|Home | About | Journals | Submit | Contact Us | Français|
Dendritic epidermal T cells (DETCs) are a well-studied population of γδ T cells that play important roles in wound repair. Here we characterize a second major population of γδ T cells in the skin that is present in the dermis. In contrast to DETCs, these Vγ5-negative cells are IL-7Rhi CCR6hi RORγt+ and are pre-committed to IL-17 production. Dermal γδ T cells fail to reconstitute following irradiation and bone marrow transplantation unless the mice also receive a transfer of neonatal thymocytes. Real-time intravital imaging of CXCR6-GFP/+ mouse skin reveals dermal γδ T cells migrate at ~4μm/min while DETCs are immobile. Like their counterparts in peripheral lymph nodes, dermal γδ T cells rapidly produce IL-17 following exposure to IL-1β plus IL-23. We have characterized a major population of skin γδ T cells and propose that these cells are a key source of IL-17 in the early hours after skin infection.
The skin has an important barrier function against infection. Part of this function is mediated by resident TCR gammadelta (γδ) T cells known as dendritic epidermal T cells (DETCs) because of their morphology and location. Seeded to the epidermis during fetal development, they express Vγ5 (according to Heilig and Tonegawa nomenclature) (1) and produce keratinocyte growth factors following wounding (2). In contrast to this well recognized subset, the types of T cells resident in the dermis have been less studied.
IL-17 plays an important role in host defense at epithelial surfaces, including the gut, lung, and skin (3). IL-17A is rapidly induced in the skin in response to intradermal Staphylococcus aureus (S. aureus) and Candida albicans (C. albicans), and IL-17A deficient mice fail to optimally resolve infection (4, 5). The cellular source of IL-17 in the first hours to days after skin infection has not been well defined, although it is unlikely to be conventional Th17 cells as they take 3 or more days to differentiate. Recently, populations of innate IL-17 producing γδ T cells have been described in lymphoid tissues, peritoneum and intestinal mucosa and appear to arise from IL-17 committed γδ thymocytes (6–10). The presence of IL-17 producing γδ T cells in total skin preparations has also recently been demonstrated (11, 12) but it has been unclear whether these cells correspond to DETCs (4) or to dermal cells (13).
Here we characterize a population IL-17 pre-committed Vγ5-γδ T cells in the dermis. We provide evidence that these cells derive from perinatal thymocytes and show by intravital imaging that, in contrast to the DETCs, these cells are motile. We propose that dermal γδ T cells are the major source of skin IL-17 during the early stages of infection.
C57BL/6 (CD45.2+), Boy/J (CD45.1+), Cxcr6gfp/+ ((14); 005693, B6.129P2-Cxcr6tm1Litt/J), Rorcgfp/+ ((15); 007572, B6.129P2(Cg)-Rorctm2Litt/J) and TCRδ−/− ((16); B6.129P2-Tcrdtm1Mom/J) mice were from The Jackson Laboratory (Jax) or National Cancer Institute. A trend towards fewer dermal and PLN CCR6+ γδ T cells was noted in mice shipped from NCI compared to animals raised in our internal (Jax-based) colony. All experiments conformed to ethical principles and guidelines approved by the UCSF Institutional Animal Care and Use Committee
LNs were digested as described (17) with 67μg/ml Liberase TM (Roche). Epidermal and dermal sheets were separated with dispase (Gibco) and digested with 85ug/ml Liberase TM (Roche) in DMEM as described (18). Cells were stained (17) with antibodies to TCRγδ (GL3), TCRβ (H57-597), IL-17A (eBio17B7), CD196/CCR6 (140706), IL-7Rα (A7R34), CD103 (2E7), Vγ5 (536), CD45.2 (clone 104) and CD45.1 (clone A20) (Biolegend, BD Biosciences, or eBioscience). Biotin-conjugates were detected with streptavidin Qdot® 605 (Invitrogen). For dermal cell sorting, ear tissue was digested as above, enriched for lymphocytes using Lympholyte®-M (Cedarlane Laboratories), and TCRγδ+IL-7Ra+CCR6+TCRβ-CD11b- cells were sorted (17). To detect IL-17A, cells were stimulated for 2 h with 50ng/ml PMA (Sigma) and 1μg/ml Ionomycin (I, EMD Biosciences) in Brefeldin A (BD Biosciences), stained for surface antigens, treated with BD Cytofix Buffer and Perm/Wash reagent (BD Biosciences), and stained with anti-IL-17A.
Digested ear skin was stimulated with 10ng/ml rmIL-1b and rmIL-23 (eBioscience) for 8 or 18 h. Brefeldin A was added during the last 6 h. Cell were stained with fixable viability dye (eFluor®780, eBioscience) to exclude dead cells and stained as above.
5–10×106 thymocytes harvested 0–48 h after birth were transferred i.v. to congenic recipients lethally irradiated with a split dose of 1300 rads. The next day, 3–6×106 congenic BM cells were transferred. Recipient mice were analyzed at least 8 weeks later.
The mouse was anesthetized with ketamine/xylazine and the dorsal side of the ear was attached on a plastic coverslip mounted on a 37°C heating stage. The ventral side of the ear was imaged with a Zeiss LSM 7MP equipped with a Chameleon laser (Coherent). GFP was excited at 910 nm and detected with a 500–550 nm emission filter. Images were acquired with Zen (Zeiss), and time-lapse images and cell tracks generated with Imaris 5.7.2 (Bitplane). The velocities and turning angles of cells were calculated with MATLAB (The MathWorks, Inc.). Annotation and final compilation of videos were with After Effects 7.0 software (Adobe).
In the course of studies characterizing LN subcapsular and interfollicular regions (17), we identified a population of IL-7Rhi cells enriched in proximity to CD169+ macrophages (Suppl. Fig. 1A). By flow cytometric analysis, the IL-7Rhi cells in peripheral LNs (PLNs) were also high for CXCR6, detected using CXCR6-GFP/+ reporter mice (14), mostly CCR6hi, and could be further divided into populations of TCRγδ+, TCRαβ+ and non-T cells (Suppl. Fig. 1B). The γδ T cells represented about 0.1% of PLN cells but were much less abundant in mesenteric LNs and spleen (Suppl. Fig. 1C). A similar population of PLN enriched γδ T cells was recently reported by a number of groups and shown to correspond to an innate IL-17 producing cell population (6, 7, 9, 13, 19). However, the basis for the enrichment of these cells in skin draining LNs compared to mucosal LNs and spleen was not clear and led us to ask whether there was a related cell population in the skin. One study characterizing a novel marker, SCART2, present on PLN γδ T cells noted the presence of SCART2+ Vγ5− γδ T cells in the dermis, providing support for this possibility though the cells were not characterized further (13). Flow cytometric analysis of enzyme-digested epidermis revealed the dominant presence of the expected TCRγδhi Vγ5+ DETCs and few other T cells (Fig. 1A). Strikingly, the dermis also contained a large population of γδ T cells but these cells were TCRγδint, lacked Vγ5 (Fig. 1A) and instead had a surface phenotype similar to the cells in PLNs: they were CCR6hi, IL-7Rhi, CXCR6hi and high for αE integrin (CD103; Fig. 1A, B). DETCs differed from dermal γδ T cells in being CCR6lo and IL-7Rlo (Fig. 1A, B). The DETC-phenotype cells present in dermal sheet preparations (Fig. 1A) were most likely due to contamination by small amounts of epidermis.
The close phenotypic similarity between dermal γδ T cells and PLN γδ T cells, together with the high expression of CCR6, a receptor associated with IL-17 producing cells (20), led us to ask whether dermal γδ T cells were pre-committed to IL-17 production. Indeed, when cells prepared from dermal and epidermal sheets were activated by PMA/I, dermal γδ T cells promptly produced large amounts of IL-17 whereas DETCs did not (Fig. 2A). Consistent with this pre-commitment, dermal γδ T cells had high constitutive expression of IL-17A, IL-17F, and IL-22 transcripts (Fig. 2B). Correspondingly, these transcripts were more abundant in dermis than epidermis (Fig. 2B). RORγt promotes IL-17 expression and has been detected in γδ T cells isolated from total skin (11). Analysis of separated epidermal and dermal sheets from RORγt-GFP mice revealed reporter expression in dermal γδ T cells but no expression in DETCs (Fig. 2A).
Previous work has shown that IL-1β and IL-23 are sufficient to promote IL-17 production by PLN γδ T cells after 3 days of in vitro culture (6). We found that both CCR6+ TCRγδint dermal and PLN cells, but not CCR6− TCRγδhi DETCs, began making IL-17 within 8 h of incubation with IL-1β and IL-23 (Fig. 2C). Our findings contrast with a recent report suggesting IL-17 was made selectively by DETCs following incubation with IL-1β and IL-23 (4). We suspect that this discrepancy reflects contamination of the DETC preparations with dermal γδ T cells.
It is well established that Vγ5+ precursors of DETCs are produced in a wave within the e14–17 fetal thymus and are not further produced in the adult thymus (2). Thus, Vγ5+ γδ T cells seed the epidermis in fetal life and give rise to a population of DETCs that are locally maintained (21). Recent studies have demonstrated that IL-17 committed Vγ5- γδ T cells appear in the thymus near the time of birth but remain detectable in significant numbers in the adult thymus (22, 23). Based on these observations we anticipated that dermal and PLN γδ T cells might be replaced by BM-derived cells following irradiation and reconstitution of adult mice, as recently observed for intestinal γδ T cells (24). However, after 8–26 weeks of reconstitution by BM cells, there was little if any reconstitution of dermal and PLN CCR6hi γδ T cells (Fig. 3A and Suppl. Fig. 2A), despite ablation of the majority of the endogenous cells by irradiation. The small number of CCR6hiγδ T cells we did observe were mostly radiation-resistant host CCR6hiγδ T cells. Transferring BM into irradiated TCRδ−/− mice did not improve reconstitution (Suppl. Fig. 2A). We then asked whether cells with reconstituting potential might be most prominent in the thymus during the perinatal window when they first appear (Fig. 3B and (22, 23)). When irradiated adult mice were given a combination of BM and neonatal thymocytes, we observed significant reconstitution of CCR6hi dermal and PLN γδ T cells that produced IL-17 when stimulated with PMA/I (Fig. 3C and Suppl. Fig. 2A). These findings suggest that IL-17 producing dermal and PLN γδ T cells can be regenerated by precursors arising in the perinatal thymus, but not adult BM. It will be interesting in future studies to characterize the TCR repertoire of these cells to determine whether perinatal thymic precursors reconstitute the full repertoire of IL-17 producing γδ T cells.
Taking advantage of the abundant expression of CXCR6 in skin γδ T cells (Fig. 1C) we used two-photon laser scanning microscopy (TPLSM) to examine their migration dynamics, distinguishing cells in the epidermis and dermis based on their location with respect to the collagen (visible from second harmonic generation emission) that separates these layers (Fig. 4A). In the dermis, 20–35% of CXCR6-GFP+ cells were TCRγδint cells, whose GFP expression was approximately two-fold more abundant than the TCRβ+ and TCR- GFP+ cells (Suppl. Fig. 2B). CXCR6-GFP+ cells in epidermal suspensions were almost exclusively TCRγδhi DETCs and had the same GFP intensity as the dermal γδ T cells (Suppl. Fig. 2B). In the imaging analysis, we compared GFP bright cells in the epidermis (DETC) to GFP bright cells in the dermis, reasoning that the latter would be highly enriched for dermal γδ T cells (Fig. 4A, B, Suppl. Videos 1–3). As a second approach to allow imaging-based identification of dermal γδ T cells we reconstituted WT (wild type) mice with CXCR6-GFP/+ neonatal thymocytes and WT BM. In these chimeric mice, at least 80% of the GFP+ dermal cells were TCRγδint T cells (Suppl. Fig. 2B) whereas the DETCs remained GFP- host derived cells, as expected (2). Imaging the dermis of these mice revealed similar GFP+ cell behaviors to the GFPhi dermal cells in the CXCR6-GFP/+ non-chimeric mice (Fig. 4C and Suppl. Video 4). DETCs and dermal γδ T cells exhibited marked differences in behavior (Fig. 4D–F and Suppl. Videos 1–4): while DETCs had a highly dendritic morphology and were immobile (Fig. 4D and Suppl. Video 5), dermal γδ T cells were more rounded and moved at average speeds of 3–5 μm/min (range 1.4–7.6 μm/min) (Fig. 4F and Suppl. Fig. 2C), pausing occasionally and turning in various directions with a mean angle of ~85 degrees (Suppl. Fig. 2D). Dermal γδ T cells showed a similar extent of displacement over time whether tracked in CXCR6-GFP/+ or chimeric mice, whereas DETCs showed no displacement over time (Fig. 4G, and Suppl. Fig 2E). We suggest that the dermal environment is more permissive for cell motility than the tightly adherent epidermis and dermal γδ T cells take advantage of this property to achieve surveillance coverage of a larger 3-dimensional area.
In summary, although several studies have identified a critical role for IL-17 in the protective response mounted in the early hours to days of cutaneous infections (4, 5, 12), the precise cellular source of this IL-17 was not defined (5, 12) or was suggested to be DETCs (4). Our work demonstrates that dermal γδ T cells are a major subset of IL-17 pre-committed cells in the skin. Given the ability of these cells to secrete IL-17 within hours of IL-1β and IL-23 exposure, we propose that they contribute to the rapid IL-17 production following infections. These cells may also contribute to the IL-17 dependent psoriasis-like skin pathology that occurs following repeated intradermal IL-23 injection (25).
The reconstitution of CCR6+ dermal γδ T cells by neonatal thymocytes but not BM suggests that these cells are seeded from the perinatal thymus and subsequently maintained in the periphery. Consistent with this model, BrdU-labeling studies suggest these cells divide at a low rate in situ (data not shown). How they are maintained in the dermis remains to be explored, although given their high IL-7Rα expression (Fig. 1B), IL-7 is a candidate trophic factor. Early studies of transgenic mice over-expressing IL-7 in keratinocytes revealed a marked expansion of Vγ5-TCRγδint T cells in the dermis, and it was suggested that local γδ T cells were responding to the cytokine (26, 27). It is likely that the major population expanded in the skin of these mice was the IL-17 pre-committed, CCR6+ γδ T cells; overproduction of IL-17 may well account for the dermatitis suffered by some of the transgenic animals (26, 27). Future studies should examine the possibility that IL-7 production by keratinocytes is important for the local maintenance of dermal γδ T cells. Alteration of IL-7 abundance may be a method to increase or decrease dermal γδ T cell function for therapeutic benefit.
We thank Lisa Kelly for cell sorting and helpful discussion, Irina Grigorova for Matlab code used in data analysis, Andrea Reboldi and Oliver Bannard for helpful discussion, Ying Xu and Jinping An for expert technical assistance and Jennifer Bando and Richard Locksley for RORγt-GFP mice. J.G.C is an Investigator of the Howard Hughes Medical Institute.
1This work was supported by NIH grant AI40573.