IL-1β and IL-23 Promote Extrathymic Commitment of CD27⁺CD122⁻ γδ T Cells to γδT17 Cells Open Access

This article is featured in In This Issue, p.2609

Functional phenotypes of γδ T cells, i.e., IFN-γ–producing γδT1 cells and IL-17–producing γδT17 cells, are generated during thymic development. γδT17 cells can be identified by a series of surface markers. IL-23R expression strictly segregates with IL-17 expression in γδ T cells (1). In addition, all IL-17–producing γδ T cells are contained within the CD27⁻ γδ T cell compartment and express CCR6 (2, 3). Whether a thymocyte is committed to the γδT1 versus the γδT17 lineage is most likely dependent on various cues. γδ TCR engagement has been suggested to dictate the commitment of thymocytes to the γδT1 lineage (4). In contrast, γδT17 cell development appears to be independent of TCR engagement as IL-17 production is detected in a subset of thymocytes that are poised to become γδ T cells but have not yet rearranged their δ and γ TCR genes (5). Thus, other cues, including Notch-dependent pathways but also cytokine signals, might be determinants of γδT17 commitment. For example, the Notch–Hes1 pathway and the noncanonical (RelB) NF-κB pathway as well as, in a γδ thymocyte–extrinsic manner, NIK and RelA expression in the thymus are required for the generation of γδT17 cells (6–8). In contrast, IRF4, which is critically required for Th17 cell differentiation, is dispensable for the development of γδT17 cells (9). Although probably irrelevant for the de novo generation of γδT17 cells in the thymus, STAT3 is required for the expansion of γδT17 cells, and the selective expansion of γδT17 cells by IL-7 was reported to depend on STAT3 (10). In contrast, IL-17 production most likely from αβ thymocytes might negatively regulate γδT17 generation and was proposed to be responsible for shutting down γδT17 generation in the adult thymus (5).

Once exported from the thymus, γδT1 cells and γδT17 cells populate secondary lymphoid organs and certain peripheral tissue niches. For example, the dermis harbors γδT17 cells that expand upon exposure of the skin to imiquimod cream (Aldara) and cause a psoriasis-like pathology (11). Recently, it was demonstrated that early after birth, the dermis mostly contains TCR-Vγ6⁺ γδT17 cells, whereas in adult individuals more TCR-Vγ4⁺ than TCR-Vγ6⁺ γδT17 cells populate the dermis (12). TCR-Vγ6⁺ γδT17 cells are thymically imprinted to home to the dermis, whereas TCR-Vγ4⁺ γδT17 cells acquire their skin-homing properties extrathymically. Yet, both subsets of dermal γδ T cells are preimprinted to produce IL-17 and not IFN-γ. γδT17 cells exhibit a permissive epigenetic landscape and can convert into IFN-γ producers in vitro and in vivo in the context of distinct inflammatory environments (13). In contrast, IFN-γ–producing γδT1 cells have been proposed to exhibit an epigenetic signature indicative of a “terminally committed” phenotype which precludes plasticity to convert into IL-17 producers (13). It is commonly believed that γδT17 cells develop in the thymus during an early ontogenetic window shortly before birth and, therefore, the possibility of a secondary de novo generation of γδT17 cells outside the thymus has thus far been dismissed. Rather, it is thought that postnatally, the initial pool of γδT17 cells is maintained through self-renewal, perhaps in response to homeostatic cytokine cues including IL-7 (10).

γδT17 cells have most recently been shown to participate in the immunopathology of a variety of inflammatory diseases in epithelial barrier tissues, including inflammatory bowel disease, psoriasis, and asthma (11, 14, 15), but also in spondylarthropathies and CNS inflammation in mouse models (1, 16) and in humans (17–19). Because the presence of γδT17 cells in certain tissues might set the threshold for these tissues to respond directly to pathogen-associated molecular patterns and indirectly to the proinflammatory cytokine IL-23, the regulation of γδT17 cell populations during homeostasis and in inflammation is highly relevant in the context of immune interventions in autoimmune diseases, chronic inflammation, and inflammation associated with malignant disease. In this study, we show that under homeostatic conditions, the adult thymus has largely lost its capacity to produce γδT17 cells. We found that thymic development of γδT17 cells may be partially restricted by sensing IL-2. However, under inflammatory conditions, γδT17 cells can also be de novo generated in the peripheral immune compartment from IL-23R⁻ precursor γδ T cells. Notably, IL-23 is indispensable for this process. De novo–generated γδT17 cells are bona fide γδT17 cells and participate in inducing immunopathology in an in vivo model of psoriasis.

Nur77^GFP reporter mice (20) and Il23r^gfp/+ mice (21) were previously described. Because the GFP-reporter cassette is knocked into the endogenous Il23r locus, Il23r^gfp/gfp mice are functional Il23r^−/− mice. The Il17f-Cre^eyfp fate mapping system to track current and historic IL-17F–producing cells in mice has been published (22, 23). C57BL/6 mice, CD45.1 congenic C57BL/6 mice (B6.SJL-Ptprc^aPepc^b/BoyJ, stock number 002014), Il2ra^−/− mice, and Rag1^−/− mice were obtained from the Jackson Laboratory. CD45.1 congenic mice from the Jackson Laboratory do not carry the Sox13 mutation that is associated with decreased endogenous frequencies of γδT17 cells (24). Analysis of fetal thymic Il2 transcript levels was performed in embryos of timed-pregnant mice. All animals were kept in a conventional, specific pathogen-free facility at the Technical University of Munich, and all experiments were carried out in accordance with the guidelines prescribed by the Bavarian state authorities (Az 55.2-1-54-2532-29-13 and 55.2-1-54-2532-1-15).

For the generation of bone marrow chimeras or fetal liver chimeras, Rag1^−/− recipient mice were lethally irradiated. The 7 Gy total dose was delivered as two 3.5-Gy doses 3 h apart. The donor bone marrow was always depleted of CD90⁺ cells prior to transfer to get rid of residual mature αβ and γδ T cells. A total of 5 × 10⁶ (or 1 × 10⁷ when indicated) bone marrow cells were injected i.v. into recipients 1 d after irradiation. For fetal liver cell transplantation, fetal mice were prepared at day 13.5 of gestation. Liver cells (9 × 10⁶) were transferred 1 d after irradiation. Reconstituted mice were maintained on antibiotic water (Enrofloxacin, Bayer, 0.1 mg/ml) for 2 wk after cell transfer. Reconstitution of the hematopoietic system was tested in the peripheral blood.

The mouse back was shaved and depilated (Veet; Reckitt Benckiser). A dose of 50 mg of Aldara (5% IMQ cream; 3M Pharmaceuticals) or vehicle control (Vaseline [white], European Pharmacopoeia; Bombastus Werke, Freital, Germany) was applied on the back for 4–5 consecutive days. Back skin thickness was measured daily in triplicates using a micrometer (Mitutoyo).

To obtain single-cell suspensions from thymi, spleens, and peripheral lymph nodes, tissues were dissected and grated through a nylon mesh with 70-μm pores (Falcon). RBC lysis was performed using BD Pharmlyse Buffer (BD Biosciences).

Single-cell suspensions were incubated with Fc-blocking Abs (αCD16/αCD32; BD Biosciences) and subsequently stained with 1:100 dilutions of the respective surface marker Abs for 30 min at 4°C. Staining was carried out with Abs to CD3 (145-2C11), CD4 (RM4-5), and CD27 (LG.3A10) from BD Biosciences; CD45.1 (A20), CD45.2 (104), and CD122 (TM-β1) from eBioscience; CCR6 (140706, R&D), TCR γδ (GL-3), TCR-Vγ1 (2.11), and TCR-Vγ4 (UC3-10A6) from BioLegend. The Tonegawa nomenclature was used to designate TCR-Vγ chains throughout the text. The staining panels always included dead cell staining dyes (7-AAD [BD Biosciences] or Aqua [Invitrogen]). After staining, cells were washed and analyzed using a CyAn FACS machine (Beckman Coulter) or a BD FACSAria III (BD Biosciences).

Cells were stimulated in culture medium containing PMA (50 ng/ml; Sigma), ionomycin (1 μg/ml; Sigma-Aldrich), and monensin (GolgiStop, 1 μl/ml; BD Biosciences) at 37°C and 10% CO₂ for 4 h. After staining of surface markers, cells were fixed and permeabilized (Cytofix/Cytoperm and Perm/Wash Buffer; BD Biosciences), followed by staining with mAbs to mouse IL-17A (TC11-18H10) (BD Biosciences) and to mouse IFN-γ (XMG1.2) and fluorocytometric analysis (CYAN; Beckmann Coulter). For ROR-γt staining, cells were fixed on ice in 2% formaldehyde (diluted from Histofix 4%; Carl Roth, Karlsruhe, Germany) for 1 h followed by staining with anti-mouse ROR-γt Abs (B2D; eBioscience) diluted in 1× eBioscience Permeabilization Buffer (eBioscience), as previously described (25).

For real-time PCR analysis, cells were homogenized using QiaShredder, and RNA was isolated using the RNeasy Mini or RNeasy Micro Kit (all Qiagen) and transcribed into cDNA using the TaqMan Reverse Transcription Reagents Kit (Life Technologies) according to the manufacturer’s instructions. Primers and probes were purchased from Life Technologies; the assays were performed on 96-well reaction plates (Life Technologies) and analyzed using a StepOne System from Life Technologies. In all experiments, Actb was used as the reference gene to normalize gene expression.

Pooled cell suspensions derived from peripheral lymph nodes and spleen of Il23r^gfp/+ mice were sorted (based on GFP expression) into IL-23R⁺ and IL-23R⁻ γδ T cells; or, as indicated, into IL-23R⁻CD122⁺ or IL-23R⁻CD122⁻ γδ T cells; or according to TCR-Vγ expression by FACS (BD FACSAria III). Purified γδ T cells were stimulated in 96-well U-bottom plates (20,000 cells per well) for 5 d with 10,000 mouse anti-CD3/-CD28–coated T-Activator Dynabeads (Life Technologies) in 10% FCS containing DMEM supplemented with 5 × 10⁻⁵ M 2-ME, 1 mM sodium pyruvate, nonessential amino acids, l-glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Recombinant cytokines were added to the cultures as indicated: 10 ng/ml mouse IL-1β, 50 ng/ml mouse IL-6, 10 ng/ml mouse IL-23, and 100 ng/ml mouse IL-21 (R&D Systems).

Statistical evaluations were performed using GraphPad Prism 6 software. The unpaired Student t test was used for comparison of two populations. The paired Student t test was used for mean fluorescence intensity comparison between different TCR-Vγ chain subsets derived from the same mouse. Multiple comparisons were performed using one-way or two-way ANOVA, followed by post hoc tests (Tukey). A p value <0.05 was considered significant.

During thymic development, γδT17 cells are precommitted to express the IL-23R and acquire the potential to produce IL-17, whereas γδT1 cells are believed to be terminally committed to produce IFN-γ but not IL-17. γδT17 cells leave the thymus of mice in a “wave” around embryonic day 18. After birth the thymic export of γδT17 cells has been reported to wane (2). To challenge this observation in a bone marrow transplantation setting, we grafted Il23r^gfp/+ bone marrow, which reports IL-23R expression by expression of GFP into irradiated adult Rag1^−/− recipient mice. Although these bone marrow chimeras were not deficient in Th17 cells and exhibited regular fractions of γδ T cells within the CD3⁺ T cell compartment (Fig. 1A, 1B), γδT17 cells were essentially absent in the thymus, lymph nodes, or spleen (Fig. 1C). To test whether the thymic milieu was repulsive for the development of γδT17 cells because of the ontogenetic “age” of thymocytes, we compared the development of γδT17 cells in the lymphoid compartments of Rag1^−/− mice transplanted with adult bone marrow or fetal liver cells (26). We found that Rag1^−/− recipient mice failed to generate γδT17 cells irrespective of whether they were reconstituted with adult bone marrow or fetal liver cells (Fig. 1D). In contrast to bone marrow or fetal liver precursor cells, cotransferred neonatal thymocytes, which had already received their thymic imprinting, contained γδT17 cells and were readily able to populate the secondary lymphoid tissue of Rag1^−/− recipient mice (Supplemental Fig. 1, see also Ref. 26).

IL-2 plays an inhibitory role during early commitment of conventional naive CD4⁺ T cells to the Th17 lineage in secondary lymphoid organs (27). Because the transcriptional profiles of Th17 cells and γδT17 cells are similar in many aspects (28, 29), we investigated whether IL-2 was a relevant determinant of thymic γδT17 development. Thymic IL-2 expression decreases before birth and resumes postnatally (Supplemental Fig. 2A). As the prenatal trough in thymic IL-2 availability coincides with the proposed time window of output of TCR-Vγ4⁺ γδT17 cells (30), it was possible that increased sensing of IL-2 within the adult thymus would prevent the commitment of γδ thymocytes toward the γδT17 lineage. Therefore, we generated mixed bone marrow chimeric mice by transferring wild-type plus Il2ra^−/− bone marrow into irradiated Rag1^−/− recipients. Indeed, after full reconstitution, the fraction of thymic γδT17 cells (as measured by CCR6 expression in CD27⁻ γδ T cells [2, 3]) was slightly increased in the IL-2Rα–deficient cell compartment as compared with its wild-type counterpart (Supplemental Fig. 2B). However, despite this facilitated commitment of Il2ra^−/− thymocytes to the γδT17 phenotype, the peripheral lymph node γδT17 population was not rescued in mixed bone marrow chimeras in the steady state (Supplemental Fig. 2B). Together, these data suggest that IL-2 may contribute to restraining the commitment of γδ thymocytes to the γδT17 lineage in the adult thymus, but may also be required for their maintenance in secondary lymphoid tissues.

We and others have previously reported that γδT17 cells massively expand under inflammatory conditions and contribute to IL-17 production in various organ systems (1, 31). Because adult Rag1^−/− recipients of Il23r^gfp/+ bone marrow lacked γδT17 cells under homeostatic conditions, we expected these mice to be largely resistant to γδT17-dependent, psoriasis-like skin inflammation induced by application of the imiquimod-containing cream Aldara. Surprisingly, we found that Il23r^gfp/+ → Rag1^−/− bone marrow chimeric mice were susceptible to Aldara-induced skin inflammation almost to the same extent as non–bone marrow chimeric Il23r^gfp/+ mutant mice, as indicated by a substantial increase in skin thickness and formation of psoriasis-like lesions (Fig. 2A). Importantly, in contrast to vehicle-treated controls, the skin-draining lymph nodes of Aldara-treated bone marrow chimeras harbored marked fractions and absolute numbers of IL-23R (GFP)⁺ γδ T cells (Fig. 2B), which were bona fide γδT17 cells because they expressed ROR-γt and produced IL-17 (Fig. 2C). These findings prompted the hypothesis that the inflammatory milieu might promote the de novo generation of γδT17 cells in the peripheral immune compartment, either through imprinting of the γδT17 transcriptional profile in uncommitted precursors or through reprogramming of γδT1 cells (contained within the IL-23R⁻ γδ T cell compartment).

To test the possibility of de novo generation of γδT17 cells, we isolated IL-23R⁻ γδ T cells from lymph nodes and spleen of naive Il23r^gfp/+ mice by flow cytometric sorting to high purity (Fig. 3A) and stimulated these γδ T cells in the presence of anti-CD3/anti-CD28–coated beads under neutral conditions, or in the presence of various cytokines. Similar to any individual cytokine tested, IL-23 alone failed to induce a stable population of γδT17 cells, whereas only the combination of IL-1β plus IL-23 resulted in the sustained induction of IL-23R⁺ γδ T cells in the precursor IL-23R⁻ γδ T cell population (Fig. 3B). “Induced” γδ T cells were bona fide γδT17 cells because they expressed ROR-γt and IL-17 (Fig. 3C). IL-1β alone was sufficient to induce minute amounts of IL-23R (GFP) in IL-23R⁻ γδ T cells in the presence and absence of TCR triggering (Fig. 3D), suggesting that after initial induction of IL-23 responsiveness in IL-23R⁻ γδ T cells by IL-1β, IL-23 then expanded this population and perhaps also stabilized their functional phenotype.

We consider IL-23 as a growth factor for IL-23R⁺ γδ T cells because Il23r^gfp/gfp mice that entirely lack a functional IL-23R still harbor GFP⁺ (IL-23R “wannabe”–expressing) γδ T cells, which are equally capable of producing IL-17 as their GFP⁺ Il23r^gfp/+ counterparts (Supplemental Fig. 3). Although the fraction of GFP⁺IL-17⁺ γδ T cells in the thymus is similar in Il23r^gfp/+ and Il23r^gfp/gfp (= Il23r^−/−) mice, the fraction of γδT17 cells in secondary lymphoid tissue is much higher in Il23r^gfp/+ than in Il23r^gfp/gfp mice (Supplemental Fig. 3). Together with our in vitro data, these findings suggest that the function of IL-23 is essentially to expand γδ T cells which have been previously induced to express IL-23R.

To directly test this idea in vivo, we established a competitive setting comparing the efficacy of induction/expansion of γδT17 cells in Il23r^gfp/+ (IL-23–responsive) versus Il23r^gfp/gfp (IL-23–nonresponsive) precursors in the context of skin inflammation induced by Aldara, which leads to the production of IL-23 in the skin (32). Mixed bone marrow chimeras (Il23r^gfp/+ plus Il23r^gfp/gfp → Rag1^−/−) were treated with Aldara. As expected, the mixed bone marrow chimeric animals developed significant skin thickening as compared with Vaseline-treated controls (Fig. 4A). Notably, the large majority of γδT17 cells in the skin-draining lymph nodes were recruited from GFP⁺ Il23r^gfp/+ γδ T cells, whereas GFP⁺ Il23r^gfp/gfp γδ T cells were induced but failed to expand; resulting in a large bias of the chimerism in the GFP⁺ γδ T cell compartment (as well as the IL-17⁺ γδ T cell compartment) in the skin-draining lymph nodes in favor of Il23r^gfp/+ and at the expense of Il23r^gfp/gfp γδ T cells (Fig. 4B, 4C). Accordingly, the absolute number of GFP⁺ γδ T cells in the draining lymph nodes was significantly smaller in the CD45.1⁺ (Il23r^gfp/gfp) compartment as compared with the CD45.2⁺ (Il23r^gfp/+) compartment (Fig. 4D). Together, these data indicate that, after establishment of IL-23 responsiveness in γδ T cells, IL-23 is crucial for the expansion of γδT17 cells in vivo.

γδT1 cells were previously characterized as CD27⁺ γδ T cells (2), whereas another report referred to them as CD122⁺ γδ T cells (33). When we sorted IL-23R⁻CD122⁺ and IL-23R⁻CD122⁻ γδ T cells from the peripheral immune compartment of Il23r^gfp/+ mice to high purity (Fig. 5A) and compared their capacity to “convert” into IL-23R (GFP)⁺ γδT17 cells, we found that only CD122⁻ γδ T cells but not their CD122⁺ counterparts could be induced to express IL-23R (GFP) and IL-17 in response to IL-1β plus IL-23 (Fig. 5A). To test whether CD122⁻ γδ T cells also represented the γδ T cell population that was induced to express IL-23R and IL-17 in our bone marrow chimeric mice (Fig. 1), we sorted IL-23R (GFP)⁻CD122⁺ and IL-23R (GFP)⁻CD122⁻ γδ T cells from the secondary lymphoid tissue of unmanipulated Il23r^gfp/+ → Rag1^−/− bone marrow chimeras and exposed them to IL-1β and IL-23 in vitro. Similar as in γδ T cells directly isolated from Il23r^gfp/+ mice, only CD122⁻ but not CD122⁺ γδ T cells from bone marrow chimeras were induced to express IL-17 (Fig. 5B). Thus, IL-23R⁻CD122⁻ γδ T cells compose a population of γδ T cells that can be induced to express IL-17 and commit to the γδT17 lineage outside the thymus.

Within the in vitro–induced γδT17 cells, TCR-Vγ4⁺ and TCR-Vγ6⁺ γδ T cells (identified in this article as Vγ1⁻Vγ4⁻) were significantly enriched over Vγ1⁺ γδ T cells, which dominated the pool of IL-23R⁻CD122⁻ γδ T cells ex vivo (Fig. 6A, 6B). To test whether the possibility to differentiate into γδT17 cells indeed segregated with the expression of distinct TCR-Vγ chains, we sorted Vγ1⁺ and Vγ4⁺ as well as “Vγ double-negative (DN)” cells from the peripheral IL-23R (GFP)⁻CD122⁻ γδ T cell compartment of Il23r^gfp/+ mice and cultured them in the presence of IL-1β plus IL-23. Here, Vγ1⁺ γδ T cells were only poorly induced to express IL-23R (GFP), whereas the majority of Vγ4⁺ γδ T cells as well as Vγ DN gained IL-23R and IL-17 expression upon stimulation with IL-1β and IL-23 (Fig. 6C). Consistent with this finding, IL-23R (GFP)⁺ γδ T cells converted by in vitro culture with IL-1β and IL-23 from Il23r^gfp/+ → Rag1^−/− bone marrow chimera–derived IL-23R⁻CD122⁻ γδ T cells, as well as IL-23R (GFP)⁺ γδ T cells induced outside the thymus in vivo in Il23r^gfp/+ → Rag1^−/− bone marrow chimeras, were highly biased toward the expression of Vγ4 at the expense of Vγ1, which was the dominating Vγ chain in IL-23R⁻CD122⁻ precursor γδ T cells (Fig. 6D). Taken together, these data suggest that IL-23R⁻CD122⁻Vγ4⁺ γδ T cells, but not their Vγ1⁺ counterparts, can be committed to the γδT17 lineage in the peripheral immune compartment.

To further investigate the idea that the Vγ repertoire was a determinant of the commitment of γδ T cells to either γδT1 or γδT17 cells, we assessed the expression of key transcription factors in highly purified CD122⁻Vγ1⁺ versus CD122⁻Vγ4⁺ γδ T cells. We found that Vγ1⁺ γδ T cells resembled CD122⁺ γδ T cells (bona fide γδT1 cells), whereas Vγ4⁺ γδ T cells expressed low levels of ROR-γt and the γδT17-associated scavenger receptor Scart2 (34) (Fig. 7A). Because we did not observe current or historic production of IL-17 in CCR6⁻CD122⁻ γδ T cells (corresponding to the IL-23R⁻CD122⁻ γδ T cell fraction in Il23r^gfp/+ mice) of unmanipulated Il17f-Cre^eyfp reporter mice ex vivo (Supplemental Fig. 4A), it is unlikely that IL-23R⁻CD122⁻Vγ4⁺ or Vγ6⁺ γδ T cells descend from previous γδT17 cells. None of the auxilliary transcription factors of the Th17 profile, which have recently been described to be dispensable for the development of γδT17 cells (29), were regulated (Fig. 7A).

Finally, we directly tested the correlation of differential ROR-γt expression in CD122⁻Vγ4⁺ versus CD122⁻Vγ1⁺ γδ T cells with previous TCR triggering by investigating adult Nur77^GFP reporter mice. We did not detect any differences in the extent of TCR engagement between these two subsets in the peripheral immune compartment. However, in the thymus, CD122⁻Vγ1⁺ γδ T cells showed a high fraction of Nur77^GFP+ cells, which was essentially absent in CD122⁻Vγ4⁺ γδ T cells (Fig. 7B, Supplemental Fig. 4B). Thus, TCR engagement in CD122⁻Vγ1⁺ γδ thymocytes might prevent the expression of ROR-γt and poise these cells for the γδT1 lineage. Conversely, lack of TCR engagement in CD122⁻Vγ4⁺ γδ thymocytes saves these cells from premature lineage commitment in the thymus and allows for the induction of a γδT17 transcriptional program outside the thymus.

In summary, although the adult thymus ceases producing γδT17 cells, bona fide γδT17 cells can be generated in the peripheral immune compartment from those IL-23R⁻ precursors that are not already poised for IFN-γ production by TCR engagement in the thymus.

γδT17 cells are gatekeepers of inflammatory responses at epithelial surfaces and perhaps also in the CNS (35, 36). It is an emerging concept that the steady-state population of certain tissue niches with γδT17 cells might dictate their responsiveness to pathogen-associated molecular patterns, IL-23, eventually the extent of immunopathology resulting from inflammation. In this study, we confirmed that γδT17 cells are not produced by the adult thymus. In contrast to Th17 cell commitment, IL-2 has only a minor inhibitory function during thymic development of γδT17 cells. However, γδT17 cells can be de novo generated extrathymically by conversion of IL-23R⁻ γδ T cells into bona fide γδT17 cells in response to IL-1β plus IL-23, indicating an unanticipated plasticity of a subset of IL-23R⁻ γδ T cells that are CD122⁻ and express TCR-Vγ4 or TCR-Vγ6.

Thymic development of γδ T cells occurs according to waves of γδ TCR rearrangement during embryogenesis. Thymic generation of TCR-Vγ5–positive γδ T cells peaks at embryonic day 14 and TCR-Vγ5–positive γδ T cells populate the epidermis of mice as dendritic epidermal T cells. TCR-Vγ6–positive γδ T cells are generated around embryonic day 17 and home to urothelium and dermis, and TCR-Vγ4–positive γδ T cells peak at embryonic day 18.5 and home to bronchial epithelium as well as secondary lymphoid tissues (37). An alternative concept uses functional definitions for different developmental waves of γδ T cells. Here, thymic export of dendritic epidermal T cells is followed by the γδT17 wave and the γδT1 wave (30, 38, 39). Notably, the time window of γδT17 export largely coincides with the TCR-Vγ4 wave and TCR-Vγ4 is overrepresented in γδT17 cells in secondary lymphoid tissues. More recently, populations of pathogenic TCR-Vγ6⁺ γδT17 cells were described in the dermis and joints of mice (12, 40). Although the TCR-Vγ chain repertoire of γδT17 cells appears to be restricted, it is unlikely that commitment of γδ T cells to the production of IL-17 is based on the thymic engagement of distinct γδ TCRs because the ability to produce IL-17 from γδ thymocytes is imprinted prior to TCR-Vγ chain rearrangement (5).

The molecular underpinning of γδT17 commitment in the thymus is only partly understood. In this study, we found that IL-2 partially constrains γδT17 cell development in the thymus. In a previous report, IL-2 has been proposed to support (rather than constrain) the maintenance of γδT17 cells (33). However, these observations were made in IL-2–deficient or IL-2Rα–deficient mice that suffer from a lymphoproliferative syndrome, which represents a systemic inflammatory scenario where IL-2 might indeed promote the expansion of γδT17 cells in the peripheral immune compartment. In addition, as in conventional αβ T cells, thymic egress of γδ T cells is controlled by S1P1 in a KLF2-dependent manner (41). As IL-2 stimulation in vitro abrogates S1P1 expression in γδ T cells (data not shown), the inability to sense IL-2 could result in increased thymic egress due to sustained surface expression of S1P1. However, in mixed bone marrow chimeras, the fractions the γδT17 cells in the thymus were higher within the IL-2Rα–deficient than the wild-type compartment, suggesting that IL-2 restricts the thymic development of γδT17 cells, for example by binding of STAT5 to γδT17 hallmark genes as has been demonstrated in Th17 cells (27). Recently, IL-15 has been reported to suppress the generation of γδT17 cells. IL-15 shares CD122 and CD132 as receptor molecules with IL-2 and thus, like IL-2, also belongs to the family of common γ-chain cytokines. While suppressing γδT17 cells, IL-15 has been shown to promote γδT1 development in the thymus by induction of T-bet in CD8αα precursors (42). Altogether, it is intriguing to speculate that common γ-chain cytokines might be an important determinant in dictating the balance of γδT1 versus γδT17 cells during thymic development.

In contrast to the current understanding that γδ T cells are committed to their cytokine phenotype in the thymus with limited plasticity to revert to IL-17 production, we found in this study that IL-17–producing γδ T cells can be de novo induced in secondary lymphoid tissues. Under inflammatory conditions, i.e., in the presence of IL-1β and IL-23, IL-23R⁻ γδ T cells can be induced to express IL-23R and respond to IL-23 by expansion. In adult individuals, TCR-Vγ4⁺ γδT17 cells are strongly expanded in the presence of IL-23. The de novo induction of γδT17 cells in the peripheral immune compartment would be in line with the idea that the γδ T cell population in secondary lymphoid tissue contains uncommitted γδ T cells. It has recently been suggested that naive (uncommitted) γδ T cells are continuously produced from embryonic day 16 and, upon recognition of the “model Ag” PE, might be able to initiate IL-17 production without prior commitment to the γδT17 lineage in the thymus (30, 43). Uncommitted γδ T cells in secondary lymphoid organs are CD44^lowCD62L^high. All of the CD44^low γδ T cells in naive mice lack IL-23R expression and CD44^low γδ T cells can indeed be induced to express IL-23R (data not shown).

IL-23R⁻TCR-Vγ1⁺ γδ T cells largely resist conversion into γδT17 cells, whereas CD122⁻IL-23R⁻ TCR-Vγ4⁺ or Vγ6⁺ γδ T cells readily adopt a γδT17 phenotype in response to IL-1β plus IL-23. The mechanistic underpinning of the differential potential of IL-23R⁻ γδ T cells with various TCR-Vγ chains to acquire a γδT17 transcriptional profile is currently unclear. Interestingly, TCR-Vγ1⁺ but not TCR-Vγ4⁺ γδ thymocytes appear to have sensed a TCR signal in the thymus, suggesting that TCR ligation in the thymus skews γδ thymocytes toward an IFN-γ–producing phenotype. Similar observations were made previously in an Ag-specific system (4) and in a recent study, CD3γ and CD3δ double haploinsufficient mice, which exhibit decreased signaling strength downstream of the TCR, lose a large fraction of their IFN-γ–producing γδ T cells during embryonic development, which is sustained during adulthood (44).

TCR-Vγ4⁺ γδ T cells expand during imiquimod-induced skin inflammation and redistribute to previously noninflamed skin sites where they were more responsive to IL-1β than nonimiquimod-sensitized γδT17 cells due to higher expression of IL-1R1 (45). The readily inducible expression of Il1r1 in TCR-Vγ4⁺ γδ T cells in a TCR-dependent or TLR1- or TLR2-dependent (46) manner, and thus the direct engagement of the Vγ4⁺ γδ-TCR by ligands in the inflamed skin, might explain their exquisite responsiveness to IL-1β plus IL-23 and subsequently result in the preponderance of TCR-Vγ4 over TCR-Vγ6 in induced γδT17 cells in vivo as compared with their in vitro–induced counterparts. It remains to be determined whether the synergistic action of IL-1β plus IL-23 in TCR-Vγ4⁺ and TCR-Vγ6⁺ γδ T cells is due to similar molecular events in these subsets. Nevertheless, exposure to IL-1β and subsequent expansion in response to IL-23 must overcome “epigenetic” fixation of CD27⁺ γδ T cells, which have been considered as poised for the production of IFN-γ (13).

Tissue-resident γδT17 cells have been described in the gut lamina propria, in the dermis, in entheseal tissue (21, 47–49), and are likely present in a variety of other epithelial barriers. Their physiological role in these niches is not entirely clear. However, in inflammation they respond to IL-23 by expansion and production of IL-17. In this article, we propose that the equipment of a specific niche with IL-23R⁺ γδ T cells may not be a fixed determinant, and the quality and extent of the γδ T cell response might easily be changed by inflammatory mediators that de novo induce IL-17 in a subset of γδ T cells believed to be poised for the production of IFN-γ. Thus, the cellular sources of mediators like IL-1β and IL-23, which are not well defined in nonlymphoid tissue and which might be induced by unexpected mechanisms including triggering of nociceptive nerve fibers (50), might be more important for the properties of γδ T cell responses than the steady-state composition of the γδ T cell population in a particular niche per se. The unraveling of the cytokine cues that dictate the abundance of γδT17 cells in steady state and in inflammation will be highly relevant for designing preventive and long-lasting strategies to limit immunopathology in autoimmunity and chronic inflammation in nonlymphoid tissues.

We thank S. Woeste and V. Husterer for skillful technical assistance. Dr. Reinhard Obst, Dr. Marc Schmidt-Supprian, and Dr. Ari Waisman kindly provided us with Nur77^GFP reporter mice and Il17f-Cre^eyfp reporter mice, respectively.

The authors have no financial conflicts of interest.