A Keratinocyte-Responsive γδ TCR Is Necessary for Dendritic Epidermal T Cell Activation by Damaged Keratinocytes and Maintenance in the Epidermis¹

Although αβ T lymphocytes comprise most of the T cells in the peripheral blood and lymphoid system, γδ T lymphocytes make up the majority in certain epithelial tissues such as the skin (1, 2). A unique population of γδ T lymphocytes, designated dendritic epidermal T cells (DETC),³resides in the murine skin. DETC were first discovered as Thy-1 positive cells (3, 4) and were later found to have a canonical Vγ3Jγ1Cγ1/Vδ1Dδ2Jδ2Cδ TCR with no junctional diversity (5). Because DETC have such a conserved TCR, it has been suggested that they have a limited Ag repertoire. DETC respond to Ag expressed on damaged, stressed, or transformed keratinocytes in vitro by means of their γδ TCR (6, 7). DETC and keratinocytes also interact in vivo, because γδ DETC play roles in wound repair through the production of keratinocyte growth factors (8). The TCR is likely used in the activation of DETC during wound healing, because TCRδ^−/− mice have defects in wound repair (8). However, the actual role of the γδ TCR in DETC localization to the skin, maintenance in the epidermis, and activation in vivo is still unknown.

During development, γδ T cells go through a programmed rearrangement of the γδ TCR and subsequently migrate to various epithelial tissues (5, 9, 10, 11, 12, 13, 14, 15). The first wave of rearrangements occurs at fetal days 14–16, resulting in Vγ3/Vδ1 TCR-expressing thymocytes that migrate to the epidermis (5, 9). This precisely timed rearrangement and subsequent migration suggests that the TCR may be important in the final localization of T cells to the skin. However, DETC expressing a unique γδ TCR were still detected in the skin of Vδ1^−/− and Vγ3^−/− mice, suggesting that the canonical Vγ3/Vδ1 TCR is not necessary for DETC migration (16, 17). Interestingly, these replacement DETC retained keratinocyte reactivity, and some could still bind the Vγ3/Vδ1-specific Ab, indicating that TCR conformation may be important in the localization of DETC to the skin. Yet Vγ2/Vδ5 TCR transgenic mice have DETC expressing the transgene in the skin, and these DETC are unresponsive to keratinocytes (18). These results suggest that neither the Vγ3/Vδ1 TCR nor a TCR with keratinocyte specificity is necessary for localization to the epidermis.

The importance of the DETC TCR in homing was revisited when it was found that endogenous δ-chain rearrangement can still occur in TCR transgenic mice (19). Ferrero et al. (20) proposed that the endogenous Vδ1 rearrangement may permit DETC to migrate to the skin even in TCR transgenic mice. They did not detect DETC in TCRδ^−/− or TCR transgenic TCRδ^−/− mice and concluded that Vδ1 is necessary for localization to the skin. Clearly, the role of the TCR and TCR specificity in migration to the epidermis is still unresolved.

Wild-type DETC home to the epidermis and reside there over the life of the mouse. It is not known whether keratinocyte Ag recognition by the TCR is necessary for DETC to be maintained in the epidermis over time. When TdT expression was forced early in the fetal thymus, DETC with variable junctional segments were detected in the epidermis of adult mice (21). However, the wild-type invariant Vγ3/Vδ1 DETC had a much greater proliferative advantage in adults, suggesting that DETC need the keratinocyte-responsive TCR to be maintained in the epidermis over time. This is also supported by the findings that skin from Vγ3^−/− or Vδ1^−/− mice still contains DETC with keratinocyte-responsive TCRs (16, 17). Therefore, γδ TCR recognition of the keratinocyte Ag may be important for DETC maintenance in the epidermis.

DETC play a critical role in wound repair (8). The functional response of DETC to keratinocyte damage likely requires the TCR, because TCRδ^−/− mice have a delay in wound repair, and DETC are known to require the TCR for recognition of damaged keratinocytes in vitro (6, 8). Therefore, the γδ TCR may be used by DETC to become activated during wound-healing processes in vivo. To examine whether αβ DETC reside in the epidermis of TCRδ^−/− mice, whether they are retained over time, and whether they are activated by damaged keratinocytes during wound repair, we examined DETC in TCRδ^−/− mice. Our results suggest that a keratinocyte Ag-responsive γδ TCR is not necessary for DETC to home to the epidermis, but is necessary for DETC to become activated during wound repair and to be maintained in the epidermis.

TCRδ^−/− mice on the C57BL/6 background were purchased from The Jackson Laboratory (Bar Harbor, ME). Both TCRδ^−/− and C57BL/6 mice were bred in our facility. All mice were housed in specific pathogen-free conditions at The Scripps Research Institute animal facility. Mice were used at 10–14 wk of age.

The following Abs were used to stain epidermal sheets or epidermal cell suspensions: PE-γδ TCR (GL3), PE-CD3ε (145-2C11), PE-CD103 (M290), PE-CD49b (HM-a2), PE-CD49d (R1-2), PE-Vα2 (B20.1), PE-CD25 (PC61), PE-CD69 (H1.2F3), PE-CD24 (M1/69), PE-CD4 (GK1.5), PE-TNF-α (MP6-XT22), FITC-αβ TCR (H57-597), FITC-CD44 (IM7), FITC-CD8α (53-6.7), FITC-Thy1.2 (53-2.1), FITC-Vβ screening panel (catalog no. 0143KK), and allophycocyanin-Thy1.2 (53-2.1) (BD PharMingen, La Jolla, CA). FITC-CD5 (53-70.3) was obtained from eBioscience (San Diego, CA). Anti-FcεRγ chain (FcεRIγ) sera was collected from peptide-inoculated rabbits and used with FITC-goat anti-rabbit secondary Ab (Jackson ImmunoResearch, West Grove, PA). Preimmune sera from the rabbits were collected and used as a control.

Mice were anesthetized with isoflurane. The mouse backs were shaved, back skin was pulled up, and a sterile 3-mm punch tool was used to create three sets of full-thickness wounds as previously described (8). Mice were caged individually, and wounds were left uncovered.

Epidermal sheets were prepared as previously reported (8). Briefly, following hair removal, ears were excised and split into dorsal and ventral halves. The dermal side was exposed to 3.8% ammonium thiocyanate in PBS for 15 min, and epidermal sheets were separated from dermis and then washed in PBS. The epidermal sheets were fixed in acetone and incubated in 1–2 μg/ml Ab for 1 h at 37°C. Sheets stained with cholera toxin B (Sigma-Aldrich, St. Louis, MO) were incubated in 0.0625 μg/ml fluorescein-labeled cholera toxin B. The stained sheets were rinsed in PBS and mounted on slides with anti-fade mounting medium (DAKO, Carpinteria, CA).

Epidermal cells were prepared from the skin of wild-type and TCRδ^−/− mice (8). Skin from wounded mice was excised including a 3-mm border around the wound. Briefly, skin was incubated on 0.3% trypsin/GNK solution for 1–2 h at 37°C, and the epidermis was separated from the dermis. Epidermis was incubated for 10 min in 0.3% trypsin/GNK solution at 37°C for two sequential treatments. The epidermal preparation was enriched for live cells by centrifugation over lympholyte M (Cedarlane Laboratories, Hornby, Ontario, Canada), and cells recovered from the interface were either immediately stained for FACS analysis or incubated overnight with or without 5 μg/ml Con A (Sigma-Aldrich).

Expression of surface markers was examined by staining cells with 0.5 μg/ml Ab in PBS containing 2% FCS and 0.2% NaN₃ for 20 min at 4°C. Intracellular staining was performed using a kit (Caltag, Burlingame, CA) following manufacturer’s directions. Cells were analyzed on a FACSCalibur flow cytometer using CellQuest software (BD Biosciences, Franklin Lakes, NJ).

For the enumeration of DETC, ×200 images of at least seven random fields were taken of each epidermal sheet, and at least four were counted in a grid. Three mice were examined from each strain per time point. Mean values ± SD were calculated, and density was expressed as the number of DETC per square millimeter.

To quantify lipid raft polarization, full-thickness wounds were made in each ear with a punch biopsy tool, and epidermal sheets were prepared 15–20 h later. Epidermal sheets were stained with fluorescein-labeled cholera toxin B and Abs specific for the γδ TCR. Images of lipid rafts were taken with a digital fluorescence microscopy system using SPOT 32 software (Diagnostic Instruments, Sterling Heights, MI). Random cells near the wound site were examined at ×1000 magnification and recorded as polarized or not polarized. Over 100 cells from at least six wounded ears were counted for each mouse strain. Cells away from the wound site or observed in nonwounded mice were not round and did not have polarized lipid rafts.

To examine whether DETC are present in the epidermis of TCRδ^−/− mice, epidermal sheets were prepared and stained with Abs specific for the γδ or αβ TCR. As previously reported (22, 23), DETC that express the αβ TCR reside in the epidermis of TCRδ^−/− mice (Fig. 1). This contradicts findings by Ferrero et al. (20) in which no DETC were detected in the epidermis of TCRδ^−/− mice. To verify our results, we performed these experiments on two different sets of mice obtained directly from The Jackson Laboratory and detected αβ DETC by both immunofluorescent staining of epidermal sheets (Fig. 1) and flow cytometry of isolated epidermal cell suspensions (Fig. 2). All αβ DETC expressed CD3ε, and all CD3ε-expressing DETC were αβ TCR positive (data not shown). TCRδ^−/− mice have approximately half the number of DETC as compared with wild-type mice. Furthermore, the αβ DETC appear slightly less dendritic in morphology and reside in clumps of cells that do not uniformly cover the epidermis as compared with those of wild-type animals (Figs. 1 and 3).

There may be other surface proteins such as integrins that play a key role in the localization of DETC to the skin. Like wild-type DETC, the αβ DETC express CD103 (Figs. 1 and 2), which has been suggested to play a key role in the homing of intraepithelial lymphocytes (IELs) to the epithelia of the gut (24) and skin (25). We determined that wild-type and αβ DETC down-regulate CD103 upon in vitro stimulation (data not shown). Therefore, CD103 with or without other surface proteins such as E-cadherin may be an alternative mechanism for the attachment and homing of DETC to the epidermis, whereas the Vγ3/Vδ1 TCR is not.

αβ DETC are phenotypically similar to wild-type DETC as determined by the analysis of epidermal sheets and epidermal cells by immunofluorescence or flow cytometry (Figs. 1 and 2). One interesting difference between the two populations of DETC is the expression of CD8αα on a subset of the αβ DETC, whereas γδ DETC are CD8⁻ (Fig. 2,A). CD8αα is also commonly expressed on gut γδ IELs (26). Like DETC from wild-type mice, DETC from TCRδ^−/− mice have constitutive expression of CD44 and CD3ε, no expression of CD5, and low expression of CD24 (Figs. 1 and 2, B and C; data not shown). Taken together, phenotypic analysis of αβ DETC suggests that they express similar surface molecules to wild-type DETC, except for αβ TCR expression and a subpopulation with CD8αα expression.

We examined Vβ chain usage by αβ DETC isolated from TCRδ^−/− mice to determine whether they retain a conserved TCR repertoire that may recognize damaged keratinocytes. Vβ chain usage by αβ DETC was examined by flow cytometry in comparison to splenic cells isolated from the same mouse. In all mice except one, individual Vβ chains were expressed proportionally by DETC and splenic T cells (Fig. 3,A). This indicates that the DETC Vβ repertoire is as polyclonal and diverse as the spleen repertoire. For example, the major Vβ chains used in the spleen are Vβ8.1 and -8.2, and they are also used most frequently by the DETC isolated from the TCRδ^−/− mice (Fig. 3,A). One mouse demonstrated preferential usage of the Vβ8.1/8.2 family in the epidermis that differed from the spleen (Fig. 3 A3). However, this skewing was not detected in any of the other seven mice tested. Therefore, a conserved conformation of TCR is not necessary for the homing of DETC to the epidermis. Furthermore, the highly diverse population of αβ DETC suggests that they do not recognize one conserved keratinocyte Ag like the wild-type DETC.

To determine whether αβ DETC can proliferate and be maintained in the epidermis, we analyzed the density of DETC in young and aged mice. The number of DETC per square millimeter was examined by immunofluorescence and counted in a grid. Like wild-type DETC, the αβ DETC in TCRδ^−/− mice appear in clusters in the epidermis soon after birth. The cells within each cluster express the same αβ TCR (Fig. 3,B) and increase in number over the first few months (C). This suggests that the canonical γδ TCR is not necessary for DETC to proliferate in the epidermis, even though it takes longer for the αβ DETC to accumulate, and the number of αβ DETC never reaches that of γδ DETC in wild-type mice (Fig. 3,C). Adult TCRδ^−/− mice have about half the number of DETC as compared with wild-type (Fig. 3 C). Thus, initial proliferation allows for the accumulation of αβ DETC; however, the wild-type TCR is necessary for full establishment of the population.

The number of DETC in wild-type mouse skin remained constant in young and aged animals (Fig. 3,C). In contrast, the number of DETC in TCRδ^−/− skin decreased significantly by 1 year of age (Fig. 3 C). By immunofluorescence, we detected fewer cells in the clusters and more distance between each cluster. This suggests that, even though the wild-type TCR is not necessary for DETC to localize to the epidermis, it is necessary for DETC to be retained there over the life of the animal. Our results indicate that the maintenance of DETC in the skin may require a TCR that recognizes the Ag expressed by damaged or stressed keratinocytes.

To determine whether the αβ DETC had a functional TCR, we examined the TCR complex and activation potential of αβ DETC. Unlike peripheral αβ T cells, wild-type DETC express FcεRIγ as a homodimer or in association with the ζ-chain of the CD3 complex (27, 28). To find out whether αβ DETC have the same TCR components as wild-type DETC, we stained epidermal cells and epidermal sheets with Abs to FcεRIγ. We discovered that the majority of αβ DETC express this signaling component (Fig. 4). There was a small population of TCR-expressing DETC that did not express FcεRIγ, and a small population of non-CD3ε-expressing cells that expressed FcεRIγ (Fig. 4). The non-CD3ε-expressing cells are not dendritic cells (data not shown) and were not detected in wild-type mice. Thus, they make-up a Thy1.2-positive, TCR-negative subset we have previously detected in the TCRδ^−/− epidermis (data not shown) that may be similar to cells reported in nude mice (29). The majority of αβ DETC were FcεRIγ positive; this suggests that any defect in αβ DETC activation is not due to the lack of TCR-associated signaling proteins on the DETC.

To examine the functionality of the αβ TCR on DETC, we isolated epidermal cells and stimulated them with either Con A or Abs to the CD3 complex. After stimulation, αβ DETC increased expression of early activation markers such as CD25 and CD69 (Fig. 5,A, lower panels), which are normally up-regulated on activated T cells including wild-type DETC (upper panels). The αβ DETC also up-regulate TNF-α production in response to the direct stimulation (Fig. 5 B). These experiments indicate that the αβ TCR complex on αβ DETC functions in response to direct stimulation; therefore, any defects in αβ DETC activation in vivo may be attributed to the lack of keratinocyte responsiveness by the TCR.

Because the αβ DETC express a functional TCR, we examined whether this TCR can be used to recognize and respond to damaged keratinocytes like the wild-type TCR. Epidermal cells were isolated from wild-type and TCRδ^−/− mice following wounding and examined for up-regulation of the early activation marker CD25. Wild-type DETC up-regulated CD25 in response to keratinocyte damage, most notably at 1 day postwounding, whereas CD25 expression on αβ DETC was unchanged (Fig. 5 C). Similar data was obtained at 48-h postwounding; however, there were fewer cells up-regulating CD25 in wild-type mice. As a further measure of activation, cytokine production during wound repair was examined. TNF-α production and integrin (CD49b and CD49d) up-regulation were also impaired in αβ DETC (data not shown), as compared with wild-type DETC during wound repair. These results demonstrate that, although the αβ TCR is functional, it cannot be used to recognize damaged keratinocytes.

To analyze the responsiveness of αβ DETC directly in the tissue, we prepared epidermal sheets of wounded ears from wild-type and TCRδ^−/− mice. Cholesterol-enriched membrane microdomains called lipid rafts are regions used for cell signaling. Lipid rafts form platforms for signaling during T cell activation, which contain the TCR and localize to the site of contact. We examined lipid raft polarization by DETC in response to damaged keratinocytes by wounding ears and staining the wounded epidermis in situ with cholera toxin B, which labels sphingomyelin/cholesterol lipid domains. The wild-type DETC clearly polarized lipid rafts at the wound site in response to the damaged keratinocytes (Fig. 6). The cholera toxin B colocalized with the γδ TCR at these polarized sites. Wild-type DETC away from the wound site did not have polarized lipid rafts. In contrast, the αβ DETC from TCRδ^−/− mice did not polarize lipid rafts either at or away from the wound site (Fig. 6). This clearly indicates that the αβ DETC do not respond to damaged keratinocytes in vivo. Taken together, these data suggest that DETC need more than a functional TCR to respond to damaged keratinocytes during wound repair. The TCR must also have a conformation that recognizes Ag expressed by damaged keratinocytes in order for the cells to become activated and produce cytokines during wound repair.

In this study, we examined the involvement of the keratinocyte-responsive γδ TCR in DETC localization to the epidermis, maintenance in the skin, and activation by damaged keratinocytes during wound repair. DETC express a canonical γδ TCR that recognizes Ag expressed on damaged keratinocytes (6, 7). Several groups have proposed that the TCR plays an important role in the homing of DETC to the epidermis. Ferrero et al. (20) suggested that the Vδ1 chain is not allelically excluded in TCR transgenic mice and is used by DETC to home to the epidermis. This group did not find αβ DETC in the TCRδ^−/− or the TCR transgenic TCRδ^−/− epidermis. However, we readily observed αβ T cells in the epidermis of TCRδ^−/− mice. Similar results have been observed by other groups (22, 23). Our data provide further evidence that neither a Vγ3/Vδ1 TCR nor a keratinocyte-responsive TCR is necessary for DETC to localize to the epidermis.

An alternative mechanism for DETC homing to the epidermis may be expression of integrins that direct DETC to the epidermis. We analyzed expression of CD103, which is an integrin suggested to have a role in the homing of IELs to the gut epithelia (24) and skin (25). We found that both wild-type and αβ DETC express CD103. CD103 is down-regulated upon DETC activation (data not shown). This correlates with our findings that DETC activated by keratinocyte damage in vivo will round-up, which may be due to a decrease in CD103 expression as well (8). Taken together, this may indicate that CD103 may be involved in the attachment and retention of DETC within the epidermis. Because CD103 can bind to E-cadherin, these integrins may have a role together in DETC homing to the skin.

After DETC localize to the epidermis in wild-type mice, they proliferate rapidly and reach a uniform density by 3 wk. We detected an increase in the number of αβ DETC between 3 wk and 3 mo of age, suggesting that the cells can proliferate. The DETC appeared to be proliferating in situ and not simply continually migrating to the skin, because clusters of cells with the same Vα or Vβ increased in cell number. Thus, the Vγ3/Vδ1 TCR is not necessary for slow proliferation of DETC in the epidermis. However, the αβ DETC do not reach maximal density like wild-type DETC; therefore, a keratinocyte-responsive TCR may be necessary for DETC maintenance in the skin. Once DETC localize to the epidermis in wild-type mice, they maintain a population there for the life of the mouse. In contrast, the number of αβ DETC in the TCRδ^−/− mice decreases by almost 50% within 1 year of age. It is possible that, due to a lack of TCR stimulation, αβ DETC do not keep up with the loss of epidermal cells over time. This supports the hypothesis that a keratinocyte-responsive TCR may be necessary for the maintenance of DETC in the epidermis. The TCR may provide the stimulus necessary for proliferation and survival in the epidermis.

We analyzed the components of the αβ TCR complex on DETC and found expression of the FcεRIγ chain as seen in the wild-type DETC TCR complex. In rare cases, FcεRIγ has been associated with the TCR in αβ T cells. For example, early αβ TCR expression in some TCR transgenic mice causes double-negative αβ T cells to express the FcεRIγ as part of the CD3 signaling complex (30). The lack of δ-chain rearrangement may have caused early αβ TCR expression and allowed the FcεRIγ to be a part of the epidermal T cell signaling complex. The αβ DETC can be activated through their TCR and use the same TCR components as wild-type DETC. Therefore, the αβ DETC have a functional TCR complex that can respond to direct stimulation.

In mice lacking Vδ1 or Vγ3, DETC in the epidermis have a surprisingly conserved TCR repertoire, and many cells retain a TCR conformation specific for stressed keratinocytes (16, 17). αβ DETC express a broad range of TCRs and do not become activated by damaged keratinocytes that normally stimulate wild-type Vγ3Vδ1 DETC, suggesting that αβ DETC do not have the TCR conformation used by wild-type DETC. We have previously shown that TCRδ^−/− mice have defects in wound repair and a lack of keratinocyte growth factor production at the wound site (8). Our results suggest that, although αβ DETC have the ability to become activated like wild-type DETC, without the keratinocyte-responsive TCR they may not be able to respond to keratinocyte damage in vivo. We have a unique epidermal peeling procedure that allows us to examine the DETC as they function in the tissue itself. Lipid rafts polarize to the site of T cell stimulation upon activation by an APC. The TCR and associated complex are translocated into the lipid rafts to strengthen the signal. Using epidermal peels, we were able to identify this early stage of DETC activation as it occurs temporally during wound repair. Translocation of the lipid rafts to the site of keratinocyte damage was defective in αβ DETC from TCRδ^−/− mouse wounds. Other obvious signs of T cell activation, such as IL-2R up-regulation and cytokine production, were also defective in DETC isolated directly from the wound sites. Therefore, our results suggest that a keratinocyte-responsive TCR is necessary for DETC activation by damaged keratinocytes during wound repair and to be maintained in the epidermis.

DETC have intimate interactions with epithelial cells in vivo and have been shown to play a role in keratinocyte proliferation during wound repair (8). DETC are also involved in the regression of skin tumors (31) and in the suppression of cutaneous graft-vs-host disease (23). DETC are representative of epithelial γδ T cells in other tissues such as the gastrointestinal tract where γδ IELs are important in repair from dextran sodium sulfate-induced colitis (32). We have shown that, although the TCR may not be involved in the homing of DETC to the epidermis, it plays a key role in activation in vivo and maintenance of DETC in the epidermis. Our results establish a role for the DETC TCR in vivo that may be useful in the future design of therapies that involve intraepithelial γδ T lymphocytes.

We thank Marieke Svoboda and Dr. Stephanie Rieder for laboratory assistance and manuscript review. We also thank Dr. R. Ulevitch for use of his immunofluorescent microscope and digital camera.