Retinoic Acid Modulates Hyperactive T Cell Responses and Protects Vitamin A–Deficient Mice against Persistent Lymphocytic Choriomeningitis Virus Infection

Vitamin A is one of the essential nutrients, playing an important role in physiological functions, including vision, growth, reproduction, immunity and cellular integrity (1). Vitamin A deficiency (VAD) is a significant public health problem worldwide, especially in low-income countries. Although the prevalence of VAD has declined in the past decade, which is attributed to vitamin A supplements, there is still a high prevalence in sub-Saharan Africa and south Asia (48 and 44%, respectively) among children aged 6–59 mo in 2013 (2). More than 100,000 deaths among under-5-y-olds from either diarrhea or measles are attributed to VAD in low-income and middle-income countries in 2013 (2). VAD is known to increase the risk of disease and death from severe and chronic viral infections. For instance, clinical studies have indicated that HIV-positive patients have lower serum retinol concentrations compared with HIV-negative individuals (3, 4). A high prevalence of VAD is also observed in hepatitis C patients throughout all stages of chronic liver disease, and the serum retinol concentration is related to the severity of disease development, complications and mortality (5, 6). Although the correlation of vitamin A and disease severity is observed in clinical studies, infectious diseases can precipitate VAD by decreasing intake and increasing excretion, raising a question as to whether VAD contributes to disease progress and eventual outcome (7). At present, the mechanism of vitamin A in determining anti-infectious immunity is not entirely understood.

The strategy to deliver vitamin A supplements to infants and children with or without HIV infection is performed in many countries, where VAD is a public health problem. Among HIV-positive children, vitamin A supplements reduced the mortality and morbidity; however, no beneficial effect was found for HIV-infected adults (1, 8, 9). These clinical trials suggest that vitamin A may not have direct anti-HIV effects. However, vitamin A and other retinoids have recently been demonstrated to inhibit measles virus and hepatitis C virus in vitro (10). The all-trans retinoic acid (RA), which is a predominate and natural metabolite of vitamin A, exhibited synergistic effects with PegIFN-α2a treatment in reducing viral load in HCV patients (11). These findings indicate that vitamin A and RA may regulate antiviral immune responses and benefit the host in viral infections. Recent studies highlight the role of RA in immunity and tolerance, especially in the intestinal tract. For instance, GALT CD103⁺ dendritic cells promote expression of homing markers α4β7 and CCR9 on effector T cells via RA (12, 13). The addition of RA to the spleen dendritic cell culture significantly enhanced regulatory T (Treg) cell induction in TGF-β– and retinoic acid receptor (RAR) α–dependent ways (14, 15). Th17 cells were ablated in the GALT of VAD mice at steady state; however, RA treatment suppressed Th17 responses and reduced the pathology from bacterial infection (16). It has been reported that VAD mice have aberrant immunity, including increased Th1, but limited Th2 responses (17). Vitamin A and RA can downregulate IFN-γ production through modulating IFN-γ promoters and costimulating signals (18, 19). Thus, these findings indicate that RA orchestrates T cell activation and differentiation, contributing to T cell–mediated inflammation and tissue pathology. We previously reported that adenovirus infection increased the transcript level and enzyme activity of RALDH in hepatic stellate cells, indicating that endogenous RA may play a role in immune regulation during viral infection (20). Indeed, RA treatment attenuated liver injury and modulated T cell activation in acute viral infection, highlighting the therapeutic potential of RA in modulating immune responses and limiting tissue damage (20). However, the innate and adaptive immunity in VAD mice during different stages of viral infection remains not clear.

In this study, we established a systemic viral infection in VAD mice using a widely used model pathogen, lymphocytic choriomeningitis virus (LCMV), and evaluated both innate and adaptive antiviral responses at acute, contraction, and persistent stages. We also restored the appropriate T cell responses in VAD mice by RA treatment in persistent infection. To dissect the mechanism of RA on T cell activation, we treated mice with RA and measured the levels of cytokine production and transcriptional factor expression. Mechanistically, we analyzed the phosphorylation of TCR signaling and downstream molecules in T cells by RA treatment in vitro. Our study demonstrated that VAD contributes to a hyperactive T cell response to viral infection and that RA may have the potential to serve as a therapeutic agent for excessive immune responses and tissue injury caused by infectious diseases.

C57BL/6 (B6) mice from the Jackson Laboratory were bred and maintained under specific pathogen-free conditions in the University of Texas Medical Branch (UTMB) animal care facility and used at 10–12 wk of age for breeding. Vitamin A–deficient (TD.10991) and control (20,000 IU vitamin A/kg, TD.10992) diets were purchased from Envigo (Huntingdon, U.K.). At day 14.5 of gestation, pregnant B6 females were fed with either vitamin A–deficient or a control diet and maintained on the relevant diet until weaning of the litter. Weanlings were maintained on a special diet throughout the study. Vitamin A–deficient and control mice (6–7 wk of age) were i.v. injected with 2 × 10⁶ and 1 × 10⁵ focus forming units (FFU) of LCMV strain Clone 13 and Armstrong, respectively. RA (Sigma-Aldrich, St. Louis, MO) was prepared in corn oil and DMSO was used as a control for RA. For long-term RA treatment in persistent infection, mice were i.p. injected with 25 μg RA daily. For RA treatment in acute infection, mice were i.p. injected with 200 μg RA at 1, 3, and 5 d postinfection (dpi). All procedures were approved by UTMB’s Institutional Animal Care and Use Committee and performed according to National Institutes of Health Guidelines.

Liver specimens were fixed in 10% buffered formalin. Paraffin-embedded sections were stained with H&E for histological evaluation using a modified Knodell scoring system (21).

Lymphocytes were isolated according to our previous method (22). Briefly, the liver and lungs were perfused 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-Aldrich). Spleen and lymph nodes were removed from mice and gently meshed in the RPMI 1640 medium through a cell strainer. RBCs were removed by using Red Cell Lysis Buffer (Sigma-Aldrich). Naive CD4⁺ and CD8 T⁺ cells were isolated from spleens of naive mice and purified using Naive CD4⁺ and CD8⁺ T cell isolation kits, respectively, based on the CD44⁻ phenotype (Miltenyi Biotec, Auburn, CA). The purities of the target cells were higher than 95%.

Splenocytes were cultured in an anti-CD3 Ab (5 μg/ml; BioLegend)–coated plate with soluble anti-CD28 Ab (4 μg/ml; BioLegend) for 3 or 4 d. RA was added into the culture at the beginning, and DMSO was used as a control. Brefeldin A (Thermo Fisher Scientific, Waltham, MA) was added into each well for the last 5 h of culture, followed by intracellular cytokine staining. For the T cell signaling assay, splenocytes were incubated with RA for 16 h in complete RPMI medium plus 10% FBS. Cells were collected and incubated on ice for 15 min with the anti-mouse CD3 (1 μg/ml) and anti-mouse CD28 (1 μg/ml). Cells were then stimulated by cross-linking the receptor-bound anti-CD3 and -CD28 for 10 min in a 37°C water bath with 1 μg/ml of goat anti-hamster Ig (Thermo Fisher Scientific), followed by phosflow analysis.

Naive CD4⁺ T cells and CD8⁺ T cells were labeled with CFSE (Sigma-Aldrich) and washed three times. Cells (2 × 10⁵/well) were seeded in an anti-CD3 Ab (5 μg/ml)–coated plate with soluble anti-CD28 Ab (4 μg/ml). After 4 d, cell proliferation was evaluated by flow cytometry.

At 1 d prior to viral infection, splenocytes (1× 10⁷ cells) from either naive vitamin A control (VAC) or VAD mice were adoptively transferred into CD45.1 transgenic donor mice. In the parallel experiment, splenocytes (1× 10⁷ cells) from naive CD45.1 transgenic mice were adoptively transferred into control or VAD donor mice. Mice were i.v. infected with 2 × 10⁶ FFU of LCMV Clone 13 and sacrificed at 6 dpi.

The following Abs were purchased from Thermo Fisher Scientific (San Diego, CA): Allophycocyanin–anti–IFN-γ (XMG1.2), PerCP-efluor 710–anti–TNF-α (MP6-XT22), FITC–anti-CD107a (eBio1D4B), PE–anti-Eomes (Dan11mag), allophycocyanin–anti–IL-10 (JES5-16E3), anti-Percp–Cy5.5 T-bet (ebio4B10), allophycocyanin–anti-CXCR5 (SPRCL5), PE–anti-Foxp3 (FJK-16s), and Fixable Viability Dye eFluor 506. The following Abs were purchased from BioLegend: PE-Cy7–anti-CD3 (17A2), allophycocyanin-Cy7–anti-CD8 (53-6.7), Pacific Blue–anti-CD4 (GK1.5), FITC–anti-CD19 (1D3), PE–anti–IL-2 (JES6-5H4), FITC–anti–PD-1 (J43), PE–anti-CD44 (IM3), Percp-Cy5.5–anti-CD45.1 (A20), FITC–anti-CD45.2 (104), PE–anti-NFATc1 (7A6), allophycocyanin–anti-NK1.1 (PK136), Percp-Cy5.5–anti-CTLA4 (UC10-4B9), and purified anti-CD16/32 (2.4G2). The following phosflow Abs were purchased from Cell Signaling Technology (Danvers, MA): -Phospho-Zap-70 (Tyr319)/Syk (Tyr352) (65E4), Phospho-p38 MAPK (Thr180/Tyr182) (D3F9), Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (D13.14.4E), Phospho-Akt (Ser473) (D9E), and Phospho-S6 Ribosomal Protein (Ser235/236) (D57.2.2E). The secondary Ab is anti-mouse IgG (H+L), F(ab')2 Fragment (Alexa Fluor 594 Conjugate). The PE-conjugated H-2D^b/GP33 MHC tetramer was generated by the Advanced Technology Core of MHC Tetramer at Baylor College of Medicine. The goat anti–hamster Ig was from Southern Biotech (Birmingham, AL). The RAR α-selective antagonist BMS 195614 was from Cayman Chemical (Ann Arbor, MI). For surface staining, cells were first incubated with FcγR blocker (CD16/32), followed by fluorochrome-labeled Abs of surface markers. For intracellular cytokine staining, GP33 and GP61 peptides (5 μg/ml; AnaSpec, Fremont, CA) were used in the presence of brefeldin A (BD Bioscience, San Jose, CA) to stimulate virus-specific CD8⁺ and CD4⁺ T cell responses, respectively. PMA (50 ng/ml) and ionomycin (750 ng/ml) from Sigma-Aldrich were also used in some experiments. After incubation, cells were stained for surface markers first, fixed by using an IC fixation buffer, and stained for intracellular cytokines (Thermo Fisher Scientific). For transcriptional factor staining, cells were fixed and permeabilized by eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific). The phosflow experiments were performed according to the protocol from BD Bioscience (23). Briefly, cells were fixed immediately at the end of stimulation using a prewarmed BD Phosflow Lyse/Fix Buffer at 37°C for 12 min. Cells were permeabilized using chilled BD Phosflow Perm Buffer III for 1 h on ice, then washed and stained with Abs. Samples were processed on an LSRII FACSFortessa (Becton Dickinson, San Jose, CA) and analyzed by using FlowJo software (TreeStar, Ashland, OR).

Tissue protein was extracted in RIPA buffer (Cell Signaling Technology) and quantified using a BCA kit (Thermo Fisher Scientific). Liver cytokine profiles and serum IFN-α/β were characterized respectively using Cytokine 17-Plex Mouse ProcartaPlex Panel and IFN-α/IFN-β 2-Plex Mouse ProcartaPlex Panel (Thermo Fisher Scientific). The samples were read on a Bio-Rad Bio-Plex 200 System. Raw data were measured as the relative fluorescence intensity and then converted to the concentration according to the standard curve. For detecting IFN-γ and TNF-α in various organs, ELISA kits from BD Bioscience were also used.

The LCMV stocks were prepared and titrated according to a modified method. Briefly, virus was incubated with baby hamster kidney cells for 72 h. The culture fluid was centrifuged for 10 min at 350 × g, 4°C and stored at −70°C. For quantitation of the virus, Vero cells were cultured with a series of 10-fold virus dilutions for 90 min, followed by a methylcellulose overlay. After 4 d of culture, cells were first incubated with mouse anti-LCMV polyclonal Ab (Fitzgerald, Acton, MA), followed by incubation with peroxidase (HRP)-conjugated anti-mouse IgG (Southern Biotech). The AEC HRP Substrate Kit (Enzo Life Sciences, Farmingdale, NY) was used for immunocytochemical procedures. Viral titers were calculated by counting the numbers of positive clusters.

Data were shown as mean ± SEM and analyzed by using the two-tailed Student t test when compared between two groups. One-way ANOVA was used for statistical analysis of more than two groups. For analysis of the histological scores, the nonparametric Mann–Whitney U test was used, and *, **, and *** indicate p values <0.05, <0.01, and <0.001, respectively. Statistical analyses were performed using Prism 5.0 (GraphPad Software, San Diego, CA).

To investigate whether VAD affects type I IFN responses, we measured serum levels of IFN-α and IFN-β of LCMV-infected mice at 12, 24, and 48 h postinfection (hpi). Comparable levels of IFN-α and IFN-β, as well as viremia, were observed in VAD and control mice at these time points (Fig. 1A, 1B), indicative of relatively normal innate antiviral responses in VAD mice during acute infection. Vitamin A is important for immune cell development and activation (24, 25). Naive VAD and control mice had comparable T and B cell compartments in the spleen. However, VAD mice had lower percentages of B cells in the liver (Supplemental Fig. 1A), which stores most of the body’s vitamin A. These aberrant T and B cell ratios were also observed in VAD mice in the liver, lung, and mesenteric lymph nodes but not in the spleen, inguinal, or cervical lymph nodes at 2 dpi (Fig. 1C, 1D, Supplemental Fig. 1B). In addition to the altered cell ratios, T cells in the liver, lung, and mesenteric lymph nodes of VAD mice displayed higher activation status with upregulated IFN-γ expression (Fig. 1E). VAD mice exhibited increased liver injuries, as evidenced by higher serum alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, and pathologic changes at 7 dpi (Fig. 2A, 2B), which were associated with overzealous LCMV-specific T cell responses manifested by increased frequencies and numbers of IFN-γ⁺TNF-α⁺ T cells in the spleen and liver 7 dpi (Fig. 2C). The IFN-γ and TNF-α protein levels were also higher in the liver of VAD mice compared with that in control mice (Fig. 2D, Supplemental Fig. 1C). We further analyzed virus-specific CD8⁺ T cells and found increased GP33-tetramer⁺ CD8⁺ T cells in the liver of VAD mice. Importantly, VAD resulted in downregulation of PD-1 on virus-specific CTLs (Fig. 2E). These exaggerated T cell responses in VAD animals were detrimental and did not lead to enhanced viral clearance at 7 and 15 dpi, as similar levels of viremia were observed in VAD and control mice (Supplemental Fig. 1D). We also measured transcriptional factor T-bet and Eomes as well as CXCR5 on CD8⁺ T cells. The levels of T-bet were higher in the spleen of VAD mice, whereas T-bet was expressed on most of liver CD8⁺ T cells and its level was comparable between VAC and VAD samples (Supplemental Fig. 2A). Expression of CXCR5 on CD8⁺ T cells were relative low and comparable in two groups. Although no difference was observed by Treg cell numbers, lower IL-10, but not CTLA-4, expression was found in VAD mice (Supplemental Fig. 2B). We also infected mice with LCMV Armstrong and found the higher levels of ALT and AST in VAD mice at 6 dpi (Supplemental Fig. 2C). Consistent with LCMV Clone 13 infection, the LCMV Armstrong–infected VAD mice displayed stronger immune responses, as evidenced by increased numbers of cytokine-producing T cells and higher expression of granzyme B and T-bet in spleen CD8⁺ T cells (Supplemental Fig. 2D, 2E).

The normal immune system enters the contraction phase following clonal expansion and viral reduction in the acute phase (26). In this study, we examined the effect of VAD on immune regulation at 30 d after LCMV infection. We found that the levels of TNF-α and IL-2 in liver tissues were significantly higher in VAD mice compared with those in control mice (Fig. 3A). In addition, other proinflammatory cytokines, including IL-1β, IL-6, IL-18, and IL-23, as well as anti-inflammatory IL-10 and IL-13, were also elevated in VAD mice (Fig. 3A). PD-1 is an important checkpoint regulator (27) and plays a crucial role in immune contraction, T cell exhaustion and viral persistence during infection. Flow cytometry data showed that effector T cells of VAD mice displayed increased frequencies of IFN-γ⁺TNF-α⁺ and IFN-γ⁺IL-2⁺ T cells, with significantly lower frequencies of PD-1⁺ T cells in the spleen, liver, and lungs at 30 dpi (Fig. 3B–D). We further found the increased percentages and numbers of cytokine-expressing GP33 tetramer⁺ CD8⁺ T cells in spleen and livers of VAD mice (Fig. 3E). In line with the stronger T cell immune response in VAD mice at the contraction stage, their viremia was lower compared with that of control mice (Supplemental Fig. 1D). Despite exaggerated responses of the T cell compartment in this phase, the percentages of B lymphocytes in the spleen, liver, and lungs of VAD mice were significantly lower than those in control animals (data not shown). Together, these results demonstrated that VAD can cause profound dysregulation of adaptive immune responses to LCMV infection. Increased T cell activation and proinflammatory cytokine production can lead to tissue injury in viral infection.

RA is one of the metabolites of vitamin A, and it plays an important role in immune function and regulation (28). We reported previously that RA treatment can attenuate liver inflammation and injury by restraining T cell functions in acute viral infection (20). However, whether RA can modulate exaggerated T cell responses during persistent infection is not known. In this study, we treated the infected VAD mice with RA daily starting from 12 dpi and found that RA-treated VAD mice all survived, whereas 60% of VAD mice without RA treatment succumbed at 48 dpi (Fig. 3F). The histological results showed that the liver inflammation was mild and comparable in two groups; however, RA-treated VAD mice had lower pathogenic inflammation in the lungs at 30 dpi (Fig. 3G). Although RA treatment did not significantly inhibit viral loads in organs (Fig. 3H), it decreased the percentages and numbers of hyperactive effector cells, including IFN-γ⁺TNF-α⁺ and IFN-γ⁺IL-2⁺ CD8⁺ T cells at 30 dpi (Fig. 3I). These results demonstrated that RA treatment can partially restore hyperactive T cell functions and rescue VAD mice in persistent LCMV infection.

Normal mice reduce viral burden and survive persistent infection (29) (Fig. 4A). However, LCMV-infected VAD mice succumbed to the infection beginning at 32 dpi, and all died at 52 dpi (Fig. 4A). At 48 dpi, the moribund VAD mice exhibited a 3-fold increase of PD-1 molecules on splenic CD8⁺ T cells compared with those of control mice (Fig. 4B). Additionally, around 80% of CD8⁺ T cells in the liver of VAD mice displayed a PD-1⁺ exhausted phenotype, compared with only half as much in control animals (Fig. 4B). The LCMV-specific multifunctional CTL responses were much lower in VAD mice, as evidenced by lower proportions of triple (IFN-γ⁺TNF-α⁺IL-2⁺ and IFN-γ⁺TNF-α⁺CD107a⁺), double (IFN-γ⁺CD107a⁺ and IFN-γ⁺TNF-α⁺), and single (IFN-γ⁺ and CD107a⁺) positive CD8⁺ T cells (Fig. 4C). Moreover, significant amounts of virus were detectable in the serum and lungs of VAD mice at 48 dpi (Fig. 4D). These data demonstrated that VAD resulted in T cell exhaustion, uncontrolled viremia, and animal death in persistent LCMV infection.

To determine whether VAD causes anomalous thymic development of T cells or affects T cell functionality in the periphery, we evaluated the T cell activation and effector functions in vitro. We found that naive T cells from both VAD and control mice performed equally well in proliferation and IFN-γ production in response to CD3/CD28 stimulation (Fig. 5A, 5B). To further determine the effect of VAD on T cell development in vivo, we isolated splenocytes from VAD and control mice on the CD45.2 background and adoptively transferred them into congenic CD45.1 mice, followed by LCMV infection. We found that donor cells from VAD or control mice displayed similar activation status in recipient animals, as evidenced by comparable cell percentages and cytokine production at 6 dpi (Fig. 5C). These results indicated that VAD did not grossly impair the thymic development of T cells.

To test if VAD affects T cell functionality in the periphery, we harvested naive splenocytes from CD45.1 congenic mice and transferred them into CD45.2 recipient mice in either aVAD or control environment, followed by LCMV infection. We found that effector functions of donor T cells were elevated in VAD recipients compared with those in control mice, as evidenced by increased percentages of effector cytokine-producing CD45.1⁺ T cells in VAD mice (Fig. 5D). These data suggested that vitamin A in the peripheral organs played a critical role in regulating T cell functions and maintaining tissue homeostasis during viral infection.

The mechanism by which RA modulates T cell function in viral infection is not fully elucidated. TCR signaling plays key roles in T cell activation and differentiation of effector functions (30). To test the hypothesis that RA can regulate T cell responses through modulating TCR signaling, we cultured spleen cells of naive mice in the presence of RA for 16 h. First, we found that RA of <10 μM concentrations was not cytotoxic. A higher dose of RA (20 μM) was toxic to T cells and reduced cell viability (Supplemental Fig. 3A). By using Phosflow analysis, we found that RA treatment significantly attenuated TCR/CD28-mediated phosphorylation of several downstream signaling molecules of TCR, including Zap70, ERK, AKT, and p38. p-S6, a marker for mammalian target of rapamycin (mTOR) signaling pathway activation, was also reduced in T cells by RA treatment (Fig. 6A, Supplemental Fig. 3B). Collectively, these data suggest a crucial role for RA in regulating TCR-proximal signaling events in both CD4⁺ and CD8⁺ T cells.

NFAT is a family of transcription factors important in T cell responses (31). As NFAT proteins are downstream of the TCR signaling pathway, we treated the LCMV-infected mice with RA and analyzed T cell activation at 7 dpi. Consistent with our previous finding (20), RA treatment significantly inhibited T cell activation in vivo as evidenced by decreased frequencies of IFN-γ– and IFN-γ TNF-α–producing T cells in both the spleen and liver (Fig. 6B). NFATc1 expression was also significantly reduced in CD8⁺ T cells of RA-treated mice (Fig. 6C). To confirm this finding, we cultured spleen cells with anti-CD3/CD28 Abs in the presence of RA for 3 d. RA treatment significantly reduced the percentages of IFN-γ– and IFN-γ/TNF-α–expressing CD8⁺ and CD4⁺ T cells, as well as frequencies and levels of NFATc1 expression in these cells (Fig. 6D). A reduction of NFATc2 was also observed in RA-treated T cells (data not shown). Furthermore, the neutral RAR α-selective antagonist BMS 195614 (1 μM) completely restored frequencies and levels of NFATc1 expression in RA-treated CD8⁺ and CD4⁺ T cells (Supplemental Fig. 4). These data suggest that RA can restrain T cell effector functions by inhibiting a number of TCR downstream signaling molecules.

VAD is a public health problem, especially in sub-Saharan Africa and South Asia, leading to an increased risk of disease and death from severe infection (2). Vitamin A plays an essential role in immune processes, including immune cell development, proliferation, differentiation, migration, and Ab production; however, its role and mechanism in viral infection are not entirely understood. VAD is linked with an abnormality of immune cell development. For instance, vitamin A is critical for B1 cell development through regulating the transcriptional factor NFATc1 (25). We observed decreased percentages of B cells in livers and mesenteric lymph nodes but not in the spleen, inguinal, or cervical lymph nodes at early stages during viral infection (Fig. 1C, 1D). Notably, aged VAD mice also displayed decreased numbers of B cells in the spleen (25). Because vitamin A is mainly stored in the liver, our results indicate that the impaired immune cell development by VAD may occur first in the gut–liver axis (32). Moreover, we provided both in vitro and in vivo evidence that T cells from VAD mice displayed comparable capabilities for proliferation and activation as those from control mice (Fig. 5). Other researchers also reported that VAD results in more myeloid cells in the periphery (33). These findings suggest that vitamin A may differently regulate the development of immune cell subpopulations.

Malnutrition may lead to abnormal type I IFN responses (34), which is essential for viral control and following adaptive virus-specific immunity (35). Vitamin A and RA have been demonstrated to inhibit viral replication through induction of IFN-stimulated genes and upregulation of IFN-αR in vitro (10, 36); however, we found comparable type I IFN production and viremia in VAD mice at 12, 24, and 48 hpi (Fig. 1A, 1B), indicative of the dispensable role of endogenous vitamin A and retinols for the type I IFN response and early viral control. In addition to innate immunity, vitamin A and its metabolite RA are essential for T cell activation and differentiation (Figs. 2, 3), which determine the outcome of viral infection. Several reports demonstrated recently that RA inhibits Th17 polarization via inhibiting IL-6 signaling (16, 37, 38) but increases TGF-β–mediated Treg cell expansion and Foxp3 expression in naive T cells through TGF-β–enhanced RAR (15, 39). Overproduction of IFN-γ was reported in VAD mice, whereas dietary vitamin A supplements downregulated IFN-γ expression in Th1 cells (17, 18, 40). These studies suggest vitamin A and RA act as potent immunomodulators for T cell responses. In this study, we found significantly increased numbers and cytokine production of T cells in the liver, lung, and mesenteric lymph nodes of VAD mice at 2 dpi (Fig. 1C–E). This overzealous T cell response was further observed in both acute and contraction stages, resulting in increased inflammatory cytokines and severe tissue damage (Figs. 2, 3). Vitamin D is reported to upregulate PD-1/PD-L1 signaling in Crohn disease patients (41), indicating that vitamin might be involved in checkpoint regulation. In our study, the lower levels of PD-1 expression in various organs of VAD mice may suggest the essential role of vitamin A in the regulation of immune checkpoint molecules. However, the underlying mechanism is still not well known and needs further investigation. Interestingly, naive T cells isolated from either VAD or control mice displayed a similar capacity for proliferation and IFN-γ production in vitro and in vivo. However, donor T cells in adoptive transfer experiments exhibited a more activation phenotype in VAD recipient mice (Fig. 5), suggesting that the upregulated T cell response in VAD mice was attributed to the lack of environmental vitamin A. In addition, we also found decreased B cell percentages in VAD mice at 30 dpi (data not shown), suggesting that VAD mice may have impaired Ab production in persistent infections (17). Our results highlight the role of vitamin A in regulating antiviral T cell responses and subsequent tissue pathogenesis following viral infection. However, the increased T cell responses in VAD mice may also contribute to viral clearance in the contraction stage (Supplemental Fig. 1D). Interestingly, VAD led to impaired multifunctional T cell responses, an inability to clear the virus, and animal death in the persistent infection (Fig. 4). Moreover, VAD may also lead to impaired T regulatory cell function (Supplemental Fig. 2B). These findings indicate that the overzealous T cell activity and immune-mediated tissue damage may ultimately cause T cell exhaustion and viral relapse in chronic infection. Importantly, RA treatment delayed VAD mice death in the persistent infection (Fig. 3F). It is reported that LCMV-infected mice may die because of the leakage of exudate into the lung that compromised respiration (29). We observed that LCMV-infected VAD mice did not have any obvious signs of illness (data not shown). Because vitamin A plays a key role in maintaining lung epithelial integrity, it is highly possible that VAD mice may succumb to infection due to lung pathology. Our histological staining results showed that RA treatment may rescue mice through decreasing lung inflammation (Fig. 3G). Further analysis showed that RA treatment decreased the numbers of cytokine-producing T cells during the T cell contraction stage (Fig. 3I). Therefore, our results suggest to us that dietary RA supplements may inhibit the overactivation of T cells and protect against T cell exhaustion in chronic infections.

Our previous and current studies have demonstrated that RA treatment inhibited T cell activation both in vitro and in vivo (Fig. 6) (20). RA contributes to T cell responses through several ways, including modulating dendritic cell migration, inducing immunosuppressive molecule arginase-I, regulating cytokine production, and determining T cell–homing markers (20, 28). The strength of the TCR signal is pivotal for T cell activation and differentiation. TCR signaling is initiated by the activation of the protein tyrosine kinase Lck and the phosphorylation of the TCR-signaling chain CD3ζ, followed by recruitment and phosphorylation of the tyrosine kinase Zap70 (42, 43). Activated Zap70 then phosphorylates several other signaling molecules, including p38, ERK, NF-κB, AKT, and mTOR, transducing the TCR signals to downstream signaling pathways. In this study, we found that RA can target TCR signaling and associated downstream molecules. By using the CD3/CD28 cross-linking stimulation, we found that RA negatively regulated the TCR-proximal signaling, as evidenced by decreased phosphorylation of Zap70. Consistently, the phosphorylation of several downstream signaling molecules, including p38, ERK, AKT, and S6, were all downregulated by RA in a dose-dependent manner (Fig. 6). It has been demonstrated that the strength of the TCR signal controls Th1/Th2 polarization (44–46). A weak TCR signal primes IL-4–producing Th2 cells, whereas a strong TCR signal drives Th1 differentiation. Accordingly, upon TCR engagement, the phosphorylation profiles were weaker in Th2 than that in Th1 clones. Interestingly, RA treatment induces a Th2 polarization and impairs Th1 responses in both human and mouse purified T cells under TCR stimulation (47, 48). Thus, RA may contribute to T cell differentiation by regulating TCR signals. Notably, it is reported that RARα-deficient CD4⁺ T cells have poor effector responses and impaired signaling activation by anti-CD3 stimulation, indicating that RA is necessary for T cell activation (49). Although the reason for the discrepancies from different models is not known at present, we demonstrated in this study that exogenous RA played a critical role in regulating TCR signals.

The transcriptional factor NFATs are downstream of the TCR, playing a critical role in the development and differentiation of immune cells (31). T cells can express NFATc1 (NFAT2), NFATc2 (NFAT1), and NFATc3 (NFAT4). The expression of NFTAc1 is inducible and can be upregulated by TCR stimulation, resulting in the activation of T cells (50). We demonstrated in this study that RA treatment can decrease the expression of NFATc1 in CD8⁺ T cells in vivo (Fig. 6C). An in vitro study showed that RA inhibited NFATc1 expression in both CD4⁺ and CD8⁺ T cells, whereas the selective RARα antagonist restored NFATc1 expression in the presence of RA (Fig. 6D, 6E, Supplemental Fig. 4). Our results may imply that RA modulates T cell function through regulating NFATs (51, 52). However, it is reported that VAD mice have decreased NFATc1 expression in B1 cells, resulting in an inability to mount an Ab response against bacterial infection (25). The discrepancies in these studies indicate that RA may diversely regulate NFAT functions in different immune cells.

In conclusion, our study revealed that VAD caused an aberrant T cell response to viral infection, resulting in increased tissue damage and host mortality, whereas RA can downregulate T cell activation and rescue mice from death in the persistent infection. VAD did not cause an intrinsic deficiency of T cell function, but the microenvironmental vitamin A or its metabolites played an essential role in T cell function during viral infection. Moreover, RA may modulate T cell activation by regulating TCR signals and NFAT transcriptional factors, indicating that RA may represent a potential immunomodulator in infectious diseases.

We thank Dr. Sherry Haller for assistance with manuscript preparation. The authors wish to express gratitude to other members of the UTMB Joint Immunology Working Group (Drs. Cong, Stephens, Rajsbaum, Hu, and their trainees) for many helpful discussions.

The authors have no financial conflicts of interest.