BATF Regulates Innate Lymphoid Cell Hematopoiesis and Homeostasis

Summary

Early hematopoietic progenitors undergo sophisticated developmental processes to become committed innate lymphoid cell (ILC) progenitors and ultimately mature ILC subsets in the periphery. Basic leucine zipper ATF-Like transcription factor (Batf) plays important roles in lymphocyte biology. We report here that Batf regulates the production of bone marrow ILC progenitors and maintenance of peripheral ILCs. The expression of Batf is induced during ILC development at the α-lymphoid progenitor (αLP) stage in response to the cytokine IL-7. As a potential mechanism, upregulated Batf binds and activates transcription of the Nfil3 gene to promote ILC hematopoiesis. Batf is necessary to maintain normal numbers of early and late ILC progenitors in the bone marrow and mature ILC1, ILC2, ILC3, and NK cells in most peripheral tissues. Batf deficiency causes ILC lymphopenia, leading to defective ILC responses to inflammatory cytokines and defective immunity to enteric bacterial infections. Thus, Batf plays critical roles in bone marrow hematopoiesis, peripheral homeostasis, and effector functions of ILCs.

One sentence summary:

Batf supports differentiation of ILC progenitors and optimal population of NK, ILC1, ILC2, and ILC3 in peripheral tissues.

Introduction

Innate lymphoid cells (ILCs) are lymphocytes with effector functions similar to T cells but lack antigen receptors. ILCs produce effector molecules that can kill or regulate various cell types to fight pathogens or promote tolerance (1–6). They also play important roles in host metabolism and tissue repair (7). ILCs are divided into the ILC1, ILC2, and ILC3 groups. The ILC1 group includes NK cells and non-NK ILC1, which have functions similar to CD8 T cells and Th1 cells, respectively. ILC2 are similar to Th2 cells and ILC3 resemble Th17 cells in terms of cytokine production, but they have other important functions beyond immune regulation (8, 9). The three ILC subsets are distinguished by their master transcription factors, including T-bet (ILC1), Gata3 (ILC2), RORγt (ILC3), and Eomes (NK cells) (8–13).

ILCs are generated from progenitors that originate from the fetal liver during embryo development and from lymphoid progenitors in the bone marrow (BM) (14, 15). ILCs originate from the common lymphoid progenitors (CLP), and α-lymphoid progenitors (αLP) have been identified to contain common progenitors for all ILC groups, including NK cells (16, 17). Common helper ILC progenitors (CHILP) generate non-NK ILC1, ILC2 and ILC3 cells (15). Among the transcriptional factors implicated in ILC development (18–20), TOX and Nfil3 support CLP commitment to early ILC progenitors (21, 22). Nfil3 is also important for NK cell progenitor (NKP) generation (23, 24). Id2 is required for the development of common ILC progenitors (25–27). Gata3 is not only important for emergence of ILC2P but also for common ILC progenitors (9, 28).

Batf is a member of the AP1 family of transcription factors that heterodimerizes with JUN proteins and often works together with IRF4 or IRF8 to regulate the functional maturation of lymphocytes (29, 30). Batf is known to regulate gene expression by acting as a pioneer transcription factor in lymphocytes (31, 32) and supports the differentiation of and cytokine production by Th2, Th17, Tfh, and Th9 cells (33–36). Batf is also required for T cell expression of gut-homing receptors and B cell immunoglobulin class recombination (37–39), and a recent report suggests that Batf is implicated in the ILC2 response to IL-25 (40).

Here we report that Batf regulates ILC hematopoiesis, peripheral homeostasis, and effector functions. Batf supports ILC hematopoiesis by promoting the commitment of CLP into ILC progenitors and is required for normal development of most ILC groups, including NK cells, ILC1, ILC2, and ILC3. Moreover, Batf is required for peripheral ILC proliferation and effector cytokine production.

Results

IL-7 receptor signaling induces Batf expression in early ILC progenitors

We examined the expression of Batf at mRNA and protein levels in ILC precursors and mature ILCs (Figure 1A and 1B). We examined publicly available bulk RNA-seq and scRNA-seq data and performed qRT-PCR to evaluate RNA expression (Figure 1A, supplementary Figure 1A–D). These data indicate that Batf is highly expressed by ILC progenitors and mature ILC subsets, and expression was highest in αLP and ILCP (Figure 1A). The expression in ILC1/2/3 subsets was higher relative to other immune cell types, including CD4/CD8 T cells, γδ T cells, B cells, DCs, and macrophages (supplementary Figure 1A, B, D). We compared the expression of Batf with that of other transcription factors, and expression of Nfil3, Id2, and Maf was higher in ILCs and their progenitors (supplementary Figure 1A–D). Batf protein expression was highest at the αLP stage (Figure 1B). CHILP and ILCP, along with ILC1, ILC2 and ILC3, expressed Batf at levels higher than DCs, B cells, and CD8 T cells.

Figure 1. Batf expression in ILCs and ILC-lymphopenia in Batf-deficient mice.

(A) Transcription activity of the Batf gene in BM ILC progenitors and small intestinal ILC subsets. (B) Expression of Batf protein in immature and mature ILCs versus other immune cells. Flow cytometry was performed. Percent positive cells and mean fluorescence intensity (MFI) are shown. (C) Induction of Batf protein expression by IL-7 in cultured CLPs in a Stat3-dependent manner. CLPs were cultured with IL-7 on OP9-DL1 cells for 4 days in the presence of IL-7 and/or a Stat3 inhibitor (Stat3-i). Levels of Batf protein and pStat3 (Tyr705) in αLPs (Lin^-CD127⁺ SCA-1⁺ Flt3^- α4β7⁺) were determined by flow cytometry. (D) Numbers of ILC1, ILC2, and ILC3 subsets in the colon of Rag1^−/− vs. Batf^−/− Rag1^−/− mice. (E) Numbers of ILC2 in the gonadal white adipose tissue (WAT) and ear skin (two per mouse). (F) Numbers of NK cells. Pooled data obtained from at least 3 different experiments (n=4–5) are shown. ^1,2Significant differences (p<0.05) from the isotype control¹ or αLP² (B). *Significant differences (p<0.05, Student t-test) from control or between groups (C-F).

To understand the mechanism of Batf upregulation in αLPs, we cultured CLPs in the presence of OP9 stromal cells expressing DL1 (OP9-DL1). These conditions induced Batf expression at a low level in αLPs (Figure 1C). IL-7 signaling promotes ILCP generation (11, 41), so we investigated whether IL-7 induces Batf expression. Addition of IL-7 to the cell culture induced Batf expression in the majority of αLPs. Stat3 is implicated in Batf expression, and IL-7R signaling activates Stat3 (42, 43), so we examined if Batf induction in ILC progenitors is Stat3-dependent. IL-7 increased the activation of Stat3, as measured by Tyr705 phosphorylation, in αLPs, and induction of Batf expression was limited by a Stat3 inhibitor (Figure 1C). The presence of Stat6 binding sites on the Batf promoter has been reported (44), and inhibition of Stat6 was effective in suppressing Batf upregulation in αLPs (supplementary Figure 2). These results indicate that IL-7 signaling induces Batf expression in ILC progenitors in a Stat3/6-dependent manner.

Batf deficiency causes ILC lymphopenia in peripheral tissues

We examined the frequency of ILC subsets in various tissues of WT versus Batf^−/− mice (supplementary Figure 3A and 3B) to assess the function of Batf. The numbers of ILC1, ILC2, T-bet^- ILC3, and T-bet⁺ ILC3 were lower in the spleen, liver, lung, MLN, colon, and small intestine (SI) of Batf^−/− mice relative to WT mice (supplementary Figure 3A–C). The numbers of NK cells were also lower in these tissues from Batf^−/− mice (supplementary Figure 3D).

Previous studies have shown that the functional maturation of T and B cells is regulated by Batf (33, 35, 37, 38). Batf deficiency could indirectly affect ILCs through its effect on T and B cells, so we generated Batf^−/− mice in the Rag1^−/− background to examine ILC subsets in the absence of T and B cells. Batf^−/− Rag1^−/− mice also had fewer ILC1, ILC2, T-bet^- ILC3, and T-bet⁺ ILC3 (Figure 1D, supplementary Figure 4A–C) relative to WT mice. ILC2 are the dominant ILC subset in adipose and skin tissues (45–47), and the number of ILC2 was lower in these tissues of Batf^−/− Rag1^−/− mice (Figure 1E). The frequency of NK cells was lower in Batf^−/− Rag1^−/− mice compared with control Rag1^−/− mice (Figure 1F, supplementary Figure 4D).

Defective fetal LTi cell development in Batf-deficient mice

LTi cells induce the formation of secondary lymphoid tissues during fetal development (48). We observed that Peyer’s Patches in Batf^−/− mice were smaller than those in WT mice, with considerable underdevelopment of both B and T cell areas (Figure 2A). No changes in lymph nodes and intestinal isolated lymphoid follicles were observed in Batf^−/− mice, confirming a previous report by others (35). Because the production of LTi cells is high during the fetal stage (8), we examined the numbers of LTi cells in the fetal liver and intestine. The frequency of LTi cells was dramatically lower in the liver and intestine of Batf^−/− feti (E13.5–14.5) relative to WT feti (Figure 2B). These results indicate that Batf supports the development of LTi cells.

Figure 2. LTi insufficiency in Batf^−/− mice.

(A) The size and numbers of Peyer’s patches in WT and Batf^−/− mice. (B) Frequencies of liver and SI LTi cells in WT and Batf^−/− fetal tissues at embryonic day 13.5–14.5. Pooled data obtained from 4 different experiments are shown. *Significant differences (p<0.05) from control.

Batf deficiency leads to decreased ILC hematopoiesis

The high expression of Batf by BM ILC progenitors raises the possibility that Batf has a critical regulatory function in ILC hematopoiesis. We examined the numbers of CLP and ILC progenitors, including αLP, CHILP, ILC2P, ILC1P, ILCP, and NK progenitor (NKP) cells (49, 50), in WT and Batf^−/− mice to better understand the regulatory role of Batf. Whereas the number of CLP was higher, the numbers of all ILC progenitors decreased from the αLP stage in Batf^−/− mice (Figure 3A). We also examined the frequency of BM B or T/B progenitors, including all lymphoid progenitors (ALP) and B lymphoid progenitors (BLP) (supplementary Figure 5A) in WT versus Batf^−/− mice. We found that the frequency of BLP along with pro/pre-B cells was abnormally high in Batf^−/− mice (supplementary Figure 5B). The frequency of LBP (lymphoid-biased progenitors), which are defined as Kit⁻Sca-1^highFlt3⁺ CD127^dim cells that are known to rapidly produce B and T cells (51), was also higher in the BM of Batf^−/− mice. Together these data indicate abnormalities in early lymphoid hematopoiesis in Batf^−/−mice.

Figure 3. Defective ILC hematopoiesis in Batf^−/− mice.

(A) Frequencies and absolute numbers of ILC progenitors in the BM from 2 femurs of WT and Batf^−/− mice. (B) In vitro differentiation of ILC1, ILC2, and ILC3 from sorted WT and Batf^−/− BM αLPs. WT and Batf^−/− BM αLPs were cultured on OP9-DL1 cells for 11 days. (C) Competitive population of BM ILC progenitors derived from the BM of WT versus Batf^−/− mice. Competitive population of NK (D) and non-NK ILC subsets (E) derived from the BM progenitors of WT versus Batf^−/− mice. Pooled data obtained from at least 3 different experiments (n=4–6) are shown. *Significant differences (p<0.05) from WT.

To assess the role of Batf in ILC differentiation, we performed an in vitro differentiation experiment for WT and Batf^−/− BM progenitors on OP9-DL1 cells. Compared with WT αLP progenitors, Batf^−/− αLP progenitors were inefficient in generating ILC1, ILC2, and ILC3 on OP9-DL1 cells (Figure 3B). We next performed a competitive BM reconstitution experiment by co-transferring equal numbers of WT and Batf^−/− BM cells into lethally irradiated mice. Compared with WT controls, Batf^−/− BM cells were less efficient in making αLP, CHILP, ILCP, ILC2P, and ILC1P cells (Figure 3C) in the bone marrow and mature ILC1, ILC3, NK cells, and ILC2 cells in the spleen, liver, lung, MLN, and SI (Figure 3D and 3E). Moreover, Batf^−/− ILC2Ps were less efficient in populating most peripheral tissues (supplementary Figure 6A). Similar deficiencies were observed for Batf^−/− ILC2 in the gonadal white adipose tissue (WAT) and skin following BM transplantation (supplementary Figure 6B and 6C). These results indicate that Batf supports the generation of ILC progenitors in the BM and mature ILC subsets in the periphery.

Comparison of the tissue populating ability of WT and Batf^−/− ILCs in parabiosis mice

ILCs maintain high levels of tissue residency in peripheral tissues and, unlike the relatively migratory T cells, tissue ILCs are barely replaced in parabiosis mice (47). To understand the impact of Batf on tissue population of ILCs, we performed a parabiosis study pairing WT and Batf^−/− mice (Figure.4A). It is known that mature ILCs in peripheral tissues have low levels of cell replacement in parabiosis mice (52).The parabiosis approach provides information regarding combined results of cell turnover (differentiation, proliferation, and death) and migration expressed as percent host-derived cells for each cell population. We examined the ILC replacement rates for WT versus Batf^−/− parabiosis mice. In BM, WT ILC progenitors in the WT mice were hardly replaced by Batf ^−/− ILC progenitors, whereas Batf ^−/− ILC progenitors in the Batf ^−/− mice were more readily replaced by their WT counterparts (supplementary Figure 7A and 7B). This occurred for all stages of BM ILC progenitors but was most apparent for αLPs and CHILPs, which indicates that Batf is important for the population stability of progenitors. However, there was no difference between WT and Batf ^−/− CLPs, suggesting that Batf does not affect the population of CLPs. In the periphery, Batf ^−/− NK cells, ILC1, ILC2, and ILC3 were replaced more readily than their WT counterparts (Figure 4B–E). It was notable that the effect of Batf was different depending on the tissue type and ILC subset, as ILCs in some tissues (i.e. colon NK, liver ILC2, and colon ILC2) did not show significant differences in cell replacement between WT and Batf ^−/− mice. Overall, the parabiosis data suggest that Batf is required for the maintenance of ILC progenitors in the BM and mature ILC populations in certain peripheral tissues.

Figure 4. Batf is required to maintain tissue population of mature ILCs in parabiosis mice.

(A) Parabiosis between WT and Batf^−/− mice was maintained for 55–60 days and examined for relative population of WT versus Batf^−/− ILC subsets in various tissues. (B-E) Frequencies of indicated host-derived ILC subsets in WT versus Batf^−/− mice. Pooled data obtained from 4–6 different pairs are shown. *Significant differences (p<0.05) from WT.

Batf regulates the expression of Nfil3 in ILC progenitors

Nfil3 is a transcription factor that is important for early ILC development and was significantly down-regulated in Batf^−/− αLPs. qRT-PCR analysis confirmed that the expression of Nfil3 was significantly lower in Batf^−/− αLPs (Figure 5A). The expression of Id2 and Ets1 was also decreased but other transcription factors did not differ greatly between WT and Batf^−/− αLPs. The defective Nfil3 expression in αLP and CHILP, but not CLP or ILC2P, was also confirmed at the protein level by flow cytometry (Figure 5B).

Figure 5. Batf induces the expression of Nfil3 transcription during ILC hematopoiesis.

(A) Expression of transcriptional factor genes at mRNA level in αLP from WT and Batf^−/− mice. qRT-PCR analysis was performed. (B) Nfil3 protein expression in CLP, αLP, CHILP, and ILC2Ps in the BM. (C) Batf binding activity in WT versus Batf^−/−αLPs. ChIP-seq profiles of Batf binding and H3K27ac in CD4 T cells, retrieved from publicly available data (32), are shown to outline Nfil3 gene structure and putative Batf binding sites. ChIP-PCR on WT and Batf^−/− αLPs was performed. (D) Regulation of Nfil3 and Id2 expression by Batf. (E) Restoration of ILC hematopoiesis in Batf^−/− progenitors by retroviral transfer of Nfil3 or Id2 gene. Batf^−/− CLPs were cultured on OP9-DL1 with IL-3, IL-6, SCF and Flt3L overnight and the cells were transduced with retroviral vectors followed by culturing for 7 (mRNA expression by total cells) or 11 days in the presence of IL-7, SCF and IL-15/IL-33/IL-23 (respectively for ILC1/ILC2/ILC3) for ILC production by GFP⁺ (Nfil3), Thy1.1⁺ (Batf), or NGFR⁺ (Id2) transduced cells. Pooled data obtained from at least 3 different experiments (n=3–4) are shown. *Significant differences (p<0.05) from WT.

To gain insights into the regulation of Nfil3 gene by Batf, we examined the Batf binding sites on the Nfil3 gene. Batf-JUN-IRF4/IRF8 protein complexes recognize both composite TRE/IRF sequences referred to as activating protein 1 (AP-1)–IRF composite elements (AICEs) and IRF-independent non-AICE elements (53–55). We found 3 non-AICE elements in the regulatory region upstream of exon 1 and exon 2 (Figure 5C). These sites display Batf binding and H3K27 acetylation activity in T lymphocytes (32). We performed a ChIP assay to detect Batf binding activity on the predicted sites on the Nfil3 gene in WT and Batf^−/− αLP cells. High levels of Batf binding activity were detected on these sites in WT but not Batf^−/− αLPs (Figure 5C). Id2 expression was also decreased in ILC progenitors (Figure 5A, supplementary Figure 8A) but we did not find any significant Batf binding activity on the Id2 gene using a similar approach (supplementary Figure 8B). This indicates that Nfil3, but not Id2 gene expression is directly regulated by Batf binding.

Next, we examined if restoration of Batf expression induces Nfil3 gene expression in Batf^−/− progenitor cells. We performed retroviral Batf gene transfer into Batf^−/− CLP cells and transduced cells were cultured with OP9-DL1 cells and IL-7. Batf gene complementation increased the expression of Nfil3 and Id2 (Figure 5D). Retroviral restoration of Nfil3 or Id2 expression in Batf^−/− CLPs was also performed to determine if this can complement the Batf deficiency during ILC differentiation. Considerable increases in ILC differentiation were detected following the Nfil3 or Id2 gene complementation (Figure 5E), indicating that Batf-dependent Nfil3 expression is functionally important for ILC differentiation.

Defective peripheral ILC maintenance and responses in Batf deficiency

We examined the impact of Batf deficiency on the peripheral ILC2 response to cytokines. We determined the population expansion and cytokine expression of ILC2 in response to IL-25 and IL-33, which are typically produced during helminth infection or fat browning (45, 56). WT and Batf^−/− mice were injected with IL-25 or IL-33 daily for 3 days, and ILC2 numbers were determined in various tissues. Treatment with these cytokines resulted in increased ILC2 numbers in the lungs, SI, WAT, and skin of both WT and Batf^−/− mice, but the numbers of ILC2 in Batf^−/− mice were lower than that of WT mice (Figure 6A). We also examined the expression of the ILC2 effector cytokines IL-5 and IL-13. Both IL-25 and IL-33 increased the production of IL-5 and/or IL-13 by ILC2 in WT mice, but the induction of these cytokines was largely defective in Batf^−/− mice (Figure 6B). We also observed a defective lung eosinophilic response in Batf^−/− mice in response to IL-33/IL-25 (Figure 6C). These results suggest that Batf is required for a normal ILC2 effector response.

Figure 6. Batf is required for normal ILC2 responses to IL-25 and IL-33.

(A) ILC2 numbers in indicated tissues following IL-25 or IL-33 administration. (B) Frequency of cytokine (IL-5 and IL-13) expressing ILC2 following IL-25 or IL-33 administration. WT and Batf^−/− mice were injected i.p. with IL-25 or IL-33 daily for 3 days and mice were sacrificed 20 h and 60 h after the last challenge with IL-33 and IL-25, respectively. (C) Absolute number of eosinophils in lung tissues in response to IL-25 or IL-33 administration. Pooled data obtained from at least 3 different experiments (n=4) are shown. *Significant differences (p<0.05) from WT.

We examined the numbers and cytokine expression by ILC3in Batf-deficient mice during infection by Citrobacter rodentium. We infected WT and Batf^−/− mice, both in the Rag1^−/− background, and all Rag1^−/− mice survived infection, whereas Batf^−/− Rag1^−/− mice succumbed to infection with greater weight loss and pathogen burdens in feces (Figure 7A). In addition, the expression of major antibacterial cytokine and effector molecules (IL-22, RegIIIβ, and RegIIIγ) expressed in the colon during infection was defective in Batf^−/− Rag1^−/− mice (Figure 7B). There was no difference in the expression level of IL-23R, but the expression of IL-22 by ILC3 in response to IL-23 decreased by Batf deficiency (supplementary Figure 9A). A similar defect in IL-22 production by ILC3 was observed in vivo in Batf^−/− mice (supplementary Figure 9B) and the level of CD127/IL-7R expression also decreased in Batf^−/− ILC3 (supplementary Figure 9C). The expression of IFN-γ by ILC1 in the spleen and colon was not affected, but ILC1 in the liver and SI and NK cells in the liver and SI had lower expression of IFN-γ (supplementary Figure 9D). The numbers of RORγt⁺ ILC3 increased during C. rodentium infection, but this was defective in Batf^−/− mice (Figure 7C). We also examined the effector function of ILC3 after cell transfer into Rag2^−/−IL2rg^−/− mice, and these results indicate that Batf^−/− ILC3 were less efficient than their WT counterparts in controlling C. rodentium infection (supplementary Figure 10). Thus, Batf is required for the optimal effector function of ILC3 during infection.

Figure 7. Increased bacterial infection and defective ILC immunity in Batf^−/− mice.

(A) Responses of WT and Batf^−/− mice in the Rag1^−/− background to infection by C. rodentium. (B) Expression of ILC3-associated intestinal effector molecules (IL-22, RegIIIβ, and RegIIIγ). RT-PCR was performed on colonic tissues on day 32 post-infection. (C) Numbers of RORγt⁺ ILC3 in the colon of WT and Batf^−/− mice infected with C. rodentium. (D) Representative plots for Ki-67 expression by colonic-ILC3 in Rag1^−/− and Rag2^−/−IL2rg^−/− with or without infection. Pooled data obtained from at least 3 different experiments (n=3 for A,B; n=4 for C; n=5 for D) are shown. *Significant differences (p<0.05) from control.

Batf is known to regulate many genes beyond effector cytokines (36, 37, 39, 53, 54, 57), so it is possible that Batf is required for the activation of tissue ILC3. Stat5 phosphorylation is induced upon cytokine stimulation in ILC3 for IL-22 expression, cell survival, and proliferation (58, 59). Stat5 T694 phosphorylation was decreased in colonic ILC3 from Batf^−/− mice (supplementary Figure 11A). Along with decreased Stat5 activation, Batf^−/− ILC3 had decreased cell survival rates and moderately decreased Ki-67 expression in the presence of ILC3-stimulating cytokines, compared with WT counterparts (supplementary Figure 11B and 11C). Thus, Batf is important for ILC3 activation and homeostasis in peripheral tissues.

Discussion

We have studied the role of Batf in regulating ILC development and maintenance and have defined critical roles that Batf plays in regulating ILCs. Our results indicate that Batf supports the generation of ILC progenitors in the BM. We also provide evidence that Batf is necessary for normal ILC survival and cytokine responses important to effector functions.

In our study, ILCs had the highest Batf expression levels among all immune cell types examined, and the level of Batf expression was particularly high in ILCs compared with other cell types such as T and B cells, where Batf is known to play important roles. Previous studies have shown that Batf is upregulated at the single positive stage in the thymus and is required for CD4 T cell differentiation into Th2, Th9, and Th17 cells (33–36). Batf is also important for B cell activation and class switch recombination needed for antibody production (37, 38). Batf expression is detected in a small subset of CLPs, but the majority of αLPs express Batf. This suggests that Batf is upregulated at a late CLP-to-αLP differentiation stage. Batf expression is somewhat lower, but still elevated, in downstream ILC progenitors, including CHILP, ILCP, ILC2P, and ILC1P and mature ILC subsets (ILC1, ILC2 and ILC3) in the periphery. Thus, high expression of Batf is a defining feature of the ILC lineage progenitor cells. IL-7 is known to be critical to the induction of all lymphoid linage cells, and here we observed that IL-7 can induce Batf expression in pre-ILC progenitors such as CLP as well as ILC-committed progenitors such as αLPs in a Stat3/6-dependent manner. The mechanism involved in this induction needs further investigation, and the positive roles of Stat3 and Stat6 in Batf expression in ILC progenitors do not necessarily suggest that these Stats are directly activated by IL-7 signaling in ILC progenitors. Rather, Stat3 and Stat6 may prepare progenitor cells to respond to IL-7 and mediate the upregulation of Batf expression during ILC hematopoiesis in the bone marrow. A similar role for Stat3 has been reported for B cell lymphopoiesis (60).

Our results indicate that Batf promotes the commitment of common progenitors, such as CLPs, into ILC-lineage committed progenitors, producing normal numbers of ILC progenitors in BM. Defects in ILC hematopoiesis are detectable at the αLP stage when Batf is most highly expressed during ILC hematopoiesis. It appears that Batf promotes the differentiation of αLPs to downstream ILC progenitors and mature ILC subsets. Because of its early action during ILC hematopoiesis, Batf affects the differentiation of all subsets of ILCs, including NK cells, non-NK ILC1, ILC2 and ILC3 subsets.

Batf-JUN heterodimers recognize AP-1 binding sites such as the TPA response elements (TRE: TGA(C/G)TCA) (29). Many effector cytokine genes, such as IL-4, IL-5, IL-9, IL-10, IL-13, IL-17a, and IL-22, have the AP-1–IRF Composite Elements (AICEs) in their genes so that expression of these cytokines is dependent on Batf (53, 54), including the whole Th2 cytokine locus control region (61). This explains, in part, the defective function of peripheral ILC2 and ILC3. Batf also regulates other genes, including certain transcriptional factors. For example, the Nfil3 gene has at least three active binding sites for Batf in αLP cells and is transcriptionally regulated by Batf, as shown in this study. Because Nfil3 is critical for ILC hematopoiesis (21, 24, 62–64), this function of Batf explains the decreased ILC hematopoiesis in Batf^−/− mice. The expression of Id2 gene is also increased by Batf but this is likely to be indirectly mediated through other genes such as Nfil3 because we failed to detect Batf binding activity on the Id2 gene.

LTi cells belong to the ILC lineage, generated from common ILC progenitors, and, therefore, they are also regulated by Batf. LTi cells, the cell type that induces lymphoid tissues during fetal development, also have diverse functions from inducing the autoimmune regulator (AIRE) gene in the thymus to enhancing B cell responses and antigen-presentation (65). They enter lymphoid tissue anlagens and induce the expression of lymphotoxin α1β2, TRANCE, CXCL13, CCL19, CCL21 and adhesion molecules by stromal cells, leading to recruitment of more LTi as well as T and B cells to the anlagen sites. LTi cells are made in the fetal liver and seed the fetal spleen, SI and thymus (48, 66). Our results indicate that Batf is required for optimal generation of fetal LTi cells. This is explained by the fact that Nfil3, which is regulated by Batf, is also required for LTi cell development (62, 64).

We found considerable deficiency in the numbers of all types of mature ILCs in peripheral tissues in Batf^−/− mice. A likely reason for this ILC deficiency is the low production of ILC progenitors and mature ILCs as discussed above. Another potential reason for ILC lymphopenia in Batf^−/− mice is the defective maintenance of ILCs in peripheral tissues. Independent of its role in ILC hematopoiesis, Batf affects the maintenance of peripheral ILCs, as evidenced by increased death and decreased survival of Batf^−/− ILC3. The positive effect of Batf on cell survival may be explained, in part, by the function of Batf as a negative regulator of Bim transcription (67). Our parabiosis study between WT and Batf^−/− mice revealed that ILC progenitors and mature ILCs require Batf to maintain homeostasis throughout the body. Without Batf, the BM niche for ILC progenitors is defective, as evidenced by increased replacement with WT ILC progenitors. This phenotype can be due in part to reduced generation of ILC progenitors. Additionally, peripheral Batf^−/− ILCs fail to maintain their hemostasis, resulting in inability to compete for niches against their WT counterparts in parabiosis mice. This phenotype of Batf^−/− ILCs is likely due to decreased ILC proliferation and increased cell death. Thus, Batf deficiency causes decreased numbers of peripheral ILCs, leading to defective ILC responses to immunological challenges.

Despite the broad effect of Batf on ILC differentiation, ILCs are still made at reduced numbers in Batf^−/− mice. This indicates that, rather than functioning as an essential master transcription factor, Batf plays a boosting role in ILC differentiation. Batf regulates many genes beyond the Nfil3 gene, and therefore additional genes or mechanisms would be involved in regulating ILC differentiation by Batf. Batf functions as a pioneer factor by binding to Batf-dependent gene loci and recruiting other important transcription factors that induce chromatin restructuring and the looping of distal regulatory elements (31, 32). In this regard, in depth studies of molecular mechanisms for regulation of ILC lineage specification by Batf should be performed.

In sum, our results establish essential roles of Batf that support the ILC system, including progenitor differentiation, survival, proliferation, and effector cytokine production. Our findings establish Batf as an important regulator of ILC homeostasis and effector functions.

Materials and Methods

Study design

The overall objective of this study was to compare the numbers and activities of ILC subsets between WT and Batf^−/− mice under various in vitro and in vivo conditions. Sample sizes vary per experiment, and power analysis was used to predict sample numbers when applicable. All relevant data were included, and the study was not blinded or randomized. Experiments were repeated three times for most experiments with some exceptions, which are specified with sample numbers in the figure legends.

Animals

Animal protocols were approved by the Animal Care and Use Committees at University of Michigan and Purdue University. C57BL/6, CD45.1, Rag1^−/−, and Rag2^−/−IL2rg^−/− mice were originally from the Jackson Laboratory and housed at University of Michigan. and Batf^−/− mice were previously described (38). Rag1^−/− mice were crossed with Batf^−/− mice to generate Rag1^−/− Batf^−/− mice.

Immune cell isolation from tissues

BM cells were obtained by flushing femurs and tibias with a syringe containing MACS buffer. Red blood cells were lysed with hypotonic ACK buffer. For intestinal cells, tissues were cut open longitudinally and washed with cold PBS. SI tissues were processed after removing Peyer’s patches. The tissues were further cut into 1–2-cm-long pieces and treated three times with HBSS containing 1 mM EDTA, 2% HEPES, and 0.35 g/L NaHCO3 to remove epithelial cells. The tissues were digested by collagenase IV (1.5 mg/ml, Worthington, Lakewood, NJ) containing 10% newborn calf serum for 45 min at 37°C to make lamina propria cell suspensions. Lung and liver tissues were cut into small pieces with scissors and digested with collagenase IV (1.5 mg/ml, Worthington, Lakewood, NJ) in RPMI-1640 media containing 10% newborn calf serum for 30 min at 37°C with continuous agitation. The digested solution was further homogenized through iron meshes and lysed with ammonium-chloride-potassium (ACK) buffer. After digestion, intestinal and lung lymphocytes were obtained by centrifuge on 40/80% Percoll gradients for 30 min. Spleen and mesenteric lymph node (MLN) tissues were made into cell suspensions by filtration through fine meshes. Splenocytes were treated with a red blood cell lysis buffer. For ILC2, skin and gonadal white adipose tissue (WAT) were minced and digested with collagenase IV (1.5 mg/ml, Worthington, Lakewood, NJ).

Flow cytometry and cell sorting

Gating information for ILC progenitors and mature ILCs (NK, ILC1, ILC2 and ILC3) is described in Supplementary Table 1. Cells were stained with antibodies to surface antigens, such as CD25/3C7, CD45.2/104, CD45.1/A20, CD90.2/53–2.1, CD127/A7R34, Sca-1/D7, KLRG-1/2F1, α4β7/DATK32, Flt3/A2F10, c-kit/2B8, NKp46/29A1.4, IL-23R/12B2B64 and lineage antigens. The lineage-specific antibody cocktail (Lin) included antibodies to CD3ε, CD4, CD8, TCRγδ, CD11b, CD11c, CD19, B220, Gr-1, NK1.1, and Ter119. For ILC1P and NKP cells, anti-NK1.1 was omitted. Cells were fixed and permeabilized with Transcription Factor Staining Buffer Kit (Tonbo) for further staining for intracellular antigens (T-bet/eBio-4B10, GATA-3/TWAJ, RORγt/AFKJS-9, PLZF/9E12, Eomes/Dan11mag, Batf/D7C5, Phospho-Stat3/D3A7, and/or Phospho-Stat5/C11C5). Most of the antibodies were from BioLegend or eBioscience unless indicated otherwise. Antibodies to detect protein phosphorylation were from Cell Signaling Technologies. Stained cells were acquired on a NovoCyte Flow Cytometer (ACEA Biosciences, Inc.), and data were analyzed with FlowJo V10.0.7 (FlowJo). For intracellular staining of cytokines, such as IL-22, IL-13, IL-5, and IFN-γ, cells, isolated from various tissues and stained for surface antigens, were activated with phorbol myristate acetate (PMA) and ionomycin in the presence of monensin for 4h, and the cells were fixed in 1% paraformaldehyde and permeabilized with saponin buffer followed by staining with antibodies to cytokines (BioLegend). For cell sorting, Lin^-CD127⁺SCA-1⁺Flt3⁺ common lymphoid progenitors (CLP) and Lin^-CD127⁺SCA-1⁺Flt3^-α4β7⁺ α-lymphoid precursor (αLP) were sorted from BM cells and Lin^-CD127⁺CD90⁺KLRG-1^- ILC3 were sorted (~95% pure) from intestinal lamina propria cells by a flow cytometry sorter (Melody, BD Biosciences).

OP9-DL1 culture for ILCs

For in vitro ILC differentiation, sorted CLPs or αLPs (1 × 10⁶ cells/well) were co-cultured with OP9-DL1 cells (pre-treated with 50 ng/ml mitomycin C for 25 min) in completed DMEM in the presence of cytokines in 6-wells plates. Recombinant murine IL-7, stem cell factor (SCF), and IL-15 (all 20 ng/ml) were used for ILC1 differentiation; IL-7, SCF, and IL-33 were used for ILC2 differentiation; IL-7, SCF, and IL-23 were used for ILC3 differentiation in 6-well plates. Half of the culture medium was replaced every 3 days, and cultured cells were harvested on day 10–12. When indicated, Stat3 inhibitor VI (100 μM, Santa Cruz), Stat5 inhibitor (200 μM, Cayman Chemical), and Stat6 inhibitor (AS1517499, 250 nM, Cayman Chemical) were added to culture.

RNA expression

For qRT-PCR analysis, total RNA from sorted ILC progenitors, retrovirus-infected cells and the indicated tissues was extracted using the RNeasy micro Kit (Qiagen) and reverse transcribed with Sensiscript RT Kit (Qiagen). Gene expression was analyzed by qRT-PCR with SYBR™ Green PCR Master Mix (ThermoFisher). Oligonucleotide primers used are listed in Suppl. Table 2.

Retroviral gene transfer into CLPs

Retroviral vectors (LZRS-IRES-GFP-Batf, MSCV-IRES-Thy1.1-Nfil3, and LZRS-IRES-NGFR-Id2) have been previously described (39, 68). Flow-sorted αLPs were cultured in DMEM (10% FBS) with IL-3 (10 ng/ml), IL-6 (10 ng/ml), SCF (20 ng/ml) and Flt3-ligand (20 ng/ml) for 18 hours. These cells were spin-infected (2300 rpm at 37°C for 90 min) with Platinum-E-conditioned medium containing retrovirus particles supplemented with 8 μg/ml polybrene. The cells were rested for 6 h and subsequently cultured for 7–13 days on mitomycin C (50 ng/ml)-treated OP9-DL1 cells with IL-7 (10 ng/ml), SCF, and Flt3-ligand. Transduced cells were cultured for 7 days and examined for mRNA expression by SYBR Green qRT-PCR. For flow cytometry of protein expression, transduced cells were cultured for 11 days, and positively transduced cells based on GFP, Thy1.1, or NGFR expression were analyzed for numbers of produced ILC1/2/3.

BM transplantation

A 1:1 mixture of BM cells (5×10⁶ each) from CD45.1⁺CD45.2⁺ WT mice and CD45.2⁺ Batf^−/− mice were co-injected i.v. into irradiated (2×500 cGy with a 3 h interval) CD45.1⁺ congenic mice. The mice were sacrificed 8 weeks later for evaluation of the frequencies and numbers of donor-derived ILC1/2/3 subsets and ILC progenitors in various organs.

Parabiosis study

Upon weaning at 3 weeks of age, CD45.1 WT and CD45.2 WT or CD45.2 Batf^−/− were cohoused for 2 weeks and then surgically paired to make parabiosis mice as described previously (69). Parabiosis mice were sacrificed 50–55 days after surgery.

Confocal microscopy

Peyer’s Patches were frozen in OCT compound (Sakura). Frozen tissue blocks were cut into 7 μm sections, fixed in acetone, and stained with antibodies to CD3 and B220. The images were obtained with a Nikon A1 confocal system equipped with GaAsP detectors.

ILC transfer and infection with C. rodentium

ILC3 (10⁶ cells per mouse), isolated from the spleen of Rag1^−/− or Rag1^−/−Batf^−/− mice (70), were transferred i.v. into Rag2^−/−IL2rg^−/− mice, and these mice were infected with C. rodentium (DBS100, 10⁹ CFU/mouse) via oral gavage and monitored for weight change, stool consistency, and C. rodentium load until day 34.

Chromatin immunoprecipitation (ChIP) assay

Putative Batf binding sites on the Nfil3 and Id2 genes were identified based on the following sequences: TGANTCRAAA (Batf), DHTGASTCA (Batf:Jun), TTTCnnnnTGASTCA (AICE1), and GAAATGASTCA (AICE2). ChIP PCR primers were designed for the putative Batf binding motifs corresponding to the known H3K27Ac and Batf binding peaks (GSE123209) visualized with Integrated Genome Browser (BioViz) (Supplementary Table 2). The ChIP assay was performed with the SimpleChiP Kit (Cell Signaling Technologies, Danvers, MA). Immunoprecipitation was performed using rabbit control immunoglobulin G or Batf (D7C5) Rabbit mAb (Cell Signaling Technology, Danvers, MA). qPCR analysis was performed using SYBR™ Green PCR Master Mix (ThermoFisher).

Statistical analysis

Student’s t-test (paired two-tailed) was used to determine the significance of differences between WT and Batf^−/− groups with Prism (GraphPad). P values are shown in Supplementary Table 3 (raw data file) and P values < or = 0.05 were considered significant.

Supplementary Material

Supplemental material

Supplementary Figure 1. Expression of Batf and other transcription factors at mRNA level in ILC versus other immune cells.

Supplementary Figure 2. Impact of Stat proteins on Batf expression in αLP cells.

Supplementary Figure 3. Comparison of ILC frequencies in peripheral tissues of WT versus Batf^−/− mice.

Supplementary Figure 4. Comparison of ILC frequencies and numbers in peripheral tissues of WT versus Batf^−/− mice on the Rag1^−/− background.

Supplementary Figure 5. Frequency of early T-B progenitors in the BM of WT versus Batf^−/− mice.

Supplementary Figure 6. Defective generation of ILC2P and ILC2 from Batf^−/− progenitors following BM transplantation.

Supplementary Figure 7. Comparison of WT and Batf^−/− progenitors in the BM of parabiosis mice.

Supplementary Figure 8. Id2 protein expression in αLPs from WT and Batf^−/− mice and Batf binding activity on Id2 gene.

Supplementary Figure 9. Effects of Batf on the expression of IL-22, IL-22R, and IFNγ by WT and Batf^−/− ILCs.

Supplementary Figure 10. Batf^−/− ILCs are defective in fighting C. rodentium infection.

Supplementary Figure 11. Impact of Batf on Stat5 activation, cell death and Ki-67 expression of colonic ILC3.

Supplementary Table 1: Surface markers of ILC subsets used for flow cytometry

Supplementary Table 2: Primers used in this study.

Supplementary Table 3. Raw data file (Excel spreadsheet).