BATF is required for normal expression of gut-homing receptors by T helper cells in response to retinoic acid

BATF KO mice generated by targeted deletion of either exons one and two or exon three of the Batf gene have been previously described to have relatively normal numbers of T cells in secondary lymphoid tissues (Schraml et al., 2009; Betz et al., 2010). When we examined the intestine by immunohistochemistry, CD4⁺ T cells were numerically deficient in the LP compartment of the small intestinal villi (Fig. 1 A). CD8⁺ T cells were present but also decreased in numbers in the small intestinal villi of the BATF KO mice. Flow cytometry analysis revealed that the frequency of CD4⁺ T cells was decreased significantly in the colon and the small intestinal LP (Fig. 1 B). CD4⁺ T cells were decreased also in PPs but not in the mesenteric LN (MLN). When FoxP3⁺ CD4⁺ T cells were examined, they were more decreased in the colon and the small intestine than in MLN or PPs (Fig. 1, B and C). CD8⁺ T cells were significantly decreased in PPs and the small intestine (Fig. 1, A–C).

In an effort to provide an explanation for the deficiency of CD4⁺ T cells in the intestine, we examined the expression of CCR9 and α4β7 (Fig. 2 A). Both FoxP3⁺ and FoxP3⁻ CD4⁺ T cells in the small intestine of BATF KO mice expressed CCR9 at reduced levels compared with WT mice (Fig. 2, A and B). Numbers of CD4⁺ FoxP3⁺ T cells positive for α4β7 were reduced in the small intestinal LP of BATF KO mice (Fig. 2, A and B). A considerable decrease in CCR9 and α4β7 expression was also detected in CD8⁺ T cells in the small intestine but not in other organs (Fig. 2 C).

Expression of CCR9 and α4β7 by T cells is induced by RA (Iwata et al., 2004). We next examined whether BATF KO naive T cells are defective in up-regulating trafficking receptors in response to T cell activation in the presence of RA in vitro. Compared with WT CD4⁺ T cells, BATF KO naive T cells showed less up-regulation of CCR9 in response to RA (Fig. 3 A). Similarly, BATF KO naive T cells were unable to up-regulate α4β7 to WT levels. Similar results were obtained with concanavalin A (Fig. 3 A) or with OVA_323–339 when OT-II BATF KO T cells were used (not depicted). At the mRNA level, expression of both CCR9 and Itg-α4 in response to RA was defective in BATF KO T cells (Fig. 3 B). However, expression of Itg-β7 mRNA by BATF KO T cells was comparable with WT T cells. We examined other chemokine receptors such as CCR4, CCR6, CCR7, and CXCR4 (Fig. 3 C) and did not detect any significant differences in expression of these receptors between WT and BATF KO T cells cultured with or without RA. RA enhances the induction of FoxP3⁺ T cells by TGFβ1 (Kang et al., 2007; Mucida et al., 2007), and this response was intact in BATF KO T cells (Fig. 3 D). However, BATF KO T regulatory cells (iT_reg cells) induced with RA and TGFβ1 were deficient in expression of CCR9 and α4β7 (Fig. 3 E). These results indicate that expression of gut-homing receptors in BATF KO T helper cells to RA is defective, whereas differentiation of T_reg cells was normal.

It is possible that BATF KO T cells are defective in expression of CCR9 and α4β7 as the result of a block in cell development rather than the absence of functional BATF. This was addressed using retroviral gene complementation in BATF KO T cells. We constructed retroviral vectors expressing the full-length and two truncated versions of BATF and used these to complement the BATF deficiency in T cells (Fig. 4 A). Enforced BATF expression induced a low level of α4β7 but did not induce CCR9 in the absence of exogenous RA (Fig. 4, B and C). This modest induction of α4β7 is likely caused by residual RA that is present in the culture medium. BATF KO CD4⁺ T cells were able to normally express CCR9 and α4β7 in response to RA when the full-length BATF protein was expressed (Fig. 4, B and C). A low but detectable activity was observed for a BATF mutant deleted for the first exon (BATFΔ2), and this was predicted because the protein retains the Jun dimerization and DNA-binding domains of BATF (Fig. 4, B and C). In contrast, a nonfunctional BATF protein deleted for the first and second exons (BATFΔ1) did not have such activities. These results demonstrate that BATF KO CD4⁺ T cells are not developmentally defective but rather are functionally defective in expression of the gut-homing receptors.

To gain insights into the defective expression of CCR9 and α4β7 in BATF deficiency, the expression of RARα, RARβ, and RARγ was examined and was found to be normal in BATF KO T cells cultured in the presence and absence of RA (Fig. 5 A). We examined the sequence structure of the regulatory region of the CCR9 and Itgα4 genes. We found several RARα and BATF binding sites in the 5′ upstream regions of the mouse CCR9 and Itgα4 genes (Fig. S1 and Table S1). Chromatin immunoprecipitation (ChIP) followed by genomic DNA PCR on WT T cells activated in the presence of RA revealed the binding of RARα and BATF to 18 regions in the 44-kbp 5′ upstream region of the CCR9 gene as well as to the 5′ upstream region of the Itgα4 gene (Fig. 5, B and C). The binding of RARα to the CCR9 and Itgα4 genes was abnormally low in BATF KO T cells (Fig. 5, B and C). Moreover, the level of acetylated histone 4 protein was very low in the regulatory regions of CCR9 and Itgα4 genes. In line with this, the promoter activity of a 1.3-kbp region spanning the 5′ upstream regulatory region of Itg-α4 gene was significantly lower in BATF KO than WT T cells (Fig. 5 D).

Nuclear hormone receptor function involves a series of enzymatic modifications of proteins including the acetylation of histone proteins for gene expression (Rosenfeld et al., 2006). It has been reported that expression of Sirt1, a histone deacetylase (HDAC) class III enzyme, is induced in BATF deficiency to decrease acetylation of histones in CD8⁺ T cells (Kuroda et al., 2011). To determine the role of histone acetylation in expression of CCR9 and α4β7 in BATF deficiency, we used inhibitors of HDACs (trichostatin A and BML-210 for class I/II; Ex-527 for Sirt1). Increased acetylation through class I and II deacetylase inhibitors restored the expression of the gut-homing receptors (Fig. 5 E). However, inhibition of Sirt1 was not effective.

Defective expression of CCR9 and α4β7 is expected to negatively influence the homing of BATF KO T cells to the gut. We determined the short-term homing ability of BATF KO T cells in vivo. Naive CD4⁺ T cells from WT and BATF KO mice were activated in the presence of RA for 6 d and examined for chemotaxis to the CCR9 ligand CCL25 in vitro (Fig. 6 A). Chemotaxis by RA-treated BATF KO CD4⁺ T cells was inefficient compared with WT T cells (Fig. 6 A), which is explained by the CCR9 deficiency. When introduced via a tail vein injection and examined 20 h later by flow cytometry, RA-treated BATF KO CD4⁺ T cells were poorly represented in the small intestine compared with WT T cells (Fig. 6 B). The flow cytometry technique is useful for assessing cell migration into an organ but cannot accurately evaluate cell migration into specific microenvironments within an organ. To examine cell migration to specific tissue sites, frozen sections of the tissues were examined by confocal microscopy (Fig. 6 C). We found that migration of CD4⁺ T cells into the small and large intestinal LP and PPs was defective. Migration to MLN was decreased by 30% compared with WT T cells. One disadvantage of confocal imaging on tissue sections is that the areas viewed are limited by sectioning, and it is often difficult to locate tissue sites such as small lymphoid patches in the intestine. To overcome this issue, we used multiphoton microscopy to image whole-mounted tissues (Fig. 6 D). Using this technique, we confirmed that the migration of BATF KO T cells into the small intestine and colon was severely compromised. Additionally, we found that migration of RA-conditioned BATF KO T cells to both colonic patches and PPs was defective as well. Three-dimensional (3D) images of the differential migration into the indicated whole-mounted tissues are shown in Videos 1–3.

Oral immunization increases antigen-specific T cells, particularly FoxP3⁺ T cells, in the intestine. We examined whether antigen-specific (OT-II) BATF KO T cells can populate the small intestine after mucosal immunization with OVA. OVA-specific BATF KO FoxP3⁺ CD4⁺ T cells failed to populate the small intestine (Fig. 7, A and B). However, there was no significant difference in the numbers of these cells populating other organs such as PPs, MLN, spleen, and peripheral LN (PLN). OVA-specific BATF KO FoxP3⁻ non-T_reg cells also failed to populate the small intestine after immunization, whereas their population in other organs except PPs were not affected significantly (Fig. 7, A and B).

BATF KO CD4⁺ T cells may not effectively induce inflammation in the intestine because of their poor migration to the tissue. We tested this possibility using the colitis model in Rag1^−/− mice. Rag1^−/− mice were injected with T_reg cell–depleted naive CD4⁺ T cells isolated from WT or BATF KO mice. Naive BATF KO CD4⁺ T cells did not decrease the weight of Rag1^−/− mice as effectively as WT T cells (Fig. 8 A). Histological examination revealed only mild inflammation in the Rag1^−/− mice injected with naive BATF KO CD4⁺ T cells (Fig. 8 B). This is in contrast to the severe inflammation seen in both the proximal and distal colon of the Rag1^−/− mice injected with naive WT T cells. The numbers of Th1 cells, which are the major effector T cells to induce inflammation in this model, were decreased in the colon of Rag1^−/− mice injected with naive BATF KO CD4⁺ T cells (Fig. 8 C). The specific decrease of Th1 cells in the colon indicates that BATF is required for Th1 effector T cell population in the inflamed gut tissue. Also decreased were the numbers of Th17 and T_reg cells in the colon and the small intestine (Fig. 8 C). This decrease in Th17 cell numbers in all tissues, including the spleen, is in line with the previous reports demonstrating that BATF KO T cells are impaired in their ability to differentiate into Th17 cells (Schraml et al., 2009; Betz et al., 2010). Overall, these results demonstrate the defective population and inflammatory function of BATF KO T cells in the intestine.

The defective effector function of BATF KO CD4⁺ T cells may also be caused by problems in cell activation, death, or proliferation. In this regard, a lymphoproliferative disorder caused by increased T cell survival has been observed in aged mice overexpressing BATF (Logan et al., 2012). We observed that BATF KO CD4⁺ T cells were comparable with WT T cells in loss of CD62L and up-regulation of CD69 in response to anti-CD3/28 + IL-2 (Fig. 9 A). Moreover, BATF KO CD4⁺ T cells did not display significant differences in cell death after T cell activation (Fig. 9 B). Expansion of OT-II CD4⁺ T cells after parenteral immunization was also largely intact (Fig. 9 C). Similarly, homeostatic CD4⁺ T cell expansion in lymphopenic Rag1^−/− mice was slightly decreased but was not defective (Fig. 9 D). Absolute numbers of BATF KO CD4⁺ T cells in the spleen and PLN were comparable with WT, but the numbers in the intestinal LP and the intestine-draining LN (MLN) were decreased as expected (Fig. 9 E). These results indicate that general cell processes such as cell activation, death, and proliferation appear largely intact in BATF KO T cells.

As shown in Fig. 3 E, iT_reg cells derived from naive BATF KO CD4⁺ T cells were deficient in the expression of CCR9 and α4β7. We examined whether this deficiency would affect the in vivo suppressive function of BATF KO iT_reg cells. RA-treated BATF KO iT_reg cells were as effective as WT iT_reg cells in suppressing responder T cells in vitro (Fig. 10 A). Thus, function of the BATF KO iT_reg cells is intact in vitro. We next compared the suppressive functions of CD45.2⁺ iT_reg cells generated from naive WT and BATF KO CD4⁺ T cells in vivo in the T cell–induced colitis model. RA-treated BATF KO iT_reg cells were not as effective as WT T_reg cells in suppressing the weight loss induced by the inflammatory activity of WT T cells (Fig. 10 B). Histological examination revealed that only WT iT_reg cells, but not BATF KO iT_reg cells, effectively suppressed colitis (Fig. 10 C). BATF KO iT_reg cells were also less effective than WT iT_reg cells in decreasing the number of CD45.1⁺ (WT naive T cell derived) Th1 effector cells in the MLN and colon (Fig. 10 D). Th17 cells, which are generally suppressed by Th1 cells, were reciprocally increased with injection of WT iT_reg cells but not with BATF KO iT_reg cells. Fewer CD45.2⁺ BATF KO iT_reg cells than WT iT_reg cells were found in the colon but not in the MLN of the Rag1^−/− mice (Fig. 10 E), indicating defective population of the transferred BATF KO iT_reg cells in the colon but not in the MLN in the inflammatory condition. Overall, these results demonstrate that BATF is required for the in vivo population of T_reg cells in the gut and their regulatory activity on the Th1 cell activity and tissue inflammation.

Our study identified BATF as a cellular factor required for the optimal expression of CCR9 and α4β7 by T cells, particularly CD4⁺ T cells, in response to RA. We provided evidence that naive CD4⁺ T cells require BATF to up-regulate the gut-homing receptors during T cell activation and differentiation. BATF is important for the expression of gut-homing receptors by both effector cells and FoxP3⁺ T_reg cells. Moreover, this role of BATF in regulating gut-homing receptor expression is important not only for effector T cells in inducing intestinal inflammation but also for T_reg cells in controlling inflammation. The findings implicate BATF as a key regulator of T cell–mediated immune responses and inflammation in the intestine.

Our results indicate that BATF deficiency makes CD4⁺ T cells less responsive to the RA signal in up-regulating the gut-homing receptors. RA, produced by gut epithelial cells and dendritic cells, is the major inducer of CCR9 and α4β7. In this process, RARα plays an important role as the receptor for RA and works together with the TCR activation signal to up-regulate CCR9 (Iwata et al., 2004). RA also plays an important role in up-regulating the Itg-α4 subunit of the α4β7 integrin (Kang et al., 2011). The Itg-β7 subunit of α4β7 is expressed constitutively and increased further by TGFβ1 (Kang et al., 2011). Our results indicate that expression of Itg-α4, but not Itg-β7, is defective at the mRNA level in BATF deficiency.

Defective expression of the gut-homing receptors by BATF KO CD4⁺ T cells can result in inefficient migration of these T cells into the intestine. The CCR9 ligand, CCL25 (also called TECK), is highly expressed in the epithelial cells of the small intestine (Kunkel et al., 2000). CCR9 deficiency leads to defective homing of memory T cells into the small intestinal LP (Stenstad et al., 2006). α4β7 is required for migration into the whole intestine because its adhesion receptor, MAdCAM-1, is expressed on intestinal endothelial cells (Sigmundsdottir and Butcher, 2008). MAdCAM-1 is expressed also on the endothelial cells in PPs (Streeter et al., 1988), and therefore, BATF KO T cells with low α4β7 expression are inefficient for migration into PPs. Additionally, migration of BATF KO T cells to colonic patches was defective, indicating that BATF expression is important for T cell migration into the gut-associated lymphoid tissues and nonlymphoid tissue areas.

Oral or intragastric immunization induces immune tolerance in the intestine in part by generating gut-homing T_reg cells in the intestine (Thorstenson and Khoruts, 2001; Nagatani et al., 2004). Gut-homing receptors such as CCR9 and α4β7 are required for the proper induction of immune tolerance in the intestine, probably because of the need for efficient migration of T_reg cells into the intestine in the process (Cassani et al., 2011). Our results revealed that BATF deficiency leads to defective population of T_reg cells in the small intestine after oral immunization. Also defective was the population of non-T_reg cells in the intestine. Although the short-term homing of T cells into PPs is somewhat defective, oral immunization still normally induced population of T cells in PPs and other lymphoid tissues such as spleen and PLN. This is perhaps because T cell migration to secondary lymphoid tissues involves trafficking receptors that are not regulated by BATF. In this regard, it has been shown that T cell migration into PPs can be mediated also by an α4β7-independent mechanism mediated by L-selectin (Bargatze et al., 1995; Warnock et al., 2000).

Our results indicate that BATF deficiency has a significant functional consequence on the induction and regulation of tissue inflammation in the intestine. BATF KO T cells fail to induce colitis in Rag1^−/− mice. Also, BATF KO FoxP3⁺ T cells fail to suppress colitis induced by inflammatory T cells. In an inflammatory condition, BATF KO effector Th1 cells and FoxP3⁺ T cells failed to populate the colon. These results suggest that BATF is important for the generation of functionally competent effector and suppressor T cells with a normal gut migratory capacity. Although this study focuses on the migration aspect, BATF is known to have additional functions in regulation of T cell function. Although BATF is not required to generate Th1 cells and FoxP3⁺ T cells, BATF is required to generate Th17 cells and T-Fh cells. It has been reported that BATF is required for generation of effector CD8⁺ T cells (Kuroda et al., 2011) and for a check point in the self-renewal of hematopoietic stem cells in response to DNA damage (Wang et al., 2012). Thus, other phenotypes of BATF KO mice in addition to the migration defect could contribute to the ineffective function of effector and suppressor T cells in control of tissue inflammation in the intestine. Nevertheless, we did confirm that the basic activation, survival, and expansion of BATF KO T cells in response to antigens and in a lymphopenic condition are not impaired.

The result of our retroviral gene transfer study demonstrated that BATF expression allows BATF KO T cells to regain sensitivity to the RA signal and up-regulate the two homing receptors. How BATF regulates gene expression and cell function is an active area of research. The in vitro analysis of BATF function indicates a role for the protein in the suppression of AP-1 transcriptional activity (Echlin et al., 2000). BATF inhibits expression of the nicotinamide adenine dinucleotide (NAD⁺)–dependent deacetylase Sirt1, which in turn results in increased T-bet expression (Kuroda et al., 2011). In contrast, BATF has been shown to be associated with transcriptional activation of several genes in Th17 cells and with the induction of AID (Schraml et al., 2009; Ise et al., 2011). BATF and IRF-4 can cooperate to induce the IL10, IL-17a, and IL-21 genes in T cells and to regulate the development of CD8α⁺ dendritic cells (Glasmacher et al., 2012; Li et al., 2012; Tussiwand et al., 2012). Several BATF- and RARα-binding sites are present in the 5′ regulatory regions of the CCR9 and Itg-α4 genes. With one exception, all of these binding sites on the CCR9 gene and all of the BATF-binding sites on the Itg-α4 gene do not possess the features of AP-1–IRF composite elements, suggesting that the expression of the gut-trafficking receptors would likely involve a different regulatory mechanism. Histone modification by acetylation is critical for chromatin remodeling for gene expression in response to nuclear hormone receptor ligands (Rosenfeld et al., 2006). Our results indicate that RAR binding and histone acetylation at the CCR9 and Itgα4 genes are important for expression of the gut-homing receptors and are regulated by BATF. The detailed interaction underlying the regulation of the trafficking receptor genes will require further study as it may involve cross talk between BATF and the RAR expression machinery and the activities of several coactivators and repressors. In conclusion, this study identified a novel function for BATF in up-regulating the major gut-homing receptors on T helper cells.

Batf ΔZ/ΔZ mice with the third exon deleted in the Batf gene have been described previously (Betz et al., 2010) and will be referred to as BATF KO mice in this article. C57BL/6 mice breeding pairs were purchased from Harlan. BATF KO mice were backcrossed on the C57BL/6 background for >10 generations. Congenic CD45.1 C57BL/6, OT-II (C57BL/6-Tg(TcraTcrb)425Cbn/J), and Rag1^−/− mice (B6.129S7-Rag1tm1Mom/J) were obtained from the Jackson Laboratory. OT-II mice were crossed to BATF KO mice to generate OT-II BATF KO mice. All mice used in this study were maintained at Purdue University, and experiments were performed according to approved protocols by the Purdue University Animal Care and Use Committee. Generally, the mice were used between 6 and 8 wk of age at the start of each experiment.

Single cell suspensions of LNs (mesenteric, inguinal, auxiliary, and brachial) and spleens were prepared by grinding tissues through an iron mesh. CD4⁺ T cells were isolated by a CD4⁺ T cell isolation kit (Miltenyi Biotec). The cells were further processed for isolation of T_reg cell–depleted CD4⁺CD25⁻CD44⁻CD69⁻ naive T cells with an AutoMACS separator (Miltenyi Biotec) as previously described (Kang et al., 2011). The cells were cultured for 5–6 d in complete RPMI 1640 medium (10% FBS) supplemented with 100 U/ml hIL-2. When indicated, 1 ng/ml hTGF-β1 and/or 10 nM RA (all-trans RA) was added. As T cell activators, 2.5 µg/ml concanavalin A or OVA_323–339 peptide (at 1 µg/ml with 3× irradiated splenocytes for OT-II cells) was used. All cytokines were obtained from R&D Systems or PeproTech. When indicated, HDAC inhibitors such as 10 µM EX-527 (Sigma-Aldrich), 5 nM trichostatin (Enzo Life Sciences), and 10 µM BML-210 (Sigma-Aldrich) were added to the T cell culture.

Expression of chemokine receptors and integrins was examined as previously described (Kang et al., 2011). Single cell suspensions, isolated from various organs of mice, were stained with antibodies to CCR9 (clone 242503) and α4β7 (clone DATK32). For cells stained with unconjugated antibodies, biotin-labeled secondary antibodies and tertiary PE/PerCP/APC-streptavidin (BD) were used along with antibodies to CD44 and CD4. When necessary, the cells were further stained with antibodies to FoxP3 (FJK-16s; eBioscience) according to the manufacturer’s protocol.

The chemotaxis assay was performed and analyzed as described previously (Wang et al., 2010). In brief, 5 × 10⁵ T cells, activated in the presence or absence of 10 nM RA for 5 d, were added to the upper chamber of Transwell inserts (Corning) and allowed to migrate to the lower chamber containing 3 µg/ml murine CCL25 (R&D Systems) for 3 h. The cells in the lower chambers were collected, stained, and quantitated by flow cytometry.

The full-length and truncated coding regions of the mouse BATF cDNA were amplified by PCR using the primers described in Table S2. Amplified DNA fragments were cloned into a retroviral vector with GFP expression to help identify transduced cells (Wang et al., 2009) and sequenced for accuracy. A retrovirus packaging cell line (Phoenix-ECO) was transfected with the vectors, and activated T cells were infected with culture supernatant containing the recombinant virus. The T cells were previously activated for 36 h with IL-2 and concanavalin A in the presence or absence of 10 nM RA. The virus-infected T cells were harvested 4 d later, and GFP⁺ T cells were examined for CCR9 and α4β7 expression.

The VISTA bioinformatics tool was used to determine conserved noncoding sequences. ChIP assays were performed as described previously (Kang et al., 2011). CD4⁺ T cells (5 × 10⁶ naive per sample), cultured in the presence of 100 U/ml IL-2, 1.5 µg/ml concanavalin A, and 10 nM RA for 4–5 d, were used for ChIP assays. DNA protein complexes were immunoprecipitated using 4 µg anti-BATF (rabbit polyclonal; Cell Signaling Technology), anti-RARα (rabbit polyclonal; Santa Cruz Biotechnology, Inc.), and anti–acetyl-histone H4 antibodies (EMD Millipore). Real-time PCR detection using the primers described in Table S3 was conducted with a 7500 Sequence Detection System using the SYBR green Master Mix (Applied Biosystems).

Before transfection, T cells were cultured with 10 nM RA and 100 U/ml IL-2 with plate-bound 5 µg/ml anti-CD3 and 2 µg/ml anti-CD28 for 4 d in complete RPMI medium replaced every 2 d. Activated CD4⁺ T cells (5 × 10⁶ cells per sample) were transfected with 20 µg pGL4.1-5′-Itg-α4 (pGL4.1 harboring the 1.3 kbp 5′ regulatory region of the mouse Itg-α4 gene) and 3.5 µg pRL with the Mouse T Cell Nucleofector kit (Lonza). The cells were rested at 37°C for 4–5 h and activated with plate-bound anti-CD3, anti-CD28, and 100 U/ml IL-2 in the presence 100 nM RA for 16 h before measurement with the Dual Luciferase Assay System (Promega). Relative light units (RLU) after normalization of firefly luciferase activity with Renilla luciferase (pRL) activity are shown.

For homing experiments, naive CD4⁺ T cells isolated from splenocytes and LN cells of WT and BATF KO mice were cultured for 5–6 d in complete RPMI 1640 medium (10% FBS) supplemented with 2.5 µg/ml concanavalin A, 25 U/ml hIL-2, and 10 nM RA. WT cells (10⁷ cells/mouse; labeled with CFSE; Invitrogen) and BATF KO cells (10⁷ cells/mouse; labeled with tetramethylrhodamine isothiocyanate; Invitrogen) were coinjected i.v. into normal C57BL/6 mice. The host mice were sacrificed ∼20 h later, and indicated organs were harvested. The numbers of injected CD4⁺ cells migrated into each organ were determined with flow cytometry. Absolute numbers of T cells that migrated into various organs and homing index were determined as described previously (Wang et al., 2010).

Spleen, MLN, small intestine, colon, and PPs were harvested from 8–10-wk-old mice and frozen in Tissue-Tek OCT Compound (Sakura). The tissue blocks were cut into 6-µm sections, fixed in cold acetone, and stained with combinations of FITC/PE/APC-conjugated antibodies to CD4 (RM4-5), cytokeratin (AE1/AE3), and CD8 (53–6.7). The images were collected with an LSM 710 (Carl Zeiss) or SP5 II (Leica) confocal microscope. The 3D images were acquired on paraformaldehyde-fixed whole-mounted tissues with an SP5 multiphoton System (Leica) with Mai Tai Deep See Tunable IR Laser (IR laser set at 800 nm; filters were 430–480 for second harmonic generation, 500–595 for CFSE, and 595–605 for TRITC). Acquisition volume was ∼300 × 300 × 100 µm, and z-axis resolution was 0.5 µm.

T_reg cell–depleted CD4⁺CD25⁻ T cells were isolated from WT OT-II and BATF KO OT-II (CD45.2) mice. The OT-II T cells (10⁷ cells/mouse) were adoptively transferred i.v. into CD45.1 congenic mice. 2 d later, the recipient mice were injected with OVA protein (100 mg/injection in 200 µl of PBS; Sigma-Aldrich) intragastrically with a round-tip needle. On the next day, this immunization was repeated once. 12 d later, mice were sacrificed to examine the numbers of CD4⁺ FoxP3⁺ and FoxP3⁻ T cells in indicated organs.

For assessing the T_reg cell activity, CFSE-labeled CD4⁺CD25⁻ T cells (target cells, 3 × 10⁴ cells/well) and iT_reg cells as suppressors were co-cultured in round-bottom 96-well plates for 3 d at the indicated ratios with 2 µg/ml anti-CD3 antibody and irradiated splenic cells (9 × 10⁴ cells/well). iT_reg cells were generated in vitro with culturing CD4⁺CD25⁻ naive T cells for 6 d in the presence of 1 ng/ml TGFβ1, 10 nM RA, and 2.5 µg/ml concanavalin A. Dilution of CFSE indicating cell proliferation was determined by flow cytometry.

For assessment of effector T cell function, Rag1^−/− mice were injected i.p. with 0.6 × 10⁶ WT or BATF KO naive T cells. For comparison of T_reg cell function, 0.6 × 10⁶ WT naive T cells were injected together with 1.2 × 10⁶ WT or BATF KO iT_reg cells. Weight change was monitored, and the mice were sacrificed on day 25–28 after T cell transfer when some mice lost >20% of their original weight. Intestinal inflammation in Rag1^−/− mice was scored as previously described (Wang et al., 2010) based on the degree of leukocyte infiltration, mucosal hyperplasia, and loss of villi on a scale of 0–4. The histological images were obtained with a wide-field DM2000 microscope equipped with a DFC295 color camera (Leica) at 100 magnification.

For assessment of T cell proliferation in vivo, 10⁷ CFSE-labeled CD4⁺ T cells were injected i.v. into Rag1^−/− mice, and the host mice were sacrificed on day 14. For assessment of antigen-dependent T cell proliferation, CD45.1 congenic mice were injected i.v. with 10⁷ CFSE-labeled WT or BATF KO OT-II CD4⁺ T cells and immunized i.p. with 250 µg OVA in complete Freund’s adjuvant. The host mice were sacrificed 5 d later. CFSE dilution and absolute numbers of donor CD4⁺ T cells in indicated tissues were examined.

Means with SEM are shown in most of the figures. Student’s paired t test (two tailed) was used to determine the significance of differences between two groups. The weight change data in Figs. 8 and 10 was examined with a repeated measures ANOVA (SAS version 9.2; SAS Institute Inc.). P-values ≤ 0.05 were considered significant.

Fig. S1 shows the regulatory regions of CCR9 and Itg-α4 genes and the binding sites for RARα and BATF examined in this study. Videos 1–3 show 3D images of the WT and BATF KO T cells migrated into the spleen and PLN (Video 1), a PP and colonic patch (Video 2), and the small intestinal LP and colon (Video 3). Tables S1–S3 list primer sequences and putative binding sites for RAR and BATF.

This study was supported, in part, by grants from the National Institutes of Health (NIH; R01AI074745, R01DK076616, 1S10RR02829, and R01AI080769), Crohn’s and Colitis Foundation of America, and National Multiple Sclerosis Foundation to C.H. Kim and the NIH (R01CA114381) to E.J. Taparowsky.

The authors declare there are no competing financial interests.