Type 1 Diabetes-associated IL2RAvariationlowers IL-2 signaling and contributes to diminished CD4⁺CD25⁺ regulatory T-cell function

Abstract

Numerous reports have demonstrated that CD4⁺CD25⁺regulatory T cells (Tregs) from individuals with a range of human autoimmune diseases, including Type 1 diabetes (T1D),are deficient in theirability to control autologous pro-inflammatory responses when compared to non-diseased, control individuals. Treg dysfunction could be a primary, causal event or may result from perturbations in the immune system during disease development.Polymorphisms in genes associated with Treg function, such as IL2RA, confer a higher risk of autoimmune disease. Although this suggests a primary role for defective Tregs in autoimmunity, a link between IL2RA gene polymorphisms and Treg function has not been examined. We addressed this by examining the impact of an IL2RA haplotype associated with T1D on Treg fitness and suppressive function. Studies were conducted using healthy human subjects to avoid any confounding effects of disease. We demonstrated that the presence of an autoimmune disease-associated IL2RA haplotype correlates with diminished interleukin (IL)-2-responsiveness in antigen-experienced CD4⁺ T cells, as measured by phosphorylation of STAT5a, and is associated with lower levels of FoxP3 expression by Tregs, and a reduction in their ability to suppress proliferation of autologous effector T cells. These data offer a rationale that contributes to the molecular and cellular mechanisms through which polymorphisms in the IL-2RA gene impact upon immune regulation, and consequently upon susceptibility to autoimmune and inflammatory diseases.

Introduction

Type 1 diabetes (T1D) is characterised by autoimmune destruction of pancreatic beta cells, a process in which autoreactive T cells play a pivotal role(1-3). There is now a growing body of evidence to suggest that in T1D, this pathological autoimmunity is the direct result of a failure of immune regulation(4). This includesthe defective function of various populations of regulatory T cells (Tregs), especially those characterised as CD4⁺CD25^hiFoxP3⁺. In support of this, we and others have demonstrated that the suppression of autologous responder T cells by CD4⁺CD25^hi Tregs in individuals with newly diagnosed T1D is reduced significantly compared with that observed in age-matched control subjects(5-8). Importantly, we also demonstratedin an independent cohort of patients that defective suppression is not only present close to diagnosis, but also in individuals who have had T1D for over 20 years, suggesting that the functional defect represents a phenotype that is stable over time and most likelyunder genetic control(9).

A candidate gene study identified an association between T1D and the IL-2RAgene (encoding the IL-2 receptor alpha chain, CD25) (10, 11). There are three protective IL2RA haplotypes, one of which is marked by the SNP rs12722495,where the protective allele confers a relative risk for T1D of 0.65. Recently, the disease-associated IL2RArs12722495haplotype was correlated with several distinct cellular immunophenotypes, most notably the higher expression of CD25 on memory CD4⁺ T cells and higher levels of IL-2 secretion from these memory T cells (12). IL-2RA is part of the high affinity IL-2 receptor complex and is constitutively expressed at high levels on both naturally occurring and peripherally induced FoxP3⁺ Tregs(13, 14). Numerous linesof evidence from both mouse and man have demonstrated that IL-2 plays a key role in both the generation and function of FoxP3⁺Tregs (15-20).Studies invitro have demonstrated that activation of CD4⁺CD25⁺ T-cell suppressor function requires IL-2 (21) and in vivothat IL-2 is required for peripheral survival and expansion of CD4⁺CD25⁺ Tregs (reviewed in (13)). Furthermore, signaling via common γ-chain cytokines, of which IL-2 is a key member, is required for the maintained expression of FoxP3 by Tregs, which is essential for their suppressive function (19, 22).

These findings invoke the hypothesis that gene polymorphisms in the IL-2/IL-2RA pathway exert their influence on T1D risk via effects on the number or functional ability of FoxP3⁺ Tregs. In support of this, recent studies show that Tregs from individuals with T1D are more prone to apoptosis and more easily lose expression of FoxP3 and that these phenotypes may be linked to a relative defect in signaling via the IL-2 pathway(7, 23, 24). To date, however, direct evidence linking polymorphisms in IL2RA with altered Treg function is lacking. To address this knowledge gap, we examined whether the T1D susceptibility allele defined by the IL2RASNP rs12722495is associated with a reduction in the functional capacity of Tregs. In order to avoid any potential confounding effects of disease on Treg phenotype, we conducted these studies in individuals without disease. Our studies show that that the T1D-susceptibility IL2RA haplotype identified by rs12722495 is associated with decreased signaling via the IL-2 pathway in both memory T cells and Tregs and that this is linked to diminished Treg function.

Materials and Methods

Subjects and study design

Individuals homozygous for the T1D protective rs12722495 IL2RA haplotype and the fully susceptible IL2RA haplotype (denoted as P1P1 and SS, respectively) were recruited for the present study. The levels of CD25 expression on CD4⁺ T-cell subsets have previously been studied in these individuals and pairs of P1P1 and SS individuals were pre-selected who showed typical haplotype-specific patterns of CD25 expression on the conventional memory CD4⁺ T-cell subset (i.e. a relatively higher level of CD25 expression in P1P1 individuals compared to SS). For peripheral blood mononuclear cell (PBMC) isolation, blood samples were collected in vacutainers containing sodium heparin anti-coagulant, diluted 1:1 with RPMI 1640 supplemented with 100 μg/ml penicillin/streptomycin (Invitrogen Ltd, Paisley, UK) and stored overnight under constant rotation. The following day, PBMCs were isolated from whole blood by density gradient centrifugation (Lymphoprep, Axis-Shield PoC AS, Oslo, Norway). For whole blood staining, blood was collected into vacutainers containing sodium heparin anti-coagulant and used on the day of collection. Due to the day-to-day variation inherent in intracellular staining protocols, it was not possible to normalize pSTAT5a and CD25 MFI values through time as we were able to do previously using a cell surface staining protocol (12). Therefore all analyses were performed in a paired manner with one P1P1 and one SS individuals analyzed on a single day.Ethical approval for this study was granted by the local Ethics Committee and informed consent obtained. All experiments were performed in a blinded manner without knowledge of genotype of the individual.

Monoclonal antibodies and reagents

The following antibodies were used in these studies as indicated: phycoerythrin (PE)-conjugated monoclonal anti-CD25 (clones M-A251 and 2A3), fluorescein isothiocyanate (FITC)-conjugated anti-CD4 (clone SK3), AlexaFluor647-labelled anti-STAT5a (pY694) and AlexaFluor488-labelled anti-STAT5a (pY694) were obtained from BD Biosciences (Oxford, UK); eFluor450 labelled anti-CD4 (clone SK3), anti-CD45RA (clone HI100) and peridin-chlorophyll protein-cyanine 5.5 (PerCP-Cy5.5)-conjugated anti-CD127 (clone eBioRDR5) were obtained from eBiosciences (Hatfield, UK); and AlexaFluor700 labelled anti-CD45RA (clone HI100), allophycocyanin-cyanine 7 (APC-Cy7)-conjugated anti-CD14 (clone HCD14), fluorescein isothiocyanate (FITC)-conjugated anti-Helios (clone 22F6), AlexaFluor647-labelled anti-FoxP3 (clone 259D), AlexaFluor700-labelled anti-CD4 (clone RPA-T4), Pacific blue-conjugated anti-CD45RA (clone H100) and PE-conjugated anti-FoxP3 (clone 259D) were obtained from BioLegend Ltd, (Cambridge UK). Antibody concentrations used were based upon manufacturers’ recommendations and optimization studies. Complete media for functional studies was X-Vivo-15 media (Lonza Ltd., Wokingham, UK) supplemented with 5% human pooled AB+ sera (PAA Laboratories, Lutterworth, UK) and 100 μg/ml penicillin/streptomycin (Invitrogen, Paisley, UK). Proleukin (Chiron Corporation, Emeryville, USA) was used as a source of human recombinant IL-2.

Flow cytometric analysis for phosphorylated STAT5a

Analysis of phosphorylated STAT5a in PBMCs was performed using BD Phosphoflow reagents according to the manufacturers’ instructions. Briefly, 1×10⁶ PBMCs were incubated with various concentrations of IL-2 for 10 minutes at 37°C, fixed with Phosflow buffer I, permabilized with Perm Buffer III, stained with anti-CD4-FITC, anti-CD25-PE, anti-CD127-PerCP-Cy5.5, anti-CD45RA-eFluor450 and anti-STAT5a-pY694-APC and analysed in a BD FACSCantoII. Due to a technical failure in one sample, only nine pairs of PBMC samples were analyzed for expression of pSTAT5a. For simultaneous detection of pSTAT5aand FoxP3, 500 μl of fresh blood was incubated with 500 μl of X-Vivo media containing various concentrations of IL-2 for 10 minutes at 37°C, fixed with warm BD Lyse/Fix buffer and permeabilized with 100% methanol for 20 minutes on ice. After extensive washing with PBS containing 0.2% BSA, cells were stained with anti-CD4-AlexaFluor700, anti-CD25-APC, anti-CD45RA-Pacific Blue, anti-FoxP3-PE and anti-STAT5a-pY694-AlexaFluor488 and analyzed using a BD Fortessa. Results are expressed as the median fluorescence intensity (MFI) of pSTAT5a staining for all cells within a particular T-cell subset.

Isolation and analysis of cell populations for functional studies

PBMCs were stained with anti-CD4-eFluor450, anti-CD25-PE, anti-CD127-PerCP-Cy5.5, anti-CD45RA-AlexaFluor700 and anti-CD14-APC-Cy7 and control populations stained with anti-CD4-eFluor450, IgG1-PE, IgG1-PerCP-Cy5.5, anti-CD45RA-AlexaFluor700 and anti-CD14-APC-Cy7. Lymphocytes were identified based on forward and side-scatter parameters and populations isolated for functional analysis using a BD FACS-Aria II flow cytometer and FACS Diva Software (BD Biosciences Oxford, UK). Flow cytometry data was analyzed using FlowJo software (Treestar Inc., Ashland, OR, USA).

Flow cytometric analysis for FoxP3 and Helios

FoxP3 and Helios staining was performed on cells immediately post-sorting and after 48 hours of culture with various concentrations of IL-2 using the BioLegend FoxP3 Fix/Perm buffer set according to the manufacturers’ instructions. For analysis of FoxP3 and Helios in cultured cells, 10⁴ sorted Tregs were incubated in complete media supplemented with IL-2 for 48 hours prior to analysis.

In vitro co-culture suppression assays

Suppression assays were performed by culturing memory or naïve conventional T-cell populations (2.5×10³/well) in the presence or absence of either autologous Tregs or a third party Treg cell line at the ratios indicated. Cells were activated by the addition of Dynabeads® Human T-Activator anti-CD3/anti-CD28 beads (Invitrogen, Paisley, UK) at a bead:conventional-cell ratio of 1:1. All conditions were conducted in triplicate. The third party Treg cell line was generated by expanding FACSorted Tregs from a single donor for 14 days in 600 U/ml IL-2. Expanded Tregs were cryopreserved and a single aliquot was thawed for use with each pair of samples analyzed. After 5 days of culture, 100 μl of supernatant was removed and stored at −80°C for later cytokine analysis. Proliferation was assessed by the addition of 0.5 μCi/well 3[H]-Thymidine (Perkin Elmer, Waltham, MA, USA) for the final 18 hours of co-culture. The percentage of suppression was calculated using the following formula: %suppression = 100-(counts per minute (cpm) in the presence of Tregs ÷ cpm in the absence of Tregs) × 100).

Statistics

The normality of all datasets was tested using D’Agostino-Pearson omnibus normality test. Where data did not significantly deviate from the normal distribution, either an independent or paired Student t test was used to test for significance as indicated. Where one or more datasets were found to significantly deviate from the normal distribution, statistical significance was determined using Wilcoxon matched-pairs signed rank test for paired data. All statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc., La Jolla, CA.).

Results

We have previously reported that the IL2RArs12722495 protective haplotype is associated withsignificantly higher levels of expression of CD25 on conventional memory CD4⁺ T cells (mTconv) and we observed a similar trend for FoxP3⁺ Tregs (12). Therefore, we first sought to investigatewhether increased levels of CD25 expression from individuals with the P1P1 IL2RA haplotype resulted in altered responsiveness to IL-2 signaling, measured via phosphorylation of STAT5a in all T-cell populations after brief in vitro exposure to IL-2. As fixation precluded the use of CD127 as a surface marker of Tregs and FoxP3 co-staining was found to be un-reproducible using the pSTAT5a staining protocol for PBMC samples, Tregs wereinitiallyidentified based on a high level of CD25 staining and reduced CD4 staining as previously described by other investigators and illustrated in Figure 1A (the frequency of CD4^loCD25⁺Tregs defined using this method and via CD25⁺CD127^lo expression was highly correlated, R²=0.87, P<0.0001, Figure S1). In addition, a more stringent definition was applied to identify Tregs expressing very high levels of CD25 (CD25^hi Tregs) by gating on the top 1% of CD25-staining CD4⁺ T cells as previously described by other investigators (23, 25) (gating on an identical population in unfixed cells confirmed that >98% of these cells were CD127^lo/-vein all individuals examined). Conventional T cells were subdivided based on expression of CD45RA to delineate memory (CD45RA⁻, mTconv) and naïve (CD45RA⁺, nTconv) (Figure 1B). Similarly, Tregs were subdivided into CD45RA⁻ Tregs and CD45RA⁺ (Figure 1C). As CD25^hi Tregs were >90% CD45RA⁻ (Figure 1D; i.e. consisted mainly of antigen experienced Tregs) this population was not sub-divided for analysis. In all data presented, a pair of subjects differing at the rs12722495 SNP has been studied in a discrete, simultaneous experiment; each pair is denoted by a distinct symbol that is consistent throughout all graphs in the results section. Details of the gender and age bands of subjects and the symbols used to denote the results are shown in Supplemental Table I.

Figure 1. Example of the gating used for the identification of T-cell populations for STAT5a phosphorylation studies using PBMCs.

A. Lymphocytes identified by their forward and side scatter properties were gated for CD4⁺ expression and then examined for expression of CD25. Tregs were identified using two different gating strategies; firstly based on a high level of CD25 staining and reduced CD4 staining (CD4^loCD25⁺ Tregs) and secondly gating on the top 1% of CD25-staining CD4⁺ cells (CD25^hi Tregs). Conventional T cells were identified by low/intermediate levels of CD25 staining. B-D.All populations were analysed for expression of CD45RA to delineate populations of CD45RA⁺ and CD45RA⁻ conventional T cells and Tregs.

Relationship between IL2RA haplotype and expression of CD25 on CD4⁺ Tconv and Treg populations

Analysis of isolated PBMCs indicated that individuals with the P1P1 IL2RA haplotype express significantly more CD25 on mTconv (P=0.0002) and Treg (P=0.01) compared to individuals with the SS haplotype (Figures 2A-B). The higher levels of CD25 were observed both in the CD45RA⁺ (P=0.005) and CD45RA⁻ CD4^loCD25⁺ Tregs (P=0.008) (Figures 2C-D). We observed no difference in the frequency of mTconv between the P1P1 and SS individuals (mean 51.5% and 42.7%, respectively), the proportion of CD4⁺ T cells classified as Tregs (mean 4.28% and 4.38%, respectively) or the percentage of CD45RA⁻ Tregs (mean 69.2% and 65.2%, respectively). These observations are all in keeping with our previous report (12) even though CD25 levels were quantified using an intracellular staining protocol in the current study.

Figure 2. Relationship between IL2RA haplotype and CD25 expression on CD4⁺ Tconv and Treg populations from PBMCs.

PBMC from donors with the protective rs12722495 IL2RA haplotype (P1P1) and donors with the susceptible IL2RA haplotype (SS) were isolated and analyzed as matched pairs by flow cytometry as shown in Figure 1. Each symbol type represents a pair of individuals analyzed on the same day and are joined by a horizontal line in paired analysis. Median fluorescence intensities (MFI) of CD25 stainingon (A) mTconv, (B) CD4^loCD25⁺ Tregs, (C) CD4^loCD25⁺CD45RA⁻ Tregs and (D) CD4^loCD25⁺CD45RA⁺ Tregs. Statistical significance was determined using two-tailed paired Student t test.

mTconv and Tregs from individuals with the protective P1P1 IL2RA haplotype show increased sensitivity to IL-2

Sensitivity to IL-2 was assessed by measuring phosphorylation of STAT5a in all T-cell populations after brief in vitro exposure to IL-2 (Figures 3A-B). As might be predicted from their CD25 levels, sensitivity to IL-2 in this assay was lowest for nTconv, with mTconv, CD4^loCD25⁺CD45RA⁺ Treg, CD4^loCD25⁺CD45RA⁻Treg and CD25^hi Treg showing successively higher sensitivities, especially at very low concentrations of IL-2 (Figure 3C).

Figure 3. Example of phospho-STAT5a staining in Tconv and Treg populations from PBMCs.

PBMCs were incubated with various concentrations of IL-2 for 10 minutes, fixed and stained for CD4, CD25, CD45RA and pSTAT5a (Y694). A+B.Representative plot of CD4⁺ cells form one individual incubated with (A) 0 U/ml and (B) 10 U/ml IL-2. C. Representative example of a dose response curve in Tconv and Treg populations from one individual; open triangles represent nTconv, closed triangles represent mTconv, filled squares represent CD4^loCD25⁺CD45RA⁺ Treg, closed circles represent CD4^loCD25⁺CD45RA⁻ Treg and open squares represent CD25^hi Tregs.

Memory Tconv from individuals with the P1P1 IL2RA haplotype were more responsive to IL-2 at both 10 U/ml (P=0.01) and 100 U/ml (P=0.007) compared to individuals with the SS haplotype, however no such difference was observed in nTconv (Figures 4A-D). In addition, we observed that at low concentrations of IL-2 CD4^loCD25⁺Tregs from P1P1 individuals were more responsive to IL-2compared to individuals with the SS haplotype (Figure 4E, 0.1 U/ml, P=0.01). This difference was observed in both the CD45RA⁺ and CD45RA⁻populations (Figures 4F-G, P=0.005 and P=0.03, respectively). Interestingly, these significant differences were only observed at the lowest concentration of IL-2 but not at higher concentrations of 1, 10 or 100 U/ml (data not shown). Similarly, we observed that CD25^hi Tregs from P1P1 individuals were more responsive to IL-2compared to individuals with the SS haplotype Figure 4H). Again, this was true at low concentrations of IL-2 (0.1 U/ml, P=0.02) but not at higher concentrations (data not shown).

Figure 4. Relationship between IL2RA haplotype and STAT5a phosphorylation in Tconv and Treg populations from PBMCs.

PBMCs from donors with the protective rs12722495 IL2RA haplotype (P1P1) and donors with the susceptible IL2RA haplotype (SS) were isolated and analyzed as described in Figure 3. Populations of conventional and regulatory cells were defined as described in Figure 1. The % of cells positive for pSTAT5a in nTconv (A+B), mTconv(C+D), CD4^loCD25⁺ Treg(E), CD4^loCD25⁺CD45RA⁺ Treg(F), CD4^loCD25⁺CD45RA⁻ Treg (G) and CD4⁺CD25^hi Treg(H) following exposure to IL-2 as indicated. Each symbol type represents a pair of individuals analyzed on the same day and are joined by a horizontal line. Results are expressed as the median fluorescence intensity of pSTAT5a staining for all cells within a given population following exposure to IL-2 as indicated. Statistical significance was determined using a two-tailed paired Student t test.

In order to confirm and extend these findings we developed a whole blood assay incorporating additional bona fide markers of Tregs, i.e. the transcription factor FoxP3, to examine IL-2 responsiveness (Supplemental Figure 2). In addition to providing a more definitive marker for Tregs, examination of the expression level of FoxP3 combined with expression of CD45RA also allowed the delineation of three different populations of FoxP3⁺ T cells as described by Sakaguchi and colleagues (26).

An example of the gating strategy used to identify FoxP3⁺ Tregs, the division of this population into resting Tregs (rTregs, FoxP3⁺CD45RA⁺), memory Tregs (mTregs, FoxP3⁺CD45RA⁻) and activated Tregs (aTregs, FoxP3^hiCD45RA⁻) and the relative levels of CD25 expression on these populations is shown in Figures 5A-C. Again sensitivity to IL-2 in this assaywas related to the relative expression levels of CD25 within each T-cell population, especially at very low concentrations of IL-2 (Figure 5D and Supplemental Figure 2). This assay was then deployed on a fresh cohort of 13 pairs of individuals with the P1P1 or SS IL2RA haplotype consisting of six of the original pairs studied above and seven new pairs. Consistent with the results in isolated PBMCs, staining in whole blood demonstrated that P1P1 individuals express significantly more CD25 on mTconv (P=0.001) and antigen experienced FoxP3⁺mTreg (P=0.045) and FoxP3^hiaTreg (P=0.0004) compared to individuals with the SS haplotype, however, no such difference was observed in rTregs(Figures 6A-D).Similarly, results from the whole blood assay again demonstrated that mTconv and FoxP3^hiaTreg from individuals with the P1P1 IL2RA haplotype were more responsive to IL-2 at non-saturating doses (mTconv 4 U/ml, P=0.04; aTreg 0.3 U/ml P=0.02) compared to individuals with the SS haplotype (Figures 6E and 6H, respectively).However, no significant difference was observed in these populations at higher concentrations of IL-2 (mTconv 10 or 100 U/ml; aTreg 1-100 U/ml, data not shown).The lack of a significant difference between the P1P1 and SS donors in rTregs and mTegs (Figures 6F and 6G, respectively) reflects the fact that in three pairs the SS cells were more responsive to IL-2 than the P1P1 cells. This reflects the fact that in addition to IL2RA, many other genes, including PTPN2 (26), contribute to variation in the signalling pathways that mediate the phosphorylation and dephosphorylation of STAT5a in response to IL-2.

Figure 5. Example of FoxP3-pSTAT5a staining in Tconv and Treg populations from whole blood.

Whole bloodwas incubated with various concentrations of IL-2 for 10 minutes, fixed, permeabilizedand stained for CD4, CD25, CD45RA, FoxP3 and pSTAT5a (Y694). A+B. Example of the gating strategy to identify Tconv and FoxP3⁺ Treg subsets. Lymphocytes identified by their forward and side scatter properties were gated for CD4⁺ expression and then examined for expression of CD25 and FoxP3. FoxP3⁺ Tregs were identified by co-expression of CD25 and FoxP3 (A) and then divided into resting Tregs (rTregs, FoxP3⁺CD45RA⁺), memory Tregs (mTreg, FoxP3⁺CD45RA⁻) and activated Tregs (aTregs, FoxP3^hiCD45RA⁻) (B). C. Analysis of CD25 expression levels from whole blood staining in mTconv (dotted line), FoxP3⁺rTregs (dashed line), FoxP3⁺mTregs (solid line) and FoxP3^hi aTregs (filled histogram). D. Representative example of a dose response curve in mTconv and FoxP3⁺ Treg populations from one individual; closed circles represent mTconv, filled squares represent FoxP3⁺rTregs, open/closed squares represent FoxP3⁺ mTregs and open squares represent FoxP3^hi aTregs.

Figure 6. Relationship between IL2RA haplotype, CD25 expression and STAT5a phosphorylation in response to IL-2 in T-cell subsets identified by FoxP3-pSTAT5a staining in whole blood.

Whole blood from donors with the protective rs12722495 IL2RA haplotype (P1P1) and donors with the susceptible IL2RA haplotype (SS) was stimulated and analyzed as described in Figure 5. A-D.Relationship between IL2RA haplotype and CD25 expression.MFI of CD25 stainingon (A) mTconv, (B) FoxP3⁺rTregs, (C) FoxP3⁺mTregs and (D) FoxP3^hi aTregs. E-HRelationship between IL2RA haplotype and pSTAT5a expression. MFI of pSTAT5a on (E) mTconv, (F) FoxP3⁺rTregs, (G) FoxP3⁺mTregs and (H) FoxP3^hi aTregs following exposure to IL-2 as indicated. Each symbol type represents a pair of individuals analyzed on the same day and are joined by a horizontal line. Results are expressed as the median fluorescence intensity of pSTAT5a staining for all cells within a given population. Statistical significance was determined using a two-tailed paired Student t test.

Summarising these findings, our results show that antigen-experienced CD4⁺conventional T cells and Tregs from individuals with the protective P1P1 IL2RA haplotype showed increased sensitivity to IL-2.

Tregs from individuals with the protective P1P1 IL2RA haplotype maintained higher levels of FoxP3 in the presence of IL-2

Lymphocytes were gated for CD4⁺CD14⁻ expression, examined for expression of CD25 and CD127 and sorted into regulatory (CD4⁺CD14⁻CD25⁺CD127^−/lo) and conventional T-cell populations (non-Treg gate) as previously described (28) and shown in Figure 7A. We then measured the expression of FoxP3 in isolated Tregs both immediately post-flow cytometric sorting and after culture for 48hours in various concentrations of IL-2. Tregs were also stained for Helios, a member of the Ikaros family of zinc finger transcription factors that is expressed at high levels in thymic-derived Treg cells but not peripherally induced Tregs or activated Tconv (29). An example of the staining is shown in Figure 7B. Experimental conditions were selected that represent non-saturating (2U/ml) and saturating concentrations (20U/ml) of IL-2 as determined in preliminary experiments (Figure 8A-D). Immediately post-sort there was no significant difference in the proportion of Tregs that expressed FoxP3 (mean 80.1 ± 7.8% standard deviation (SD) and 79.6 ± 7.9%SD, respectively) or those that expressed Helios (mean 73.3 ± 7.4%SD and 77.6 ± 6.1%SD, respectively) between individuals with the P1P1 or SS IL2RA haplotype (data not shown). As previously reported, culture of Tregs in the absence of IL-2 resulted in a decrease in both the percentage of cells expressing FoxP3 and the level of expression. FoxP3 expression can be “rescued” by addition of IL-2 (22, 24). Our results indicated that FoxP3 maintenance under these conditions is significantly dependent upon IL2RA haplotype. We show that Helios⁺ Tregs from individuals with the protective P1P1 IL2RA haplotype expressed significantly higher levels of FoxP3 under conditions of limiting IL-2 (0U/ml, P=0.03 and 2U/ml, P=0.04; Figure 8E-H). Taken together, these data indicate that Tregs from individuals with the protective P1P1 IL2RAhaplotype maintained higher levels of FoxP3 in the presence of limiting concentrations of IL-2.

Figure 7. Example of the gating used for the isolation of CD4⁺ T-cell populations for functional studies and staining of isolated Treg populations to investigate maintenance of expression of FoxP3 and Helios.

A. Lymphocytes identified by their forward and side scatter properties were gated for CD4⁺CD14⁻ expression and then examined for expression of CD25 and CD127 (inset plot shows isotype control staining). Regulatory T cells were isolated based on CD4⁺CD14⁻CD25⁺CD127^−/lo. B. Isolated Tregs were fixed, permeabilized andstained for expression of the transcription factors FoxP3 and Helios (dot plot) or relevant isotype control (density plot).

Figure 8. Helios and FoxP3 staining in isolated Treg populations cultured with limiting concentrations of IL-2.

A-D. Example of Helios and FoxP3 staining in freshly isolated and cultured Tregs:Tregs were isolated by flow cytometry sorting as described in Figure 7 andfollowing fixation,stained for expression of the transcription factors FoxP3 and Helios either immediately post-sorting (A)or following culture for 48 hr in 0 U/ml IL-2 (B), 2 U/ml IL-2 (C) or 20 U/ml IL-2 (D). Quadrant gates were set based on staining with the relevant isotype controls (99^th centile) and FoxP3 MFI indicates the median fluorescence intensity of FoxP3 staining in the Helios⁺ Tregs. E-H. FoxP3 expression in Helios⁺ Tregs under suboptimal IL-2 concentrations:Tregs were isolated fromdonors with the protective rs12722495 IL2RA haplotype (P1P1) and donors with the susceptible IL2RA haplotype (SS) and analyzed for expression of FoxP3 and Helios either immediately post-sorting (E)or following culture for 48 hr in 0 U/ml IL-2 (F), 2 U/ml IL-2 (G) or 20 U/ml IL-2 (H) as described above. Each symbol type represents a pair of individuals analyzed on the same day and are joined by a horizontal line. Statistical significance was determined using one-tailed Wilcoxon matched-pairs signed rank test.

Tregs from individuals with the protective P1P1 IL2RA haplotype show increased suppression of autologous mTconv

In conventional “Shevach” suppression assays, Tregs from individuals with the protective P1P1 IL2RA haplotype displayed higher levels of suppression of autologous mTconv compared with the SS IL2RA haplotype (Figure 9A). A similar difference in suppression was also observed at lower ratios of Treg:mTconv in pairs of individuals who demonstrated a high degree of difference at the 1:1 ratio (Figure S3), however, the difference between the two groups of pairs did not reach significance at lower ratios (data not shown). Importantly, the difference in suppression between P1P1 and SS groups only seen in co-cultures containing mTconv and autologous Treg but was not seen when a standard third party population of Tregs was used in the place of autologous Tregs (Figure 9B) or when nTconv were used in the place of mTconv (Figure 9C).

Figure 9. Percent suppression of mTconv proliferation by autologous or third party Tregs.

Tregs and Tconv were isolated fromdonors with the protective rs12722495 IL2RA haplotype (P1P1) and donors with the susceptible IL2RA haplotype (SS). The suppression of proliferation of Tconv by Tregs was measured by in vitro co-culture. Plots shows (A) suppression of mTconv by autologous Tregs, (B) suppression of mTconv by standard third party Tregs and (C) suppression of nTconv by autologous Tregs. Each symbol type represents a pair of individuals analyzed on the same day and are joined by a horizontal line. Statistical significance was determined using one-tailed Wilcoxon matched-pairs signed rank test.

Discussion

In the present study we addressed the impact of an IL2RA haplotype associated with autoimmune T1D on Treg fitness and suppressive function. We showed that the presence of a disease-associated (SS) IL2RA haplotype leads to diminished IL-2-responsiveness, resulting in lower levels of FoxP3 expression by Tregs, and a reduction in their ability to suppress proliferation of autologous mTconv cells. These data offer a rationale that potentially accounts for the molecular and cellular mechanisms through which polymorphisms in the IL-2RA gene impact upon immune regulation, and consequently upon susceptibility to autoimmune and inflammatory diseases.

Numerous reports have demonstrated that CD4⁺CD25⁺ Tregs from individuals with a range of human autoimmune diseases, including T1D,are deficient either in their frequency or in theirability to control autologous pro-inflammatory responses when compared to non-diseased, control individuals (5, 6, 30-36). However, there is a major gap in current knowledge as to how CD4⁺CD25⁺ Treg function relates to the development of human autoimmune disease in terms of causality: is Treg dysfunction a primary, causal event or itself a result of alterations in the immune system due to the disease process? We hypothesized that if a diabetes-susceptibility haplotypecould be linked to altered (decreased) Treg function in individuals with no history of autoimmune disease, this would support the proposal that Treg dysfunction is causal in T1D. Several genes within the IL-2/IL-2RA pathway have been identified which influence susceptibility to human autoimmune diseases (e.g.IL2, IL2RA, IL2RBand PTPN2) (11, 37-42), and given the vital importance of IL-2 and IL-2 signaling to the generation and function of CD4⁺CD25⁺FoxP3⁺ Tregs, it is likely that that these genes exert their effects by altering Treg frequency or functional ability. In this and our previous report (12), the diabetes susceptible IL2RA haplotype (rs12722495) is shown to confer reduced CD25 expression and IL-2 secretion by mTconv cells. Although the frequency of Tregs detectable in the peripheral blood is not influenced by rs12722495, the ability of Tregs to signal via IL-2 is altered by this polymorphism and function is thereby impaired. This constellation of findings is resonant with observations made in patients with T1D: the frequency of Treg populations is normal, but Treg function is reduced (5, 6, 9). In addition, a number of recent studies show that patients with T1D have impaired IL-2 signaling and increased Treg apoptosis (7, 24). This implies that impairment of the IL-2 pathway, through its effect on Treg fitness and immune regulation, is a key pathway in autoimmune disease pathogenesis.

Binding of IL-2 to its high affinity receptor complex leads to a cascade of signaling events including activation of the Ras/MAPK, JAK/STAT and PI 3-kinase/Akt pathways. In Tregs a major cellular consequence of IL-2 signaling is the phosphorylation and activation of STAT5a, which binds to the FOXP3 promoter leading to sustained FoxP3 expression and enhanced suppressive capacity (19, 21, 43). IL-2 signaling and STAT5a phosphorylation are also vital for the generation of induced FoxP3⁺ Tregs (iTregs) (22). In our studies, mTconv T cells from individuals with the susceptible IL2RA haplotype had a lower level of pSTAT5a in response to IL-2 stimulation. A key question is the extent to which this might also impact upon the ability of an individual to respond to immune regulatory cues and generate iTregs. Naïve conventional T cells (which express lower levels of CD25and in whom the majority of IL-2 induced STAT5a phosphorylation will occur in a CD25 independent manner) from IL2RA susceptible individuals did not show a reduced pSTAT5a response. However, our in vitro studies did not take into account of the fact that priming of naïve T cells in vivo requires TCR ligation, a process which alters CD25 expression and hence IL-2 responsiveness. It remains to be established, therefore, whether lower levels of IL-2 sensitivity can result in reduced ability to generate iTregs from naïve T cells.

The definitive identification of Tregs by flow cytometry is challenging as activated T cells share many of the phenotypic characteristics of Tregs. As fixation precluded the use of CD127 when measuring pSTAT5a in PBMC samples, we adopted a highly conservative approach to identify CD25^hi Tregs by gating on the top 1% of CD25-staining CD4⁺ T cells, an approach adopted by other researchers (7, 25). Gating on an analogous population in the unfixed samples demonstrated this population consisted of almost exclusively CD127^−ve cells, and we were therefore confident that the observed difference in STAT5a phosphorylation is due to a difference in sensitivity to IL-2 mediated signaling in Tregs rather than in any contaminating conventional T cells. These findings were then confirmed in a separate cohort of individuals using a whole blood assay that facilitated the simultaneous detection of FoxP3 and pSTAT5a to allow for more definitive gating of bona fideTregs based on expression of FoxP3. Importantly, in both groups of individuals tested, this difference was only observed under non-saturating concentrations of IL-2, suggesting that there was not an inherent difference in the ability of STAT5a to be phosphorylated in individuals with the diabetes susceptible IL2RA haplotype, but rather a reduced sensitivity to IL-2 signaling that is only apparent under conditions of limiting IL-2.

The decreased STAT5a phosphorylation observed in individuals with the susceptible IL2RA haplotype also translated to a relative inability to maintain high levels of FoxP3 expression under limiting conditions of IL-2. Maintained expression of high levels of FoxP3 are a requirement for sustained suppressive function and down regulation of FoxP3 has been associated with a rapid loss of regulatory function (44, 45). This difference was only observed in Tregs co-expressing Helios, recently identified as a transcription factor that is proposed to identify thymic-derived FoxP3⁺ Tregs (nTregs) and distinguish them from those that are generated in the periphery (iTregs) (29). It is possible that nTregs and iTregs have different requirements for sustained IL-2 signaling to maintain high levels of FoxP3 and that, at the concentrations of IL-2 chosen for this study, a difference could only be observed in the nTregs.

In the light of these and other studies, a model can be elaborated to explain the link between the IL-2 pathway gene polymorphisms and the development of human autoimmunity. This proposes that diminished Treg function is influenced by the IL-2/IL-2RA signaling pathway via two distinct immunophenotypes: lower levels of IL-2 production by conventional T cells and a lower level of responsiveness to signaling via IL-2 in Tregs, the combined result of which is impaired immune regulation (Figure 10). This ‘dual effect’ may not grossly affect Treg populations under steady state conditions (e.g. as measured in peripheral blood), but may only be revealed under conditions of inflammation, such as those present in the pancreas and pancreatic lymph nodes during development of T1D (46). The net effect would be an inability to sustain the suppressive action of activated Tregs and a subsequent failure to suppress pathogenic autoimmunity.

Figure 10. A model to explain the link between IL-2 pathway gene polymorphisms and the development of autoimmunity.

Data from the present study and from Dendrou and colleagues (12) suggests that the rs12722495IL2RA susceptibility genotype may influence two distinct immunophenotypes: less IL-2 production from conventional memory T cells and a reduced sensitivity to IL-2 signaling in Tregs resulting in lower levels of pSTAT5a and reduced expression levels of FoxP3. We propose that that these two immunophenotypes could act synergistically to reduce Treg function.

In summary, we have demonstrated that a disease-susceptible IL2RAhaplotype is associated with reduced Treg fitness and suppressive function in vitro. This represents the first demonstration that a gene polymorphism that increases susceptibility to autoimmune diseases such as T1D is associated with altered Treg function. Further studies will be required to determine whether other susceptibility genes also contribute to the defective Treg function that characterizes individuals with autoimmunity.