Analysis of the T Cell Response to Zika Virus and Identification of a Novel CD8+ T Cell Epitope in Immunocompetent Mice

Zika virus (ZIKV) is an emerging arbovirus of the Flaviviridae family. Although ZIKV infection is typically mild and self-limiting in healthy adults, infection has been associated with neurological symptoms such as Guillain-Barré syndrome, and a causal link has been established between fetal microcephaly and ZIKV infection during pregnancy. These risks, and the magnitude of the ongoing ZIKV pandemic, have created an urgent need for the development of animal models to study the immune response to ZIKV infection. Previous animal models have primarily focused on pathogenesis in immunocompromised mice. In this study, we provide a model of ZIKV infection in wild-type immunocompetent C57BL/6 mice, and have provided an analysis of the immune response to infection. We evaluated the activation of several innate immune cell types, and studied the kinetics, phenotype, and functionality of T cell responses to ZIKV infection. Our results demonstrate that ZIKV infection is mild in wild-type immunocompetent C57BL/6 mice, resulting in minimal morbidity. Our data establish that at the peak of the adaptive response, antigen-experienced CD4+ T cells polarize to a Th1 phenotype, and antigen-experienced CD8+ T cells exhibit an activated effector phenotype, producing both effector cytokines and cytolytic molecules. Furthermore, we have identified a novel ZIKV CD8+ T cell epitope in the envelope protein that is recognized by the majority of responding cells. Our model provides an important reference point that will help dissect the impact of polymorphisms in the circulating ZIKV strains on the immune response and ZIKV pathogenesis. In addition, the identification of a ZIKV epitope will allow for the design of tetramers to study epitope-specific T cell responses, and will have important implications for the design and development of ZIKV vaccine strategies.


Introduction
Zika virus (ZIKV) is an emerging mosquito-borne pathogen that belongs to the flavivirus genus of the Flaviviridae family, and is related to other globally relevant human pathogens including Dengue, West Nile and Yellow Fever viruses. ZIKV was first described in 1947 after it was isolated from a febrile sentinel monkey in the Zika forest region of Uganda [1,2]. Although human infection was reported as early as 1964, the first major ZIKV outbreak did not occur until 2007, when nearly 75% of the population of Yap Island, Federated States of Micronesia became infected in a period of 4 months [3][4][5]. This was followed by an outbreak in French Polynesia in 2013, which marked the first reports of infection-associated neurological symptoms such as Guillain-Barré syndrome and fetal microcephaly [4,5]. The current ongoing pandemic in the Americas has already seen millions of people infected, and an alarming increase in cases of fetal microcephaly in babies born to mothers infected during pregnancy [4,5]. Furthermore, the mounting evidence linking ZIKV infection to birth defects has been deemed sufficient to establish a causal relationship [6]. This includes the attenuation of human neural progenitor cell growth in vitro, detection of ZIKV in the blood and tissues of microcephalic fetuses, the detection of ZIKV-specific IgM antibodies in the cerebrospinal fluid of microcephalic infants, and intrauterine growth restriction and microcephaly in fetuses of pregnant SJL mice infected with a Brazilian ZIKV isolate [7][8][9][10]. Thus, due to the magnitude of the current pandemic and the increase in fetal microcephaly associated with infection, there is an urgent need to develop experimental models for ZIKV to understand infection, tissue tropism, pathogenesis, and the immune response.
Previous studies establishing mouse models of ZIKV infection have primarily been performed in immunocompromised mice; for example, via intracranial infection of suckling mice or in mice deficient in the interferon (IFN)-α/β receptor (IFNAR) or both IFNAR and the IFN-γ receptor (IFNGR) [2,[11][12][13][14]. Although very useful for the characterization of pathogenesis, these mice lack an effective immune response, and as such, have limited potential for analysis of the overall immune response to ZIKV infection and the efficacy of vaccine strategies. Furthermore, in some of these models inoculation with as little as one plaque forming unit (PFU) of ZIKV can be lethal [13], which is not reflective of the natural course of infection in immunocompetent adults, where disease is typically mild and self-limiting [4,5]. The lack of type I IFN also leads to a severely attenuated T cell response, as type I IFN is required for optimal T cell accumulation [15]. A more recent study has successfully demonstrated fetal abnormalities after infection of pregnant wild-type (WT) SJL mice (but not C57BL/6 mice) with a Brazilian ZIKV isolate [10]. Intravaginal infection of WT C57BL/6 mice with ZIKV early in pregnancy has also been demonstrated to cause fetal growth restriction, as has infection of IFNAR +/heterozygous fetuses [16,17]. However, analysis of the immune response to ZIKV infection has not been performed to date using these models [10,16,17].
Efforts to evaluate the immune response to ZIKV infection have been hampered by a lack of knowledge of specific ZIKV epitopes, which precludes the analysis of antigen-specific T cells using tetramers or peptide restimulation. However, previous studies have established that total T cell responses can be tracked without a priori knowledge of specific epitopes using surrogate markers of activation [18,19]. These studies have thoroughly demonstrated that antigen encounter, but not inflammatory cytokines or bystander infection, leads to a decrease in cell surface expression of CD8α and an increase in surface expression of the integrin CD11a (CD8α lo CD11a hi ) on CD8 + T cells [19]. This phenotype is strictly dependent on antigenencounter and is maintained for the lifetime of antigen-experienced CD8 + T cells, allowing this population of antigen-specific cells to be tracked at late time-points following infection with a variety of pathogens [19,20]. Similarly, antigen-experienced CD4 + T cells can be tracked using surface expression of CD11a and CD49d (CD11a + CD49d + ) following various infections [18,21,22]. Thus, this provides a powerful approach to analyze the global T cell response to bacterial, parasitic or viral infections without knowledge of specific epitopes. As such, this is an ideal strategy for studying the immune response to emerging pathogens, such as ZIKV.
The recent demonstration that the ZIKV NS5 protein can block type I IFN production in human cells but not in mouse cells suggests that the IFN response in WT mice may prevent the establishment of infection [23]. In this study, we sought to determine whether immunocompetent WT C57BL/6 adult mice could be infected with ZIKV, and if so, whether the total CD4 + and CD8 + T cell responses could be tracked using surrogate markers of activation. We observed that infection of C57BL/6 mice induced an antiviral immune response reminiscent of the immune response observed in a variety of other viral infections [24][25][26]. Furthermore, we have identified a novel, potentially immunodominant, CD8 + T cell epitope in the ZIKV envelope protein. Thus, this will serve as an important point of reference for future studies to analyze the immune response to ZIKV infection across virus isolates (to determine the impact of viral evolution), in other mouse strains (to determine the role of host genetics), and will be useful in the identification of additional viral epitopes and in the design and development of effective ZIKV vaccine strategies.

ZIKV-Induced T Cell Responses Can Be Tracked in Immunocompetent Mice Using a Surrogate Marker Approach
Previous studies have established that antigen-experienced T cells induced following infection can be tracked without a priori knowledge of their specificity using surrogate markers [18,19]. This approach has allowed analysis of the immune response to infections with a variety of pathogens [18][19][20][21][22]. To further confirm this approach in our mouse colony, we adoptively transferred a low number of congenically marked Thy1.1/1.2 CD4 + T cell receptor (TCR) transgenic cells (SMARTA cells, specific to the lymphocytic choriomeningitis virus (LCMV) glycoprotein (GP) 61-80 peptide) [27] and congenically marked Thy1.2/1.2 CD8 + TCR transgenic cells (P14 cells, specific to GP 33-41 of LCMV) [28], into the same WT Thy1.1/1.1 C57BL/ 6 recipient mouse and infected the mice intraperitoneally (i.p.) with 2×10 5 PFU of LCMV Armstrong (acute viral infection that has been extensively used to study T cell responses) the following day. This approach allows us to track cells of known antigen specificity as well as assess both their surrogate marker expression and that of endogenous T cells. We determined the percentage of CD11a + CD49d + CD4 + T cells and CD8α lo CD11a hi CD8 + T cells at baseline (day -1) and at the peak of the T cell response, 8 days post-infection (dpi; S1 Fig). As expected, the TCR transgenic CD4 + and CD8 + T cell populations expanded sufficiently to be detected by flow cytometry, which indicates they have encountered their respective cognate antigens and have been activated. Further, and as described previously, virtually all of the antigen-experienced TCR transgenic CD4 + and CD8 + T cells responding to infection were identified using the surrogate marker approach (S1C-S1G Fig) [18,19]. Importantly, and in concurrence with the literature, this method also allowed for the tracking of endogenous antigen-experienced T cells responding to infection [18][19][20][21][22]. Thus, this approach allows us to track endogenous T cell responses to ZIKV infection.
WT C57BL/6 adult mice were infected with ZIKV either intravenously (i.v.) or i.p. (Fig 1) with 10 6 PFU of ZIKV, strain PLCal_ZV, an isolate that is phylogenetically closely related to the French Polynesian strain (Asian lineage), isolated from a Canadian patient who acquired the infection while travelling in Thailand (Genbank accession KF993678) [29]. Although we did not observe any gross behavioural or physical anomalies, ZIKV infection induced mild but detectable morbidity, with mice initially losing weight on day 1 post-infection, before recovering and gradually gaining weight over the following 9 days (Fig 1A). While this represents relatively mild morbidity, the observed weight loss is greater than what is typically observed following LCMV Armstrong infection, which is asymptomatic in mice despite establishing a robust acute infection (S1B Fig). Using the surrogate marker approach, we tracked the induction of ZIKV-specific CD4 + and CD8 + T cells at various dpi by conducting a longitudinal analysis of the peripheral blood lymphocytes of infected animals. While infection did not dramatically alter the overall proportion or number of total CD4 + and CD8 + T cells in the blood over time with either route of infection ( Fig 1B and 1C and Fig 1F and 1G), we observed the induction of a T cell response that peaked at day 7 post-infection ( Fig 1D and 1E and Fig 1H  and 1I). Compared to baseline, the proportion of CD11a + CD49d + CD4 + T cells in i.v.-infected mice increased from approximately 4% (which represents virtual memory cells present in all adult WT mice) [30] to approximately 10% at the peak of the response (7 dpi; Fig 1D). In addition, we observed a greater increase in the proportion of CD8α lo CD11a hi CD8 + T cells expanding from approximately 6% at baseline (representing the virtual memory population) to approximately 38% on day 7 post-infection ( Fig 1H). These were paralleled by a three-fold increase in the number of CD11a + CD49d + CD4 + T cells, and an eight-fold increase in the number of CD8α lo CD11a hi CD8 + T cells at the peak of the response (Fig 1E and Fig 1I). The proportions of expanded antigen-experienced T cells underwent a contraction phase, resulting in a pool of presumptive memory CD4 + and CD8 + T cells, which maintained the CD11a + CD49d + or CD8α lo CD11a hi phenotype, respectively (Fig 1D and 1E and Fig 1H and 1I). We observed a similar T cell response following i.p. infection, albeit with a slightly reduced magnitude ( Fig 1D  and 1E and Fig 1H and 1I). Taken together, these results demonstrate that ZIKV infection induces a T cell response in WT immunocompetent mice that can be tracked using the surrogate marker approach, suggesting that this represents a valid model for the analysis of the overall T cell responses to ZIKV infection.

Kinetics of ZIKV Infection in Adult WT Mice
Although the observed T cell response was strongly suggestive of an active viral infection, we sought to determine the kinetics of ZIKV infection by quantifying ZIKV RNA in the spleen at various time points post-infection i.v. with 10 6 PFU of ZIKV. Spleens were harvested at 6, 12, 24, 48 and 72 hours (h) post-infection, weighed, and total RNA was extracted. Viral burden was assessed using quantitative real-time PCR (qRT-PCR) and the data is expressed as PFU equivalents per gram of tissue (Fig 2A). PFU equivalents were generated by comparing C t values obtained to a standard curve produced using qRT-PCR analysis of ZIKV RNA extracted from titrated viral supernatants previously quantified by plaque assay (S2 Fig). ZIKV RNA is detectable at 6 h post-infection, with approximately 1.5×10 6 PFU equivalents/g in the spleen (Fig 2A). ZIKV RNA levels peaked in the spleen at 12 and 24 h postinfection, with an average of approximately 2.8×10 7 PFU equivalents/g and 2.7×10 7 PFU equivalents/g, respectively. The level of ZIKV RNA declined steadily thereafter to 2.2×10 6 PFU equivalents/g at 72 h post-infection. As an additional control, viral RNA was quantified in the spleens of C57BL/6 mice infected i.v. with 10 6 PFU of ZIKV or an equivalent dose of UV-inactivated virus at 12 h post-infection (peak viral load, Fig 2A) using qRT-PCR ( Fig 2B). ZIKV RNA was present at an average of 7×10 7 PFU equivalents/g in the ZIKV-infected group, and was virtually undetectable in the UV-inactivated group ( Fig 2B). This indicates that infection with live virus is required to detect viral RNA in the spleen at the tested time points (Fig 2). Thus, our data demonstrates that ZIKV can infect WT immunocompetent mice resulting in the accumulation of viral RNA in the spleen that peaks at 12-24 h post-infection.

ZIKV Infection Induces Dendritic Cell and Natural Killer Cell Activation
As our data demonstrated that ZIKV was able to infect WT mice, we next sought to analyze the innate immune response to infection by measuring dendritic cell (DC) and natural killer (NK) cell activation. C57BL/6 mice were infected i.v. with 10 6 PFU of ZIKV, an equivalent dose of UV-inactivated ZIKV, or mock-infected, and spleens were harvested at 2 dpi. ZIKV induced a significant increase in the expression of the co-stimulatory molecules CD40, CD80, and CD86 compared to mock-or UV-inactivated virus-infected mice (Fig 3A-3F), indicating that DC activation is dependent on infection with live virus (the gating strategy used to identify DCs is indicated in S3A-S3D Fig). Furthermore, NK cells in the spleens of ZIKV-infected mice also presented with an activated phenotype as measured by a significant increase in the percentage and number of NK cells expressing the activation marker CD69 (Fig 3G-3I). Finally, we assessed early total NK1.1 -CD3 + T cell activation using expression of the early activation marker CD69 at 2 dpi. In ZIKV-infected mice, but not those infected with UV-inactivated virus or mock-infected, we observed a significant increase in the percentage and number of CD69 + total T cells, suggestive of early T cell activation (S3E- S3G Fig). However, it is important to note that this could be a result of antigen or cytokine encounter at this early time-point. Thus, ZIKV infection of WT mice induces innate immune cell activation that is dependent upon infection with live virus.
The importance of the type I IFN response in ZIKV immunity has been demonstrated clearly by the severe morbidity and mortality of mice lacking IFNAR or both IFNAR and IFNGR in previous models of ZIKV infection [11][12][13][14]. To assess type I IFN production in response to ZIKV infection, we infected mice i.v. with 10 6 PFU of ZIKV or an equivalent dose of UV-inactivated virus and harvested spleens at 12 h post-infection. Total RNA was extracted, reverse-transcribed, and analyzed for IFN-α and IFN-β gene expression using qRT-PCR. We observed induction of both IFN-α and IFN-β in both groups compared to mock-infected mice, suggesting that, as opposed to DC and NK cell activation, infection with live virus is not CD11a + CD49d + antigen-experienced CD4 + T cells (D-E), total CD8 + T cells (F-G), and CD8α lo CD11a hi antigen-experienced CD8 + T cells (H-I) at baseline (day 0) and on indicated dpi in the peripheral blood of C57BL/6 mice infected i.p. (open circles) or i.v. (closed squares) with 10 6 PFU of ZIKV. Error bars represent mean ± SEM. Data has been pooled from two independent experiments, n = 3-4 mice per group per experiment.
doi:10.1371/journal.ppat.1006184.g001 a strict requirement for the induction of a type I IFN response ( Fig 3J). Thus, ZIKV infection induces a rapid type I IFN response that likely plays a role in limiting viral replication in this model.

CD4 + T Cells Polarize to a Th1 Phenotype in Response to ZIKV Infection
To further characterize the nature of the T cell response induced following ZIKV infection, we infected mice i.v. with 10 6 PFU of ZIKV, an equivalent dose of UV-inactivated virus or mockinfected, and harvested the spleens at 7 dpi (peak of the T cell response, Fig 1). Similar to what we observed in the peripheral blood, ZIKV infection induced a CD4 + T cell response in the spleen with a significant increase in both the frequency and number of antigen-experienced (CD11a + CD49d + ) CD4 + T cells (Fig 4A-4E), compared to UV-inactivated virus-and mockinfected mice (27% versus 12% and 11%, respectively).
Additionally, we examined the CD4 + T cell response to varying doses of ZIKV to determine whether the observed immune response is dose-dependent (S4 Fig). Mice were infected i.v. with 10 4 , 10 5 or 10 6 PFU of ZIKV or mock-infected, and CD4 + T cell activation was assessed 7 dpi. ZIKV-infected mice exhibited significantly higher CD4 + T cell activation than mockinfected mice at all doses tested. Although there were no significant differences in the proportions of antigen-experienced CD4 + T cells in response to the various doses of ZIKV, we observed a significant reduction in the number of CD11a + CD49d + CD4 + T cells in the mice infected with 10 5 PFU of ZIKV compared to those infected with 10 6 PFU (S4A- S4F Fig). Thus, the magnitude of the CD4 + T cell response to ZIKV is dependent on the dose of inoculation. This further validates our use of 10 6 PFU, which ensured infection and induced a more robust immune response than the lower doses tested.
In the presence of different inflammatory cytokines, CD4 + T cells can be polarized into one of several T helper (Th) subtypes. Briefly, in the presence of interleukin (IL)-12 and IFN-γ, which are typically produced in response to viral infection, effector CD4 + T cells polarize to  the Th1 subset, characterized by production of IFN-γ, tumour necrosis factor (TNF)-α and IL-2, and expression of the T-box transcription factor, T-bet [31]. In contrast, IL-4 is critical for the development of Th2 cells, which are characterized by the production of IL-5, among other cytokines. Finally, both IL-23 and tumour growth factor (TGF)-β are important for the development and maintenance of to Th17 cells, which produce mainly IL-17 [31].
To determine the differentiation status of the effector CD4 + T cells induced by ZIKV infection, we stimulated total splenocytes from ZIKV-infected mice with phorbol myristate acetate (PMA) and ionomycin in the presence of Brefeldin A for 3 h and analyzed cytokine production with intracellular cytokine staining via flow cytometry ( Fig 5). We observed that antigenexperienced CD4 + T cells exhibited a typical Th1 cytokine profile characterized by the production of IFN-γ, TNF-α, and IL-2 in response to PMA and ionomycin stimulation (but not in the absence of stimulation) (Fig 5A-5F). As an additional control, we also stimulated total splenocytes with plate-bound anti-CD3 antibody for 5 h in the presence of Brefeldin A. Since naïve T cells are dependent on CD28 co-stimulation for cytokine production, only effector T cells are able to respond to CD3 cross-linking [32]. As expected, IFN-γ production was only detectable from CD11a + CD49d + CD4 + T cells, and not naïve CD11a -CD49d -CD4 + T cells (S5A- S5C Fig). In addition, a significantly higher proportion of CD11a + CD49d + CD4 + T cells expressed the Th1 transcription factor T-bet, and significantly more T-bet on a per-cell basis when compared with mock-infected mice (Fig 5G-5I).
Although the production of Th1, but not Th2, cytokines has previously been described in response to LCMV infection [25], we sought to characterize the cytokine profile of antigenexperienced CD4 + T cells in LCMV-infected mice in order to compare what was observed in ZIKV-infected mice to a known model pathogen. As expected, and in concordance with the literature [25], antigen-experienced CD4 + T cells from LCMV-infected mice produced only IFN-γ, TNF-α and IL-2, but not the Th2 cytokine IL-5 or Th17 cytokine IL-17, in response to PMA and ionomycin stimulation (but not in the absence of stimulation) (S8 Fig). Taken together, these results establish that ZIKV infection induces a prototypical Th1 CD4 + T cell response, without inducing substantial Th2 or Th17 responses.
In addition to inducing effector CD4 + T cells, viral infection can have an impact on regulatory T cells (Tregs), a T cell subset that plays a key role in the maintenance of immune homeostasis [33]. ZIKV infection led to a significant decrease in the percentage of splenic FoxP3 + Tregs, although this was accompanied by a significant increase in the total number of Tregs present in the spleen (S7F- S7I Fig). Thus, ZIKV infection may have an impact on Treg numbers and it will be of interest to determine what role, if any, this plays in the natural course of ZIKV infection.

ZIKV Infection Induces a Robust Effector CD8 + T Cell Response
To assess the function and phenotype of the ZIKV-specific CD8 + T cells at the peak of the response, we first assessed the proportion of antigen-experienced (CD8α lo CD11a hi ) CD8 + T cells in the spleen at 7 dpi using the surrogate marker approach. Similar to what we observed in the peripheral blood, ZIKV infection induces a robust CD8 + T cell response with a significant increase in the proportion of antigen-experienced CD8 + T cells compared to UV-inactivated virus or mock-infection (41% versus 7% and 8% respectively, Fig 4F-4I), resulting in an approximately eight-fold increase in total number of CD8α lo CD11a hi CD8 + T cells (Fig 4J). This response was also dose-dependent, as infecting with 10 4 PFU of ZIKV induced a lower number of CD8α lo CD11a hi CD8 + T cells than infection with 10 5 or 10 6 PFU, despite no difference in the proportion of antigen-experienced CD8 + T cells across doses (S4G- S4L Fig).
Next, we determined the capacity of the CD8α lo CD11a hi CD8 + T cells to produce key effector cytokines such as IFN-γ and TNF-α, as well as IL-2. Following PMA and ionomycin stimulation, antigen-experienced CD8 + T cells responded by producing IFN-γ, with nearly 20% of cells maintaining some level of poly-functionality, producing both IFN-γ and TNF-α (Fig 6A,  6B and 6D-6F and S6E and S6F Fig). Furthermore, plate-bound anti-CD3 stimulation led to cytokine production only from the antigen experienced CD8α lo CD11a hi CD8 + T cells, and not from naïve CD8α hi CD11a lo CD8 + T cells (S5D- S5F Fig). As expected for effector CD8 + T cells, we did not observe any IL-2 production from the antigen-experienced CD8 + T cells ( Fig  6C-6F). These results are highly reminiscent to what has been described in the immune response to LCMV infection, which induces production of IFN-γ and TNF-α, and minimal IL-2 production [24]. We also confirmed that this cytokine profile is indeed observed in CD8 + T cells exhibiting the antigen-experienced phenotype in LCMV infection (S9 Fig). Finally, the majority of antigen-experienced CD8 + T cells expressed the transcription factor T-bet, which is critical for effector CD8 + T cell function (Fig 6G-6I) [34].
Viral infection typically induces CD8 + T cell populations that can be broadly classified into either short-lived effector cells (SLECs) that are critical for the acute immune response or memory precursor effector cells (MPECs) that preferentially survive contraction and go on to seed the memory CD8 + T cell pool [34]. As expected for the peak of the immune response where SLECs are critical for protective responses, the majority of cells polarized to a SLEC phenotype following ZIKV infection, expressing high levels of killer cell lectin-like receptor G1 (KLRG1) and down-regulating CD127 (IL-7 Receptor α chain); while only approximately 25% of cells maintained an MPEC phenotype (Fig 7D-7F). As demonstrated previously [34], the balance of these two cell populations is dynamic and changes over time post-infection, with the majority of CD8α lo CD11a hi CD8 + T cells expressing an MPEC phenotype, and less than 25% retaining the SLEC phenotype at >100 dpi (S10 Fig). Furthermore, antigen-experienced CD8 + T cells significantly down-regulate CD62L (L-selectin) (Fig 7G-7I), a lymph node homing marker that is characteristic of central memory CD8 + T cells [35], and ZIKV infection results in significantly more granzyme B (an important cytolytic molecule) expression on a per-cell basis than mock-infected mice (Fig 7J and 7K) at the peak of the T cell response. In contrast, the small population of CD8α lo CD11a hi CD8 + T cells from mock-infected mice retain (D) and number (E) of CD11a + CD49d + antigen-experienced CD4 + T cells from spleens of mock-, UV-inactivated ZIKV-and ZIKVinfected mice 7 dpi. Representative plots of CD8α lo CD11a hi antigen-experienced CD8 + T cells from the spleens of mock-(F), UVinactivated ZIKV-(G) and ZIKV-infected (H) mice 7 dpi. Percentage (I) and number (J) of CD8α lo CD11a hi antigen-experienced CD8 + T cells from the spleens of mock-, UV-inactivated ZIKV-and ZIKV-infected mice 7 dpi. Error bars represent mean ± SEM. Data are pooled from two independent experiments, n = 3 mice per group per experiment. Data in (D, E, I and J) were analyzed by one-way ANOVA with Tukey's post-test of multiple comparisons. ****p<0.0001. doi:10.1371/journal.ppat.1006184.g004 Analysis of the T Cell Response to ZIKV Infection a memory-like phenotype (likely virtual memory), expressing little KLRG1 and high levels of CD127 and CD62L, demonstrating that these cells are not activated during mock-infection (Fig 7A-7C and 7G-7I). Importantly, not all antigen-experienced effector cells can be classified as CD62Lor granzyme B + further emphasizing the power of the surrogate marker approach able to track all antigen-experienced cells regardless of their activation phenotype. In addition, the proportion of SLEC/MPECs at the peak of infection is reminiscent of the profile reported for CD8 + T cells responding to LCMV infection [34]. Taken together, our data indicates that ZIKV infection is able to induce a prototypical antiviral CD8 + T cell response in WT immunocompetent mice characterized by the majority of the antigen-specific CD8 + T cells developing critical effector functions.

Antigen-Experienced CD8 + T Cells Recognize an Epitope in the ZIKV Envelope Protein
In order to provide further validation of the surrogate marker approach in the analysis of the immune response to ZIKV infection, as well as to potentially identify ZIKV-specific CD8 + T cell epitopes, we used the Immune Epitope Database (IEDB) analysis resource consensus tool to predict CD8 + T cell epitopes within the ZIKV polyprotein [36]. This tool identifies peptides that are likely to bind the H-2-Db allele of class I major histocompatibility complex (MHC-I) in mice. From this list, we chose to further analyze the top six predicted peptides, including Env 294-302 , NS2A 1272-1280 , NS2A-NS2B 1350-1358 , NS4B 2329-2337 , NS4B 2371-2379 , and NS5 3066-3074 (numbered for their position in the ZIKV MR766 polyprotein, Genbank accession LC002520).
Next, we infected mice with 10 6 PFU of ZIKV, an equivalent dose of UV-inactivated virus, mock infection, or 2x10 5 PFU of LCMV Armstrong. At the peak of the T cell response to ZIKV (7 dpi) we sacrificed the mice, harvested spleens, and restimulated total splenocytes for 5.5 h in the presence of Brefeldin A with media alone, GP [33][34][35][36][37][38][39][40][41] (an immunodominant epitope of LCMV), or with one of the ZIKV-specific peptides. After incubation, the capacity of each peptide to stimulate IFN-γ production from the antigen-experienced (CD8α lo CD11a hi ) and naïve (CD8α hi CD11a lo ) CD8 + T cells was assessed by intracellular staining using flow cytometry (Fig 8, S11 Fig and S12 Fig). In the LCMV-infected group, IFN-γ production from antigenexperienced (CD8α lo CD11a hi ) CD8 + T cells was only observed after restimulation with GP 33-41 , and not with any of the ZIKV peptides (Fig 8B-8D and S12B-S12G Fig). In the ZIKV-infected group, we observed a strong induction of IFN-γ production from antigen-experienced (CD8α lo CD11a hi ) CD8 + T cells only when stimulated with the Env 294-302 peptide, with an average of approximately 56% of cells producing IFN-γ (Fig 8A, 8C and 8D and S12A, S12C-S12G Fig). Neither the mock-infected group, nor the group infected with UV-inactivated virus, responded to stimulation with the Env 294-302 peptide (Fig 8C and 8D and S11 Fig). Importantly, we observed no production of IFN-γ from naïve CD8α hi CD11a lo CD8 + T cells in response to any peptides, in any of the infection groups (Fig 8, S11 Fig and S12 Fig). This indicates that Env 294-302 is a bona fide ZIKV epitope recognized by CD8 + T cells, and suggests that it is the immunodominant epitope recognized by CD8 + T cells in this model. This data also further validates the use of the surrogate marker approach for tracking the total T cell response to ZIKV infection and directly identifies a novel ZIKV epitope targeted by CD8 + T cells.

Discussion
A rapid and effective global response will be required for the development of vaccines and antiviral therapies to combat the ongoing ZIKV epidemic in the Americas where viral pathogenesis appears to be increasing in severity. Paramount to this goal is the establishment of suitable small animal models to study ZIKV pathogenesis and the immune response to infection. Despite the development of several small animal models to study ZIKV infection and pathogenesis, most of these models lack critical innate immune components and succumb rapidly to ZIKV infection, precluding an analysis of the immune response to infection [11-14, 16, 17]. As such, the model developed and data presented herein represent an important reference point for the analysis of the immune response to ZIKV infection in adult, immunocompetent mice, and has uncovered a novel epitope of ZIKV recognized by CD8 + T cells.
The recent demonstration that the ZIKV NS5 protein is capable of interrupting type I IFN production in human, but not mouse cells has led to the suggestion that ZIKV is unable to cause disease in mice [23]. However, our data demonstrates that ZIKV establishes a limited, but detectable, infection in immunocompetent WT C57BL/6 adult mice resulting in minimal morbidity. This is supported by data from other studies where ZIKV RNA was detectable in the spleen, albeit at low levels, following infection of either 129Ev/Sv or C57BL/6 mice, both immunocompetent mouse strains [11,13]. More importantly, we show that ZIKV infection of adult WT C57BL/6 mice is sufficient to induce a T cell response, which peaks at day 7 before contracting to potentially form a pool of memory T cells. This course of infection in immunocompetent mice is potentially more representative of typical ZIKV pathogenesis in immunocompetent adult humans, which is typically mild and self-limiting [4,5]. Furthermore, since type I IFN signalling is required for initiation of optimal T cell responses [15], immunocompetent mouse models will be extremely valuable for vaccine strategies aimed at inducing T cell responses, particularly as this model has enabled us to identify a novel ZIKV epitope that could be used in vaccination approaches.
Our model of ZIKV infection combined with the surrogate marker approach provides a valuable reference point that is required to begin interrogating whether pathogenic isolates impact the host immune response to establish a more severe infection. Additionally, the current approach will allow us to compare the immune response induced by infection in a variety of mouse strains (where epitopes are unlikely to be conserved) and potentially link the immune response back to pathogenesis. The recent demonstration that infection of pregnant SJL, but not C57BL/6, mice with a Brazilian ZIKV isolate results in intrauterine growth restriction and microcephaly raises several questions regarding ZIKV evolution and host genetics that can be addressed in our model [10]. Importantly, the approaches developed by Rai et al. and McDermott et al. allow us to measure and compare the immune response in strains with different genetic backgrounds [18,19]. Thus, the combination of this approach, which we have now thoroughly validated for use in ZIKV infection models, and the use of reverse genetics systems to manipulate viral sequences will allow us to answer a variety of important questions in the future [37,38].
The identification of a specific, potentially immunodominant, ZIKV epitope in the envelope protein provides an important advancement for ZIKV immunology. Interestingly, this (D), number (E), and gMFI (F) of cytokine-producing CD8α lo CD11a hi CD8 + T cells from the spleens of ZIKV-infected mice 7 dpi. (G) Representative histograms of T-bet expression in splenic CD8α lo CD11a hi CD8 + T cells from mock-and ZIKV-infected mice 7 dpi. Shaded histograms represent isotype control. Percentage (H) and gMFI (I) of T-bet expression in splenic CD8α lo CD11a hi CD8 + T cells from mock-and ZIKV-infected mice 7 dpi. Error bars represent mean ± SEM. Data are pooled from two independent experiments, n = 3-5 mice per group per experiment. Data in (H and I) were analyzed with a two-tailed, unpaired Student's t test. ****p<0.0001. epitope is conserved across 99% (103/104) of all full-length ZIKV polyprotein sequences available on Genbank to date, making it a powerful tool for future studies. Knowledge of the epitopes to which the immune system responds will allow for the design of ZIKV vaccine candidates aimed at inducing a potent T cell response, and will lead to the establishment of tetramers to identify and track Env 294-302 -specific CD8 + T cells. It will also be of interest to determine whether this epitope is immunogenic in human samples from the ongoing epidemic.
In conclusion, our findings demonstrate that ZIKV infection of immunocompetent adult C57BL/6 mice is a self-limiting infection inducing mild, but detectable, morbidity. This induces innate immune activation, leading to a Th1-polarized CD4 + T cell response and an effector CD8 + T cell response. By tracking the total T cell response using the surrogate marker approach, we have identified a novel CD8 + T cell epitope in the ZIKV envelope protein that will have important implications for ZIKV immunology and the design and testing of ZIKV vaccine candidates.

Methods Cells
Vero cells (African Green monkey kidney epithelial cells, kindly provided by Steven Varga, University of Iowa) were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS), 1% L-glutamine and 1% non-essential amino acids at 37˚C and 5% CO 2 .

Viruses
Low passage (p.4) infectious ZIKV (PLCal_ZV, Genbank accession KF99378) derived from a ZIKV-infected traveller returning to Canada was kindly provided by the National Microbiology Laboratory (Winnipeg, Canada) [29]. ZIKV stocks were propagated in Vero cells after infecting at a multiplicity of infection (MOI) of 0.5. Supernatants from both mock-infected as well as ZIKV-infected cells were harvested after 72 h post-infection, filtered, and viral stocks were titrated by plaque assay on Vero cells. Briefly, Vero cells were seeded at a density of 1.5×10 6 cells per 60 mm dish. Twenty-four hours later, viral stocks were serially diluted (10-fold) in EMEM and 2 mL of each dilution was used for infection. At 2 h post-infection, cells were overlaid with EMEM (Wisent), 1.2% Carboxymethyl cellulose (Sigma-Aldrich), and 2% heat-inactivated FBS for 4 days prior to fixation with 5% formaldehyde and 0.1% crystal violet stain. ZIKV was UV-inactivated by transferring 1 mL of ZIKV into one well of a 6-well plate and exposing it to 3 Joules/cm 2 of UV irradiation in a UVC 500 Crosslinker (Hoefer). Dose of UV irradiation was chosen based on previous work in related flaviviruses [39]. LCMV Armstrong was kindly provided by John Harty (University of Iowa) and propagated as described [40].

Mouse Experiments
C57BL/6 mice (WT) were purchased from Charles River laboratories or bred at McGill University. Infected mice were housed at the appropriate biosafety level and used at 6-9 weeks of spleens of ZIKV-infected mice 7 dpi. Percent (E) and number (F) of SLECs and MPECs on CD8α lo CD11a hi CD8 + T cells from ZIKV-infected mice. Representative histogram (G), percentage (H), and number (I) of CD62L + and CD62L -CD8α lo CD11a hi CD8 + T cells from spleens of mock-(shaded histogram) and ZIKV-infected (open histogram) mice 7 dpi. Numbers on histogram indicate percentage of CD62L -(left gate) and CD62L + (right gate) cells from ZIKV-infected sample. (J) Representative histograms of granzyme B expression in CD8α lo CD11a hi CD8 + T cells from spleens of mock-(left) and ZIKV-infected (right) mice 7 dpi. Shaded histograms represent isotype control. (K) gMFI of granzyme B expression. Error bars represent mean ± SEM. Data are pooled from two independent experiments, n = 3-5 mice per group per experiment. Data in (H, I and K) were analyzed with a two-tailed, unpaired Student's t test. ***p<0.0005; ****p<0.0001. doi:10.1371/journal.ppat.1006184.g007

Ex Vivo Stimulation
At 7 dpi spleens were isolated and mechanically disrupted to generate single-cell suspensions. Erythrocytes were lysed with ACK buffer, and stimulated for 3 h at 37˚C/5% CO 2 in 5 ng/mL indicated peptide in the presence of Brefeldin A. (C-D) Percentage of IFN-γ + antigen-experienced CD8α lo CD11a hi or naïve CD8α hi CD11a lo CD8 + T cells from mock-, UV-inactivated ZIKV-, ZIKV-or LCMV-infected mice 7 dpi after restimulation with media alone or 200nM of GP [33][34][35][36][37][38][39][40][41] (LCMV peptide) (C) or Env 294-302 (ZIKV peptide) (D) for 5.5 h at 37˚C in the presence of Brefeldin A. Error bars represent mean ± SEM. Data are pooled from two independent experiments, n = 3 mice per group per experiment. Data in (C and D) were analyzed with a two-tailed, paired Student's t test. ****p<0.0001.  [33][34][35][36][37][38][39][40][41] in the presence of Brefeldin A. Type of stimulation used is indicated in the corresponding figure legend. Anti-CD3 antibody was bound to 96-well plates by incubating in 100 μL PBS for 2 h at 37˚C/5% CO 2 .  [11,41]. A standard curve was generated of C t value versus log 10 PFU using serial 10-fold dilutions of ZIKV RNA extracted from previously titrated viral stocks. Viral burden is expressed as PFU equivalents per gram of tissue after comparison with the standard curve (S3 Fig). The limit of detection was set as the average PFU equivalent per gram of tissue from the mock-infected samples.

qRT-PCR Analysis
Total RNA was extracted as described above and reverse transcribed using the iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad) according to the manufacturer's instructions. qRT-PCR was performed in duplicate in 96-well PCR plates (Bio-Rad) using SensiFAST SYBR Lo-ROX mix (Bioline) in a Bio-Rad CFX96 Touch Real-Time System under standard cycling conditions. Relative mRNA levels were calculated using the ΔΔCT method [42] using TATAbinding protein (TBP) expression as an internal control, and plotted as fold change by normalizing to mock-infected samples. The following primers were used to detect IFN-α, IFN-β and TBP mRNA: IFN-α Forward

Statistical Analyses
Data were analyzed using GraphPad Prism7 software. Specific tests for determining statistical significance are indicated in the figure legends. P values of less than 0.05 were considered statistically significant.

Ethics Statement
All animal procedures were carried out in accordance with the Canadian Council on Animal Care and were approved by the McGill University Animal Care Committee (Protocol #7800).