MicroRNAs regulate T‐cell production of interleukin‐9 and identify hypoxia‐inducible factor‐2α as an important regulator of T helper 9 and regulatory T‐cell differentiation

Summary

MicroRNAs (miRNAs) regulate many aspects of helper T cell (Th) development and function. Here we found that they are required for the suppression of interleukin‐9 (IL‐9) expression in Th9 cells and other Th subsets. Two highly related miRNAs (miR‐15b and miR‐16) that we previously found to play an important role in regulatory T (Treg) cell differentiation were capable of suppressing IL‐9 expression when they were over‐expressed in Th9 cells. We used these miRNAs as tools to identify novel regulators of IL‐9 expression and found that they could regulate the expression of Epas1, which encodes hypoxia‐inducible factor (HIF)‐2α. HIF proteins regulate metabolic pathway usage that is important in determining appropriate Th differentiation. The related protein, HIF‐1α enhances Th17 differentiation and inhibits Treg cell differentiation. Here we found that HIF‐2α was required for IL‐9 expression in Th9 cells, but its expression was not sufficient in other Th subsets. Furthermore, HIF‐2α suppressed Treg cell differentiation like HIF‐1α, demonstrating both similar and distinct roles of the HIF proteins in Th differentiation and adding a further dimension to their function. Ironically, even though miR‐15b and miR‐16 suppressed HIF‐2α expression in Treg cells, inhibiting their function in Treg cells did not lead to an increase in IL‐9 expression. Therefore, the physiologically relevant miRNAs that regulate IL‐9 expression in Treg cells and other subsets remain unknown. Nevertheless, the analysis of miR‐15b and miR‐16 function led to the discovery of the importance of HIF‐2α so this work demonstrated the utility of studying miRNA function to identify novel regulatory pathways in helper T‐cell development.

Abbreviations

HIF

hypoxia‐inducible factor

IFN‐γ

interferon‐γ

IL‐9

interleukin‐9

iTreg

in vitro‐induced Treg

miRNAs

microRNAs

TGF‐β

transforming growth factor‐β

Th

helper T cell

Treg

regulatory T cell

UTR

untranslated region

Introduction

Interleukin‐9 (IL‐9) is a cytokine that has recently received a significant amount of attention for its role in tumour immunity,1, 2 the immune clearance of certain helminths,3 and its role in autoimmune responses leading to allergic airway inflammation4 and colitis.5 Interleukin‐9 can be produced by mast cells, eosinophils and type two innate lymphoid cells, but it is primarily produced by T helper (Th) cells.6 These include Th2, Th17 and regulatory T (Treg) cells7, 8, 9, 10; however, the principal source of IL‐9 is thought to be Th9 cells,11, 12 which are a newly identified subset defined by the production of IL‐9 without the cytokines characteristic of other subsets. Th9 cells can be produced in vitro by the activation of naive CD4⁺ T cells in the presence of IL‐4 and transforming growth factor‐β (TGF‐β),11, 12, 13 but their role in vivo has been hard to define. Some of the best evidence for their importance has been demonstrated in mice with a T‐cell‐specific deletion of PU.1, which is a transcription factor required for IL‐9 expression.7 These mice lack Th9 cells, but the development of other Th subsets remains largely unaffected. They are resistant to airway inflammation in a model of asthma7 and also to ulcerative colitis in a model of inflammatory bowel disease.5 Understanding the development of Th9 cells and the regulation of IL‐9 expression could provide new inroads towards clinical approaches to many immune‐related diseases.

Transcriptional regulation plays an important role in IL‐9 expression.14 Signalling through signal transducer and activator of transcription 6 is essential11, 12, 15 and so are the transcription factors PU.1,7, 16, 17 IRF4,16 GATA311 and BATF,18 whereas Id3 inhibits transcription.19 Outside transcriptional regulation, it is not known if other mechanisms are important for IL‐9 expression. MicroRNAs (miRNAs) are one such mechanism. These are double‐stranded RNAs of approximately 23 bp that post‐transcriptionally regulate gene expression by inhibiting translation and inducing message instability.20, 21 Several studies have identified their roles in Th development. First of all, the T‐cell‐specific deletion of Drosha or Dicer (RNases required for miRNA synthesis) results in a propensity towards Th1 development and a significant decrease in the number of Treg cells.22, 23, 24 In addition, multiple individual miRNAs regulate various activities in Th development and function. They are thought to fine‐tune the expression of genes important for the development and maintenance of the stability of Th cells.25 In this study we found that miRNAs play an important role in IL‐9 expression and Th9 development and function, and we used them as a tool to discover the significance of hypoxia‐inducible factor‐2α (HIF‐2α).

Materials and methods

Mouse strains and isolation of naive T cells

C57BL/6 (Charles River, Kent, UK), Rag2^−/− and CD4Cre Dicer ^lox/lox mice were kept in a conventional specific pathogen‐free facility. All the animal work was performed according to the Animals Scientific Procedures Act, UK under the animal Project Licence 70/6965. Naive T cells (CD4⁺ CD62L^high CD25⁻) were isolated from pooled spleen and lymph node cells of 8‐ to 16‐week‐old mice, first using the Dynabeads^® Untouched^™ Mouse CD4 cells kit (Invitrogen, Carlsbad, CA), followed by biotinylated‐anti‐CD25 (7D4; BD Biosciences, Franklin Lakes, NJ), and finally biotinylated‐anti‐CD62L (MEL‐14; BD Biosciences), with the last two steps using streptavidin‐MicroBeads (Miltenyi Biotech, Bergisch Gladbach, Germany). Purity of cells was > 90% as determined by flow cytometry.

Th cell development and analysis

Naive T cells were activated in the presence of plate‐bound anti‐CD3/anti‐CD28 antibodies (eBioscience, San Diego, CA) with 1 μg/ml anti‐CD3 and 2 μg/ml anti‐CD28 for Th0, Th1, Th2 and induced Treg (iTreg) differentiation, or 10 μg/ml anti‐CD28 for Th17 and Th9 differentiation. T cells were differentiated into Th0 using 5 μg/ml anti‐interferon‐γ (IFN‐γ) (BD Biosciences) and 5 μg/ml anti‐IL‐4 (BD Biosciences); Th1 using 20 ng/ml recombinant IL‐12 (eBioscience) and 5 μg/ml anti‐IL‐4; Th2 using 20 ng/ml recombinant IL‐4 (BD Biosciences) and 5 μg/ml anti‐IFN‐γ; Th17 using 2·5 ng/ml recombinant TGF‐β, 50 ng/ml recombinant IL‐6 (eBioscience), 5 μg/ml anti‐IFN‐γ, 5 μg/ml anti‐IL‐4 and 5 μg/ml anti‐IL‐2 (BD Biosciences); Th9 using 2·5 ng/ml recombinant TGF‐β (eBioscience), 40 ng/ml recombinant IL‐4 and 10 μg/ml anti‐IFN‐γ; and iTreg cells using 2·5 ng/ml recombinant TGF‐β and 5 ng/ml recombinant‐IL‐2 (eBioscience). Cells were cultured for 3–4 days before analysis. For intracellular staining of cytokines, cells were treated with 1 μg/ml each of PMA and ionomycin (Sigma, St Louis, MO) for 4 hr and 1 μg/ml of brefeldin A (eBioscience) for 2 hr before staining. For Foxp3 staining, cells were fixed with Foxp3 fixation/permeabilization buffer (eBioscience) for 30 min before staining, and if green fluorescent protein (GFP) was also detected, cells were fixed with 2% paraformaldehyde for 5 min before fixation/permeabilization. Antibodies for flow cytometry experiments were: CD4‐FITC/phycoerythrin (PE)/Peridinin chlorophyll protein/allophycocyanin (GK1.5), CD8α‐Peridinin chlorophyll protein (53‐6.7), CD25‐PE (PC61.5), Foxp3‐allophycocyanin (FJK‐16s), IFN‐γ‐FITC (XMG1.2), IL‐17‐PE and IL‐9‐eFluor^® 660 (RM9A4) (all from eBioscience), and IL‐17a‐PE (TC11‐18H10) and IL‐4‐PE (11B11) (BD Biosciences). Data were acquired using a FACS Canto II or FACS Calibur (BD Bioscience) and analysed using flowjo software (Tree Star, Ashland, OR). Compensation and gate lines for cytokine and Foxp3 staining were set in each experiment using wild‐type cells activated under Th0 polarizing conditions or as stated for the individual experiment.

Retroviral and lentiviral production

The miR‐15b/16 expression and decoy vectors were described previously.26 The expression vector was derived by cloning the genomic region encoding the miRNAs into the retroviral pMIG vector. The decoy vector was designed as described previously27 to express a short RNA containing miR‐15b/16 target sites. The decoy sequence was cloned into the pSIF lentiviral vector for Pol III expression. Retrovirus and lentiviruses were produced by calcium phosphate transfection of human embryonic kidney (HEK) 293T cells with helper virus vectors. Culture supernatants were harvested and used to spin‐infect naive CD4⁺ T cells that were activated overnight by plate‐bound anti‐CD3/anti‐CD28 antibodies. Cells were then differentiated into the different Th subsets as described above.

Adoptive T‐cell‐transfer colitis model

Naive T cells were activated and retrovirally transduced then FACS‐sorted for GFP⁺ cells. These were differentiated under Th9 polarizing conditions, and 10⁶ cells were then injected into the peritoneal cavity of C57BL/6 Rag2 ^−/− mice. Weight was monitored every week, and mice were killed after 8 weeks (or when weight decreased by 20%) for analysis. Histopathological review was performed blindly as previously described.28

Other protocols

Interleukin‐9 was measured in culture supernatants of Th differentiated cells after 48 hr of culture using murine IL‐9 ELISA ready‐SET‐Go! second‐generation kit (eBioscience). For quantitative RT‐PCR, RNA was isolated from CD4⁺ T cells using miRNAeasy (for miRNAs) or mRNAeasy (for mRNAs) isolation kits (QIAGEN, Manchester, UK). For miRNA detection, cDNA synthesis and subsequent quantitative RT‐PCR with locked nucleic acid primers for specific miRNAs was performed using miRCURY LNA^™ universal reverse transcriptase microRNA cDNA synthesis and a quantitative RT‐PCR kit (EXIQON). For mRNA detection, cDNA was prepared using the miRscript cDNA synthesis kit (QIAGEN). The qPCR was performed using IQ^™ BioRad SYBR green master mix (BioRad, Hertfordshire, UK). Primer sequences are available upon request. For Western blotting, rabbit anti‐HIF‐2α (ab179825) (Abcam, Cambridge, UK) was used. For siRNA knockdown of Epas1 expression, naive T cells were transfected with siRNAs using Accell deliver medium (all from Dharmacon, Lafayette, CO) following the manufacturer's protocols. Luciferase assays were performed using the Dual‐Luciferase reporter assay system (Promega, Madison, WI) from extracts prepared from HEK293T cells transfected with a control renilla vector and the firefly reporter containing the 3′ untranslated region of Epas1 downstream of the luciferase gene.

Statistical analysis

prism software (GraphPad, San Diego, CA) or excel was used for statistical analyses to calculate mean and standard deviation values from independent experiments using different pools of isolated T cells or in the case of the adoptive transfer experiments, individual mice. The Student's t‐test was used to calculate significance. P‐values ≤ 0·05 were considered significant.

Results

IL‐9 expression is enhanced in Dicer^−/− T cells

To analyse miRNA control of IL‐9 expression, we used mice with a T‐cell‐specific deletion of Dicer containing a floxed allele of Dicer and a CD4‐Cre transgene.23 As a starting point, IL‐9 FACS staining was optimized using naive Th cells (CD4⁺ CD62L^high CD25⁻) activated under different polarization conditions. Only under Th9 polarizing conditions were substantial numbers of IL‐9‐expressing cells observed (see Supplementary material, Fig. S1). Therefore, we used these staining conditions in subsequent experiments. Next, IL‐9 expression was examined in the CD4⁺ population of spleen and lymph node cells (or isolated CD4⁺ cells) that were activated by PMA and ionomycin and that were from either CD4‐Cre Dicer ^+/lox or CD4‐Cre Dicer ^lox/lox mice. In the absence of miRNAs, there was approximately a five‐fold increase in the percentage of IL‐9‐producing cells on average, whereas consistent with our previous results,23 there was a decrease in the numbers of Foxp3⁺ cells (Fig. 1a). Substantiating this increase in IL‐9‐expressing cells was a 12‐fold increase in Il9 mRNA in Dicer ^−/− ex vivo isolated and unstimulated CD4⁺ CD25⁻ cells (Fig 1a, far right). Since Th9 cells are the main producers of IL‐9,11, 12 we examined the importance of miRNAs on Th9 differentiation in vitro by activating naive Th cells under Th9 polarizing conditions using TGF‐β and IL‐4. This results in reduction in the number of Foxp3⁺ cells that would be induced by TGF‐β alone, and it produces a population of IL‐9⁺ Foxp3⁻ cells classically defined as Th9.12 Dicer ^−/− Th cells had a significant decrease in the percentage of Foxp3⁺ cells and a significant increase in the percentage of IL‐9⁺ Foxp3⁻ cells, (Fig. 1b; and see Supplementary material, Fig. S2). This increase in IL‐9‐expressing cells was also substantiated by an approximately twofold increase in Il9 mRNA in Dicer ^−/− cells (Fig. 1b far right). Therefore, miRNAs appear to suppress Th9 differentiation. Interestingly, Dicer ^−/−, ex vivo isolated Treg cells (CD4⁺ CD25⁺), which are primarily thymus‐derived Treg cells, also had approximately a twofold increase in the percentage of IL‐9‐producing cells and an approximately 3·5‐fold increase in Il9 mRNA (Fig. 1c). Therefore, miRNAs also appear to suppress IL‐9 expression in Treg cells.

Figure 1.

miRNAs suppress IL‐9 expression in Th cells. (a) Spleen and lymph node cells from CD4 Cre Dicer ^+/lox or CD4 Cre Dicer ^lox/lox mice were activated by PMA and ionomycin for 4 hr with brefeldin A. The analysis of Foxp3‐APC and either IL‐9‐PE or IL‐9‐FMO control is shown on the CD4⁺ population of activated (as determined by SSC and FSC) cells. To the right, the mean and standard deviation values are shown from seven independent experiments (using either spleen and lymph node cells or isolated CD4⁺ cells) indicating a significant increase in IL‐9‐expressing cells when Dicer is absent (**P = 0·002). Substantiating this and shown at the far right was an increase in the relative level of Il9 mRNA in ex vivo‐isolated CD4⁺ CD25⁻ cells (as measured by quantitative RT‐PCR and normalized to GAPDH) (*P = 0·02 from five independent experiments). (b) Naive CD4⁺ T cells were activated under Th9‐polarization conditions, and IL‐9 and Foxp3 expression was analysed by FACS as in (a). There was a significant increase in IL‐9‐expressing cells when miRNAs were absent (*P = 0·02 from four independent experiments). Likewise, there was a significant increase in the relative level of Il9 mRNA (*P = 0·02 from five independent experiments). (c) Isolated regulatory T (Treg) cells (CD4⁺ CD25⁺) also displayed an increase in IL‐9 expression in Dicer ^−/− cells (*P = 0·01 for FACS and *P = 0·04 for Il9 mRNA from three or five independent experiments, respectively).

To examine if miRNAs suppress IL‐9 production in other Th subsets, naive Th cells were activated under different polarization conditions, and the level of IL‐9 production was measured in culture supernatants (Fig. 2a). Dicer ^−/− Th cells had an increase in IL‐9 production under all polarizing conditions tested, particularly in Th2 and iTreg conditions where the increase was to the same level found in Th9 cells. Therefore, miRNAs play a general role in suppressing IL‐9 expression in all Th polarization conditions, and they are most important in Th2 and iTreg conditions. To determine if miRNAs suppress the expression of IL‐9 within each subset or alternatively prevent the differentiation towards a Th9‐like phenotype, we examined IL‐9 expression in combination with a signature marker of a specific Th subset (IFN‐γ for Th1, IL‐4 for Th2, IL‐17 for Th17 and Foxp3 for Treg cells). Consistent with previous reports,22, 23, 29 loss of miRNAs resulted in increased production of IFN‐γ in Th0 and Th1 conditions and increased IL‐4 in Th2 conditions, whereas there was a slight loss of IL‐17 expression under Th17 conditions and a significant decrease of Foxp3 expression in induced Treg (iTreg) cells (Fig. 2b). The increase in IL‐9 expression in all subsets but iTreg cells occurred in both populations of cells expressing or not expressing the signature cytokine. Therefore, miRNAs appeared to be not only important regulators of IL‐9 expression within these subsets but also important for inhibiting differentiation into a Th9‐like phenotype. In contrast, the increase in IL‐9 expression in iTreg induction occurred primarily in cells not expressing Foxp3. Therefore, miRNAs appeared to be solely important for the lineage choice in these conditions but not for the regulation of IL‐9 expression in cells with an iTreg phenotype, which was unlike ex vivo‐isolated Treg cells that expressed more IL‐9 in the absence of miRNAs.

Figure 2.

miRNAs suppress IL‐9 expression in all Th subsets. (a) Naive CD4⁺ cells containing (Dicer ^+/−) or lacking (Dicer ^−/−) miRNAs were activated under the indicated polarization conditions for 48 hr, and the level of IL‐9 was determined by ELISA from culture supernatants. Values are from three independent experiments. Cells lacking miRNAs expressed significantly more IL‐9: Th0 (***P = 0·0003), Th1 (*P = 0·03), Th2 (******P = 0·0000008), Th9 (*P = 0·05), Th17 (**P = 0·002) and iTreg cells (***P = 0·0003). (b) Representative profiles from three or four independent experiments showing expression of signature cytokines (or Foxp3 for iTreg cells) and IL‐9 for indicated polarization conditions.

miR‐15b/16 suppresses IL‐9 production in Th9 cells

To begin to understand miRNA regulation of IL‐9 expression, we first wanted to identify individual miRNAs that could function in suppression. Subsequently, relevant targets could be determined whose regulation impacted IL‐9 expression. Because loss of miRNAs had an opposite effect on the differentiation of iTreg cells and Th9 cells, we decided to test the function of miRNAs that we previously found to be important for iTreg induction26 and determine if they could additionally suppress Th9 differentiation. Two of these miRNAs were miR‐15b and miR‐16. They are highly related miRNAs encoded on the same primary transcript that target the same messages. They enhance iTreg induction through their suppression of mTOR and Rictor, which inhibits mTOR signalling and directs cells into a Treg differentiation pathway. We previously found that miR‐15b and miR‐16 were more abundantly expressed in iTreg cells than in naive T cells or those activated under Th0, Th1, Th2 or Th17 polarizing conditions, which was consistent with their importance in iTreg induction.26 Here we show those data again and add the miR‐15b and miR‐16 expression levels in Th9 cells, which were also much lower than in iTreg cells (Fig. 3a). Therefore, miR‐15b/16 might regulate IL‐9 expression in iTreg cells, but they would be unlikely to be important in other subsets. Nevertheless, to determine if miR‐15b/16 were capable of suppressing IL‐9 expression, we examined if their over‐expression could inhibit IL‐9 expression in Th9 polarization conditions. Over‐expression was achieved using a GFP‐expressing retrovirus that contained the genomic sequence encoding both miRNAs. This or a control, empty retrovirus was transduced into naive CD4⁺ T cells after their activation, which gave on average twice the expression level of these miRNAs in iTreg cells (see Supplementary material, Fig. S3). Cells were subsequently activated under Th9 polarizing conditions, and IL‐9 expression was examined in transduced, GFP⁺ cells. Over‐expression of miR‐15b/16 resulted in a significant reduction in the percentage of IL‐9‐producing cells and the level of IL‐9 production (Fig. 3b,c). In contrast, miR‐15b/16 over‐expression did not have an effect on the low numbers of cells expressing IL‐9 in Th1, Th2, Th17 or iTreg polarizing conditions (Fig. 3d–g). Likewise, miR‐15b/16 over‐expression did not affect the development of Th1, Th2 or Th17 cells as measured by the expression of signature cytokines, but consistent with our previous findings, it did enhance iTreg induction.26 Therefore, miR‐15b/16 appear to be capable of regulating Th9 and iTreg differentiation but not other subsets.

Figure 3.

Over‐expression of miRNAs miR‐15b and miR‐16 suppresses the expression of IL‐9 during Th9 polarization but has no effect on Th1, Th2 or Th17 polarization. (a) Relative miRNA levels of miR‐15b and miR‐16 normalized to 5S rRNA (as determined by quantitative RT‐PCR) from naive CD4⁺ T cells activated under the indicated polarization conditions. Th9 levels are added for comparison to data from our previous report26 of levels in other subsets. (b–g) T cells transduced with a control or miR‐15b/16 over‐expressing retrovirus were activated under the indicated polarization conditions. T helper differentiation in the GFP ⁺ cell population was measured by the percentage of cells expressing the signature cytokine (b, d–f) or Foxp3 for iTreg cells (g). A representative experiment is on the left, and the mean and standard deviation values are on the right for four independent experiments. Only in Th9 polarized cells was the percentage of IL‐9‐expressing cells significantly different (*P = 0·02). (c) mR‐15b/16 over‐expressing cells produce less IL‐9 during Th9 polarization, as measured by ELISA from culture supernatants. Data are from three independent experiments (**P = 0·001).

miR‐15b/16 inhibits the inflammatory response of Th9 cells in vivo

Because Th9 cells are important in autoimmunity, we examined if miR‐15b/16 could inhibit their inflammatory response. In vitro‐derived Th9 cells will produce an autoimmune response in the colon when they are adoptively transferred into Rag2 ^−/− mice.11 Therefore, we tested if miR‐15b/16 over‐expression could inhibit this activity. Naive CD4⁺ CD25⁻ T cells were activated then transduced with control or miR‐15b/16‐expressing retroviruses. GFP⁺ cells were sorted then further activated under Th9 polarizing conditions before their adoptive transfer into Rag2 ^−/− mice. miR‐15b/16 over‐expression significantly inhibited weight loss (Fig. 3a) and reduced the resultant colitis as measured by colon thickness (Fig. 4b). Analysis of GFP⁺ cells recovered from the spleen revealed that miR‐15b/16 over‐expression reduced the number of IL‐9‐producing CD4⁺ cells (Fig. 4c). However, miR‐15b/16 over‐expression also resulted in an increase in the number of Treg cells (Fig. 4d) so it was uncertain if the reduced inflammatory response was due to the reduction in Th9 cells or an increase in Treg cells. Nevertheless, the effects of miR‐15b/16 over‐expression on IL‐9 expression in Th9 polarization observed in vitro were maintained in vivo over the time course of this experiment.

Figure 4.

Over‐expression of the miRNAs miR‐15b and miR‐16 suppresses the inflammatory response of Th9 cells. (a, b) Naive CD4⁺ cells were activated then transduced with control or miR‐15b/16‐expressing retroviruses. GFP ⁺ cells were sorted then differentiated under Th9 polarization conditions and adoptively transferred into Rag2 ^−/− mice as described in the Materials and methods. Weight loss (a) was significantly inhibited when Th9 differentiated cells over‐expressed miR‐15b/16 (**P = 0·002), as was the resultant colitis (b) as measured by colon thickness (*P = 0·03). (c, d) Spleen cells were isolated from the above mice and activated by anti‐CD3 and anti‐CD28 then GFP ⁺ cells (which derived from the original adoptively transferred cells) were analysed for expression of IL‐9 (c) or Foxp3 (d) in combination with CD4. A representative experiment is shown on the left and the mean and standard deviation values on the right, which show a significant decrease in IL‐9‐expressing cells (**P = 0·005) and a significant increase in Foxp3‐expressing cells (**P = 0·006). Data are derived from four mice per group.

Epas1, encoding the hypoxia transcription factor HIF‐2α, can be regulated by miR‐15b/16

Because miR‐15b/16 could regulate the expression of IL‐9, and because these miRNAs are highly expressed only in Treg cells, we hypothesized that miR‐15b/16 might be important for regulating IL‐9 expression in Treg cells. Therefore, we searched for potential target genes using target prediction algorithms.30 Because true target genes would most likely be more highly expressed in Dicer ^−/− Treg cells, potential targets of miR‐15b/16 were manually examined for those that were more highly expressed in Dicer ^−/− compared with Dicer ^+/+ Treg cells using gene expression array data.31 Epas1 was one gene found in this analysis, which encodes HIF‐2α. HIF‐2α or its related gene product, HIF‐1α, associates with HIF‐1β (also known as the aryl hydrocarbon receptor nuclear translocator, ARNT). These heterodimers act as transcription factors to regulate gene expression for cell survival in low‐oxygen environments.32 HIFs play important roles in cancer biology because of the hypoxic conditions that develop as tumours grow.33 Originally, the HIF‐α subunits were thought to be interchangeable, but recent evidence has found differences in their function.33 HIF‐1α plays a complicated role in Th development in that it has been found to be important for the development of Th17 cells over Treg cells,34, 35 but it is also important for Treg function.36 In contrast, virtually nothing is known about the function of HIF‐2α in Th development. Therefore, its regulation by miRNAs was examined.

Epas1 RNA was abundantly expressed in Th9‐polarized cells, which was consistent with a role in IL‐9 expression. However, it was also abundant in Th2‐polarized cells (Fig. 5a), but whereas the highest RNA level was found in Th2‐polarized cells, the highest protein level was found in Th9‐polarized cells (Fig. 5c). In either case, it was less expressed in Treg cells and other Th subsets. Therefore, its abundant expression in Th9 cells was consistent with a role in IL‐9 expression. However, its expression in Th2 cells indicated that it most likely would not be sufficient.

Figure 5.

HIF‐2α is regulated by over‐expression of the miRNAs miR‐15b and miR‐16. (a) Epas1 mRNA levels normalized to GAPDH (as measured by quantitative RT‐PCR) and relative to naive T cells are highest in T cells differentiated under Th2 and Th9 conditions. Thymus‐derived regulatory T (tTreg) cells are ex vivo isolated Treg cells. Data are from four independent experiments. (b) Dicer ^−/− T cells have significantly increased Epas1 mRNA levels in both conventional T cells (CD25⁻) (**P = 0·002) and Treg cells (CD25⁺) (**P = 0·003). Data are from four independent experiments. (c) HIF‐2α levels are highest in Th9 and Th2 differentiated cells, and levels increase in all helper subsets when miRNAs are absent. On the left is a representative Western blot, and on the right are the values normalized to GAPDH and relative to Dicer ^+/− Th0 cells as determined by densitometry from four independent experiments. (d) Epas1 contains an miR‐15b/16 target site in its 3′ untranslated region (UTR). The sequence of the region containing the miR‐15b/16 target site in a luciferase reporter vector with the Epas1 3′ UTR is shown as well as the sequence of a mutated reporter. Below is the sequence of miR‐15b showing the homology through the seed sequence. When the wild‐type reporter was transfected into HEK 293T cells, it was significantly repressed by the co‐expression of miR‐15b/16 (****P = 0·00003), and suppression was significantly reduced with the reporter containing the mutated miR‐15b/16 target site (**P = 0·002). (e) Over‐expression of miR‐15b/16 suppressed Epas1 mRNA and HIF‐2α protein expression in Th9 differentiated cells. mRNA levels are derived from three independent experiments (*P = 0·04), and the Western blot is representative of two independent experiments with the level of HIF‐2α relative to GAPDH in this blot (as determined by densitometry) shown between.

The expression of Epas1 was regulated by miRNAs, as there was a significant increase in mRNA levels in ex vivo‐isolated CD4⁺ conventional T cells (CD25⁻) and Treg cells (CD25⁺) that lacked miRNAs (Fig. 5b). Likewise, HIF‐2α protein levels were increased in Dicer ^−/− T cells that were activated under all polarization conditions tested (Fig. 5c). Therefore, expression of HIF‐2α could be regulated by miR‐15b/16 in Treg cells, but other miRNAs would most likely be important in other subsets because, as stated above, miR‐15b/16 are only abundantly expressed in Treg cells. Nevertheless, we examined regulation of HIF‐2α expression by miR‐15b/16. The activity of a luciferase reporter gene containing the Epas1 3′ untranslated region was suppressed by the co‐expression of miR‐15b/16, and this suppression was significantly reduced if the predicted miR‐15b/16 target site was mutated (Fig. 5d). In addition, expression of Epas1 RNA and HIF‐2α protein was suppressed in Th9‐polarized cells by the over‐expression of miR‐15b/16 (Fig. 5e). Therefore, miR‐15b/16 can regulate the expression of Epas1.

HIF‐2α regulates the development of Th9 cells and iTreg cells

To test the importance of HIF‐2α in the development of Th9 cells, the effect of altering its expression was analysed. Transfection of an siRNA against Epas1 reduced the percentage of IL‐9‐producing cells when activated under Th9 polarizing conditions, whereas it enhanced the production of Foxp3⁺ cells (Fig. 6a). The observed decrease in the numbers of IL‐9‐expressing cells was substantiated by the reduced production of IL‐9 in culture supernatants from these cultures (Fig. 6a). In contrast to the siRNA knockdown, over‐expression of HIF‐2α using a retrovirus containing the Epas1 cDNA, enhanced the percentage of IL‐9‐producing cells and decreased the production of Foxp3⁺ cells (Fig. 6b). Over‐expression of Epas1 also inhibited the induction of iTreg cells in iTreg conditions, but as in the Foxp3⁺ cells in Th9 polarization conditions, it did not increase IL‐9 expression. Therefore, expression of HIF‐2α is also not sufficient for IL‐9 expression in iTreg cells. Other regulatory factors specific to Th9 cells must also be required. To test the importance of miR‐15b/16 suppression of HIF‐2α expression in Treg cells, naive T cells were activated and transduced with a lentivirus expressing an miR‐16 decoy that acted as a competitive inhibitor of miR‐15b/16. These cells were then differentiated under iTreg conditions. Inhibiting the high levels of miR‐16 in Treg cells increased the expression of Epas1 mRNA and HIF‐2α protein (Fig 6d), and consistent with our previous results26 it inhibited iTreg induction, but ironically it did not increase IL‐9 expression (Fig. 6e). Therefore, the loss of miR‐15b/16 expression in Dicer ^−/− iTreg cells is not sufficient for the developmental changes that lead to some cells with a Th9 like phenotype. Other miRNAs must be important in iTreg cells and also for regulation of IL‐9 expression in other subsets. Nevertheless, miR‐15b/16 appears to be one means of suppressing the expression of HIF‐2α in iTreg cells and preventing its inhibitory effects on iTreg induction.

Figure 6.

HIF‐2α enhances Th9 development and inhibits iTreg cell induction. Naive CD4⁺ T cells were activated and transfected with control or Epas1 small interfering (si) RNAs (a) or transduced with control or Epas1‐expressing retroviruses (b, c) or with a control or an miR‐16‐decoy‐expressing lentivirus (d, e). Cells were then activated under Th9 polarization (a, b) or iTreg induction (c–e) conditions. (a) An Epas1 siRNA inhibits Th9 differentiation (*P = 0·02) and also increases the number of Foxp3⁺ IL‐9⁻ cells in these conditions. It also inhibits the production of IL‐9 in culture supernatants as measured by ELISA (*P = 0·02). The levels of HIF‐2α as measured in a Western blot are displayed on the far right. (b, c) Over‐expression of Epas1 enhances Th9 differentiation (*P = 0·01) and inhibits the induction of iTreg cells (**P = 0·004). (d) Expression of miR‐16 decoy during the induction of iTreg cells enhances the expression of Epas1 (as measured by quantitative RT‐PCR and normalized to GAPDH *P = 0·02) and also HIF‐2α (as measured by Western blot and also normalized to GAPDH). (e) Despite the inhibition of iTreg induction by the miR‐16 decoy (*P = 0·01), it does not increase IL‐9 expression in iTreg cells. For FACS data, a representative experiment is shown with the mean and standard deviation values from four to six experiments. The ELISA and Epas1 mRNA data are derived from four or three independent experiments respectively, and the HIF‐2α Western blots are representative of two independent experiments with similar results. The relative levels of HIF‐2α normalized to GAPDH for these blots as determined by densitometry is shown.

Discussion

In this report we have demonstrated that miRNAs regulate the expression of IL‐9 in Th cells. Two highly related miRNAs, miR‐15b and miR‐16, were found capable of suppressing IL‐9 expression when over‐expressed in Th9 cells. These miRNAs were able to regulate the hypoxia transcription factor gene Epas1 (encoding HIF‐2α), which was found to be important for IL‐9 expression in Th9 cells.

This study demonstrated the utility of examining miRNA function to identify new regulatory pathways in Th development and function. However, the irony is that the miRNAs analysed (miR‐15b/16) are probably not directly relevant for regulating IL‐9 expression in vivo because (i) they could only suppress IL‐9 expression when they were over‐expressed in Th9 cells; (ii) they are not abundantly expressed in Th0, Th1, Th2 and Th17 cells, in which IL‐9 expression is low, so they are unlikely to be important regulators in these subsets; and (iii) in iTreg cells where miR‐15b/16 are abundantly expressed, blocking their function did not induce IL‐9 expression. Therefore, other miRNAs must be responsible for the increased IL‐9 expression observed in Dicer ^−/− T cells. Nevertheless, miR‐15b/16 regulation of HIF‐2α expression in iTreg cells appears to be one mechanism of suppressing the expression of HIF‐2α and preventing its inhibitory function on iTreg induction. Therefore, HIF‐2α appears to be an additional target of miR‐15b/16 in iTreg cells that explains the importance of these two miRNAs in iTreg induction. All these findings are summarized in the diagram in Fig. 7.

Figure 7.

Summary diagram of miRNA and HIF‐2α regulation of Th9 differentiation and IL‐9 expression. The miRNAs miR‐15b/16 play important regulatory roles in iTreg cells. Our previous work demonstrated that these miRNAs enhance iTreg induction through suppressing mTOR signalling. In this work we have found that they also enhance iTreg induction through suppressing HIF‐2α expression, which like HIF‐1α inhibits iTreg induction. In contrast, HIF‐2α is important for Th9 differentiation but not sufficient for IL‐9 expression in other subsets. Finally, miR‐15b/16 do not appear to be the important miRNAs for regulating IL‐9 expression in all Th subsets. Therefore, as yet undefined miRNAs must be involved in suppressing Th9 differentiation and IL‐9 expression in all Th subsets.

The suppression of iTreg induction by HIF‐2α and its importance for the expression of IL‐9 in Th9 cells adds new information to the function of the HIF proteins in Th development. As mentioned above, HIF‐1α plays a complicated role in Th development with many questions regarding its functions.37 HIF‐1α is reported to enhance Th17 over Treg development,34, 35 but it is also required for Treg function.36 In Th17 cells, HIF‐1α associates with the transcription factor RORγt to regulate the transcription of genes important for Th17 development. It also inhibits Treg development through its association with Foxp3, which targets it for degradation through the proteasome.35 However, hypoxia also directly activates Foxp3 expression and the induction of Treg cells.36 Therefore, the details of HIF‐1α function need further understanding. HIF‐2α appears to regulate the differentiation between Th9 and Treg cells analogous to HIF‐1α in the Th17–Treg axis. However, the mechanisms of HIF‐2α activating IL‐9 expression and suppressing Treg induction will need to be explored. With IL‐9 expression, clearly HIF‐2α is not sufficient by itself because its relevantly abundant expression in Th2 cells or its over‐expression in iTreg cells does not lead to high expression of IL‐9. Other factors specific to Th9 cells must also be required. Finally, both HIF‐1α and HIF‐2α are regulated by oxygen levels through hydroxylation of key proline residues, which leads to their degradation by the proteasome. In T cells HIF‐1α is also stabilized under normal oxygen conditions by TCR stimulation or IL‐6 signalling through signal transducer and activator of transcription 3.35, 38 Therefore, regulation of HIF protein expression plays an important role in the diverse environments encountered by T cells in vivo.

Regulation of IL‐9 expression is controlled by many mechanisms, and this work has illustrated the importance of miRNAs and HIF‐2α. Therefore, it will be important to understand how all these combine to regulate IL‐9 expression in an immune response.

Author contributions

YS and BC planned the experiments, and YG performed the vast majority. BC constructed some of the recombinant vectors and OG scored the pathology of the adoptively transferred mice. FL provided financial support. YS and BC wrote the manuscript with editing by OG and FL.

Disclosures

The authors have no conflicting interest in the publication of this work.

Supporting information

Figure S1. Verification of IL‐9 detection in FACS staining.

Figure S2. Live/dead and unstimulated cell controls for IL‐9 detection in Th type 9‐differentiated T cells.

Figure S3. Expression levels of miRNAs miR‐15b and miR‐16 in retrovirally transduced T cells relative to induced regulatory T cells.