Hypoxia-sensing by the Histone Demethylase UTX (KDM6A) Limits Colitogenic CD4+ T cells in Mucosal Inflammation

Abstract

Hypoxia is a hallmark of inflammatory conditions [e.g., IBD (inflammatory bowel disease)], and adaptive responses have consequently evolved to protect against hypoxia-associated tissue injury. Since augmenting hypoxia-induced protective responses is a promising therapeutic approach for IBD, a more complete understanding of these pathways is needed. Recent work has demonstrated that the histone demethylase UTX is oxygen-sensitive, but its role in IBD is unclear. Here, we show that hypoxia-induced deactivation of UTX downregulates T cell responses in mucosal inflammation. Hypoxia results in decreased T cell pro-inflammatory cytokine production and increased immunosuppressive regulatory T cells, and these findings are recapitulated by UTX deficiency. Hypoxia leads to T cell accumulation of H3K27me3 histone modifications, suggesting that hypoxia impairs UTX’s histone demethylase activity to dampen T cell colitogenic activity. Finally, T cell-specific UTX deletion ameliorates colonic inflammation in an IBD mouse model, implicating UTX’s oxygen-sensitive demethylase activity in counteracting hypoxic inflammation.

Introduction

Hypoxia occurs when oxygen demand outstrips supply and is a common feature of IBD (1, 2) and other inflammatory diseases. The consequences of hypoxia can be dire, leading to metabolic crisis and cell death at pathologic sites. To adapt to these hypoxic threats, mechanisms have evolved to mediate a protective response in low oxygen conditions, which represent potential pathways which may be harnessed for therapies (2). Thus, understanding mediators of hypoxia-induced protective responses is critical for the development of new treatments that bolster these pathways. Here, we identify a key role for the oxygen-sensitive histone demethylase UTX (encoded by Kdm6a) in controlling colitogenic T cell responses.

Materials and Methods

Mice

Mouse procedures were approved by UCLA IACUC. Mice were housed and bred in sterile, specific pathogen-free facilities. UTX^TCD mice were generated as previously described (3). 6–8 week old female littermates were used.

Experimental colitis model

Enriched WT or UTX^TCD T cells were transferred into Rag2^−/−mice as previously described (4). Recipients were euthanized at weight loss>20% or 10 weeks post-transfer. For colon histopathology, specimens were fixed in 10% formalin, sectioned (5 μm), and stained with H&E. Blind scoring for inflammation was performed as described (4, 5) by a veterinary pathologist.

T cell isolation and stimulation

Splenic naïve T cells were enriched (StemCell:19765) and cultured with anti-CD28 (0.5mg/ml), and plate-bound anti-CD3 (5ug/ml) for 2d. In vivo T cell activation: mice were injected with 50ug of anti-CD3 (2C11) for 2d. Small intestines were harvested for flow cytometry as previously described (6).

T helper 1 (Th1) skewing assay

Enriched splenic CD4⁺CD25⁻ naïve T cells (StemCell) were cultured with cytokines (10ng/mL IL-2 and 5ng/ml of IL-12), anti-CD28 (0.5 mg/ml), and plate-bound anti-CD3 (5ug/ml). Cells were moved into a new well with media and cytokines (2d) and analyzed by flow cytometry (4d). Cells were incubated in normoxia (37°C; 5% CO₂) or hypoxia conditions, using a modular hypoxia incubator chamber (Billups Rothenburg; MIC-101) per manufacturer instructions.

Flow cytometry

Flow cytometry was performed using the Attune NxT Acoustic Focusing cytometer and data were analyzed with FlowJo (10.7.2). eBioscience Foxp3/Transcription Factor kit was used for nuclear proteins or BD Cytofix/Cytoperm kit for cytokines.

RNA Isolation and Quantitative PCR

5×10⁵ naïve CD4⁺ T cells were isolated (StemCell: 19765) followed by RNA isolation (Zymo: R1051). cDNA (ThermoFisher: 4368813) was used in SybrGreen Real-Time PCR assay. Expression was normalized to β-actin, a housekeeping gene, and then to WT for each gene.

RNA-Sequencing

RNA samples with a RINe >6.0 were sequenced using Illumina HighSeq 4000 platform (single end, 50bp). RNA sequencing analysis was analyzed by mapping to mm10 genome (HISAT2 (2.2.1)), counts per gene identified (HTSeq), and differential expression analyses was performed (DESeq2 (1.24.0)). Pathway analysis of clustered RNA-seq data was performed using DAVID, g:profiler, and Enrichr.

CUT&Tag

Anti-H3K27me3 CUT&Tag library preparation was performed as previously described (7). H3K27me3 CUT&Tag fastq files were aligned to mm10 genome (Bowtie2 (2.2.9)), peaks were called (MACS2 (version 2.1.1)), counts per peak were identified (HTSeq) and annotated using the annotatepeaks.pl (HOMER), and visualized (Integrated Genome Browser (9.1.8)). Peaks with p-value<0.05 were considered significant.

Statistical Analyses

Data are shown as mean±SEM. Statistical differences were evaluated using a Student’s t test. Log-rank (Mantel-Cox) test with correction was used for Kaplan-Meier survival curve. Graphs were produced and statistical analyses were performed using GraphPad Prism and ggplot2 in R.

Data Availability

Sequencing Datasets: Gene Expression Omnibus (GEO) (GSE236301: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE236301).

Results

Decreased CD4⁺ T cell IFN-γ production in hypoxia is recapitulated by UTX deficiency.

In accordance with prior studies (8), CD4⁺ T cells stimulated with proinflammatory cytokines IL-2 and IL-12 produced lower levels of IFN-γ in hypoxic (1% O₂), compared to normoxic (20% O₂), conditions (Fig. 1a–c). However, the mechanism underlying hypoxia-induced changes are unclear. UTX is a ubiquitously-expressed epigenetic regulator that is oxygen-sensitive (9). We therefore hypothesized that UTX may play a role in hypoxia-induced T cell changes. To test this, we generated mice with a T cell specific deletion of UTX (UTX^TCD; Kdm6a^flox/flox Lck^cre+) and compared IFN-γ production by CD4⁺ T cells from UTX^TCD mice vs. wildtype (WT; Kdm6a^flox/flox). UTX deficiency phenocopied hypoxia conditions, with significantly decreased IFN-γ production by UTX^TCD CD4⁺ T cells (Fig. 1d–f). Notably, the hypoxia-induced decrease in IFN-γ production was eliminated by UTX deficiency (Fig. 1g–i), implicating a requirement for UTX in oxygen sensitivity in CD4⁺ T cells. To determine effects of UTX deficiency in vivo, we injected anti-CD3 antibody to trigger inflammation in the small intestine of UTX^TCD vs. WT mice (6) (Supplemental Fig. 1a). A lower frequency of IFN-γ production was detected in CD4⁺ T cells in the small intestine of UTX^TCD mice (Supplemental Fig. 1b, c). Thus, recapitulating the effects of hypoxia, UTX deficiency impairs CD4⁺ T cell IFN-γ production, both in vitro and in vivo.

Figure 1: Attenuated CD4⁺ T cell IFN-γ production in hypoxia is recapitulated by UTX deficiency.

a - i) Flow cytometry of splenic CD4⁺ T cells from WT and UTX^TCD mice stimulated with IL-12 (10ng/ml), IL-2 (5ng/ml), anti-CD28 (0.5 mg/ml), and anti-CD3 (10 ug/ml) for 4 days. a) Representative contour plots and b) %IFN-γ⁺ and c) IFN-γ MFI in WT CD4⁺ T cells cultured in normoxia vs. hypoxia (n=5). d) Contour plots and e) %IFN-γ⁺ and f) IFN-γ MFI in WT vs. UTX^TCD CD4⁺ T cells in normoxia (WT:n=5; UTX^TCD:n=3). g) Representative contour plots and h) %IFN-γ⁺ and i) IFN-γ MFI in UTX^TCD CD4⁺ T cells in normoxia vs. hypoxia (n=8). Unpaired two-tailed Student’s t-test performed with data points as individual mice (N.S.; Not Significant; *, p<0.05; **, p<0.01).

Both hypoxia and UTX deficiency are associated with H3K27me3 accumulation in CD4⁺ T cells.

We next sought to understand the mechanism by which UTX loss-of-function may underlie hypoxia-induced changes. UTX is a histone demethylase that removes repressive H3K27me3 histone modifications to activate gene transcription (Supplemental Fig. 1d). Since UTX’s demethylase activity is oxygen-sensitive (9), we hypothesized that the observed hypoxia-induced T cell changes may be due to UTX’s impaired demethylase function. Indeed, hypoxia was associated with H3K27me3 accumulation in mouse CD4⁺ T cells (Supplemental Fig. 1e,f). In parallel, H3K27me3 levels were higher in CD4⁺ T cells from UTX^TCD mice vs. WT (Supplemental Fig. 1g,h). Additionally, H3K27me3 levels were also increased in WT CD4⁺ T cells treated with an H3K27me3 demethylase inhibitor (GSK-J4) (10) vs. vehicle control (Supplemental Fig. 1i,j). These findings suggest hypoxia-induced alterations in CD4⁺ T cells are mediated by UTX’s demethylase function.

To identify genetic loci modulated by UTX’s H3K27me3 demethylase activity, we performed H3K27me3 CUT&Tag in splenic CD4⁺ T cells isolated from UTX^TCD vs. WT mice. A majority of sequenced reads mapped to intergenic (52.92%) and intronic (39.93%) regions, with a subset in promoter-TSS (transcription start site) regions (2.2%) (Supplemental Fig. 1k). Additionally, H3K27me3 accumulation was evident in UTX^TCD CD4⁺ T cells, with 2288 increased vs. 867 decreased peaks (p-value<0.05; FDR<0.05; FC>±0.5) (Fig. 2a; Supplemental Table 1). Gene pathway analysis of increased H3K27me3-associated genes revealed inflammatory and Th1-associated pathways (e.g. “JAK-STAT signaling” and “Th1-Cell Activation”) in UTX^TCD CD4⁺ T cells (Fig. 2b). Specifically, increased H3K27me3 levels were observed at key Th1 genes (e.g. Ifng, Tbx21, Il18r1) in UTX^TCD CD4⁺ T cells vs. WT (Fig. 2c). Together, these data demonstrate that UTX removes repressive H3K27me3 at Th1-associated gene loci in CD4⁺ T cells.

Figure 2: UTX promotes CD4⁺ T cell inflammatory pathways through histone demethylation of Th1 genes.

(a-c) H3K27me3 CUT&Tag of WT and UTX^TCD CD4⁺ T cells stimulated with anti-CD28 (0.5mg/ml), and plate-bound anti-CD3 (10ug/ml) for 48h. a) Volcano plot of differential H3K27me3 in WT vs. UTX^TCD CD4⁺ T cells plotted as Log₂ Fold Change (FC) (y-axis; Log₂ FC> 0.5) and Log₂ Mean H3K27me3 Counts (x-axis; p-value<0.05). Red dots: significantly increased peaks (Log₂FC>0.5); blue dots: significantly decreased peaks (Log₂FC>−0.5); gray dots: not significant (p>0.05). b) Pathway analysis of loci with significantly increased H3K27me3 (log₁₀(p-value): x-axis). c) Heatmap of target genes in Th1 gene pathways with differential H3K27me3. d-f) Bulk RNA-seq of WT or UTX^TCD CD4⁺ T cells after 48h of stimulation with anti-CD28 (0.5 mg/ml) and plate-bound anti-CD3 (10ug/ml) (n=3). d) Volcano plot of differentially expressed genes in WT vs. UTX^TCD CD4⁺ T cells (y-axis: Log₂FC>0.5; x-axis: Log₂ Mean Expression). Red dots: significantly increased peaks (Log₂FC>0.5;p<0.05), blue dots: significantly decreased (Log₂FC>−0.5), and gray dots: not significant (p>0.05). e) Pathway analysis of significantly downregulated RNA-seq genes (x-axis:-log₁₀(p-value)). f) Heatmap of Th1 genes that are significantly differentially expressed by RNA-seq in WT or UTX^TCD CD4⁺ T cells. g) qRT-PCR of relative Th1 gene expression in UTX^TCD CD4⁺ T cells normalized to WT. h) Log₂FC in UTX^TCD vs. WT of H3K27me3 (y axis) plotted by either decreased (>−0.5 Log2FC) (blue) or increased (>+0.5 Log2FC) (purple) expression by RNA-seq (x-axis). i) Gene tracks from of H3K27me3 CUT&Tag and RNAseq of Ifng, Tbx21, and Il18r in WT and UTX^TCD CD4⁺ T cells; Y-axis: counts per million (CPM). a,b,g). Unpaired two-tailed Student’s t-test performed with data points as individual mice (N.S.; Not Significant; *, p<0.05; **, p<0.01; **, p<0.01).

Concomitant bulk RNA sequencing (RNA-seq) identified global changes in transcription, with 1084 downregulated and 660 upregulated genes (p-value<0.05; FDR<0.05; FC>±0.5) in UTX^TCD splenic CD4⁺ T cells (Fig. 2d; Supplemental Table 2). Pathway analysis of genes downregulated in UTX^TCD CD4⁺ T cells revealed Th1 pathways (e.g. “IL-12 and Stat4 dependent signaling,” and “Pathways of Th1 Development”) (Fig. 2e). Differentially downregulated genes in UTX^TCD included multiple Th1 genes (e.g. Tbx21, Ifng, and Tnf) (Fig. 2f). Decreased expression of Th1 genes in UTX^TCD CD4⁺ T cells was verified by qRT-PCR (Fig. 2g). These data implicate UTX loss-of-function in decreased Th1 gene transcription in CD4⁺ T cells.

Multiple Th1 genes with decreased expression in UTX^TCD CD4⁺ T cells correlated with repressive H3K27me3 (Fig. 2h). For instance, significantly higher H3K27me3 accumulation was seen at Tbx21, Ifng, and Il18r1 by CUT&Tag in UTX^TCD vs. WT CD4⁺ T cells, which was accompanied by significantly decreased transcription by RNA-seq (Fig. 2i). Thus, these data suggest that UTX catalyzes the removal of repressive H3K27me3 marks to activate transcription at Th1 loci.

Hypoxia and T cell specific UTX deficiency is associated with increased Tregs

Previous work demonstrated that hypoxia not only impairs IFN-γ production by CD4⁺ T cells, but also increases suppressive CD4⁺ CD25⁺ FOXP3⁺ T regulatory cells (Tregs) (11, 12). We confirmed these findings using mouse primary naive CD4⁺ T cells stimulated with anti-CD3/CD28 antibodies in hypoxia vs. normoxia (Supplemental Fig. 2a). We also noted an increased frequency of Tregs in CD4⁺ T cells isolated from UTX^TCD mice vs. WT (Supplemental Fig. 2b). Moreover, Treg frequencies were also increased in vivo in the lamina propria (LP), mesenteric lymph nodes (MLN), small intestine (SI), and spleen of UTX^TCD vs. WT mice, after anti-CD3 antibody injection (Supplemental Fig. 2c,d). Increased Tregs in both hypoxia and UTX deficiency suggest that loss of T cell UTX function may be responsible for hypoxia-associated accumulation of Tregs.

Bulk RNA sequencing revealed significantly higher expression of Treg-associated genes (e.g. Foxp3, Socs2, Il2ra) in UTX^TCD CD4⁺ T cells (Supplemental Fig. 2e). Moreover, Treg-associated pathways (e.g. “IL-2 Receptor Beta Chain in T cell activation” and “TGF-β signaling”) were upregulated by pathway analysis (Supplemental Fig. 2f). In contrast to Th1 genes, because UTX removes repressive H3K27me3 marks to activate transcription (Supplemental Fig. 1d), it is unexpected that loss of UTX’s demethylase activity would enhance Treg gene transcription. Moreover, no significant differences in H3K27me3 deposition were seen at Treg-associated genes (e.g. Foxp3 and Ahr) in WT vs. UTX^TCD CD4⁺ T cells (Supplemental Fig. 1l,m). Thus, increased Treg gene transcription may instead be through an indirect mechanism.

T cell specific UTX deficiency protects against colitis.

Increased T cell IFN-γ production is linked to IBD, specifically, CD4⁺ T cell IFN-γ deficiency is protective in mouse models of colitis (13, 14). Additionally, decreased Tregs and immunosuppressive function are features of IBD, with Treg transfer protecting mice from colitis (15). Thus, CD4⁺ T cell IFN-γ overproduction and lack of Tregs are key pathogenic mechanisms in IBD. Based on our data that UTX deficiency lowers CD4⁺ T cell IFN-γ production and increases Treg numbers, we reasoned that impairment of T cell UTX may restore immune homeostasis and protect against colitis.

To test this, we used a CD4⁺ T cell adoptive transfer model of inflammatory colitis in which naïve CD4⁺ T cells (CD3⁺TCRβ⁺CD4⁺CD45RB^hi) from either WT or UTX^TCD mice were transferred into immunodeficient recipients (Fig. 3a) (4). Notably, recipients of UTX^TCD CD4⁺ T cells were protected from weight loss and lethality from colitis (Fig. 3b, c). Hematoxylin and Eosin (H&E) staining of colons revealed diffuse inflammation (cecum to rectum) and immune infiltration into the LP of WT recipients (Fig. 3d, left, and Fig. 3e). In contrast, recipients of UTX^TCD cells had decreased infiltration into the LP, mucosal thickening, and abnormal crypt formation, measured by blinded histopathological scoring (Fig. 3d, right, and Fig. 3e). Moreover, UTX^TCD recipients had decreased colon weight to length ratios (Fig. 3f), dampened IFN-γ production by CD4⁺ T cells (Fig. 3g), and an increased proportion of Tregs in the LP (Fig. 3h) vs. WT. Together, these data demonstrate that UTX loss in CD4⁺ T cells protects against colitis through suppressed IFN-γ production and increased Treg numbers.

Figure 3: T cell UTX deficiency protects against colitis.

A) Schematic of colitis transfer experiment (see Materials and Methods). b) % weight loss in recipients of WT vs. UTX^TCD transferred cells (n=7–8). c) Kaplan-Meier survival curve of recipients of WT vs. UTX^TCD transferred cells (n=7–8). d) Representative H&E staining of colon tissues. Arrows denote extensive infiltration in the LP in WT recipients (left), vs. mild infiltration in UTX^TCD recipients (right). Scale bar=50um. e) Average histopathological scores of colon tissues. f) Weight-to-length ratios of colons measured as weight (g) divided by length (cm) from the cecum to rectum (n=10). g,h) Flow cytometric quantification of %IFN-γ⁺ (g) and CD25⁺FOXP3⁺ Treg cells (h) in CD4⁺ T cells from the LP of WT vs. UTX^TCD recipients (n=16). Unpaired two-tailed Student’s t-test performed with data points as individual mice (*, p<0.05; ****, p<0.0001).

Discussion

Efforts are underway to treat IBD by harnessing adaptive hypoxia pathways modulated by hypoxia-inducible factor (HIF) transcription factors, which protect mucosal barrier function (16, 17). A gut epithelium-targeted HIF-1α stabilizer (GB004) has shown efficacy in early clinical trials (18). Here, we identify a hypoxia-sensitive pathway mediated by UTX that plays a protective role in colitis through its effects on T cells. Given their distinct mechanisms of action, pharmacological interventions that target HIF and UTX may have additive benefits for IBD treatment.

Key Points.

1. Hypoxia impairs the activity of the epigenetic regulator UTX in CD4⁺ T cells.

2. UTX impairment protects against colonic inflammation in an IBD mouse model.

3. UTX loss downregulates inflammatory Th1 and increases suppressive Treg responses.