IL-22 hinders antiviral T cell responses and exacerbates neurological disorders in ZIKV-infected immunocompetent neonatal mice

The Zika virus (ZIKV) outbreak that occurred in multiple countries was linked to increased risk of neurological disorders and congenital defects. However, host immunity and immune-mediated pathogenesis in ZIKV infection are not well understood. Interleukin-22 (IL-22) is a crucial cytokine for regulating host immunity in infectious diseases. Whether IL-22 plays a role in ZIKV infection is unknown. The cellular of was in IFNAR −/− mice and To determine the role of IL-22, we challenged 1-day-old wild-type (WT) and IL-22 −/− mice with ZIKV and monitored clinical manifestations. Glial cell activation in the brain was assessed by confocal imaging. ZIKV-specic CD8 + T cell responses in both the spleen and brain were analyzed by ow cytometry. In addition, we infected mouse primary astrocytes in vitro, and characterized the reactive astrocyte phenotype. Human glial cell line was also infected with ZIKV in the presence of IL-22, followed by the evaluation of cell proliferation, cytokine expression and viral loads.

junctions [17]. These studies suggest that IL-22 may play a detrimental role in the central nervous system (CNS) during viral infection. However, whether IL-22 contributes to ZIKV-induced encephalitis and the underlying mechanisms are not entirely known.
In this study, we found that γδ T cells were the main source of IL-22 in the brain and spleen. We found increased animal survival, alleviated clinical symptoms and reduced viral burden in the neonatal IL-22 −/− mice compared with those in WT mice. Interestingly, ZIKV infection induced microglia activation and promoted the polarization of A1-prone astrocytes, which were considered neurotoxic [18]. Although IL-22 exhibited a dispensable role for glial cell activation and infection in vitro, the lack of IL-22 resulted in decreased microglia activation in vivo. Importantly, IL-22 −/− mice mounted more effective ZIKV-speci c T cell responses in vivo, while recombinant IL-22 treatment hindered these responses. Together, our study indicates that IL-22 signaling may play a detrimental role in encephalitis in the ZIKV-infected neonatal mice.
ZIKV (Asian lineage FSS13025) was obtained from the World Reference Center for Emerging Viruses and Arboviruses (Galveston, TX). Virus was ampli ed in Vero cells and viral titer was calculated as uorescent focus units (FFU) per ml. All newborn mice were born from pathogen-free parents and inoculated one day after birth with 4 × 10 3 FFU ZIKV in 2 µL by Hamilton microliter syringes through subcutaneous (s.c.) inoculation at the lateral side of body part. For in vivo treatment, neonatal WT mice were s.c. injected with recombinant IL-22 treatment (1 µg in 5 µL, s.c.) every other day. In some experiments, mice that were 3 weeks old were intraperitoneally infected with 1 × 10 5 FFU ZIKV. Animals were monitored daily for bodyweight changes and clinical signs of pathology. Moribund animals were euthanized in accordance with the UTMB IACUC guidelines.

Clinical Symptom Evaluation
The clinical evaluation of infected neonatal mice was modi ed according to another report [19]. Brie y, Mice were weighed and examined for signs of infection daily. Examination criteria included appearance, stance, and motility. The description of clinical presentations included stagger step (increased spread of hind legs and unusual pauses during movement), paralysis (loss of muscle function of one or two hind legs), and seizure (sudden stiffening of muscles in the back and legs).

Cell Lines And Primary Mouse Glial Cells
Vero and U-87 MG cells were cultured in Minimum Essential Medium (Thermo Fisher Scienti c) supplemented with 10% fetal bovine serum. All cells were cultured in the presence of 100 U/mL penicillin and 100 µg/mL streptomycin in 37 °C incubator with 5% CO 2 and 95% humidity control.
For primary mouse glial cell preparation, B6 mouse pups (1-4 days old) were used for mixed cortical cell isolation [20]. Brie y, mouse brains were taken out and placed into a dish containing cold HBSS. The meninges from the cortex hemispheres were pulled with the ne forceps under a stereomicroscope to avoid contamination with meningeal cells and broblasts. The cortex hemispheres were cut into small pieces, followed by 2.5% trypsin digestion for 30 min at 37 °C. Cortex tissue pieces were harvested and dissociated into a single-cell suspension. Mixed cortical cells were counted and cultured with DMEM/F12 (1:1) medium plus 10% fetal bovine serum, 100 U/mL penicillin and 100 µg/mL streptomycin in a T75 culture ask. Medium was replaced every 5 days. For microglia isolation, the mild trypsinization method was used. Brie y, mixed cortical cells were cultured for 12-15 days and incubated with a trypsin solution (0.25% trypsin, 1 mM EDTA in HBSS) diluted 1:4 in DMEM/F12 medium. The upper layer of cells was detached in one piece and removed from the ask. Microglia remained attached to the bottom and were harvested by further trypsinization for 15-20 min. The purity of microglia was 98.5% as con rmed by CD11b, CX3CR1 and CD45 ow cytometric analysis (Fig. S1A). For astrocyte isolation [21], after 7-8 days culture, the ask was shaken at 180 rpm for 30 min to remove microglia. New medium was added into the ask followed by shaking at 240 rpm for 6 hours (hrs) to remove oligodendrocyte precursor cells.
Astrocytes were detached by trypsin-EDTA, washed with PBS, and plated into two T75 culture asks. Medium was changed every three days and astrocytes were harvested at 12-14 days after the rst split.
Astrocytes were then identi ed using IF staining of GFAP (Fig. S1B). For in vitro infections, a single dose of ZIKV (MOI of 1) was used.
RNA Isolation And Quantitative Real-time PCR (qRT-PCR) RNA was isolated using Rneasy mini kit (Qiagen) according to the manufacturer's instructions. cDNA was synthesized using the iScript cDNA synthesis kit (Bio-Rad). The abundance of target genes was measured by qRT-PCR using a Bio-Rad CFX96 real-time PCR apparatus. SYBR Green Master Mix was from Bio-Rad, and TaqMan Universal Master Mix, including gene-speci c probes and primers were from Integrated DNA Technologies (IDT). The relative level of gene expression was calculated using the 2 −ΔΔct method. The sequences of primers and probes are listed in Supplementary table 1.

Measurement Of Viral Burden
For measuring viral burden in vivo, mouse tissues were weighed and homogenized using a Tissue-Tearor (BioSpec). ZIKV RNA levels were determined by TaqMan quantitative reverse transcriptase PCR on the real-time PCR detection system (Bio-Rad). Virus burden was determined by interpolation onto an internal standard curve composed of serial 10-fold dilutions of a synthetic ZIKV RNA fragment. A previously published primer set was used to detect ZIKV RNA [22]. For measuring viral burden in infected cells in vitro, the relative level of gene expression was calculated based on C t values using GAPDH as a housekeeping gene.

Lymphocytes Isolation And Puri cation
Lymphocytes were isolated according to our previously reported method [23]. Brie y, brains were cut into pieces and digested with 0.05% collagenase IV (Roche, Indianapolis, IN) at 37 °C for 30 min. Cell suspensions were passed through a 70-µm nylon cell strainer to yield single-cell suspensions. Lymphocytes were enriched by centrifugation (400 g) at room temperature for 30 min over a 30/70% discontinuous Percoll gradient (Sigma). Spleens were collected from mice and gently mashed in the RPMI-1640 medium through a cell strainer. Red blood cells were removed by using Red Cell Lysis Buffer (Sigma, St. Louis, MO). Cells were harvested by centrifugation (300 g, 10 min, 4 °C) and resuspended in RPMI-1640 medium plus 10% fetal bovine serum.

Flow Cytometry
Intracellular staining was performed with ow cytometry as in our previous report [23]. Brie y, for IL-22 and IL-17A detection, lymphocytes were cultured with rIL-23 (20 ng/ml) for 12 hrs. Brefeldin A solution (eBioscience) was added for the last 4 hrs of culture. For detecting IFN-γ and TNF-α in ZIKV-speci c CD8 T cells, lymphocytes were incubated with ZIKV peptide E 294 − 302 (1 mg/mL, GenScript) in the presence of Brefeldin A solution for 5 hrs. Cells were then stained for anti-CD16/32 (Clone 2.4G2) and surface markers, xed by using an IC xation buffer, and followed by staining for intracellular cytokines (Thermo

Immuno uorescence Staining And Confocal Microscopy
Mice were euthanized with CO 2 and perfused transcardially with cold PBS. Frontal cortices were collected and were immediately placed in 4% PFA in PBS at 4 °C overnight and then cryoprotected in a 30% sucrose solution in PBS for at least 24 hrs at 4 °C. Tissues were embedded in optimal cutting temperature compound (Sakura Finetek, Torrance, CA). Transverse sections (35 µm) were prepared on a cryostat (Leica CM 1900). The sections were kept in Hito oating section storage solution (Hitobiotec Corp) at -20 °C until they were stained for immunocytochemistry. For immunostaining, tissue sections were rinsed with PBS twice to remove the storage solution and blocked with 5% BSA and 0.3% Triton X-100 in PBS for 2 hrs at room temperature, followed by 48 hrs incubation with primary antibodies. After ve washes with PBS, the sections were incubated with uorophore-conjugated secondary antibodies at 4 °C overnight prior to section mounting. Confocal Z-stacks images were captured within the layer I-II of the cortex using a confocal microscope (Nikon A1). For each mouse, at least 3 xed-frozen sections were included for each experiment, and at least 3 z-stacks images at 20 x, 40 x or 60 x magni cation were taken. Thirty to fty consecutive optical sections with 1 µm interval thickness at 40 x and 60 x magni cation were captured for each Z-stacks image. To process images, "Subtract Background" (50 pixels) was applied to remove the background and a threshold (50-225) was set to remove outliers. The analysis of positive areas and cell sizes were performed using Image J software. The rabbit-anti-Iba1 (Abcam, ab178846) and goat-anti-GFAP (Abcam, ab53554) antibodies were purchased from Abcam. The secondary antibodies, including goat anti-rabbit IgG (H + L) Alexa Fluor 488 and donkey anti-goat IgG (H + L) Alexa Fluor 555, were purchased from Thermo Fisher Scienti c.
Western Blot Analysis Total brain protein was homogenized in RIPA buffer including a 1% protease inhibitor cocktail (Sigma-Aldrich), and protein concentrations of the lysates were quanti ed using a BCA kit (Thermo Fisher Scienti c). Five to twenty µg of proteins per sample were loaded onto a 12% Novex Tris-Glycine Gel and subsequently transferred to a PVDF membrane. The membrane was blotted with primary antibodies at 4 °C overnight. Antibody detection was accomplished using horseradish peroxidase conjugated secondary antibodies and visualized with ECL. Markers were used to identify the target protein band. As loading control, the expression of GAPDH was also measured. The signal intensity was quanti ed with Image Studio Lite. The primary antibodies anti-VE-cadherin, anti-ZO1, anti-Occluding, anti-Claudin 1 and anti-Claudin 3 were from Abcam. Anti-GAPDH was purchased from Cell Signaling Technology.

Statistical Analyses
Data were shown as mean ± SEM and analyzed using the two-tailed Student's t-test when compared between two groups. One-way ANOVA was used for statistical analysis of more than two groups. A logrank (Mantel-Cox) test was used for survival curve analysis. *, ** or *** means P-value < 0.05, < 0.01 or < 0.001, respectively. Statistical analyses were operated by GraphPad Prism software 7.0 (GraphPad Software Inc., San Diego, CA).

ZIKV infection induced γδ T cell-derived IL-22
To investigate whether ZIKV infection can induce IL-22 expression, we infected 3-week-old IFNAR −/− mice with ZIKV as previously reported [24]. We observed that the infected mice started to lose weight at 4 days post infection (dpi) and all mice died at 6 dpi ( Fig. S2A and B). Severe in ammatory in ltration appeared in the liver, lung and brain at 5 dpi, accompanied with increased expression of in ammatory genes, including IL-6, IL-1β, TNF-α and IFN-γ ( Fig. S2C and D). We also found that IL-22 mRNA expression in the brain increased at 3 dpi and reached a peak at 5 dpi (Fig. 1A). On the contrary, splenic IL-22 mRNA was upregulated signi cantly at 3 dpi, and then returned to the baseline at 5 dpi. No considerable IL-22 upregulation was observed in the lung and liver following ZIKV infection (Fig. 1A). To de ne the source of IL-22 following infection, we isolated lymphocytes from the brain and spleen, and analyzed the IL-22-expressing subpopulations. We found that IL-22 was mainly produced by γδ T cells, which also produced IL-17 (Fig. 1B). NK or conventional T cells had minimal or no detectable levels of IL-22 (data not shown).
Consistently, the kinetic pro le of IL-22 + γδ T cells in the brain and spleen showed the similar patterns as IL-22 transcript levels (Fig. 1B). We con rmed this nding using a WT (B6 background) mouse model by subcutaneously challenging one-day old neonatal mice with ZIKV (4 × 10 3 FFU). The infection of neonates resulted in an elevation of IL-22 expression in the spleen and brain at 2 and 13 dpi, respectively (Fig. 1C). Similarly, we showed that γδ T cells in these immunocompetent mice were the main producers of IL-22 and IL-17 (Fig. 1D). IL-22 protein levels were also increased in the brain, but not in the spleen at 13 dpi (Fig. 1E). Thus, our results indicated that ZIKV infection can induce IL-22 expression in γδ T cells.

IL-22 De ciency Alleviated ZIKV-induced Neurological Disease In ZIKV Infection
To elucidate the role of IL-22 in ZIKV infection, we s.c. infected one-day old WT and IL-22 −/− mice and monitored their bodyweight changes, survival rates and clinical manifestations. We found that ZIKVinfected IL-22 −/− mice maintained signi cantly higher body weights starting at 8 dpi and continuing to the end of the observation period compared with that of infected WT mice ( Fig. 2A). No signi cant difference in bodyweight was observed between WT and IL-22 −/− mice without ZIKV infection ( Fig. 2A). All IL-22 −/− mice survived, while about 40% of the WT mice succumbed at 20 dpi (Fig. 2B). Consistently, the rate of paralysis for IL-22 −/− mice was lower than that of WT mice (Fig. 2C). In the context of clinical manifestations, WT mice began to display staggered steps at 10 dpi, and around 60% of mice developed paralysis or seizure symptoms at 15 dpi. On the contrary, only 10% of IL-22 −/− mice developed paralysis during the infection (Fig. 2D); neither seizure nor death was observed in the absence of IL-22. Full recovery was observed in about 90% of IL-22 −/− mice by the end of the study, while none of the WT mice got full recovery at 20 dpi (Fig. 2D). To further con rm the detrimental role of IL-22, one-day old WT mice were infected with ZIKV, followed by recombinant IL-22 treatment every other day. We found that IL-22treated mice exhibited reduced weight gain at 11 and 13 dpi compared with those of PBS-injected mice (Fig. 2E). Moreover, IL-22 treatment resulted in increased rates of paralysis at the same times (Fig. 2F). Collectively, we demonstrated that IL-22 plays a pathogenic role in ZIKV encephalitis.

IL-22 De ciency Reduced Microglia Activation In ZIKVinfected Brain
Microglia cells are considered to be brain residential macrophages that are responsible for the clearance of invading pathogens and damaged neuronal cells [25]. However hyperactivation of microglia induces chronic in ammation and neurodegeneration [26]. ZIKV-infected microglia exhibit an activated phenotype, characterized by upregulation of several in ammatory cytokines and may also transmit virus to other target cells in the brain [9,27]. To determine the role of IL-22 in microglial pro le during ZIKV infection, we analyzed the microglial phenotype in both IL-22 −/− and WT neonatal mice. We found that ZIKV-infected IL-22 −/− neonatal mice showed less number of activated microglia, as evidenced by decreased numbers of IBA-1 + cells in the cortex compared to WT control mice (Fig. 3A). Further analysis revealed that the amoeba-like microglia in IL-22 −/− mice were smaller than those in WT mice, indicating there is reduced microglia activation in the absence of IL-22 (Fig. 3B). Similarly, the numbers of activated astrocytes, which were characterized as GFAP + , were also lower in the cortex of IL-22 −/− mice (Fig. 3A).
We also examined the gene expression of in ammatory cytokines in the brain using qPCR. The mRNA levels of IFN-γ, as well as its inducible chemokines, CXCL9 and CXCL10, were not changed signi cantly. However, the level of brain TNF-α, which is the main cytokine produced by microglia [27,28], was signi cantly decreased in IL-22 −/− mice (Fig. S3). Therefore, these data suggested that IL-22-de ciency in ZIKV-infected neonatal mice resulted in reduced microglia activation and decreased pro-in ammatory TNF-α expression.

IL-22 played a dispensable role in ZIKV-induced glial cell activation in vitro
To determine whether IL-22 has a direct effect on glial cell activation and viral inoculation, mixed cortical cells were isolated from mouse pups and cultured in vitro for generating astrocytes and microglia (Fig.  S1). Reactive astrocytes mainly display two polarizations, termed A1 and A2, which play neurotoxic and protective roles, respectively [18]. We found that ZIKV infection induced astrocyte activation and promoted an A1-prone phenotype. However, IL-22 did not show any effect on ZIKV-induced astrocyte activation (Fig. 4A). To determine whether IL-22 contributes to cell apoptosis and proliferation, we analyzed BCL2 and Ki67 transcript levels in ZIKV-infected astrocytes. Although ZIKV infection resulted in decreased cell survival and proliferation as evidenced by downregulation of BCL2 and Ki67 expression, no effect of IL-22 was observed on astrocytes in vitro (Fig. 4B). Finally, we measured viral loads in ZIKVinfected astrocytes in the presence of IL-22. Our data showed that ZIKV infected astrocytes e ciently, and IFN-γ signi cantly decreased viral burden at 24 hrs. However, either IL-22 alone or synergized with IFN-γ did not alter viral burdens (Fig. 4C). To further con rm our results, we infected human glial cell U-87MG with ZIKV, followed by rIL-22 treatment. Our results showed that IL-22 did not rescue cell death or promote cell proliferation (Fig. S4 A and B). The comparable transcript levels of CXCL10, CCL2 and BCL2, as well as viral loads were observed between control and rIL-22-treated groups (Fig. S4 C and D), indicating that IL-22 was dispensable for the growth, activation and viral infection of human glial cells. Microglia, as resident macrophages in the brain, have a high ability to secret in ammatory cytokines (e.g. TNF-α), which induce astrocyte A1 polarization and lead to neuron death [18]. Our qPCR data showed that ZIKV-infected mouse primary microglia increased the expression of several in ammatory cytokine genes, including TNF-α, IL-1β, CXCL2 and CCL2 (Fig. 4D). Arginase-1 (Arg-1) can compete with nitric oxide synthase in the brain, playing a neuroprotective role [29]. We found reduced Arg-1 expression in microglia following ZIKV infection, suggesting to us that ZIKV may cause damage in the brain through inhibiting neuroprotective factors. In addition, ZIKV-infected microglia downregulated anti-apoptotic gene BCL2 expression, indicating that ZIKV infection may promote glial cell apoptosis (Fig. 4D). Although microglia were infected by ZIKV, IL-22 did not contribute to this infection or cell activation, as evidenced by similar levels of viral loads and comparable gene expression of activation markers following IL-22 supplementation in vitro (Fig. 4D). Since human Th17 cells are known to promote BBB disruption and CNS in ammation [17], we measured several tight junction proteins, including ZO-1, VE-cadherin, Occludin, Claudin-1, and Claudin-3 in the brain of both ZIKV-infected WT and IL-22 −/− mice. Similar levels of these tight junction proteins were detected, implying that IL-22 may not disrupt the BBB during ZIKV infection (Fig. S5). Therefore, our data demonstrated that ZIKV infection induced neurotoxic reactive astrocytes and activated microglia, leading to brain in ammation, and that IL-22 was dispensable for glial cell activation, polarization and viral clearance in vitro.

IL-22 Hindered Anti-ZIKV CD8 + T Cell Responses
Although immune cells do not express IL-22 receptor, IL-22 can regulate T cell responses in both viral and parasitic infection, probably via indirect ways [14]. The adaptive immune response, especially the anti-ZIKV cytotoxic CD8 T cell response, has been demonstrated to play a protective role against ZIKV infection [30,31]. However, excessive CD8 + T cell in ltration in the brain can cause paralysis in mice with ZIKV infection [32]. Here, we speculated that the absence of IL-22 resulted in more robust CD8 + T cell responses which e ciently controlled ZIKV infection. To test this hypothesis, we s.c. infected one-day old WT and IL-22 −/− mice, and interrogated viral burdens and anti-ZIKV CD8 + T cell responses. We found that tissue viral load spiked at 13 dpi in WT mice, with much higher levels in the brain compared with those in the spleen (Fig. 5A and B). Importantly, IL-22 −/− mice displayed signi cantly lower viral loads in both the spleen and brain at the peak of viral infection (13 dpi), but not at other time-points including 2, 7 and 20 dpi ( Fig. 5A and B). IL-22 −/− mice displayed more effective anti-ZIKV CD8 + T cell responses, as evidenced by increased numbers of IFN-γ + CD8 + T cells (Fig. 5C). However, comparable numbers of in ltrated lymphocytes and IFN-γ + CD8 + T cells were found in the brains of WT and IL-22 −/− mice at 13 dpi (Fig. 5D).
These results may indicate that IL-22 de ciency promoted effector functions of antiviral CD8 + T cells, without increasing in ammatory in ltration to the brain. Consistently, exogenous IL-22 treatment increased brain viral loads and impaired anti-ZIKV CD8 + T cell responses in both the spleen and brain ( Fig. 5E and F). We also con rmed our nding using a three-week old immunocompetent mouse model. Again, IL-22-de ciency resulted in increased cytokine-producing CD8 + T cells in the spleen at 7 dpi (Fig.  S6). Collectively, our ndings suggested that IL-22 dampens anti-ZIKV T cell responses in the periphery and exacerbates viral infection in the brain, leading to profound cerebral in ammation and animal paralysis and death.

Discussion
The roles of IL-22 in various diseases are diverse. In mucosal disorders, IL-22 plays a protective role by preserving epithelial integrity [33,34], promoting antibacterial peptides and proteins [35], and inducing mucins [36,37]. However, IL-22 is pathogenic in some in ammatory settings, such as psoriasis [38], allergic airway in ammation [39] and collagen-induced arthritis [40]. The role of IL-22 in viral infection is enigmatic. Our previous research indicated that IL-22 contributes to antiviral immune responses and determines viral clearance in LCMV infection [14]. In this study with ZIKV infection, we found that de ciency of IL-22 resulted in decreased viral loads, alleviated clinical manifestations, and increased survival rates in neonatal mice ( Fig. 1-3). Our in vitro results showed that ZIKV infection promoted the polarization of neurotoxic reactive astrocytes and the activation of microglia, which may induce neuroin ammation in the brain [18]. Although IL-22 did not directly exert its effects on brain glial cells in vitro, the absence of IL-22 led to reduced glial cell activation in vivo (Fig. 4). More importantly, IL-22 −/− mice elicited qualitatively better ZIKV-speci c T cell responses compared with those in WT mice. Therefore, our results demonstrated a pathogenic role of IL-22 in ZIKV encephalitis of neonatal mice.
Upon viral infection, IL-22 induction is organ-speci c. NK and NKT cells can produce IL-22 in response to murine cytomegalovirus and in uenza virus infection [41][42][43]. Intrahepatic γδ T cells are also the source of IL-22 in hepatitis B virus-infected patients [44]. We have previously reported that intrahepatic γδ T cells are the main immune cells to produce IL-22 by IL-23 stimulation in an LCMV-infected mouse model [14]. Similarly, we found in this study that γδ T cells were the main source of IL-22 in the spleen and brain in both IFNAR −/− and WT neonatal mouse models (Fig. 1). The peak of IL-22 expression in the spleen was as early as 2 dpi, while the time-point for peak expression of IL-22 in the brain was delayed. Since ZIKV initially infected lymphoid organs and subsequently invaded the CNS [24], this dynamic pattern of IL-22 suggests that IL-22 might be driven by the virus or virus-induced innate immune responses. Indeed, ZIKV infection induced high levels of brain in ammatory cytokines, including IL-1β (Fig. S1D), which may facilitate the expression of IL-22 from γδ T cells [45]. In addition, high level of IL-6 in the brain, but not in the liver and lung, also suggest that IL-6 may be required for IL-22 production perhaps through the induction of downstream IL-21 [46].
Both WNV and ZIKV belong to the aviviridae family and cause severe encephalitis, yet the IL-22mediated immune regulation seems to be different in the CNS. It has been reported that IL-22 −/− mice were resistant to lethal WNV infection due to reduced in ammatory in ltration and decreased viral load in the CNS [15]. We found that in IL-22 −/− mice, infection of both WNV and ZIKV led to alleviated clinical manifestations (Fig. 2) with decreased viral load and elevated pro-in ammatory TNF-α expression in the brain ( Fig. 5A and S3) [15], whereas recombinant IL-22 cytokine treatment in vivo restored the detrimental role of IL-22 (Fig. 2E). Interestingly, there were several distinct aspects between WNV and ZIKV infection in IL-22 −/− mice. First, IL-22 was critical for virus-carrying neutrophil migration through the blood brain barrier, leading to severe WNV infection in the brain [15]; however, ZIKV-infected IL-22 −/− mice showed similar levels of lymphocyte in ltration, chemokine expression and tight junction proteins in the brain (Fig. 5D, S3 and S5). It is reported that neutrophil migration from the blood into the brain was strikingly reduced in WNV-infected IL-22 −/− mice [15]. Although we observed neutrophil in ltration in the brain during ZIKV infection, the absence of IL-22 did not change the number of in ltrated neutrophil (data not shown). These results indicated that the in ammatory in ltration and BBB may not be the main reason for the alleviated clinical manifestations in IL-22 −/− mice during ZIKV infection. Secondly, IL-22 de ciency did not contribute to the anti-WNV immunity in the periphery [15], but actually resulted in more vigorous ZIKV-speci c CD8 + T cell responses in the spleen (Fig. 5C and S6). In line with the recent ndings that CD8 + T cells protected against ZIKV infection in the CNS [31,47,48], our study suggests that IL-22 may contribute to ZIKV encephalitis pathogenesis by modulating periphery CD8 + T cell responses. Additional evidence is that IL-22 de ciency did not in uence WNV burdens in the spleen [15]; however, IL-22 −/− mice had lower viral loads in the spleen following ZIKV infection (Fig. 5A). The possible reasons for these discrepancies could be related to diversity of the two viruses and differences of the animal models.
ZIKV can target several types of glial cells, including astrocytes and microglia, leading to intracranial viral spreading, brain in ammation and fetal congenital malformations [2,9,10]. Microglia-derived TNF-α plays a critical role as an in ammatory mediator and can further activate microglia through an autocrine manner [28]. Neutralization of TNF-α or depletion of microglia prevents memory impairment in ZIKVinfected mice [49], indicating that microglia and TNF-α play detrimental roles in ZIKV infection. We showed in this study that ZIKV infection caused the activation of microglia, as evidenced by the formation of amoeba-like shapes (Fig. 3B) [25]. We also found that IL-22 de ciency resulted in reduced microglia numbers as well as cell activation ( Fig. 3A and B). In addition, decreased TNF-α expression was observed in the brain of IL-22 −/− mice (Fig. S3). Our in vivo results suggested that IL-22 may contribute to glial cell activation and induce brain in ammation, leading to cerebral pathogenesis. In addition to microglia, astrocytes are considered the initial target of ZIKV infection immediately following viral inoculation of newborn mice [10]. A recent study explored whether activated microglia can produce in ammatory factors (e.g. TNF-α, IL-1β and complement C1q) to induce astrocyte polarization into neurotoxic A1 astrocytes, which secret neurotoxins and rapidly degenerate neurons and mature differentiated oligodendrocytes [18]. Our nding that the absence of IL-22 resulted in decreased astrocyte activation (Fig. 3A), leads us to examine whether ZIKV infection promotes neurotoxic A1 polarization. Our in vitro data comprehensively demonstrated the pan-reactive and A1-prone phenotype of astrocytes during ZIKV infection, although some A2 typical gene transcripts were also upregulated (Fig. 4A). Our data indicated that ZIKV not only infects astrocytes for replication, but also directly induces astrocytes to shape a neurotoxic phenotype. Moreover, ZIKV-activated microglia may also facilitate A1 polarization through producing TNF-α and IL-1β, which were highly expressed in the brain following ZIKV infection (Fig. S1D).
Whether IL-22 can directly affect glial cells and regulate cell function is not entirely clear. IL-22 receptor was detected in human astrocytes of both healthy controls and multiple sclerosis patients, and IL-22 treatment reduced TNF-α-induced apoptosis of astrocytes [16]. In addition, exogenous IL-22 also promoted the proliferation of human glial cells accompanied by an anti-apoptotic effect [50]. To our surprise, such effects of IL-22 on astrocytes or microglia were not observed in our study using a human glial cell line, and mouse primary astrocytes or microglia ( Fig. 4 and S4). Although ZIKV infection signi cantly inhibited cell growth and elevated in ammatory gene expression, supplementing with IL-22 showed a dispensable role in cell proliferation and activation. Moreover, neither IL-22 alone nor synergizing with IFN-γ contributed to ZIKV replication in astrocytes and microglia (Fig. 4C). In addition, we were unable to detect IL-22R1 transcript in mouse primary astrocytes and microglia by qPCR (data not shown). These data suggest that IL-22 may not be capable of directly regulating glial cell function in the brain; instead, IL-22 appears to dampen antiviral T cell responses and delay viral clearance in the periphery, probably leading to more ZIKV invasion of the brain, increased glial cell activation and increased disease severity. However, we could not exclude the possibility that IL-22R1 may be inducible and upregulated by the in ammatory cytokines in the brain due to the disease status. Additionally, IL-22 may need a synergetic mechanism with critical cytokines, such as IFN-λ, to amplify the downstream signals and execute the function [51]. Although the reason for the discrepancies in ours and others' studies is not known at present, further investigation is needed to clarify the unique role of IL-22 in the CNS among distinct disease animal models.

Ethics statement
This study was carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal procedures were performed in compliance with the Institutional Animal Care and Use Committee (protocol # 1606029) at the University of Texas Medical Branch (UTMB) in Galveston, TX. All of efforts were exerted to minimize the suffering of experimental animals.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no con ict of interest.

Availability of data and materials
All data generated or analysed during this study are included in this published article and its supplementary information les.

Funding
This work was supported in part by grants from the NIH (EY028773 to JC and JS; AI132674 to LS). TG was the recipient of summer internships from an NIAID T35 training grant (AI078878, PI: LS). independently. A two-tailed Student's t-test was used to compare the two groups. One-way ANOVA was used to compare more than two groups. Log-rank (Mantel-Cox) test was used for survival curve analysis.

Author contributions
* P<0.05, ** P<0.01, *** P<0.001.   Lymphocytes were harvested from the spleen and brain at 13 dpi and stimulated with ZIKV peptide for 5 hrs in the presence of Brefeldin A. ZIKV-speci c CD8+ T cells were quanti ed by intracellular ow cytometry staining. (E) Neonatal WT mice were s.c. infected with ZIKV, followed by rIL-22 treatment as indicated in Figure 2E. Viral loads of brains were measured at 13 dpi, and (F) ZIKV-speci c CD8+ T cells were quanti ed in the spleen and brains. All experiments were repeated three times independently. Data are shown as means ± SEM and a two-tailed Student's t-test was used for statistical analysis. * P<0.05, ** P<0.01, *** P<0.001.