Proenkephalin⁺ regulatory T cells expanded by ultraviolet B exposure maintain skin homeostasis with a healing function

Significance

Regulatory T (Treg) cells, expressing CD25 and Foxp3, constitute about 5 to 10% of peripheral CD4⁺ T cells and maintain immunological self-tolerance through the suppression of various immune responses. Here we found that skin Treg cells expanded by UVB exposure possess a unique TCR repertoire and a healing function. UVB-expanded skin Treg (UVB-skin Treg) cells expressed proenkephalin (PENK), an endogenous opioid precursor, and amphiregulin (AREG), the epidermal growth factor receptor ligand. The Treg-derived PENK and AREG promoted keratinocyte outgrowth in a skin explant assay. Moreover, UVB-expanded skin Treg cells played a key role in promoting wound healing in vivo. Our results provide a new implication in developing a therapy using PENK⁺UVB-skin Treg cells.

Abstract

Regulatory T (Treg) cells, expressing CD25 (interleukin-2 receptor α chain) and Foxp3 transcription factor, maintain immunological self-tolerance and suppress various immune responses. Here we report a feature of skin Treg cells expanded by ultraviolet B (UVB) exposure. We found that skin Treg cells possessing a healing function are expanded by UVB exposure with the expression of an endogenous opioid precursor, proenkephalin (PENK). Upon UVB exposure, skin Treg cells were expanded with a unique TCR repertoire. Also, they highly expressed a distinctive set of genes enriched in “wound healing involved in inflammatory responses” and the “neuropeptide signaling pathway,” as indicated by the high expression of Penk. We found that not only was PENK expression at the protein level detected in the UVB-expanded skin Treg (UVB-skin Treg) cells, but that a PENK-derived neuropeptide, methionine enkephalin (Met-ENK), from Treg cells promoted the outgrowth of epidermal keratinocytes in an ex vivo skin explant assay. Notably, UVB-skin Treg cells also promoted wound healing in an in vivo wound closure assay. In addition, UVB-skin Treg cells produced amphiregulin (AREG), which plays a key role in Treg-mediated tissue repair. Identification of a unique function of PENK⁺ UVB-skin Treg cells provides a mechanism for maintaining skin homeostasis.

Regulatory T (Treg) cells were first defined as CD25 (interleukin (IL)-2 receptor α chain)⁺ CD4⁺ T cells, which suppress multiple organ-targeted autoimmune diseases (1). CD25⁺CD4⁺ Treg cells constitute about 5 to 10% of peripheral CD4⁺ T cells and the development and function of Treg cells are controlled by the Foxp3 transcription factor (2–4). Now it is well established that Treg cells play an important role, not only in maintaining immune tolerance, but also in suppressing various immune responses in both mice and humans (5–7). Moreover, emerging evidence highlights a critical role of tissue Treg cells with tissue homeostasis and tissue-specific function (8). Control of tissue Treg cells is therefore likely to be important in maintaining tissue homeostasis.

The skin is the largest organ in the human body, containing many immune cells (9–12). Skin Treg cells are controlled by environmental factors such as the microbiota and ultraviolet (UV) irradiation (13, 14). UVB irradiation expands functional skin Treg cells up to ∼60% of skin CD4⁺ T cells, resulting in Treg cell homeostasis in peripheral lymphoid organs (14). The origin of Treg expansion is mostly from the in situ proliferation of thymic-derived Treg cells in the UVB-exposed skin, not the influx of T cells (14). A recent study showed new functions of skin Treg cells such as promoting hair follicle-stem cell proliferation and differentiation (15). However, functional characteristics of UVB-expanded skin Treg (UVB-skin Treg) cells remain to be elucidated and how they are implicated in the regulation of skin homeostasis is still unclear.

To explore the function of UVB-skin Treg cells, their gene expression was investigated using RNA sequencing (RNA-seq). Here we show that UVB irradiation expands skin Treg cells expressing a unique T cell receptor (TCR) repertoire and a unique gene set related to a healing function including Penk and Areg. These expanded Treg cells play an important role in promoting wound healing via the production of proenkephalin (PENK), an endogenous opioid precursor, and amphiregulin (AREG), the epidermal growth factor receptor ligand involved in tissue repair. These results indicate that UVB-skin Treg cells with a healing function act to maintain skin homeostasis.

Results

UVB-Skin Treg Cells Possess a Unique T Cell Receptor Repertoire and Gene Expressions Enriched in the Neuropeptide Signaling Pathway.

The expansion of skin Treg cells by UVB irradiation is dependent on the CD11b-type langerin⁻ dermal dendritic cell (DC) subset (16). As it’s possible that the dermal DC subset presents antigens to induce clonal expansions of skin Treg cells upon UVB irradiation, we investigated if UVB-skin Treg cells possess a unique TCR repertoire. There was a significant increase of CD25^high Foxp3⁺CD4⁺ Treg cells in the skin of UVB-irradiated mice compared to those in UVB-nonirradiated mice (Fig. 1A and SI Appendix, Fig. S1A). Since CD25^high CD4⁺ T cells sorted from wild-type mice contained >90% of Foxp3⁺CD4⁺ T cells (14, 16), we sorted CD25^highCD4⁺ as Treg cells and CD25⁻CD4⁺ as conventional T (Tconv) cells from the skin, or lymph nodes (LNs) from UVB-irradiated mice (Fig. 1A and SI Appendix, Fig. S1B). The complementarity-determining region (CDR)3 of the TCRα and TCRβ chain of individually sorted cells were sequenced (Fig. 1B and SI Appendix, Fig. S1 C and D). The advantages of single-cell TCR repertoire analysis is that clones with identical paired TCRα and TCRβ can be determined. Notably, in UVB-skin Treg cells, there were 5 identically paired TCRα and TCRβ with 11 per 79 pairs, which constitutes 13.9% of the sequenced paired clones (Fig. 1B, violet slices). On the other hand, Tconv cells from UVB-irradiated skin (UVB-skin Tconv) had only one identical paired clone with 2 per 87 pairs, which represents 2.3% of the sequenced paired clones (Fig. 1B, green slices). It is of note that the expanded clones in UVB-skin Treg cells and UVB-skin Tconv cells did not overlap (SI Appendix, Fig. S2). Moreover, there was no identical paired clone per 96 sequenced paired clones of LN Treg cells from UVB-irradiated mice (UVB-LN Treg), or 120 sequenced paired clones of LN Tconv cells from UVB-irradiated mice (UVB-LN Tconv) (Fig. 1B). Also, the expanded clones in UVB-skin Treg cells did not overlap with those in Treg cells from visceral adipose tissue (VAT), muscle, and brain, as reported previously (17–19). One clone expanded in UVB-skin Treg cells (the third one from the top in SI Appendix, Fig. S2) was found in UVB-LN, suggesting the possible migration of the UVB-skin Treg clone to LNs. TCR repertoire analysis of CD25^high Treg cells from normal skin could not be performed due to low cell yield as there were fewer CD25^high Treg cells in the skin without UVB irradiation (Fig. 1A and SI Appendix, Fig. S1A). Although the same diversity may exist prior to UVB irradiation, skin Treg cells in UVB-irradiated mice may be clonally expanded with a unique TCR repertoire.

Fig. 1.

UVB-skin Treg cells expanded with distinctive gene expression pattern and express PENK. (A) Experimental scheme of UVB irradiation and representative fluorescence-activated cell sorting (FACS) plots of CD4⁺ T cells from control and UVB-exposed skin. B6 mice were irradiated with 500 mJ/cm² UVB as previously reported (14). After 7 d, ear skin was analyzed for Foxp3 and CD25 expression gated on live CD45⁺CD4⁺ cells. The graphic is the summary of four experiments showing the frequency of Foxp3⁺CD25^high Treg cells/CD4⁺T cells in the skin. Each circle indicates a mixture of two or four mice (UVB, n = 4; control [UVB−], n = 4). **P < 0.01. (B) Single-cell TCR repertoire analysis of paired TCRα and TCRβ. Pie charts showing frequencies of expanded clones from UVB-skin Treg, UVB-skin Tconv, UVB-LN Treg, or UVB-LN Tconv cells. n, number of paired analyzed sequences per each group. The widths of the violet or green slice illustrate the frequency of a clone’s total repeats. Total frequency of expanded clones is shown at the Top Right. The result is combined data from two separate experiments. (C) The scatterplots represent the ratio of the expression level of UVB-skin Treg cells to UVB-LN Treg cells (x axis) and those to UVB-skin Tconv cells (y axis). Red dots indicate the 15 genes identified as highly expressed genes in UVB-skin Treg cells by the analysis by the Venn diagram from SI Appendix, Fig. S4A. (D) Heatmap showing the genes that are highly expressed in UVB-skin Treg cells compared with UVB-skin Tconv cells. Thirty-one genes were extracted under the following conditions to generate the heatmap. First, highly expressed genes in UVB-skin Treg cells (fragments per kilobase of exon per million mapped fragments [FPKM] >150) were chosen. Then, FPKM value of UVB-skin Treg cells more than fourfold than those of UVB-skin Tconv cells (P < 0.05) were chosen. The genes are ranked in descending order of fold changes. The heatmap was generated using z-score. The result from one experiment is shown. (E) The neuropeptide signaling pathway is enriched by GSEA using gene expression datasets of Treg and Tconv from UVB-irradiated skins. Top five ranked gene ontology terms of biological process in normalized enrichment score (NES) are shown (Left in table). The graph represents the enrichment plot of the neuropeptide signaling pathway gene set (Right); x axis is rank order of genes from the most up-regulated to the most down-regulated between UVB-skin Treg and UVB-skin Tconv cells. (F) UVB-skin Treg cells express PENK. B6 mice were irradiated with (UVB) or without UVB (control) and after 7 d, cells were prepared from ear skin, stimulated with PMA and ionomycin for 3.5 h, and stained with surface markers. Then, cells were divided into three and stained intracellularly with anti-Foxp3, anti-PENK Ab, or isotype control, respectively. Representative FACS plots are shown. Cells were gated on live CD45⁺CD4⁺ cells. The frequency of PENK⁺ CD25^high Treg/CD4⁺ T cells from the Bottom plots is summarized in the graphic on the Right. Each circle indicates a mixture of two to six mice (n = 8). Data are presented as a summary of four separate experiments. ****P < 0.0001. (G) UVB-skin Treg cells express AREG. As in F, but cells from ear skin or LN were stained with anti-AREG Ab and anti-Foxp3 without PMA and ionomycin. Representative FACS plots are shown. Cells were gated on live CD45⁺ CD4⁺ cells. The frequency of AREG⁺ Foxp3⁺ Treg/CD4⁺ T cells is summarized in the graphic on the Right. Each circle indicates an individual mouse for skin and LN (n = 4). Data are presented as a summary of four separate experiments.***P < 0.001, N.S., not significant.

To gain insight into the physiological roles of UVB-skin Treg cells, we next performed RNA-seq analysis. We found that UVB-skin Treg cells highly expressed a unique set of genes compared to UVB-skin Tconv cells and UVB-LN Treg cells (Fig. 1 C and D and SI Appendix, Fig. S3). The highest fold increase relative to UVB-skin Tconv cells was Penk, which encodes proenkephalin (PENK) (Fig. 1D), and the highest fold increase relative to UVB-LN Treg cells was Areg, which encodes amphiregulin (AREG) (SI Appendix, Fig. S3). There were 15 genes that overlapped and were identified as highly expressed genes in UVB-skin Treg cells (Fig. 1C and SI Appendix, Fig. S4 A, Left). UVB-skin Treg cells also highly expressed Treg-signature genes including Foxp3, Klrg1, Tnfrsf4, and Ctla4 compared to UVB-skin Tconv and UVB-LN Treg cells (SI Appendix, Fig. S5) (8, 19). To compare the gene expression of skin Treg cells from UVB-nonirradiated mice (normal-skin Treg), we analyzed the database (20). Normal-skin Treg cells highly expressed Treg-signature genes as observed in UVB-skin Treg cells (SI Appendix, Fig. S6A). Notably, normal-skin Treg cells did not highly express Penk compared to skin Tconv (normal-skin Tconv) or LN Treg (normal-LN Treg) cells from UVB-nonirradiated mice (SI Appendix, Fig. S6 B–D). There were 14 genes that overlapped and were identified as highly expressed genes in normal-skin Treg cells (SI Appendix, Fig. S6 B and E). There were three commonly highly expressed genes both in UVB-skin Treg and normal-skin Treg cells, i.e., Klrg1, Areg, and Gzmb (Fig. 1C and SI Appendix, Fig. S6B). To further analyze the unique feature in UVB-skin Treg cells, gene set enrichment analysis (GSEA) with all genes was performed. We found that the most significant enriched pathway in UVB-skin Treg cells was the “neuropeptide signaling pathway” (false discovery rate [FDR] = 0.02098) and the core-enriched gene identified from this pathway was Penk (Fig. 1E). On the other hand, normal-skin Treg cells showed that the most enriched pathway was “tolerance induction” (FDR = 0.104079, SI Appendix, Fig. S6F), and that the neuropeptide signaling pathway was not enriched (ranked at 744, FDR = 0.615320).

To investigate if UVB-skin Treg cells express PENK and AREG at protein levels, flow cytometric analysis was performed. The simultaneous staining of PENK and Foxp3 was impossible due to the different fixation/permeabilization conditions. Therefore, the samples stained with CD25 were divided and further stained with Foxp3, PENK, or isotype. Compared to normal-skin Treg cells (Fig. 1F, control), PENK⁺CD25^high UVB-skin Treg cells, which corresponded to Foxp3⁺, were expanded significantly (Fig. 1F, UVB). In the case of AREG, the simultaneous staining with Foxp3 was successful. Therefore, we used Foxp3 because it is a more specific marker than CD25 for Treg cells. Notably, upon UVB irradiation, there were about 15% of AREG⁺ Treg/CD4⁺ T cells without additional TCR stimulation by phorbol-12-myristate-13-acetate (PMA) plus ionomycin (Fig. 1 G, Top). This was surprising because AREG production was generally detected with PMA plus ionomycin (21). Also, a significant increase of AREG⁺ skin Treg cells was observed with UVB irradiation (Fig. 1 G, Top). On the other hand, there were only about 1% AREG⁺Treg/CD4⁺ T cells both in normal-LNs and UVB-LNs without PMA plus ionomycin (Fig. 1 G, Bottom). Therefore, UVB-skin Treg cells, which were Foxp3⁺CD25^high, produced PENK and AREG at the protein levels.

We next analyzed the gene expression of Treg cells from VAT, muscle, and brain using the database (18, 19, 22). GSEA showed that VAT, muscle, and brain Treg cells were enriched in distinct pathways (SI Appendix, Fig. S7). The data from each tissue Treg cell were also compared to the corresponding tissue Tconv cells and Treg cells from the spleen, or LNs. VAT Treg, muscle Treg, and brain Treg cells highly expressed distinct sets of genes compared to corresponding tissue Tconv and spleen, or LN Treg cells (SI Appendix, Figs. S4A and S8). When the highly expressed genes in muscle, VAT, brain, and UVB-skin Treg cells were analyzed in a Venn diagram, each tissue Treg cell highly expressed unique sets of genes (SI Appendix, Fig. S4B). There were only three common genes highly expressed in muscle, VAT, brain, and UVB-irradiated skin, which are Il1rl1, Areg, and Klrg1 (SI Appendix, Fig. S4B).

Il1rl1 is one of the three common genes (SI Appendix, Fig. S4B). Although UVB-skin Treg expressed suppression of tumorigenicity 2 (ST2) (IL-33 receptor), encoded by Il1rl1, we found that UVB-skin Treg cells were not reduced in ST2 knockout (KO) mice (23) (SI Appendix, Fig. S9). UVB irradiation did not affect the Treg cells in LNs and spleen in ST2 KO mice, either (SI Appendix, Fig. S9B). Therefore, the IL-33–ST2 axis is not required for the expansion of skin Treg cells by UVB irradiation, whereas it is essential for Treg expansion in VAT, muscle, intestine, and brain (19, 24–26). UVB-skin Treg cells also expressed CD44, killer cell lectin-like receptor 1 (KLRG1), inducible costimulator (ICOS), which indicate an activated phenotype (27, 28) and 4-1BB (CD137, tumor necrosis factor superfamily member 9), which is encoded by Tnfrsf9 and known to be important for Treg expansion (29) (SI Appendix, Fig. S10A). UVB-skin Treg cells were mostly of the CD44^highCD62L^low effector or memory phenotype (27) (SI Appendix, Fig. S10B).

To investigate if Treg cells in other barrier organs are also expanded by UVB irradiation, Treg cells in the intestine were chosen to be compared because the intestine and skin are constantly in contact with the outside, such as microbiota and other environmental factors (10). In contrast to skin, Treg cells in the intestine did not increase in the UVB-irradiated mice (SI Appendix, Fig. S11). Therefore, the UVB-induced Treg expansion with the unique feature is specific to skin as the barrier organ.

These results indicate that UVB-skin Treg cells are different from other tissue Treg cells regarding their requirement of the IL-33–ST2 axis and high expression of PENK and AREG.

Treg Cells Promote Keratinocyte Outgrowth via Production of PENK and AREG upon UVB Irradiation.

A PENK-derived neuropeptide, Met-ENK, stimulates opioid δ-receptors on keratinocytes, which accelerates cell migration and wound repair (30). AREG is the epidermal growth factor receptor ligand and involved in repair of injured lung, skeletal muscle, and ischemic brain via Treg cells (18, 19, 21). To investigate its direct effect on keratinocyte outgrowth, Met-ENK or AREG was added to the skin explant assay (31, 32). Notably, we found that Met-ENK or AREG significantly enhanced keratinocyte outgrowth when added to the skin explants from naïve mice (Fig. 2 A and B). Thus, it raised a hypothesis that PENK and AREG derived from UVB-skin Treg cells play a key role in wound healing.

Fig. 2.

Treg cells enhance keratinocyte outgrowth by producing PENK and AREG. (A) The PENK derivate, Met-ENK, promotes keratinocyte (KC) outgrowth. Skin explants from naïve mice were excised from ear skin by a 3-mm punch biopsy and cultured in collagen-coated wells with (Met-ENK) or without (control) Met-ENK (500 nM). After 2-d culture, whole skin explants were photographed, and length and area of keratinocyte outgrowth were measured in a blind manner. If the skin explants had parts lifted up during the culture, keratinocyte outgrowth was measured by length, but not the area. Data are presented as a summary of three separate experiments (control n = 10 for area, n = 13 for length; Met-ENK n = 9 for area, n = 14 for length). Each square indicates an individual skin explant. *P < 0.05. (B) AREG promotes keratinocyte outgrowth. As in A, but skin explants were cultured with (AREG) or without (control) recombinant AREG (100 ng/mL). Data are presented as a summary of three separate experiments (control n = 8 for area, n = 9 for length; AREG n = 6 for area, n = 9 for length). Each circle indicates an individual skin explant. *P < 0.05. (C) UVB-LN Treg cells express PENK. B6 mice were irradiated with UVB and after 7 d, cells were prepared from ear skin, stimulated with PMA and ionomycin for 3.5 h, and stained with surface markers and then stained with anti-PENK Ab or isotype intracellularly. Representative FACS plots are shown. In Left contour plots, cells were gated on live CD45⁺CD4⁺ cells. In histograms, cells were gated on CD25^highCD4⁺Treg cells (Treg) or CD25⁻CD4⁺T cells (Tconv). The frequency and mean fluorescent intensity (MFI) for PENK was summarized in the graphics. Data are from one experiment and each circle indicates an individual mouse (n = 4). ***P < 0.001, ****P < 0.0001. (D) UVB-LN Treg cells express AREG with TCR stimulation. As in Fig. 1G, but UVB-LN cells were stimulated with PMA and ionomycin in the presence of brefeldin A for 3.5 h. In contour plots, cells were gated on live CD45⁺ CD4⁺ cells. In histograms, cells were gated on Foxp3⁺CD4⁺Treg cells (Treg) or Foxp3⁻CD4⁺T cells (Tconv). The frequency and MFI for the AREG expression are summarized in the Right graph. Data are presented as a summary of two separate experiments and each circle indicates an individual mouse (n = 5). ****P < 0.0001. (E) Experimental scheme of skin explant assay with Treg cells from UVB-irradiated mice and representative pictures. As in A, but skin explants from naïve B6 mice were cocultured with CD25^highCD4⁺ Treg or CD25⁻CD4⁺ Tconv cells from UVB-LN cells (2 × 10⁵/well). They were cultured with anti-CD3/CD28 beads plus IL-2 for 2 d. Control indicates no T cells, no anti-CD3/CD28 beads, and no IL-2. A part of representative pictures is shown. (Scale bar, 200 μm.) Dashed lines represent radical distances migrated and the borders of keratinocyte outgrowth. (F) As in E, but naltrindole (5 μM) was added to the skin explants with or without CD25^highCD4⁺ T cells in the presence of anti-CD3/CD28 beads plus IL-2. Control indicates no T cells, no anti-CD3/CD28 beads, and no IL-2. Data are presented as a summary of three separate experiments (control, n = 8; CD25^high, n = 10; CD25^high + naltrindole; n = 10; naltrindole; n = 9). Each mark indicates an individual skin explant. Statistics were analyzed by Dunnett’s multiple comparisons test. *P < 0.05, N.S., not significant. (G) As in F, but anti-AREG Ab (10 μg/mL) was added. Data are presented as a summary of two separate experiments (control, n = 7; CD25^high, n = 7; CD25^high + anti-AREG Ab; n = 6; CD25⁻, n = 4). Each mark indicates an individual skin explant. Statistics were analyzed by Dunnett’s multiple comparisons test. *P < 0.05, N.S., not significant.

Next, to examine if Treg cells can promote keratinocyte outgrowth, Treg cells from UVB-irradiated mice were added to the skin explants from naïve mice. It was difficult to obtain sufficient numbers of viable UVB-skin Treg cells to add into the skin explants. Therefore, UVB-LN Treg cells were added to the skin explants with anti-CD3/CD28 beads because UVB-LN Treg cells produced PENK and AREG with TCR stimulation (Fig. 2 C and D). UVB-LN Treg cells produced a small amount of PENK (Fig. 2C), which might not be specific to UVB condition (SI Appendix, Fig. S12). In contrast, the higher production of PENK from skin Treg cells was specific to UVB (Fig. 1F). Upon TCR stimulation by PMA plus ionomycin, most UVB-LN Treg cells also produced AREG (Fig. 2D), in contrast to those without TCR stimulation (Fig. 1 G, Bottom). Notably, in the skin explant assay, we observed that the presence of UVB-LN Treg cells promoted keratinocyte outgrowth (Fig. 2 E–G). Fewer numbers of UVB-LN Treg cells also showed the similar effect on the keratinocyte outgrowth (SI Appendix, Fig. S13A). To investigate if promoting keratinocyte outgrowth is dependent on PENK or AREG from Treg cells, we used a specific δ-opioid receptor antagonist, naltrindole (33) or anti-AREG neutralizing antibody (Ab) (19). We found that the presence of naltrindole or anti-AREG neutralizing Ab in the skin explant culture blocked the effect of promoting keratinocyte outgrowth by UVB-LN Treg cells (Fig. 2 F and G and SI Appendix, Fig. S13 B and C).

These data indicate that, upon UVB irradiation, Treg cells produce PENK and AREG to promote keratinocyte outgrowth, suggesting a healing function of UVB-skin Treg cells.

UVB-Skin Treg Cells Highly Express Wound Healing-Related Genes and Suppress Inflammation Which Delays Wound Healing.

The above data showed the unique role of UVB-skin Treg cells in keratinocyte outgrowth for healing. Indeed, we found that UVB-skin Treg cells highly expressed genes related to keratinocyte proliferation (34, 35) (Fig. 3 A, Left). UVB-skin Treg cells also highly expressed wound healing-related genes relating to angiogenesis, chemotaxis, and extracellular matrix formation (Fig. 3A). To further confirm the healing function of UVB-skin Treg cells, the gene ontology (GO) analysis using the list of 74 genes extracted in SI Appendix, Fig. S3 was performed. The most enriched biological process was “wound healing involved in inflammatory responses” in UVB-skin Treg cells (Fig. 3B). Moreover, the GO analysis using the list of 31 genes extracted in Fig. 1D showed that “negative regulation of IFN-γ production” and “negative regulation of inflammatory response” were among the 10 most enriched biological processes in UVB-skin Treg cells (Fig. 3C).

Fig. 3.

UVB-skin Treg cells are enriched in genes with healing function and suppress inflammation in the UVB-irradiated skin. (A) UVB-skin Treg cells highly express genes related to wound healing. Scatterplots of gene expression in UVB-skin Treg vs. UVB-LN Treg cells (Top), or UVB-skin Treg vs. UVB-skin Tconv cells (Bottom). Genes highly expressed in UVB-skin Treg cells (more than twofold, P < 0.05) compared with UVB-LN Treg or UVB-skin Tconv cells, and those related to keratinocyte proliferation (34, 35) are presented in red dots (Left). Genes highly expressed in UVB-skin Treg cells (more than twofold, P < 0.05) compared with UVB-LN Treg or UVB-skin Tconv cells, and those related to wound healing such as angiogenesis, chemotaxis, or extracellular matrix in the list of GO terms in gene database from National Center for Biotechnology Information (NCBI) (https://www.ncbi.nlm.nih.gov/gene/) are presented by red dots and their gene names are shown if FPKM values were >150. (B) Top 10 GO biological processes in UVB-skin Treg vs. UVB-LN Treg cells. The GO enrichment analysis with the PANTHER classification system uses the list of 74 genes in SI Appendix, Fig. S3. (C) The top 10 GO biological processes in UVB-skin Treg vs. UVB-skin Tconv cells. The GO enrichment analysis with the PANTHER classification system uses the list of 31 genes in Fig. 1D. (D) Treg cells suppress IFN-γ production in the UVB-irradiated skin. Real-time PCR analysis of IFN-γ gene expression in ear skin from control naïve mice (control), UVB-irradiated Foxp3^DTR mice treated with (UVB + DT) or without DT (UVB) after day 7 after UVB irradiation. Data are representative of two separate experiments (ear numbers, n = 6 per group). Each circle indicates an individual mouse ear. The levels of indicated gene expression relative to the mean value of control is shown. (E) The expression of T-bet on skin CD4⁺ T cells. Ear skin from control naïve mice (control), or UVB-irradiated Foxp3^DTR mice treated with (UVB + DT) or without DT (UVB) on day 7 after UVB irradiation was analyzed. Each circle indicates an individual mouse. Data are representative of two separate experiments (mouse numbers, n = 3 per group). (F) Treg cells suppress proinflammatory macrophage accumulation in the UVB-irradiated skin. As in E, but expression of Ly6C on CD45.2⁺CD11b⁺F4/80⁺ macrophages in the ear skin from control naïve mice (control), or UVB-irradiated Foxp3^DTR mice treated with (UVB + DT) or without DT (UVB) was examined. Representative FACS plots are shown. Cells were gated on live CD45⁺CD11b⁺ cells. Each circle indicates an individual mouse. Data are representative of three separate experiments (mouse numbers, n = 3 per group).

To investigate if our UVB irradiation protocol induced skin inflammation, ear swelling was measured. The ear thickness was increased at day 7 and returned to normal levels at day 16 (SI Appendix, Fig. S14A). Real-time PCR analysis of whole skin showed that mRNA expression of inflammatory cytokines, i.e., IL-1β and tumor necrosis factor (TNF)-α, was up-regulated after UVB irradiation (SI Appendix, Fig. S14B). As inflammation delays wound healing (36, 37), we investigated if Treg cell-depletion enhanced inflammation in the UVB-irradiated skin. Using Foxp3-IRES-GFP-DTR (Foxp3^DTR) mice expressing the diphtheria toxin receptor (DTR), in which Foxp3⁺ Treg cells can be depleted by diphtheria toxin (DT) injection (38), we found that depletion of Treg cells enhanced the expression of IFN-γ, IL-1β, and TNF-α, expansion of T-bet⁺Foxp3⁻CD4⁺ T effector cells, and accumulation of CD11b⁺F4/80⁺Ly6C^high proinflammatory macrophages in the skin of UVB-irradiated mice (Fig. 3 D–F and SI Appendix, Fig. S14C). The ear swelling induced by UVB irradiation was also enhanced by Treg depletion (SI Appendix, Fig. S14D). Therefore, Treg cells play an important role in inhibiting UVB-induced skin inflammation.

These data indicated that UVB-skin Treg cells enriched with wound healing-related genes may promote healing by suppressing the inflammation in the UVB-irradiated skin, as well as through the production of PENK and AREG.

UVB Promotes Wound Healing through Treg Cells In Vivo.

Finally, we examined whether UVB irradiation itself enhanced the response to cutaneous wounds in mice. On day 6 after UVB irradiation when skin Treg expansion has already occurred (14), we made full-thickness skin wounds on the back of mice and investigated wound closure (Fig. 4A). Compared with control naïve mice, we found that UVB irradiation significantly promoted wound healing (Fig. 4A). To detect the epithelialization under the crust migrating into the wound bed, the length of the epidermal tongue, detected by keratin-14⁺ keratinocytes, was visualized by immunofluorescent microscopy (32, 34). Epidermal tongues in UVB-treated mice were significantly longer compared to those in control naïve mice in wounds at days 3 and 5 (Fig. 4B). To deny the effect of stress, cortisol levels were measured in the wounded mice with or without UVB irradiation. There was no significant increase of cortisol secretion (SI Appendix, Fig. S15). To further confirm the effect of UVB irradiation, keratinocyte outgrowth of skin explants from mice with or without UVB irradiation were examined. Skin explants were excised from UVB-irradiated mice on day 6 after UVB irradiation when Treg cell expansion had already occurred (Fig. 4C). Compared with skin explants from control naïve mice, keratinocyte outgrowth was significantly enhanced in those from UVB-irradiated mice (Fig. 4C). These results show UVB irradiation promotes wound healing in mice.

Fig. 4.

UVB irradiation promotes wound healing, which is canceled by Treg depletion. (A) UVB irradiation promotes wound healing in mice. B6 mice (7w) were irradiated with or without UVB on day −6 (UVB and control, respectively). On day 0 when skin Treg cells were expanded, two to three wounds were made on the back skin of each mouse. Wound closure is pictured with a ruler at the indicated time point and was measured in a blind manner. Representative pictures are shown. Data are presented as a summary of four separate experiments (wound numbers: control, n = 45, except 39 on day 6; UVB, n = 47, except 39 on day 6). Each symbol indicates an individual wound and each color indicates the data from individual experiments. Statistics were analyzed by mixed effect analysis with Bonferroni multiple comparison test (Prism 8). *P < 0.05, ****P < 0.0001. (B) UVB irradiation promotes elongation of epidermal tongues. As in A, but length of epidermal tongues stained by anti-keratin 14 Ab (green) was measured in day 3 or day 5 wounds of UVB-irradiated or control B6 WT mice (UVB and control, respectively). A representative picture is shown. (Scale bar, 500 μm.) Data are presented as a summary of two separate experiments (epidermal tongue numbers: day 3, control n = 10, UVB n = 14; day 5, control n = 14, UVB n = 19). Each circle indicates an individual epidermal tongue. **P < 0.01, ***P < 0.001. (C) UVB irradiation promotes keratinocyte outgrowth in skin explant assay. As in Fig. 2A, but B6 mice were irradiated with or without UVB on day −6 (UVB and control, respectively), and skin explants were excised from ear skin on day 0. Representative pictures of skin explants are shown. (Scale bars, 200 μm.) Dashed lines represent radical distances migrated and the borders of keratinocyte outgrowth. Data are presented as a summary of two separate experiments (control n = 13 for length, n = 4 for area; UVB n = 14 for length, n = 5 for area). Each circle indicates an individual skin explant. *P < 0.05, ****P < 0.0001. (D) Treg cells play an important role in promoting wound healing in the UVB-irradiated mice. As in A, but Foxp3^DTR mice were irradiated with UVB on day −6 and injected intraperitoneally (i.p.) with DT on days −6, −4, −2, 0, 2, and 4 (UVB + DT). Control Foxp3^DTR mice were i.p. injected with PBS on days −6, −4, −2, 0, 2, and 4 (control). Data are presented as a summary of two separate experiments (wound numbers: control, n = 15; UVB, n = 15). Each symbol indicates an individual wound. Statistics were analyzed by two-way ANOVA with Bonferroni multiple comparison test. *P < 0.05, ***P < 0.001. (E) Elongation of epidermal tongues promoted by UVB irradiation was canceled in Treg-depleted mice. As in D, but length of epidermal tongues stained by anti-keratin 14 Ab was measured in day 5 wounds. Data are presented as a summary of two separate experiments (epidermal tongue numbers: control, n = 11; UVB + DT, n = 16). N.S., not significant. (F) Keratinocyte outgrowth promoted by UVB irradiation was canceled in Treg-depleted skin explants. As in C, but skin explants were obtained from Foxp3^DTR mice that had been treated with UVB and DT (UVB + DT) or treated only with PBS without UVB (control). Data are presented as a summary of two separate experiments (control n = 17 for length, n = 10 for area; UVB + DT n = 17 for length, n = 8 for area). Each triangle indicates an individual skin explant. N.S., not significant.

To evaluate the importance of Treg cells in UVB-promoted wound healing, we next used Foxp3^DTR mice. Notably, compared with control Treg-sufficient naïve mice, UVB irradiation failed to promote wound healing and elongation of the epidermal tongue in Treg-depleted mice (Fig. 4 D and E). The promotion of keratinocyte outgrowth was also negated in UVB-irradiated Treg-depleted mice (Fig. 4F). Also, UVB-irradiated Treg-depleted Foxp3^DTR mice delayed wound healing significantly compared to UVB-irradiated Treg-sufficient Foxp3^DTR mice (SI Appendix, Fig. S16A). DT injection to wild-type B6 mice did not affect the wound healing (SI Appendix, Fig. S16B). Therefore, Treg cells play a key role in promoting wound healing in UVB-irradiated mice.

To examine if UVB has an effect on wound healing in human skin, the database (39) was analyzed. We found that gene sets related to wound healing were one of the enriched gene sets in the UVB-treated human skin (SI Appendix, Fig. S17). Therefore, UVB irradiation may also play a role in promoting wound healing in humans. Further investigation is required in human samples.

Discussion

In this study, we demonstrated that skin Treg cells expanded by UVB irradiation expressed PENK, which promoted keratinocyte outgrowth. UVB-skin Treg cells possessed a distinct pattern of gene expression and also expressed AREG. Moreover, UVB-skin Treg cells suppressed skin inflammation and promoted wound healing. Collectively, our results indicate that UVB-skin Treg cells support wound healing to maintain skin homeostasis.

UVB-skin Treg cells exhibit distinctive features compared to other tissue Treg cells for the following three reasons. First, UVB-skin Treg cells express PENK and a unique set of genes related to wound healing (Fig. 1D and SI Appendix, Figs. S3 and S4). CD25^highFoxp3⁺ Treg cells producing PENK expanded in the UVB-irradiated skin (Fig. 1F). It is quite interesting that Treg cells express PENK in the skin, which is not part of the central nervous system. In this study, we propose that the PENK-derived neuropeptide, Met-ENK, could be one of the Treg-derived factors responsible for wound healing. Mice deleted with the δ-opioid receptor, Met-ENK receptor, show delayed wound healing (40). Therefore, PENK produced from UVB-skin Treg cells can also contribute to wound healing in vivo as well. Since Met-ENK has an effect on pain, it is also possible that UVB-expanded Treg cells may provide a healing effect on pain in the skin. It is reported that brain Treg cells express Penk and Htr7, encoding serotonin receptor 7 (19). Although serotonin receptor 7 is shown to be important for brain Treg cell expansion, the expression and function of PENK in brain Treg is still unknown. Htr7 was not highly expressed in UVB-skin Treg cells (Fig. 1D and SI Appendix, Figs. S3 and S4); therefore, expression of PENK might be a special feature of UVB-skin Treg cells.

Second, UVB-skin Treg cells possess a unique TCR repertoire (Fig. 1B). Compared to Treg cells from VAT, muscle, and brain (17–19), the expanded clones in UVB-skin Treg cells were different. It has been recently reported that myosin-specific Treg cells mediate cardioprotection in myocardial infarction (41). Similarly, it is possible that upon UVB irradiation, skin Treg cells recognize unique antigens released from UVB-damaged skin presented by CD11b-type dermal DCs (16). Further investigation is required to identify the antigens recognized by UVB-skin Treg cells.

Third, in contrast to the requirement of the ST2 for Treg expansion in intestine, muscle, VAT, and brain (19, 24–26, 42), the ST2–IL-33 axis is not essential for Treg expansion in the UVB-irradiated skin, even though ST2 is expressed abundantly on the surface of UVB-skin Treg cells. Therefore, UVB-skin Treg cells are different not only in gene expression, but also in mechanisms of their proliferation and survival. The proliferation of Treg cells in the skin can also be affected by skin commensals (13). However, it is unlikely, because mice were kept in the same specific pathogen-free (SPF) condition for a week after UVB irradiation in which the original commensals should recover. Rather, it is possible that Treg cells proliferate in response to self-antigens released from UVB-irradiated skin since the expansion of UVB-skin Treg cells requires antigen-presenting DCs (16). Consistent with this idea, UVB-skin Treg cells showed a CD44^highCD62L^low activated phenotype (SI Appendix, Fig. S10B) and produced AREG without additional TCR stimulation in vitro (Fig. 1 G, Top), indicating that their TCRs have already been stimulated with cognitive antigens in the UVB-exposed skin. In contrast, there were only a few AREG⁺ Treg cells both in normal LNs and UVB-LNs without TCR stimulation in vitro (Fig. 1 G, Bottom), but upon additional TCR stimulation, AREG was produced in most UVB-LN Treg cells (Fig. 2 D). Thus, TCRs of UVB-LN Treg cells may not be stimulated enough in UVB-irradiated mice, which could contribute to the different gene expression between UVB-skin Treg and UVB-LN Treg cells (SI Appendix, Fig. S3). It is also reported that Treg frequency in the skin is correlated to hair cycles (15), which are synchronized with mice age (43). We used 7-wk-old B6 mice whose hair cycles are synchronized in the end of anagen (43). Therefore, it is unlikely that hair cycles influence the results. The precise mechanism by which UVB expands skin Treg cells with a healing function remains to be determined.

Our findings imply that UVB exposure plays a beneficial role in wound healing. A well-known positive effect of UVB is immunosuppression. It is shown that UVB-induced immunosuppression is mediated by Treg cells in mouse models of contact hypersensitivity (44, 45), experimental autoimmune encephalomyelitis, and atherosclerosis (46, 47). UVB therapy is very effective for treatment of skin immunological disorders such as psoriasis and atopic dermatitis (48). In addition, here we found that UVB supports wound healing via expansion of skin Treg cells. The wound healing process is mediated by keratinocyte migration and proliferation, which is regulated by multiple factors including growth factors and cytokines (49). Our data indicated that UVB-skin Treg cells may promote wound healing directly by producing PENK and AREG (Figs. 1 and 2), and also indirectly by suppressing skin inflammation (Fig. 3 D–F and SI Appendix, Fig. S14). AREG from Treg cells has already been reported to have an effect on tissue repair in the mouse model of influenza infection and brain infarction (19, 21). Therefore, it is possible that AREG from UVB-skin Treg cells similarly contributes to wound healing in vivo since UVB-skin Treg cells express AREG without additional TCR stimulation in vitro (Fig. 1 G, Top). Our UVB irradiation protocol, a single exposure of 500 mJ/cm², which is 2× minimal erythema dose (MED) to B6 mice, induced skin inflammation (SI Appendix, Fig. S14) and a lesser dose did not expand skin Treg cells (14). A higher dose of UVB was used in other reports to induce Treg cells (50, 51). Therefore, skin Treg cells may be expanded to suppress the UVB-induced skin inflammation, and this suppression mediated by the expanded Treg cells helps promote the wound healing as well. In other words, some degree of inflammation may be necessary to expand Treg cells in the UVB-irradiated skin. Likewise, Treg cells suppress imiquimod-induced psoriatic skin inflammation in mice (52, 53). Psoriasis is a chronic skin inflammation caused by IL-23/IL-17 axis-induced keratinocyte hypergrowth (54). In the UVB-treated human skin from the database, one of the enriched genes was a gene set related to wound healing (SI Appendix, Fig. S17). As a possible mechanism for the effective therapy in psoriasis, UVB irradiation may tune the healing process, i.e., refraining keratinocyte psoriatic hypergrowth. Wound healing is also controlled by many factors, so it is possible other variables are involved as well. Further investigation will be required to determine whether UVB therapy could contribute to promoting would healing in humans.

In conclusion, this study highlights a function of UVB-skin Treg cells. UVB expands skin Treg cells expressing PENK and AREG, which support wound healing in the skin. Our data might add an insight to the role of UVB therapy, which is the expansion of Treg cells with a healing effect. UVB is efficiently provided from daily sunlight. It may be important to be exposed to sunlight in a proper way to maintain skin homeostasis and healing ability.

Materials and Methods

Mice.

C57BL/6J (B6) mice (7 wk of age) were purchased from Japan SLC, Inc. or CLEA Japan, Inc. Foxp3^DTR B6 background mice were from The Jackson Laboratory (38). ST2 KO B6 background mice were kindly provided by Koubun Yasuda and Tomohiro Yoshimoto, Hyogo College of Medicine, Hyogo, Japan (23). Mice were maintained at the Nagoya City University animal facility. All mice were kept under the same SPF conditions. The institutional animal care and use committees of Nagoya City University approved this study (approval no. H29M-01).

Reagents.

Details are presented in SI Appendix, Materials and Methods.

UVB Irradiation.

UVB was produced by a FL20S-E-30 lamp (TOSHIBA), which emits most of its energy within the UVB range (290 to 320 nm; emission peak 305 nm). Seven-week-old mice were shaved and exposed to UVB at 500 mJ/cm² as previously reported (14, 45). A single exposure of UVB at 500 mJ/cm² is 2× MED in B6 mice per our irradiation protocol. We previously published that a single dose of 150 mJ/cm² did not expand Treg cells in the skin and that multiple UVB exposures of 150 mJ/cm² daily on 4 consecutive days expanded skin Treg cells similarly to a single UVB exposure of 500 mJ/cm² (14).

Treg Depletion in Foxp3^DTR Mice.

Details are presented in SI Appendix, Materials and Methods.

Cell Isolation.

Ear skin was mechanically disrupted with scissors or incubated with Liberase TL and DNase I, followed by digestion with a gentleMACS dissociator (Miltenyi Biotec), as reported previously (14, 16). Inguinal and axillary lymph nodes were mixed and used as skin-draining LNs (14). Cells from the small intestine, such as intraepithelial lymphocytes and lamina propria, were prepared using a lamina propria dissociation kit (Miltenyi) according to the manufacturer’s protocol.

Flow Cytometry.

Single-cell suspensions were incubated with anti-CD16/32 Ab (93) to block Fc receptors and stained with Abs specific for CD45.2 (104), CD4 (RM4-5), CD11b (M1/70), F4/80 (BM8), Ly6C (HK1.4), I-A/I-E (M5/114.15.2), ICOS (7E.17G9), KLRG1 (2F1), CD44 (IM7), ST2 (DIH9), CD25 (PC61) from Biolegend or eBioscience and Live/Dead Fixable aqua from Thermo Fisher Scientific. Abs specific for Foxp3 (FJK-16s), T-bet (4B10), and AREG (AF989) were used for intracellular staining with an eBioscience Foxp3/Transcription Factor Staining Buffer Set (eBioscience). For AREG staining in Fig. 1G, cells were stained right after preparation without further stimulation. In staining for AREG in Fig. 2D, cells were incubated at 37 °C in a CO₂ incubator with PMA 50 ng/mL, ionomycin 750 ng/mL, and brefeldin A 5 μg/mL for 3.5 h before staining of surface molecules with Abs.

For intracellular staining of PENK, ear skin suspensions were incubated at 37 °C in a CO₂ incubator in the presence of PMA 50 ng/mL and ionomycin 750 ng/mL for 3.5 h. Then, cells were stained with surface markers, fixed, and permeabilized using BD CytoFix/Cytoperm from BD Bioscience because the fixation/permeabilization condition for Foxp3 did not work with simultaneous staining of PENK. Fixed/permeabilized cells were stained with anti-PENK polyclonal Ab (Thermo Fisher Scientific) followed by PE-donkey anti-rabbit IgG (BioLegend). Cells were analyzed on a FACS Verse flow cytometer (BD Biosciences) and data analysis was performed using FlowJo software (FlowJo).

Ear skin was used for flow cytometry analysis. However, we previously reported that Foxp3⁺ Treg cells were also expanded in the back skin of UVB-irradiated mice by confocal microscopy (14).

Cell Sorting.

Because Foxp3 is an intracellular staining and cannot be used to sort live Treg cells, CD25^highCD4⁺ T cells from wild-type mice were sorted as Treg cells using a FACS Aria II (BD Bioscience) as previously reported (>90% Foxp3⁺) (14, 16, 55). CD25⁻CD4⁺ were sorted as Tconv cells. The dead cells were eliminated by live/dead fixable aqua or near-IR (SI Appendix, Fig. S1B, Far Right). The sorted T cells were used for TCR repertoire analysis, RNA-seq, and skin explant experiments as described below.

Quantitative (Real-Time) PCR.

Details are presented in SI Appendix, Materials and Methods.

RNA-Seq Analysis.

Details are presented in SI Appendix, Materials and Methods.

Database Analysis.

Details are presented in SI Appendix, Materials and Methods.

TCR Repertoire Analysis.

Four to six B6 mice were irradiated with UVB. After 7 d, ear skin or LNs were prepared as described in Cell Isolation. Single-cell sorting of CD25^high or CD25⁻CD4⁺CD45⁺ cells was performed using the Auto Cell Deposit Unit of FACS Aria II (BD Bioscience). Unbiased TCR repertoire analysis was performed with modifications of the procedures from refs. 56, 57. Briefly, cDNA synthesis was performed directly from single cells using SuperScript IV reverse transcriptase (Thermo Fisher Scientific). cDNA of TCRα and TCRβ genes was amplified by two round PCR using HotStarTaq Plus DNA polymerase (QIAGEN). Amplified PCR products were sequenced and searched for in the IgBLAST database (https://www.ncbi.nlm.nih.gov/igblast/).

Skin Wound Healing Assay.

Mice were anesthetized, shaved on the back, and wounds were generated in mice with or without UVB irradiation. UVB irradiation was performed 6 d before wounding. Other details are presented in SI Appendix, Materials and Methods.

Measurements of Cortisol.

Details are presented in SI Appendix, Materials and Methods.

Skin Explant Assay.

Skin explant assay was performed as previously reported with some modification (31). The skin explants were cultured in conditions that keratinocytes normally grow in 2 d. Ear skin of UVB-treated or untreated mice was excised using 3-mm punch biopsy. Skin explants were placed into wells of collagen I-coated 24 or 48-well plates (Corning), dermis side down. After 3 to 5 min, 200 µL or 70 µL of Dulbecco’s modified eagle medium (DMEM)/F12-DMEM mixture (1:1) (Thermo Fisher Scientific, GIBCO) containing heat-inactivated 10% fetal bovine serum (FBS) (HyClone, GE Healthcare Life Science), 10 ng/mL epidermal growth factor, 2 nM triiodothyronine, 5 µg/mL apo-transferrin, 5 µg/mL insulin, 0.4 µM hydrocortisone, penicillin-streptomycin, and gentamicin, 10 ng/mL cholera toxin were added (31). After a 2-d incubation at 37 °C, 5% CO₂, skin explants were photographed using a Keyence microscope with ×4 or ×20 objective lenses (BZ-X800, Keyence). Sometimes, skin explants were washed with phosphate buffered saline (PBS) and fixed with 4% paraformaldehyde-phosphate buffer (Wako) for 10 min and kept at 4 °C until taking photographs. Keratinocyte outgrowth length from the edge of skin was measured at the longest part of each sample with ImageJ software (ImageJ Ver. 1.51) or Photoshop CC (Adobe Inc.) in a blind manner, in which researchers who did not perform the explant experiments analyzed the pictures without any information of the indicated pictures. The area of keratinocyte outgrowth was measured with Photoshop CC in a blind manner. If the skin explants had technical problems such as parts lifted up during the culture, they were eliminated for the measuring area of keratinocyte outgrowth.

To examine the effect of following reagents, recombinant AREG (100 ng/mL), antigen affinity-purified polyclonal goat anti-mouse AREG Ab (R&D Systems) (19) at 10 μg/mL, or naltrindole at 5 μM were added on day 0.

To investigate the effect of Treg cells, CD25^high or CD25⁻CD4⁺CD45⁺ cells were purified from LNs from day 7 UVB-irradiated mice as described in Cell Sorting. CD25^high or CD25⁻CD4⁺CD45⁺ cells (0.4 to 2 × 10⁵/well) were added into the skin explant from normal, naïve mice on day 0 of skin explant with Dynabeads Mouse T-Activator CD3/CD28 at a T-bead ratio of 1:1, plus human IL-2 100 U/mL in a 48-well plate. Controls were no T cells, no IL-2, and anti-CD3/CD28 beads.

Immunofluorescence Microscopy for the Epidermal Tongue Analysis.

Details are presented in SI Appendix, Materials and Methods.

Statistics.

All statistics were analyzed with GraphPad Prism 7 or 8. All numerical data were summarized using mean ± SD (SD). P values were determined using Student’s t test unless indicated in each figure legend. P values <0.05 were considered statistically significant. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. N.S. indicates not significant.

Data and Materials Availability.

All RNA-seq data have been deposited in the DDBJ databases with an accession ID DRA009184 (https://www.ddbj.nig.ac.jp/index-e.html). All TCR repertoire sequences are in SI Datasets S1.