RNA binding protein DDX5 restricts RORγt+ Treg suppressor function to promote intestine inflammation

Abstract

Retinoid-related orphan receptor (RAR) gamma (RORγt)-expressing regulatory T cells (RORγt⁺ T_regs) play pivotal roles in preventing T cell hyperactivation and maintaining tissue homeostasis, in part by secreting the anti-inflammation cytokine interleukin-10 (IL-10). Here, we report that hypoxia-induced factor 1α (HIF1α) is the master transcription factor for Il10 in RORγt⁺ T_regs. This critical anti-inflammatory pathway is negatively regulated by an RNA binding protein DEAD box helicase 5 (DDX5). As a transcriptional corepressor, DDX5 restricts the expression of HIF1α and its downstream target gene Il10 in RORγt⁺ T_regs. T cell-specific Ddx5 knockout (DDX5^ΔT) mice have augmented RORγt⁺ T_reg suppressor activities and are better protected from intestinal inflammation. Genetic ablation or pharmacologic inhibition of HIF1α restores enteropathy susceptibility in DDX5^ΔT mice. The DDX5-HIF1α-IL-10 pathway is conserved in mice and humans. These findings reveal potential therapeutic targets for intestinal inflammatory diseases.

Fig. 1.. DDX5 regulates homeostatic CD4⁺ T cell heterogeneity and cytokine production capacity in the intestine lamina propria.

(A) Representative images from immunohistochemistry analysis of DDX5 in the ileum and colon of wild-type (WT) mice. Enlarged images are shown on the right. White stars: intestinal epithelial cells; magenta arrows: intestinal infiltrated immune cells. IHC, immunohistochemistry. (B) Representative histogram of DDX5 protein expression in CD4⁺ T cells from the ileal lamina propria (iLP). (C) Representative flow cytometry analysis of total live CD4⁺ T cells. Gating strategy for RORγt⁺ T_reg (RORγt⁺Foxp3⁺), cT_reg (RORγt⁻Foxp3⁺), T_H17 (RORγt⁺Foxp3⁻), and other (RORγt⁻Foxp3⁻) cells in the iLP from wild-type mice under steady state. (D) Geometric mean fluorescence intensity (gMFI) of DDX5 in RORγt⁺ T_reg, cT_reg, and T_H17 from iLP and colonic lamina propria (cLP). Each dot represents the result from one mouse. *P < 0.05 and **P < 0.01; ns, not significant (multiple paired t test, n = 4). (E) Proportions of the indicated subsets among CD4⁺ T cells and their IL-10 and IL-17A production potential in iLP from steady-state CTL and DDX5^ΔT littermates. Each dot represents the result from one mouse. (F) Proportions of the indicated subsets among CD4⁺ T cells and their IL-10 and IL-17A production potential in cLP from steady-state CTL and DDX5^ΔT littermates. Each dot represents the result from one mouse. *P < 0.05 (multiple paired t test, n = 8).

Fig. 2.. DDX5 in T cells promotes intestinal inflammation.

(A) gMFI of DDX5 in the indicated T cell subsets from iLP and cLP of steady-state or anti-CD3ε–challenged wild-type mice. Each dot represents the result from one mouse. *P < 0.05 and **P < 0.01 (paired multiple t test, n = 4). (B) Weight changes of CTL and DDX5^ΔT gender-matched cohoused littermates intraperitoneally challenged with anti-CD3ε monoclonal antibodies (mAb) on days 0, 2, and 4 as indicated. **P < 0.01, ***P < 0.001, and ****P < 0.0001 (paired multiple t test, n = 14, combined from three independent experiments). (C) Proportions of the indicated subsets among CD4⁺ T cells and their IL-10 and IL-17A production potential in iLP from inflamed CTL and DDX5^ΔT littermates. *P < 0.05 (paired multiple t test, n = 5). (D) Representative flow cytometry analysis of IL-10 in ileal CD4⁺ T cell subsets from CTL and DDX5^ΔT mice administrated with anti-CD3ε. SSC, side scatter as an index of cell granularity. (E) Proportions of the indicated subsets among CD4⁺ T cells and their IL-10 and IL-17A production potential in cLP from inflamed CTL and DDX5^ΔT littermates. *P < 0.05 (paired multiple t test, n = 5).

Fig. 3.. DDX5 negatively regulates RORγt⁺ T_reg suppressor function in vitro and in vivo.

(A) Gating strategy for FACS sorting the CCR6⁻CD25⁺ (enriched with 65 to 75% cT_regs) and CCR6⁺CTLA4⁺ (enriched with 45 to 50% RORγt⁺ T_regs) populations. (B) Proliferation index of wild-type naïve cells (CD4⁺CDD62L^hiCD44^lo) cocultured with CCR6⁻CD25⁺ and CCR6⁺CTLA4⁺ cells from CTL or DDX5^ΔT mice in the indicated ratio for 72 hours. Average and SD from three replicates are shown. **P < 0.01 and ***P < 0.005 (multiple t tests). (C) Left: Experiment scheme. Right: Weight changes of RAG1^−/− recipients of wild-type naïve CD4⁺ T cells (white circles, n = 2), naïve cells together with CTL CCR6⁺CTLA4⁺ (black circles, n = 4), or naïve cells together with DDX5^ΔT CCR6⁺CTLA4⁺ cells (blue circles, n = 4). *P <0.05 and **P < 0.01 (multiple t tests). (D) Total cLP mononuclear cell counts and CD4⁺ T effector (T_eff) cell counts derived from the wild-type naïve donor on day 36 after T cell transfer in mice described in (C). (E) Normalized mRNA expression of the indicated cytokine genes in cLP mononuclear cell lysates from mice described in (C). *P < 0.05 (multiple t tests). (F) Proportions of IL-10 production potential in transferred T_reg (CD90.1⁻) and T effector (CD90.1⁺) populations on day 36 after T cell transfer in mice described in (C). Each dot represents the result from one mouse. *P <0.05 and ***P < 0.005 (multiple t tests).

Fig. 4.. Identifying DDX5 targets in T cells.

(A) Weight changes of cohoused CTL and DDX5^ΔT littermates administered with anti-CD3ε and anti-rat IgG (isotype control) or anti–IL-10R intraperitoneally on days 0, 2, and 4. *P < 0.05 (paired multiple t tests, n = 4). (B) CD90.1⁺ CTL or tdTomato⁺ DDX5^ΔT naïve CD4⁺ T cells cotransfer into wild-type mice. iLP T cell cytokine levels were analyzed 5 days after anti-CD3ε challenge. *P < 0.05 (t test). (C) IL-10 levels in cultured RORγt⁺ T_reg-like cells (n = 6). (D) Distribution of DDX5-bound RNA regions in cultured T cells. 5′-UTR, 5′ untranslated region. (E) Expression of select mRNAs. Log₂ fold changes were defined by qRT-PCR results normalized to Gapdh from T cells cultured in the TGFβ (5 ng/ml, high) over those under the TGFβ (0.1 ng/ml, low) condition (multiple t tests, n = 4). Black dot: Enriched with DDX5 eCLIP-seq signal; gray dot: DDX5 eCLIP-seq signal absent. (F) Normalized mRNA levels of two Hif1a isoforms in cultured RORγt⁺ T_reg-like cells. Each dot represents the result from one mouse. **P < 0.01 (paired multiple t test, n = 6). (G) Representative histogram of isotype or HIF1α in cultured RORγt⁺ T_reg-like cells. (H) Representative flow plots (left) and proportions (right, each dot represents the result from one mouse) of HIF1α and IL-10 expression in cultured RORγt⁺ T_reg-like cells. ***P < 0.001 (paired multiple t test, n = 6).

Fig. 5.. HIF1α regulates RORγt⁺ T_reg suppressor function.

(A) gMFI of HIF1α in the indicated iLP subsets from steady-state CTL and DDX5^Δ mice. Each dot represents the result from one mouse. *P < 0.05 (paired multiple t test, n = 8). (B) HIF1α⁺IL-10⁺ proportions among the indicated iLP subsets from steady-state mice. Each dot represents the result from one mouse. *P < 0.05 (paired multiple t test, n = 8). (C) HIF1α⁺IL-10⁺ proportions among the indicated iLP subsets from anti-CD3ε–challenged mice. Each dot represents the result from one mouse. *P < 0.05 (paired multiple t tests, n = 5). (D) Left: Workflow for Cas9-sgRNA–mediated knockdown of HIF1α. Right: gMFI of HIF1α in transduced cultured Cas9⁺ T cells normalized to results from sgEmpty. MSCV, Murine Stem Cell Virus. Each dot represents the result from one mouse. **P < 0.01 (t test, n = 4). (E) RORγt⁺ T_reg-like cells proportions among transduced cells from (D) normalized to results from sgEmpty. ns, not significant (t test, n = 4). (F) IL-10⁺ proportion among the transduced RORγt⁺ T_reg-like cells normalized to results from sgEmpty. Each dot represents the result from one mouse. ***P < 0.005 and ****P < 0.001 (t test, n = 4). (G) HIF1α occupancy on the Il10 locus in cultured RORγt⁺ T_reg-like cells. mIgG, mouse immunoglobulin G. Each dot represents the result from one mouse. **P < 0.01 (paired multiple t tests, n = 4). (H) Weight changes of indicated mice after anti-CD3ε mAb challenge. **P < 0.01 (paired multiple t tests, n = 6). (I) Proportion of IL-10–expressing RORγt⁺ T_regs from steady-state or anti-CD3ε–challenged mice. FSC, forward scatter as an index of cell granularity. *P < 0.05 (paired multiple t test).

Fig. 6.. Genetic ablation of HIF1α or inhibition of HIF1α DNA binding restores enteropathy susceptibility in DDX5^ΔT mice.

(A) Weight changes of three pairs of CTL, DDX5^ΔT, and DDX5^ΔHIF1α^ΔT mice administered with anti-CD3ε mAb. *P < 0.05 (multiple t tests, n = 3). (B) Proportions (left) and number (right) of ileal CD4⁺ T cell in the mice as shown in (A). *P < 0.05 (t test, n = 3). (C) Proportions of IL-10⁺ ileal CD4⁺ T cell subsets from (B). Each dot represents the result from one mouse. *P < 0.05 (t test). (D) Representative flow cytometry analysis of IL-10 in ileal CD4⁺ T cell subsets from indicated mice administrated with anti-CD3ε. (E) Weight changes of five pairs of CTL and DDX5^ΔT mice administered with anti-CD3ε mAb or anti-CD3ε mAb and echinomycin (100 μg/kg) or echinomycin alone (100 μg/kg). **P < 0.01, ***P < 0.005, and ****P < 0.001 (multiple t tests). (F) Disease score of ileal sections from (E). (G) Ileal CD4⁺ T cell count from (E). Each dot represents the result from one mouse. *P < 0.05 (t test). (H) Proportions of IL-10⁺ ileal CD4⁺ T cell subsets from (E). Each dot represents results from one mouse. *P < 0.05 (t test). (I) Working model: DDX5 inhibits HIF1α-mediated IL-10 expression and contact-dependent suppressor function in RORγt⁺ T_reg and promotes T cell–mediated inflammation in the intestine.

Fig. 7.. DDX5 promotes R loop disassembly and restricts RNA Pol2 recruitment to repress Hif1a transcription.

(A) DDX5-bound regions on the Hif1a transcript. (B) HIF1α⁺IL-10⁺ proportions in cultured Cas9⁺ RORγt⁺ T_reg-like cells transduced with the indicated retroviruses. Each dot represents the result from one mouse. *P < 0.05 (t test). (C) DRIP-qRT detection of RNA-DNA hybrid structures in cultured RORγt⁺ T_reg-like cells. TSS, Transcription Start Site; TTS, Transcription Termination Site. Each dot represents the result from one mouse. *P < 0.05 (paired multiple t tests, n = 4). RNaseH1, ribonuclease H1. (D) RNA Pol2 and H3K4me3 ChIP-qPCR analysis at the Hif1a locus in cultured RORγt⁺ T_reg-like cells. Each dot represents the result from one mouse. *P < 0.05 (paired multiple t test, n = 3). (E) Working model: DDX5 resolves the R loop to block RNA Pol2 loading and inhibit Hif1a transcription. (F) HIF1α⁺IL-10⁺ proportions in cultured RORγt⁺ T_reg-like cells treated with vehicle [dimethyl sulfoxide (DMSO)] or the DDX5 inhibitor (RX-5902) for 72 hours. Each dot represents the result from an independent experiment. *P < 0.05 (t test). (G) HIF1α⁺IL-10⁺ proportions in human Jurkat T cells treated with vehicle (DMSO) or the DDX5 inhibitor (RX-5902) for 72 hours. Each dot represents the result of an independent experiment. ***P < 0.001 (t test, n = 4). (H) Representative staining of HIF1α, IL-10, DDX5, and isotype (negative) in human Jurkat T cells. h, human. (I) HIF1A and IL10 expressions in human ileal RORγt⁺ T_regs as determined by single-cell RNA sequencing [GSE134809 (53) and GSE162335 (54)]. Each dot represents the average expression of these genes from one donor.