Nutritional Immunology

Vitamin A Deﬁciency Increases the In Vivo Development of IL-10–Positive Th2 Cells and Decreases Development of Th1 Cells in Mice1,2 Charles B. Stephensen,3 Xiaowen Jiang, and Tammy Freytag U.S. Department of Agriculture, Western Human Nutrition Research Center at the University of California, Davis, and Nutrition Department, University of California, Davis, CA 95616

ABSTRACT Vitamin A deﬁciency impairs both T helper type 1 (Th1)- and type 2 (Th2)-mediated immune

responses, although Th2 responses seem to be principally affected. Multiple mechanisms are involved in this Downloaded from immune suppression, but the hypothesis that deﬁciency affects development of Th1/Th2 memory cell phenotype has not been tested directly in vivo. To do so, lymphocytes from DO11.10 T cell receptor (TCR)-transgenic mice were transferred to vitamin A–deﬁcient or control BALB/c recipients. Recipients were then immunized with the

cognate peptide antigen for the TCR-transgenic DO11.10 T cells (OVA323–339). After 2–5 wk, the transferred OVA323–339–speciﬁc T cells were identiﬁed from draining lymph nodes with the TCR-clonotypic antibody KJ1–26,

and their Th1/Th2 phenotype was characterized by intracellular cytokine staining after in vitro stimulation with jn.nutrition.org phorbol myristate acetate and ionomycin. The percentage of CD4ϩKJ1–26ϩ cells positive for IL-10 was 100% greater in vitamin A–deﬁcient mice (3.49 Ϯ 0.41%; mean Ϯ SE) than in control mice (1.74 Ϯ 0.37%). IL-4 did not differ between groups. In addition, the percentages of CD4ϩKJ1–26ϩ cells from vitamin A–deﬁcient mice that were positive for interferon (IFN)-␥ (8.8 Ϯ 0.73%) and interleukin (IL)-2 (39.5 Ϯ 3.1%) were both lower than the

percentages in control mice (11.4 Ϯ 0.67 and 47.0 Ϯ 2.8%, respectively). Thus vitamin A deﬁciency, at the time of at USDA, National Agricultural Library on December 11, 2007 initial antigen exposure, enhances the development of IL-10–producing Th2 or T regulatory cells and diminishes the development of Th1 memory cells. J. Nutr. 134: 2660–2666, 2004.

KEY WORDS: ● rodent ● Th1/Th2 ● immunomodulators ● transgenic/knockout ● vitamin A deﬁciency

Population-based studies showed that vitamin A supple- IgG1, IgE, and IgA antibody responses, are impaired by vita- ments decrease the risk of death from infectious diseases (1), min A deficiency, whereas some aspects of Th1-mediated presumably by restoring innate and adaptive immune func- responses may be enhanced. In particular, production of in- tions compromised by vitamin A deficiency (2). However, terferon (IFN)-␥ (a Th1 cytokine) on a per cell basis is some community and clinical studies found that supplements increased during vitamin A deficiency and can be directly may increase the severity of respiratory infections (3,4) or the decreased by treatment with retinoic acid, the principal active risk of vertical transmission of HIV (5). The reasons for these metabolite of vitamin A (9,10). Because IFN-␥ can impede disparities are unclear but highlight the need to understand the development of Th2 responses, this may be one mecha- the mechanism by which vitamin A affects immune function. nism by which vitamin A deficiency decreases T-lymphocyte– Mechanistic studies are difficult to pursue in human sub- mediated antibody responses. jects, and rodent models have provided valuable insights into Because Th2 responses are diminished by vitamin A defi- the effect of vitamin A deficiency on immune function. For ciency, it is logical to ask whether the number of Th2 memory example, vitamin A deficiency decreases T-lymphocyte–medi- cells may be lower in vitamin A–deficient mice. The answer to ated antibody responses (6,7), a finding that was replicated in this question has implications for long-term immunity. For humans (8). These and other studies indicate that immune example, if the number of Th2 cells is decreased, this could responses mediated by T-helper type 2 (Th2)4 cells, such as mean that the memory response would remain impaired even after correction of vitamin A deficiency. Two principal ap- proaches have been taken to address this question. First, lymphocyte populations from vitamin A–deficient and control 1 Presented in part at Experimental Biology 04, April 2004, Washington, DC [Stephensen, C. B., Jiang, X. & Freytag, T. L. (2004) Vitamin A deficiency mice were analyzed for the presence of cells (presumably T increases development of IL-10-positive Th2 or Treg cells. Experimental Biology lymphocytes) capable of providing help to IgG1-producing B meeting abstracts (accessed at http://www.biosis-select.org/faseb/index.html). lymphocytes. Such functional studies indicate that vitamin A FASEB J. 18: 37.2 (abs.)]. 2 Supported by U.S. Department of Agriculture National Research Initiative grant #97–35200–4229, USDA CRIS Project #5306–51530–006-00D and Na- tional Institutes of Health grant #1 R01 AI 50863. incomplete Freund’s adjuvant; IFN, interferon; IL, interleukin; M, male; OVA323–339, 3 To whom correspondence should be addressed. peptide comprised of amino acids 323 through 339 of ovalbumin; PE, phycoerythrin; E-mail: [email protected]. PerCP, peridinin chlorophyll protein; PF, paraformaldehyde; PMA, phorbol myristate 4 Abbreviations used: APC, allophycocyanin; CFA, complete Freund’s adju- acetate; TCR, T cell receptor; Th1, T helper type 1; Th2, T helper type 2; Treg, T vant; F, female; FBS, fetal bovine serum; FITC, fluorescein isothiocyanate; IFA, regulatory cell.

2660 VITAMIN A DEFICIENCY AND Th1/Th2 DEVELOPMENT 2661 deficiency decreases the number of such cells, indirectly dem- axillary, popliteal) were collected immediately after killing into cold onstrating a reduction in Th2 numbers (11) (12). A limitation HBSS containing 2% fetal bovine serum (FBS). Tissues were dis- of such studies is that increased production of IFN-␥ could also rupted with a few strokes of a Duall glass tissue homogenizer (Kontes have the same effect as decreased production of Th2 cytokines Glass) in HBSS:2% FBS and filtered using filter-top 5-mL tubes such as interleukin (IL)-4. Another study used an Elispot (Falcon #352063) to eliminate debris. Cells were centrifuged for 10 min at 200 ϫ g at 4°C and resuspended in Russ-10 media, made as method to demonstrate that the number of IL-5 producing described (16). The retinol concentration in the lot of FBS used at cells was decreased by vitamin A deficiency although the 10% in the Russ-10 media was 0.77 Ϯ 0.011 ␮mol/L; thus, the number of IFN-␥ producing cells was not affected (10). Al- concentration of retinol in the cell culture media was 0.077 ␮mol/L. though the type of Th1/Th2 cytokine was identified in this Splenocytes were treated with lysis buffer to eliminate erythrocytes study, the specific identity of the cells was not established. (17). Cells were resuspend in Russ-10 at 2 ϫ 106 cells/0.5 mL, and 0.5 In the present study, we addressed the question whether mL/well was placed in a 48-well plate followed by 0.5 mL of stimu- vitamin A deficiency alters the development of Th1 and Th2 lation medium. Stimulation medium consisted of Russ-10 medium memory cells using an adoptive transfer model. Na¨ıve T lym- plus PMA (Sigma; 40 ␮g/L) and ionomycin (Sigma; 2 ␮mol/L). Cells phocytes from DO11.10 ␣/␤-T-cell receptor (TCR) transgenic were then cultured at 37°C in 5% humidified CO2. Brefeldin A was mice (13) were transferred into vitamin A–deficient and con- added to block cytokine secretion after2hatafinal concentration of 10 mg/L. Incubation was continued for an additional 3 h. trol BALB/c mice. The DO11.10 TCR recognizes amino acids Downloaded from After incubation, the contents of each well were then placed in 5 323–339 of ovalbumin (OVA323–339) and can be identified mL polystyrene tubes (Falcon) with 2 mL Dulbecco’s PBS containing using the clonotypic monoclonal antibody KJ1–26 (14). After 2% FBS and centrifuged at 200 ϫ g for 10 min at 4°C. PBS:2% FBS transfer, recipient mice were immunized with OVA323–339 in was used for antibody dilutions and subsequent wash steps. The incomplete Freund’s adjuvant (IFA) to stimulate development ␮ supernatant was then gently poured off and 1 gFc-block/million of memory T lymphocytes. The Th1/Th2 phenotype was de- cells (Pharmingen #01241A) was added in a total volume of 10 termined 2–5 wk after immunization by intracellular cytokine ␮L/million cells. Cells were incubated for 15 min at 4°C. Antibodies staining. Vitamin A–deficient mice had a higher frequency of for surface staining were then added in a total volume of 10 ␮L/ jn.nutrition.org IL-10–positive Th2 memory [or T regulatory (Treg)] cells and million cells at a concentration of 0.2 ␮g/million cells and incubated a lower frequency of IFN-␥- and IL-2–positive Th1 memory for 20 min at 4°C in the dark. These antibodies included peridinin cells than did control mice. A similar difference in supernatant chlorophyll protein (PerCP)-labeled rat anti-CD4 (Pharmingen #553052), a PerCP-labeled rat IgG2a control antibody (Pharmingen cytokine concentrations was found in splenocyte cultures re- at USDA, National Agricultural Library on December 11, 2007 #553933), allophycocyanin (APC)-labeled clonotypic TCR-specific stimulated with OVA323–339 peptide. mouse antibody KJ1–26 (Caltag #MM7505) and an APC-labeled mouse IgG2a control antibody (Caltag #MG2905). After surface MATERIALS AND METHODS staining, cells were washed once and then fixed with 0.5 mL 1% paraformaldehyde (PF) in calcium and magnesium-free PBS (1% PF). Mice. DO11.10 mice were a gift from Casey Weaver, University Cells were then stored at 4°C in the dark overnight. of Alabama at Birmingham, and were bred in our facility. BALB/c On the following morning, cells were centrifuged at 200 ϫ g at mice were purchased from Charles River. This research was approved ␮ by the U.C. Davis Institutional Animal Care and Use Committee. 4°C, the 1% PF buffer gently poured off, and 250 L Cytofix/ Diet protocol. Pregnant mice consumed a semipurified vitamin Cytoperm buffer (Pharmingen) was added. Cells were incubated at A–deficient diet (#88407; Teklad Diets) ad libitum beginning on 4°C in the dark for 15 min. Cells were then washed once with 2 mL gestation d 14. The number of pups per litter was balanced within Perm/Wash buffer (Pharmingen) and stained with cytokine-specific and control antibodies (0.5 ␮g/million cells) diluted in Perm/Wash 24 h of birth so that litter sizes were similar between diet groups and ␮ so that no litter had Ͼ5 pups. Half of the litters in each experiment buffer (10 L/million cells) for 60 min at 4°C in the dark. Cells were were then switched to the control diet containing 4000 IU vitamin stained with phycoerythrin (PE)-labeled anti-IL-4 (Pharmingen #554435) or a PE-labeled control rat IgG1 (Pharmingen #553925) A/kg as retinyl palmitate (Teklad diet #88406). Pups were weaned at ␥ 21 d of age to the same diet fed to the dams. Serum retinol was plus fluorescein (FITC)-labeled anti-IFN- (Pharmingen #554411) or measured at 7 wk of age in 4 mice per group to determine whether the FITC-labeled rat IgG1 control antibody (Pharmingen #553995). In mice fed the vitamin A–deficient diet had serum retinol that was addition, a separate tube of cells was stained with PE-labeled anti- Ͻ0.50 ␮mol/L and significantly lower than that of the control diet IL-10 (Pharmingen #554467), or a PE-labeled control rat IgG2b group. This protocol and the diet composition were described previ- (Pharmingen #553989) plus FITC-labeled anti-IL-2 (Pharmingen ously (15). #554427), or a FITC-labeled rat IgG2b control antibody (Pharmin- Two experiments with identical protocols were conducted. Intra- gen #553988). Cells were then washed with 2 mL Perm/Wash buffer cellular cytokine staining data were combined from the 2 experiments and resuspended in 0.5 mL 1% PF buffer. after statistical comparison showed that the results from these exper- Cells were analyzed using a FACSCalibur 4-color flow cytometer iments were essentially identical. Lymph node and spleen cultures (Becton Dickenson) as outlined in Figure 1. Duplicate cultures were were performed in the second experiment for measurement of cyto- prepared from each mouse for each pair of cytokine antibodies, and a kine concentrations by ELISA. single culture was used for each isotype control pair; Adoptive transfer and immunization. After vitamin A defi- 200,000–300,000 events in the lymphocyte gate were collected for ciency was confirmed using serum retinol concentration, mice in the each culture. To correct for nonspecific staining, the percentage of 2 diet groups received 5 ϫ 106 CD4ϩKJ1–26ϩ cells by tail vein cells positive with the appropriate isotype control antibody was injection. Splenocytes and lymph node cells from DO11.10 mice subtracted from the percentage of cells positive with the cytokine- were pooled and analyzed by flow cytometry, as described below, to specific antibody. Means of the duplicates were determined and used determine the number of total cells to be injected. The CD4ϩKJ1– for statistical analysis. Four mice (2 deficient and 2 control) were 26ϩ cells were not separated before transfer. Cells were suspended in analyzed on a given day to help control for possible day-to-day HBSS. The total injection volume was 0.15 mL. variation. During wk 3 and 4, values for IL-4 from most mice were Three days after transfer, OVA323–339 peptide (Alpha Diagnostic negative (after correction for the isotype control), and these values International) was diluted in PBS and emulsified with incomplete were adjusted to zero. Freund’s adjuvant (IFA; Sigma). The final injection volume was 0.3 CD62L staining. In the second experiment, lymph node cells mL/mouse and contained 300 ␮g peptide. Mice were injected s.c. in from selected mice after immunization were stained with PE-labeled both hind footpads (50 ␮L per site) at the base of the tail (100 ␮L) anti-CD4 (Pharmingen #553730), FITC-labeled KJ1–26 (Caltag and the back of the neck (100 ␮L). #MM7501–3) and biotin-labeled anti-CD62L (Pharmingen Cell culture and intracellular cytokine staining. Two to 5 weeks #01262D). CyChrome-labeled strepavidin was used as the second after immunization, the spleen and 6 draining lymph nodes (brachial, reagent (Pharmingen #13038A). The percentage of CD4ϩKJ1–26ϩ 2662 STEPHENSEN ET AL.

FIGURE 1 Flow cytometric analysis for identiﬁcation of cytokine-producing CD4ϩKJ1–26ϩ T-cells in draining lymph nodes of BALB/c mice after adoptive transfer of lymphocytes from DO11.10 mice and immunization with

OVA323–339 peptide. Cells were treated with PMA and ionomycin to stimulate cytokine production. Lymphocytes were identiﬁed by forward and side scatter (A, D). CD4ϩKJ1–26ϩ T cells from the lymphocyte gate were then identiﬁed as double-positive cells (B,

E). The percentages of such double- Downloaded from positive cells positive for IL-10, IL-2 (C), IL-4 or IFN-␥ (F) were then enu- merated. jn.nutrition.org

cells with a memory phenotype (CD62Llo) was determined by a observations were available in each diet group on all culture days (d fluorescence-activated cell sorter. The cutoff level for CD62Llo cells 2, 4/5 and 7) for OVA323–339–stimulated as well as unstimulated was determined on each day using the predominant CD62Lhi popu- cultures, for both males and females. Because of this balance, 3-way at USDA, National Agricultural Library on December 11, 2007 lation from the CD4ϩKJ1–26Ϫ cells from the same sample as a ANOVA could be used to compare cytokine concentrations among reference. these groupings. The initial analysis simultaneously compared males Cell culture and cytokine ELISA. Lymph node and spleen cells vs. females, days in culture, and stimulated vs. unstimulated cultures. were cultured at 2 ϫ 106 cells/mL in Russ-10 media containing either Cytokine concentrations did not differ between males and females; 6 mg/L OVA323–339 or an equal volume of diluent. Cultures were thus, the data from males and females were analyzed together to then placed in wells of a 48-well (0.5 mL) plate (Nunc) and cultured determine the effects of diet, days in culture, and antigen-specific at 37°C in 5% humidified CO2 for 7 d. Individual wells were har- stimulation. vested at 2, 4 or 5, and 7 d. Data from d 4 and 5 were analyzed Backwards, stepwise regression analysis was used to identify sig- together as a single day. Cells were pelleted by centrifugation at 200 nificant predictors of the percentage of cytokine-producing ϫ g at 4°C for 10 min. The supernatant was frozen for later analysis. CD4ϩKJ1–26ϩ cells in lymph nodes using the categorical variables IL-2, IL-4, IL-10, and IFN-␥ were measured in cell culture superna- “gender” (female ϭ 1, male ϭ 0), “vitamin A deficiency” (control tants by capture ELISA using antibodies, purified standards, and diet ϭ 0, deficient diet ϭ 1) and “experiment” (first experiment ϭ 1, protocols suggested by the manufacturer (Pharmingen). The limit of second experiment ϭ 2), as well as the continuous variable “days detection of the 4 ELISAs was 0.03 ␮g/L. postimmunization.” Statistical analysis. Statistical analysis was performed with the Results of 2- and 3-way ANOVAs are presented as means Ϯ SE. SigmaStat program (Jandel Scientific). A two-tailed P-value of 0.05 Summary statistics and the results of t tests and 1-way ANOVA are was used to determine statistical significance unless otherwise indi- presented in the text and figures as means Ϯ SD. cated. Serum retinol concentrations were compared between diet groups by 2-way ANOVA to control for gender, and then by Stu- dent’s t test because gender was not a significant factor. Body weights RESULTS were measured weekly, and weights for males (M) and females (F) were compared separately between diet groups by 2-way repeated- Confirmation of vitamin A deficiency using serum retinol. measures ANOVA to control for changes in weight over time. ϭ ϩ ϩ Serum retinol was measured at 7 wk of age (n 2 M and 2 The percentage of CD4 KJ1–26 cells in lymph nodes, the F/diet group in each experiment). In both experiments, serum percentage of cells producing cytokines, and the concentration of retinol was higher in control mice (1.15 Ϯ 0.49 and 0.91 cytokines from lymph node cultures were compared by two-way Ϯ 0.37 ␮mol/L) than in mice fed the vitamin A–deficient diet ANOVA with mice categorized by time after immunization and Ϯ ␮ ϭ Ϯ ␮ gender. Gender was a significant factor for predicting the percentage (0.36 0.11 mol/L, P 0.020; 0.30 0.071 mol/L P ϩ ϩ ϭ of CD4 KJ1–26 cells in lymph nodes; thus, diet groups were com- 0.018, respectively). pared separately for males and females by 2-way ANOVA controlling Body weights of vitamin A–deficient and control mice. for time after immunization. Gender was not a significant predictor Body weights were compared by diet for the 45 mice in the 2 for cytokine-positive cells and cytokine concentrations in lymph experiments (including 5 unimmunized controls). In the first node cultures. Thus, these variables were compared between diet experiment (deficient diet; n ϭ 4 M, 6 F; control diet; n ϭ 3 groups using 2-way ANOVA to control for time after immunization. M, 5 F), body weights were compared at 8, 9, 10 and 11 wk of Cytokine concentrations from lymph node cultures were compared age and mice were killed for analysis beginning at 11 wk of age. between diet groups only on days when an antigen-specific response Males did not differ (P ϭ 0.41) on the basis of diet, but there occurred. When the OVA –treated culture had a significantly 323–339 was an interaction of diet with age (P ϭ 0.011) reflecting the higher cytokine concentration (P Ͻ 0.05 by one-tailed t test) than did the unstimulated culture from the same mouse, the response was fact that male mice fed the deficient diet began the study considered antigen specific. slightly heavier than control mice (mean of 24.3 vs. 23.4 g) Data from antigen-stimulated and unstimulated splenocyte cul- but ended the study slightly lighter. Body weights did not differ tures were available from 24 mice (deficient, n ϭ 6 F, 6 M; control in females due to diet (P ϭ 0.68). In the second adoptive diet n ϭ 6 F, 6 M) from the second experiment. Equal numbers of transfer study (deficient diet; n ϭ 6 M, 8 F; control diet; n ϭ 7 VITAMIN A DEFICIENCY AND Th1/Th2 DEVELOPMENT 2663

M, 6 F) body weights were compared at 5, 7, 8, 9, and 10 wk of age. Mice were killed beginning at 10 wk of age for analysis. Males did not differ (P ϭ 0.81) on the basis of diet, but there was an interaction of diet with age (P ϭ 0.013), again reflecting the fact that male mice fed the deficient diet began the study slightly heavier than control mice (mean of 22.2 vs. 20.7 g) but ended the study slightly lighter (24.6 vs. 25.5 g). Body weights did not differ in female mice due to diet (P ϭ 0.062), although the deficient females were slightly heavier than the control mice at all time points. Percentages of CD4؉-KJ1–26؉ donor cells in draining lymph nodes. After adoptive transfer, CD4ϩ-KJ1–26ϩ cells accounted for 3.2 Ϯ 0.8% of total CD4ϩ cells in the lymph nodes of unimmunized mice fed the vitamin A–deficient diet (n ϭ 2 M, 3 F). After immunization, the percentage of CD4ϩ- ϩ Ϯ ϭ KJ1–26 cells decreased from 3.2 1.2% (n 9F,9M) Downloaded from during wk 3 to 1.9 Ϯ 1.0% (n ϭ 6 F, 6 M) during wk 4 and remained stable at 1.6 Ϯ 0.8% during wk 5 (n ϭ 7F,3M)(P ϭ 0.002, wk 3 vs. 5; P ϭ 0.015, wk 3 vs. 4). The percentages did not differ on the basis of diet, but male recipients had higher percentages of CD4ϩ-KJ1–26ϩ cells than females at all time points: 3.5 vs. 2.5% in wk 3, 2.3 vs. 1.4% in wk 4 and 1.9 vs. 1.3% in wk 5 (P ϭ 0.019 for gender comparison by 2-way jn.nutrition.org ANOVA, comparing weeks and gender). Staining with anti- CD62L demonstrated that the mean percentage of CD4ϩ- FIGURE 2 The effect of vitamin A deficiency on the percentage of ϩ antigen-specific CD4ϩKJ1–26ϩ T lymphocytes producing the Th2 cy- KJ1–26 cells with a memory phenotype (CD62Llo) was 73.5 Ϯ ϭ tokines IL-10 (A) and IL-4 (B), and the Th1 cytokines IL-2 (C) and IFN-␥ 5.4% after immunization (n 11 mice). at USDA, National Agricultural Library on December 11, 2007 ؉ ؉ (D). Percentages were determined by intracellular cytokine staining of Intracellular cytokine staining of CD4 -KJ1–26 donor draining lymph node cells treated with PMA and ionomycin nodes after cells from draining lymph nodes. The mean percentages of adoptive transfer of lymphocytes from DO11.10 mice and immunization cytokine-positive cells in unimmunized mice (n ϭ 2M,3F; with OVA323–339 peptide. The numbers of vitamin A–deficient and con- vitamin A–deficient diet) were as follows: IL-4, 0.27 Ϯ 0.40%; trol mice in wk 3, 4 and 5 were 9, 6, and 6, and 9, 6 and 4, respectively. IFN-␥, 0.81 Ϯ 0.32%; IL-10, 0.20 Ϯ 0.30%; and IL-2, 22.9 Diet treatments were significantly different for IL-10 and IFN-␥,as Ϯ 7.4%. indicated by the asterisks (*). Values are means Ϯ SE. Vitamin A deficiency increased the percentage of memory T lymphocytes positive for the Th2 cytokine IL-10 (P ϭ 0.003) (Fig. 2A). The overall percentage of IL-10 positive sponse did not occur for IL-4 but was present for IL-2 on d 2 cells was 100% higher in vitamin A–deficient mice (3.49 and for IFN-␥ and IL-10 on both d 4/5 and 7. Cytokine Ϯ Ϯ 0.41%) than in control mice (1.74 0.37%). This differ- concentrations did not differ between the vitamin A–deficient ence occurred during wk 3 (90% higher), wk 4 (170%), and and control groups (data not shown). wk 5 (70%) after immunization. This positive association of Cytokine concentrations after antigenic stimulation of the percentage IL-10–positive cells with vitamin A deficiency spleen cells. Concentrations of the Th2 cytokines IL-4 and was also present in the multiple regression analysis (Table 1). IL-10 were both higher in antigen-stimulated splenocyte cul- The overall percentage of IL-4–positive cells was quite low tures from vitamin A–deficient mice than in cultures from throughout the experiment, and the diet groups did not differ control mice, whereas concentrations of the Th1 cytokines ϭ (P 0.27; Fig. 2B, Table 1). IL-2 and IFN-␥ were both lower (Fig. 3). Although IL-10 Vitamin A deficiency decreased the percentage of memory concentrations did not differ by diet group when all days were ␥ T lymphocytes positive for the Th1 cytokines IL-2 and IFN- . compared (P ϭ 0.41), within-day comparisons showed that The overall percentage of IL-2–positive cells in the vitamin the IL-10 concentration on d 4/5 was 19% higher in cultures Ϯ A–deficient mice (39.5 3.1%) tended to be lower (16%) from deficient mice than in cultures from control mice (P Ϯ ϭ than in control mice (47.0 2.8%; Fig. 2C)(P 0.081). ϭ 0.043) (Fig. 3A). The IL-4 concentration on d 2 was higher However, a significant, negative association was found be- in cultures from the deficient mice than from the control mice tween vitamin A deficiency and the percentage of IL-2–posi- (P Ͻ 0.001) (Fig. 3B) but subsequent days did not differ. On tive cells when multiple regression analysis was used to control bothd2(P ϭ 0.01) and d 4/5 (P Ͻ 0.001), IL-2 concentra- for variation due to experiment (1st vs. 2nd experiment) and tions were lower in cultures from vitamin A–deficient mice time after immunization (Table 1). The overall percentage of than in those from control mice (Fig. 3C), whereas IFN-␥ ␥ IFN- –positive cells in the vitamin A–deficient mice (8.8 concentrations were lower in cultures from deficient mice than Ϯ Ϯ 0.73%) was 23% lower than in control mice (11.4 0.67%; from control mice (P Ͻ 0.001) on all days (Fig. 3D). P ϭ 0.013) (Fig. 2D). This negative association of vitamin A ␥ deficiency with the percentage of IFN- –positive cells was also DISCUSSION seen by multiple regression analysis (Table 1). Cytokine concentrations after antigenic stimulation of The purpose of this study was to determine whether vita- draining lymph node cells. Draining lymph node cells not minAdeficiency at the time of immunization affects the needed for flow cytometric analysis were used to assess antigen- development of Th1 and Th2 memory cells. The DO11.10 ϭ specific cytokine production by ELISA (n 10 control and 8 adoptive-transfer model was used, in which na¨ıve OVA323–339– deficient mice). Unstimulated control cultures were included specific T lymphocytes are transferred to recipient mice in when cells were available (n ϭ 5). An antigen-specific re- numbers sufficient to allow monitoring of T-lymphocyte de- 2664 STEPHENSEN ET AL.

TABLE 1 Association of vitamin A deﬁciency with the percentage of cytokine-producing CD4ϩ-KJ1-26ϩ T lymphocytes isolated from draining lymph nodes of OVA323–339-immunized mice as assessed by backward, stepwise multiple regression analysis1

IL-4 IFN-␥ IL-2 IL-10 R2 ϭ 0.456 R2 ϭ 0.238 R2 ϭ 0.871 R2 ϭ 0.423

Variable ␤ P ␤ P ␤ P ␤ P

Vitamin A deﬁciency Ϫ0.135 0.034 Ϫ0.179 0.005 0.430 0.001 Experiment 0.369 0.014 0.600 Ͻ0.001 Days postimmunization 0.675 Ͻ0.001 Ϫ0.619 Ͻ0.001 0.518 Ͻ0.001

1 Correlation coefficients (R2), standardized regression coefficients (␤), and P-values (P) are shown for equations predicting the percentage of cytokine-positive cells. Downloaded from velopment, but not at levels high enough to produce an To minimize the effect of antigen-presenting cells and “unusual” abundance of responding cells in lymph nodes drain- bystander cells on the assessment of Th1/Th2 phenotype after ing the site of OVA323–339 immunization (18–20). This model immunization, we chose to stimulate T-cell cytokine produc- allows the direct enumeration of responding Th1 and Th2 tion directly using PMA and ionomycin. This approach was cells using the clonotypic KJ1–26 monoclonal antibody and taken because vitamin A has multiple effects on antigen- intracellular cytokine staining. IFA was used so as not to presenting cell development and function. For example, vita- jn.nutrition.org strongly bias the cytokine response in a Th1 or Th2 direction. minAdeficiency can increase IL-12 production by antigen- IFA induces memory T cells that produce both Th1 and Th2 presenting cells and IFN-␥ production by T cells (10). On the cytokines, although markedly less IFN-␥ than complete other hand, retinoic acid treatment during in vitro develop- Freund’s adjuvant (CFA), which contains mycobacterial cell ment of dendritic cells from peripheral blood monocytes can wall components (21–23). However, some studies suggested also result in the development of IL-12–producing antigen- at USDA, National Agricultural Library on December 11, 2007 that IFA produces a Th2 bias (i.e., inducing Th2 cytokines presenting cells that promote Th1 development (25). Using rather than simply failing to induce IFN-␥). For example, PMA and ionomycin to stimulate cytokine production directly when IFA and CFA were compared using intraperitoneal bypassed these potentially confounding effects on assessment immunization with multiple antigens in different mouse of Th1/Th2 phenotype. Many stimuli could have been used to strains, IFA typically induced IL-5 but little IFN-␥, whereas stimulate T-cell cytokine production directly. We evaluated CFA had the opposite effect (24). Interestingly, the use of the anti-CD3 plus anti-CD28 antibodies in addition to PMA plus OVA323–339 peptide in BALB/c mice produced a balanced ionomycin and chose the latter because anti-CD3/anti-CD28 IL-5/IFN-␥ response in that study. Thus, it is likely that the stimulation interfered with identification of CD4ϩKJ1–26ϩ immunization protocol used in this study did not strongly favor cells by flow cytometry, presumably due to internalization of either a Th1 or Th2 response. the TCR complex after antibody binding. Immunization with the OVA323–339 peptide transformed na¨ıve DO11.10 T lymphocytes into memory cells, as shown by CD62L staining and Th1/Th2 cytokine profiles. We chose to immunize with the peptide not only to provide a potent antigenic stimulus for the adoptively transferred transgenic T lymphocytes, but also to minimize the response by host lymphocytes to other epitopes that might occur had we used intact ovalbumin protein as the antigen. The percentage of CD4ϩKJ1–26ϩ cells expressing Th1/Th2 cytokines from unimmunized mice after adoptive transfer was very low and increased after immunization, as expected with the development of a memory cell phenotype. The percentage of cells positive for the Th1 cytokine IFN-␥ was ϳ10%, although the percentage of IL-2–positive cells was much higher. Higher percentages of IFN-␥–positive cells (20–40% of CD4ϩKJ1– 26ϩ cells) occurred in our laboratory after 2 immunizations using CFA followed by IFA, or using alum, or after in vitro stimulation with IL-12 treatment (60%), but the lower percentage response seen here was reproducible and appears to represent a robust memory response to a single immunization. FIGURE 3 The effect of vitamin A deficiency on the concentration Previous work comparing ovalbumin protein to OVA323–339 of the Th2 cytokines IL-10 (A) and IL-4 (B) and the Th1 cytokines IL-2 (C) immunization in BALB/c mice showed similar T-cell cytokine ␥ and IFN- (D) in supernatants from splenocyte cultures from vitamin recall responses (24), supporting our conclusion that a peptide A–deficient and control mice following adoptive transfer of lymphocytes from DO11.10 mice and immunization with OVA peptide. immunization produced a recall response similar to that in- 323–339 duced by intact protein. The percentage of cells positive for Cells were restimulated with OVA323–339 peptide (OVA) or were treated with a vehicle control (no OVA). Diet treatments were significantly the Th2 cytokine IL-4 was quite low throughout the study. different for all cytokines on the days indicated by asterisks (*). Values Similar percentages were found using 2 immunizations, or after are means Ϯ SE, n ϭ 12 deficient and 12 control. oral immunization with cholera toxin as an adjuvant (data not VITAMIN A DEFICIENCY AND Th1/Th2 DEVELOPMENT 2665 shown), although never as high as that after in vitro stimula- in vivo effects of vitamin A deficiency on Th1/Th2 develop- tion with IL-4 (30%). The percentage of IL-10–positive cells ment. For example, direct addition of retinoic acid to cell increased from wk 3 to 5 after immunization, as might be culture may by-pass the normal regulation of retinoic acid expected for a regulatory cytokine that can dampen T-cell– production that exists in vivo (34), thereby producing a tem- mediated immune responses (26). porary excess in the in vitro studies. Thus, in vitro treatment The mice consuming the vitamin A–deficient diet in this with retinoids may be more representative of a condition of study presented biochemical evidence of deficiency by the vitamin A excess vs. deficiency, rather than adequacy vs. time of immunization, as indicated by low serum retinol levels, deficiency. This question could be addressed by assessing the but they were not so deficient that body weight was affected. effect of high-level dietary vitamin A on Th1/Th2 develop- Although the rate of weight gain in vitamin A–deficient ment using the DO11.10 adoptive-transfer model. males was lower than in males fed the control diet, the When cultures from draining lymph nodes were restimu- potential confounding effect of inanition on immune function lated with peptide antigen, IFN-␥ and IL-2 concentrations did was avoided in this study because significant weight differences not differ, even though the frequencies of IFN-␥–producing were not present at the time of immunization. CD4ϩKJ1–26ϩ cells differed between the diet groups. How- A principal finding of this study was that vitamin A defi- ever, the IL-2 concentration in the deficient cultures at 2 d

ciency doubles the frequency of antigen-specific, IL-10–pro- was 21% lower than in the control cultures. This value is Downloaded from ducing CD4ϩ T cells. The magnitude of the difference be- similar to the difference seen with intracellular cytokine stain- tween the diet groups increased over time and may represent ing; however, given the SD in these ELISA data, a much larger a relative increase in abundance of T cells, Th2 or Treg, sample size (n ϭ 64) would be required to identify a 20% capable of downregulating immune responses (26). This is a difference as significant (assuming power ϭ 0.80 and P-value novel observation and may represent a mechanism by which ϭ 0.05 for Student’s t test). A similarly large sample size would vitamin A deficiency could diminish the immune response to be required to detect a 23% difference in IFN-␥ concentra- infectious or inflammatory diseases. This could include Th2- tions. Cytokines produced by non-CD4ϩKJ1–26ϩ cells would jn.nutrition.org mediated responses, which are diminished by vitamin A defi- also be detected by ELISA, but not by flow cytometry, and may ciency (2), as well as Th1-mediated immune responses, which also be a factor in explaining the different findings with the 2 have received less attention, such as cytotoxic T-lymphocyte methods.

or delayed-type hypersensitivity responses (27–29). Research Splenocyte cultures from vitamin A–deficient mice in the at USDA, National Agricultural Library on December 11, 2007 on regulatory T cells is in its infancy in many regards, and even present study had lower antigen-specific IFN-␥ and IL-2 con- the identification of which T cells are “regulatory T cells,” i.e., centrations and higher IL-4 and IL-10 concentrations than did induced vs. naturally occurring Treg cells, Tr1 cells, or Th3 cultures from control mice. When data from all 3 poststimu- cells, is still a point of contention (26,30,31). Thus, it is not lation days were considered, the IFN-␥ and IL-2 concentra- clear whether the IL-10—positive cells identified in this study tions in cultures from the deficient mice were 41 and 60% of fall into this category of regulatory T cells. Clearly, some the levels in the control cultures, whereas the concentration of CD4ϩ T-cell types with regulatory activity act by producing IL-4 was 181% of control levels (using data from d 2). These IL-10 that downregulates immune responses. Thus, it will be of deficits for IFN-␥ and IL-2 are greater than the 20% lower interest to determine whether the IL-10–positive cells iden- frequency of IFN-␥– and IL-2–positive CD4ϩKJ1–26ϩ cells in tified here do indeed play a regulatory role. lymph nodes from deficient mice, but the direction of the How vitamin A deficiency affects the development of IL- difference is the same. It is possible that the differences in 10–producing T cells is not clear. The development of Treg splenocyte cultures reflect even greater differences in frequen- cells, which produce IL-10 and little IL-2 (as seen in the cies of cytokine-producing CD4ϩKJ1–26ϩ cells than in drain- present study), is facilitated by an early exposure to IL-10. ing lymph nodes. Higher IL-10 concentrations occurred in However, previous work demonstrated lower antigen-specific splenocyte cultures from deficient mice on 1 of3dinculture, IL-10 production in immunized, vitamin A–deficient mice which is consistent with the higher frequency of IL-10–pro- (10). In addition, work from this laboratory showed no effect ducing CD4ϩKJ1–26ϩ cells in lymph nodes from deficient of retinoic acid treatment on IL-10 production during devel- mice, although the magnitude of the difference is smaller. opment of DO11.10 memory T cells cultured in vitro (16). Earlier studies showed that vitamin A deficiency increases Further characterization of these cells is required to clarify antigen-specific production of IFN-␥ in antigen-restimulated their role in immune suppression caused by vitamin A defi- lymphocyte cultures on an absolute and per cell basis (10,35). ciency. The higher rate of IFN-␥ was diminished by the addition of Vitamin A–deficient mice had 23% fewer IFN-␥—positive exogenous vitamin A. In contrast, in the present study, vita- and 16% fewer IL-2–positive cells in draining lymph nodes minAdeficiency decreased the production of IFN-␥ in restim- than did mice consuming the control diet with adequate ulated splenocyte cultures. One possible explanation for this vitamin A. These differences indicate that vitamin A defi- phenomenon is that the use of 10% FBS in the culture media ciency decreased development of Th1 memory cells, although in the present study produced a less Th1-biased environment the magnitude of the decrease was not large. Previous work in vitro than in previous studies using serum-free medium, due using an Elispot method to enumerate IFN-␥—positive cells in to the presence of vitamin A or other nutrients or hormones vitamin A–deficient and control mice after immunization in FBS that support lymphocyte and antigen-presenting cell showed no difference between the diet groups, but the vari- function. Differences between the DO11.10 adoptive transfer ability of the Elispot method (based on reported SE from that system and previous methods may also be involved. For exam- study) was such that a 20% difference would not have been ple, previous studies immunized with intact proteins or infec- detected (10). In vitro studies from our laboratory (16) with tious agents, whereas the present study used a peptide. Because na¨ıve DO11.10 T cells indicated increased Th2 development intact proteins require processing by antigen-presenting cells with 9-cis but not all-trans retinoic acid, although other groups before presentation by major histocompatibility complex mol- reported increased Th2 development with all-trans retinoic ecules, and these processes are affected by vitamin A (36), acid using similar methods (32,33). The present findings sug- initial antigen presentation to na¨ıve T lymphocytes could gest that in vitro methods may not adequately recapitulate the have been less affected by vitamin A deficiency in the present 2666 STEPHENSEN ET AL. study than in previous studies. In addition, the present study 14. Marrack, P., Shimonkevitz, R., Hannum, C., Haskins, K. & Kappler, J. (1983) The major histocompatibility complex-restricted antigen receptor on T used IFA as the adjuvant, whereas previous studies have used cells. IV. An anti-idiotypic antibody predicts both antigen and I-specificity. J. Exp. adjuvants such as CFA, which has greater Th1-inducing ac- Med. 158: 1635–1646. tivity (23). Use of such an adjuvant may interact with vitamin 15. Stephensen, C. B., Blount, S. R., Schoeb, T. R. & Park, J. Y. (1993) Adeficiency to affect Th1/Th2 development. Vitamin A deficiency impairs some aspects of the host response to influenza A virus infection in BALB/c mice. J. Nutr. 123: 823–833. In summary, our data demonstrate that vitamin A defi- 16. Stephensen, C. B., Rasooly, R., Jiang, X., Ceddia, M. A., Weaver, C. T., ciency at the time of initial antigen exposure significantly Chandraratna, R. A. & Bucy, R. P. (2002) Vitamin A enhances in vitro Th2 enhanced the development of IL-10–producing Th2 or Treg development via retinoid X receptor pathway. J. Immunol. 168: 4495–4503. cells while diminishing the development of Th1 memory cells. 17. Coligan, J. E., Kruisbeek, A. M., Margulies, D. H., Shevach, E. M. & Strober, W., eds. (2001) Current Protocols in Immunology. John Wiley and This observation adds to our knowledge of how vitamin A Sons, New York, NY. deficiency affects Th1/Th2-mediated immune responses and 18. Kearney, E. R., Pape, K. A., Loh, D. Y. & Jenkins, M. K. (1994) Visu- suggests a novel mechanism by which vitamin A deficiency alization of peptide-specific T cell immunity and peripheral tolerance induction in vivo. Immunity 1: 327–339. may diminish such responses. Such decreases in immune func- 19. Pape, K. A., Kearney, E. R., Khoruts, A., Mondino, A., Merica, R., Chen, tion may contribute to the increased risk of death from infec- Z. M., Ingulli, E., White, J., Johnson, J. G. & Jenkins, M. K. (1997) Use of tious diseases that is a hallmark of human vitamin A defi- adoptive transfer of T-cell-antigen-receptor-transgenic T cell for the study of T-cell activation in vivo. Immunol. Rev. 156: 67–78. ciency. Downloaded from 20. Rogers, W. O., Weaver, C. T., Kraus, L. A., Li, J., Li, L. & Bucy, R. P. (1997) Visualization of antigen-specific T cell activation and cytokine expression ACKNOWLEDGMENTS in vivo. J. Immunol. 158: 649–657. 21. Heeger, P. S., Forsthuber, T., Shive, C., Biekert, E., Genain, C., Hofstetter, The authors thank Daniel Hwang, Darshan Kelley, and Susan H. H., Karulin, A. & Lehmann, P. V. (2000) Revisiting tolerance induced by Zunino for reading the manuscript and making helpful suggestions, autoantigen in incomplete Freund’s adjuvant. J. Immunol. 164: 5771–5781. Alina Wettstein for performing retinol analysis, and Pat Bucy for help 22. Shibaki, A. & Katz, S. I. (2002) Induction of skewed Th1/Th2 T-cell differentiation via subcutaneous immunization with Freund’s adjuvant. Exp. Der- with setting up the adoptive transfer model. jn.nutrition.org matol. 11: 126–134. 23. Billiau, A. & Matthys, P. (2001) Modes of action of Freund’s adjuvants LITERATURE CITED in experimental models of autoimmune diseases. J. Leukoc. Biol. 70: 849–860. 24. Yip, H. C., Karulin, A. Y., Tary-Lehmann, M., Hesse, M. D., Radeke, H., 1. Fawzi, W. W., Chalmers, T. C., Herrera, M. G., & Mosteller, F. (1993) Heeger, P. S., Trezza, R. P., Heinzel, F. P., Forsthuber, T. & Lehmann, P. V. Vitamin A supplementation and child mortality. A meta-analysis. J. Am. Med. (1999) Adjuvant-guided type-1 and type-2 immunity: infectious/noninfectious Assoc. 269: 898–903. dichotomy defines the class of response. J. Immunol. 162: 3942–3949. at USDA, National Agricultural Library on December 11, 2007 2. Stephensen, C. B. (2001) Vitamin A, infection, and immune function. 25. Mohty, M., Morbelli, S., Isnardon, D., Sainty, D., Arnoulet, C., Gaugler, B. Annu. Rev. Nutr. 21: 167–192. & Olive, D. (2003) All-trans retinoic acid skews monocyte differentiation into 3. Villamor, E. & Fawzi, W. W. (2000) Vitamin A supplementation: impli- interleukin-12-secreting dendritic-like cells. Br. J. Haematol. 122: 829–836. cations for morbidity and mortality in children. J. Infect. Dis. 182 (suppl. 1): 26. Groux, H. (2003) Type 1 T-regulatory cells: their role in the control of S122–S133. immune responses. Transplantation 75: 8S–12S. 4. Stephensen, C. B., Franchi, L. M., Hernandez, H., Campos, M., Gilman, 27. Ahmed, F., Jones, D. B. & Jackson, A. A. (1991) Effect of vitamin A R. H. & Alvarez, J. O. (1998) Adverse effects of high-dose vitamin A supple- deficiency on the immune response to epizootic diarrhoea of infant mice (EDIM) ments in children hospitalized with pneumonia. Pediatrics 101: E3. rotavirus infection in mice. Br. J. Nutr. 65: 475–485. 5. Fawzi, W. W., Msamanga, G. I., Hunter, D., Renjifo, B., Antelman, G., 28. Sijtsma, S. R., Rombout, J. H., West, C. E. & van der Zijpp, A. J. (1990) Bang, H., Manji, K., Kapiga, S., Mwakagile, D., Essex, M. & Spiegelman, D. Vitamin A deficiency impairs cytotoxic T lymphocyte activity in Newcastle disease (2002) Randomized trial of vitamin supplements in relation to transmission of virus-infected chickens. Vet. Immunol. Immunopathol. 26: 191–201. HIV-1 through breastfeeding and early child mortality. AIDS 16: 1935–1944. 29. Wiedermann, U., Hanson, L. A., Kahu, H. & Dahlgren, U. I. (1993) 6. Smith, S. M. & Hayes, C. E. (1987) Contrasting impairments in IgM and Aberrant T-cell function in vitro and impaired T-cell dependent antibody response IgG responses of vitamin A-deficient mice. Proc. Natl. Acad. Sci. U.S.A. 84: in vivo in vitamin A-deficient rats. Immunology 80: 581–586. 5878–5882. 30. O’Garra, A. & Barrat, F. J. (2003) In vitro generation of IL-10-producing 7. Pasatiempo, A. M., Kinoshita, M., Taylor, C. E. & Ross, A. C. (1990) regulatory CD4ϩ T cells is induced by immunosuppressive drugs and inhibited by Antibody production in vitamin A-depleted rats is impaired after immunization Th1- and Th2-inducing cytokines. Immunol. Lett. 85: 135–139. with bacterial polysaccharide or protein antigens. FASEB J. 4: 2518–2527. 31. Mittrucker, H. W. & Kaufmann, S. H. (2004) Mini-review: regulatory T 8. Semba, R. D., Muhilal, Scott, A. L., Natadisastra, G., Wirasasmita, S., Mele, L., Ridwan, E., West, K. P., Jr. & Sommer, A. (1992) Depressed immune cells and infection: suppression revisited. Eur. J. Immunol. 34: 306–312. response to tetanus in children with vitamin A deficiency. J. Nutr. 122:101–107. 32. Iwata, M., Eshima, Y. & Kagechika, H. (2003) Retinoic acids exert 9. Cantorna, M. T., Nashold, F. E., Chun, T. Y. & Hayes, C. E. (1996) direct effects on T cells to suppress Th1 development and enhance Th2 devel- Vitamin A down-regulation of IFN-gamma synthesis in cloned mouse Th1 lym- opment via retinoic acid receptors. Int. Immunol. 15: 1017–1025. phocytes depends on the CD28 costimulatory pathway. J. Immunol. 156: 2674– 33. Hoag, K. A., Nashold, F. E., Goverman, J. & Hayes, C. E. (2002) 2679. Retinoic acid enhances the T helper 2 cell development that is essential for robust 10. Cantorna, M. T., Nashold, F. E. & Hayes, C. E. (1994) In vitamin A antibody responses through its action on antigen-presenting cells. J. Nutr. 132: deficiency multiple mechanisms establish a regulatory T helper cell imbalance 3736–3739. with excess Th1 and insufficient Th2 function. J. Immunol. 152: 1515–1522. 34. Duester, G., Mic, F. A. & Molotkov, A. (2003) Cytosolic retinoid dehy- 11. Chun, T. Y., Carman, J. A. & Hayes, C. E. (1992) Retinoid repletion of drogenases govern ubiquitous metabolism of retinol to retinaldehyde followed by vitamin A-deficient mice restores IgG responses. J. Nutr. 122: 1062–1069. tissue-specific metabolism to retinoic acid. Chem.-Biol. Interact. 143–144: 201– 12. Carman, J. A., Smith, S. M. & Hayes, C. E. (1989) Characterization of 210. a helper T lymphocyte defect in vitamin A-deficient mice. J. Immunol. 142: 35. Carman, J. A. & Hayes, C. E. (1991) Abnormal regulation of IFN- 388–393. gamma secretion in vitamin A deficiency. J. Immunol. 147: 1247–1252. 13. Murphy, K. M., Heimberger, A. B. & Loh, D. Y. (1990) Induction by 36. Geissmann, F., Revy, P., Brousse, N., Lepelletier, Y., Folli, C., Durandy, antigen of intrathymic apoptosis of CD4ϩCD8ϩTCRlo thymocytes in vivo. Sci- A., Chambon, P. & Dy, M. (2003) Retinoids regulate survival and antigen ence 250: 1720–1723. presentation by immature dendritic cells. J. Exp. Med. 198: 623–634.