IL-9, produced mainly by specialized T cells, mast cells, and group 2 innate lymphoid cells, regulates immune responses, including anti-helminth and allergic responses. Polarization of naive CD4 T cells into IL-9–producing T cells (Th9s) is induced by IL-4 and TGF-β1 or IL-1β. In this article, we report that the transcription factor growth factor–independent 1 transcriptional repressor (GFI1) plays a negative role in mouse Th9 polarization. Moreover, the expression of GFI1 is controlled by liganded RARα, allowing GFI1 to mediate the negative effect of retinoic acid on IL-9 expression. The Gfi1 gene has multiple RARα binding sites in the promoter region for recruiting nuclear coactivator steroid receptor coactivator-3 and p300 for histone epigenetic modifications in a retinoic acid–dependent manner. Retinoic acid–induced GFI1 binds the Il9 gene and suppresses its expression. Thus, GFI1 is a novel negative regulator of Il9 gene expression. The negative GFI1 pathway for IL-9 regulation provides a potential control point for Th9 activity.
Interleukin-9 is a cytokine produced during helminth infection and allergy responses. IL-9 activates mast cells, T cells, B cells, epithelial cells, goblet cells, and smooth muscle cells, among others (1–3). It promotes IgE production by B cells and also supports erythroid cell formation. The major cell source of IL-9 in the immune system includes T cells, group 2 innate lymphoid cells, and mast cells (4). In T cells, IL-9 expression is induced by TCR activation and cytokines such as IL-2, IL-4, TGF-β1, IL-1β, and IL-25 (5–7). Although transcription factors, such as STAT4, SMAD2, SAMD3, SMAD4, PU.1, IRF4, BATF, STAT5, ITK, STAT6, IRF8, HIF-1α, NF-AT, NOTCH, and NF-κB, mediate the signals from TCR and cytokine receptors to induce IL-9 expression, certain transcriptional regulators, such as STAT3, ID3, and T-BET, negatively regulate the IL-9 expression (8, 9).
Vitamin A and its metabolites have profound effects on the immune system and support oral tolerance development to Ags (10). Vitamin A metabolites, such as all-trans retinoic acid (RA) and 9-Cis-retinoic acid, promote naive CD4 T cell polarization to induced regulatory T cells (Tregs), but suppress it to Th17 cells (11). RAs bind RARα, RARβ, and RARγ, which are complexed with retinoid X receptor isotypes and bind retinoic acid response elements (RAREs) in regulatory regions of many genes (12). A major mechanism for the gene expression regulation by liganded RARs is mediated by their recruitment of coactivators (13), such as steroid receptor coactivator-3 (SRC3) and p300.
Growth factor–independent 1 transcriptional repressor (GFI1), a nuclear zinc-finger transcriptional regulator, regulates the development and functions of lymphoid and myeloid cells (14). GFI1 interacts with the histone demethylase lysine-specific demethylase 1, methyl transferases such as protein arginine methyltransferase 1, and histone deacetylases for transcriptional suppression (15). We studied the interactive effects of RA and RARα on an early Th cell transcriptome using robust loss- and gain-of-function approaches and found that Gfi1 is one of the genes that are coordinately and reciprocally regulated by the RA-RARα axis. We report, to our knowledge, a novel role of GFI1 in negatively regulating IL-9–producing T cell (Th9) polarization.
Materials and Methods
C57BL/6J mice (002216; Jackson), ΔRARαLck (Rara knockout [KO]), RARα-transgenic (TG), and Gfi1fl/fl CD4-cre (Gfi1−/−) mice were previously described (16, 17). All animal protocols were approved by the University of Michigan Institutional Animal Care and Use Committee. Wild-type (WT) and Gfi1−/− mice were injected i.p. with anti-CD3 (145-2C11, 20 μg/mouse) and sacrificed 2 d later to examine Th9 cells in the lung.
Cell isolation and culture
Naive CD4 T cells were isolated and activated using plate-bound anti-CD3 (145-2C11, 5 μg/ml) and anti-CD28 (37.51, 2 μg/ml) as previously described (17). For Th9-polarizing culture, human (h) IL-2 (100 U/ml), hIL-4 (30 ng/ml), hTGF-β1 (2 ng/ml), and anti–IFN-γ (10 μg/ml) were used. When indicated, mouse (m) IL-1β (10 ng/ml) and mTNF-α (10 ng/ml) were additionally used. RAs (10–20 nM) were added when indicated. The culture condition for Th17 cells was described previously (17). In brief, naive CD4 T cells were activated with Abs to CD3 and CD28 for indicated periods in the presence and absence of RA (10 nM) with IL-1β, hTGF-β1, IL-6, IL-21, IL-23, TNF-α, anti–IFN-γ, and anti–IL-4.
Single-cell suspensions from collagenase-digested tissues or cultures were assessed by flow cytometry as described previously (17). Abs used were CD4 (Clone RM4-5), CD44 (Clone IM7), IL-9 (Clone RM9A4), and IL-17a (Clone TC11-18H10.1).
RNA sequencing, chromatin immunoprecipitation sequencing, chromatin immunoprecipitation–PCR, and quantitative RT-PCR analyses
For bulk RNA sequencing (RNA-seq), total RNA was extracted from 24-h activated naive CD4+ T cells, cultured in the presence and absence of RA (10 nM) with IL-1β, hTGF-β1, IL-6, IL-21, IL-23, TNF-α, anti–IFN-γ, and anti–IL-4, using the RNeasy Mini spin kit (Qiagen, Venlo, the Netherlands) (Supplemental Fig. 1A). For chromatin immunoprecipitation (ChIP) sequencing (ChIP-seq) analysis of H3K27Ac, activated naive CD4+ T cells cultured for 16 h were processed and immunoprecipitated with Ab to H3K27Ac (clone D5E4) (Supplemental Fig. 1A). The sequencing data are deposited in NCBI GEO as GSE201728 and GSE201729. ChIP-PCR analyses (SimpleChIP; Cell Signaling Technology) for SRC3 binding on the Gfi1 gene and GFI1 binding on the Il9 gene were respectively performed on 16-h and 40-h activated CD4 T cells, which were cultured in a Th9 condition. Anti-GFI1 (E5J6J), anti-SRC3 (clone 5E11), or rabbit control IgG (clone DA1E) and the primers shown in Supplemental Table I were used. ChIP sites on the Gfi1 and Il9 genes were chosen on the basis of containing putative RARα ((G/A)(G/T)TCA) or GFI1 (AATC(A/C/T)N(A/G/T)N(C/G/T)) binding motifs (https://motifmap.ics.uci.edu/). SYBR Green real-time PCR analysis was performed for ChIP and quantitative RT-PCR assays, and the data were normalized to 2% input samples or control IgG signals.
Dual-luciferase reporter assay for Gfi1 promoter
Gibson cloning was used to generate pGL4.10 plasmids containing the original promoter region (−10 to −1085 of TSS) of the Gfi1 gene (NM_001267621) and its mutated version containing mutations ((G/A)(G/T)TCA to (G/A)(G/T)ACA) in the three putative RARα binding sites corresponding to ChIP sites b and c (Fig. 2B, Supplemental Fig. 3C). Naive T cells were preactivated for 16 h and transfected with pGL4.10_Gfi1 plasmids and pRL-CMV (Promega) using a Nucleofector (Lonza, Houston, TX). The cells were rested for 4 h and restimulated for 6 h in the Th9 polarizing condition with or without 10 nM RA. The activity of Renilla and Firefly luciferases was measured using the Dual Luciferase kit (Promega, Madison, WI) on a Synergy HT (BioTek, Winooski, VT) plate reader.
Retroviral expression of Gfi1 gene
Control and GFI1-expressing pMSCV-IRES-GFP vectors (91891 and 20672; Addgene) were used to produce retroviral supernatant from Platinum E cells. Cells were preactivated, infected with viral supernatant, and cultured in a fresh Th9-polarizing culture medium with or without 10 nM RA for 3 d after transduction. Cells were surface stained, activated for 4 h, and stained for cytokine expression (17).
Statistical significance was tested using GraphPad Prism v8.0. Differences between two groups were compared using t test or Wilcoxon test. For three or more groups, one-way ANOVA with Bonferroni’s multiple testing correction was used. The p values <0.05 were considered significant. All error bars indicate SEM.
RARα and RA have both distinct and synergistic effects on the early transcriptome of CD4 T cells
To better understand the impact of RARα on the early T cell transcriptome in RA high and low conditions, we studied the transcriptome of 24-h activated naive CD4 T cells isolated from WT, hypo-RA–sensing RARα KO (ΔRaraLck), and hyper-RA–sensing pCD2-RARα-TG mice (17) at 0 and 10 nM RA (Supplemental Fig. 1A). We performed bulk RNA-seq analysis and identified various gene groups that were induced or suppressed by RA and/or RARα synergistically or independently. The transcriptome Treeview software identified multiple gene groups that are distinctively regulated (Fig. 1A). The RA effect on gene expression was amplified in TG but decreased in KO T cells (Supplemental Fig. 1B, 1C). It is apparent that the high RARα gene dose in TG T cells augmented the effect of RA not only in a positive but also in a negative fashion. Some genes were more highly expressed in RARα KO T cells, while their expression was decreased in TG T cells, suggesting that RARα can negatively regulate many genes.
Gene set enrichment and pathway analyses predicted that the differentially regulated genes are involved in T cell responses such as proliferation, cytokine expression, and chemotaxis, among others (Supplemental Fig. 1D, 1E). Particularly, the analysis also highlighted the effect on differentiation and cytokine production related to Th17, Th22, and Th9 cells. Cytokine expression networks, such as RORγt-IL-17, AHR-IL-22, and IRF4-Il9-Il10, were predicted to be affected (Supplemental Fig. 1F).
Regulation of Il9 gene expression and Th9 cells by RARα and RA
Among the genes and pathways that were predicted to be regulated by the RA system were IL-9 expression and Th9 polarization, which have been described in part in a recent report (18). At the 24-h time point under an RA-deficient Th17 polarization condition, RARα-TG T cells expressed Il9 mRNA at the highest level among the three groups, whereas the KO T cells had the lowest expression (Fig. 1B). This suggests that RARα plays an early positive, rather than negative, role in Il9 gene transcription in low RA conditions.
At the protein level after culture for 4 d in a Th9 polarization condition, we found that the level of IL-9 expression in KO and WT T cells was comparable (Fig. 1C). TG T cells had the lowest expression of IL-9. RA was suppressive on Th9 differentiation of naive CD4 T cells from all three lines, but this suppressive effect was significantly weaker for KO T cells (Fig. 1C). We examined the regulation of Th9 polarization in four different cytokine conditions (Supplemental Fig. 2). The suppressive function of RA was detectable in all four conditions. Although less affected than WT T cells, RARα-deficient T cells were still suppressed by RA, suggesting the presence of RARα-independent mechanisms for the suppression as well.
To account for the suppressive effect of liganded RARα on IL-9 expression, we examined the expression of transcriptional factors, including those implicated in Th9 polarization (8, 9) (Fig. 1D). The expression of STAT3, ITK, IRF8, and GFI1 was increased by RA in WT T cells. This pattern of expression was clear for TG but unclear for KO T cells. In addition, we detected RA-induced SMAD3, ITK, and TBX21 in TG T cells. In contrast, the expression of IRF4, STAT5A, and BATF was downregulated by RA in WT and TG T cells. The regulation of these genes by RA was largely abolished in KO T cells, suggesting that the regulation is dependent on RARα. Gfi1 was one of the most clearly regulated by RARα and RA (Fig. 1D) and has not been associated with Th9 polarization. Interestingly, the expression of Gfi1 was increased by RA in an RARα dose-dependent manner. The regulation pattern of the Il9 gene by RA and RARα was exactly the opposite to that of Gfi1.
RA and RARα cooperatively induce the expression of GFI1
A potential mechanism for the RA suppression of IL-9 expression would be induced expression of inhibitory transcription factors. GFI1 fits this description because it is a transcription repressor (19). Our RNA expression data (Fig. 2A, Supplemental Fig. 3A) indicate that RA robustly induces Gfi1 gene transcription in T cells early at 24 h during T cell activation. The expression of Gfi1 mRNA in response to RA (10 nM) waned over time in WT T cells, but it was maintained at a high level even after 72 h in TG T cells (Supplemental Fig. 3A). This suggests that liganded RARα induces the expression of Gfi1.
Our ChIP-seq analysis for histone modification (H3K27Ac) revealed superenhancer regions on the Gfi1 locus with increased H3K27Ac modification in the presence of RA in all three groups of T cells but more robustly in TG T cells (Fig. 2A). RA has a positive effect on the H3K27Ac modification in this region, and the basal level of H3K27Ac modification was decreased in TG T cells, but it was greatly increased by RA in TG T cells (10 nM) (Fig. 2A). We further confirmed that H3K27Ac was increased in the regulatory region around exon 1 of the Gfi1 gene, which also contains RARα binding sites (Fig. 2B, Supplemental Fig. 3B). Analysis of publicly available ChIP-seq data (20) revealed RARα binding activity in the promoter and H3K27Ac peaks on the promoter/enhancer of the Gfi1 gene, and this was abolished in T cells with defective RARα function (i.e., dominant negative RARα-expressing T cells) (Fig. 2B). H3K27Ac modification is increased by histone acetyl transferases such as SRC3 (also called nuclear receptor coactivator 3), which has nuclear receptor interacting domains and an intrinsic histone acetyltransferase activity (21). Therefore, we examined SRC3 binding activity on the promoter/enhancer region of Gfi1 that contains putative RARα binding motifs (Fig. 2B). SRC3 binding to a RARE site of the Gfi1 promoter was increased by RA (Fig. 2C). SRC3 recruits the histone acetyl transferase p300 to form the SRC-p300 coactivator complex, which is required for nuclear hormone-regulated gene expression (22). p300 binding to this region was also dependent on RARα binding because it was decreased in T cells expressing a dominant-negative form of RARα (Fig. 2B) (23). We performed a RARE promoter reporter assay and found that RA increased the transcription activity of the Gfi1 promoter (Fig. 2D). When the putative RARE motifs were mutated, the regulation by RA was lost with moderately elevated or dysregulated promoter activity. Thus, the expression of the Gfi1 gene is regulated by RA and RARα.
Impact of GFI1 on Th9 polarization
We searched for canonical GFI1 binding motifs (AATC(A/C/T)N(A/G/T)N(C/G/T)) on the Il9 locus and found many of them on the Il9 gene, as well as in both the upstream and downstream regions (Fig. 2E). Our ChIP-PCR assay on CD4 T cells cultured in a Th9 polarization condition detected GFI1 binding activities on the Il9 gene, which was increased by RA (Fig. 2F). Thus, GFI1 binds the Il9 gene in an RA-dependent manner.
To determine the role of GFI1 in Th9 polarization, we used a retroviral gene transfer approach to overexpress GFI1 in CD4 T cells. GFI1 overexpression significantly decreased IL-9 expression (Fig. 3A), suggesting that GFI1 plays a negative role in Th9 polarization.
To further confirm the suppressive function of GFI1, we polarized naive CD4 T cells isolated from WT and Gfi1−/− mice and compared their IL-9–producing ability. Th9 polarization was more efficient for ΔGfi1CD4 (Gfi1−/−) T cells compared with their WT counterparts (Fig. 3B). Relatively more Gfi1−/− T cells became IL-9 producers in the presence of exogenous RA than WT T cells, further corroborating the negative role of GFI1 in Th9 polarization. To confirm the negative role of Gfi1 in vivo, we examined the frequency of Th9 cells in the lung of WT versus Gfi1−/− mice with or without the administration of anti-CD3 to boost CD4 T cell activation and differentiation. Gfi1−/− mice had increased numbers and frequency of lung Th9 cells compared with WT control mice (Fig. 3C). Anti-CD3 treatment increased the difference in Th9 numbers in the lung between WT and Gfi1−/− mice.
We studied the expression regulation of the transcription repressor GFI1 and its impact on Th9 polarization. RARα and RA synergistically, as well as independently, shape the early transcriptome of activated naive T cells to affect their differentiation. The impact of RARα and RA was unexpectedly comprehensive, potentially affecting the differentiation of all known Th subsets, including Th1, Th2, Th17, Th9, and Tregs. One of the genes that were highly regulated by RA was Gfi1. As a transcription repressor, GFI1 plays important roles in developing Th2, and its downregulation promotes the generation of Th17, induced Tregs, and Th1 cells (24, 25). In this study, we found a negative role of RA-induced GFI1 in Th9 polarization. The presence of this regulatory pathway from RA to GFI1 expression and Th9 polarization implies the possibility that Th9 activity is restrained by tissue RA levels via the functions of RARα and GFI1. We propose a novel mechanism for IL-9 regulation where RA and RARα cooperatively induce GFI1 expression and GFI1 binds the IL-9 gene to transcriptionally suppress its expression (Supplemental Fig. 3D). This pathway is thought to involve the recruitment of SRC3 and p300 acetyltransferases to the Gfi1 gene to increase histone acetylation for epigenetic regulation. The molecular mechanism by which GFI1 suppresses Th9 differentiation needs additional studies. One potential mechanism could be mediated by enhancing the activation of STAT3 by the GFI1-mediated suppression of PIAS3 (26). In this regard, STAT3 is implicated in suppressing Th9 differentiation (27). Because GFI1 expression is increased by IL-4 and suppressed by TGF-β1 and IFN-γ (24, 28), RA would cooperate or antagonize the functions of these cytokines in regulating GFI1 expression and Th cell differentiation.
We thank Phong Los, Jessica Suan, and Yoo-Jin Kim for technical assistance.
This work was supported by the National Institute of Allergy and Infectious Diseases, National Institutes of Health (NIH) (Grants R21AI14889801, R01AI074745, and 1R01AI080769 to C.H.K.) and National Institute of Allergy, NIH (Grant R01AI121302 to C.H.K.).
The online version of this article contains supplemental material.
L.F., R.K., and Q.L. carried out experiments and plotted the data. L.F. and H.Y. analyzed the RNA sequencing data. J.Z. provided Gfi1mice, and N.L. provided expertise in interpretation of the lung data. C.H.K. conceived and directed the study, obtained funding, and drafted the manuscript. All participated in making the final manuscript for submission.
The sequencing data presented in this article have been submitted to the National Center for Biotechnology Information Gene Expression Omnibus under accession numbers GSE201728 and GSE201729.
Abbreviations used in this article:
chromatin immunoprecipitation sequencing
growth factor–independent 1 transcriptional repressor
all-trans retinoic acid
retinoic acid response element
steroid receptor coactivator-3
IL-9–producing T cell
regulatory T cell
The authors have no financial conflicts of interest.