Differentiation of IL-17-producing effector and regulatory human T cells from lineage-committed naïve precursors

Abstract

A subset of human regulatory T cells (Tregs) secretes IL-17 and thus resembles Th17 effector cells. How IL-17+Treg cells differentiate from naïve precursors remains unclear. Here, we show that IL-17-producing T-cells can differentiate from CCR6+ naïve T cell precursors in the presence of IL-2, IL-1β, TGF-β, and IL-23. CCR6+ naïve T cells are present in adult peripheral and umbilical cord blood and in both conventional T naïve (T_N) and FOXP3+ naïve Treg (T_Nreg) subsets. IL-17+ cells derived from CCR6+ T_Nreg cells (referred to as IL-17+Treg) express FOXP3, but not HELIOS, another Treg-associated transcription factor, and these cells display suppressor capacity and a surface phenotype resembling memory Tregs. Remarkably, the IL-17+Treg compartment was preferentially reduced relative to the canonical Th17 and Treg compartments in a subset of HIV+ subjects, suggesting a specific perturbation of this subset during the course of disease. Our findings that CCR6+ naïve precursors contain a predetermined reservoir to replenish IL-17-secreting cells, may have implications in balancing the Th17 and IL-17+Treg compartments that are perturbed during HIV infection and potentially in other inflammatory diseases.

Introduction

Regulatory T cells (Tregs) mediate immunological tolerance, curbing autoimmunity and over-exuberant immune responses. Manipulation of Treg responses and numbers in inflammatory disorders, cancer and transplantation settings is a highly sought-after therapeutic strategy (1-3). It is now clear that Tregs are a phenotypically and functionally heterogeneous subset, which can suppress a wide range of immune responses (4, 5). Of particular interest, some Tregs can produce the inflammatory cytokine IL-17A (6-8), and are herein referred as IL-17+Tregs. Recent studies suggest that IL-17+Tregs may also have pathogenic potential (7-9), emphasizing the need for a better understanding of Treg cell sub-specialization. However, the precursor populations and signals that lead to functionally diverse Treg cell subsets are not yet fully elucidated.

Thymus-derived, or “natural” Tregs, (nTregs) express both the FOXP3 and HELIOS transcription factors (10-15). In vitro, nTregs can differentiate and expand from naïve T cells expressing CD25 (T_Nreg) (16-18). Tregs with suppressive capacity can be “induced” (iTreg) from conventional CD25- T_N cells through TGF-β signaling or ectopic expression of FOXP3 (1). However, FOXP3 is also expressed transiently upon TCR activation in the presence of TGF-β, and does not confer suppressive ability (19-21), thus confounding the discrimination and analysis of Treg subsets ex vivo. In addition, TGF-β is crucial for the differentiation of both Treg and Th17 subsets, in a concentration dependent manner (22), and the Treg and Th17 developmental pathways are putatively linked (23). However, to what extent Treg and Th17 lineages share a common precursor or lineage-commitment program is not yet clear.

Human Th17 cells, defined by secretion of IL-17, differentiate from naïve T cells in the presence IL-1β, IL-6, and low concentrations of TGF-β; IL-23 is also required for expansion and maintenance of Th17 cells (24). The Th17 lineage specialization depends on the induction of transcription factor RORγt (25), and all Th17 cells also express chemokine receptor CCR6 for their localization (24). Th17 cells are crucial for host defense against fungal and extracellular bacterial pathogens but are also implicated in the pathology of several autoimmune diseases (24, 26). However, the function and role of IL-17+Treg cells in health and disease remain unclear.

In this study, we sought to determine how human IL-17+Tregs differentiate from precursor naïve T cells. We determined that a portion of the T_Nreg cells expressing CCR6 have a predetermined capacity to differentiate into IL-17+Treg cells in vitro. Remarkably, a small portion of T_N cells expressing CCR6 also has a propensity to develop into Th17 cells. The IL-17+Tregs differentiated from CCR6+ T_Nreg cells could be partly discriminated from their conventional Th17 counterparts by their suppressive activity and surface phenotype, which resembled memory Tregs. Further, we show that the IL-17+Treg compartment is selectively reduced in a subset of HIV-infected individuals with suppressed viral loads through antiretroviral treatment. Together, these findings establish a framework to dissect the developmental program of naïve T cells poised to differentiate into Th17 or IL-17+Tregs, and suggest potential therapeutic strategies to reconstitute these populations after perturbation in diseases such as HIV infection.

Materials and Methods

Cell purification and activation

Blood samples were obtained from anonymous consenting healthy donors as buffy coats (New York Blood Center). Peripheral Blood Mononuclear Cells (PBMC) were isolated from umbilical cord blood or from adult peripheral blood using Ficoll-Paque PLUS gradient (GE Healthcare). CD4⁺ T cells were isolated using Dynal CD4 Positive Isolation Kit (Invitrogen) and further sorted into different naïve and memory subsets using BD FACS Aria (BD Biosciences). Monocyte derived dendritic cells (DC) were generated from CD14+ cells as previously described (27). Purified cells were cultured in RPMI (Life Technologies, Carlsbad, CA) media containing 10% FBS (Fetal Bovine Serum) (Atlanta Biologicals, Lawrenceville, GA) as previously described (27). To activate cells for expansion in vitro and in experiments other than suppression assays, anti-CD3 and anti-CD28 coated beads (Invitrogen) were used at a bead: cell ratio of 1:4 in media containing IL-2 (27).

FACS staining and analysis

Cells were stained in complete RPMI media or PBS+2% FCS and 0.1% sodium azide for 30 minutes at 4°C and washed before running on BD LSR-II flow cytometer. Staining for chemokine receptors was done at room temperature for 45 minutes. Data was analyzed using FlowJo software (Tree Star) and gated on live cells based on fixable viability dye eFluor 780 (eBioscience). The following antibodies were used in stains and sorts: CD45RO, CCR6 (biotinylated), CD161, CD49d, CD25, GARP, CD127, HLA-A2, IL-17A, IFNγ, FOXP3, HELIOS, CCR4, CD3, CD4 (Biolegend), CTLA-4 (BD Pharmingen) and IL-1R1-PE (R&D systems). For intracellular cytokine staining, cells were activated with PMA (20ng/ml for CD4+ T cells and 40ng/ml for PBMC) and Ionomycin (500ng/ml) (Sigma Aldrich) in the presence of GolgiStop protein transport inhibitor (BD) for 4-6 hours. Cells were then stained with fixable viability dye and surface markers, then fixed and permeabilized using ebioscience Fixation/permeabilization buffers according to the manufacturer's instructions, before staining for cytokines and transcription factors. PBMC were pre-cultured in IL-7 (20ng/ml) (Biolegend) for 1 day to enhance Th17 phenotype (28).

In vitro cytokine polarization assay

Sorted T_N and T_Nreg were activated with anti-CD3 and anti-CD28 beads and cultured in media containing IL-2 10ng/ml (Chiron). The next day, IL-1β (10ng/ml), TGF-β (10ng/ml), and IL-23 (100ng/ml) (R&D Systems) were added. Cells were expanded for 2 weeks in media replenished for IL-2 only. For mixed-donor seeding experiments, donor A and donor B were chosen as HLA-A2+ or HLA-A2-, as determined by antibody staining and T_N or T_Nreg from each donor were isolated on the same day. 5,000 cells from donor A were combined with 45,000 cells from donor B. On day 14, HLA-A2 antibody was added to the cytokine stains to determine donor origin. In IL-1R1/CD161 sorting experiments, to enhance expression of Th17 cell- phenotype markers,T cells were pre-cultured in IL-2, IL-7 or IL-15 (20ng/ml, Biolegend) prior to sorting, as described (29).

Real-time PCR analysis

Total RNA was isolated from flash-frozen cells using Qiagen RNeasy© mini kit, and cDNA generated using High capacity reverse transcriptase kit (Applied Biosystems). Taqman primer/probe mixtures were purchased from Applied Biosystems: RORC (Hs01076112_m1) β-Actin (Hs99999903_ml). Samples were run on Applied Biosystems 7300 apparatus. Data were normalized to β-Actin for each sample.

Statistical analysis

All statistics were done using GraphPad Prism software. Two-tailed t-test was used in all figures except Figure 5, in which the non-parametric Mann-Whitney U test and Spearman's rank correlation were used.

Figure 5. The IL-17+Treg compartment is perturbed in HIV+ individuals.

PBMCs from HIV+ subjects on antiretroviral therapy or HIV negative controls were stained for proteins including CD4, FOXP3, HELIOS and IL-17. (A, B) The percent of memory CD4+ T cells that are (A) FOXP3+HELIOS+ or (B) FOXP3+HELIOS- in controls or HIV+ subjects. (C, D) PBMCs were cultured overnight in IL-7 then stimulated with PMA and Ionomycin in the presence of GolgiStop then stained for surface markers and intracellular proteins. The proportion within total CCR6+ memory cells of (C) total IL-17+IFNγ- cells or (D) FOXP3+IL-17+IFNγ- cells. (E, F) Correlation between the proportion of FOXP3+HELIOS- T cells in the total FOXP3+ compartment shown on the x-axis and the proportion of IL-17+Tregs within the total IL-17+ compartment on the y-axis among (E) the controls and (F) HIV infected cohort

Human Subjects

Twenty HIV negative controls and 18 HIV-infected individuals were recruited in accordance with an IRB approved protocol and consent from NYU School of Medicine. All HIV positive subjects were well controlled on antiretroviral therapy with HIV viral load less than 100 copies/mL. HIV infected subjects had median (range) CD4 T cell counts of 524 (220-1281) cells/mm³ and median (range) CD4 percent of 31% (11-44%). Uninfected controls were subjects recruited from an IRB approved protocol or random blood samples that were obtained from the blood bank.

Treg suppression assay

Resting Naïve CD4+ cells (targets) were labeled with CFSE (Invitrogen) and plated at 3 x 10⁴/ well in 96-well U-bottom plates. Suppressor cells were added at various ratios, and cells were activated using DC (generated as described above) at a DC: target cell ratio of 1:10 and anti-CD3 antibody (ATCC clone OKT-3) at 10-100ng/ml. All cells were washed at least twice in complete media to remove any cytokines. CFSE dilution was measured on BD LSRII flow cytometer 4 and 5 days later.

Results

Characterization of the Treg and Th17 compartments of human T cells

Tregs are generally identified by resting expression of CD25 and FOXP3 (1). Human Tregs can also be subdivided into T_Nreg and Treg subsets based on expression of CD45RO (Fig. 1A) and (17, 18, 30). Recently, concomitant expression of HELIOS and FOXP3, in both humans and mice, was suggested as a specific marker of bona fide Tregs (12, 31). Indeed, we observed significant heterogeneity within FOXP3+ Tregs with regards to expression of HELIOS (Fig. 1B), in line with previous reports (12, 15). We found that only about half of the memory FOXP3+HELIOS+ T cells expressed CD25, whereas within naïve T cells, more than 90% of FOXP3+HELIOS+ cells were CD25 positive (Fig. S1A and S1B). Similarly, the total FOXP3+HELIOS+ within memory T cells, was typically twice the percent of the CD25bright subset (Fig. S1C), suggesting that at least about half of all memory Treg cells were low for CD25 expression. As such, to ensure CD25low Tregs were not excluded from our analysis, we utilized FOXP3 and HELIOS as markers to define human Treg subsets throughout the study.

Figure 1. Defining IL-17-secreting human Treg subsets.

CD4+ T cells were sorted from PBMCs and stained with antibodies against CD25, CD45RO (surface) and FOXP3, HELIOS (intracellular). (A) Surface phenotype, and (B) FOXP3/HELIOS expression within CD25 and CD45RO populations are shown. (C) CD4+ cells cultured overnight in IL-2 containing media were stimulated with PMA/Ionomycin and then intracellularly stained for IL-17, IL-22 and IFNγ within FOXP3/HELIOS populations of memory CCR6+ cells. (D) Percent of cytokine expressing cells from several donors is shown.

To further characterize the previously described IL-17+Treg subset (6, 32), we determined expression of HELIOS and effector cytokines on T-cells derived from healthy human blood. Among CCR6+ T cells, which contain all IL-17+ cells, similar levels of IL-17, IL-22 and IFNγ were found in FOXP3+HELIOS- cells compared to the FOXP3-HELIOS-subset. However, FOXP3+HELIOS+ cells did not secrete any of these cytokines (Fig. 1C and D).

Differentiation of human IL-17+Tregs

It was recently reported that IL-17-producing cells preferentially arise from T_Nreg cells in the presence of the Th17-polarizing-cytokines IL-1β, IL-23 and TGF-β (33). Using this cytokine polarization protocol (Fig. 2A) we differentiated and expanded IL-17-secreting cells from highly purified T_Nreg precursors (Fig. 2B and C). Similar to ex vivo analysis, in vitro-generated FOXP3+IL-17+ cells did not express HELIOS (Fig. 2D and E). Although we noticed significant donor-to-donor variability in deriving IL-17+Tregs from T_Nreg precursors, each individual displayed consistent potential to produce the same relative amount, when the experiment was performed on cells purified from blood of the same donor collected at different time points (Fig. S2A). As previously shown, expanded T_N cells also expressed FOXP3, which is induced by TGF-β during T cell activation (34, 35).

Figure 2. In vitro generation of IL-17-producing cells from T_Nreg cells.

(A) Polarization scheme with T_N or T_Nreg cells. (B) Representative FOXP3/HELIOS expression in day 14 polarization cultures. (C) Averages of IL-17 production from T_N or T_Nreg cells cultured in IL-2, compared to polarizing cytokines. The data are average of 3 donors from 3 independent experiments. (D) Representative intracellular IL-17/IFNγ within FOXP3/HELIOS gated populations shown in B. (E) IL-17 expression in polarized T_Nreg cells gated as shown in B and C, for multiple donors/ independent experiments. (F) T_N cells were seeded at a 1:10 ratio with T_N or T_Nreg cells from a different donor, with discrete HLA-A2 phenotype, and polarized as shown in A. (G) T_Nreg cells were seeded with different donor T_N or T_Nreg cells, as in F, and polarized as shown in A. Data in F and G are levels of IL-17 produced in gated seed donor cells. 3 donors/ independent experiments are shown.

Because expanded T_Nreg cells contained both FOXP3- and FOXP3+HELIOS- cells among the canonical HELIOS+FOXP3+ Tregs, we designed an experimental approach to rule out that T_Nreg cultures could be providing an in trans polarization signal to a small amount of “contaminant” non-T_Nreg cells therein. We therefore intentionally seeded FOXP3- T_N cells into T_Nreg polarization cultures. To ensure identification of the seeded cells at the end of the expansion, we used two donors in each experiment, one for seeded and the other for a driver population, which were discerned by presence or absence of MHC HLA-A2 expression. We mixed cells at a 1:10 seed:driver ratio, and stimulated them in the presence of polarizing cytokines as in Figure 2A. We found that IL-17 was produced only from seeded cells with a T_Nreg origin, even when they were mixed at 1:10 ratio with T_N cells (Fig. 2F), and that seeded T_N cells did not differentiate into IL-17-secreting cells, even in the presence of T_Nregs (Fig. 2G). Thus, T_Nregs have a preferential propensity to differentiate into IL-17-secreting cells.

IL-17+Tregs are derived from CCR6+ naïve precursors

We next asked if there was a specific subset within T_Nregs that was poised for IL-17+Treg differentiation. We first assessed the previously reported IL-1R1 and CD161 as potential markers for IL-17+Treg precursors (36-38). While sorting cells based on these markers did enhance IL-17-secreting-cell differentiation (Fig. S2B), when CCR6+ cells were excluded, IL-1R1+CD161+ cells did not give rise to IL-17+Tregs (Fig. S2C). Therefore we sorted the CCR6+ cells present within the adult naïve T cell compartment (Fig. 3A). Almost all CCR6+ TNreg cells expressed FOXP3, and about 75% were also HELIOS+, very similar to CCR6- TNreg cells, whereas CD25-CCR6+ T_N cells were FOXP3 negative (Fig. S3D). Importantly, CCR6+ naïve T-cells also expressed higher levels of RORC compared to CCR6- naïve T cells (Fig. S3E). Although RORC expression within resting CCR6+ naive T cells was about 10-fold lower compared to memory CCR6+ cells (Fig S3E), after TCR activation and expansion of CCR6+ TN and TNreg subsets in polarization conditions, their RORC expression became comparable to memory T cells (Fig. S3F). In addition, to ascertain whether these CCR6+ naïve cells were recent migrants from the thymus, we determined their numbers in umbilical cord blood (Fig. 3B). Although this population was relatively small, it was consistently present in healthy donors and fewer in cord blood (Fig. 3C). We then compared the differentiation of naïve CCR6+ and CCR6-T cells into IL-17+ cells using the polarization scheme described above (Fig. 2A). We found that only CCR6+, but not CCR6- cells, were capable of differentiating into IL-17+ cells, regardless of whether they were purified from adult T_N, adult T_Nreg, or umbilical cord blood naive cells (Fig. 3D and E). However, CCR6+ naïve T cells sorted from the adult blood displayed increased capacity to differentiate into IL-17+ cells (Fig. 3F), relative to CCR6+ naïve T cells sorted from the cord blood (Fig. 3G). This finding suggests that the CCR6+ naïve T cell subset may be both developmentally and environmentally regulated, which is also consistent with significant adult donor-to-donor variation (Fig. S2A).

Figure 3. A subset of naïve T cells that express CCR6 develop into IL-17 producers.

CCR6 surface expression was assessed in total CD4+ T cells, along with CD25 and CD45RO, for gating on Treg populations as shown in Fig 1A. Representative CCR6 expression versus CD45RO expression is shown for each subset of (A) adult blood or (B) umbilical cord blood. (C) Multiple donors/ independent stainings are shown. (D) Day 14 cultures of polarized CCR6+ or CCR6- cells were re-stimulated with PMA/Ionomycin in the presence of GolgiStop and stained for FOXP3 and HELIOS. (E) Populations gated as in D were assessed for IL-17 and IFNγ production. (F) Multiple donors/ independent experiments shown for representative plots for adult blood in E. (G) Multiple donors/independent experiments shown for representative plots for umbilical cord blood in E.

While we observed FOXP3+IL-17+ cells were derived from both CCR6+ T_N and T_Nreg populations, we hypothesized that only CCR6+ T_Nreg give rise to bona fide IL-17+Tregs. To address this question, we compared both in vitro differentiated subsets to ex vivo Tregs. We first performed in vitro suppression assays, the gold standard for Treg function, and found that our T_Nreg-derived cells were similarly suppressive to ex vivo CD45RO+ memory Tregs (Fig. 4A). Because, it was not technically feasible to determine the suppressive capacity of only IL-17-secreting cells within our bulk cultures, in lieu, we assessed a set of Treg surface markers (1, 39, 40), on these cells. We found that FOXP3+IL-17+ cells generated from the CCR6+ T_Nreg subset, but not from CCR6+ T_N cells (Fig. 4B), closely resembled ex vivo-isolated IL-17+Tregs in expression of CD25, IL-1R1, CCR4, CTLA-4, and CD49d, when compared to FOXP3-IL-17+ cells within the same cultures (Fig. 4C). Similarly, the FOXP3+ but IL-17-negative cells expanded and polarized from CCR6+ T_Nreg cells expressed Treg markers not seen on FOXP3- cells. However, among CCR6+ T_N cells, there was no difference between FOXP3+ IL-17- or FOXP3- IL-17- cells within the same cultures (Fig. S3). Taken together, these results corroborate the notion that IL-17+Tregs mostly develop from CCR6+ T_Nreg precursors.

Figure 4. Characterization of in vitro derived IL-17+Tregs.

(A) Expanded and polarized CCR6+ subsets from figure 3 were subjected to an in vitro assay for suppression of T-cell proliferation. Ex vivo mature Tregs were used as a positive control. Target fresh T_N cells were labeled with a cell trace, and stimulated with DCs and anti-CD3 antibody, in the presence or absence of candidate “suppressor” cells at various suppressor: target ratios noted. After 5 days, dilution of cell trace in the target cells was assessed by flow cytometry. (B) Expanded T_Nreg or T_N cells were stimulated with PMA and Ionomoycin and stained for surface markers, followed by intracellular staining for FOXP3 and IL-17. IL-17+ subsets were gated based on their FOXP3 expression, and the FOXP3+ versus FOXP3- portions of each are compared with histogram overlaps. One donor representative of 3 is shown. (C) Ex vivo sorted CD4+ T cells were cultured overnight in IL-2 containing media and then stimulated, stained, and subjected to the same analysis as in B.

IL-17+Tregs are decreased in an HIV infected cohort

The homeostasis of Treg and Th17 populations during HIV infection is of keen interest since both are integral to proper immune function in the gut mucosa, and both subsets have been reported to be perturbed at different stages of the infection (41, 42). Given our findings that IL-17+Tregs derive from a predetermined naïve precursor, we asked whether this subset differed in HIV+ subjects compared to HIV- controls. We profiled PBMCs from a subset of HIV-infected individuals who were on antiretroviral therapy and had very low or undetectable viremia. HIV+ patients and HIV- controls were analyzed for T cells that express Treg markers and/or secrete Th17 cytokines. We found that there was a specific and significant increase in total HELIOS+FOXP3+ Tregs in HIV+ subjects (Fig. 5A), whereas FOXP3+HELIOS- T cells did not differ from healthy controls (Fig. 5B). Thus, as a proportion of total FOXP3+ T cells, the HELIOS- subset was significantly lower in HIV+ subjects (Fig. S4A). Importantly, while the percent of CCR6+ memory T cells that were IL-17+ IFNγ- were not different in HIV+ subjects compared to healthy controls (Fig. 5C), the IL-17+Tregs (defined as FOXP3+IL-17+IFNγ- T cells) were greatly reduced among this cohort of HIV+ individuals (Fig. 5D). Additionally, the proportion of FOXP3+IL-17+ cells within the Th17 compartment was also significantly decreased in HIV+ subjects compared to controls (Fig. S4B). Remarkably, the proportion of FOXP3+HELIOS- T cells in the total FOXP3+ compartment positively correlated with the proportion of IL-17+Tregs within the total IL-17+ compartment in HIV+ subjects but not in healthy controls (Fig. 5E and F). These data suggest a functional association between Treg and IL-17+Treg perturbation during HIV disease and identify a specific “immunological hole” of IL-17+Tregs in some of the infected individuals.

Discussion

Here, we show that CCR6+ human naïve T cells selectively differentiate into IL-17 secreting cells; CCR6+ T_Nreg precursors are lineage-committed towards IL-17+Treg cells, whereas CCR6+ T_N cells differentiate into conventional Th17 cells. Corroborating these findings, the in vitro generated human IL-17+Tregs phenotypically and functionally resemble endogenous memory IL-17+Tregs.

These findings may reconcile previously conflicting reports about whether the origin of IL-17 producers is CD25+ or CD25- T cells (33, 43, 44). We demonstrate that both types of naïve cells may be differentiated to produce IL-17, however only from the respective CCR6+ fractions. IL-17-producers within the naïve human T cell differentiation cultures were probably overlooked (33) since <1% of human T_N cells express CCR6 (Fig. 3A), (45). In addition, the two previous studies that reported Th17 differentiation from CD25-T_N precursors did not examine T_Nreg cells, and may have utilized culture growth that selected for outgrowth of CCR6+ cells (43, 44).

Furthermore, the presence of CCR6+ naïve T cells, which are poised to differentiate into Th17 cells in cord blood supports the notion that a subset of Th17 lineage cells is developmentally regulated. Indeed, similar to bona fide Tregs, some Th17-like cells have been shown to emerge directly from the thymus and to be activated by self-antigen in the periphery (46). While we were constrained by limited amounts of CCR6+ cells in umbilical cord blood samples and were not able to further subdivide them into CD25+CCR6+ and CD25-CCR6+ cells, a portion of these cells reproducibly differentiated into FOXP3+IL-17+ cells. However, it is not clear why there are more CCR6+ naïve precursors of Th17 and IL-17+Treg cells in adult blood. It is conceivable that this is due to preferential homeostatic expansion of CCR6+ T-cells due to the environmental cytokine milieu. It is also possible that naïve T cells can be conditioned toward a lineage commitment program, for example based on the cytokine milieu, thus expressing CCR6 in the periphery.

A major difficulty in defining Treg subsets, especially in humans, has been the expression of FOXP3 in non-Treg cells and the similarity of other Treg markers (such as CD25) to the phenotype of activated T cells (19, 35). However, single-cell cloning of FOXP3+IL-17+ cells show that ex vivo IL-17+Treg cells are indeed suppressive (6, 8, 47). Furthermore, suppression assays performed by Valmori and colleagues demonstrate the suppressive activity of polarized T_Nreg cultures (33). Consistent with these findings, polarized CCR6+ T_Nreg cells displayed potent suppressive activity and the IL-17+FOXP3+ cells therein, were strikingly different from FOXP3+IL-17+ cells from T_N cultures with regard to expression of the Treg markers (Fig. 4 A and B). In addition, these markers exclude cells that produce IFNγ (data not shown), which is associated with another described subset of Th17 cells (24). Future studies assessing the methylation pattern at the Treg-specific-demethylation region (TSDR) will be useful in assessing the stability of CCR6+ T_Nregs and IL-17+Tregs.

Both Treg and Th17 cells are crucial for maintaining mucosal barrier integrity and a balanced immune response in the lamina propria (24, 42). Quantitative assessments of IL-17+Treg cells in gut tissues are difficult to make in humans, and limited data in healthy donors is available. However, several studies now demonstrate an accumulation of IL-17+Tregs in the gut during inflammatory states (7-9), when their function may be more pertinent. One possible role for IL-17+Treg cells could be both suppressing the formation of professional memory responses by T-cells, while also securing barrier integrity and ensuring a basal level of microbial control via secretion of IL-17 (42, 48). For example, IL-17+Tregs could prevent adaptive immune responses against commensal bacteria, while using IL-17 to keep the growth of commensals to an acceptable level and to repair epithelial damage caused by pathogens. Indeed, host tolerance of pathogens through employment of basic repair mechanisms, may allow the immune response to proceed in a less vigorous manner, thus minimizing immunopathology (49). Conversely, in specific contexts, IL-17+Tregs may also contribute to mucosal disease, as they were recently implicated in development of colon cancer (7, 9) and possibly also inflammatory bowel disease (8).

In the setting of HIV infection, we found that IL-17+Treg cells in HIV+ individuals on active antiretroviral therapy with suppressed viremia were significantly reduced. Interestingly, in these HIV+ subjects the levels of FOXP3+HELIOS- cells were also greatly decreased as a proportion of total FOXP3+ cells and this correlated a with decreased proportion of IL-17+Tregs in the Th17 subset. In contrast, FOXP3+HELIOS+ Tregs were overall higher in these HIV+ subjects compared to healthy controls. As such, these findings reveal an “immunological hole” within the IL-17-secreting and Treg cell compartments in vivo, despite effective antiretroviral therapy. These results also support our in vitro evidence that the HELIOS- IL-17+Treg cells represent a separate compartment within Treg or Th17 subsets. It is unclear why these cells are preferentially reduced in the blood during HIV infection and what biological outcome this may have in HIV+ individuals. Given that the gut mucosal barrier is likely to be breached during HIV infection (41, 42) and there is a propensity of IL-17-secreting cells to migrate to the gut mucosa, it is conceivable that IL-17+Tregs are preferentially targeted by the virus and, at least in some individuals, do not recover. Thus, it is tempting to speculate that disruption of IL-17+Tregs in certain HIV+ subjects contributes to disruption of mucosal barriers and may have unknown long-term adverse consequences. In future studies, it may be important to assess the consequences of IL-17+Treg deficiency in the context of HIV infection in a humanized mouse models. Ultimately, our findings that IL-17+Tregs differentiate from a discreet subset of precursor naïve T cells may enable better manipulation of these cells as a therapeutic approach during HIV infection and other disease states where they are perturbed.