BATF regulates the development and function of IL-17 producing iNKT cells

Background BATF plays important roles in the function of the immune system. Batf null mice are deficient in both CD4+ Th17 cells and T follicular helper cells and possess an intrinsic B cell defect that leads to the complete absence of class switched Ig. In this study, Tg mice overexpressing BATF in T cells were used together with Batf null mice to investigate how altering levels of BATF expression in T cells impacts the development and function of a recently characterized population of iNKT cells expressing IL-17 (iNKT-17). Results BATF has a direct impact on IL-17 expression by iNKT cells. However, in contrast to the Th17 lineage where BATF activates IL-17 expression and leads to the expansion of the lineage, BATF overexpression restricts overall iNKT cell numbers while skewing the compartment in vivo and in vitro toward an iNKT-17 phenotype. Conclusions This work is the first to demonstrate that BATF joins RORγt as the molecular signature for all IL-17 producing cells in vivo and identifies BATF as a component of the nuclear protein network that could be targeted to regulate IL-17-mediated disease. Interestingly, these studies also reveal that while the Il17a gene is a common target for BATF regulation in Th17 and iNKT-17 cells, this regulation is accompanied by opposite effects on the growth and expansion of these two cell lineages.


Background
BATF is a basic leucine zipper transcription factor that dimerizes with the JUN proteins to direct patterns of activator protein-1 (AP-1)-mediated gene expression in the immune system [1]. The impact of disrupting BATF function in vivo has been examined by several groups [2][3][4][5]. Mice in which BATF is overexpressed using a T cell-specific promoter display a reduced number of iNKT cells [6], an increased number of CD4 + T cells expressing IL-17 (Th17) [5] and an altered cytokine environment that promotes the gross overproduction of class switched Ig by B cells [7]. Batf null mice are viable, yet display a severe deficiency in Th17 and T follicular helper cells [2,3,5]. The T cell deficiencies are combined with an intrinsic B cell defect blocking the production of class switched Ig to impair the immune response of these animals to antigen challenge [2,3]. The dramatic consequences of altering BATF expression in vivo provides evidence that BATF functions to coordinate immune system activities critical in autoimmunity, inflammation and the host response to pathogens.
The ability of BATF to promote the differentiation of naïve CD4 + T cells to the Th17 lineage has been shown to rely on the formation of IRF4/BATF protein complexes that bind and transactivate a number of genes, including Il17a/f [8]. Interestingly, we have observed a negative influence of BATF on the development of iNKT cells [6,9] and therefore sought to examine how these two opposing activities of BATF may influence the development of a recently identified subset of iNKT cells that expresses IL-17 [10][11][12][13]. Murine iNKT-17 cells are a CD4 -NK1.1population that is enriched in peripheral LN (PLN) and respond following stimulation by rapidly secreting IL-17. iNKT-17 cells express RORγt and additional markers that define them as a lineage distinct from classic iNKT cells. A role for iNKT-17 cells has been demonstrated in experimental models of airway disease, asthma and collagen-induced arthritis [12,14,15]. iNKT-17 cells are over-represented in NOD mice where their influence in the pancreas exacerbates the development of diabetes [16]. In the present study, using mouse models of BATF overexpression (CD2-HA-BATF) and deficiency (Batf ΔZ/ΔZ ), we demonstrate the importance of BATF to the development of iNKT-17 cells. Despite the overall reduction in the number of iNKT cells in CD2-HA-BATF mice, the majority express IL-17. Likewise, while peripheral iNKT cell numbers are increased in Batf ΔZ/ΔZ mice, the cells are deficient in IL-17 production. These data are consistent with results in Th17 cells [5,17] and suggest that BATF-containing protein complexes transactivate the Il17a/f gene in NKT cells as well. The novel finding is that the function of BATF as an IL-17 inducer is separate from its effect on cell growth since Th17 cell numbers are expanded in the presence of BATF, while iNKT cell numbers are reduced. To identify an in vitro system that would facilitate the study of BATF-mediated gene regulatory events relevant to the iNKT cell lineage, we describe features of the DN32.D3 hybridoma [9] that indicate similarity to the iNKT-17 lineage, including the BATF-dependent expression of Il17a mRNA. We conclude that BATF joins RORγt as the molecular signature for all IL-17 producing cells in vivo and represents an essential component of a nuclear protein network that could be targeted to regulate IL-17-mediated disease.

BATF controls CD4 + Th17 cell differentiation
To test if CD2-HA-BATF Tg mice display the increase in IL-17 expression reported for a separate model of BATF overexpression in vivo [5], splenocytes were stimulated with anti-CD3 and anti-CD28 antibodies and RNA isolated after 48 h. For comparison, cells from non-Tg (WT) and Batf ΔZ/ΔZ mice were used. Gene expression analysis by quantitative (q)RT-PCR revealed the expected high level of Il17a mRNA in Tg cells and undetectable levels of Il17a mRNA in Batf ΔZ/ΔZ cells ( Figure 1A). To implicate altered Th17 development as the major contributor to this change in Il17a expression, naïve CD4 + splenocytes from Tg, WT and Batf ΔZ/ΔZ mice were subjected to in vitro protocols skewing differentiation toward the Th17 lineage or toward Th1 cells as a control. Differentiation was assessed by measuring the expression of genes specific for these CD4 + T cell subsets. As shown in Figure 1B, levels of Th1 associated transcripts (Tbet and Ifnγ) were statistically similar across all samples, while levels of Th17 associated transcripts (Rorγt, Il17a, Il23R and Il21) were elevated in Tg samples and essentially undetectable in Batf ΔZ/ΔZ samples ( Figure 1C). These findings in CD2-HA-BATF mice confirm a positive role for BATF in the differentiation of Th17 cells.

Levels of BATF correlate with altered iNKT cell numbers in vivo
Mice overexpressing BATF in thymic T cells (p56 lck HA-BATF) display a reduced number of iNKT cells [6].

Interestingly, quantification of thymic iNKT cells in
Batf ΔZ/ΔZ mice revealed a number equivalent to WT mice [2]. As the CD2-HA-BATF transgene directs BATF overexpression to all T cells (thymic and peripheral) [7], iNKT cells were re-examined and compared to numbers in WT and Batf ΔZ/ΔZ mice. The percentage of T cells positive for interaction with glycolipid-loaded CD1d tetramers was assessed by flow cytometry using single cell suspensions from thymus, spleen and PLN. Whether calculated on the basis of percentage ( A RNA prepared from stimulated, CD4 + T cells of the indicated genotypes was analyzed for Th17-associated transcripts by qRT-PCR. B and C Naïve CD4 + T cells of the indicated genotypes were cultured in vitro under conditions skewing differentiation to Th1 (B) or Th17 (C) cells. After 5 days, RNA was analyzed for the indicated transcripts by qRT-PCR. Data were averaged from 6 (n = 6) (A) or 3 (n = 3) (B and C) mice per genotype. Bars indicate standard error. *p values < 0.05. population in the thymus, yet do reveal, for the first time, that Batf ΔZ/ΔZ mice possess a statistically significant increase in the number of peripheral iNKT cells.
In previous studies, we reported that the residual iNKT cells in p56 lck HA-BATF mice were disproportionately CD44 + NK1.1and concluded that BATF negatively influences iNKT cell expansion and maturation [9]. However, recent studies have identified an important sub-population of NK1.1 -iNKT cells that are fully mature and function in the periphery to regulate autoimmunity, inflammation and the host response to infection [13]. To profile NK1.1versus NK1.1 + iNKT cells in Tg, WT and Batf ΔZ/ΔZ mice, the flow analysis was repeated to include an evaluation of NK1.1. As shown in Figure 3A, Tg mice with a reduced number of iNKT cells show a higher percentage of NK1.1cells in all tissues. Conversely, Batf ΔZ/ΔZ mice show no bias in the distribution of NK1.1versus NK1.1 + iNKT cells in thymus and spleen, but a bias toward NK1.1 + in PLN. A graphic representation of the findings from PLN highlight the influence of BATF levels on the distribution of NK1.1 + versus NK1.1 -iNKT cells in the periphery ( Figure 3B).

Batf regulates the production of IL-17 by iNKT cells
Peripheral NK1.1 -iNKT cells respond rapidly following stimulation and produce large quantities of IL-17 [10][11][12]. These cells are considered a distinct lineage and are designated as iNKT-17 cells [13]. The fact that iNKT cells in CD2-HA-BATF mice are predominantly NK1.1 -, together with the observation that Tg animals display increased IL-17 expression and CD4 + Th17 cell numbers, suggest that BATF may regulate the differentiation of iNKT-17 cells. To examine cytokine expression by iNKT cells, splenocytes from Tg, WT and Batf ΔZ/ΔZ mice were stimulated in vitro with glycolipid antigen (αGalCer) and ELISA were performed ( Figure 4A). As expected, IL-4 is secreted efficiently by iNKT cells in WT and in Batf ΔZ/ΔZ cultures, but accumulates to a lower level in Tg cultures where there are significantly fewer iNKT cells (Figure 2). For IL-17, the level of IL-17A secreted by Tg cultures was statistically indistinguishable from WT ( Figure 4A), indicating that NK1.1cells overexpressing BATF readily produce this cytokine. On the other hand, IL-17A production by Batf ΔZ/ΔZ cells, where NK1.1and NK1.1 + iNKT cells are present in roughly equal amounts, is dramatically less than IL-17A production by WT cells ( Figure 4A). These data indicate that not only does BATF influence iNKT cell numbers, it is a critical determinant of IL-17 expression by iNKT cells. Flow cytometric analysis using intracellular staining to detect IL-17A in purified, stimulated iNKT cells from the PNL of Tg, WT and Batf ΔZ/ΔZ mice provided additional evidence in support of this conclusion ( Figure 4B).

DN32.D3 cells model IL-17 production by iNKT cells
IRF4/BATF protein complexes bind to DNA within the mouse Il17a/f locus and are essential for the transactivation of the Il17a/f gene in Th17 cells [8]. However, the mechanism by which BATF functions to regulate gene expression in different cellular contexts, including iNKT-17 cells where IRF4 does not play a transcriptional role [18], continues to be investigated. Additionally, BATF and its interaction partner proteins such as JUNB and the IRF4 and 8 proteins [8,17], are expressed independently of IL-17 status in other T cell lineages [1,9] and BATF-containing protein complexes bind directly to genes where the transcriptional outcome is the inhibition of gene expression [1,4,19,20]. These facts complicate proposing a unified model to explain how BATF regulates its target genes. Therefore, to address the role of BATF in regulating Il17a and other target genes in iNKT cells, we sought to identify an in vitro system that could be used for this purpose. DN32.D3 cells are a murine CD4 -CD8 -iNKT cell hybridoma [21] that have been used to investigate the ligand specificity of the iNKT TCR [22]. In response to stimulation, DN32.D3 cells express IL-2 [23] and upregulate genes (Il4, Il10, Il13) encoding cytokines that define classic, iNKT cells ( Figure 5A, top panel). Stimulation with αGalCer also leads to the induction of several genes essential for iNKT cell development (Zbtb16 (Plzf ), Nrp1, Egr1, Egr 2) [13,24] (Figure 5A, middle panel). Interestingly, stimulation with glycolipid does not change the level of endogenous Batf gene expression, but instead increases expression of the genes encoding the basic leucine zipper dimerization partners of BATF (c-Jun, JunB and JunD) ( Figure 5A, bottom panel). This up-regulation of JUN family members explains the increase in AP-1 DNA binding observed previously in αGalCer stimulated DN32.D3 cells [9]. Stimulated DN32.D3 cells express Rorγt mRNA ( Figure 5A) and a very high level of Il17a mRNA ( Figure 5B) which are shared features of the Th17 and iNKT-17 lineages. However, DN32.D3 cannot be classified as true iNKT-17 cells, as they are NK1.1 + [21], express Ifnγ mRNA when stimulated and do not express Il23R ( Figure 5A and C). Nevertheless, it is worth testing the role of BATF as a regulator of IL-17A expression in these cells.
Toward that goal, DN32.D3 cells were treated with Batf siRNA or control siRNA, stimulated with αGalCer loaded dimers and after 2 hr, RNA was analyzed by qRT-PCR. Batf siRNA resulted in a 50% reduction in Batf mRNA expression as well as a corresponding reduction in the level of Il17a mRNA ( Figure 5C). These data support a critical role for BATF in Il17a gene regulation while demonstrating that signaling through the iNKT cell TCR expressed by DN32.D3 cells triggers cooperating molecular events that are required for efficient IL-17 induction. We conclude that the DN32.D3 iNKT cell line is a convenient in vitro model system in which the details of these molecular events can be investigated further.

Conclusions
Our studies have identified BATF as a common regulator of lineage decisions in the murine immune system that involve the expression of IL-17. BATF joins RORγt in the transcription factor network that promotes the differentiation of IL-17 expressing cells downstream of pathways triggered by TGFβ and IL-6 in T cells [25] and by a pathway dependent on TGFβ, but not IL-6, in iNKT cells [11,26]. These studies have characterized the DN32. D3 iNKT cell line as a model in which the regulation of Il17a gene and protein expression by this lineage can be investigated further. As the molecular details controlling IL-17 production in vivo continue to emerge, new approaches to control autoimmunity, inflammation and infectious disease will become a reality.

Mice
Batf ΔZ/ΔZ and CD2-HA-BATF mice expressing human, HA-tagged BATF were described previously [2,7] and were maintained by breeding to C57Bl/6 mice (Harlan, Indianapolis, IN). Experiments were performed using sex-matched littermates between 6 and 12 wk of age. Mice were housed in a specific, pathogen-free facility. All protocols were approved by the Purdue University Animal Care and Use Committee.

Antibodies and reagents
All antibodies, cytokines and the mouse CD1d dimers were obtained from BD Biosciences unless otherwise specified. αGalCer was obtained from Axxora (Farmingdale, NY). PE-labeled, CD1d tetramers, empty or loaded with PBS-57, were obtained from the NIH tetramer core facility.
RNA was isolated using Trizol and analyzed by qRT-PCR using SYBR green (Roche Diagnostics, Indianapolis, IN) and an ABI 7300 real-time PCR system. Data were normalized to Hprt expression and relative mRNA levels calculated using ΔΔCt values. The primers for Hprt, Actin, Batf, Il4, Il23R, Il17a and Il21 have been described [2]. Il10 primers were from Qiagen (Gaithersburg, MD) and additional primers (  For BATF knock-down, DN32.D3 cells were cultured for 24 h in siRNA delivery media containing 1 μM BATF targeting (Accell siRNA smart pool) or non-targeting siRNA (Accell non-targeting siRNA 1) (Dharmacon/ Thermo Fisher Scientific, Waltham, MA). Cells were washed and stimulated for 2 h prior to analysis as described above.