Immune cell dynamics in rhesus macaques infected with a Brazilian strain of Zika virus

Abstract

Zika virus (ZIKV) is a mosquito-borne and sexually transmitted flavivirus, associated with fetal CNS-damaging malformations during pregnancy in humans. This study documents the viral kinetics, and immune responses in rhesus macaques infected with a clinical ZIKV Brazilian isolate. We evaluated the viral kinetics and immune responses induced after an i.v. infection with a Brazilian ZIKV clinical isolate (HS-2015-BA-01) in rhesus macaques for up to 142 days. ZIKV-specific antibody-secreting cells (ASCs), germinal center (GC) reactions as well as monocyte, DC, NK and T cell frequencies were monitored. ZIKV loads were readily detected in plasma (until day 5 or 7), semen and urine (until day 7 and 14), and saliva (until day 42), but the viremia was rapidly controlled. No detectable clinical manifestations were observed. However, lymph node (LN) hyperplasia was clearly visible post viremia, but associated with low frequencies of ZIKV-specific ASCs in LNs and bone marrow (BM), correlating with low antibody titers. CD14+/CD16- monocytes and myeloid CD11c^hi DCs decreased in blood, while NK and T cell numbers were only marginally altered during the course of the study. ZIKV infection caused a significant lymphoid tissue activation but limited induction of ZIKV-specific B cells, suggesting that these parameters need to be considered for ZIKV vaccine design.

INTRODUCTION

Zika virus (ZIKV) transmission to humans occurs primarily via mosquito bite, though sexual transmission has been documented (1–3), broadening concerns about the potential of a developing epidemic. The virus was first reported in sentinel rhesus macaques from the Zika forest, Uganda in 1947. A few years later, serologic evidence of ZIKV human infection was detected in several African countries. Due to the few symptomatic cases listed, little clinical importance was given to this infection until the 2007 and 2014 outbreaks in French Polynesia and Brazil respectively (reviewed in (4)).

During those recent ZIKV outbreaks, an increased number of clinical manifestations have been reported, such as microcephaly and arthrogryposis in newborns (5–7) as well as Guillain-Barré syndrome and macula atrophy in adults (8–10). Accurate diagnostics are critical to measure the size of the epidemic and to validate strategies to inhibit viral transmission. Diagnostic kits based on ZIKV-specific monoclonal antibodies have been manufactured, but while positive serology has been an important retrospective diagnostic tool (11), quantitative RT-PCR has been preferably chosen to confirm acute ZIKV infection (12, 13). This strategy has allowed for detection of the viral ability to infect multiple tissues with preferential tropism for myeloid cells (14–16).

Animal models of Zika infection have been established in rodents and nonhuman primates (16–21). Moreover, ZIKV isolates from different locations have been used to characterize animal models and these isolates appear to present distinct patterns of pathogenicity. Recently, it was demonstrated that a ZIKV Brazilian isolate induced significantly higher pathogenicity in neural cells than an African strain (23). Sequencing data revealed that American isolates are more similar to Asian than to African isolates (24). Recent findings in experimental infections of macaque monkeys, show that ZIKV RNA is found in diverse fluids and tissues, but are short-lasting in blood. However, ZIKV loads may last longer in saliva, urine or semen and tissues (16–18, 21, 22). Recently, the extended persistence of ZIKV particles in the testes and epididymidis of IFNar1 KO infected mice was correlated with inflammation, damage of the male genital tract, culminating with lower testosterone levels and infertility (19, 21).

Several experimental ZIKV-specific vaccines and therapeutics have already been developed (25–27)(25–29). Nonetheless, ZIKV-infection elicited immune responses have not been fully described yet. Recent studies showed monocyte and NK cell responses in the first week after infection in macaques as well as ZIKV-specific T cells. Among B cells, total plasmablasts had increased frequency between days 7 and 10 of infection (17, 18). At that time, serum ZIKV NS1-specific IgM and IgG titers were already detected (16, 30).

Thus, in this manuscript we evaluated the unexplored aspects of the innate and adaptive immune responses induced by infection with a primary clinical ZIKV Brazilian isolate in rhesus macaques.

METHODS

Animals and virus challenge

Four male rhesus macaques of Indian origin (A7R078, A7R014, A7L045 and 00-R027) were selected from the breeding colonies of the New Iberia Research Center (NIRC), University of Louisiana at Lafayette (ULL, USA) and housed there for the entire study. Animal ages ranged between 9 (A7R014, A7R078 and A7L045) and 17 (00-R027) years. They were cared for in conformance to the guidelines of the Committee on the Care and Use of Laboratory Animals. All experimental protocols and procedures were reviewed and approved by the ULL Animal Care and Use Committee.

Brazilian ZIKV stocks were provided by Dr. Mauro M. Teixeira, Federal University of Minas Gerais, Brazil, and the challenge strain represented passage number 1 from the clinical sample. C6/36 cells were used to expand that virus and its initial concentration was 1.75 × 10⁹ PFU /mL before challenge. Five × 10⁷ PFUs were diluted in 5 mL final volume of PBS and each animal received 1mL via i.v. injection. Peripheral blood mononuclear cells (PBMCs), bone marrow (BM) aspirates and lymph node (LN) and spleen biopsies were obtained at various time points (Figure 1a). Mucosal fluids were collected via Weck-cel collectors as previously described (31).

Figure 1.

Scheme of ZIKV virus challenge and tissue sampling and short ZIKV detection rhesus macaque plasma. A) 107 PFUs of ZIKV (HS-2015-BA-01 strain - GenBank Acession Number - KX520666.1) were administrated via intravenous route in 4 rhesus macaques. Multiple tissues and fluids were collected in indicated time points. B) Isolated plasma samples had their total RNA reverse-transcribed before performance of ZIKV-specific quantitative PCR assay at different time points.

Viral load determinations

ZIKV viral loads were measured based on ZIKV RNA copies per milliliter of plasma, urine and semen. Total RNA was extracted with QIAamp Viral RNA MiniKit (Qiagen, Valencia, CA). Quantitation of viral RNA was performed using the Taqman RNA-to-Ct 1-Step Kit (Thermo Fisher Scientific, Waltham, MA), according to the manufacturer’s protocol on an Applied Biosystems 7500 Real-Time PCR system (Thermo Fisher) using ZIKV-specific primers and probes as previously described (13). Viral loads (copy/ml) were estimated from a standard curve generated using genomic RNA of a Puerto Rican ZIKV strain (PRVABC59 - Centers for Disease Control and Prevention) (BEI resources).

ELISPOT and ELISA

ELISPOT and ELISA were performed as previously described (32)(32). Ninety six well plates were coated with 100 μl of 2.5 μg/mL and 0.5 μg/mL of ZIKV NS1 and domain III of Env (EDIII) proteins (supplied by Luis C. S. Ferreira from The Vaccine Development Laboratory at the University of São Paulo, Brazil) for ELISPOT and ELISA respectively.

Cell staining

Surface and intracellular staining were performed in PBMCs and LN cells as previously described with the following antibodies (clones): Live/Dead (Thermo Fisher); anti-CD3 (SP34-2), anti-CD14 (M5E2), anti-CD123 (7G3), anti-CD80 (L307.4), anti-Ki67 (B56), anti-CD8 (RPA-T8), anti-CD4 (L200), anti-CCR7 (3D12), anti-CD21 (B-Iy4), anti-CD56 (B159), anti-CD95 (DX2), anti-IFNγ (B27) (BD); anti-CD16 (3G8), anti-CD11c (3.9) and anti-PD1 (EH12.2H7) (Biolegend); anti-CD20 (Leu-16) (Becton Dickinson); anti-HLA-DR (Tu36) (Life Technologies); anti-CD19 (CB19) (Abcam); anti-CD38 (AT-1) (Stem Cell Technologies); anti-CADM1 (3E1) (MBL International Corp); anti-CD28 (CD28.2), anti-NKG2D and anti-NKp44(Z199) (Z231) (Beckman Coulter); anti-CXCR5 (MU5UBEE) (eBioscience); Donkey anti-chicken IgG (polyclonal) (Jackson Immunoresearch Labs); anti-IgD (polyclonal) (Southern Biotech); anti-Perforin (Pf-344) (Mabtech). Samples were acquired on a BD Fusion 14 color cytometer (Mountain View, CA) and analyzed using FlowJo software version 9.9.3 (Tree Star, Ashland, OR).

Immunohistochemistry

Immunofluorescence staining was performed on serial sections for CD20, Ki67, CD3, PD1, CXCR5 and DAPI dye, using a modified protocol previously described (33). Briefly, 4 μm paraffin-embedded tissue sections were subjected to deparaffinization in xylene, rehydration in graded series of ethanol, and rinsing with distilled water. Heat-mediated epitope retrieval was performed with DIVA decloaker, followed by blocking with 10% normal donkey serum (Jackson ImmunoResearch) for 1 hour. Sections were incubated with optimized concentrations of rabbit antihuman-Ki67 (clone SP6, Abcam), mouse antihuman CD20 (clone L26, Dako), Rat anti Human CD3 (clone CD3-12, BIO-RAD), goat polyclonal antihuman PD-1 (AF1086, R& D systems) and rabbit polyclonal anti CXCR5 (HPA042432, Sigma) overnight. Thereafter, the sections were washed and incubated with conjugated secondary Abs (Alexa Fluor 488/568/647, Abcam) at room temperature for 1 hour. Following incubations, the slides were washed twice with PBS-FSG-Tx100 for ten minutes. Upon completion of immunofluorescence staining, the sections were mounted with ProLong® Gold antifade reagent with DAPI (4′,6-diamidino-2-phenylindole) (Life Technologies) as a nuclear counterstain and coverslipped. Images were captured using a Leica SP8 confocal microscope.

RESULTS

Clinical assessment of rhesus macaques challenged with a Brazilian ZIKV isolate

In order to study ZIKV pathogenicity, immune responses and potential clinical symptoms, we inoculated a relatively high dose (10⁷ PFUs) of a Brazilian viral strain (HS-2015-BA-01: GenBank Accession Number - KX520666.1, www.ncbi.nlm.nih.gov/genbank) i.v., similar to our dengue infection model (34) in 4 male Indian rhesus macaques. Multiple samples/tissues, including peripheral blood, saliva, urine, semen, LN biopsies and BM aspirates were collected at different time points (Figure 1A) to monitor viral and immune response dynamics. In spite of the high-dose inoculum, clinical or neurologic symptoms were never detected during the whole period of study and body weights remained stable (data not shown). Hematology and serum chemistry profiles showed individual fluctuations but no clear trend (Supplementary Figure 1).

ZIKV loads detection in blood and other body fluids

Plasma viral loads were positive already on day 1 post infection in 3 of 4 animals (00-R027, A7R014 and A7R078) peaking on day 3–4 with an average 23,379 copies/mL of plasma (10,870 to 74,945 copies/mL, Figure 1B). However, plasma viremia was rapidly resolved and became undetectable at 5–10 days post infection. In fact, only monkey A7L045 exhibited positive viremia from day 5 to 7. Although ZIKV has been reported to be detectable in semen and urine long after clearance of viremia, this was not the case here. ZIKV RNA was found in semen and urine samples until day 7 and 14 respectively (Figures 1C–D). In fact, ZIKV RNA was detectable for extended periods of time in saliva samples (Figure 1E). Aside from plasma samples, ZIKV detection ranged between 70 to 2,200 copies/mL in the other body fluids.

ZIKV-eliciting antibody-secreting cell response and rapid GC formation

The kinetics of various B cell subsets were assessed post-infection including total, memory B cells and ASCs. Total B cells (CD3⁻/CD20⁺) numbers increased only on day 3. Within total B cells, activated (CD3⁻/CD20⁺/HLADR⁺ /CD21⁻/CD27⁺ (35)) and resting (CD3⁻/CD20⁺/HLADR⁺/CD21⁺ /CD27⁺ (35)) memory B cell numbers remained constant until day 70. The overall plasmablast (CD3⁻/CD20⁻/CD14⁻/CD16⁻/CD11c⁻/CD123⁻ /HLADR⁺/CD80⁺ (32)) frequencies in blood, representative of an acute response peaked on day 5 post infection and again on day 42 (Figure 2A). The early plasmablast kinetic contrasts with those reported for infection of monkeys with an Asian (18) or African (17) ZIKV strains, which showed peak responses on days 7 or 10 respectively. The origin of the late plasmablast peak was not fully identified. However, ZIKV positive viral loads were detected in saliva samples from animals A7R014 and 00-R027, on day 42 (160 and 270 copies/mL respectively), suggesting that the second plasmablast response was also ZIKV-related. In an attempt to estimate ASC and antibody specificities, ZIKV NS1 and EDIII recombinant proteins were assayed by ELISPOT and ELISA respectively. Although the Env protein was available only for later time point analyses by ELISPOT, ZIKV NS1- and EDIII-specific ASCs could be enumerated from the LNs, BM and spleen within a month after infection (Figures 2B–C) and at day 142 (Figure 3E), but not from the peripheral blood (data not shown). However, on average, less than 50 NS1-specific IgG-, IgA- and IgM ASCs were detected per million LN and BM cells respectively (Figures 2B–C), which is considerably lower than the values obtained post simian immunodeficiency virus infection (144 and 1102 IgG ASCs in the blood and LNs respectively (Silveira, unpublished data)). Percentages of antigen-specific ASCs relative to total ASCs in LNs represented, on average, 1.74%, 4.9% and 1.3% of IgG, IgA and IgM-secreting cells respectively at day 14. BM ASCs detected on day 28 represented, on average, 0.64%, 0.3% and 0.35% of IgG+, IgA+ and IgM+ ASCs respectively. NS1-specific ASC frequency was maintained in similar values at day 142 in the axillary LNs, BM and spleen (Figure 3E). ZIKV NS1-specific IgG titers became detectable by day 28 in serum samples from 2 of 4 monkeys and increased, on average, almost 2-fold until day 70. Nevertheless, EDIII-specific IgG titers became detectable only by day 56 and remained stable until day 70 (Figure 2D). In this case, the detection of EDIII-specific antibodies may be a consequence of B cell restimulation since ZIKV loads were still found in saliva samples (Figure 1E) from some animals by day 42.

Figure 2.

Total B cells and plasmablasts (blood ASCs) have their frequency increased at early time points after ZIKV challenge with a Brazilian isolate in rhesus macaques. Analytical flow cytometry was performed in total, activated memory, resting memory B cells and total plasmablasts from PBMCs. The enumeration of ZIKV-specific antibody-secreting cells were done at different time points with LN (B) and BM cells (C) based on the recognition of NS1 recombinant protein by ELISPOT. NS1-specific LN wells shown in B (left) were plated with 1.08 × 10⁵ cells, while total Ig wells had 0.12 × 10⁵ cells respectively. BM wells shown in C (left) were plated with 1.125 × 10⁵ cells for NS1 and 0.04 × 10⁵ cells for total Ig wells. E) Representative images showing ASC quantification at day 142 after ZIKV infection, wells were plated with 1.08 × 10⁵ cells for axillary LN cells, while total Ig wells from bone marrow and spleen received 0.36 × 10⁵ cells.

Figure 3.

Rapid germinal center (GC) reactions LNs after ZIKV challenge. A) Representative flow cytometry data showing a sequential gating strategy to enrich for proliferating B cells: 1) Live cells; 2) Lymphocytes; 3–4) Singlets; 5) CD3⁻/CD20⁺ cells; and 6) Ki67+ cells at days 7 and 14 after ZIKV challenge. Proliferating B cells (CD3⁻/CD20⁺/Ki67⁺) and formation of B cell follicles presented increased frequency in the LNs at day 14 by flow cytometry (A) and tissue immunofluorescence (B). C) Representative flow cytometry data showing a Tfh cell gating. During formation of B cell follicles, increased frequency of CD4 and CD8 Tfh cells (CD3⁺/CD20⁻/CD8^−/+/PD1^hi/CXCR5⁺/CCR7⁻) were detected in the LNs at day 14 of challenge by flow cytometry (C) and tissue immunofluorescence assays (D).

Since ZIKV-specific ASC and antibody responses were detected, we investigated their potential origin by evaluating the kinetics of germinal center (GCs) formation in LNs. GCs are characterized by the presence of proliferative B cells (CD3⁻/CD20⁻/Ki67⁺), as the source of newly minted plasmablasts (32), and T follicular helper (Tfh) cells (CD3⁺/CD20⁻/PD1^hi/CXCR5⁺) (36–38). The numbers of proliferating B cells were found to increase from day 7 to 14 (Figure 3A), leading to the rapid formation of B cell follicles (Figure 3B). Simultaneously, higher numbers of CD4⁺ and CD8⁺ PD-1^hi Tfh cells (39) were observed (Figure 3C) in close proximity to the B cell follicle areas (data not shown), occupying relatively larger areas within the LNs (Figure 3D).

ZIKV alters the frequency of monocytes subsets and myeloid dendritic cells

Recently, CD14+/CD16+ monocytes were shown to elicit B cell differentiation into ASCs in the context of Dengue virus (DENV) infection in vitro (40). Thus, 3 monocyte populations (CD14⁺/CD16⁺, CD14⁺/CD16⁻ and CD14⁻/CD16⁺) were assessed in peripheral blood by flow cytometry. Percentages of CD14⁺/CD16⁺ cells increased almost 3-fold on days 2 to 5, in comparison to baseline levels. On the other hand, CD14⁺/CD16⁻ cells displayed increased cell frequencies on day 1, followed by much variation and a drop on day 28, to about half the baseline numbers. Only by day 70 post-infection these cells did return to their baseline frequencies. Of note, CD14⁻/CD16⁺ cells presented only a 3-fold percentage increase in frequency between days 28 and 42 (Figure 4A).

Figure 4.

ZIKV challenge presented differential impact over monocyte and DC subsets from rhesus macaque PBMC. A) After staining PBMCs for CD14⁺/CD16⁻ (left plot), CD14⁺/CD16⁺ (center plot) and CD14⁺/CD16⁺ (right plot) monocytes, their respective percentages were evaluated by flow cytometry at early time points after ZIKV challenge. A) Representative flow cytometry data showing a monocyte (Live⁺/CD3⁻/CD20⁻/HLA-DR⁺) gate based on CD14 and CD16 expressions (left plot). Whereas ZIKV infection does not change the cell frequencies of CD14⁻/CD16⁺ monocytes, there was an increase for CD14⁺/CD16⁺ cells from day 2 to 5 and a continuous decrease for CD14⁺/CD16⁻ monocytes from day 1 to day 28 upon viral challenge. B) Representative flow cytometry data showing a DC (Live CD3⁻/CD20⁻/CD8⁻/CD14⁻/HLA-DR⁺) gate based on CD11c and CD123 expressions. The kinetics of different dendritic cell subsets, total myeloid, myeloid (CD123⁻/CD11c^hi) and (CD123⁻/CD11c^int) and plasmacytoid (CD123⁺/CD11c⁻) were assessed upon ZIKV challenge. Whereas ZIKV infection punctually decreased the total and CD11c^hi mDC frequencies at day 2, the opposite was observed for pDCs after day 10. C) Representative flow cytometry data showing distinct myeloid DCs based on CD16⁺ or CADM1⁺ expression (upper plots) within the respective DC subsets, followed by their kinetics data (lower plots).

Both myeloid and plasmacytoid DC subsets were investigated. Total myeloid CD16⁻ DCs (mDCs) exhibited a temporary decrease in frequency only on day 2, represented mainly by CD11c^hi cells that almost disappeared from peripheral blood (Figure 4B). The frequencies of CD16⁺ or CADM1⁺ myeloid DCs were also assessed. Whereas CD16⁺/CD11c^hi mDC numbers had a continuous drop from day 10 to 42 after major fluctuations, CD16⁺/CD11c^int mDCs displayed increased frequencies from day 3 to 10, followed by a persistent marked reduction. A distinct DC subset (CD11c⁻/CD123⁻/CADM1⁺), described in macaques as equivalent to murine CD8a⁺ cells, sheep CD26⁺ DCs and human CD141⁺ cells (reviewed by (41)(41)) showed increased frequencies on days 3, 5, 21, 42 and 70 in comparison to baseline numbers (Figure 4C). Plasmacytoid DC (pDCs) numbers increased appreciably after day 10 (Figure 4B).

Absolute numbers of both monocyte and DC subsets varied only slightly up to day 28 (last time point for CBC data) (Supplemental Figures 2A–B), suggesting that the kinetics for the various subset based on relative values (percentage) reflected true changes induce by ZIKV infection.

Limited ZIKV effect on NK and total T cells

Since NK cells (CD3⁻/NKG2A⁺/CD8⁺/HLADR⁻) represent one of the initial anti-viral mechanisms, we evaluated the kinetics of 3 distinct subsets (CD16⁻/CD56⁺, CD16⁺/CD56⁻ and CD16⁺/CD56⁺) after infection. However, no major changes were seen for any subset (Figure 5A). A modest decrease in Perforin+/IFNγ- cell numbers within the CD16⁻/CD56⁺ subset from day 1 to 10 (Figures 5B–C) and a transient increase in perforin expression on CD16⁺/CD56⁻ cells were detected on day 14 (Figure 5D).

Figure 5.

ZIKV challenge had limited impact on the frequency of NK cell subsets and their functionality in rhesus macaque PBMC. A) Representative flow cytometry data (left plot) showing a NK cell (CD3⁻/NKG2A⁺/CD8⁺/HLA-DR⁻) gate based on CD16 and CD56 expression as well as their respective kinetics data for the following cell subsets CD16⁻/CD56⁺, CD16⁺/CD56⁻ and CD16⁻/CD56⁻. B) Representative flow cytometry data for Perforin and IFNγ-secreting NK cells and the kinetics data for Perforin⁺ IFNγ⁻ NK cell subsets (C). The level of Perforin expression (MFI values) was measured within the distinct NK cell subsets (D).

In contrast, variations in the subsets of CD4+ and CD8+ T cell frequencies were limited, with exception of naive and memory cells showing opposite transient trends on day 7 (Supplemental Figure 3A). Activated (HLA-DR⁺) memory and effector CD4+ T cells presented marginal, but persistent increases in frequencies from days 3 and 5 on respectively (Supplemental Figures 3B–C). Proliferating (Ki67⁺) CD4⁺ T cells tended to decrease overtime though (Supplemental Figures 3D–E). For CD8+ T cell frequencies, all subsets remained stable throughout the study (Supplemental Figure 4A). Activated and proliferating CD8+ T cells exhibited marginal changes with inverse patterns in comparison to their CD4+ counterparts (Supplemental Figures 4B–E).

DISCUSSION

Rhesus and cynomolgus macaques have been used to study ZIKV pathogenesis induced by this flavivirus infection. Distinct ZIKV isolates, such as African, Asian and South/Central American, have been chosen for animal challenges. As a standard pattern, viral RNA copies have been detected in plasma until day 7 to 14 upon viral challenges (16–18, 25, 30). Although transient viremia spikes were occasionally detected up to day 30 of infection in macaque plasma (22), ZIKV detection in fluids, such as urine or semen, has been negative after that period, in contrast to humans for which ZIKV persisted in these 2 compartments beyond plasma viremia. In fact, our data suggest that saliva may be an important site for longer maintenance of ZIKV particles in macaques. As hypothesized, the relatively high ZIKV inoculum delivered intravenously appeared to result in a faster viremia kinetic, but also resulted in earlier viral clearance from the blood (16, 18), in spite of a viral inoculum 10 times higher than any previously administered to rhesus macaques (16–18, 25, 30). The short viremia observed may also be secondary to the fact that the virus was a primary clinical isolate. It remains to be determined whether the route of infection and inoculum size (challenge dose) influence the magnitude and kinetics of viral replication in the host.

Besides viral loads, we and others showed that ZIKV infection can modulate the relative frequencies of several immune cell subsets (16–18), though variations may be more pronounced in lymphoid organs than in blood. Among B cells, we determined an earlier and a later plasmablast peak for a Brazilian ZIKV strain (Figure 2A) than for African and Asian strains (17, 18) which may be linked to the viral kinetics (Figure 1). However, plasmablast specificity has not been explored upon ZIKV infection yet. It is well known that DENV infection induces a robust, but transient plasmablast response in humans. Moreover, the majority of those plasmablasts are Env-specific and represent more than 50% of all circulating B cells post acute infection (42, 43). Nonetheless, DENV2 infection in monkeys leads to markedly higher viral loads than those observed in the current study (34). However, a direct comparison of plasmablasts from both infections was not possible as the procurement of ZIKV Env was delayed and evaluations of these responses were delayed in our ZIKV study. Conversely, NS1-specific ASCs were not tested in the context of our DENV infections and it is possible that that blood DENV NS1-specific ASCs frequencies be also low to undetectable during that infection. The low percentages of NS1-specific ASCs detected in tissues, such as LN, BM and spleen, suggest that ZIKV NS1 is probably not the most immunogenic antigen for antibody responses. ZIKV NS1-specific ASCs may also be mostly short-lived, which would impact their frequencies. Another potential explanation for the low responses may be the existence of inhibitory mechanisms induced by ZIKV infection. The expression of TACI and BCMA proteins, BAFF and APRIL receptors, have been demonstrated to be critical for the maintenance of ASC and antibody responses (44, 45). RNA-seq data from ZIKV infected neural progenitor cells showed downregulation of TNFRSF13C, another BAFF receptor (46). Considering that ZIKV loads can present a long duration in body fluids and antigen-specific ASCs are still found in low frequencies, it remains to be elucidated whether ZIKV infection also dampens the expression of BAFF receptors in B cells in vivo, impacting the titers of anti-ZIKV antibodies.

Although marked GC formation was noted in LNs at day 14 after viral challenge (Figure 3), the antigen probably does not persist long enough, creating an asynchronous B cell selection that limits the magnitudes of ZIKV specific-ASC and antibody responses. DENV proteins and ZIKV RNA have been found in LNs at early time points after infection (16, 18, 47) concomitantly with GC B cells (47). To the best of our knowledge, our data represents the first description of GC cellular responses for ZIKV infection (Figure 3).

Recently, a monocyte subset (CD14⁺/CD16⁺) was described as able to elicit B cell differentiation into ASCs in vitro in the context of DENV infection (40). Although ZIKV infection increased the frequency of those cells from day 2 to 5 (Figure 4A), though this increase did not correlate with prominent ZIKV-specific ASC response (Figures 2B–C).

Although DCs have been also been reported as ZIKV permissive cells (15, 48, 49), the infection mainly affected CD11c^hi mDCs in vivo (Figure 4B). Among receptors that mediate ZIKV cell entry, DC-SIGN is expressed only in mDCs (reviewed by (50)). pDCs on the other hand secrete high amounts of type I IFN after sensing virus entry reviewed by (51), which inhibits viral replication (15). Recently, pDC-derived type I IFN has been associated with expansion of B regulatory cells (Bregs) (52). Whether Bregs are generated during ZIKV infection in vivo, restricting antigen-specific antibody responses, remains to be investigated.

NK cells have been mainly associated with anti-viral responses through perforin and granzyme secretion to eliminate infected cells. In ZIKV challenged macaques, the frequency of activated (Ki67+) NK cells were reported as increased during peak viremia (16–18). However, we found no significant change in the kinetics of 3 distinct NK cell subsets in our study (Figure 5).

Blood T cell subsets, (naive and memory) showed surprisingly little variation post infection though increased levels of proliferation suggesting a moderate, but active response corroborating previous reports (17, 18). As a matter of fact, ZIKV Capsid and Env-specific cytokine-secreting T cells would represent less than 1% of all T cells (16). This could be secondary to the choice of antigen, as data from DENV infected patients point out that non-structural proteins contain more immunodominant T cell epitopes than structural proteins (reviewed in (53)).

In summary, our ZIKV animal model shows good evidence of virus infection and replication with a Brazilian ZIKV isolate, and the rapid viral kinetics concur with previous reports (16–18). The detailed analysis of multiple immune cell lineages, should contribute to the definition of immune correlates and the model provide a tool for testing the efficacy of vaccines and therapeutics against this viral infection.