Demethylation of the RORC2 and IL17A in Human CD4⁺ T Lymphocytes Defines Th17 Origin of Nonclassic Th1 Cells

Activated CD4⁺ T cells can be divided into functionally distinct subsets on the basis of their cytokine production and their specific transcription factor expression: Th1 secrete IFN-γ, express the transcription factor T-bet (T-box expressed in T cells, TBX21), and protect the host against intracellular infections; Th2 cells secrete IL-4, IL-5, and IL-13, express the transcription factor GATA-3, and mediate host defense against helminths; and Th17 cells, which selectively produce IL-17A, express the transcription factor retinoic acid orphan receptor (ROR)C2 and are critical for the host defense against extracellular pathogens (1–4). Moreover, a subset of human IL-17A–producing CD4⁺ T cells was found also to produce IFN-γ (Th17/Th1), and both Th17 and Th17/Th1 cells exhibited plasticity toward the Th1 profile when cultured in the presence of IL-12 (4). Similar findings were then also reported in some murine models (5–7), suggesting that both human and murine Th17 cells probably represent a transient phenotype (8). In a previous study, we also showed that virtually all human memory Th17 cells are contained within the CD161⁺ fraction of both circulating and tissue-infiltrating CD4⁺ T cells, and originate from CD161⁺ precursors present in umbilical cord blood (UCB) and newborn thymus (9), a finding that has not been reported in mice. In humans, Th1 cells that derive from the shifting of Th17 cells were defined as nonclassic Th1 cells (8, 10) because they express RORC2, CD161, and CCR6, as do Th17 cells (9–11).

Tissue- and cell-specific transcriptional regulatory programs are acquired during various stages of cellular differentiation and must persist into the cellular progeny to maintain the specialized functions of the cells in a particular tissue. Covalent modifications to histones and DNA, referred to as epigenetic modifications, are used to regulate transcription factor and polymerase access to transcriptional regulatory elements in chromatin. Epigenetic modification of chromatin permits the cells to retain acquired transcriptional regulation throughout cell division, providing a cellular identity to the progeny (12, 13). In addition, the development of distinct Th cell lineages has been shown to involve heritable epigenetic changes that regulate the expression of both lineage-specific transcription factors (including TBX21, RORC2, and GATA3) and cytokines (including IFN-γ, IL-17A, and IL-4) (14–16).

In this study, we examined the methylation patterns of regions of interest (ROI), identified by sequence homology analysis between mice and humans, of the human genes RORC2 (17), TBX21, IL17A (17), IL17F (18), and IFNG (19) in either human T cell clones or circulating freshly isolated CD4⁺ T cells, isolated from both peripheral blood (PB) of healthy subjects and synovial fluid (SF) of juvenile idiopathic arthritis (JIA) patients, characterized by the following phenotypes: Th17, Th17/Th1, nonclassic (Th17-derived), and classic Th1. We further analyzed the methylation changes of the above-mentioned ROI in Th17 clones during the in vitro polarization of Th17 into the nonclassic Th1 phenotype. Our results demonstrate the existence of a strict relationship between Th17 and nonclassic Th1 cells, and support the concept that nonclassic and classic Th1 cells belong to two distinct cell lineages. These findings also identify the demethylated status of RORC2 and IL17A ROI as useful markers of the “Th17 membership.” Finally, the study of RORC2 downstream ROI methylation status, as well as the PCR analysis at the single-cell level of the CD4⁺CD161⁺ T cell subset that contains all Th17 precursors (9), demonstrated the heterogeneity of this cell population, only partially composed of cells expressing RORC2, that is, potentially able to differentiate into Th17 cells. Altogether our data provide other important pieces in the puzzle of human Th17 cell physiology and heterogeneity.

The PB samples were obtained from seven healthy donors. SF samples were collected from four JIA patients during routine therapeutic arthrocentesis. UCB samples were obtained from eight donors. The procedures followed in this study were in accordance with the ethical standards of the Regional Committee on Human Experimentation.

The medium used was RPMI 1640 (Seromed, Berlin, Germany), supplemented with 2 mM l-glutamine, 1% nonessential amino acids, 1% pyruvate, 2 × 10-5 M 2-ME (Life Technologies Laboratories, Grand Island, NY), and 10% FCS (Euroclone). The following unlabeled or fluorochrome-conjugated mouse anti-human mAbs were used to stain cells: CD3, CD4, CD8, CD161, CCR6, CD45RA, CD45RO, CD31, IFN-γ, and isotype-matched control Abs (BD Biosciences, Mountain View, CA), IL-17A (eBioscience, San Diego, CA), and CD161 (Miltenyi Biotec, Bergisch Gladbach, Germany). PMA, ionomycin, brefeldin A, and saponin were from Sigma-Aldrich (St. Louis, MO). IL-1β, IL-6, IL-12, IL-23, and TGF-β were purchased from R&D Systems.

Mononuclear cell (MNC) suspensions were obtained from PB, SF, and UCB by centrifugation on Ficoll-Hypaque gradient. PB CD4⁺ T cells were negatively selected by high-gradient magnetic cell sorting (Miltenyi Biotec), as described by Cosmi et al. (9) and were then further divided into CD161⁺ and CD161⁻ by immunomagnetic cell sorting (Miltenyi Biotec). PB CD4⁺CD161⁺ and CD4⁺CD161⁻ T cell populations were further subdivided in CCR6⁺ and CCR6⁻ cell fractions by FACSAria (BD Biosciences) and then cultured under limiting dilution to obtain Th17 and Th1 clones, classified by flow cytometry for intracellular staining of IL-17A or IFN-γ cytokines (4). We stimulated in vitro four selected pure Th17 cell clones in the presence of allogeneic feeder cells, PHA (0.001% v/v) plus IL-2 (50 IU/ml), and the Th1-polarizing cytokine IL-12 (2 ng/ml). After 2 wk of in vitro activation, a significant proportion of cells started to produce IFN-γ. The Th17/Th1-polarized Th17 clones were then sorted by FACSAria into CD4⁺IL-17A⁺IFN-γ⁻, CD4⁺IL-17A⁺IFN-γ⁺, CD4⁺IL-17A⁻IFN-γ⁺ T cell subpopulations, using the cytokine secretion assay, as described below. The CD4⁺IL-17A⁻IFN-γ⁺ cell fraction of three of these polarized clones was then cultured again for 2 wk in the presence of IL-2 (50 IU/ml) alone, or with the addition of IL-1β (10 ng/ml) and IL-23 (20 ng/ml) or with IL-1β, IL-6 (2 ng/ml), IL-23, and TGF-β (5 ng/ml). PB CD4⁺ naive T cells were obtained as CD45RO⁻CD45RA⁺CD31⁺ by FACSAria. UCB CD4⁺ T cells, negatively selected by high-gradient magnetic cell sorting (Miltenyi Biotec), were then pooled from eight samples and further divided into CD161⁺ and CD161⁻ T cell fractions by FACSAria. The same sorting strategy was followed to obtain CD4⁺CD161⁺ and CD4⁺CD161⁻ single cells from two independent UCB samples.

PB and SF CD4⁺ were negatively selected by high-gradient magnetic cell sorting and further divided into CCR6⁺ and CCR6⁻ cells by immunomagnetic cell sorting (Miltenyi Biotec), then were stained with anti-CD161 PE-Cy7, stimulated with PMA–ionomycin, recovered after 3.5 h, washed, and then stained with IFN-γ and IL-17A catch reagents (Miltenyi Biotec), according to the manufacturer’s instructions. Following an additional 45 min of incubation (37°C, 5% CO₂), cells were stained with anti-CD3–Pacific Blue, -IL-17A–APC, and -IFN-γ–FITC; analyzed; and sorted by FACSAria. PB samples were divided into CD4⁺CCR6⁺CD161⁺IL-17A⁺IFN-γ⁻, CD4⁺CCR6⁺CD161⁺IL-17A⁺IFN-γ⁺, CD4⁺CCR6⁺CD161⁺IL-17A⁻IFN-γ⁺, and CD4⁺CCR6⁻CD161⁻IL-17A⁻IFN-γ⁺ T cell subsets, whereas SF samples were divided into CD4⁺CCR6⁺CD161⁺IL-17A⁻IFN-γ⁺ and CD4⁺CCR6⁻CD161⁻IL-17A⁻IFN-γ⁺ T cell subsets. Because of the low numbers of recovered cells, SF cell fractions were then expanded in vitro in the presence of IL-2 (50 IU/ml).

Genomic DNA was extracted and subsequently bisulfite was converted using the DNA Blood Mini Kit (QIAGEN, Hilden, Germany) and EZ DNA Methylation-Gold Kit (Zymo Research, Irvine, CA), respectively, according to the manufacturer’s instructions. DNA methylation of the ROI was assessed by pyrosequencing of bisulfite-specific PCR products using standard procedures (Varionostic, Ulm, Germany), as described previously by Dong et al. (19). Briefly, 40 ng bisulfite-converted DNA was PCR amplified for the ROI of the IL17A, IL17F, IFNG, RORC2, and TBX21 genes using bisulfite-DNA–specific primers (Table I) in a PCR using the following program: 95°C for 3 min; 50 cycles of 95°C for 35 s, annealing temperature (Table I) for 35 s, and 72°C for 40 s; and a final extension at 72°C for 10 min. Bisulfite-PCR products were then subjected to pyrosequencing with the primers listed in Table I. Methylation levels at CpG sites were obtained using Pyro Q-CpG software.

Total RNA was extracted using the RNeasy Micro Kit (QIAGEN) and treated with DNase I to eliminate possible genomic DNA contamination. TaqMan RT-PCR was performed as described by Santarlasci et al. (20). Primers and probes used were purchased from Life Technologies. Quantification was performed on cell number.

Single cells were sorted by FACSAria and then processed using a Single Cell to CT Kit (Life Technologies), according to the manufacturer’s instructions. Briefly, single cells were lysed in the presence of DNase I prior to RNA reverse transcription. cDNA was then subjected to a preamplification step, using a panel of Taqman Gene Expression Assays (Life Technologies) of our interest. Preamplified cDNA was then quantitated by real-time PCR using Taqman Gene Expression Master Mix and Taqman Gene Expression Assays (Life Technologies) on a 7900HT Instrument (Life Technologies).

The Mann–Whitney U test was used for DNA methylation analysis; the paired Student t test was used for mRNA expression analysis. A p value ≤ 0.05 was considered significant.

We have recently shown that human Th17 cells can be shifted in vitro into IFN-γ–producing cells (that we named as nonclassic Th1 cells) in the presence of IL-12 and/or TNF-α (4, 21, 22). We have also found increased numbers of nonclassic Th1 cells in the gut of patients with Crohn’s disease, the skin of psoriatic patients, and the SF of patients suffering from JIA (9, 21, 23). Human nonclassic Th1 can be distinguished from classic Th1 cells because of the expression of Th17 cell–associated markers, such as CD161, CCR6, and RORC2 (10).

To define the epigenetic signatures of pure CD4⁺ Th cell subsets, we decided to analyze DNA methylation of TBX21, RORC2, IFNG, IL17A, and IL17F gene loci (Table I), selected on the basis of evolutionary conservation among species and CpG site density in noncoding sequences (promoter and CNS) (24), in Th17, Th17/Th1, classic Th1, and nonclassic Th1 (Th17 derived), clones (Figs. 1, 2). As shown in Fig. 1A, we selected two ROI in the promoter of the IL17A and IL17F genes, respectively. The analysis of DNA methylation status of the two ROI by bisulfite pyrosequencing showed that Th17 and Th17/Th1 cells had a marked demethylation, and nonclassic Th1 cells exhibited a significant hypomethylation. In contrast, classic Th1 cells, as well as naive CD4⁺ Th cells, were characterized by a complete methylation of the above-mentioned gene loci (Fig. 1A). As shown in Fig. 1B, two ROI in the IFNG gene locus were selected, the promoter and CNS-1 (located 4.2 kb upstream of the transcriptional start site) (19). The analysis of DNA methylation status of the two ROI by bisulfite pyrosequencing showed that whereas Th17/Th1, nonclassic Th1, and classic Th1 cells had a marked demethylation, Th17 cells and naive CD4⁺ Th cells were characterized by a complete methylation (Fig. 1B). We also looked at the transcription factors involved in Th17 and Th1 development, and to this end, we analyzed one ROI for RORC2 (RORC2 downstream (17)) and one for TBX21 (CNS-1). The analysis of DNA methylation status of the RORC2 downstream showed that whereas Th17, Th17/Th1, and nonclassic Th1 cells had a marked demethylation, naive CD4⁺ Th cells and classic Th1 cells were characterized by a complete methylation of this DNA region (Fig. 2A). Of interest, the analysis of DNA methylation status of the TBX21 CNS-1 (Fig. 2B) showed that whereas nonclassic and classic Th1 cells had a marked demethylation, Th17 and Th17/Th1 cells were hypermethylated, compared with a complete methylation in naive CD4⁺ Th cells. These findings strongly support the concept that nonclassic Th1 cells originate from Th17 cells.

To provide additional evidence for this possibility, we decided to stimulate in vitro four Th17 clones in the presence of allogeneic feeder cells and IL-12. As expected on the basis of the results of our previous study (4), after 2 wk of activation, a significant proportion of clonal cells started to produce IFN-γ (Fig. 3). IL-17A⁺IFN-γ⁻, IL-17A⁺IFN-γ⁺, and IL-17A⁻IFN-γ⁺ cell fractions were then obtained by cell sorting, and DNA methylation of TBX21, RORC2, IFNG, IL17A, and IL17F gene loci was analyzed. As indicated in Fig. 3B and 3D, all three cell fractions showed, independently of their ability to produce IL-17A, complete demethylation of IL17A and RORC2, and hypomethylation of IL17F ROI, demonstrating an epigenetic imprinting of Th17 origin in these cells. Of note, only IFN-γ–producing cell subsets (IL-17A⁺IFN-γ⁺ and IL-17A⁻IFN-γ⁺ cells) were characterized by a significant hypomethylation of the IFNG promoter, when compared with cells producing IL-17A alone (Fig. 3C). Moreover, all three cell subsets also showed complete demethylation of the ROI in TBX21 (Fig. 3E).

To understand whether the acquisition of the nonclassic Th1 phenotype by Th17 cells was only a transient or a stable phenomenon, IL-17A⁻IFN-γ⁺ cell fractions were FACS sorted from in vitro polarized Th17 clones and then cultured for an additional 2 wk under neutral (medium alone) or Th17-polarizing (IL-1β plus IL-23 or IL-1β plus IL-6 plus IL-23 plus TGF-β) conditions. Irrespective of the different culture conditions, the totality of the cells was able to produce IFN-γ, as assessed at the single-cell level upon PMA plus ionomycin stimulation (data not shown). Accordingly, as shown in Fig. 3F, methylation analysis of the IFNG promoter and CNS-1 of cells cultured under Th17 polarizing conditions clearly showed stability of the demethylated status.

To confirm these findings obtained on T cell clones, we extended our analysis in different ex vivo CD4⁺ Th subsets isolated from PB by separation procedures based on both the expression of surface markers and the ability of cells to produce different cytokines. Th17 cells were obtained as CD4⁺CD161⁺CCR6⁺IL-17A⁺IFN-γ⁻, Th17/Th1 cells as CD4⁺CD161⁺CCR6⁺IL-17A⁺IFN-γ⁺, nonclassic Th1 cells as CD4⁺CD161⁺CCR6⁺IL-17A⁻IFN-γ⁺, and classic Th1 cells as CD4⁺CD161⁻CCR6⁻IL-17A⁻IFN-γ⁺ cells (Fig. 4A). Because of the paucity of circulating CD4⁺CD161⁺ T cells, and to avoid a possible modulation of the Th cell phenotype owing to in vitro expansion, the above-mentioned cell populations were collected from pooled buffy coats of five individual donors. To verify the quality of cell sorting, the ex vivo–derived Th17, Th17/Th1, nonclassic, and classic Th1 cell fractions were then analyzed for RORC2, TBX21, IL17A, and IFNG mRNA expression. As shown in Fig. 4B, ex vivo–derived Th17 and Th17/Th1 cells expressed high RORC2 and IL17A, but low IFNG, as well as intermediate TBX21 mRNA levels, whereas classic Th1 cells were characterized by high TBX21 and IFNG, but low RORC2 and IL17A, mRNA levels. In contrast to classic Th1 cells, nonclassic Th1 cells exhibited intermediate RORC2 mRNA levels (Fig. 4B). Next, we analyzed DNA methylation of TBX21, RORC2, IFNG, IL17A, and IL17F gene loci in these ex vivo Th17, Th17/Th1, classic, and nonclassic Th1 cell fractions. As shown in Fig. 4, ex vivo Th17 cells exhibited a complete demethylation of the RORC2 (Fig. 4E) and IL17A (Fig. 4C) ROI, and hypomethylation of the IL17F (Fig. 4C) and TBX21 (Fig. 4F) ROI, but hypermethylation of the IFNG (Fig. 4D) ROI, when compared with CD4⁺ naive Th cells. Ex vivo Th17/Th1 cells were characterized by hypomethylation of all the RORC2, IL17A, IL17F, TBX21, and IFNG ROI, when compared with CD4⁺ naive Th cells (Fig. 4C–F), in line with their mixed phenotype. Strikingly, in contrast to classic Th1 cells, nonclassic Th1 cells were characterized by significant hypomethylation of both RORC2 (Fig. 4E) and IL17A ROI (Fig. 4C). In contrast, IFNG and TBX21 ROI were almost equally demethylated on both nonclassic and classic Th1 cells (Fig. 4D, 4F), whereas IL17F ROI was hypermethylated on both nonclassic and classic Th1 cells (Fig. 4C).

To assess the methylation status of nonclassic and classic Th1 cells in diseased patients, we isolated CD4⁺CD161⁺CCR6⁺IL-17A⁻IFN-γ⁺ (nonclassic Th1) cells and CD4⁺CD161⁻CCR6⁻IL-17A⁻IFN-γ⁺ (classic Th1) cells from the SF of four JIA patients. Unfortunately, the number of cells was insufficient to make direct methylation analysis. For this reason, we performed quantitative PCR analysis of TBX21, RORC2, IL17A, and IFNG on all ex vivo isolated classic and nonclassic Th1 cells. Moreover, we analyzed DNA methylation of TBX21, RORC2, IFNG, IL17A, and IL17F gene loci on in vitro expanded classic and nonclassic Th1 cells derived from two of the four JIA patients. As shown in Fig. 5A, ex vivo nonclassic Th1 cells expressed higher IL17A and RORC2 (with the latter reaching statistical significance) but equal IFNG and TBX21 mRNA levels, when compared with classic Th1 cells. Accordingly, as shown in Fig. 5, 2-wk in vitro expanded nonclassic Th1 cells exhibited reduced methylation of the RORC2 (Fig. 5D), IL17A, and IL17F promoters (Fig. 5B) and IFNG CNS-1 (Fig. 5C) ROI, when compared with classic Th1 cells. No differences in the methylation of IFNG promoter (Fig. 5C) and TBX21 CNS-1(Fig. 5F) ROI were appreciable between nonclassic and classic Th1 cells (Fig. 5).

In a previous study, we showed that human Th17 cells originate from a small population of CD4⁺CD161⁺ Th cell precursors present in both UCB and newborn thymus (9). This small population of naive CD4⁺CD161⁺ T cells exhibited high RORC2, IL23R, CCR6, and IL1R1 mRNA levels, whereas the much more abundant CD4⁺CD161⁻ naive Th cell counterpart virtually did not. However, the CD4⁺CD161⁺ naive Th cells were unable to produce IL-17A, unless they were activated in vitro in the presence of IL-23 and IL-1β, whereas the CD4⁺CD161⁻ naive Th cells could never be induced to produce IL-17A (9). On the basis of these findings, we decided to analyze DNA methylation of the RORC2 gene locus in these two populations as a tool to better define the nature of human Th17 precursors. To this end, ex vivo UCB-derived CD4⁺CD161⁺ and CD4⁺CD161⁻ T cell fractions were separated by cell sorting and first analyzed for RORC2 mRNA expression. As expected, ex vivo–derived UCB-derived CD4⁺CD161⁺ cells expressed high RORC2 mRNA levels, whereas RORC2 mRNA levels in UCB-derived CD4⁺CD161⁻ cells were virtually undetectable (Fig. 6A). Surprisingly, however, RORC2 ROI was hypermethylated similarly in both CD4⁺CD161⁺ and CD4⁺CD161⁻ UCB-derived populations (Fig. 6B).

To explain this apparent contradiction, we hypothesized that only a small minority of the whole population of CD4⁺CD161⁺ naive Th cells was expressing RORC2. To provide experimental support for this hypothesis, we performed PCR analysis of RORC2, IL23R, IL1R1, and on ex vivo UCB-derived CD4⁺CD161⁺ and CD4⁺CD161⁻ cells even at the single-cell level.

This approach was used because bisulfite analysis cannot be performed on single cells; however, it is reasonable to speculate that the detection on a single cell of a given mRNA can be considered a sign of permissive methylation status of the gene locus in the same cell. Therefore, single-cell PCR analysis was performed on a total number of 100 cells from ex vivo UCB-derived CD4⁺CD161⁺ and CD4⁺CD161⁻ cells (50 cells for each cell subset from two different donors). Data obtained on UCB bulk cells subsets confirmed our previous observations (9, 11), that is, that RORC2, IL23R, CCR6, and IL1R1 mRNA was highly expressed on CD4⁺CD161⁺ compared with CD4⁺CD161⁻ cell fractions (Fig. 6C). Moreover, single-cell PCR analysis revealed that only a fraction of CD4⁺CD161⁺ cells expressed the analyzed mRNA and indicated a frequency of RORC2-expressing cells of 25% (CB1 = 18%, CB2 = 32%) in the CD4⁺CD161⁺ cell fraction and of 0% (CB1 = 0%, CB2 = 0%) in the CD4⁺CD161⁻ counterpart (Fig. 6D and data not shown). In addition, 5% (CB1 = 4%, CB2 = 6%) of the CD4⁺CD161⁺ cells analyzed expressed IL23R and none of them expressed IL1R1 and CCR6 (Fig. 6D and data not shown). None of the CD4⁺CD161⁻ T cells expressed IL23R, IL1R1, or CCR6 (Fig. 6D and data not shown). Of note, all the few cells expressing IL23R were contained in the RORC2-positive cell fraction (Fig. 7).

For many years, in both murine experimental models and human diseases, Th1 cells have been thought to be responsible for chronic inflammation and organ-specific autoimmune disorders (25). However, this scenario has been completely changed by the recent discovery, first in mice and then in humans, of a new lineage of CD4⁺ T helper cells that produce IL-17A and express the transcription factor RORγt (RORC2 in humans), named Th17 (3). Several models of knockout mice provided evidence that supports the pathogenicity of Th17 cells, whereas Th1 cells were considered to be mainly protective at the sites of inflammation (26–28). A possible answer to this question can come from the observation that Th17 and Th1 phenotypes are not mutually exclusive. Studies performed in our laboratory demonstrated first that a portion of human IL-17A–secreting cells are also able to simultaneously produce IFN-γ (Th17/Th1 cells) and provided evidence that both Th17 and Th17/Th1 cells can be shifted to the production of IFN-γ alone (named as nonclassic Th1 cells) (4, 10). Following these initial findings in humans, the Th17 cell plasticity was demonstrated even in mice (29).

Despite the findings that support the Th17 origin of nonclassic Th1 cells, it was recently shown that in vitro expanded human Th1 and Th17 cells, as well as intermediate Th17/Th1 cells, have distinct epigenetic signatures at cytokine (IFNG and IL17A) and transcription factor RORC2, but not at transcription factor TBX21, gene loci (17). The results of this study were based on the isolation of different CD4⁺ Th subsets from PB, thanks to the development of a flow cytometry–based method that took advantage of the different expressions of CCR6, CD161, and CXCR3. CCR6⁺CD161⁺CXCR3⁻ CD4⁺ T cells were defined as Th17 cells, CCR6⁺CXCR3⁺ cells as Th17/Th1 cells, and CCR6⁻CXCR3⁺ as Th1 cells (17). However, the Th17/Th1 cells, as defined by applying the Cohen (17) flow cytometry–based method, indeed represented a mixed population, containing not only cells producing both IL-17A and IFN-γ but also Th17-derived (nonclassic) Th1 cells. In addition, owing to the low numbers of cells collected from a single donor, the different cell fractions required in vitro expansions in the presence of autologous feeder cells for 2 wk, thus possibly causing undesired effects of cell polarization and/or modulation.

To identify a more convincing epigenetic signature of pure Th17, Th17/Th1, classic, and nonclassic Th1 cells, we decided to evaluate DNA methylation levels at cytokine (IFNG, IL17A, and IL17F) and transcription factor (TBX21 and RORC2) gene loci of the above-mentioned Th cell subsets obtained from four different sources: 1) Th17, Th17/Th1, classic, and nonclassic Th1 clones; 2) IL-17A⁺IFN-γ⁻ (Th17), IL-17A⁺IFN-γ⁺ (Th17/Th1), and IL-17A⁻IFN-γ⁺ (nonclassic Th1) cells derived from the in vitro polarization of a Th17 cell clone; 3) PB MNC ex vivo isolated CD4⁺CCR6⁺CD161⁺IL-17A⁺IFN-γ⁻ (Th17) cells, CD4⁺CCR6⁺CD161⁺IL-17A⁺IFN-γ⁺ (Th17/Th1) cells, CD4⁺CCR6⁺CD161⁺IL-17A⁻IFN-γ⁺ (nonclassic Th1) cells, and CD4⁺CCR6⁻CD161⁻IL-17A⁻IFN-γ⁺ (classic Th1) cells; and 4) in vitro expanded CD4⁺CCR6⁺CD161⁺IL-17A⁻IFN-γ⁺ (nonclassic Th1) and CD4⁺CCR6⁻CD161⁻IL-17A⁻IFN-γ⁺ (classic Th1) cells derived from the SF of JIA patients. The data clearly demonstrate that Th17 cells, in agreement with their mRNA expression profile, were characterized by low methylation levels at RORC2, IL17A, and IL17F, and by high methylation levels at IFNG, gene loci. In accordance with their mixed phenotype, in addition to hypomethylation of RORC2, IL17A, and IL17F gene loci, Th17/Th1 cells showed higher demethylation levels of TBX21 and IFNG gene loci when compared with Th17 cells. Of note, nonclassic Th1, derived from both PBMC of healthy subjects and SF of JIA-affected patients, could be distinguished from classic Th1 cells not only because of the demethylation of the RORC2 gene locus but also because of their lower methylation levels at the IL17A gene promoter. Finally, and in support of the data obtained on purified Th cell subsets, we found that the IFN-γ–secreting cell fractions derived from the in vitro polarization of Th17 clones were characterized by progressive IFNG promoter demethylation, whereas the RORC2 gene locus and IL17A promoter maintained a complete demethylated status. More importantly, we demonstrated that the acquired IFNG promoter demethylation was maintained after 2 wk of in vitro cultures of the IFN-γ–producing cell fraction in the presence of Th17-polarizing cytokines.

Taken together, these data clearly establish at the molecular level the existence of a strict relationship between Th17 and nonclassic Th1 cells, supporting the concept that nonclassic and classic Th1 cells belong to two distinct cell lineages. Obviously, it cannot be excluded that the activation of naive UCB Th17 precursors (CD4⁺CD161⁺ T cells) in the presence of Th1-polarizing conditions may induce a direct polarization of these cells toward the Th1 phenotype (Fig. 7). These findings also identify the demethylated status of RORC2 and IL17A ROI as useful stable epigenetic markers of the “Th17 membership,” in addition to the expression of CD161 and CCR6 (8). Increased numbers of nonclassic Th1 cells have been found in patients with JIA or Crohn’s disease (21, 23), suggesting that acquisition of the ability to produce IFN-γ by Th17 cells occurs in vivo and associates with chronic inflammatory responses. It seems likely that nonclassic Th1 cells can play a pivotal role in the establishment and persistence of the disease because their number at the level of the inflammatory sites directly correlates with parameters of disease activity (21). Therefore, further examination of the epigenetic profile of Th1 cells from patients with chronic inflammatory disorders will be key to a better understanding of their origin. This issue is also very important from a practical point of view because the knowledge of whether these cells are primarily pathogenic mediators or have protective functions will help to identify which cell types should be therapeutically targeted in a variety of diseases.

In this study, we also investigated the DNA methylation signature at the RORC2 locus in human Th17 cell precursors, which are known to be contained within the CD4⁺CD161⁺ population of UCB cells (9). This population displays RORC2, IL23R, CCR6, and IL1R1 gene expression, whereas its CD161⁻ counterpart virtually does not. Therefore, we assessed DNA methylation levels at the RORC2 promoter in CD4⁺CD161⁺ and CD4⁺CD161⁻ UCB cells. Unexpectedly, comparable methylation levels between these two cell sources were found, allowing us to hypothesize that Th17 cell precursors expressing RORC2 are only a small fraction of the entire CD4⁺CD161⁺ UCB population. Therefore, owing to the impossibility of performing bisulfite sequencing on single cells, we performed single-cell real-time PCR on CD4⁺CD161⁺ and CD4⁺CD161⁻ UCB cells. In agreement with our hypothesis, we found that only 25% of the CD4⁺CD161⁺ cells and none of the CD4⁺CD161⁻ cells exhibited RORC2 mRNA. Accordingly, only 20% of clones expressing RORC2 mRNA could be obtained by culturing single CD4⁺CD161⁺ naive T cells in the presence of IL-1β and IL-23 (data not shown). The analysis of IL23R, CCR6, and IL1R1 mRNA levels on single cells showed that they were expressed by none of the CD161⁻ cells, whereas only 5% of the CD161⁺ cells expressed IL23R. Of note, all these IL23R-positive cells were expressing RORC2. Although we were able to detect CCR6 and IL1R1 mRNA expression on the CD161⁺ bulk population, we found no single cells expressing them, thus suggesting a frequency of positive cells <1%. These data identify one more “tile in the puzzle” of Th17 cell physiology. Taken together, they confirm that only CD4⁺CD161⁺, but not CD4⁺CD161⁻, naive Th cells contain RORC2-expressing Th17 precursors and demonstrate that these latter cells represent only a minority of the entire CD4⁺CD161⁺ T cell population. This finding explains why in our previous studies (9, 11) we were unable to polarize the entire UCB-derived CD4⁺CD161⁺ cells toward the Th17 phenotype following their in vitro activation in the presence of IL-1β plus IL-23. This small subset of CD4⁺CD161⁺RORC2⁺ cells may express IL-23R, and perhaps also CCR6 and IL-1R1, probably in later phases of their development, whereas they acquire the ability to produce IL-17A only following exposure to IL-1β and IL-23 in the tissue environment. However, currently we cannot exclude the possibility that the CD4⁺CD161⁺RORC2⁻ cells represent a still immature stage, which, under certain microenviromental conditions, may be induced to acquire the RORC2 expression (Fig. 7).

We thank M.K. Levings for providing RORC2 downstream sequence and primers.

The authors have no financial conflicts of interest.