IL‐23 drives differentiation of peripheral γδ17 T cells from adult bone marrow‐derived precursors

Abstract Pro‐inflammatory interleukin (IL)‐17‐producing γδ (γδ17) T cells are thought to develop exclusively in the thymus during fetal/perinatal life, as adult bone marrow precursors fail to generate γδ17 T cells under homeostatic conditions. Here, we employ a model of experimental autoimmune encephalomyelitis (EAE) in which hematopoiesis is reset by bone marrow transplantation and demonstrate unequivocally that Vγ4+ γδ17 T cells can develop de novo in draining lymph nodes in response to innate stimuli. In vitro, γδ T cells from IL‐17 fate‐mapping reporter mice that had never activated the Il17 locus acquire IL‐17 expression upon stimulation with IL‐1β and IL‐23. Furthermore, IL‐23R (but not IL‐1R1) deficiency severely compromises the induction of γδ17 T cells in EAE, demonstrating the key role of IL‐23 in the process. Finally, we show, in a composite model involving transfers of both adult bone marrow and neonatal thymocytes, that induced γδ17 T cells make up a substantial fraction of the total IL‐17‐producing Vγ4+ T‐cell pool upon inflammation, which attests the relevance of this novel pathway of peripheral γδ17 T‐cell differentiation.


Introduction
Interleukin (IL)-17A (IL-17 herein) is a major promoter of antimicrobial peptide production and neutrophil mobilization, which likely accounts for its conservation across evolution of the vertebrate immune system [1]. On the other hand, the contributions of IL-17 to inflammatory and autoimmune diseases make it a hot target for current and upcoming immunotherapeutic strategies [2].
cd17 T cells are also major sources of IL-17 in steady-state conditions [23], likely due to their "developmental pre-programming" in the thymus [24]. Thus, we and others have shown that mouse cd thymocytes can acquire the capacity to produce IL-17, which associates with the upregulation of CCR6 and the loss of CD27 expression [25,26]. Importantly, the development of cd17 T cells is believed to be restricted to fetal/perinatal life, as transplantation of adult bone marrow, or induction of Rag1 activity after birth, failed to generate cd17 T cells [27]. According to this model, steady-state cd17 T cells are only generated in fetal and neonatal thymus, persisting thereafter as self-renewing and long-lived cells in the thymus and in peripheral organs [27,28], where they can engage in immune responses. Whether cd T cells derived from adult bone marrow precursors can be induced to express IL-17 in peripheral lymphoid organs under inflammatory conditions still remains unresolved. Indeed, since a substantial fraction of cd T cells exit the adult thymus as functionally immature ("naïve") T cells, they could differentiate into IL-17 producers upon activation, alike conventional ab T H 17 cells. While this has been shown for a very small (~0.4%) population of cd T cells whose TCR recognizes the algae protein phycoerythrin (PE) [28,29], it remains unknown whether (and to what extent) such peripheral differentiation occurs in pathophysiological settings. To address this important question, we turned here to the experimental autoimmune encephalomyelitis (EAE) mouse model of multiple sclerosis.
cd T cells significantly accumulate during the acute phase of EAE [30]; most of these cells bear a Vc4 + TCR and make IL-17 [22,31].
While EAE clearly constitutes an appropriate model to address peripheral cd17 T-cell differentiation under inflammatory conditions, there is a major confounding factor-the sizeable "natural", that is, thymic-derived cd17 T-cell pool established in steady-state secondary lymphoid organs since birth. To overcome this problem, we have here induced EAE after resetting hematopoiesis through lethal irradiation followed by bone marrow transplantation. Since adult bone marrow precursors cannot generate thymic cd17 T cells [27], the transplanted mice are devoid of thymic-derived peripheral cd17 T cells before EAE induction. This allowed us to unequivocally demonstrate the differentiation of cd17 T cells from "naïve" cd T cells in draining lymph nodes in response to inflammatory IL-23 signals.

Peripheral differentiation of cd17 T cells upon EAE inflammation
We established bone marrow chimeras (BMCs) using a congenic marker (Thy1.1/Thy1.2) to distinguish donor and host hematopoietic cells and TCRd À/À recipients, to guarantee the absence of any host cd T cells that might resist the irradiation protocol ( Fig 1A). As expected [27], after 8 weeks of reconstitution, cd T cells lacked IL-17 but expressed IFN-c in peripheral organs ( Fig 1B; Fig EV1). EAE was induced by injection of myelin oligodendrocyte glycoprotein (MOG) peptide, complete Freund's adjuvant (CFA) and pertussis toxin, as widely established [22]. The BMCs developed severe pathology, comparable to unmanipulated C57Bl/6 mice, with slightly delayed onset ( Fig 1C). When we analyzed the BMCs at the peak of disease (day 14 post-induction; p.i.), we found striking proportions of IL-17 + cd T cells in the brain, lymph nodes, and spleen, in stark contrast with naïve BMCs (Fig 1B and D). As expected in EAE [22], these cd17 T cells expressed almost exclusively Vc4 + TCRs ( Fig 1E). Importantly, they also expressed the master transcription factor RORct, but not T-bet (Fig 1F), the cytokine receptor IL-1R1 ( Fig 1G) and the surface molecule CD44 (Fig 1H). These data demonstrate that bona fide cd17 T cells can differentiate in the periphery under inflammatory conditions.
MOG and TLR-independent peripheral cd17 T-cell differentiation in lymph nodes Next, we investigated the generation site of the induced cd17 T cells in EAE by sacrificing the animals at an early time point (day 7 p.i.), before the appearance of the first clinical signs of the disease. We examined lymphoid organs, the target tissue, and other non-lymphoid tissues implicated in the generation of encephalitogenic cells [36,37] and found cd17 T cells mainly in the draining lymph nodes (Fig 2A and B), where they actively proliferated, as shown by Ki67 staining (Fig 2C). In some mice, we detected small frequencies of cd17 T cells also in the cervical lymph nodes (cLN), spleen, and lungs ( Fig 2D). While it is possible that these cells can differentiate outside the immunization area (due to propagation of inflammatory signals), they could, alternatively, be recirculating to get licensed to enter the CNS [37]. Importantly, these cells were not found in the brain, which still did not show an inflammatory infiltrate at this time point, nor in the lamina propria, mesenteric lymph nodes (mLN), or thymus (Fig 2A and D).
Given that the EAE induction protocol comprises both myelinspecific antigen (MOG peptide) and innate stimuli derived from CFA and pertussis toxin (PTx), we next administered (subcutaneously) different combinations of the adjuvants in the absence of MOG peptide ( Fig 3A). BMCs immunized with CFA plus PTx showed substantial pools of cd17 T cells in draining lymph nodes, in stark contrast to IFA plus PTx or CFA alone (Fig 3A and B; Fig EV2A). Since the peripheral generation of cd17 T cells did not require myelin-specific antigens but rather CFA and PTx, we hypothesized that innate cytokine stimuli, rather than recognition of ▸ Figure 1. Peripheral differentiation of cd17 T cells upon EAE inflammation.

IL-23-dependent peripheral cd17 T-cell differentiation
In order to obtain a reliable source of uncommitted cd T cells, we employed an IL-17 fate-mapping reporter mouse line where eYFP expression permanently marks the activation of the Il17 locus [23]. We cultured highly purified (> 99%) eYFP(À) cd T cells with various activation/differentiation cocktails (Fig 4A). IL-1b and IL-23 were found to be sufficient to elicit de novo cd17 T-cell generation (Fig 4A and B). In vitro stimulation with IL-1b and IL-23 (but not TGF-b or IL-6) had been shown to trigger abundant IL-17 secretion by peripheral CD27-CCR6 + cd T cells [22,26,38,39], but since these cells contained thymic-derived cd17 T cells, it was not possible to distinguish between expansion of pre-differentiated versus induction of cd17 T cells. In our in vitro system, although TCR stimulation was not essential, it synergized with these cytokines to greatly enhance the frequency of eYFP + cells (Fig 4A and B). Unexpectedly, addition of IL-6 and TGF-b decreased the mean fluorescence intensity (MFI) of eYFP ( Fig 4C). As our data (Fig 3C and D) argued against a non-redundant role for MyD88-dependent IL-1b/IL-1R1 signaling, we next investigated whether either MyD88-independent IL-1b/IL-1R1 or IL-23/IL-23R signals drove peripheral cd17 T-cell differentiation in vivo. For this, we generated mixed BMC using Thy1.1 + and IL-23R À/À or IL-1R1 À/À as donor cells (in a 1:1 ratio), and after 8 weeks immunized them with CFA plus PTx ( Fig 4D). As expected, cd17 T cells were found in the draining lymph nodes of IL-23R À/À mixed BMCs, but not in their naïve counterparts (Fig 4E and F). Importantly, the vast majority were of Thy1.1 (IL-23R +/+ ) origin ( Fig 4G). Moreover, we observed a marked shift in the Thy1.1:IL-23R À/À ratio among total cd T cells (Fig EV3A and B), which further attests the   impact of the CFA-induced and IL-23-dependent cd T-cell response. As for Thy1.1:IL-1R1 À/À mixed BMCs, while they harbored cd17 T cells after immunization (Fig 4H and I) and displayed a shift in Thy1.1:IL-1R1 À/À ratio (Fig EV3C and D), they contained a substantial fraction of cd17 T cells derived from IL-1R1 À/À progenitors ( Fig 4J). These data collectively suggest that IL-23R (rather than IL-1R1) signaling is the key orchestrator of peripheral cd17 T-cell differentiation in vivo.

Induced cd17 T cells make a large contribution to the total cd17 T-cell pool in EAE
Finally, we aimed to establish whether peripheral cd17 T-cell differentiation would occur in the presence of "natural" (thymicderived) cd17 T cells-and, if so, to determine the relative contributions of the two pools in EAE. To answer these questions, we transplanted neonatal thymocytes (expressing both Thy1. Thy1.2) and bone marrow cells (Thy1.1 + ) into TCRd À/À mice ( Fig 5A). As expected [17,27], we could observe "natural" cd17 T cells of neonatal thymic origin in the lymph nodes of naïve mice (Fig 5B; Fig EV4A). However, upon EAE induction (Fig EV4B), cd17 T cells were found also in the brain and spinal cord (Fig 5B and C). Of note, these mice presented increased frequencies of cd17 T cells in cervical LN and spleen, but decreased in the draining LN (Fig 5B  and C), probably due to their migration to the central nervous system (CNS). Of interest, the dominant cd17 T-cell subset in this model switched from Vc1 -Vc4to Vc4 + cells (Fig EV4C and D). Critically, around half of the Vc4 + cd17 T cells in the lymph nodes and CNS during EAE derived from adult bone marrow precursors, whereas Vc4 -cd17 T cells were mainly of neonatal thymic origin (Fig 5D and E). These data clearly demonstrate that peripheral cd17 T-cell differentiation accounts for a large fraction of the total Vc4 + cd17 T-cell pool in EAE.
In summary, our study identifies a peripheral pathway of differentiation of bona fide RORct + cd17 T cells derived from adult bone marrow precursors, which occurs mainly in the draining lymph nodes upon inflammation, including EAE. We further demonstrate that this pathway does not depend on specific myelin antigens but rather on innate stimuli, as those contained in CFA and PTx. By combining in vitro and in vivo approaches and gene-targeted mice, we ascribe a key role to IL-23 in the de novo differentiation of peripheral cd17 T cells.
Even more complex is the role of the TCR in cd17 T-cell differentiation. Although TCR engagement synergized with IL-1b plus IL-23 stimulation (to enhance cd17 T-cell induction), it was per se not required for acquisition of IL-17 expression by cells that had never activated the Il17 locus before. This is consistent with the overall impact of the TCR on peripheral cd17 T-cell responses [22]; but also with its dispensable role in the development of "natural" Vc4 + cd17 T cells in the thymus [20,44]. Interestingly, strong TCR signals promote the development of thymic-derived Vc6 + cd17 T cells, which are the other main subset of cd17 T cells. Whereas in EAE the main responsive cd17 T-cell subset is Vc4 + , other inflammatory diseases (also) engage Vc6 + cd17 T cells. That is the notable case of imiquimod-induced psoriasiform inflammation in the dermis, where Vc4 + and Vc6 + cd17 T cells are differentially involved [17,18]. Therefore, it will be interesting to investigate the peripheral differentiation of Vc4 + versus Vc6 + cd17 T cells in this model.
In a previous study, a discrete population (~0.4%) of cd T cells was shown to recognize the algae antigen PE via the TCR and differentiate into IL-17 producers [29], although the fetal versus adult origin of the "naive" precursors was not addressed. Most interestingly, cognate PE interactions were shown to upregulate IL-23R (as well as IL-1R1) expression, suggesting that TCR signals were required to "license" the cells to respond to IL-23 (and IL-1b) [29]. We therefore suggest that the mechanisms of peripheral cd17 T-cell differentiation converge on IL-23R signaling. Consistent with this, cd T cells constitutively expressing IL-23R are known to be the first cells to respond to IL-23 during EAE development [21]. Moreover, IL-23 was shown to greatly enhance IL-17 production by cd T cells triggered by stimulation with TLR2 and Dectin-1 ligands in vitro [45]. The presence of pathogen-associated molecular patterns of M. tuberculosis and subsequent IL-23 production may thus underlie the expansion of the cd17 T-cell pool in response to CFA in our and previous studies [45,46]. However, the need for PTx signals in our model points to different requirements and/or thresholds of response to innate cytokines in peripheral cd17 T cells compared to their thymic counterparts. Thus, PTx is likely required to maximize the innate stimuli provided by CFA (but not IFA) through the production of innate cytokines, as previously described [47,48].
Peripheral cd17 T-cell differentiation is especially relevant in humans, where cd thymocytes are functionally immature [49] and require activation under inflammatory conditions to produce IL-17 [50], with IL-23 playing a major role in the process [51]. This reinforces the rational for targeting the IL-23/IL-17 axis in autoimmune diseases, which has already produced remarkable results in psoriasis and shows great potential in multiple sclerosis [52].

Mice
All mice used were adults 6-18 weeks of age. C57BL/6J.Thy1.1 and C57BL/6J.MyD88 À/À (hereafter referred as Thy1.1 + and MyD88 À/À , respectively) mice were obtained from Instituto Gulbenkian de Ciências (Oeiras, Portugal), the latter with permission from Dr. Shizuo Akira (Osaka University, Osaka, Japan). C57BL/6J.TCRd À/À and C57Bl/6J.IL-1R1 À/À mice were purchased from The Jackson Laboratory. C57BL/6.IL-23R À/À (hereafter referred as IL-23R À/À ) were obtained from Dr. Fiona Powrie (University of Oxford, Oxford, UK) with permission from Dr. Mohamed Oukka (University of Washington, Seattle, USA). Mice were bred and maintained in the specific pathogen-free animal facilities of Instituto de Medicina Molecular (Lisbon, Portugal). Il17a Cre R26R eYFP (referred to as IL-17 fatemapping reporter) mice were bred in the MRC National Institute for Medical Research (Mill Hill, London, UK) animal facility under specified pathogen-free conditions. All experiments involving animals were done in compliance with the relevant laws and institutional guidelines and were approved by local and European ethics committees.

Bone marrow chimeras
TCRd À/À mice were lethally irradiated (950 rad), and the next day injected intravenously with a total of 5-10 × 10 6 whole bone marrow cells from Thy1.1 + donor mice. For mixed BMCs, a total of 10 7 whole bone marrow cells of mixed (following a 1:1 ratio) Thy1.1 + and IL-23R À/À (Thy1.2 + ) origin, from age-matched animals, were injected in previously lethally irradiated TCRd À/À hosts. In some experiments, BMCs supplemented with neonatal thymocytes were generated as previously described [17]. In brief, lethally irradiated TCRd À/À hosts (Thy1.2 + ) were injected, after 6 h, with neonatal thymocytes from pups (Thy1.1 + Thy1.2 + ) within 48 h of birth. After 24 h, the host received 5-10 × 10 6 bone marrow cells from C57BL/6J (Thy1.1 + ) donors. All BMCs were kept on antibiotics-containing water (2% Bactrim; Roche) for the first EMBO [53]. In brief, the score system ranged from 0 to 5, with 0.5 increments, being score attributed to animals with no clinical signs of EAE and five representative of death. Score 1 consisted in limp tail; score 2 consisted in limp tail together with hind legs weakness; score 3 consisted in partial limb paralysis; and finally, score 4 consisted in complete hind leg paralysis. Cell preparation, flow cytometry, cell sorting, and analysis For cell surface staining, single-cell suspensions were incubated in the presence of anti-CD16/CD32 (eBioscience) with saturating concentrations of combinations of the mAbs listed above. For the preparation of brain, spinal cord, lungs, and lamina propria cells, mice were perfused through the left cardiac ventricle with cold PBS. Lungs, spinal cord, and brain were dissected, and tissue was cut into pieces, and digested with collagenase type IV (0.5 mg/ml; Roche) and DNase I (0.10 mg/ml) (Sigma-Aldrich) in RPMI 1640 containing 5% fetal bovine serum (FBS) at 37°C for 30 min. For lamina propria cell preparation, small intestines were dissected, washed in ice-cold PBS, and Peyer's patches were excised. The organ was then cut into pieces and incubated with EDTA 0.05 M at 37°C for 20 min; cells were then washed and passed through a 100lm cell strainer and then digested as the other organs. Mononuclear cells were isolated by passing the tissue through a 40-lm cell strainer, followed by a 33% Percoll (Sigma-Aldrich) gradient and 30-min centrifugation at 1,160 g. Mononuclear cells were recovered from the pellet, resuspended, and used for further analysis. For intracellular cytokine staining, cells were stimulated with PMA (phorbol 12-myristate 13-acetate) (50 ng/ml) and ionomycin (1 lg/ml), in the presence of Brefeldin A (10 lg/ml) (all from Sigma) for 3 h at 37°C. Cells were stained for the identified above cell surface markers, fixed 30 min at 4°C and permeabilized with the Foxp3/Transcription Factor Staining Buffer set (eBioscience) in the presence of anti-CD16/CD32 (eBioscience) for 10 min at 4°C, and finally incubated for 1 h at room temperature with identified above cytokine-specific Abs in permeabilization buffer. Cells were analyzed using FACSFortessa (BD Biosciences) and FlowJo software (Tree Star).
For cell sorting, peripheral lymph nodes (pLN) were prepared and stained for cell surface markers as mentioned above and then electronically sorted on a FACSAria (BD Biosciences).

Statistical analysis
The statistical significance of differences between populations was assessed with the Kruskal-Wallis test (nonparametric one-way ANOVA) or by using a two-tailed nonparametric Mann-Whitney U-test, when applicable. The P-values < 0.05 were considered significant and are indicated on the figures.
Expanded View for this article is available online.
technical supervision and assisted in the experimental design; BS-S supervised the research and wrote the manuscript.