Occludin Expression in Epidermal γδ T Cells in Response to Epidermal Stress Causes Them To Migrate into Draining Lymph Nodes Free

The epidermis provides a barrier against insults from the environment, including pathogens, UV light, and chemicals. There are two types of immune cells in murine epidermis: Langerhans cells (LCs), which capture invading pathogens and bring them to the draining lymph nodes, leading to immune responses, and dendritic epidermal γδ T cells (DETCs), which recognize stress- or damage-induced molecules in keratinocytes (1, 2). DETCs have been proposed to eliminate stressed or DNA-damaged keratinocytes, thereby helping to prevent tumor development (3). In addition, in wounded epidermis, DETCs secrete cytokines and play key roles in wound healing, thereby contributing to epidermal homeostasis (4, 5). Recent work showed that γδ T cells promote contact hypersensitivity (CHS) responses (6, 7), skin graft rejection (8), and Ab production (9, 10). However, the molecular mechanisms by which DETCs activate these immune responses are not well defined.

DETCs are generated in the first wave of Vγ locus rearrangements, expressing canonical Vγ5Vδ1 TCR [nomenclature of Heilig and Tonegawa (11)], and subsequently migrate to the epidermis. These cells develop in the thymus only during the early embryonic period and must be selected via mechanisms involving Skint-1, which is expressed in thymic epithelial cells and keratinocytes (12–14). Thymocytes expressing this canonical γδTCR engage Skint-1⁺ cells and upregulate Egr-3 expression, thus becoming IFN-γ–producing γδ T cells (15). In the epidermis at steady-state, γδTCRs expressed in DETCs signal constitutively and form the immunological synapse at squamous keratinocyte tight junctions (16). This semiactivated state may be important for interactions between DETCs and keratinocytes.

In stressed epidermis, DETCs are further activated in response to molecular stress signals expressed in that layer. Several costimulatory molecules involved in the interaction of DETCs and stressed keratinocytes have been identified. For example, DETCs express junctional adhesion molecule–like protein (JAML), which interacts with Coxsackie and adenovirus receptor (17). In addition, DETCs upregulate CD100, which induces costimulatory signals by interacting with plexin B2 on keratinocytes (18). Activated DETCs undergo morphological changes and detach from tight junctions, subsequently reorienting toward basal layers and LCs (16). Pathogens and other insults, including chemicals that disrupt tight junctions, are also sensed by γδTCRs expressed in DETCs. Although there is some evidence that γδ T cells function as innate-like cells that augment adaptive immune responses, it remains unclear how activated γδ T cells interact with other immune cells and whether they migrate out of the epidermis. In this study, we found that DETCs express occludin, a tight junction molecule, in UVB-irradiated or allergen-treated epidermis. Occludin-deficient DETCs exhibited impairments in morphological changes and motility upon activation. Furthermore, allergen-induced CHS responses in occludin-deficient mice were much weaker than those in wild-type mice. These results suggested that DETCs exert their functions by expressing occludin, causing them to become motile and ultimately leading to the activation of other immune cells.

C57BL/6J mice were purchased from Sankyo Laboratories (Shizuoka, Japan). Occludin-deficient mice (Ocln^−/−) (19) on the C57BL/6 background were provided by M. Furuse and were obtained from the Center for Animal Resources and Development at Kumamoto University. Littermates generated from Ocln^+/− × Ocln^+/− crosses were used as occludin-deficient (Ocln^−/−) or wild-type control (Ocln^+/+) mice. TCRδ^−/− mice (20) were housed in our mouse facilities. KikGR mice were described previously (21). Six- to ten-week-old male and female mice were used in all experiments. All mice were maintained under specific pathogen–free conditions in the animal facility at Tokyo University of Science, and experimental studies were approved by the university’s Animal Care and Use Committee.

Abs against TCRγδ (UC7-13D5; BioLegend), CD25 (PC61.5; eBioscience), CD69 (H1.2F3; eBioscience), CD11c (N418; BioLegend), I-A/I-E (M5/114.15.2; eBioscience), CD100 (BMA12; eBioscience), CD3 (145-2C11; eBioscience), or Vγ5 (7-17; BD Pharmingen), conjugated to FITC, PE, allophycocyanin, or biotin, were obtained from the indicated suppliers. Allophycocyanin-, PE-Cy7–, or Brilliant Violet 421–conjugated streptavidin (BioLegend) was used for the second step. CD16/32 (2.4G2) Abs were used for Fc blocking. Purified anti-TCRγδ (UC7-13D5, functional grade) and anti-CD3 (145-2C11, functional grade) were purchased from eBioscience. Rabbit anti-occludin (711500; Invitrogen) and Alexa Fluor 647– or Alexa Fluor 488–conjugated goat anti-rabbit IgG (A21245; Invitrogen) were used for histochemistry. 2,4-Dinitrofluorobenzene (DNFB) was purchased from Sigma-Aldrich. PMA (50 ng/ml) and ionomycin (0.5 μM; both from Sigma-Aldrich) were used to stimulate cells.

Epidermal γδ T cell lines were established from wild-type, Ocln^−/−, KikGR, and KikGR/Ocln^−/− mice by culturing epidermal cells in RPMI 1640 complete medium containing 5% FCS and 10 ng/ml IL-2. Epidermal stromal cells were obtained from a long-term epidermal cell culture that contained heterogeneous cell populations and maintained in RPMI 1640 complete medium containing 5% FCS.

To obtain UVB-irradiated epidermis, euthanized mice were placed in a GS Gene Linker UV Chamber (Bio-Rad), and their ear skin was irradiated at 130 mJ/cm² on the dorsal and ventral sides. In cell-migration experiments, dorsal back skin of anesthetized mice was clipped and irradiated at 130 mJ/cm². Epidermal mechanical stress was administered by stripping the epidermis five times with normal Scotch tape.

Dorsal and ventral sides of ear skin were incubated for 1 h at 37°C in 0.2% EDTA/PBS, and epidermal sheets were peeled off. These sheets were placed in slide glasses, fixed in cold acetone for 10 min, rinsed in PBS, and blocked in 5% BSA/PBS. Samples stained with Abs against Vγ5 were analyzed by fluorescence microscopy (KEYENCE BZ-9000). For staining of occludin in epidermal sheets, ventral sides of skin were fixed in cold ethanol for 30 min, rinsed in PBS, and incubated in 3.8% ammonium thiocyanate/phosphate buffer for 17 min at 37°C. These samples were rinsed in PBS, and epidermal sheets were peeled off. Samples were treated for 1 h with blocking buffer containing 5% goat serum, 10% FCS, 0.3% Triton X-100, and Fc-block in PBS and then stained overnight at 4°C with primary Abs. Following staining with secondary Abs for 1 h at room temperature, the samples were analyzed using an Olympus FluoView FV10i confocal microscope.

Epidermal cell suspensions were prepared from epidermal sheets isolated by incubating ear skin for 1 h at 37°C in 0.2% trypsin in PBS. Cells prepared from lymph nodes or epidermis were incubated with anti-CD16/CD32 to block Fc binding and then stained with anti-TCR γδ (GL3; BioLegend), anti-Vγ5 (7-17; BD Pharmingen), anti-CD11c (N418; BioLegend), anti–I-A/I-E (M5/114.15.2; eBioscience), anti-CD100 (BMA12; eBioscience), and anti-CD69 (H1.2F3; eBioscience) in various combinations. Cells were subjected to flow cytometry on a BD FACSCalibur or BD FACSCanto II, and the resultant data were analyzed with CellQuest Pro or FlowJo software.

To analyze CHS responses, we followed the protocol of Nielsen et al. (7), with minor changes. Briefly, 25 μl of 0.15% DNFB in a 1:4 olive oil/acetone mixture was painted on the dorsal side of one ear, and olive oil/acetone mixture (vehicle) alone was painted on the other ear, for three consecutive days. On day 23, the ears were painted with the same materials, and 24 h later the ear thickness of anesthetized mice was measured on both sides using an engineer’s micrometer. Data were expressed as the ear swelling response above baseline (i.e., the thickness of the DNFB-painted ear − the thickness of the vehicle-painted ear).

KikGR mice were anesthetized, and hair was removed from the back skin (3 cm × 3 cm) using a hair clipper (THRIVE model 515R). Naked skin was exposed to violet light (420 nm) for 2 min using an LED hand lamp (LED-420P; OptoCode, Tokyo, Japan).

Total RNA was isolated from cells using ISOGEN and reverse transcribed using ReverScript III (both from Wako Pure Chemical Industries). PCR was performed in triplicate using GoTaq qPCR Master Mix containing SYBR Green on an Applied Biosystems 7500 Fast Real-Time PCR System. Experiments were independently repeated two or three times. Values for each gene were normalized to the corresponding expression levels of β-actin. The specific primers used were as follows: claudin-1: 5′-ACTCCTTGCTGAATCGAACAGT-3′, 5′-GGACACAAAGATTGCGATCAG-3′; claudin-3: 5′-CTCATCGTGGTGTCCATCC-3′, 5′-ATGGTGATCTTGGCCTTGG-3′; claudin-4: 5′-GGAATCTCCTTGGCAGTCCT-3′, 5′-CACCCACGATGATGCTGAT-3′; claudin-6: 5′-TGTGTGGTTCAGAGCACTGG-3′, 5′-AGCAGACAGGAATGAGCGTC-3′; claudin-7: 5′-ACGCCCATGAACGTTAAGTACGAG-3′, 5′-CTTTGCTTTCACTGCCTGGACA-3′; claudin-8: 5′-TCAGAATGCAGTGCAAGGTC-3′, 5′-AGCCGGTGATGAAGAAGATG-3′; ZO-1: 5′-AGCGATTCAGCAGCAACA-3′, 5′-GGCTCAGAGGACCGTGTAAT-3′; ZO-2: 5′-GGAAGATGTGCTCCATTCG-3′, 5′-TGCCGACTCCTCTCACTGTA-3′; cadherin 1: 5′-CTGCCACCAGATGATGATACC-3′, 5′-CGAACACCAACAGAGAGTCGTA-3′; occludin: 5′-ATGACATGTATGGCGGAGAG-3′, 5′-ATAGCCTCTGTCCCAAGCAA-3′; tricellulin: 5′-GTACTCGTGGGGCTGGATT-3′, 5′-GAACATCGCATTCATTGGTG-3′; JAML: 5′-GTGGTCCAGGACGAATTTCA-3′, 5′-ATGCCCTGCTGACCCTTAT-3′; β-actin: 5′-GGCTGTATTCCCCTCCATCG-3′, 5′-CCAGTTGGTAACAATGCCATGT-3′; HPRT: 5′-GTAATGATCAGTCAACGGGGGAC-3′, 5′-CCAGCAAGCTTGCAACCTTAACCA-3′; and IFN-γ: 5′-ATGAACGCTACACACTGCATCTT-3′, 5′-GAATCAGCAGCGACTCCTTTTCC-3′.

Vγ5⁺ cells were isolated from wild-type and Ocln^−/− DETC lines by FACS. Cells were then suspended in 100 μl of culture medium containing IL-2 at 1 × 10⁶ per milliliter and placed into the upper chamber of a 24-well Transwell plate with a 5-μm-pore filter (Corning Costar). Cells that migrated into the lower chamber, which contained 600 μl of culture medium with or without 500 ng/ml CXCL12 (BioLegend), were harvested after 4 h and counted. Migration of cells from the skin was analyzed by isolating ear skin from wild-type and Ocln^−/− mice. In six-well plates, dorsal and ventral sides of skin were floated dermal side down on 4 ml of medium containing 100 ng/ml CCL21 (R&D Systems). Dendritic cells and γδ T cells migrating out of the dermis and epidermis were collected after 48 h and counted and analyzed by FACS.

Statistical analyses were performed using the Student t test. A p value < 0.05 was considered statistically significant.

UVB is an environmental insult that affects epidermal homeostasis. To investigate the roles of DETCs in stress surveillance in the epidermis, we irradiated C57BL/6 mice with UVB and analyzed activation of DETCs. Epidermal sheets from ear skin were isolated, and DETCs were stained with FITC-conjugated anti-Vγ5 Abs. DETCs, which were distributed evenly at steady-state, became less frequent and lost their dendrites 48 h after irradiation (Fig. 1A). These cells extended their dendrites toward the surface of the skin at steady-state, whereas they became rounded in UVB-irradiated epidermis. These morphological changes were observed previously in wounded epidermis (22) and in epidermis overexpressing an NKG2D ligand (2). DETCs sorted from UVB-irradiated epidermis expressed high levels of IFN-γ, indicating that they were activated by an unidentified molecule expressed in this tissue following insult (Fig. 1B).

DETCs survey stress-induced molecules by extending their dendrites into tight junctions and receive signals through their TCRs (16). To identify molecules expressed in DETCs that could be involved in recognition or activation, we stimulated a DETC line established from epidermal cells with PMA plus ionomycin for 24 h and then analyzed the expression of tight junction molecules by RT-PCR. Claudin-1, claudin-3, claudin-4, and claudin-8 were expressed in epidermal cells but not in DETCs. Interestingly, occludin was expressed in DETCs only after activation, whereas ZO-1, ZO-2, cadherin-1, and tricellulin were constitutively expressed in resting and activated DETCs (Fig. 1C). To determine whether occludin expression was also induced in DETCs receiving signals through the TCR, we stimulated cells on plates coated with anti-CD3 Abs or with PMA plus ionomycin and then measured relative expression levels of tight junction molecules by real-time PCR. Occludin expression was also induced by TCR engagement (Fig. 1D), suggesting that DETCs express occludin in response to Ags. Expression of JAML, an adhesion molecule that interacts with Coxsackie and adenovirus receptor in tight junctions (17), was upregulated to some extent; however, induction of occludin was much higher. To determine whether occludin was also expressed in activated DETCs in stress-induced epidermis, we stained epidermal sheets isolated from mouse ear skin, exposed or not exposed to UVB irradiation, with anti-TCRγδ and anti-occludin Abs. Epidermis obtained from Ocln^−/− mice was also used as a negative control for occludin staining. Occludin expression was detected in some cells in UVB-irradiated epidermis; these occludin⁺ cells also expressed TCRγδ. Occludin was not detected in TCRγδ⁺ cells from epidermis not exposed to UVB irradiation. The occludin staining was specific, because no staining was observed in Ocln^−/− epidermis. These results suggested that occludin expression was induced in DETCs upon activation and that occludin expressed in DETCs may have distinct functions.

Ocln^−/− mice have little functional defect in tight junctions (19). When we analyzed immune cells in the thymus and lymph node, we observed no difference in immune cell populations between Ocln^−/− and wild-type mice (Supplemental Fig. 1). We prepared epidermal cell suspensions from wild-type and Ocln^−/− mice and analyzed DETCs (TCRγδ⁺ and Vγ5⁺) and LCs (I-A⁺) by FACS. There was no difference between wild-type and Ocln^−/− epidermis for either population (Fig. 2A). Next, we estimated the frequency of DETCs in epidermal sheets by counting cells in 100 × 100-μm areas. There was no significant difference between wild-type and Ocln^−/− epidermis with regard to the average number of DETCs (Fig. 2B). To determine whether Ocln^−/− DETCs were activated in response to Ags, we stimulated DETCs from wild-type and Ocln^−/− mice on plates coated with anti-TCR γδ Abs. The activation markers CD25 and CD100 were upregulated equally in wild-type and Ocln^−/− mice (Fig. 2C). Epidermal sheets were prepared from UVB-irradiated mice, and DETCs were stained with anti-Vγ5 Abs. Morphological changes were evaluated by determining the proportions of cells with more than two dendrites, one or two dendrites, or no dendrites. After irradiation, DETCs in Ocln^−/− mice had more dendrites than did those in wild-type mice (Fig. 2D). These results suggested that occludin expressed in DETCs after activation is involved in cytoskeletal rearrangement.

To determine whether occludin expressed in DETCs, but not in keratinocytes, was responsible for these effects, we induced morphological changes in wild-type and Ocln^−/− DETCs in vitro. On wild-type epidermal stromal cells, DETCs from KikGR and KikGR/Ocln^−/− DETC lines were cultured in the presence of IL-2. DETCs expressing GFP became adherent and visible under fluorescence microscopy. The DETCs were then incubated with biotin–anti-CD3 Abs, followed by allophycocyanin-streptavidin, and their morphologies were observed after 4 h. DETCs from wild-type and Ocln^−/− mice remained attached to the stromal cells in the culture, whereas DETCs stimulated with anti-CD3 became rounded (Fig. 2E, indicated by arrows). Numbers of cells with rounded versus stretched shapes were counted in 10 areas (Supplemental Table I), and average percentages were calculated. More than 60% of wild-type DETCs became rounded, versus ∼25% of Ocln^−/− DETCs, upon stimulation (Fig. 2F). These results indicated that occludin expressed on DETCs plays a role in the morphological changes that take place in these cells.

Recent analysis by in vivo microscopy demonstrated that occludin expressed in γδ intraepithelial lymphocytes (IELs) contributes to their migration (23). Our detection of occludin expression in activated DETCs prompted us to investigate whether DETCs are also able to migrate upon activation. To this end, we analyzed migration of DETCs from the epidermis to draining lymph nodes by tracking cells that express KikGR (21). To convert KikGR from green to red fluorescence, we clipped the back skin of KikGR mice and exposed it to violet light for 2 min. Epidermal stress was induced by UVB irradiation or tape stripping, both of which caused activation of DETCs (tape stripping data not shown). Lymph node cells were collected from inguinal lymph nodes 2–5 d after treatment (Fig. 3A). Almost all DETCs in the epidermis were photoconverted to red (KikGR-Red) by violet light exposure (data not shown). In skin-draining inguinal lymph nodes, photoconverted cells were detected in 2 d. Of the photoconverted KikGR-Red cells present in inguinal lymph nodes, ∼30% were TCRγδ⁺ in mice treated with UVB irradiation or tape stripping. In contrast, mice with no epidermal stress other than violet light exposure had fewer KikGR-Red⁺ γδ T cells in draining lymph nodes in the skin (Fig. 3B). These results suggested that γδ T cells are stimulated to migrate in response to stress-induced molecules expressed following UVB irradiation or tape stripping. Cells migrating from the skin to inguinal lymph nodes were counted on days 2–5 after the photoconversion. The number of KikGR-Red cells in inguinal lymph nodes peaked on day 4 after photoconversion. More KikGR-Red cells were found in mice subjected to tape stripping, suggesting that this treatment more efficiently promoted cell migration from the skin to the draining lymph nodes (Fig. 3C). The number of γδ T cells migrating from the skin was also calculated by FACS analysis. The highest numbers of KikGR-Red⁺ γδ T cells in draining lymph nodes were found on day 4 in mice subjected to tape stripping (Fig. 3D). These results demonstrated that γδ T cells traffic from the skin to the draining lymph nodes in response to epidermal stress. We then stained with anti-Vγ5 Abs to determine whether DETCs were present among the migrated γδ T cells in the draining lymph nodes. Although cell numbers were small, we could detect TCRγδ⁺Vγ5⁺ cells among KikGR-Red cells in the inguinal lymph nodes on days 4 after tape stripping (Fig. 3E). These TCRγδ⁺Vγ5⁺ cells were only detected on days 4 and 5 (Fig. 3F).

Next, using Ocln^−/− KikGR mice, we investigated whether occludin expressed on activated DETCs was involved in their migration. Epidermal cells in wild-type and Ocln^−/− mice were photoconverted, and the cells that migrated after tape stripping were analyzed on day 4. The frequencies of photoconverted cells in draining lymph nodes were similar (Fig. 4A). The number of Vγ5⁺ KikGR-Red cells in inguinal lymph nodes varied depending on the mice used; however, we could always detect Vγ5⁺ cells in wild-type mice on day 4. Few KikGR-Red cells expressing TCRγδ and Vγ5 were found in inguinal draining lymph nodes in Ocln^−/− mice (Fig. 4A, 4B). These results indicated that the major defect in Ocln^−/− mice was impaired migration of DETCs.

Because occludin is expressed in activated DETCs and in other surrounding cells, such as keratinocytes, we then analyzed whether the defect in DETC migration in Ocln^−/− mice was due to the lack of occludin expression in DETCs but not in other cells. For this purpose, we tested wild-type and Ocln^−/− DETCs for their ability to migrate toward CXCL12. CXCL12 promoted the migration of wild-type and Ocln^−/− DETCs; however, the number of migrated cells was significantly smaller in Ocln^−/− DETCs than in the wild-type (Fig. 4C). These results indicated that occludin expressed in DETCs is involved in migration.

The impaired traffic of γδ T cells in Ocln^−/− mice may affect immune responses in the skin. Because the importance of DETCs in CHS pathogenesis has been demonstrated (7), CHS was an appropriate system for investigating whether occludin expressed in activated DETCs plays a role in immune responses. First, we investigated whether occludin was expressed in DETCs in DNFB-treated epidermis. For this purpose, epidermal sheets were isolated 24 h after DNFB administration and stained with anti-TCRγδ and anti-occludin Abs. Epidermal sheets isolated from Ocln^−/− mice were also stained as a negative control. We did not observe occludin staining in Ocln^−/− epidermis. DETCs in DNFB-treated epidermis were positive for occludin expression, whereas those in intact epidermis were not (Fig. 5A). As observed in UVB-irradiated epidermis, DETCs in Ocln^−/− mice exhibited very little morphological change 18 h after DNFB treatment, whereas those in wild-type mice had rounded shapes (Fig. 5B, 5C). Next, we investigated the involvement of occludin in γδ T cells during CHS by inducing CHS in Ocln^−/− and wild-type mice. After DNFB challenge, ear thickness was much lower in Ocln^−/− mice than in wild-type mice. The reduction in CHS in TCRδ^−/− mice was more significant, indicating that the response of Ocln^−/− γδ T cells to CHS was partially impaired (Fig. 5D). To exclude the possibility that cytokine expression by αβ or γδ T cells was impaired in Ocln^−/− mice, we stimulated cells obtained from skin-draining lymph nodes from wild-type and Ocln^−/− mice with anti-CD3 plus anti-CD28 Abs. Both types of T cells expressed IFN-γ and IL-17A at similar levels (Supplemental Fig. 2). These results suggested that Ocln^−/− DETCs, which exhibit impairments in cytoskeletal rearrangement and motility after activation, failed to migrate and activate other cells involved in CHS.

Because the dendritic cells that capture Ags and migrate into the draining lymph nodes also play a role in CHS, we analyzed chemotaxis of dendritic cells in vitro. For this purpose, ear skin from Ocln^−/− and wild-type mice was separated into dorsal and ventral sides and floated dermal side down on medium containing CCL21. After 48 h, cells migrating out of the skin were collected and analyzed by FACS (Fig. 6A). Dendritic cells, as well as γδ T cells, migrated out of skin isolated from wild-type mice. Similar numbers of dendritic cells were found in culture medium of Ocln^−/− samples, indicating that chemotaxis of dendritic cells was not affected by occludin deficiency (Fig. 6B). As we expected, the numbers of γδ T cells migrating out of the skin were very low. We also found that significantly fewer γδ T cells migrated out of Ocln^−/− skin than wild-type skin. To determine whether dermal γδ T cells express occludin constitutively, we analyzed the expression of occludin in DETCs and dermal γδ T cells. DETCs (TCR γδ⁺Vγ5⁺) and dermal γδ T cells (TCR γδ⁺Vγ5⁻) were sorted from the ear skin of wild-type mice, and relative expression levels of occludin were analyzed by quantitative real-time PCR (Supplemental Fig. 3). Occludin expression in dermal γδ T cells was three times higher than that in DETCs in steady-state, suggesting that occludin deficiency also affected the motility of dermal γδ T cells. These results suggested that the reduction in CHS responses in Ocln^−/− mice was not due to a defect in dendritic cells but, instead, was a result of the impaired motility of γδ T cells.

DETCs survey epidermal stress–induced molecules and play a pivotal role in maintaining epidermal homeostasis. In this study, we found that DETCs migrated into draining lymph nodes upon activation and promoted immune responses in the skin, indicating that DETCs function in the epidermis, as well as in lymph nodes. Morphological changes occurring upon DETC activation have been documented in wounding (1), allergen administration (7), and ligation of NKG2D (2); however, further roles of rounded DETCs are not well understood. At steady-state, dermal γδ T cells are mobile, whereas epidermal DETCs remain immobile (24, 25). However, once DETCs are activated in dysregulated epidermis, occludin expression causes them to undergo morphological changes and become motile, ultimately migrating into draining lymph nodes.

Occludin is critical for migration of γδ IELs (23), which migrate actively within the intraepithelial compartment and lamina propria. This migration is mediated via a homotypic interaction between occludin molecules expressed on the surfaces of IELs and epithelial cells. It is possible that occludin is involved in the motility of all γδ T cells. Chemotaxis toward CXCL12 was impaired in Ocln^−/− DETCs, indicating that occludin expressed in DETCs, but not in keratinocytes, is required for DETC motility. In DETCs, occludin expression was induced upon activation, whereas dermal γδ T cells, as well as γδ IELs, express occludin constitutively. DETCs express other tight junction molecules, such as ZO-1, as do γδ IELs (23); therefore, downstream signals of tight junction molecules could confer unique features to γδ T cells. Fewer dermal γδ T cells in Ocln^−/− mice compared with wild-type mice migrated out of the skin in vitro. However, in vivo analysis using KikGR mice revealed that the defect in migration from the skin to the draining lymph nodes was observed only in DETCs in Ocln^−/− mice, suggesting that occludin expression in dermal γδ T cells may not be involved in their migration into lymph nodes.

Morphological changes in DETCs after activation were impaired in Ocln^−/− mice, indicating that occludin-induced cytoskeletal rearrangement plays a pivotal role in determining cell shape. DETCs form immunological synapses at keratinocyte tight junctions; however, reorganization of synapses is induced once these cells are activated in response to keratinocyte stress (16). The microtubule-organizing center localizes toward immunological synapses, and the microtubule network is disorganized in occludin-deficient epithelial cells (26), suggesting that occludin expressed in DETCs plays a role in microtubule-organizing center reorientation and localizes to leading edges of migrating cells. Signals from CD100 expressed in activated DETCs play a pivotal role in these morphological changes. In particular, CD100 signals through an interaction with plexin B2 expressed in keratinocytes, leading to changes in morphology (18). However, occludin-deficient DETCs expressed CD100 at levels similar to those observed in wild-type DETCs after activation, suggesting that occludin expression may be downstream of CD100 signaling and that CD100-deficient DETCs might not induce occludin expression.

Dysregulation of the epidermis induced by tape stripping or UVB irradiation increased the migration of Vγ5⁺ and Vγ5⁻ γδ T cells, suggesting that these stimuli activated DETCs, as well as dermal γδ T cells. Dermal γδT17 cells migrate into draining lymph nodes in imiquimod-induced psoriasis-like dermatitis (27). γδT17 cells play an important role in IL-17–mediated inflammation; however, the roles of DETCs that normally produce IFN-γ in draining lymph nodes remain poorly understood. Nielsen et al. (7) demonstrated that DETCs are involved in CHS; DETCs are activated by IL-1β produced in the epidermis and express IL-17. Those investigators showed that CHS responses in TCRδ^−/− mice were almost equivalent to those in IL-17^−/− mice, suggesting that γδ T cells expressing IL-17 play a central role in these processes. Occludin deficiency did not alter IL-17 expression in γδ T cells. CHS responses were reduced in Ocln^−/− mice but not as much as those in TCRδ^−/− mice, suggesting that some γδ T cells that migrated into draining lymph in an occludin-independent manner also participated in CHS responses. Alternatively, a subset of responses may take place in the skin, because DETCs have been shown to express IL-17 in the epidermis, which plays an important role in CHS responses (7).

Our experiments in which we tracked cells in the skin by photoconversion clearly demonstrated that DETCs could migrate into draining lymph nodes, and these cells were observed on days 4 and 5 after photoconversion. Although we could detect very few photoconverted cells in draining lymph nodes, this system was sensitive enough to detect DETCs derived from the epidermis. Cell traffic from the skin to the draining lymph nodes, including γδ T cells and dendritic cells, was increased by epidermal stress, indicating that γδ T cells are innate immune cells that contribute to adaptive immunity. Migration of dendritic cells is regulated by CCR7, which is upregulated by activation, but the mechanisms by which γδ T cells migrate from the skin to draining lymph nodes are not well understood (28). The reduction in the number of γδ T cells in draining lymph nodes, and especially the lack of DETCs in Ocln^−/− mice, demonstrated that DETCs use occludin for cytoskeletal changes and movement. DETCs became motile only after occludin expression following activation.

CHS is a T cell–mediated inflammatory response induced by skin allergens, and IL-1β–mediated DETC activation plays an important role in CHS pathogenesis (7). Recent work showed that activation of DETCs can be induced by NKG2D ligands, which are upregulated on keratinocytes by allergens (29). Occludin expression was observed in DNFB-treated epidermis, potentially as the result of activation of DETCs in response to NKG2D ligand expressed on keratinocytes. In addition to hapten-specific T cells, DETCs are required to transfer CHS responses, and draining lymph node cells on days 4 and 5 following epicutaneous hapten administration can transfer full CHS (30). This observation is in close agreement with our analysis showing that DETCs were detected on days 4 and 5 after epidermal stress, and it seems likely that those DETCs contributed to CHS by helping hapten-specific T cells. As discussed above, γδ T cells, but not dendritic cells, exhibited impaired motility in occludin-deficient mice. These results strongly suggest that occludin expressed upon activation plays a pivotal role in DETC function in draining lymph nodes.

We thank M. Furuse for providing Ocln^−/− mice and Y. Hara for cell sorting.

The authors have no financial conflicts of interest.