Cutting Edge: Skin CCR10⁺ CD8⁺ T Cells Support Resident Regulatory T Cells through the B7.2/Receptor Axis To Regulate Local Immune Homeostasis and Response

Barrier tissues, such as the skin, are under constant assault by various environmental agents. Resident T cells in barrier tissues play important roles in protecting against assaults, such as infections (1–3), but they could be responsible for the development of tissue inflammatory diseases when their homeostasis is dysregulated (4–6). Understanding the mechanisms regulating the homeostatic presence of T cells in barrier tissues is fundamental to developing cures against tissue inflammatory diseases and infections.

The chemokine receptor CCR10 is expressed on skin-homing and resident T cells (7, 8). It was suggested that, through interaction with its skin-specific ligand CCL27, CCR10 regulates the migration of T cells during skin inflammatory responses (9, 10). However, a later study found that CCR10 knockout (KO) in T cells did not affect their migration into immunization sites of the skin (11). Using a strain of CCR10-KO/EGFP-knockin (CCR10^EGFP/EGFP) mice in which the CCR10-coding sequence is replaced with a DNA fragment coding for EGFP to report CCR10 expression (12), we demonstrated recently that CCR10 is critical for T cell migration into the noninflamed skin (13), suggesting that CCR10 might be important in the establishment of skin-resident T cells. In this article, we report that CCR10-regulated proper establishment of CD8⁺ T cells in the skin is critical for CD4⁺ T cell homeostasis to prevent overreactive inflammatory responses. Furthermore, we found that the B7.2/ligand interaction mediates CD8⁺ T cell regulation of regulatory T cells (Tregs) in the skin.

CCR10^EGFP/EGFP mice were described previously (12). Wild-type (WT) C57BL/6, Rag1^−/−, transgenic OT-I, B7.1^−/−B7.2^−/−, and Foxp3-RFP reporter mice were purchased from The Jackson Laboratory (Bar Harbor, ME). OT-I mice on the CCR10^EGFP/EGFP or B7.2^−/− background were generated by proper crossing. All animal experiments were approved by The Pennsylvania State University Institutional Animal Care and Use Committee.

Abs were purchased from BD Biosciences (San Jose, CA), eBioscience (San Diego, CA), or BioLegend (San Diego, CA). 1-Fluoro-2,4-dinitrobenzene (DNFB) and chicken OVA protein were purchased from Sigma-Aldrich (St. Louis, MO). Cholera toxin was purchased from List Biological Laboratories (Campbell, CA).

The experiments were performed as reported (13). Briefly, purified naive splenic OT-I CD8⁺ T cells (∼0.5 million) from CCR10^EGFP/EGFP, CCR10^+/EGFP, B7.2^−/−, or WT mice were injected i.p. into WT mice, which were then immunized by topical application of 50 μl PBS solution of OVA (5 mg/ml) and cholera toxin (0.5 mg/ml) on the ear or back twice, with an interval of 7 d. For challenge, immunized mice received topical application of OVA (5 mg/ml in PBS) or DNFB (0.5% in 4:1 acetone/olive oil) on unimmunized ear or torso skin 3 mo after the OVA immunization.

A total of 0.5 million sorter-purified splenic CD8⁺ T cells from CCR10^EGFP/EGFP, CCR10^+/EGFP, B7.1^−/−B7.2^−/−, or WT mice, together with 1 million purified WT splenic CD4⁺ T cells, were injected into Rag1^−/− mice. Recipients were analyzed 1–2 mo later.

Skin lymphocyte isolation, flow cytometric (FC) analysis, and cell sorting were performed as previously described (13).

RNA extracted from the skin was reverse transcribed to cDNA and analyzed by SYBR Green real-time PCR with the following primer pairs for specific cytokines: TNF-α, 5′-TTCTATGGCCCAGACCC-3′ and 5′-GGCACCACTAGTTGGTTGTC-3′; IL-1β, 5′-TCTCGCAGCAGCACATCA-3′ and 5′-CACACCAGCAGGTTATCATCAT-3′; IL-10, 5′-ACCAAAGCCACAAAGCAGCC-3′ and 5′-CCGACTGGGAAGTGGGTGC-3′; and β-actin, 5′-CCCATCTACGAGGGCTAT-3′ and 5′-TGTCACGCACGATTTCC-3′.

Skin Tregs or CD4⁺ effector T cells (Teffs) were sorter purified from Foxp3-RFP mice based on RFP signals, cocultured with purified skin CD8⁺ cells from WT or B7.1^−/−B7.2^−/− mice at a 1:2 ratio in the presence of IL-2 (2 ng/ml) for 1 d, and analyzed for survival by annexin V staining and flow cytometry. To block the B7.2/ligand interaction, 5 μg/ml anti-B7.2 Abs (GL1) was added to some cultures.

Data are expressed as mean ± SEM. A two-tailed Student t test or ANOVA test with Tukey adjustment was used to determine statistical significance between two groups or among multiple groups, respectively. The p values < 0.05 were considered significant.

To address whether CCR10 was involved in the establishment of resident CD8⁺ T cells in the skin, we used a T cell–transfer model in which naive OVA-specific transgenic OT-I CD8⁺ T cells from CCR10^+/EGFP or CCR10^EGFP/EGFP mice were transferred into WT mice (referred to as CCR10^+/EGFPOT-I^TR and CCR10^EGFP/EGFPOT-I^TR mice), followed by epicutaneous immunization of ears with OVA, which activated OT-I cells to differentiate into EGFP(CCR10)⁺ T cells (13). One week after immunization, most EGFP⁺ OT-I cells in the skin displayed a surface marker expression pattern that was typical of resident memory T cells (CD103⁺CD69⁺CD44⁺CD62L⁻) (Fig. 1A). CCR10 KO did not affect marker expression on CCR10^EGFP/EGFPEGFP⁺ OT-I cells (Fig. 1A). However, when cotransferred into WT mice, CCR10^EGFP/EGFPEGFP⁺ OT-I or polyclonal T cells migrated much less efficiently into the skin than did their corresponding CCR10^+/EGFP counterparts (Fig. 1B, 1C), supporting the notion that CCR10 is important for T cell migration into untreated skin (13).

We then assessed how CCR10 KO affected the establishment of resident OT-1 cells in the skin during the memory phase. One month after immunization, there were much fewer CCR10^EGFP/EGFP OT-I cells than CCR10^+/EGFP OT-I cells in untreated parts of the skin (torso), and the percentage of EGFP⁺CCR10^EGFP/EGFP OT-I cells was also lower than that of CCR10^+/EGFP controls (Fig. 1D, Supplemental Fig. 1A). In contrast, the percentages of CCR10^EGFP/EGFP and CCR10^+/EGFP OT-I cells at immunization sites of the skin (ear) were similar (Fig. 1D, Supplemental Fig. 1A). Three months after immunization, the percentages of CCR10^+/EGFP and CCR10^EGFP/EGFP OT-I cells in the untreated skin were reduced from their levels at 1 mo after immunization (Fig. 1D). However, the proportion of the reduction was roughly similar for CCR10^+/EGFP and CCR10^EGFP/EGFP OT-I cells (from ∼20% to ∼1% and from ∼4% to ∼0.2%, respectively) (Fig. 1D). There was still no significant difference in the percentage of CCR10^EGFP/EGFP and CCR10^+/EGFP OT-I cells at original immunization sites (Fig. 1D). These results demonstrate that CCR10-dependent migration of T cells into the untreated skin is critical for establishment of resident memory T cells, but CCR10 is not critical for their retention in the skin. In contrast, CCR10 is dispensable for migration and establishment of resident memory T cells at immunization sites of the skin.

The different requirements of CCR10 for migration of T cells into untreated and immunized skin could be due to increased inflammation and/or preferential stimulation of T cells by local Ags at immunization sites. To dissect these, CCR10^EGFP/EGFPOT-I^TR and CCR10^+/EGFPOT-I^TR mice were immunized with OVA on ears and painted with DNFB on one side of the back at the same time to induce OVA-specific and OVA-irrelevant inflammation locally. One week after immunization, OVA-immunized ears and DNFB-treated skin had more infiltration of total OT-I cells than did the untreated skin, primarily due to increased EGFP⁻ OT-I cells (Fig. 1E), indicating that inflammation overrides the requirement of CCR10 for migration of T cells into the skin.

Resident memory T cells play important roles in memory responses. We then challenged OVA-immunized CCR10^EGFP/EGFPOT-I^TR and CCR10^+/EGFPOT-I^TR mice with OVA on the unimmunized skin to induce memory responses 3 mo after the immunization. As expected, CCR10^EGFP/EGFPOT-I^TR mice still had fewer OT-I cells in the skin than did CCR10^+/EGFPOT-I^TR mice 3 d after challenge (Fig. 2A). Surprisingly, however, compared with CCR10^+/EGFPOT-I^TR mice, CCR10^EGFP/EGFPOT-I^TR mice had larger increases in thickness (Fig. 2B), increased microscopic pathology (Fig. 2C), and increased infiltration of neutrophils (Gr-1⁺F4/80⁻) and macrophages (Gr-1⁻F4/80⁺) in the challenged skin (Fig. 2D), indicating an overall enhanced inflammation. Compared with CCR10^+/EGFPOT-I^TR mice, CCR10^EGFP/EGFPOT-I^TR mice also had enhanced TNF-α and reduced IL-10 expression in the challenged skin (Fig. 2E), suggesting defective immune regulation.

Tregs are important in immune regulation. Therefore, we compared their presence in the skin of immunized and/or challenged CCR10^EGFP/EGFPOT-I^TR and CCR10^+/EGFPOT-I^TR mice. There were significantly lower percentages of Foxp3⁺ Tregs in the OVA-challenged skin of immunized CCR10^EGFP/EGFPOT-I^TR mice compared with CCR10^+/EGFPOT-I^TR mice (Fig. 3A). There were also fewer Tregs in untreated skin of immunized CCR10^EGFP/EGFPOT-I^TR mice compared with CCR10^+/EGFPOT-I^TR mice, even before OVA challenge (Fig. 3B). CD4⁺ T cells of untreated skin of immunized CCR10^EGFP/EGFPOT-I^TR mice also had lower percentages of IFNγ⁺ cells and higher percentages of IL-17⁺ cells than did CCR10^+/EGFPOT-I^TR controls (Fig. 3C). These results suggest that reduced CCR10^EGFP/EGFP OT-I cells in the skin result in defective CD4⁺ Treg and Teff homeostasis. Consistent with this, OVA-immunized CCR10^EGFP/EGFPOT-I^TR mice also had enhanced skin inflammation in response to DNFB challenge compared with CCR10^+/EGFPOT-I^TR controls (Fig. 3D, 3E). The dysregulation of Tregs and Teffs was observed in CCR10^EGFP/EGFPOT-I^TR mice as early as 1 wk after OVA immunization (Supplemental Fig. 1B, 1C), suggesting that the modulation of skin CD4⁺ T cells by infiltrating CCR10⁺ CD8⁺ OT-I cells began at the effect phase.

To confirm that regulation of CD4⁺ T cell homoeostasis by CD8⁺ T cells in the skin was a general process, we used another model in which polyclonal CD8⁺ T cells of spleens of CCR10^EGFP/EGFP or CCR10^+/EGFP mice were cotransferred with WT splenic CD4⁺ T cells into Rag1^−/− mice. There were fewer CCR10^EGFP/EGFP CD8⁺ than CCR10^+/EGFP CD8⁺ donor T cells in the skin of recipients 1 mo after transfer (Fig. 3F, Supplemental Fig. 1D), consistent with the requirement of CCR10 for T cell migration into the skin. Associated with this, donor CD4⁺ cells of the skin of recipients cotransferred with CCR10^EGFP/EGFP CD8⁺ T cells contained lower percentages of Foxp3⁺ and IFNγ⁺ cells and higher percentages of IL-17⁺ cells compared with recipients cotransferred with CCR10^+/EGFP CD8⁺ cells (Fig. 3G, Supplemental Fig. 1E, 1F). Together, these results demonstrate that CD8⁺ T cells promote CD4⁺ T cell homeostasis, including supporting Tregs in the skin to help immune homeostasis.

We searched for molecules expressed on skin CD8⁺ T cells that were involved in supporting Tregs. Notably, skin CCR10⁺ CD8⁺ T cells expressed high levels of the costimulatory molecule B7.2 but no or low levels of B7.1, PD-L1, PD-L2, B7H, B7H3, B7X, or MHC class II (Fig. 4A, Supplemental Fig. 2A). Of the two receptors for B7.2, CD28 was equally expressed on CD4⁺ skin Tregs and Teffs, whereas CTLA-4 was highly expressed only on the surface of skin Tregs (Fig. 4B). Although both receptors are involved in the maintenance and function of Tregs (14, 15), the CTLA-4/B7.2 interaction has a much higher affinity than does the CD28/B7.2 interaction (16) and is likely specifically involved in supporting skin Tregs.

We then tested whether B7.2 was required for CD8⁺ T cell modulation of skin CD4⁺ cells by coculturing skin CD4⁺ Tregs or Teffs with skin CD8⁺ cells in vitro in the presence of anti-B7.2 Abs. CD4⁺ Tregs and Teffs were purified from the skin of Foxp3-RFP mice based on the RFP reporter of Foxp3 expression (17). Based on annexin V⁺ staining and loss of RFP signal (due to leakage of cytosolic RFP), significantly higher percentages of Tregs died and fewer of them remained alive after coculture with skin CD8⁺ T cells in the presence of anti-B7.2 Abs than in the presence of isotype-control Abs (Fig. 4C). In contrast, RFP⁻ CD4⁺ Teffs cocultured with skin CD8⁺ cells in the presence of anti-B7.2 or control Abs had similar survival rates (Fig. 4C). Therefore, B7.2 expressed on CD8⁺ T cells is important to support survival of Tregs but not Teffs. Supporting this notion, fewer skin Tregs were alive 1 d after coculture with skin B7.2^−/−B7.1^−/− CD8⁺ cells than with WT CD8⁺ cells (Fig. 4D, Supplemental Fig. 2B).

To test whether B7.2 mediated CD8⁺ T cell regulation of CD4⁺ T cell homeostasis in vivo, we cotransferred B7.2^−/−B7.1^−/− CD8⁺ or WT CD8⁺ cells and WT CD4⁺ cells into Rag1^−/− mice. One month after transfer, mice receiving WT CD4⁺ and B7.2^−/−B7.1^−/− CD8⁺ T cells had fewer skin CD4⁺ Tregs than did those receiving WT CD4⁺ and WT CD8⁺ T cells (Fig. 4E, Supplemental Fig. 2C). In addition, skin CD4⁺ T cells cotransferred with B7.2^−/−B7.1^−/− CD8⁺ T cells expressed higher IL-17 and lower IFN-γ than did those cotransferred with WT CD8⁺ T cells (Fig. 4F, Supplemental Fig. 2D). There were also fewer B7.2^−/−B7.1^−/− donor CD8⁺ T cells in the skin of recipients compared with WT donor CD8⁺ T cells (Fig. 4G, Supplemental Fig. 2E), suggesting bidirectional effects of B7.2/ligand interaction on promoting homeostatic maintenance of Tregs and CD8⁺ T cells in the skin.

We further tested whether B7.2 KO in CD8⁺ OT-I cells would affect skin CD4⁺ T cell homeostasis in the OT-I cell–transfer model, in which B7.2^−/− or WT OT-I cells were transferred into WT mice (referred to as B7.2^−/−OT-I^TR and WT OT-I^TR mice). One week after OVA immunization on ears, there were slightly lower percentages of Tregs in the untreated torso skin of B7.2^−/−OT-I^TR mice compared with WT OT-I^TR mice, whereas there was no significant difference in the infiltration of B7.2^−/− and WT OT-I cells into the skin (Fig. 4H, Supplemental Fig. 2F). One month postimmunization, there were significantly lower percentages of donor OT-I cells and host Tregs in the untreated skin of B7.2^−/−OT-I^TR mice compared with WT OT-I^TR mice (Fig. 4H, Supplemental Fig. 2F). These results reveal that B7.2-mediated signals are important in the homeostatic maintenance of CD8⁺ T cells and Tregs in the skin.

In conclusion, various subsets of T cells reside in barrier tissues, such as skin. Their balanced presence is critical for local immune protection, and it prevents the development of inflammatory diseases. Our study reveals a novel process in which CD8⁺ T cells regulate CD4⁺ T cell homeostasis by supporting the maintenance of Tregs in the skin, and B7.2 expressed on CD8⁺ T cells is involved in the CD8⁺ T cell–regulated survival of Tregs. Considering that CTLA-4 is highly expressed on skin Tregs and that engagement of CTLA-4 could block activation-induced cell death of T cell hybridoma in vitro (18), it will be interesting to determine whether the B7.2/CTLA-4 interaction plays a role in CD8⁺ T cell–regulated survival and maintenance of Tregs in vivo. The general importance of the CD8⁺ T cell–regulated maintenance of Tregs in skin homeostasis and inflammation also requires further investigation.

The authors have no financial conflicts of interest.