IL-15 Promotes Inflammatory Th17 Cells in the Intestine

Abstract

Ulcerative colitis (UC) is a chronic gastrointestinal condition with high morbidity. Modern therapies have revolutionized the care of UC, but 10% to 25% of patients do not respond to treatments and many progress to surgery. Thus, developing new treatments remains an important goal in UC. T cells, especially T helper 17 (Th17) cells, are linked with the pathogenesis of UC and are thought to be major targets of medications in UC. Therefore, we considered cytokines that may regulate pathogenic T cells in UC, focusing on cytokines that regulate T cells in tissue, as the intestinal microenvironment contains specialized noncirculating T cell subsets. Using public sequencing datasets, we identified that the IL15 axis is upregulated in UC and CD4⁺ T cells that express the specificity conferring receptor for IL-15 (IL-2RB) exhibit a pathogenic Th17 signature. Using a combination of murine models and human biospecimens, we verified that pathogenic Th17 cells express IL-2RB. IL-15 was redundant for Th17 differentiation, but IL-15 could activate terminally differentiated Th17 cells in the colon in a JAK1-dependent manner. Thus, we found that the IL15 axis is upregulated in UC and that IL-15 can activate inflammatory Th17 cells in the colon, raising the possibility that IL-15 is a potential target for UC treatments.

Lay Summary

We found that the IL15 axis was upregulated in the intestine in inflammatory bowel disease (IBD) and especially in ulcerative colitis. IL-15 signaling in known to be important for natural killer cell and CD8⁺ T cell pathways, but its role in CD4⁺ T cell biology is less well characterized. We found that pathogenic Th17 cells express IL-2RB, the specificity conferring receptor for IL-15. IL-15 was redundant for Th17 differentiation but could activate terminally differentiated intestinal Th17 cells. Thereefore, our data suggest that IL-15 promotes the function of pathogenic intestinal Th17 cells, raising the possibility it may be target in IBD.

Introduction

Ulcerative colitis (UC) is a chronic, relapsing, and remitting condition that is thought to be driven by dysregulated immune responses to dysbiotic gut microbiota. UC presents as bloody diarrhea and is characterized by inflammation of the rectum which often extends to involve the proximal colon.¹ Modern therapies like biologics and small molecule inhibitors have substantially improved clinical outcomes in UC; however, up to 50% of patients have primary nonresponse or suffer secondary loss of response to these treatments and about 10% to 20% of patient still progress to surgery.^1,2 Further, flares of UC can occur unpredictably even in patients with long-standing and well-controlled disease.¹ Therefore, developing new therapies for UC remains a key goal in inflammatory bowel disease (IBD).

CD4⁺ T cells, especially interleukin (IL)-17–producing CD4⁺ T cells (T helper 17 [Th17]), are strongly implicated in the pathogenesis of UC.³ Risk-conferring loci for UC are enriched for genes that are highly expressed by Th17 cells, and Th17 cells are increased in the intestine of patients with UC.^4–7 Moreover, multiple murine models of colitis including IL-10^−/−, T cell transfer, and dextran sodium sulfate (DSS) colitis demonstrate the colitogenic potential of Th17 cells.^7,8 Th17 cells are defined by the production of IL-17A, but IL-17A is not thought to be directly pathogenic in IBD. Rather, murine models show that IL-17A protects the intestinal epithelium, partially explaining the clinical finding that anti-IL-17A treatment worsens IBD outcomes.^9–11 Instead, it increasingly appears that Th17 cells are pathogenic via IL-17A-independent mechanisms, including the production of granulocyte-macrophage colony-stimulating factor (GM-CSF) (encoded by csf2) and potentially interferon γ (IFNγ).¹² Adding to this complexity, Th17 cells exhibit plasticity and can be nonpathogenic, typified by IL-10 production, or pathogenic, characterized by the production of IFNγ and GM-CSF and the expression of the IL-23 receptor (IL-23R).^12,13 Thus, IL-17A is a marker of Th17 cells but does not define their phenotype. Nonetheless Th17 cells are major targets of anti-IL-23 therapies in UC, underscoring their pathogenicity in IBD.

Cytokines are comparatively easy targets for therapeutic manipulation. Because the colon microenvironment contains highly specialized T cell subsets, we focused on locally released cytokines that regulate T cells in situ.^14–18 In this respect, IL-15 directs the homeostatic processes of committed T cells, and small-scale studies report it is highly expressed in IBD.^18–20 We confirm that the IL-15 axis is upregulated in UC and that pathogenic Th17 cells in the colon express the specificity conferring IL-15 receptor β subunit (IL-2RB). Functionally, IL-15 is redundant for Th17 differentiation, but it acts directly on colon Th17 cells to drive the production of inflammatory cytokines via JAK1. Inflammatory cytokine production also likely involves RORC, which encodes the master transcription factor for Th17 cells retinoic acid RORγt. Thus, our data suggest that IL-15 may drive pathogenic Th17 cells in the colon in UC, raising the possibility that IL-15 blockade may be a treatment option for UC.

Methods

Human subjects

This study conforms with the Helsinki declaration for ethical human subjects’ research and was approved by the University of Michigan Institutional Review Board. All subjects were prospectively recruited and provided written informed consent. Patients with UC and non-IBD control subjects were identified by screening the IBD and colorectal surgery clinical services. Patients with UC had clinical, histologic, radiographic and laboratory evidence of UC based on standard definitions and were under the care of dedicated IBD specialists. Healthy control subjects included those undergoing surgery for colorectal cancer, diverticulitis, or other benign conditions and did not have any evidence of IBD. Surgical specimens from patients with medically refractory UC were obtained via the Tissue Procurement Core of the University of Michigan. All specimens were reviewed by a board-certified gastrointestinal pathologist to verify inflammation consistent with UC, demarcate areas of quiescence and clear surgical margins, and identify malignancy or infection. Gastrointestinal pathologists had access to all clinical information but were blind to our study groups.

Mice

Wild-type (WT) C57BL/6 mice were purchased from the Jackson Laboratory. All experiments used 5- to 8-week-old sex- and age-matched mice, and all mice were housed in specific-pathogen–free facilities at the University of Michigan. For all experiments, mice were monitored for health by weight, spontaneous movement, diarrhea, and grooming. This protocol was approved by the University of Michigan Animal Care and Use Committee.

Golimumab and RISK cohort data

These tissue bulk RNA-sequencing datasets are publicly available from the National Library of Medicine Gene Expression Omnibus datasets. The golimumab clinical trial in UC was accessed using the GSE number GSE92415 and contains sequencing from the rectal mucosa of patients enrolled in this trial. The trial has been published, and golimumab is approved for use in UC. Bulk sequencing from ileal Crohn’s was obtained from the Pediatric RISK Stratification Study (RISK) cohort, which is a well-published cohort of pediatric patients with IBD. These data were accessed as GSE93624.

Reanalysis of single-cell RNA-sequencing data

The processed single-cell RNA transcript counts were collected from inflamed tissue from UC, accessed via the Broad Single Cell Portal (https://singlecell.broadinstitute.org/single_cell).³³ These samples were collected and processed via the 10x Genomics and cell-ranger pipelines by the study investigators. CD4 and CD8 T cells were identified by selecting cells with at least 1 CD4 or CD8 transcript, respectively. The TRM subset was further identified by selecting cells that were CD69⁺, CD45RO⁺, and CCR7⁻. Finally, within the CD4 or CD8 TRM subset, IL2RB⁺ cells were identified. In the subset of CD4 or CD8 TRMs, we ordinated with Uniform Manifold Approximation and Projection on highly variable genes to confirm the proper integration of the distinct data sources. To identify differentially expressed genes comparing IL2RB⁺ to IL2RB⁻, we used generalized linear modeling, with the transcript level as the dependent variable and IL2RB⁺ to IL2RB⁻ as the (1 or 0) as the independent variable. A false discovery rate was calculated to correct for multiple comparisons with the Benjamini/Hochberg technique. A false discovery rate cutoff of 0.05 was used for significance. The Kyoto Encyclopedia of Genes and Genomes 2021 human gene sets were used for enrichment analysis.

Citrobacter rodentium infection

WT C rodentium (Cr) (ATCC 51459) from frozen stock was cultured in 5 mL of LB broth supplemented with ampicillin (100 μg/mL) overnight on a shaker at 200 rpm at 37 °C to generate viable colonies for infection. A spectrophotometer assessed the colony/forming units (CFU) per milliliter after the overnight incubation. Oral inoculum stocks of 10⁹ CFU/200 μL were prepared by diluting the overnight Cr cultures in phosphate-buffered saline (PBS). Mice were inoculated by oral gavage with 10⁹ CFU in 100 μL. Control mice received an equivalent volume of PBS.

Isolation of murine and human intestinal lamina propria mononuclear cells and naive CD4⁺ T cells

Mouse lamina propria mononuclear cells (LPMCs) were isolated with minor modifications to this published protocol. Briefly, the colon was cleared of stool by flushing with cold Hanks’ balanced salt solution (HBSS), then washed again with fresh cold HBSS. The tissue was then transferred to a “predigestion buffer” composed of warmed HBSS, 2.5% fetal bovine serum (FBS), 5 mM EDTA, and 1 mM dithiothreitol, and dissociated using the gentleMACS Dissociator (1 minutes, tissue dissociating setting; Miltenyi Biotech). The tissue was then incubated for 20 minutes at 37 °C with gentle mixing and was then vortexed (3000 rpm, 10 seconds) and washed in fresh cold RPMI 1640 media. The tissue was transferred to 50 mL conical tubes with 5 mL of digestion solution prewarmed to 37 °C and composed of 150 U/mL Collagenase type III (Worthington Biochemical) and 50 μg/mL DNase I, 2% FBS in complete RPMI-1640, and incubated at 37 °C in a shaker for 30 minutes (human tissue) or 45 minutes (mouse tissue). The digested tissue was put through a 100 μm cell strainer, washed in cold HBSS, and centrifuged (450 g, 10 minutes, 4 °C), and the pellet was resuspended in 1 mL of 40% Percoll. The solution was overlayed on to 2.5 mL of 80% Percoll to create a 40/80 interface. The samples were centrifuged (860 g, 20 minutes, 21 °C) and the LPMCs were aspirated from the 40/80 interface. The LPMCs were washed twice with HBSS and resuspended in 1640 RPMI.

Human LPMCs were isolated in a slightly modified manner to minimize tissue necrosis and LPMC apoptosis time. Human colon tissue was cleared of fat, washed with cold RPMI, and cut into ∼0.5 to 1 cm pieces. The tissue was then incubated (45 minutes, 37 °C) in 50 mL conical tubes and a digestion solution composed of HBSS with Ca+, Mg+, 2% FBS, 0.5 mg/mL DNase I, and 0.5 mg/mL collagenase IV. The resulting cells were strained, washed, and resuspended in complete RPMI. As previously published, human peripheral blood mononuclear cells were isolated from patients who provided blood samples by Percoll gradient separation. Naive human and mouse T cells were purified from (human) peripheral blood mononuclear cells or (mouse) spleen single-cell suspensions and the respective EasySep (STEMCELL Technologies) naive T cell isolation kits. Cell purity was verified by flow cytometry and was >95%. CD4⁺ TRMs were isolated using FACS from the mouse or human LPMCs by gating on live cells, followed by CD3⁺CD4⁺CD44⁺CD69⁺ (mouse) or CD3⁺CD4⁺CD45RO⁺CD69⁺ CCR7⁻ (human) fractions.

Stimulation of isolated T cells

For mouse Th17 differentiation assays, naive T cells were plated in 24- or 48-well culture wells for incubation via the T cell receptor with plate-bound anti-CD3 (10 μg/mL) and soluble anti-CD28 (2 μg/mL) with recombinant IL-15/IL-15RA complex (100 ng/mL; Thermo Fisher Scientific) or recombinant mouse IL-6 (50 ng/mL; BioLegend), TGF-β1 (1 ng/mL), IL-23 (5 ng/mL; BioLegend), anti-mouse IL-4 (10 ug/mL; BioLegend), and anti-mouse IFN-γ (10 μg/mL; BioLegend) and incubated for 4 days at 37 °C. Human Th17 differentiation assays were performed similarly except the concentrations of recombinant human IL-6 (30 ng/mL; BioLegend), TGF-β1 (2.25 ng/mL), IL-23 (30 ng/mL; BioLegend), anti-human IL-4 (2.5ug/mL), and anti-human IFN-γ (1 μg/mL), and the incubation days (10 days). Purified intestinal murine CD4⁺ TRMs were also stimulated with plate-bound anti-CD3 (10 μg/mL) and soluble anti-CD28 (2 μg/mL) with recombinant IL-15/IL-15RA complex (100 ng/mL) or IL-1β (10 ng/mL) and IL-23 (10 ng/mL) and incubated for 4 days in a 37 °C incubator. Purified human intestinal CD4⁺ TRMs were stimulated with plate-bound anti-CD3 (10 μg/mL) and soluble anti-CD28 (2 μg/mL) with recombinant IL-15/IL-15RA complex (100 ng/mL) or IL-1β (50 ng/mL) and IL-23 (50 ng/mL) and incubated for 4 days in a 37 °C incubator.³⁸ In some experiments, the JAK1 inhibitor filgotinib was also applied (200 nM; Caymen Chemical). Media was typically changed every 3 days to maintain cell viability and all supernatants were collected for enzyme-linked immunosorbent assay per kit instructions (R&D Systems) to quantify cytokine production. Cells were washed and RNA was extracted for quantitative polymerase chain reaction. In some experiments, cells were washed and rested overnight before phorbol 12-myristate 13-acetate and ionomycin stimulated assessment with flow cytometry.

Flow cytometry

Cells were rested overnight after purification or stimulation as described previously in RPMI supplemented with glutamine, sodium pyruvate, 100 units/mL penicillin, 100 μg/mL streptomycin, and 10% FBS before stimulation with 1X stimulation cocktail (eBioscience) with Golgi Plug (BD Biosciences) before flow cytometry for 4 to 6 hours. Flow cytometry was then performed on the LSR II (BD Biosciences) or the FACSAria II (BD Biosciences), and the data were analyzed with FlowJo v11 (BD Biosciences). The following antibodies were used for flow cytometry: CD44 BV 421 (BioLegend), CD103 PE-Dazzle 594 (BioLegend), IFNγ (APC) (BioLegend), CD4 FITC (Thermo Fisher Scientific), Fix Viability eFluor 780 (Thermo Fisher Scientific), CD69 PE-Cy7 (BioLegend), IL-17A PE (BioLegend), CD3ε BV 510 (BioLegend), CD8β PerCP-eFluor 710 (eBiosciences), CD62L PE (Miltenyi Biotec), and CCR7 APC (Miltenyi Biotec).

Quantitative polymerase chain reaction

T cells or homogenized colon were suspended in RLT buffer (RNAeasy kit; Qiagen), RNA was extracted (RNAeasy; Qiagen), and cDNA was synthesized per kit instructions (iScript cDNA Synthesis; Bio-Rad). All primers were from Bio-Rad. Some results are also normalized to baseline conditions as the figure legends indicate. Results were analyzed on a CFX Connect system (Bio-Rad).

Statistics

Experiments with groups of ≥3 were analyzed using 1-way analysis of variance, and experiments with fewer groups were analyzed with paired or unpaired Student t test as appropriate. All data were tested for skewing and normality, and parametric or nonparametric tests were applied as appropriate. Most data were analyzed using GraphPad Prism version 10.3.1 (GraphPad Software). Due to the complexity of RNA sequencing data, these data as well as predictive modeling were performed separately using R version 4.4.1 (R Foundation for Statistical Computing).

Predictive modeling

We again utilized data from the golimumab clinical trial in UC (GSE number GSE92415) to determine whether IL15 and related transcript levels from rectal mucosa can predict response to anti-tumor necrosis factor (TNF) biologic therapies. We performed Lasso (least absolute shrinkage and selection operator) L1-regularized classification model using the R package glmnet. Data were randomly split into training (75%) and test (25%) sets stratified by the proportion of responders. Lasso models were trained using 5-fold cross-validation to estimate the accuracy of the models and to tune hyperparameters. Model accuracy was then evaluated by calculating the area under the receiver-operating characteristic curve on the independent test set. Variable importance scores were extracted to determine which transcript levels were most important for the model. All modeling was done in R and GraphPad Prism.

Chemicals

Unless otherwise noted, all chemicals and media were obtained from Thermo Fisher Scientific.

Results

The IL-15 axis is upregulated in UC

IL-15, along with IL-2 and IL-7, is a member of the so-called common gamma chain (γ_c) family because this family has overlapping receptor complexes.¹⁸ The γ_c family is so fundamental to immune cell function that impairments in these pathways lead to severe combined immune deficiencies in humans.¹⁸ The IL-15R complex is composed of the γ_c subunit and a cytokine specific β subunit called IL-2RB (also called IL-15RB), which is also shared with the IL-2R complex.^21,22 The γ_c subunit and IL-2RB subunits signal to the downstream effectors JAK3 and JAK1, respectively. Of these pathways, JAK 1 and JAK3 activate STAT5, which can induce the master transcription factor of T regulatory cells (Tregs), FOXP3.^23,24 By comparison, the IL-2RB-JAK1-STAT3 pathway is less well characterized (Figure 1A). Most IL-15 signaling in vivo is via trans-presentation wherein APCs present IL-15 in the context of IL-15RA to responder T cells.^21,22

Figure 1.

The interleukin (IL)-15 axis is upregulated in inflammatory bowel disease. (A) Schematic of IL-15 signaling. (B) induction of the IL15 axis and IL23 in colon tissue of patients with ulcerative colitis (UC) enrolled in the golimumab clinical trial for UC. Healthy control subjects (HCs) are also from that study. These data present patients with active (complete Mayo score [cMS] ≥ 4) and quiescent UC (cMS <4) at any point in the trial. Changes between week 0 and 6 of the indicated transcripts in (C) responders and (D) nonresponders to golimumab. (E) Multivariable linear regression of colon tumor necrosis factor α (TNFA) induction in the colon (the dependent variable) against IL2RB and IL1B (the independent variables). All independent variables used in the regression are in the methods. (F) Area under the curve from the training (red) and test (blue) sets based on the trial data to assess the extent to which IL15 axis transcripts predict response to golimumab. A responder in (C) was defined based on the trial metric. Each data point in panels A to E is an individual patient. *P < .05, **P < .01, ***P < .001, and ****P < .0001 by analysis of variance (B), paired Student t test (C, D), or regression (E). Wk, week.

To determine if IL-15 is upregulated in IBD, we examined the expression of IL15, IL15RA, and IL2RB in the colon UC and Crohn’s disease (CD) using microarray data from the anti-TNFα golimumab clinical trial, and bulk RNA sequencing data from the Crohn’s and Colitis Foundation RISK cohort.^25,26 These datasets showed that IL15, IL15RA, and IL2RB are all upregulated in the colon of patients with active UC relative to control subjects to a similar degree as IL23, which is causally implicated in UC (Figure 1B). Further, these transcripts decline in golimumab responders but not nonresponders (Figure 1C, D).

We next examined if IL15 axis transcripts correlate with expression of TNFA in the colon, as TNFA is a key target of UC treatments. On multivariable regression, IL2RB was significantly correlated with expression of TNFA to a similar degree as IL1B (Figure 1E). In turn, IL15 and IL23A showed trends toward significance (Figure S1B). These data link IL15 axis transcripts with UC but are associative. To more directly test if IL15 axis transcripts predicts response to golimumab in UC, we trained a random forest model and then tested it on a holdout validation set. The model identified IL2RB as an important predictor and had good accuracy to predict response in UC (Figure 1F; Figure S1C). Finally, we verified that IL15 axis was upregulated in UC using an independently reported meta-analysis comprising 85 mucosal gene sets in UC (last accessed August 22, 2025, at https://premedibd.com/genes.html) (Figure S1D).²⁷ In CD, IL15 and IL15RA are upregulated in the colon relative to controls, but the differences are more modest (Figure 2; Figure S1A).²⁸ Given the larger differences in the IL15 axis UC than CD, we focused on UC. Collectively, these data support that the IL15 axis is upregulated in UC.

Figure 2.

The IL15 axis is upregulated in colonic Crohn’s disease (CD). (A) Bulk RNA sequencing data from 4 independent datasets showing that the expression of IL15 (top row), IL15RA (middle row), and IL2RB (bottom row) are upregulated in the colon in CD. Patient groupings are presented at the bottom of each figure and cover active CD, healthy control subjects, and inflamed and uninflamed areas of the rectum. The Gene Expression Omnibus series numbers (GSE) for each dataset are shown. All P values are Bonferroni corrected. These datasets are available via a web interface (https://abbviegrc.shinyapps.io/ibdtransdb) that curates IBD sequencing datasets and generates figures based on queried genes and pathways of interest. W, week.

IL2RB expressing CD4⁺ T cells in the colon in UC exhibit an inflammatory Th17 transcriptional signature

Given the upregulation of the IL15 axis in UC, we pivoted to intestinal T cells that express IL2RB in UC and would therefore be responsive to IL-15. The colon environment is highly specialized and the most abundant T cell subset in the colon, comprising ∼90% of colon T cells, are tissue-resident memory T cells (TRMs). Further, TRMs have been implicated in the pathogenesis of IBD.^{14,15,29–32} Because of these facts, we narrowed our focus to TRMs and examined IL2RB expressing colon CD8⁺ and CD4⁺ TRMs in UC using public single-cell RNA sequencing data.³³ We identified TRMs in the colon in UC as CD45RO⁺CCR7⁻CD69⁺ cells within the CD8⁺ and CD4⁺ fractions.^{14,15,29–32} Based on the Kyoto Encyclopedia of Genes and Genome annotation, genes expressed by IL2RB⁺ CD4⁺ TRMs are associated with IBD and pathogenic Th17 cells (IFNG, IL23R, IL17A, IL17F), inflammation (TNFRSF9, IKBKG, CCL20), and immune checkpoints (TNFRSF4, CD70, PDCD1, CTLA4) (Table 1). In contrast, intestinal IL2RB⁺ CD8⁺ TRMs are not associated with any gastrointestinal conditions, except nonalcoholic fatty liver disease (Table S1). Last, IL-15 pathways are known to be important for natural killer (NK) cell biology, and IL15 axis transcripts also colocalized with NK cell clusters (Figure S2).^34,35 These data suggest that IL2RB expressing CD4⁺ TRMs in the colon in UC may be pathogenic Th17 cells.

Table 1.

Kyoto Encyclopedia of Genes and Genomes pathway analysis of IL2RB expressing CD4⁺ tissue-resident memory cells from the colon of patients with active ulcerative colitis.

Term	q value	Genes
Inflammatory bowel disease	0.0001	IL10; IL22; MAF; IFNG; IL23R; IL17F; IL17A
Cytokine-cytokine receptor interaction	0.0007	IL10; IL22; IFNG; CD70; CCL20; IL23R; TNFRSF9; IL17F; CXCL13; TNFRSF4; IL17A
Th17 cell differentiation	0.0007	IL22; IFNG; IL23R; IL17F; IKBKG; CD3D; IL17A
T cell receptor signaling pathway	0.0046	IL10; IFNG; CTLA4; IKBKG; PDCD1; CD3D
IL-17 signaling pathway	0.0207	IFNG; CCL20; IL17F; IKBKG; IL17A
Mineral absorption	0.0269	MT2A; MT1X; ATP7A; HMOX2
Allograft rejection	0.0646	IL10; IFNG; HLA-E
PD-L1 expression and PD-1 checkpoint pathway in cancer	0.0805	IFNG; IKBKG; PDCD1; CD3D
Th1 and Th2 cell differentiation	0.0805	MAF; IFNG; IKBKG; CD3D
Rheumatoid arthritis	0.0805	IFNG; CCL20; CTLA4; IL17A

CD4⁺ tissue-resident memory cells were identified as CD45RO⁺IL2RB⁺C-CR7⁻CD69⁺ cells within the CD3⁺CD4⁺ fractions in the single-cell RNA sequencing data. The top 10 Kyoto Encyclopedia of Genes and Genomes terms linked with IL2RB⁺ CD4⁺ TRMs are listed. Analysis was done at https://maayanlab.cloud/Enrichr/.

Abbreviations: IL, interleukin; Th, T helper cell.

IL-2RB is expressed by inflammatory Th17 cells in the colon

These sequencing data from UC patients collectively raise the possibility that excess IL-15 in UC may drive pathogenic Th17 cells. To validate that p-Th17 cells express IL-2RB, we used the T cell transfer murine colitis model wherein RB^hi CD4⁺ T cells are transferred into RAG1^−/− recipients. Donor RB^hi CD4⁺ T cells then become IFNγ⁺ Th17 cells (Th1/17) that cause colitis in recipient RAG1^−/− mice. Thus, Th1/17 are a type of p-Th17 cell.⁸ In contrast, donor T cells that become IL-17A single positive (Th17) and IL-17⁻IFNγ⁻ (Th0) do not cause colitis.⁸ To track Th17 cells in recipient mice, we used donor T cells from IL-17A (GFP) fate-mapping mice (Figure 3A). Consistent with Table 1, Th1/17 (p-Th17) cells express IL-2RB more highly relative to nonpathogenic Th17 and Th0 cells (Figure 3B-D). IL-2RB is not simply a feature of cell activation because all subtypes express the classical activation markers CD44 and CD69 but differentially express IL-2RB (Figure 3E). Finally, CD4⁺ T cells did not express IL-15RA, so autocrine signaling is unlikely (Figure 3E). Thus, p-Th17 cells appear to preferentially express IL-2RB and may therefore be IL-15 responsive.

Figure 3.

Pathogenic T helper 17 (Th17) cells express interleukin (IL)-2RB. (A) Schematic of RAG1^−/− mice subject to T cell transfer (TCT) colitis with donor T cells from IL-17 fate-mapping (GFP) mice. Intestinal T cells in recipient RAG1^−/− mice were assessed at week 4 of TCT using flow cytometry. (B) Pathogenic Th17 cells were defined as IL-17 (GFP⁺), followed by IFNγ⁺ (Th1/17), as shown by the gating (red box). Comparators are GFP⁺IFNγ⁻ (Th17; orange box) and GFP⁻IFNγ⁻ (Th0; black box). (C) Histogram and (D) quantification of the mean fluorescence intensify (MFI) of subsets from panel B with the gray line indicating IL-2RB fluorescence minus one, the black line indicating Th0 cells, the orange line indicating Th17 cells, and the red line indicating Th1/17 subsets. (E) Histograms of groups from panel B for the indicated markers and color scheme as in (C) Each point is a biological replicate, and data are representative of 3 experimental replicates (B-E). *P < .01 by analysis of variance. FSC, forward scatter area; LPMC, lamina propria mononuclear cell.

IL-15 is redundant for Th17 differentiation but acts on terminally differentiated Th17 cells

These data show that p-Th17 express IL-2RB and therefore may respond to IL-15. The differentiation of naive T cells into lineages and the secondary activation of terminally differentiated T cells are distinct processes. This distinction is important for Th17 cells because Th17 cells exhibit developmental plasticity and express key receptors, like IL-23R, differentially across their development.^{12,13,36–38} Therefore, we considered if IL-15 directs the differentiation of Th17 cells or acts on terminally differentiated p-Th17 cells or does both. To assess this, we performed in vitro Th17 differentiation assays using naïve CD4⁺ T cells from mice and humans with IL-15/IL-15RA complex, which showed that IL-15 was redundant for Th17 differentiation (Figure 4A-D).

Figure 4.

Interleukin (IL)-15 is redundant for T helper 17 (Th17) differentiation but can act on terminally differentiated Th17 cells. Naive CD4⁺ T cells from mice were in vitro polarized to Th17 cells with or without murine IL-15/RA complex, and (A) flow cytometry plots and (B) quantitated data are presented. Naive CD4⁺ T cells from humans was in vitro polarized to Th17 cells with or without IL-15/RA complex, and (C) flow cytometry plots and (D) quantified data are presented. (E) Schematic of Citrobacter rodentium (Cr) infection or control phosphate-buffered saline (PBS)–gavaged mice, followed by the purification and stimulation of colon CD4⁺ T cells at day 10. (F) The purified colon CD4⁺ T cells were stimulated with anti-CD3/CD28 with or without IL-1β + IL-23 or IL-15/RA complex. Transcripts of the indicated genes were then determined after stimulation. Genes and normalized to GAPDH, and conditions are normalized to PBS mice. Each point is a biological replicate (B, D, F). Data are representative of 3 to 5 experimental replicates in panels A to F. *P < .05, **P < .01, ****P < .001, and*****P < .0001 by analysis of variance. FSC, forward scatter area.

To determine if IL-15 acts on committed Th17 cells, we used the Cr model because it generates terminally differentiated p-Th17 cells in the colon, some of which become p-Th17 TRMs.^30–32 We FACS CD4⁺ T cells from the colon of Cr-infected mice at day 10 of the Th17 response, which is the peak of the Th17 response in our hands (∼40% of intestinal CD4⁺ T cells are Th17), and stimulated them with anti-CD3/CD28, IL-1β + IL-23, or IL-15/RA (Figure 4E). The gating strategy is presented (Figure S3A). As controls, we used CD4⁺ T cells from PBS-gavaged (control) mice, which are not Th17 cells, and stimulation with IL-1β + IL-23, which canonically promotes the formation of p-Th17 cells.^12,36–38 Consistent with the sequencing data from patients with UC and the T cell transfer data, IL-15 upregulated inflammatory Th17 genes (il17a, csf2) and the Th17 master transcription factor rorc (Figure 4F). IL-15 also upregulated the Treg master transcription factor foxp3, which may explain why IL-15 can be pro- or anti-inflammatory (Figure S4). There were only minor differences between control and Cr mice with IL-1β + IL-23 stimulation. This may be because CD4⁺ T cells from control mice differentiate into Th17 cells under IL-1β + IL-23 stimulation thereby negating any differences between groups. Collectively, these data show IL-15 is redundant for Th17 differentiation but can stimulate committed p-Th17 cells.

IL-15 promotes inflammatory Th17 TRMs in UC via JAK1-dependent pathways

To validate the extent to which IL-15 can activate colon Th17 TRMs from patients with UC, we focused on colon tissue from patients with UC who were referred for surgery (Figure 5A). We FACS purified TRMs by gating on CD3⁺CD4⁺CD45RO⁺C-CR7⁻CD69⁺ to identify CD4⁺ TRMs (Figure S3B).^14,15,29 CD4⁺ TRMs from UC patients are known to exhibit a Th17 signature.¹⁶ The γ_c and IL-2RB subunits of the IL-15R signal via JAK3:STAT5 and JAK1:STAT3, respectively (Figure 1A).^21,22 STAT3 is a key regulator of RORC (the Th17 master transcription factor) and impaired activation of STAT3 leads to major deficiencies of Th17 cells.^36,39,40 Therefore, we considered whether IL-15 regulates p-Th17 via a JAK1:STAT3 to RORC pathway. To test this, we stimulated the purified CD4⁺ TRMs with anti-CD3/28 ± IL-15/RA ± JAK1 inhibitor. Like the murine data, IL-15/RA upregulated inflammatory cytokines, and this could be blocked by inhibiting JAK1 (Figure 5B). IL-15/RA also upregulated RORC, though to a lesser extent than the stronger RORC inducers IL-1β + IL-23, which could also be blocked by inhibiting JAK1 (Figure 5C). The impairments with JAK1 inhibition are not due reduced viability, as cell numbers were similar across conditions. While our results and the literature imply IL-15 signaling is via JAK1, these data are nonspecific because it is possible that JAK1 inhibition impacts IL-15 independent pathways in our assays. Further, it is important to recognize in these assays that IL-17A is a marker of Th17 cells but is not pathogenic in IBD.^9–11 However, Th17 cells are pathogenic in IBD via IL-17A–independent mechanisms like the production of GM-CSF.^12,13,41 Last, IL-15^−/− mice lost less weight with DSS colitis and anti-CD3-enteritis relative to wild-type mice. Though DSS is a common albeit nonspecific model, anti-CD3-enteritis is an enteritis specifically driven by Th17 effector memory T cells (TEMs) (Figure S5A, B).⁴² Thus, our data are consistent with the possibility that IL-15 acts on p-Th17 TRMs in UC, via a JAK1, RORC pathway.

Figure 5.

Interleukin (IL)-15 activates CD4⁺ tissue-resident memory T cells (TRMs) in ulcerative colitis (UC) via a JAK1 pathway. (A) Schematic of pathway to obtain colon CD4⁺ TRMs from patients with refractory UC referred for surgery. (B) CD4⁺ TRMs were FACS purified from active and inactive regions of the colon of UC patients and were stimulated with anti-CD3/CD28 and IL-15/RA with or without a JAK1 inhibitor as shown and the indicated cytokines were measured with enzyme-linked immunosorbent assay. (C) FACS-purified CD4⁺ TRMs were stimulated with anti-CD3/CD28 and cytokines as indicated without or without a JAK1 inhibitor, and the induction of RORC was determined. RORC was normalized to GAPDH. Each point is a biologic replicate, and data are representative of 3 experimental replicates (B, C). *P < .1, **P < .01, ***P < .001, and ****P < .0001 by analysis of variance by analysis of variance. FI, fold induction, GM-CSF, granulocyte-macrophage colony-stimulating factor.

Discussion

Herein, we corroborate that the IL15 axis is upregulated in IBD and that colonic CD4⁺ TRMs that express IL2RB (the cytokine specificity conferring subunit of the IL-15R) exhibit an inflammatory Th17 signature. Moreover, these IL2RB expressing Th17 TRMs are linked with IBD and are enriched in UC. IL-15 was redundant for the primary differentiation of murine and human Th17 cells, but IL-2RB was expressed on terminally differentiated TEMs and TRMs. In contrast to the role of IL-15 on CD8⁺ TRMs, IL-15 does not appear to be necessary to maintain CD4⁺ TRMs in situ. Instead, IL-15 acted directly on CD4⁺ TRMs to stimulate the production of inflammatory Th17 cytokines. This IL-15–driven cytokine production and upregulation of RORC, the master transcription factor for Th17 cells, and could be ameliorated by inhibiting JAK1. Finally, IL-15^−/− mice exhibited downregulation of Th17 pathways and were protected from DSS and Th17-driven enteritis.

Our data indicate that IL-15 is proinflammatory on Th17 TRMs in UC. However, IL-15 biology is complex involving cis- and trans-presentation (in NK cells, terminally differentiated CD8⁺ and CD4⁺ T cells, and CD8⁺ TRMs), and autocrine signaling (in APCs).^18,21,22,43 Prior studies on IL-15 and Treg/Th17 cell balance have used (1) in vitro Th17 differentiation assays with IL-15^−/− mice, (2) IL-15 alone without IL-15RA complex, (3) purified populations of terminally differentiated cells (specifically Tregs), (4) mixtures of cells containing multiple populations of differentiated subsets (like total CD4⁺ T cells), or (5) IL-15^−/−-deficient hosts.^23,24 These systems are all reasonable but also have some caveats.

Naïve T cells from IL-15^−/− mice consistently exhibit an enhanced capacity for Th17 differentiation in vitro.^23,24 However, in vitro Th17 differentiation assays only contain T cell receptor ligation and cytokines, but not exogenous or APC-derived IL-15. Therefore, the capacity of naïve T cells from IL-15^−/− mice to have enhanced Th17 differentiation suggests off-target effects of IL-15 deficiency, rather than a direct impact of IL-15 on Th17 differentiation.

The IL-15R is a heterotrimer composed of IL-15RA, primarily expressed by APCs and the γ_c and IL-2RB, which are expressed by responder T cells.¹⁸ The γ_c family of cytokines (IL-2, IL-7, IL-15) all share the γ_c receptor subunit but have varying α subunits.¹⁸ However, cytokine signaling specificity on responder cells is thought to be via the β subunit.^18,21 IL-2 and IL-15 share the β subunit, meaning that the receptor for IL-2 and IL-15 on responder T cells is the same (γ_c and IL-2RB).^21,44 Because of receptor complex is shared, it can be difficult to disentangle the relative contributions of IL-2 and IL-15, though these cytokines have different functions in vivo.⁴⁴ In turn, this can make it difficult to ascertain the exact function of IL-15 on specific cell types when mixed populations of cells are used in experiments. Thus, IL-15 has been ascribed pro- and anti-inflammatory roles. We attempted to mitigate these concerns by using IL-15/RA complex and purified CD4⁺ TRMs populations from mice and humans, with experiments using APC-free in vitro systems allowing us to control the presence of IL-15. Thus, one unifying explanation for the discrepant data is that IL-15 has pleiotropic functions that vary by the responder cell type.

IL-15 appears to modulate IL-17 production rather than drive it. IL-15 can alternatively restrain the differentiation of Th17 cells via STAT5, but in some contexts such as rheumatoid arthritis, IL-15 supports the survival of Th17 cells.^24,45 In contrast, IL-23 is the dominant activator and maintainer of IL-17-producing cells, including Th17 cells, γδ T cells, and ILC3, and underlies the therapeutic targeting of the IL-23/IL-17 axis in immune-mediated inflammatory diseases.⁴⁶ Conversely, there are no data to showing that IL-17 can induce IL-15.

Genome-wide association and murine studies have strongly linked T cells, particularly Th17 cells, with the pathogenesis of UC, and they are considered major targets of approved therapies like anti-IL-23 agents. Terminally differentiated T cells are functionally heterogeneous. Adding to this, T cells are also anatomically compartmentalized. In vivo, naïve CD4⁺ T cells engage cognate antigens and differentiate into subsets (Th17, Th1, Treg, Th2) to eradicate infection (or cause inflammation), and some of these cells are retained long term to provide systemic (circulating) memory as TEMs, and some reside long-term in peripheral tissues to provide tissue memory (TRMs). In general, TEMs and TRMs can be of any subtype of CD4⁺ T cells, so that circulating subtypes (Th17 TEMs) have tissue-resident counterparts (Th17 TRMs).^14,15 However, TRMs are transcriptionally and functionally distinct from circulating TEMs.^14,15 TRMs are highly enriched in tissues suffused with microbes like the intestine, skin, and pulmonary tree, implying they regulate the microbiota.

It has long been held that TEMs are the main purveyors of damage in UC. However, TRMs are the most abundant T cell subset in the intestine and are increasingly implicated in the pathogenesis of IBD.^7,14–17 Thus, we focused on TRMs given that they are abundant, closely aligned with the microbiota (and therefore an excellent candidate to link dysbiotic microbiota with dysregulated host responses), and are increasingly implicated in the pathogenesis of IBD. We specifically focused on CD4⁺ TRMs because (CD4⁺) Th17 cells are implicated in the pathogenesis of UC and are the target of efficacious therapies for UC. TRMs are less accessible than TEMs because they are relatively disconnected from the systemic circulation, which makes TRMs difficult to target with therapeutic agents. One approach that may circumvent this challenge is to target pathways that regulate the in situ biology of TRMs, hence we focused on IL-15, which along with the other γ_c cytokines regulates homeostatic processes in terminally differentiated cells. IL-15 is predominantly produced in the tissue microenvironment and acts to regulate the local function of resident intestinal CD8⁺ TRMs and CD8⁺ intraepithelial lymphocytes.⁴³ In contrast, the role of IL-15 in CD4⁺ TRM biology is relatively uncharacterized.

Finally, it is critical to recognize that while IL-17 is the defining cytokine of Th17 cells, IL-17(A) is probably not pathogenic in UC. Indeed, 2 separate clinical trials of anti-IL-17A agents in CD were stopped early due to poor outcomes in the treatment arm.⁹ Consistent with this, IL-17A is protective of the intestinal mucosal in murine models of colitis.^10,11 In addition, data from human and murine in vitro and in vivo systems (like experimental autoimmune encephalomyelitis) indicate that pathogenic Th17 cells are characterized by the coproduction of IL-17A and IFNɣ (“Th1/17” cells), and that GM-CSF is a major mediator of pathogenic Th17 cells.^8,12,37 Thus, the prevailing model is that IL-17A is a marker of Th17 cells but that Th17 cells are pathogenic via IL-17A–independent pathways. Collectively, our data suggest that IL-15 is upregulated in UC but declines with disease remission, and although IL-15 is redundant for Th17 differentiation and the maintenance of Th17 TRMs, it can promote inflammatory Th17 TRMs via JAK1 pathways.

Data availability

All data presented in this manuscript will be made available to requestors upon review of their requests.

Supplementary Material

Supplementary data is available at Inflammatory Bowel Diseases online.

Key Messages.

What is already known?

• T cells, especially T helper 17 (Th17) cells, are strongly implicated in the pathogenesis of ulcerative colitis (UC) and are thought to be important targets of anti-interleukin (IL)-23 therapies.

What is new here?

• We find that the IL15 axis is upregulated in UC and that IL-15 can activate pathogenic Th17 cells from the colon of patients with UC, in a JAK1-dependent manner.

How can this study help patient care?

• Blocking IL-15 pathways on colon Th17 cells may be a therapeutic approach in UC.