Cutting Edge: Increased IL-17–Secreting T Cells in Children with New-Onset Type 1 Diabetes

The autoimmune destruction of pancreatic β cells by self-reactive T cells leads to type 1 diabetes (T1D). Evidence that regulatory CD4⁺ T cells (Tregs) suppress the activation of autoreactive T cells and maintain self-tolerance (1) has led to the hypothesis that Treg dysfunction is a major factor underlying the development of T1D. The best characterized Tregs are those that express the FOXP3 transcription factor, but although FOXP3 appears to be an accurate marker of Tregs in mice, it is also expressed by activated nonsuppressive T cells in humans (2).

IL-17 is a proinflammatory cytokine secreted by a distinct lineage of CD4⁺ Th17 cells. Th17 cells have an established pathogenic role in several autoimmune diseases (3). In animal studies, a function for Th17 in T1D is supported by the observation that IL-17 is expressed in pancreas of NOD mice (4) and that inhibition of IL-17 in this model leads to delayed onset of T1D during the effector phase of the disease (5, 6). Also, adoptive transfer of islet-specific Th17 cells into NOD/SCID mice induces diabetes, but only after conversion into a Th1-like phenotype (7, 8). In humans, the frequency of IL-17–secreting CD4⁺ T cells in lymphocytes from established T1D patients is increased compared with healthy controls (9), but their relevance in new-onset T1D has not been shown. Recently, several groups have reported that human IL-17⁺FOXP3⁺ cells can be isolated ex vivo and that IL-17 Tregs can transform into IL-17–producing cells (10–15). Of particular interest, human FOXP3⁺CD4⁺ T cells may be divided into three phenotypically and functionally distinct subpopulations, depending on the expression of FOXP3, CD25, and the human naive cell marker, CD45RA (14). Whereas naive Tregs (CD45RA⁺CD25^intFOXP3^low) and memory Tregs (CD45RA⁻CD25^highFOXP3^high) possess suppressor function, a third group of FOXP3⁺ memory T cells (CD45RA⁻CD25^intFOXP3^low) are nonsuppressive and secrete IL-17 (14).

Earlier studies have reported that T1D subjects have decreased (16), increased (17), or equivalent (18–20) proportions of Tregs compared with controls. Given the recently recognized complexity of FOXP3⁺ T cell subsets, we hypothesized that inconsistencies in the enumeration of Tregs in T1D may have been the result of inadvertent inclusion of cells that express FOXP3 but do not exhibit the classical Treg phenotype.

In this study, we show that although there is an increase in the overall proportion of FOXP3-expressing CD4⁺ T cells in new-onset T1D subjects as compared with controls, the increase is restricted to the CD45RA⁻CD25^intFOXP3^low subset that secretes significantly more IL-17 and is not suppressive compared with other FOXP3⁺ subsets. Additionally, we report that T1D subjects with new-onset disease have a globally increased proportion of both CD4⁺ and CD8⁺ T cells that secrete IL-17.

Peripheral blood (15 ml) was obtained from 64 subjects with new-onset (<6 mo from diagnosis) T1D and from 53 age-matched controls without T1D attending British Columbia Children’s Hospital (Supplemental Table 1). The study protocol was approved by the Clinical Research Ethics Board of the University of British Columbia (certificate H07-01707), and all parents or subjects provided informed written consent or assent.

PBMCs were isolated on a Ficoll-Hypaque gradient, cryopreserved, and later stained with different combinations of Abs to CD4 (RPA-T4), CD14 (M¥P9.1), CD8 (RPA-T8), and CD127 (hIL-7R-M21) (BD Biosciences, San Jose, CA), and to CD45RA (HI100) and CD25 (BC96) (eBioscience, San Diego, CA). For intracellular staining, cells were treated with Fix/Perm buffer (eBioscience) and stained with anti-FOXP3 (259/C7; BD Biosciences) or anti–IL-17A (eBio64DEC17; eBioscience). Data were acquired with FACSAria (BD Biosciences). The unpaired Student t test was used for statistical comparisons.

Freshly isolated PBMCs were sorted into different CD25⁺ subsets (fractions [Fr.] I–III) using a FACSAria (BD Biosciences). PBMCs (2 × 10⁶ cells/well) or sorted CD25⁺ fractions (2 × 10⁵ cells/well) were incubated with human anti-CD3/CD28 Dynabeads (Invitrogen, Carlsbad, CA) at a 1:1 ratio for 72 h, without polarizing cytokines, to optimize the detection of IL-17. PBMC samples were suspended in R10 media with PMA (100 ng/ml), ionomycin (1 μg/ml), and brefeldin A (10 μg/ml) (Sigma-Aldrich, St. Louis, MO) for 5 h before intracellular staining. CD25⁺ sorted cell-free supernatants were assessed for IL-17 concentration by ELISA (eBioscience).

To assess suppressive capacity, ex vivo allogeneic CD4⁺ responder T cells were stimulated at 8000 cells/well in the presence of human anti-CD3/CD28 Dynabeads (1000 beads/well). Different subsets of CD4⁺CD25⁺FOXP3⁺ T cells (Fr. I–III) were added to the responder cells, and suppressive capacity was assessed by measuring the amount of [³H]thymidine incorporation in the final 16 h of a 6-d culture. The percentage of suppression was calculated relative to the proliferation of responder cells alone (without suppressors) using the formula: [cpm for responders alone − (cpm for responders + putative suppressors)]/cpm for responders alone.

We observed that T1D subjects exhibited an increased frequency of Treg-like cells relative to controls when determined either by FOXP3 expression or CD25/CD127 levels, respectively (Supplemental Fig. 1). Human FOXP3⁺CD4⁺ T cells may be divided into three functional subsets based on the expression of FOXP3 and CD45RA (14). Therefore, we assessed the frequency of all FOXP3⁺ subsets in our T1D subjects (Fig. 1). In T1D subjects, the proportion of CD4⁺ T cells that was CD45RA⁻FOXP3^low (Fr. III) was increased 1.5-fold relative to healthy controls (T1D = 3.9 ± 0.3% versus control = 2.6 ± 0.2%; p = 0.003). In contrast, there was no significant difference in the proportion of CD4⁺ T cells that were CD45RA⁺FOXP3^low (Fr. I, T1D = 1.2 ± 0.2% versus control = 1.2 ± 0.2%; p = 0.93) or CD45RA⁻FOXP3^high (Fr. II, T1D = 1.1 ± 0.2% versus control = 0.9 ± 0.1%; p = 0.37). Thus, in T1D subjects, the increase in CD4⁺ T cells that express FOXP3 is due solely to an increase in the CD45RA⁻FOXP3^low subset.

CD45RA⁻FOXP3^low cells are known to produce substantially more IL-17 than other Treg subsets, and they are not suppressive (14). Therefore, we sought to determine whether CD45RA⁻FOXP3^low cells present in our T1D subjects also have enhanced capability to secrete IL-17 and a reduction in suppressive capability. Because cells stained for FOXP3 are not viable, we isolated subsets of FOXP3⁺ cells on the basis of CD25 and CD45RA expression (14). This was possible because CD25 levels on T1D CD4⁺ T cells correlate strongly with FOXP3 expression, allowing the isolation of the three FOXP3⁺ (Fr. I–III) subsets by sorting for CD45RA⁺CD25^int (Fr. I), CD45RA⁻CD25^high (Fr. II), and CD45RA⁻CD25^int (Fr. III) cells, respectively (Fig. 2A).

Sorted CD4⁺ T cells bearing a CD45RA⁺CD25^int (Fr. I), CD45RA⁻CD25^int (Fr. III), or CD25⁻ (Fr. IV) phenotype were stimulated with anti-CD3/CD28 beads, and IL-17 secretion was quantified by ELISA (Fig. 2B). Consistent with previous observations (14), T1D CD45RA⁻CD25^int cells secreted considerably more IL-17 than did CD45RA⁺CD25^int (137-fold increase; 548 ± 132 pg/ml versus 4 ± 0.01 pg/ml; p = 0.001) or CD25⁻ cell subsets (6.6-fold increase; 548 ± 132 pg/ml versus 82 ± 5.5 pg/ml; p = 0.0193). IL-17 secretion could not be determined on CD45RA⁻CD25^high (Fr. II) cells because of insufficient cell numbers after sorting.

To test the suppressive capacity of subsets of CD4⁺CD25⁺ T cells, cells were sorted into Fr. I, II, and III and tested for their ability to suppress the proliferation of CD4⁺ T cells. At a 1:8 ratio (putative suppressor-to-responder cells), CD45RA⁺CD25^int (Fr. I) cells and CD45RA⁻CD25^high (Fr. II) cells suppressed CD4⁺ responder cell proliferation significantly more than did CD45RA⁻CD25^int (Fr. III) (67 ± 11% and 51 ± 10% versus 18 ± 9%; n = 5, p = 0.02 and 0.003; Fig. 3). Thus, CD4⁺ T cells from our T1D subjects contain larger fractions of CD45RA⁻CD25^intFOXP3^low cells that secrete IL-17 and have lost suppressive capability.

The observation that our T1D subjects exhibited an increase in IL-17–secreting FOXP3⁺ T cells (Supplemental Fig. 2) led us to investigate whether CD4⁺ and CD8⁺ T cells were also biased toward IL-17 secretion. PBMCs from control and T1D subjects were stimulated with anti-CD3/CD28 beads for 3 d and restimulated with PMA and ionomycin before IL-17 detection by intracellular staining (Fig. 4). The respective proportions of CD4⁺ and CD8⁺ T cells that secrete IL-17 in T1D subjects were 2.6-fold and 3.1-fold greater than controls (CD4, 0.47 ± 0.11% versus 0.18 ± 0.03%; p = 0.02; CD8, 0.22 ± 0.02% versus 0.07 ± 0.03%; p = 0.004). Thus, in new-onset T1D subjects, both CD4⁺ and CD8⁺ T cell populations are biased toward IL-17 secretion.

We believe this is the first study of children with T1D to assess whether alterations exist in the proportion of FOXP3⁺CD4⁺ cells that have a regulatory versus inflammatory (IL-17) phenotype. We observed that although our T1D subjects have an increased proportion of cells that express FOXP3, the increase is limited to a subset that coexpresses IL-17 and is not suppressive. These results suggest that earlier studies enumerating FOXP3-expressing cells in the context of T1D should be re-evaluated, and our data support the hypothesis that an increase in IL-17–producing T cells underlies the pathogenesis of T1D.

In previous studies, where the data have been inconsistent regarding the proportion of Tregs in T1D subjects (16–20), the differences seen may have been the result of using CD25 alone as a marker of Tregs (18), using a control population older than that of the T1D group (21), and using a less specific FOXP3 Ab clone (PCH101) that precluded greater discrimination of FOXP3^high versus FOXP3^low cells (22). Given these differences, preceding studies that assessed FOXP3⁺ T cells in T1D subjects would not easily have identified the CD45RA⁻FOXP3^low population (14). To our knowledge, our study is the first to compare Treg cells in a large cohort of new-onset T1D using a highly specific FOXP3 Ab clone and age-matched controls.

The data observed in this study show that the increased proportion of CD4⁺ T cells expressing FOXP3 does not represent a true increase in naive (CD45RA⁺CD25^intFOXP3^low) or memory (CD45RA⁻CD25^highFOXP3^high) Tregs that are suppressive, but rather is confined to IL-17–secreting CD45RA⁻CD25^intFOXP3^low cells. These findings recapitulate those of Miyara et al. (14), who found that CD45RA⁻CD25^intFOXP3^low cells are nonsuppressive and also increased in systemic lupus erythematosus subjects. Further investigation will be required to determine if this observation extends to other autoimmune diseases.

Several groups have identified Th cells that, like CD45RA⁻CD25^intFOXP3^low cells, are IL17⁺FOXP3⁺ (10–14). However, their origin remains unknown. They may be activated CD4⁺ effector T cells that transiently express FOXP3 (2), or they may derive from Tregs that failed to maintain sufficient levels of FOXP3 expression (“ex-FOXP3” cells) and subsequently have converted to a proinflammatory IL-17–secreting phenotype (23). The well-described plasticity of the Treg and Th17 lineages (10–15) suggests that IL-17⁺FOXP3⁺ T cells may be an intermediate lineage, with a suppressive versus proinflammatory role that depends on the local cytokine milieu. A recent study in NOD mice, immunized with the tolerogen IgG-GAD1, found that FOXP3^int T cells that express the Th17 lineage transcription factor RORγt arise before islet inflammation, and they may differentiate in vitro to either Th17 or Tregs (24). The elevated population of IL-17–secreting CD45RA⁻FOXP3^low cells described in this study in humans may be an analogous intermediate cell population that lies between “true” Tregs and Th17 and exhibits plasticity. This could be assessed once the polarization conditions required to convert human FOXP3⁺IL-17⁺ cells into Tregs or Th17 are better understood.

The observation that CD4⁺ and CD8⁺ T cells from our T1D subjects secrete increased levels of IL-17 may be ascribed to the presence of a proinflammatory polarizing cytokine milieu. In support of this possibility, Bradshaw et al. (9) showed that monocytes from T1D subjects spontaneously secrete proinflammatory cytokines, necessary for Th17 cell differentiation and expansion. The skewing toward IL-17 secretion would likely disrupt the delicate balance of islet T cells in favor of autoimmune inflammatory destruction.

This paper highlights the importance of discriminating different FOXP3-expressing subsets, and the findings demonstrate that multiple T cell subsets are biased toward IL-17 secretion in T1D. Taken together, these data imply a potentially important role for IL-17 secretion in the pathogenesis of human autoimmune diabetes and suggest the possibility of therapeutic approaches that target the IL-17 axis in T1D.

We thank Pamela Lutley and Sue Baynham for subject recruitment, the Tan Laboratory members for helpful discussion, and all study subjects for participation.

Disclosures The authors have no financial conflicts of interest.