STAT3 is a critical cell-intrinsic regulator of human unconventional T cell numbers and function

To determine whether functional STAT3 deficiency affects the numbers of circulating unconventional T cells, we analyzed PBMCs from AD-HIES patients and normal controls. This showed a 20-fold reduction in iNKT (CD3⁺Vα24⁺Vβ11⁺) cells in STAT3 mutant individuals (Fig. 1 A). Similarly, we observed a fourfold decrease in the percentage of MAIT cells as identified both by expression of the invariant Vα7.2 TCRα chain and high levels of CD161 (Fig. 1 B) or by using MR1 tetramers loaded with 5-OP-RU, the riboflavin metabolites recognized by MAIT cells (Fig. 1 C; Reantragoon et al., 2013; Corbett et al., 2014). We assessed the phenotype of the MAIT cells and observed no difference in the percentages of cells that had down-regulated CD45RA (control: 94.1 ± 1.6% [n = 11] vs. STAT3^MUT: 93.8 ± 1.6% [n = 8]), nor a selective loss of any particular MAIT cell subset in the STAT3 mutant individuals based on CD8α and CD4 expression (CD8⁺: control 84.5 ± 2.4%, STAT3^MUT 85.4 ± 2.3%; CD4⁺: control 2.1 ± 0.5%, STAT3^MUT 3.7 ± 1.6%; DN: control 12.0 ± 2.3% [n = 12], STAT3^MUT 8.6 ± 1.4% [n = 9]). This established that the reduction in MAIT cells caused by STAT3 deficiency was not caused by the loss of a particular subset, but rather by a global reduction in all subsets, at least as defined by these phenotypic characteristics. This dramatic decrease in NKT and MAIT cells suggests that STAT3 regulates the generation and/or survival of both of these unconventional T cell populations.

In contrast, the frequency of γδ T cells was not significantly different between normal controls and STAT3-deficient individuals (Fig. 1 D). As the different TCRδ chains are associated with responses to different pathogens (Li et al., 1996; Fenoglio et al., 2009; Chien et al., 2014), we also examined the relative proportions of δ2⁺ and δ1⁺ T cells to ascertain whether STAT3 deficiency selectively affects a particular γδ T cell subpopulation. However, our analysis showed that STAT3 deficiency had no significant effect on the percentage of δ1- or δ2-expressing cells (Fig. 1, E and F), suggesting STAT3 is not critical for development or survival of γδ T cells in vivo.

Although our results clearly show that STAT3 mutations result in a decrease in iNKT and MAIT cells, it does not necessarily follow that there is a cell-intrinsic requirement for STAT3 in their survival or development. Indeed, the reduction in these cells could result from a defect in supporting cells that require STAT3 or as a result of the chronic infections associated with AD-HIES.

To delineate whether the paucity of iNKT and MAIT cells in AD-HIES is caused by a cell-intrinsic function of STAT3, we took advantage of two individuals who have been identified to have an intermediate AD-HIES phenotype caused by somatic mosaicism at one STAT3 allele (Siegel et al., 2011; Hsu et al., 2013). In these individuals ∼50% of their immune cells express mutant STAT3, whereas the other half express the WT genotype and thus have normal functioning STAT3 (Siegel et al., 2011; Hsu et al., 2013). Samples from these individuals offer the unique opportunity to examine normal and STAT3-deficient unconventional T cells in the same environment, thereby allowing the comparison of cell-extrinsic and -intrinsic effects of STAT3 mutations. Thus, if STAT3 has a critical intrinsic role in development or survival of unconventional T cells, cells with the WT genotype will have a distinct selective advantage and would therefore comprise a far greater proportion of the circulating T cells. To test this hypothesis, unconventional T cells were sorted from PBMCs from two STAT3 mosaic individuals, and the frequency of expression of the WT and mutant STAT3 alleles was determined by quantitative PCR (qPCR). As it has previously been observed that conventional CD4⁺ and CD8⁺ memory, but not naive, T cells have a cell-intrinsic dependency on STAT3 (Siegel et al., 2011), we included these cell types as controls. The populations of naive T cells were composed of approximately equal numbers of cells expressing each genotype, whereas the memory populations were composed almost entirely of cells with the WT genotype (Fig. 2 A). Strikingly, the unconventional T cell populations comprised almost entirely WT cells (Fig. 2 A). We confirmed the enrichment of the WT allele in these populations from one mosaic individual by Sanger sequencing of genomic DNA at the position of the 1145G>A SNP (Fig. 2 B). Traces from the healthy control showed only the WT G at position 1145, whereas an AD-HIES patient harboring this same mutation in the germline showed equal G and A peaks consistent with the heterozygous mutations in AD-HIES. Although traces from the unconventional T cells showed the same genotype as the WT control with no A peak visible at the mutation point, naive CD4⁺ and CD8⁺ T cells were clearly mosaic at this nucleotide, confirming the results of the qPCR. Together, these results unequivocally demonstrate that STAT3 functions as a crucial intrinsic mediator of NKT and MAIT cell frequency.

Given the role of STAT3 in regulating the numbers of iNKT cells and MAIT cells, we considered whether STAT3 might control expression of transcription factors important for development of these unconventional T cells. PLZF is a transcription factor highly expressed by both MAIT and NKT cells, which controls the development and preactivated phenotype of these cells (Kovalovsky et al., 2008; Savage et al., 2008; Eidson et al., 2011). Thus, we determined whether the levels of this molecule were altered in STAT3-deficient MAIT cells. Consistent with previous results (Savage et al., 2008), MAIT cells from normal controls expressed significantly more PLZF than other T cells (Fig. 3 A). Strikingly, STAT3-deficient MAIT cells expressed similar levels of PLZF as normal controls. As a result of the paucity of iNKT cells in STAT3-deficient patients, we were not able to determine the levels of PLZF in these cells.

RORγt is a transcription factor key to the differentiation of Th17 cells and IL-17 production (Ivanov et al., 2006); however, it is also highly expressed by MAIT cells (Dusseaux et al., 2011). Furthermore, induction of RORγt in CD4⁺ T cells requires STAT3 such that patients with AD-HIES lack CD4⁺ Th17 cells in vivo and naive CD4⁺ T cells from these individuals fail to up-regulate RORγt and produce Th17-type cytokines after in vitro Th17-polarizing conditions (de Beaucoudrey et al., 2008; Ma et al., 2008; Milner et al., 2008). Thus, we reasoned that STAT3 deficiency might also ablate RORγt expression in MAIT cells, leading to their decrease in AD-HIES patients. However, RORγt was similarly expressed by normal and STAT3-deficient MAIT cells, as determined both by intracellular staining (Fig. 3 B) and qPCR for RORC transcripts (Fig. 3 C). We also observed no significant difference in expression of other transcriptional regulators of T cell differentiation, such as Eomesodermin and Tbet, between STAT3-deficient and control MAIT cells (not depicted). Thus, it is unlikely that the reduction in MAIT cells (and by extension NKT cells) is caused by a deregulation in either PLZF or RORγt. Rather it is likely that the deficiency of NKT and MAIT cells stems from defects in survival or expansion of these cells.

Furthermore, these results suggest that, unlike CD4⁺ Th17 cell differentiation, RORγt is acquired independently of STAT3 in MAIT cells. In mice, populations of RORγt⁺ IL-17⁺ NKT and γδ T cells have been described that appear to be preprogrammed in the thymus (Coquet et al., 2008; Rachitskaya et al., 2008; Shibata et al., 2011), and at least for γδ T cells, this has been shown to occur independently of STAT3 signaling (Shibata et al., 2011). Thus, it is possible that a similar STAT3-independent RORγt up-regulation occurs during thymic development of human MAIT cells.

The normal expression of RORγt by STAT3-deficient MAIT cells, coupled with normal levels of other markers of IL-17–producing cells such as CD161 and CCR6 (Fig. 1 B and not depicted) led us to investigate whether these cells were also capable of expressing normal levels of IL-17. To address this, δ2⁻ and δ2⁺ γδ T cells, MAIT cells and total CD4⁺ and CD8⁺ T cells were sorted from patients and normal controls and stimulated with PMA/ionomycin. iNKT cells could not be analyzed because of their rarity both in normal controls but more strikingly in AD-HIES patients. Consistent with a previous study (Dusseaux et al., 2011), both MAIT cells and CD4⁺ T cells from normal donors produced low levels of IL-17A. Importantly, this IL-17A production was abrogated by STAT3 deficiency (Fig. 4 A), suggesting STAT3 plays a crucial role in inducing IL-17 not only in conventional CD4⁺ T cells but also in MAIT cells. This reflected a specific defect in producing IL-17A, rather than a general dysfunction in cytokine secretion because production of IFNγ, TNF, Granzyme A and B, and IL-2 (Fig. 4, B–D; and not depicted) was comparable for STAT3 mutant and normal T cell subsets. We also analyzed γδ T cell populations to determine whether a human equivalent of the thymically programmed IL-17–producing population could be identified. However, we detected little IL-17 production from either γδ T cell population (Fig. 4 A).

Although we found that all T cells secreted high levels of IFNγ and Granzyme B (Fig. 4, B and D) in response to PMA/ionomycin, secreted levels of IL-17A and IL-17F were very low to absent (Fig. 4 A and not depicted). The STAT3-activating cytokines IL-6 and IL-23, together with TGFβ and IL-1β, induce differentiation of CD4⁺ and γδ T cells to an IL-17–secreting phenotype (Acosta-Rodriguez et al., 2007; Wilson et al., 2007; Ness-Schwickerath et al., 2010). To determine whether these cytokines promoted IL-17 production from human MAIT cells, and whether STAT3 plays a role in this process in unconventional T cells, sorted T cells (δ2⁻ and δ2⁺ γδ T, MAIT, and as controls total CD4⁺) were cultured for 5 d with T cell activation and expansion (TAE) beads (beads expressing anti-CD3, anti-CD28, and anti-CD2 mAb) alone or together with Th17-polarizing stimuli. The cells were also cultured under Th1-polarizing conditions as a control for intact responsiveness to alternative differentiation stimuli. As shown previously (de Beaucoudrey et al., 2008; Ma et al., 2008; Milner et al., 2008), Th17 culture conditions induced high levels of IL-17A and IL-17F secretion from CD4⁺ T cells, and this response was ablated in STAT3-deficient cells (Fig. 5, A and B). The Th17 culture also stimulated high levels of IL-17F and a lower level of IL-17A by MAIT cells, and this response was severely reduced in STAT3-deficient cells (Fig. 5, A and B). The level of IL-17 secretion by γδ T cells also appeared to be reduced in STAT3-deficient cells; however, this difference was not statistically significant, possibly because of the large variability among γδ T cells even from normal donors.

We also examined secretion of IFNγ and Granzyme B (Fig. 5, C and D). Interestingly, in contrast to normal secretion of IFNγ by STAT3-deficient δ2⁺ γδ T and MAIT cells in response to PMA/ionomycin stimulation (Fig. 4 B), both of these cell types displayed decreased IFNγ secretion in the nonpolarizing culture of TAE beads alone (Fig. 5 C). However, this impairment was rescued in the Th1 and Th17 cultures. The δ2⁺ γδ T also showed decreased Granzyme A and B expression (Fig. 5 D and not depicted) in Th0 cultures, which again was rescued by addition of IL-12. It is not clear what the mechanism underlying this defect in IFNγ and Granzyme secretion is as the normal production in response to PMA/ionomycin (Fig. 4) coupled with normal expression of T-bet (not depicted) suggests that the in vivo differentiation of these cells to a Th1-like phenotype is intact. Instead it may reflect the requirement for a STAT3-dependent response to endogenously produced STAT3-activating cytokines in the Th0 cultures. To determine whether STAT3 deficiency also affected T cell survival or expansion, we enumerated the number of live cells after 5 d of culture (Fig. 5 E). We saw no differences in cell recovery between controls and STAT3-deficient cultures for any cell type (Fig. 5 E). However, although we observed a significant increase in cell numbers over the starting 20,000 cells in cultures of CD4⁺ and δ2⁻ γδ T cells, consistent with substantial proliferation, very few cells were recovered from cultures of δ2⁺ T and MAIT cells. Flow cytometric analysis of forward scatter of cells from these cultures also demonstrated that only CD4⁺ and δ2⁻ γδ T cells showed characteristics of cells undergoing blastogenesis and proliferation (not depicted). Indeed, CFSE labeling of the different populations confirmed that CD4⁺ and δ2⁻ γδ T cells, but neither MAIT nor δ2⁺ γδ T cells, underwent proliferation after 5 d in culture (not depicted). Remarkably, although we recovered ∼30–40-fold fewer cells from cultures of MAIT cells compared with CD4⁺ T cells, the level of IFNγ and IL-17F secretion was only two- to fourfold lower. This may reflect the greater heterogeneity of CD4⁺ T cells. For example, IFNγ production is largely restricted to Th1 cells, whereas the majority of MAIT cells can secrete IFNγ after in vitro stimulation (not depicted).

Together these results reveal that STAT3 deficiency leads to a loss of IL-17 production by all T cell populations. It is of particular interest that although STAT3-deficient MAIT cells expressed many features of IL-17–producing cells, including CD161, CCR6, and high levels of RORγt, they were still unable to secrete appreciable levels of IL-17. This suggests that although RORγt can be up-regulated independently of STAT3 in these cells, possibly during thymic development, the action of RORγt alone is not sufficient to drive transcription of the IL17A and IL17F genes and thus induce their secretion. This is presumably the result of a requirement for STAT3 to act directly at the promoters of these genes (Durant et al., 2010).

These experiments also provide further insight into the clinical phenotype of AD-HIES, which is characterized by recurrent infection with C. albicans and S. aureus (Kane et al., 2014). Control of these infections has been associated with the actions of IL-17A and IL-17F (Puel et al., 2011; Cypowyj et al., 2012). Indeed, previous work demonstrated a lack of IL-17–producing CD4⁺ T cells in STAT3-deficient individuals (de Beaucoudrey et al., 2008; Ma et al., 2008; Milner et al., 2008), suggesting this defect in CD4⁺ T cell differentiation contributed to poor control of pathogen infection. Importantly, MAIT cells can be activated by C. albicans and S. aureus as both of these microorganisms generate the riboflavin metabolites specifically recognized by these cells (Kjer-Nielsen et al., 2012). Furthermore, γδ T cells are also implicated in responses to these organisms (Cho et al., 2010; Cua and Tato, 2010; Ness-Schwickerath et al., 2010). Thus, our results that STAT3-deficient MAIT and γδ T cells are also unable to produce IL-17 broaden these findings and suggest that functional and numerical defects in unconventional T cells contribute to the infectious susceptibility in AD-HIES. Together, these data reinforce the significance of unconventional T cells and IL-17 in the immune system and underscore the essential role STAT3 plays in their action.

STAT3 can be activated by many cytokines, including IL-6, IL-10, IL-21, and IL-23 (Kane et al., 2014). Thus, we sought to identify which cytokine signaling pathways may be active in MAIT and NKT cells. We found prominent expression of IL-12Rβ1 and IL-21R on unstimulated MAIT and NKT cells; however, these cells lacked detectable expression of IL-6R (Fig. 6 A). This suggested these cells could potentially respond to IL-12, IL-23 (both of which share IL-12Rβ1), and IL-21. To test this, we examined STAT phosphorylation in sort-purified MAIT cells that were preactivated for 48 h. IL-12 induced phosphorylation of only STAT4, and IL-21 induced phosphorylation of STAT3 (Fig. 6 B). However, IL-23 induced robust activation of both STAT3 and STAT4, implying IL-12Rβ1 is used by IL-23, rather than IL-12, to initiate STAT3-dependent signaling in human MAIT cells. To determine whether signaling through IL-21R/STAT3 or IL-23R/STAT3 was important for controlling MAIT and iNKT cells numbers, we enumerated these populations in patients with loss-of-function mutations in IL21R or IL12RB1 (Fig. 6, C and D). We also examined IFNGR1 mutant individuals as a control for immune-deficient patients. This revealed intact numbers of MAIT and iNKT cells in IFNGR1 mutant patients, but striking deficiencies of MAIT cells in individuals with IL12RB1 mutations (Fig. 6 C) and of iNKT cells in IL-21R–deficient patients (Fig. 6 D). Indeed, the proportions of MAIT and iNKT cells in these patients approximated those in STAT3-deficient individuals, implicating IL-23 and IL-21 as the major cytokines responsible for STAT3-dependent maintenance of human MAIT and iNKT cells, respectively (Fig. 6, C and D). Importantly, ∼25% of patients with IL12RB1 mutations present with Candida infections (de Beaucoudrey et al., 2010), consistent with a role for MAIT cells in protective immunity to Candida. Together, these data define a novel requirement for specific cytokine-induced STAT3 signaling pathways that control MAIT and iNKT cell numbers in humans. Importantly, this represents the first description of a cell-intrinsic regulator of human MAIT cells. Identification of these pathways potentially paves the way for manipulating the frequencies of these cell types to modulate immune responses in the setting of infection with specific pathogens.

Buffy coats from normal donors were from the Australian Red Cross Blood Service. Peripheral blood was also collected from patients with mutations in STAT3 (Holland et al., 2007; Chandesris et al., 2012; Deenick et al., 2013), IL12RB1 (de Beaucoudrey et al., 2010), IL21R (Deenick et al., 2013; Ives et al., 2013; Kotlarz et al., 2013; Stepensky et al., 2015), or IFNGR1 (Dorman et al., 2004) and individuals mosaic for STAT3 mutations (Siegel et al., 2011; Hsu et al., 2013). Approval for this study was obtained from the ethics committees of the St. Vincent’s Hospital and Sydney South West Area Health Service (Australia), Rockefeller University Institutional Review Board (New York), and National Institute of Allergy and Infectious Diseases Intramural Institutional Review Board (Bethesda, MD). All participants gave written informed consent in accordance with the Declaration of Helsinki.

PerCp-Cy5.5 anti-CD161, PE anti-RORγt, PE anti–γδ TCR, biotinylated anti–γδ TCR, PerCp-Cy5.5 anti-CD45RA, and APC anti–αβ TCR antibodies were from eBioscience. A647 anti-PLZF, APC-Cy7 anti-CD4, PE-Cy7 anti-CD8, BV421 anti-CD3, BV421 anti-CD161, PE anti-STAT4 (pY693), PE anti–IL-6R, PE anti–IL-12Rβ1, PE anti–IL-21R, PE mouse IgG1 isotype control, PerCP-Cy5.5 anti-STAT3 (pY705), and FITC anti–Vδ2 TCR antibodies were from BD. FITC anti–Vα24 TCR and PE and biotinylated anti-Vβ11 antibodies were from Beckman Coulter. FITC anti–Vδ1 TCR was from Thermo Fisher Scientific. APC or APC-Cy7 anti-Vα7.2 and PE-Cy7 anti–αβ TCR antibodies were from BioLegend. FITC anti-CCR7 antibody was from R&D Systems. MR1–5-OP-RU tetramers have been described previously (Reantragoon et al., 2013; Corbett et al., 2014).

PBMCs were stained with mAbs and sorted as below using a BD FACSAria or Aria III. For SNP analysis, MAIT cells were sorted as CD3⁺Vα7.2⁺CD161^hiγδ⁻, NKT cells as CD3⁺Vα24⁺Vβ11⁺, γδ T cells as CD3⁺Vα7.2⁻γδ⁺ δ2^+/−, and CD4⁺ and CD8⁺ naive T cell populations as CD45RA⁺CCR7⁺, with the remaining cells sorted as memory cells. For cell culture, MAIT cells were sorted as CD3⁺Vα7.2⁺CD161^hiγδ⁻, γδ T cells as CD3⁺Vα7.2⁻γδ⁺ and either δ2⁺ or δ2⁻, total CD4⁺ T cells as CD3⁺Vα7.2⁻γδ⁻CD8⁻CD4⁺, and total CD8⁺ T cells as CD3⁺Vα7.2⁻γδ⁻CD4⁻CD8⁺.

Cells were cultured at 20,000/well/100 µl media (Deenick et al., 2013; Ives et al., 2013). Cells were stimulated with either TAE beads (1 bead/2 cells; Miltenyi Biotec) alone (Th0 culture) or under Th1 (20 ng/ml IL-12; R&D Systems) or Th17 (2.5 ng/ml TGFβ, 20 ng/ml IL-1β [PeproTech], 50 ng/ml IL-6 [PeproTech], or 20 ng/ml IL-23 [eBioscience]) conditions for 5 d or 100 ng/ml PMA plus 750 ng/ml ionomycin (Sigma-Aldrich) for 3 d before harvesting supernatants. Cytokine secretion was measured by cytometric bead array (BD) as per the manufacturer’s instructions. The absolute number of cells in culture was determined by adding a known number of CaliBRITE beads (BD) before harvest, which can then be distinguished by flow cytometry based on forward and side scatter.

Cells were stained for surface markers to identify γδ T and MAIT cells and then permeabilized using the BD transcription factor buffer set and stained with PLZF- or RORγt-specific mAbs.

Genomic DNA was extracted from sorted cells (QIAGEN). Percent mosaicism was determined in different cell populations using the TaqMan SNP genotyping assay (Life Technologies; Hsu et al., 2013). This qPCR-based assay used primers for the region of DNA containing the disease-causing mutation, together with specific probes recognizing either the WT or disease SNPs. The difference between the cycle threshold for each probe was used to determine the proportion of the two alleles in each sample. In each assay a normal control and AD-HIES sample with the same mutation were used to normalize the data, with the normal control taken as 0% and the AD-HIES sample as 100%.

Exons 12–14 of STAT3 were amplified from genomic DNA (Forward: 5′-TAGTTTAAAGAAATGCCCAGGAGCACAGAG-3′ and Reverse: 5′-TTGGCCTAAGTGACTTTTTGGAATAACTACAGC-3′). Sanger sequencing of the amplified product was performed by Garvan Molecular Genetics.

RNA was isolated from sorted populations using the RNeasy kit (QIAGEN). For qPCR, total RNA was reverse transcribed with oligo-dT. Expression of genes was determined using real-time PCR with the LightCycler 480 Probe Master Mix and System (Roche) as previously described (Ives et al., 2013).

Sorted MAIT (CD3⁺Vα7.2⁺CD161⁺) and non-MAIT (CD3⁺Vα7.2⁻CD161⁻) cells were stimulated with TAE beads for 2 d and then rested for 2 h in media before being stimulated with 50 ng/ml IL-12, 100 ng/ml IL-21, or 100 ng/ml IL-23 or media alone for 20 min. Cells were fixed, permeabilized, and stained with anti–phospho-STAT3 (pY705) and STAT4 (pY693) mAbs (Deenick et al., 2013; Ives et al., 2013).

Significant differences between datasets were determined using either the unpaired Student’s t test when comparing two groups, one-way ANOVA with Dunnet post test for more than two groups, or two-way ANOVA with Šídák’s multiple comparison test for two variables (Prism; GraphPad Software).

We thank Rob Salomon, Eric Lam, and Vitri Dewi for cell sorting and all of the patients and their families for participating in this project.

This work was funded by project and program grants from the National Health and Medical Research Council (NHMRC) of Australia (1016953, 1066694, and 1027400 to E.K. Deenick and S.G. Tangye) and the Rockefeller University Center for 541 Clinical and Translational Science (5UL1RR024143 to J.-L. Casanova). C.S. Ma is a recipient of a Career Development Fellowship (1008820), and S.G. Tangye is a recipient of a Principal Research Fellowship (1042925) from the NHMRC of Australia.

The authors declare no competing financial interests.