Skin-Resident T Cells Drive Dermal Dendritic Cell Migration in Response to Tissue Self-Antigen

This article is featured in In This Issue, p.3025

The maintenance of immune homeostasis is dependent on the ability of tissue-resident immune cells to recognize and adequately respond to both foreign and self-antigens. The skin is continuously exposed to exogenous and endogenous Ags and harbors a relatively high number of tissue-resident immune cells, largely composed of dendritic cells (DCs) and T cells (1, 2). Tissue-residing DCs predominantly operate at the interface between innate and adaptive immunity, either promoting antipathogen immunity or facilitating tolerance to tissue Ags, both of which require their initial migration to tissue draining lymph nodes (DLNs).

Migration of DCs from tissues is essential for presentation of tissue-derived Ags to T cells in secondary lymphoid organs. The mechanisms that govern DC migration under inflammatory contexts has been extensively studied (reviewed in Ref. 3). Epidermal resident Langerhans cells (LC), characterized by expression of the C type Lectin CD207 (Langerin), are anchored to epithelial keratinocytes through E-cadherin–containing tight junctions, rendering them nonmotile. Upon activation, LCs detach from keratinocytes via downregulation of E-cadherin and begin to upregulate CCR7 (4–6). Although this pathway does not play a role in LC translocation from epidermis to dermis, CCR7 is required for the mobilization of all skin DC subsets toward terminal lymphatics (7). Following epicutaneous immunization, activated dermal DCs (DDC) arrive first to skin DLN (SDLNs), and localize to distinct areas relative to slower migrating LCs (8). Recent evidence suggests that migratory LCs promote the maintenance of peripheral tolerance through deletion of self-reactive T cells, induction of regulatory T cell (Tregs), and promotion of IL-10 secretion (9–11). In contrast, dermal Langerin-expressing CD103⁺ DCs are specialized for cross-presentation of viral Ags, as well as uptake and delivery of Staphylococcus epidermidis-derived Ags from skin to SDLNs for efficient induction of Ag-specific IL-17–producing CD8⁺ T cells (12–15). Although many studies have addressed the functional specialization of migratory DC populations in response to viral, bacterial, or allergic immune responses, the role of DC subsets in the early response to tissue self-antigen and the signals that drive DC emigration from tissues during the initiation of an autoimmune response are less well defined.

To begin to address this, we used a previously established experimental model of skin autoimmunity. In this system, a model self-antigen (OVA) can be inducibly expressed in keratinocytes, resulting in a productive Ag-specific T cell–mediated immune response (16–18). The long-term kinetics of T cell recruitment, activation, and regulation have been well characterized in this system, whereas the contribution of the major skin DC populations has not been assessed. This model of inducible and restricted OVA expression in epithelial cells permits detailed analysis of the accumulation and migratory behavior of different DC populations in the early events immediately following self-antigen expression in skin. In this study, we report the observation of a robust and mass emigration of CD103⁺ DDCs from skin to SDLNs as early as 48 h after initiation of self-antigen expression. Activation of skin-resident, but not SDLN-resident, Ag-specific T cells coincided precisely with the rapid departure of CD103⁺ DDCs from skin. We further demonstrate that tissue-resident, but not SDLN-resident, T cells are necessary for DDC exit from the tissue early upon self-antigen expression. These results elucidate a fundamental property of tissue-resident T cells and suggest that they play a role in shaping the migratory behavior of tissue DCs during the initiation phase of an autoimmune response.

TRE-TGO mice were crossed with K5-rtTA mice to generate mice with inducible epidermal expression of OVA. The TGO construct encodes a fusion protein linking the transferrin receptor transmembrane domain, GFP, and aa 230–359 of chicken OVA. To create K5/TGO/DO11 mice, K5-rTA/TRE-TGO mice were crossed onto the DO11.10 TCR-transgenic background, as described previously (18). To induce expression of the TGO transgene, K5/TGO/DO11 mice were starved from all chow for 12 h and then maintained on 1 g/kg doxycycline (Dox) chow (i.e., day 0) until sacrifice on day 2. Batf3^−/− mice were purchased from the Jackson Laboratory and were bred in-house. A 16-point clinical scoring scale was used to quantify skin disease. Scaling, alopecia, loss of activity, and erythema were each given a score from 0 to 4. Individual scores were summed and mean scores per group are displayed. Hosts for adoptive transfer experiments were created by crossing K5/TGO mice with TCR-α–deficient mice (TCRα^−/−) on the BALB/c background. For adoptive transfer of DO11 T cells, DO11.10 TCR-transgenic mice were crossed onto the Rag2^−/−/CD90.1⁺ background. All mice were bred and maintained in a specific pathogen-free facility in accordance with the guidelines of the Laboratory Animal Resource Center of the University of California, San Francisco. All mice used in experiments were socially housed under a 12 h light/dark cycle.

Preparation of single-cell suspensions from SDLNs or full thickness skin for flow cytometry was performed as previously described (19). In brief, isolation of cells from axillary, brachial, and inguinal LNs (referred to as SDLNs) for flow cytometry was performed by mashing tissue over 100 μm sterile filters. For all experiments, an equal area of mouse dorsal skin was harvested (∼2 × 3 cm). All experiments were designed such that negative and positive control groups were always included in parallel and in most cases data were not pooled between independent mouse harvests. For isolation of dorsal skin cells, mouse dorsal skin was harvested and lightly defatted. It was then minced finely with scissors and resuspended in 3 ml of digestion mix (composed of 2 mg/ml collagenase XI, 0.5 mg/ml hyaluronidase, and 0.1 mg/ml DNase in RPMI 1640 with 1% HEPES, 1% penicillin-streptomycin, and 10% FCS), followed by incubation in a shaking incubator at 37°C at 250 rpm for 1 h. Single-cell suspensions were then pelleted and resuspended in PBS for cell counting and staining. The following Abs were used for flow cytometry staining: Foxp3 eFluor 450, CD25 APC–eFluor 780, CD45 Alexa Fluor 700, MHC class II (I-A/I-E) eFluor 450, Langerin (L31) PE, CD103 APC, KJ (DO11 TCR) FITC, and programmed death-1 (PD-1) PE-Cy7 (all from eBioscience); Brilliant Violet 605 CD8a, Brilliant Violet 650 CD4, Brilliant Violet 711 CD11b, Brilliant Violet 650 CD11c, and CD69 APC (all from BioLegend), CD3 PerCP, purified rat anti-mouse CD16/CD32, B220 PercP, CD11b FITC, CD90.1 PerCP, CD90.2 FITC, and MHC class II (I-A/I-E) BV421 (all from BD Pharmingen); Ghost Dye Violet 510 Live/Dead Stain (Tonbo Biosciences).

FTY720 (Selleck Chemicals) was dissolved in normal saline and administered to mice via i.p. injection at a dose of 10 mg/kg on days −2, −1, and 1. Control mice (age-matched littermates) were treated with equal volumes of normal saline according to the same schedule. All mice were harvested on day 2 for analysis of skin and SDLNs by flow cytometry.

InVivoMAb anti-mouse MHC class II (I-A/I-E) blocking Ab (clone M5/114) was purchased from Bio X Cell (West Lebanon, NH). Next, 250 μg was administered i.p on days −2, −1, and 1. Mice were administered Dox chow on day 0 and tissues harvested on day 2.

Single-cell suspensions from LNs of DO11/Rag2^−/−/CD90.1⁺ (i.e., DO11 T cells) mice were prepared and 2 × 10⁶ LN cells adoptively transferred intravenously into gender-matched K5/TGO/TCRα^−/− recipient mice. Mice were started on Dox chow the same day as adoptive transfer.

Statistical analyses were performed with Prism software package version 6.0 (GraphPad). The p values were calculated using two-tailed unpaired or paired Student t test. Sample size for animal experiments was determined based upon pilot experiments. Mice cohort size was designed to be sufficient to enable accurate determination of statistical significance. No animals were excluded from the statistical analysis, unless due to technical errors. Mice were randomly assigned to treatment or control groups, although there were inclusion criteria based on gender and age. Appropriate statistical analyses were applied, assuming a normal sample distribution. All in vivo experiments were conducted with at least two independent cohorts (as indicated in the figure legends).

The experimental model of inducible self-antigen expression we have developed comprises three transgenes (18). The first expresses membrane-bound OVA under the control of a tetracycline response element, termed TGO, and the second carries a tetracycline transactivator driven by the keratin-5 (K5) promoter. These mice were further crossed with the DO11.10 TCR-transgenic strain. In the resultant K5/TGO/DO11 mice, expression of Ag in K5-expressing basal keratinocytes is dependent on administration of the tetracycline analog, Dox (Fig. 1A). We have previously defined functional OVA expression in the thymus of K5/TGO/DO11 triple transgenic mice. Deletion of CD4⁺ DO11 T cells and increased production of Foxp3⁺CD4⁺DO11 Tregs in K5/TGO/DO11 mice is a sensitive indicator of thymic OVA expression. K5/TGO/DO11 triple transgenic mice as well as TGO/DO11 double transgenic mice have modest deletion of CD4⁺ DO11 cells and a pronounced increase in DO11 Tregs in the thymus, SDLNs, and skin. Expression of OVA in this system largely mimics the pattern of tissue-restricted self-antigen expression due to its constitutive expression in the thymus and tightly regulated induction in skin (16–18).

We have previously demonstrated that induction of Ag expression in the epidermis results in a pronounced T cell–dependent autoimmune dermatitis that peaks at 10 d (18). To determine how skin DCs respond to Ag expression in this model, we began by first defining the kinetics of total DC accumulation in skin and SDLNs by flow cytometry during the early window of time following OVA induction. In skin, DCs were identified as live, CD45^Pos CD3^neg CD8^neg B220^neg Class-II^Pos CD11c^Pos cells, whereas migratory DCs present in SDLNS were defined as class-II^High CD11c^High cells, markers that have been previously shown to reliably distinguish migratory DCs from lymphoid-resident DCs (20). Full gating strategies are shown in Supplemental Fig. 1. Within 2 d of initiation of OVA expression, we observed a significant reduction in the total number of DCs present in skin with a concomitant accumulation of migratory DCs in SDLNs (Fig. 1B, 1C), suggesting that these cells rapidly exit the skin upon initiation of self-antigen expression in the epidermis. Migratory DC numbers gradually decreased in SDLNs with a concomitant reaccumulation in skin, returning to baseline levels in the tissue by day 10 (Fig. 1B, 1C). In skin, both CD103⁺ DDCs and epithelial-resident LCs are found closely localized to keratinocytes residing in the basal layer of murine epidermis (21, 22). We examined the presence of these two subsets early after induction of Ag. Epidermal LCs and migratory LCs in SDLNs were identified in the DC gate as Langerin^Pos CD11b^High CD103^neg cells, whereas CD103⁺ DDCs were defined as Langerin^Pos CD11b^Low CD103^Pos cells (Supplemental Fig. 1). Strikingly, we observed a mass emigration of CD103⁺ DDCs from skin as early as day 2, with a ∼4-fold reduction in the skin at this time point (Fig. 1D). The corresponding migratory CD103⁺ population accumulated in SDLNs at the same time point where their numbers initially peaked and returned to baseline thereafter (Fig. 1D, 1E). In contrast, the absolute number of LCs present in both compartments was unchanged during this early time period (Fig. 1E). LCs exited the tissue with slightly delayed kinetics, as shown by an appreciable reduction in their absolute numbers in the tissue at day 5 post–Ag induction and a significant increase in their migratory counterparts in SDLNs (Fig. 1E). Although the absolute number of Langerin^neg DCs was also reduced in the skin at day 2 post–Ag induction, this represented a less marked emigration relative to CD103⁺ DDCs (Supplemental Fig. 2A, 2B). Similarly, the absolute number of SDLN-resident DCs, characterized as CD11c^High MHC class II^Intermediate, remained unchanged during the first 48 h post–Ag induction. Although their numbers declined modestly thereafter, this failed to reach statistical significance (Supplemental Fig. 2C). These results demonstrate that CD103⁺ DDCs preferentially and rapidly exit the tissue early upon self-antigen expression in skin.

Given the specialized ability of migratory CD103⁺ DDCs in eliciting functional responses to skin Ags (12–15), we attempted to elucidate the cellular signals that drive the mass emigration of this population. Importantly, this phenomenon occurs at a time point where we have previously been unable to detect any signs of clinical or cellular inflammation in skin (16–18). However, during the autoimmune response to OVA in K5/TGO/D011 mice, by immunofluorescent tissue staining of skin (17) we have observed that effector DO11 T cells localize close to CD11c⁺ cells diffusely throughout the dermis. This led us to hypothesize that T cells may be involved in facilitating the early migration of CD103⁺ DDCs upon Ag expression. We have previously characterized the T cell activation status in these mice over a long time course (18), but not during the first 48 h of OVA expression, when CD103⁺ DDCs exit skin. Ag-specific DO11 T cells were gated as Singlet^PosLive^PosCD3^PosCD4^PosKJ^Pos and then Foxp3^neg or Foxp3^Pos for effector T cells (Teffs) and Tregs, respectively. Comprehensive phenotypic analysis of DO11 T cells in skin on day 0 (i.e., no Dox treatment) and 2 d after initiation of Ag induction, revealed a 2-fold expansion of Foxp3-negative Teffs with no change in the number of Foxp3-expressing Tregs (Fig. 2A, 2B). In addition, the activation status of skin DO11 cells, as evidenced by CD25, PD-1, and CD69 expression, revealed a significant upregulation of all these markers in Teffs but not Tregs (Fig. 2A, 2B). In contrast, analysis of Ag-specific DO11 cells in SDLNs between day 0 and day 2 showed no differences in the absolute numbers of Teffs and Tregs. In addition, there was no evidence of activation of either of these populations (Fig. 2C, 2D). These results suggest that early after expression of self-antigen in skin, Ag-specific Teffs are preferentially activated in the tissue with no evidence of activation of these cells in tissue DLNs.

Given the accumulation and robust activation of skin-resident Ag-specific Teffs and the concurrent exit of CD103⁺ DDCs early after Ag induction, we sought to determine if T cells are required for DC emigration from skin. To this end, we used K5/TGO mice crossed onto a TCR-α–deficient (TCRα^−/−) background to generate K5/TGO/TCRα^−/− mice. In this model, administration of Dox will induce Ag expression in the absence of functional TCR-α–expressing T cells, allowing characterization of CD103⁺ DDC migration from skin in the absence of these cells. In K5/TGO/TCRα^−/− mice, absolute numbers of CD103⁺ DDCs in both skin and SDLNs were unchanged between day 0 and day 2 post–OVA induction (Fig. 3). These results suggest that T cells are required for the migration of CD103⁺ DDCs early during the response to tissue self-antigen.

The preferential activation of T cells in the tissue relative to SDLNs and the failure of DCs to migrate in T cell–deficient mice, suggested that skin-resident Ag-specific T cells might be the dominant drivers of CD103⁺ DDCs emigration from skin early after expression of tissue self-antigen. To test the hypothesis that the accumulation of activated Ag-specific T cells in skin was not a result of a migratory T cell wave from secondary lymphoid organs, K5/TGO/DO11 mice were treated with the sphingosine 1–phosphate receptor antagonist, FTY720, from day -2 to day 1 post–Ag induction. FTY720 potently inhibits lymphocyte egress from lymphoid organs (23). In these experiments, CD103⁺ DDCs in FTY720-treated mice rapidly exited the tissue and accumulated in the SDLNs with similar kinetics to that observed in non–TY720-treated animals (Fig. 4A). Furthermore, the absolute number of CD103⁺ DDCs in skin and SDLNs of FTY720-treated mice were equivalent to control mice on day 2 post–Ag induction (Fig. 4B), demonstrating that these cells retained the ability to emigrate from skin in the presence of FTY720. These results suggest that T cell egress from secondary lymphoid organs is not required for DDC tissue emigration, implying that tissue-resident T cells may be the major drivers of this phenomenon.

Given that blockade of T cell egress from SDLNs does not impact the migration of CD103⁺ DDCs from skin, we next sought to determine if the exclusive presence of T cells in the SDLNs (with relative absence in skin) was sufficient to promote DDC emigration. To this end, we used an adoptive transfer approach. Ag-specific T cells were isolated from CD90.1/DO11/Rag^−/− mice (referred to as DO11 cells) and transferred i.v. into K5/TGO/ TCRα^−/− hosts. Upon transfer, these cells stably engraft in secondary lymphoid organs well before they can be detected in skin (Fig. 5A). Two days after adoptive transfer into K5/TGO/ TCRα^−/− hosts, DO11 T cells (defined as CD90.1^Pos CD4^Pos KJ^Pos) were readily detectable in SDLNs but undetectable in skin (Fig. 5A, 5B). Previous studies with this model have shown that DO11 cells begin to seed the skin no earlier than 5 d after OVA induction (16, 17), allowing a window for residence of Ag-specific DO11s exclusively in secondary lymphoid organs (and not in peripheral tissues). Activation profiling of adoptively transferred DO11 cells in the SDLNs on day 2 post–Ag induction showed no differences in the expression of PD-1, CD44 or CD69 between the Dox-treated group and control mice that did not receive Dox (Fig. 5C). These results are consistent with the lack of SDLN T cell activation observed in the K5/TGO/DO11 model on day 2 post–Ag expression (Fig. 2C, 2D). Notably, analysis of CD103⁺ DDCs revealed no changes in the absolute cell numbers of this population in both skin and SDLNs during this window of time (Fig. 5D, 5E), indicating a relative inability of these cells to emigrate from skin in this setting. Taken together, these findings suggest that T cell egress from SDLNs or the simple presence of Ag-specific T cells exclusively in secondary lymphoid organs is insufficient to drive migration of CD103⁺ DDCs from skin to SDLNs early after induction of tissue self-antigen expression.

The requirement of skin-resident T cells to induce CD103⁺ DDC efflux suggested that prototypical cell surface interactions between the TCR and DC-expressed MHC class II may be an important driving factor. Also, given the rapid migration of CD103⁺ DDCs, we wanted to determine if innate-derived sensing or recognition through the Ag receptor plays a dominant role in the response to cutaneous self-antigen expression in this defined time period. We therefore used an in vivo anti–MHC class II Ab blocking approach to assess the migratory ability of CD103⁺ DDCs during the first 2 d following induction of skin Ag expression. Under these conditions, the migratory ability of CD103⁺ DDCs was partially abrogated in skin, and was reflected in their reduced accumulation in SDLNs relative to control mice not receiving anti–MHC class II Ab (Fig. 6A). In contrast, the numbers of LCs and Langerin^neg DCs was unchanged in skin and SDLNs in the presence or absence of MHC class II blockade (Fig. 6B, 6C). Partial abrogation of CD103⁺ DDC migration, but not other skin DC subsets, suggests that Ag-specific recognition through the MHC class II pathway plays a role in the induction of CD103⁺ DDC mobilization during the first 48 h post–tissue self-antigen expression. Because blockade of the MHC class II pathway only partially abrogated DDC migration, other signals may be required to synergize with Ag recognition in driving the migration of these cells from skin.

To determine if Batf3-dependant CD103⁺ DDCs play a functional role in the response to cutaneous self-antigen, K5/TGO/DO11 mice were crossed onto a Batf3^−/− background. These mice lack CD103⁺ DCs (24). This allowed us to generate K5/TGO/DO11/Batf3^−/− [(Batf3 knockout (KO)] and control K5/TGO/DO11/Batf3^+/+ [Batf3 wild-type (WT)] mice. Induction of cutaneous OVA expression in Batf3 WT mice resulted in a pronounced inflammatory dermatitis, peaking by day 10 (Fig. 7). Disease was characterized and scored by the development of erythema, scaling, and alopecia (Fig. 7A). In Batf3 KO mice, disease induction in the first 5 d was similar to Batf3 WT mice, whereas clinical severity was significantly attenuated by day 10 post–Ag expression (Fig. 7A).

Attenuated severity of disease in Batf3 KO mice was associated with a reduction in the absolute number of PD-1–, CD25-, CD69-, and Ki67-expressing Ag-specific Teffs in skin (Fig. 7C). In addition, Teff production of the cytokine IFN-γ, but not IL-2, IL-4, or IL-17, was significantly reduced in Batf3 KO mice relative to control Batf3 WT mice (Fig. 7D). Taken together, these results indicate that CD103⁺ DDCs play a major role in Ag-specific T cell activation and the development of fulminant skin inflammation in our model.

To permit efficient induction of an acquired immune response, nonlymphoid tissue-resident DCs not only need to undergo a sequence of functional maturation, but must relocate from peripheral tissues to tissue DLNs. By virtue of their migratory potential and specialized capacity to acquire and process tissue-derived Ags, DCs efficiently direct the primary Ag-specific T cell response. Studies on the pathways responsible for DC mobilization from peripheral tissues have focused on experimental models of inflammation but, to date, none have elucidated the cellular signals that drive DC migration upon expression of a tissue self-antigen. To our knowledge, we demonstrate for the first time a requirement of activated Ag-specific CD4⁺ T cells to be present in the tissue to drive the emigration of CD103⁺ DDCs during the initiation of skin autoimmunity. Interestingly, this occurs at an early time point post–epidermal Ag induction where there are no signs of cellular inflammation in the tissue. This suggests that the classical signals promoting the morphological and phenotypic changes associated with DC migration, such as activation induced via the TLR pathway or cytokine-induced MHC class II upregulation, may not be the primary drivers of CD103⁺ DDC emigration from skin. In the context of OVA infection of the female reproductive tract (FRT), reactivated Ag-specific CD8⁺ T resident-memory cells are required for triggering local DC maturation, as evidenced by induced expression of costimulatory molecules CD80, CD86, CD40, as well as the lymphoid organ homing receptor CCR7 (25). This effect was largely attributed to CD8⁺ T resident-memory–derived TNF-α in the FRT. Although this study did not determine the contribution of tissue-resident CD8⁺ cells during an autoimmune response nor directly assess the migratory ability of activated DCs in the FRT, it is highly suggestive that the presence of cognate-antigen specific T cells at tissue sites can serve as potent modulators of the immune response to tissue Ag. Our study also raises the notion that the early activation of tissue-resident CD4⁺ dermal T cells may represent local sensing of previously encountered Ags other than self-antigens (i.e., commensal microbes or foreign pathogens) that precipitate innate-like signals to draw circulating Teffs into the cutaneous microenvironment.

More recently, it has been demonstrated that DDCs can form clusters with Teffs in perivascular areas of the dermis upon epicutaneous Ag challenge (26). Accordingly, we demonstrate that tissue-resident Ag-specific T cells are necessary for DC flux shortly after tissue self-antigen expression, and do so in an MHC class II–dependent manner. Whether dermal T cell–DC clusters are an essential structure for eliciting DC migration and autoimmunity in skin is yet to be determined. This also raises the question of whether T cells, or a subset thereof, residing in other tissue sites also have the capacity to regulate migratory DC mobilization in the face of self-antigen induction.

To control the development of autoimmunity, multiple mechanisms of peripheral tolerance are present, including both activating and inhibitory T cell costimulatory pathways. By definition, costimulatory ligands expressed on APCs (including DCs) engage their cognate receptors on T cells, the result of which dictates T cell differentiation and function. The costimulatory pathway consisting of the PD-1 receptor and its ligands, PD-L1 and PD-L2, delivers inhibitory signals that regulate T cell activation, tolerance, and immune-mediated tissue damage. This pathway exerts critical inhibitory functions in the context of persistent Ag expression, such as that occurring during chronic infections (27–30). We have previously shown in the K5/TGO/DO11 model that the single best indicator of Ag recognition by a tissue-resident T cell is the upregulation of PD-1. In this study, we also demonstrate that as early as 2 d post–Ag induction, Ag-specific T cells in the skin increase expression of PD-1, whereas its expression is unchanged in the SDLNs, suggesting that tissue T cell recognition of self-antigen precedes that of secondary lymphoid-residing T cells. Whether the PD-L1–PD-1 pathway is functionally relevant for DC–T cell interactions, and thus DC migration in this system, is yet to be determined.

In addition to self-Ag induced migration, the homeostatic signals that mediate DC trafficking from peripheral tissues in the steady-state is poorly understood. It will be important to determine if cutaneous T cells are also involved in the regulation of DC migration in this context, as well as in tissue sites other than skin. Elucidating the molecular pathways governing tissue-resident T cell–induced DC migration will be critical, as these may provide novel avenues for targeting the T cell priming stage of autoimmune disease.

We thank C. Benetiz for assistance with animal husbandry.

The authors have no financial conflicts of interest.