen was presented and the animal was rewarded for touching the screen. Step 3: The visual stimulus reduced in size iteratively following 10 consecutive correct responses. The animal was rewarded after each correct response. Step 4 (Motor Screening Task, MOT): A colored stimulus of fixed-size appeared on the screen in random locations and the animal received a reward after touching the stimulus. Animals were given 10 min to complete as many motor tasks as possible. Accuracy on the motor task was calculated as the percentage of correct responses divided by the total responses; and speed was determined by the number of tasks the animal was able to complete within 10 min. Step 5 (Self-Ordered Spatial Search, SOSS): A set of visual stimuli were presented together at the start of each individual trial. The animal was required to touch each stimulus without returning to a stimulus already selected. Trials were defined as completed when the animal either touched all of the targets without repetition (correct), touched a target that had previously been selected in that trial (error), or failed to touch a target within 30 s (omission). After an inter-trial interval of 5 or more seconds, a new trial was presented with stimuli randomly assigned to a new location on the screen. A blank screen was presented briefly after each response. Animals were required to complete 40 SOSS tasks. Accuracy on SOSS was calculated as the percentage of correct responses divided by the total responses; and speed was determined by the time taken to complete 40 tasks.

Sex differences were observed in the rate of learning between male and female animals for SOSS training and, thus, required a shift in the presentation of tasks and rewards to motivate the female animals.

This task assessed fine motor speed and dexterity. It involved the retrieval of raisins from a board that contained wells oriented in different directions (i.e., vertically, horizontally and diagonally) with different depths and angles [33, 50, 51]. The task was administered using two boards that differed in task complexity vis-à-vis increasing the depth of wells, altering the angle of the wells and increasing the number of wells (from 6 wells on the first board to 9 wells on the second board). Animals were assessed once pre-infection and twice per week post-infection, on the same day as the CANTAB assessment. The dependent variables included the number of raisins retrieved and the time taken to clear the board of raisins.

Hemagglutination inhibition (HAI) assays were used to screen for flavivirus sero-positivity by measuring of DENV1-4, JEV and ZIKV antibodies in serum collected 2 weeks prior to study entry. Serum was heat inactivated at 56 °C for 30 min and non-specific protein binding was reduced by acetone extraction. Two-fold serially diluted serum samples were added to individual sucrose–acetone extracted viral antigens. Hemagglutination was detected by the addition of goose red blood cells. HAI titer was defined as the highest dilution of serum that inhibited hemagglutination. Only NHPs with HAI titer < 10 for all viral antigens were included in this study.

Infectious ZIKV was quantified by plaque assay using LLC-MK2 cell. Viral supernatant was ten-fold serial diluted, starting from 1:10 to 1:10,000, with tissue culture media, Minimum Essential Media (MEM) supplemented with 10% heat inactivated FBS (Invitrogen, USA). Then, 100 µL of diluted sample was added onto triplicate wells of LLC-MK2 cell monolayer in 12-well plates. After incubation for 1 h at room temperature (RT) on a rocker platform, the excess amount of inoculum was removed. First overlay medium with 1.8% LMP was added and the agar was allowed to solidify at RT prior to incubation at 35 °C, 5% CO₂ for 4 days. Virus-infected cells or plaque forming units (PFU) were visualized after staining with second overlay medium containing 4% Neutral red (Sigma, USA) on the following day. The average number of plaque from triplicate wells in each dilution was used to calculate ZIKV titer per milliliter (PFU/mL).

ZIKV PRVABC59 strain (stock titer of 1.8 × 10⁷ PFU/mL, provided by R. De La Barrera, WRAIR) was used [52]. Animals were inoculated with 1 × 10⁶ PFU subcutaneously in the left leg at week 0. Animals were observed three times per day for clinical signs of disease including inappetence, dehydration and lethargy. Intra-rectal temperature was measured daily using pole collar chair restraint for 10 days after ZIKV inoculation.

Plasma, CSF and tissue ZIKV-RNA levels were measured using real-time quantitative PCR [53, 54]. Viral RNA was extracted from 140 µL plasma and CSF using the QIAamp viral RNA mini kit (QIAGEN, Germany). For tissue samples, RNeasy Mini Kit (QIAGEN, Germany) was used as per the manufacturer’s instruction. ZIKV real-time quantitative RT-PCR was performed using a method modified from Lanciotti et al. [55]. Two primer/probe sets were used: (1) ZIKV 1086 forward, ZIKV 1162c reverse primers and ZIKV 1107-FAM probe [56]; and (2) ZIKV 4434 forward, ZIKV 4524c reverse primes and ZIKV 4479c-FAM probe [57]. In each experiment, 3 internal controls (no template control, NTC, negative extraction control, NEC and positive extraction control, PCE) and 6 in vitro transcribed RNA standards (at 5, 50, 5 × 10², 5 × 10³, 5 × 10⁴, 5 × 10⁵ copies/reaction) were included. All assays were performed using the SuperScript III Platinum One-Step Quantitative RT-PCR kit (Invitrogen), as per manufacturer’s instructions, using the Applied Biosystems 7500 Fast Real-Time PCR systems (Life Technologies). Limit of quantification of the assay was 5 copies. The amount of ZIKV in GE/mL in controls and samples were calculated using the formula GE/ml = (copies per reaction x eluted RNA vol × 1000 ul)/ (RNA vol used in the reaction (ul) x serum vol used (ul)).

Plasma and CSF soluble markers of immune activation (including G-CSF, GM-CSF, IFNγ, IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12/23(p40), IL-13, IL-15, IL-17A, MCP-1, MIP-1β, MIP-1α, sCD40L, TGF-α, TNF-α, VEGF, and IL-18) were quantified using the MILLIPLEX MAP Non-Human Primate Cytokine Magnetic Bead Panel (EMD Millipore Corporation, Billerica, Massachusetts, USA) as per manufacturer instructions, as previously described [58].

Cells were stained with Aqua Live/Dead dye (Invitrogen, Eugene, OR, USA) in the presence of 10% mouse IgG (Fc block; Invitrogen, Eugene, OR, USA). Subsequently samples were stained for 20 min at 4 °C with the following antibodies: anti-CD3 PE-CF594 (BD Horizon, San Diego, CA, USA), anti-CD4 Pacific Blue (Biolegend, San Diego, CA, USA), anti-CD8 BV785 (Biolegend, San Diego, CA, USA), anti-CD19 BV605 (Biolegend, San Diego, CA, USA), anti-CD20 BV711 (Biolegend, San Diego, CA, USA), anti-CD16 PE-Cy5 (Biolegend, San Diego, CA, USA), anti-CD56 PE-Cy7 (BD Bioscience, San Jose, CA, USA), anti-CD14 BV650 (BD Horizon, San Diego, CA, USA), anti-CD38 PE (Caprico Biotechnologies, Norcross, GA, USA), and anti-HLA-DR APC-H7 (BD Pharmingen, San Diego, CA, USA). They were then permeabilized with 1X BD Perm/Wash buffer (BD Bioscience, San Jose, CA, USA) for 15 min at 4 °C before staining with anti-Ki67 AlexaFluor488 (BD Pharmingen, San Diego, CA, USA) for 30 min at 4 °C. After staining cells were resuspended in 1% formaldehyde and acquired immediately using a custom-built LSRFortessa flow cytometer (BD, San Jose, CA, USA) and analyzed using FlowJo software version 10.5.3 or higher (TreeStar, Ashland, OR, USA).

Tissue samples were fixed in 4% paraformaldehyde for 16–24 h, processed in an automatic tissue processor (TP1020; Leica, Buffalo Grove, IL, USA), paraffin embedded, sectioned at 5 um, and mounted onto adhesive microscope slides (BioGnost, Medugorska, Zagreb, Croatia).

Slides were stained with Harris’s hematoxylin solution (In-House Prep.) for 10 min, washed and stained with Eosin solution (In-House Prep.) for 45 s, dehydrated with ethanol, and cleared with xylene prior to mounting with toluene-based mounting medium (Thermo Fisher Scientific, Kalamazoo, MI, USA).

Slides from basal ganglia and parietal areas of the brain were reviewed by a board-certified veterinary pathologist, in comparison to similar tissue sections from control animals.

Analysis of virologic and immunologic data was performed using GraphPad Prism v8.4.3 (GraphPad Software). Comparisons between different time-points in the same animals were performed using Wilcoxon matched-pairs signed rank test. Pre-infection CANTAB performance was established by generating mean and standard deviation (SD) using the last 4 sessions of assessment prior to ZIKV inoculation (Additional file 2: Table S1). To quantify changes post-infection, individualized Z-scores (number of SD below or above the pre-infection mean) were generated using data from each animal post-ZIKV inoculation. This method of quantification of neurocognitive impairment was based on the work of Weed et al. [44] and is advantageous as it takes into account individual variability in performance in each animal pre- and post-infection. Impairment was defined as > 2.5 SD (corresponded to 99% confidence interval around the baseline mean) reduction in accuracy for both MOT and SOSS, > 2.5 SD reduction in the number of MOT performed within 10 min or > 2.5 SD increase in time to complete 40 SOSS tasks.

Eight rhesus macaques were infected with ZIKV through subcutaneous ZIKV inoculation. Apart from transient reduction in appetite, animals did not exhibit any signs of clinical illness including fever (intra-rectal temperature > 103°F).

All animals had detectable ZIKV RNA in their plasma at day 1 PI that peaked at day 2 PI (median 7.55 × 10⁵, IQR 4.1 × 10⁵–1.69 × 10⁶, GE/mL). Plasma ZIKV RNA declined rapidly and became undetectable in plasma by day 8 PI in all animals in the Early group and day 14 in all animals in the Late group (Fig. 2A). ZIKV RNA was detectable in lymph node tissues in 4 of 4 animals at day 4 (Late group) and at day 8 (Early group) and persisted in 3 of 4 animals (Late group) at day 28 PI (Fig. 2B). ZIKV RNA was not detected in CSF supernatant at any of the time points measured (day 8 in n = 4 Early group, days 4, 14 and 28 in n = 4, Late group, Fig. 2B). Of the 15 different areas of the nervous system assessed (Fig. 2C), only low levels of ZIKV RNA were detected in the meninges (R1301 at 228 GE/mg at day 28 PI and R1435 at 60 GE/mg at day 8 PI) and the parietal region (R1301 at 166 GE/mg at day 28 PI).

Significant increases in a number of soluble markers of immune activation including IL-15 (p = 0.0078), MCP-1 (p = 0.0156), IL-1 RA (p = 0.0156), and IL-2 (p = 0.0078) were observed in plasma at day 4 PI (Fig. 3), that then reduced and normalized by day 14 PI. Similar trends were observed in the CSF and were statistically significant for IL-15 (p = 0.0156), MCP-1 (p = 0.0156) and G-CSF (p = 0.0391) when results from animals in the Early (n = 4, day 8 PI) and Late (n = 4, day 4 PI) groups were combined (Fig. 3). Levels normalized by day 14 PI. In addition, a pattern of reduction in CSF IL-10 levels was also seen post-infection (Fig. 3). When compared to pre-infection, significant increases in cellular markers of immune activation were also identified in CD4 (Ki67 + , p = 0.0156 and CD38 + HLA-DR + , p = 0.0156, Fig. 4A, B) and CD8 (Ki67 + , p = 0.0234, Fig. 4C) T cells in the peripheral blood and in CD8 (CD38 + HLA-DR + , p = 0.0078) T cells in the CSF (Fig. 4D), when data from animals in the Early (n = 4, day 8 PI) and Late (n = 4, day 4 PI) groups were combined. Interestingly, CD4 and CD8 T cell activation mostly resolved in the peripheral blood by day 14 PI but CSF CD38 + HLA-DR + CD8 T cell frequency peaked at day 14 PI and remained elevated in 2 of 4 animals (when compared to each animals’ pre-infection level) at day 28 PI. Thus, ZIKV infection was associated with early immune activation in both the peripheral blood and the CSF, but largely normalized by day 28 PI, with the exception of CSF CD8 T cell activation.

Hematoxylin and eosin-stained slides from the basal ganglia and parietal areas of each animal were reviewed by a board-certified veterinary pathologist. These sections were chosen as they represent superficial and deep parts of the brain. No pathological findings were observed.

Pre-infection CANTAB performance for each animal was established as the mean of the scores from the last 4 sessions of assessment prior to ZIKV inoculation (Additional file 2: Table S1). Median (IQR) of the pre-infection scores for all 8 animals were 80.48% (72.01–88.13) for MOT accuracy, 69.38% (49.53–82.97) for SOSS accuracy, 71.25 tasks (68.25–74.75) for the number of MOT performed within 10 min and 6.54 min (6.33–7.25) for the time required to complete 40 SOSS tasks.

To assess changes post-infection, individualized Z-scores (number of SD below or above the pre-infection mean) were generated for each animal at each assessment post-ZIKV inoculation (Fig. 5A). Impairments defined as > 2.5 SD change were identified in 7 of 8 animals (Fig. 5B). Interestingly, female animals displayed more impairments (Fig. 5B); although this may be explained by learning style and performance approach rather than impairment per se. Importantly, impairment in SOSS accuracy followed the peak of CSF CD8 T cell activation at day 14 PI in 3 of 4 animals in the Late group (Fig. 5C), suggesting a potential link between neuroinflammation and neurocognitive impairment. The correlation is not statistically significant and larger experiments would be needed for additional validation of this observation.

On the modified Brinkman Board, all animals retrieved all raisins from the two boards in all assessments. No slowing was noted in the majority (6 of 8) of animals (Fig. 6). Animal R1301 at day 2 PI showed a marked increase in time to complete the more complex board with 9 wells, but that was not mirrored in the more simple board with 6 wells in the same test session. Animal R1325 showed an increase in time to complete the simple and complex boards at day 17 PI however, no impairment on CANTAB testing was noted on that day. Thus, these findings suggest that post-infection motor performance was unaffected as assessed by the modified Brinkman Board.