Normality sensing licenses local T cells for innate-like tissue surveillance

The increasing implication of lymphocytes in general physiology and immune surveillance outside of infection poses the question of how their antigen receptors might be involved. Here, we show that macromolecular aggregates of intraepidermal γδ T cell antigen receptors (TCRs) in the mouse skin aligned with and depended on Skint1, a butyrophilin-like (BTNL) protein expressed by differentiated keratinocytes (KCs) at steady state. Interruption of TCR-mediated ‘normality sensing’ had no impact on γδ T cell numbers but altered their signature phenotype, while the epidermal barrier function was compromised. In addition to the regulation of steady-state physiology, normality sensing licensed intraepidermal T cells to respond rapidly to subsequent tissue perturbation by using innate tumor necrosis factor (TNF) superfamily receptors. Thus, interfering with Skint1-dependent interactions between local γδ T cells and KCs at steady state increased the susceptibility to ultraviolet B radiation (UVR)-induced DNA damage and inflammation, two cancer-disposing factors.

T he steady-state immune system was long thought to be resting, awakened only by infection. However, this perspective has been transformed by examples of immune responses to non-microbial challenges and evidence that the steady-state immune system helps maintain physiological processes from metabolism to memory and barrier integrity [1][2][3][4] . Many such interactions happen locally, involving immune cells permanently associated with non-lymphoid tissues [5][6][7] . While macrophages are well-established regulators of tissue status, an increasing implication of lymphocytes in this process raises the question of whether antigen receptors are used to sense tissue perturbation, akin to recognition of foreign antigens.
Whereas Skint1 and Skint2 mediate V γ 5V δ 1 + precursor development in the thymus, they are also expressed by suprabasal KCs, among which mature V γ 5V δ 1 + IELs permanently reside 17 . How they might continue to influence mature IEL function is unknown. In that regard, focal clusters of V γ 5V δ 1 + TCRs constitutively engage KCs at steady state and may engage in signaling 22 , but neither the epithelial determinant(s) nor the biological implications of these engagements is understood.
Here, we focused on V γ 5V δ 1 + IELs, also termed dendritic epidermal T cells (DETCs), to investigate how tissue T cells might use TCRs to distinguish normal self from stressed self. We show that constitutive DETC TCR-containing foci 22 align with and depend upon sustained epithelial expression of Skint1. Although these interactions did not maintain DETC numbers, they preserved DETC and KC gene expression patterns and maintained steady-state epithelial barrier function. Additionally, these interactions licensed DETCs to utilize costimulatory receptors to make rapid responses to UVR and chemical irritation. Thus, interruption of steady-state, TCR-mediated DETC-KC interactions increased epidermal susceptibility to sustained DNA damage and inflammation, two factors that jointly dispose to cancer 23 . As Skint1-dependent DETC-KC interactions occur at steady state, this mode of tissue immunosurveillance can be viewed as 'normality sensing' . This offers a revised perspective of tissue immunosurveillance and of antigen receptor biology, which conventionally focuses on the recognition of molecules induced by change.

Results
DETCs sense steady-state KCs via Skint1. At steady state, V γ 5V δ 1 + TCRs focus at apical dendrite tips in supramolecular clusters that contact neighboring suprabasal KCs, seemingly inducing constitutive TCR signaling 22 . TCRs within these clusters colocalized with actin ( Fig. 1a), as reported 22 , and most DETCs in FVB mice displayed one to four clusters (Fig. 1a). For clarity, we displayed F-actin staining masked by DETC bodies, although examination of raw staining confirmed these observations (Extended Data Fig. 1a). We next examined FVB.Tac mice, which express a truncated non-functional Skint1, and Trgv5 -/-Trdv4 -/mice. These two strains lack canonical V γ 5V δ 1 + DETCs 12,16 . Relative to wild-type mice, both strains showed increased frequencies of DETCs that lacked TCR-actin colocalization ( Fig. 1a and Extended Data Fig. 1a). Transgenic expression of wild-type Skint1 in FVB.Tac mice returned TCR-actin colocalization to wild-type levels (Fig. 1a). Because DETCs in FVB. Tac and  The increasing implication of lymphocytes in general physiology and immune surveillance outside of infection poses the question of how their antigen receptors might be involved. Here, we show that macromolecular aggregates of intraepidermal γδ T cell antigen receptors (TCRs) in the mouse skin aligned with and depended on Skint1, a butyrophilin-like (BTNL) protein expressed by differentiated keratinocytes (KCs) at steady state. Interruption of TCR-mediated 'normality sensing' had no impact on γδ T cell numbers but altered their signature phenotype, while the epidermal barrier function was compromised.

In addition to the regulation of steady-state physiology, normality sensing licensed intraepidermal T cells to respond rapidly to subsequent tissue perturbation by using innate tumor necrosis factor (TNF) superfamily receptors. Thus, interfering with Skint1-dependent interactions between local γδ T cells and KCs at steady state increased the susceptibility to ultraviolet B radiation (UVR)-induced DNA damage and inflammation, two cancer-disposing factors.
Trgv5 -/-Trdv4 -/mice use non-V γ 5V δ 1 + TCRs, these results confirm reports that juxtaepithelial TCR foci require the V γ 5V δ 1 TCR 22 .
To investigate whether Skint1 protein acutely maintained TCR-KC interactions, we performed an i.d. administration of 2G2, a monoclonal antibody specific for Skint1 (ref. 19 ), to wild-type FVB mice. Within 6 h after injection, the frequency of TCR-actin foci was significantly reduced compared to mice administered an isotype control antibody (Fig. 1b). A similar impact was seen in C57BL/6 wild-type mice (Extended Data Fig. 1c). We next applied iSIM to steady-state epidermal sheets from wild-type mice using Skint1 antibody staining. Despite Skint1 expression being restricted to KCs 17 , we only detected Skint1 protein in clusters nucleated by the TCR (Fig. 1c and Extended Data Fig. 1d). Compared to wild-type mice, TCR-proximal Skint1 antibody staining was significantly reduced in FVB.Tac and Tcrd -/mice (Extended Data Fig. 1e). While truncated Skint1 in FVB.Tac mice could affect antibody staining, Tcrd -/mice expressed Skint1 mRNA at relatively normal levels in the epidermis (Extended Data Fig. 1f), indicating that wild-type Skint1 protein was not nucleated by TCRαβ + cells in the Tcrd -/epidermis 25 . Thus, Skint1 maintained steady-state TCR-mediated interactions of DETCs with KCs, which in turn shaped the epidermal localization of Skint1.

Skint1 promotes DETC maintenance of epidermal integrity.
To analyze the impact of sustained Skint1 disruption in situ, we performed an i.d. injection of Skint1 antibody or isotype control weekly into the ears of FVB mice for 10 weeks. At this time point, mice displayed comparable numbers and frequencies of V γ 5V δ 1 + DETCs and major histocompatibility complex class II + (MHCII + ) Langerhans cells (LCs) ( Fig. 2a and Extended Data Fig. 2a-c), indicating that long-term DETC population maintenance was Skint1 independent. However, relative to isotype-treated or contralateral ears, Skint1 antibody-treated ears became moderately inflamed (Fig. 2b,c), and CD5 + αβ T cells infiltrated the epidermis (Fig. 2d). Similar increased ear thickness, albeit milder, was observed in control-treated Tcrd -/mice (Extended Data Fig. 2d), as previously reported 26 . Sustained Skint1 antibody treatment did not further increase ear thickness in Tcrd -/mice (Extended Data Fig. 2d,e), indicating that Skint1 antibody treatment primarily interfered with the Skint1-TCRγδ interaction.
We additionally observed increased transepidermal water loss (TEWL) at week 5 of Skint1 antibody treatment relative to isotype treatment (Fig. 2e) or contralateral ears (Extended Data Fig. 2f). Increased baseline TEWL measurements were also detected in isotype-treated Tcrd -/mice, as reported in ref. 4 , and those were also unaffected by Skint1 antibody treatment (Fig. 2e). C57BL/6 mice lacking Skint1 or all Skint genes also had elevated baseline TEWL compared to littermate controls (Extended Data Fig. 2g). Compared to isotype-treated mice, Skint1 antibody-treated mice consistently displayed decreased epidermal expression of the DETC-derived barrier maintenance cytokine Il13 (ref. 27 ) and of KC genes involved in epidermal barrier formation (Tgm1 and Lor; Fig. 2f,g), as did baseline Tcrd -/mice relative to wild-type controls (Extended Data Fig. 2h,i). KCs also displayed increased apoptosis following Skint1 antibody treatment (Fig. 2h). While keratin gene expression was unchanged, indicating normal KC differentiation, epidermal Skint1 and Skint2 expression was upregulated following Skint1 antibody treatment (Extended Data Fig. 2j), perhaps reflecting a compensatory response to Skint1-TCR disruption. Although LC abundance was unchanged ( Fig. 2a and Extended Data Fig. 2a,c), Skint1 antibody treatment marginally increased the frequency of activated MHCII hi CD86 + LCs (Fig. 2i), perhaps reflecting ongoing epidermal inflammation. Most DETCs contact LCs at steady state, and LCs may also contribute to epidermal integrity 28,29 . However, we did not observe differences in baseline TEWL or barrier gene expression in LC-deficient human Langerin (huLangerin)-DTA mice relative to wild-type mice (Extended Data Fig. 2k,l). Thus, the functional contributions of DETCs to effective epidermal integrity were compromised by Skint1 blockade and were LC independent.

Skint1 sensing maintains DETC immunosurveillance signatures.
To investigate whether the contribution of Skint1 to skin physiology reflected downstream consequences of a primary impact on DETCs, we conducted single-cell RNA sequencing of CD45 + MHCII − EpCAMepidermal lymphocytes from wild-type FVB mice 7 d after Skint1 antibody or isotype treatment. Non-DETC populations, including αβ T cells and contaminating dermal γδ T cells, were excluded. Four DETC clusters (cluster 0cluster 3) were identified by uniform manifold projection (UMAP) of the integrated dataset (Fig. 3a), indicating that DETCs were heterogeneous, despite their near uniform anatomical localization and TCR expression. Cluster 3 appeared segregated in two dimensions ( Fig. 3a), an artifact of dimensionality reduction that could not be resolved by adjusting algorithm parameters.
Cluster 0 was enriched in transcripts related to AP1-mediated cell regulation (Fos, Junb and Jund) and the epithelia-tropic cytokines Areg (amphiregulin) and Il13 ( Fig. 3b and Extended Data Fig. 3a-c), implying a potential to regulate KCs and other immunocytes 27,[30][31][32] . Relative to other clusters, cluster 1 appeared more quiescent, with reduced expression of gene sets associated with T cell activation, immune effector processes and regulation of cell adhesion and apoptosis (Extended Data Fig. 3b,c). Nonetheless, cluster 1 was enriched in transcripts encoding galectins, S100 proteins and the aryl hydrocarbon receptor (Ahr) (Fig. 3b), of which the latter Tac mice (n = 12) are shown in a; scale bar, 20 μm. Interacting dendrites from wild-type FVB mice administered Skint1 antibody (n = 10) or isotype control (n = 9) by intradermal (i.d.) injection 6 h before are shown in b; scale bar, 20 μm. c, Instant structured illumination microscopy (iSIM) of whole-mount epidermal sheets from unchallenged wild-type FVB mice. Left, V γ 5 TCR (green) staining; scale bar, 10 μm. Right, top-down and lateral views of three-dimensional (3D) reconstructions of typical V γ 5 TCR (green) and Skint1 (magenta) staining in selected dendrite tips from the left-hand image; scale bar, 1 μm. Representative images and pooled data are from eight (a) or three (b and c) independent experiments. A matched two-way analysis of variance (ANOVA) with Dunnett's multiple comparisons test relative to wild-type (a) or a Sidak's multiple comparisons test (b) was used to analyze the data. Values in pie charts represent mean ± s.d.; NS, not significant. is essential for DETC population self-renewal 33 . Germane to that, cluster 2 was enriched in transcripts associated with proliferation ( Fig. 3b and Extended Data Fig. 3c,d). Cluster 3, while small ( Fig. 3a), was enriched in mRNAs ascribed to lymphocyte activation and costimulation, such as Cd82, Ramp1 and Tnfrsf9 (4-1BB), as well as effector molecules Xcl1, Ctsw, Gzmc (granzyme C) and Il13 (Fig. 3b,c and Extended Data Fig. 3b,c,e). 4-1BB is conventionally viewed as a costimulatory receptor expressed by recently activated T cells, but it is also induced by Skint1-and Btnl1-mediated selection of DETCs and gut V γ 7 + IEL progenitors, respectively 15  , an unpaired two-sided t-test (c, f-i), an unpaired two-sided t-test with Welch's correction (d) or an ordinary two-way ANOVA with Tukey's multiple comparisons test (e) was used to analyze the data. Error bars represent mean ± s.d.
flow cytometry sorting indicated that 4-1BB hi DETCs expressed more Gzmc and Xcl1 than 4-1BB lo DETCs (Extended Data Fig. 3f). In Nur77-GFP reporter mice, which report calcium-dependent signaling following T cell activation 36 , GFP expression was significantly higher in 4-1BB hi DETCs at steady state (Fig. 3d). These results suggested that cluster 3 might represent a subset of DETCs actively responding to minor skin perturbations, such as grooming or scratching, which occur episodically in mice at steady state. Comparison of DETCs from isotype-and Skint1 antibody-treated mice indicated that cluster 3 was selectively diminished by Skint1 antibody treatment (Fig. 3e,f), which was reflected by a loss of overall Xcl1 and Gzmc expression (Fig. 3g,h) and validated by qPCR of DETCs sorted from Skint1 antibody-and isotype-treated mice (Extended Data Fig. 3g). Consistently, 4-1BB hi DETCs were significantly reduced by Skint1 antibody treatment (Fig. 3i), and the remaining 4-1BB hi DETCs expressed less Nur77-GFP (Fig. 3d). Thus, sustained activation of a small subset of DETCs (cluster 3) requires Skint1 sensing at steady state.
We also observed broader changes in gene expression in DETCs following Skint1 antibody treatment, regardless of clusters. Ptpn6 (SHP-1 phosphatase) and overall Xcl1 expression decreased, and Rps27, Rps29 and Egr1 expression increased following Skint1 antibody treatment ( Fig. 3h and Extended Data Fig. 3h), as validated by qPCR (Extended Data Fig. 3g). This indicated that the intrinsic activation state of most DETCs depended on sustained Skint1-dependent DETC-KC interactions. Peripheral Skint1-dependent regulation of Ptpn6 expression in most DETCs (Extended Data Fig. 3i) was reminiscent of the upregulation of Ptpn6 during Skint1-dependent DETC precursor selection (Supplementary Table 1). Similarly, Skint1 antibody treatment moderately reduced DETC expression of CD122 and CD45RB proteins (Extended Data Fig. 3j), which are also upregulated during BTNL-dependent IEL selection 15,16 . Skint1 antibody-treated wild-type DETCs resembled non-V γ 5V δ 1 + DETCs in FVB.Tac mice regarding CD45RB expression (Extended Data Fig. 3k). GSEA indicated that the collective gene signature of thymic Skint1-driven DETC progenitor selection 35 was downregulated in mature DETC by Skint1 antibody treatment ( Fig. 3j and Supplementary Table 1). 4-1BB hi DETC frequency and CD122 expression were unaltered in huLangerin-DTA mice (Extended Data Fig. 3l), suggesting that Skint1 regulation of DETCs was direct and LC independent. Thus, steady-state epidermal Skint1 promoted immunosurveillance by maintaining the predetermined effector phenotype in most DETCs and by underpinning the activation of a small subset of DETCs.

Skint1 sensing drives DETC responses to UVR.
To further address the contribution of Skint1 to immunosurveillance, we irradiated the ears of C57BL/6 mice with UVR to induce DNA damage. Within 1 d, many KCs exhibited signs of DNA damage and repair, as measured by γH2A.X foci (Fig. 4a). Simultaneously, DETCs became rounded (Fig. 4a), consistent with reports of their response to diverse skin perturbations, including UVR [37][38][39] . We also observed a substantial diminution in juxtaepithelial TCR focusing (Fig. 4a) and Skint1 and Skint2 downregulation in KCs relative to unchallenged mice (Extended Data Fig. 4a). Concurrently, LCs and DETCs displayed upregulation of surface activation markers (Extended Data Fig. 4b,c).
To test whether Skint1-dependent TCR-KC interactions established and/or maintained the preparedness of DETCs to respond to UVR, Skint1 antibody was injected i.d. into wild-type C57BL/6 mice 1 d before UVR, and DETCs were sorted and analyzed by qPCR 24 h after UVR. The UVR-induced upregulation of Gzmb, Gzmf and Cxcl2 was significantly reduced in the Skint1 antibody-treated mice, while UVR-induced inhibitory receptor genes, such as Havcr2 and Tigit, remained unaffected, and Il13 was atypically upregulated compared to isotype-treated mice ( Fig. 4e and Extended Data Fig.  4e). In addition, the UVR-induced increase in 4-1BB and CD122 expression was significantly inhibited by prior Skint1 antibody administration (Fig. 4f), while LC activation was unperturbed (Extended Data Fig. 4f). In Nur77-GFP reporter mice, DETCs displayed a basal level of GFP, which increased incrementally following UVR (Fig. 4g). Skint1 antibody treatment not only diminished the basal Nur77-GFP signal in DETCs but substantially suppressed the UVR-induced increase in Nur77-GFP expression (Fig. 4h). These observations were reproduced ex vivo in an explant model, in which DETCs in Nur77-GFP ear sheets preincubated with Skint1 antibody showed impaired upregulation of Nur77-GFP expression following in vitro exposure to UVR relative to isotype control (Extended Data Fig. 4g), indicating that Skint1 antibody acted locally. Pretreatment of ex vivo ear sheets with the Src-family kinase inhibitor PP2 produced a similar outcome (Extended Data Fig. 4h). Collectively, these observations indicate that Skint1-dependent interactions licensed DETC stress responsiveness and effector functions.

Skint1 licenses DETC responsiveness to UVR via TNF receptors.
TCR signaling amplification is a defining property of costimulatory of DETCs from wild-type FVB mice administered Skint1 antibody or isotype control by i.d. injection 1 week before (n = 1 per group). b, Top 10 most significantly enriched genes per cluster in DETCs from the combined dataset, as in a. c, Expression of key cluster 3 marker genes in the combined dataset, as in a. d, Flow cytometry expression of green fluorescent protein (GFP) in DETCs segregated by 4-1BB expression from Nur77-GFP mice treated 2 d prior with Skint1 antibody or isotype control by i.d. injection (n = 5 per group); gMFI, geometric mean fluorescence intensity. e, UMAP of DETCs comparing treatments, as in a. f, Contribution of DETCs from Skint1 antibody-and isotype-treated samples to cluster composition, as in a. g, Expression of Xcl1 in DETCs, comparing Skint1 antibody and isotype treatments, as in a. h, Differential gene expression in combined DETC clusters comparing Skint1 antibody and isotype treatments, as in a; FMO, fluorescence minus one i, Frequency of 4-1BB hi DETCs by flow cytometry in wild-type FVB mice treated with Skint1 antibody or isotype control by i.d. injection 1 week prior (n = 10 per group). j, Gene set enrichment analysis (GSEA) of combined DETC clusters comparing Skint1 antibody and isotype treatments, as in a, using a Skint1 selection gene signature from DETC precursors isolated from wild-type or FVB. Tac embryonic thymi; NES, normalized enrichment score; FDR, false discovery rate. Data are representative of three independent experiments (d) or are pooled from two independent experiments (i). A Wilcoxon test with Bonferroni's correction (b and h), paired one-way ANOVA with Sidak's multiple comparisons test (d) or unpaired two-sided t-test (i) was used to analyze the data. Error bars represent mean ± s.d. molecules, including those of the TNF receptor superfamily. To investigate whether steady-state Skint1-dependent interactions licensed DETCs to respond rapidly through co-receptors such as 4-1BB, we performed i.d. injections of an agonistic 4-1BB antibody or isotype control in the ears of Nur77-GFP mice. Akin to UVR, 4-1BB antibody drove a coordinated upregulation of CD69 and Nur77-GFP by a subset of DETCs relative to isotype control (Fig. 5a). This effect was likely direct, as epidermal 4-1BB expression was exclusive to DETCs (Extended Data Fig. 5a,b); however, i.d. injection with Skint1 antibody 24 h before 4-1BB stimulation ablated CD69 and Nur77-GFP upregulation in DETCs (Fig. 5a). Additionally, 4-1BB antibody injection induced expression of DETC effector genes related to cluster 3 (Xcl1 and Gzmc) and UVR responses (Gzmb and Gzmf), which was similarly inhibited by prior Skint1 antibody treatment (Fig. 5b). These observations indicated that steady-state TCR-Skint1 interactions with KCs licensed DETCs to respond to 4-1BB, inducing qualitative changes that partially phenocopied the UVR response. Therefore, to test whether 4-1BB was an obligate contributor to UVR responses, we performed an i.d. injection of an antibody that blocks 4-1BB ligand (4-1BBL) binding to 4-1BB into the ears of wild-type C57BL/6 mice before UVR treatment. This antibody partially, but significantly, blocked the UVR-induced increase in epidermal expression of Gzmb and Gzmf relative to isotype control (Fig. 5c). Consistent with this, UVR-induced upregulation of Gzmb and Gzmf was also inhibited in 4-1BBL-deficient Tnfsf9 -/mice relative to wild-type controls (Extended Data Fig. 5c).
Similar to 4-1BB, expression of Tnfrsf18 (which encodes GITR) was also induced by Skint1 during thymic DETC progenitor selection (Supplementary Table 2). Unlike 4-1BB, Tnfrsf18 was not enriched in cluster 3 but was expressed by most DETCs (Extended Data Fig. 5d). Additionally, GITR mRNA and protein were expressed by KCs and a small fraction of LCs (Extended Data Fig.  5a,b). Regardless, i.d. injection of an agonistic GITR antibody also upregulated Nur77-GFP and CD69 expression by DETCs, which was almost entirely Skint1 dependent (Fig. 5d). Furthermore, antagonist antibodies that blocked GITR ligand (GITRL) interactions with GITR partially but significantly inhibited the UVR-induced upregulation of epidermal Gzmb and Gzmf (Fig. 5e). Thus, signaling through TNF receptors was one non-redundant component of the rapid DETC response to skin irradiation and was dependent on prior licensing by Skint1-TCR interactions.
Next, we asked whether other conduits of innate T cell activation required licensing via Skint1-TCR interactions. Extracellular ATP is an alarmin implicated in DETC responses to UVR together with interleukin-1 (IL-1) 37 . Administration (i.d.) of ATP alone induced Nur77-GFP upregulation in DETCs, which was Skint1 dependent (Fig. 5f). Thus, steady-state Skint1 sensing regulated the ability of DETCs to respond to diverse stimuli.

Skint1-licensed immunosurveillance promotes recovery from epithelial damage.
To investigate the consequences of Skint1 blockade for host physiology, we performed an i.d. injection of Skint1 antibody or isotype control 24 h before and 48 h after UVR. Mutagenic cyclobutane pyrimidine dimers (CPDs) were rapidly induced by UVR in all mice but were removed less efficiently in Skint1 antibody-treated mice than in isotype control-treated mice 72 h after UVR (Fig. 6a). Likewise, higher numbers of γH2A.X foci, more ear swelling and greater epidermal thickening were detected in Skint1 antibody-treated mice (Fig. 6b-d and Extended Data Fig.  6a). Exacerbated ear swelling was similarly observed after UVR in C57BL/6 mice deficient in γδ T cells, Skint1 and the Skint locus (Extended Data Fig. 6b-d), providing genetic evidence that the pathophysiologic impacts of acute Skint1 blockade reflected consequences of disrupted Skint1-dependent TCR-KC interactions. Finally, UVR-associated defects in barrier function, as measured by toluidine blue permeation 40 , were also significantly exacerbated in Skint1 antibody-treated mice (Fig. 6e).
Skin is commonly exposed to non-mutagenic chemical irritants. To investigate whether Skint1 licensing applied in contexts other than DNA damage, we examined a model of irritant contact dermatitis by using a single topical dose of the hapten dinitrofluorobenzene (DNFB). DETCs upregulated Gzmb and Gzmf within 48 h after DNFB application, but this was largely inhibited by prior Skint1 antibody treatment (Fig. 7a). Similarly, Skint1 blockade suppressed the DNFB-dependent induction of Nur77-GFP in DETCs (Fig. 7b). Moreover, acute DNFB-induced dermatitis was amplified in mice treated with Skint1 antibody (Fig. 7c). Thus, Skint1 licensed prompt DETC effector responses in settings of mutagenic and non-mutagenic epidermal stress, naturally limiting inflammatory lesions and impairments in functional barrier integrity.

Discussion
Here we showed that steady-state V γ 5V δ 1 + TCR-mediated interactions between DETCs and KCs involve and depend on Skint1. This crosstalk did not obviously regulate the tissue residency of DETCs nor was it required for DETCs to populate the epidermis. However, it was essential for DETC function, as disruption of Skint1 sensing impeded DETC maintenance of epidermal homeostasis, interfered with steady-state immunosurveillance signatures, and compromised efficient DETC responsiveness to epithelial stresses. Indeed, TCR-dependent, Skint1-dependent interactions with KCs licensed DETCs for innate-like responsiveness through TNF receptors, which limited tissue pathology following insults that included UVR and a chemical irritant.
Because Skint1 is expressed by differentiated suprabasal epithelial cells and is stress-labile, DETC antigen receptors sense normality. This contrasts with the conventional consideration that TCRs engage products of change, such as microbial-or cancer-associated neoantigens and stress antigens upregulated by cellular dysregulation. It is possible that the DETC TCR also transduces signals from an unelucidated stress-induced ligand distinct from Skint1 (ref. 41 ). Nonetheless, stress sensing by DETCs depended largely on innate receptors 4-1BB and GITR, drawing parallels to NKG2D-driven VST, variance-stabilizing transformation. c, GSEA of UVR-induced gene expression changes in DETCs using signatures of steady-state DETC clusters from Fig. 3. d, qPCR for DETC gene expression in the epidermis of FVB wild-type (n = 15), Tac (n = 9), Trgv5 -/-Trdv4 -/-(n = 5) and Skint1-transgenic FVB.Tac (n = 9) mice 24 h after UVR. e, qPCR for effector genes in DETCs sorted at day 1 after UVR from wild-type C57BL/6 mice pretreated with i.d. injection of Skint1 antibody or isotype control (isotype control normal, n = 8; Skint1 antibody normal, n = 9; isotype control/Skint1 antibody UVR, n = 10 per group). f, Flow cytometry of 4-1BB and CD122 expression in DETCs from wild-type C57BL/6 mice treated as in e (n = 8 isotype control; n = 9 Skint1 antibody). g, Flow cytometry of GFP expression in DETCs following UVR of Nur77-GFP reporter mice (n = 3 per time point). h, Flow cytometry of GFP expression in DETCs from Nur77-GFP reporter mice treated as in e (n = 7 per group). Data are pooled data from two (e and f) or three (d) independent experiments or are representative of two (a) or three (h) independent experiments. A paired (d, f and h) or ordinary (e) two-way ANOVA with Sidak's multiple comparisons test was performed. Error bars represent mean ± s.d.
DETC surveillance 12 . Critically, these receptors were acquired during thymic Skint1-dependent DETC progenitor selection 35 and were sustained in the periphery by ongoing Skint1-TCR-mediated interactions. In this way, among others, steady-state TCR-mediated interactions licensed DETCs to respond rapidly to gross insults.
There are several major consequences of TCR-mediated normality sensing. First, Skint1-dependent, TCR-mediated interactions regulated barrier function and tissue homeostasis, which may occur by regulation of KC-tropic growth factor expression by DETCs. While IL-13 contributes to inflammatory pathologies 42 ,  Skint1. a,b, Flow cytometry (a) of DETC CD69 and GFP expression and qPCR (b) for DETC gene expression in the epidermis from Nur77-GFP reporter mice administered agonistic 4-1BB antibody or isotype control 24 h following Skint1 antibody or isotype control administration by i.d. injection and analyzed 6 h later (n = 4 isotype/isotype; n = 3 Skint1 antibody/isotype; n = 7 isotype/4-1BB antibody; n = 7 Skint1 antibody/4-1BB antibody). c, qPCR for DETC gene expression in the epidermis at 24 h after UVR from wild-type C57BL/6 mice pretreated with 4-1BBL antibody or isotype control by i.d. injection (n = 10 per group). d, Flow cytometry of GFP and CD69 expression in DETCs from Nur77-GFP mice administered agonistic glucocorticoid-induced TNF receptor-related protein (GITR) antibody or isotype control 24 h after i.d. injection with Skint1 antibody or isotype control and analyzed 6 h later (n = 5 isotype/isotype; n = 5 isotype/GITR antibody; n = 4 Skint1 antibody/GITR antibody). e, qPCR of DETC gene expression in the epidermis at 24 h after UVR from wild-type C57BL/6 mice pretreated with GITRL antibody or isotype control by i.d. injection (n = 10 per group). f, Flow cytometry of GFP and CD69 expression in DETCs from Nur77-GFP mice administered ATP or PBS 24 h after i.d. injection with Skint1 antibody or isotype control and analyzed 9 h later (n = 5 per group). Data are pooled from two independent experiments (a-f). An ordinary (a and b) or paired (c, e and f) two-way ANOVA with Sidak's multiple comparisons test was performed or an ordinary one-way ANOVA with Tukey's multiple comparisons test (d). Error bars represent mean ± s.d.

Fig. 5 | DeTC stress surveillance via TNF superfamily receptors is licensed by
steady-state IL-13 production by DETCs reportedly benefits epidermal integrity 27 . Although mice lacking V γ 5V δ 1 + DETCs do not suffer life-threatening barrier deficiencies, they do display cutaneous inflammation that could reflect compromised barrier function 4,16 , as we observed following sustained Skint1 blockade.
Second, Skint1 sustained the signature phenotype of mature DETCs, largely recapitulating its central role in DETC progenitor selection 16,17,24,35 . This suggests that while the effector potential of DETCs, including predetermined expression of TNFRs and other receptors, is programmed by thymic Skint1 selection, maintaining this profile requires continued peripheral Skint1 sensing. Cluster 3 DETCs, displaying hallmarks of recent activation in unchallenged mice, could represent Skint1-licensed natural immunosurveillance responses to spontaneous epithelial perturbations from scratching/grooming or mutagenesis. Intriguingly, DETC-derived XCL1 reportedly recruits cross-presenting dendritic cells (DCs) to  the epidermis 43 . Thus, XCL1 release from cluster 3 DETCs responding to episodic, low-level KC dysregulation might, in cases of infection or genotoxicity, facilitate neoantigen uptake by DCs, thereby initiating cytotoxic CD8 + T cell immunosurveillance. Third, Skint1 blockade reduced the capacity of DETCs to sense tissue perturbation through innate receptors. In the context of UVR, the prime etiologic agent of human skin cancer, we observed sustained genotoxicity and inflammatory pathology, which may in part explain the regulation of cutaneous carcinogenesis by DETCs and analogous roles fulfilled by other mouse and human γδ T cell compartments 11,44 . Additionally, normality sensing primed DETC responses to an irritant and to ATP, and it is conceivable that it might prime for wound healing and tissue regeneration. Indeed, disruption of TCR signaling in mature DETCs impaired wound healing 45 , a process reportedly modulated by Skint family members other than Skint1 and Skint2 (ref. 46 ).
Normality sensing may license invariant DETCs to respond through a panoply of innate receptors that collectively offer versatility to detect myriad manifestations of tissue perturbations with which IELs have been associated. Tissue-wide DNA damage, like that induced by UVR, resulted in overt granzyme upregulation, which may facilitate DETC killing of potentially malignant KCs 47 . Granzyme B also cleaves extracellular targets, including junctional molecules, surface receptors and extracellular matrix, whereas granzyme F is cytotoxic in vitro 48,49 . Other aspects of the DETC UVR response, including rounding up and TIM-3 upregulation, did not obviously require Skint1 licensing. Possibly, licensing regulates those traits that are inherently more dangerous if precociously expressed.
Overall, our results provide experimental evidence supporting the lymphoid stress surveillance hypothesis, according to which lymphocytes display myelomonocytic traits, such as responding rapidly through innate receptors 10 . However, the results also establish the need for prior licensing through the TCR. Whereas there is no direct DETC equivalent in humans, this concept may pertain to the selection and regulation of human intestinal innate-like V γ 4 + IELs by BTNL3 and BTNL8, a bona fide TCR-V γ ligand that is similarly expressed at steady state and is stress-labile 15,20 . This normality sensing mode of TCR stimulation may be linked to the unusually high TCR activation threshold acquired during Skint1 selection 50 , which could conceivably limit overt IEL activation in the absence of stress. In contrast to natural killer cells, which sense stress through missing self, DETC Skint1 sensing is a positive regulator of many components of tissue immunosurveillance, a role that may be phenocopied by BTNL sensing by murine and human gut IELs. The clinical implication of our findings is that epithelial pathologies may not always reflect failures to detect infectious, allergic or carcinogenic challenges but rather failings in the maintenance of normal homeostatic interactions before exposure to those challenges occur.

online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41590-021-01124-8.
Neonatal DETC reconstitution. FVB Tcrd -/and FVB.Tac Tcrd -/pups at 1-3 d of age were transferred with a thymocyte suspension prepared from E16-E18 FVB wild-type fetuses by intraperitoneal injection. The equivalent of four to five fetal thymi were transferred into each pup. Reconstituted mice were analyzed at 8 weeks of age. TEWL measurement. Under inhalation anesthesia, TEWL of the ear pinnae was measured using a Tewameter TM300 probe (Ck Electronic) under constant temperature and humidity conditions. Values were recorded as the average of ten consecutive measurements once readings were stabilized using MPA software (Ck Electronic).
UVR model. Under inhalation anesthesia, mice were positioned under a cardboard box, and one or both ears were exposed and fixed flat using Blu-Tac. UVB radiation (302 nm) was administered to dorsal ear skin for a dose of 45 mJ cm -2 (C57BL/6) or 30 mJ cm -2 (FVB) using an XX-15 Series UV bench lamp (UVP) mounted on a Perspex and plastic housing.
Ex vivo ear skin culture. Ears were split and floated on complete RPMI (10% fetal calf serum, penicillin/streptomycin, 1× non-essential amino acids, 1 mM sodium pyruvate, 10 mM HEPES and 50 μM β-mercaptoethanol) containing 10 μg ml -1 anti-Skint1 or isotype control antibodies or 20 μM PP2 (Sigma) or DMSO vehicle alone. After 4 h of culture at 37 °C, ear skin was irradiated with 45 mJ cm -2 UVB as described above and returned to 37 °C for 24 h. acquired. For staining of specific γδ TCRs, epidermal cells were rested overnight in complete RPMI at 37 °C before live/dead dye staining and were stained with 17D1 hybridoma supernatant 55 at 4 °C for 20 min. Cells were then washed and stained with surface antibodies as above, including anti-V γ 5-APC (BioLegend) and anti-rat IgM-PE (eBioscience). For analysis of other trypsin-sensitive epitopes, the epidermis was separated and digested as above but with TrypLE Express (Gibco) in place of trypsin-GNK, and the epidermis was separated for 2 h. Cells were stained as above but including combinations of the following antibodies: anti-CD122-PE, anti-4-1BB-APC and anti-4-1BB-PE (all BioLegend). Cells were acquired with an LSR II or Fortessa (BD) and analyzed with FlowJo (TreeStar). Gating strategies are detailed in Supplementary Fig. 1.
Confocal microscopy. For analysis of DETC dendrites and V γ 5 staining, ears were depilated with Nair and split, and the epidermis was separated by incubation in 0.5 M NH 4 SCN for 30 min at 37 °C and fixed in ice-cold acetone for 20 min at -20 °C.
For analysis of γH2A.X staining and for KC F-actin visualization in Extended Data Fig. 1, whole depilated split ear sheets were flattened in a histology cassette and fixed in neutral buffered formalin for 2 h at room temperature and permeabilized in 0.5% Triton X-100 for at least 1 h at room temperature.
For dendrite analysis, CD3 + cells (or V γ 5 + cells in the reconstitution experiments) were identified by the surfaces tool and used to mask F-actin staining. TCR-and F-actin-rich dendrites were identified by using the spots tool to detect 3-μm-diameter volumes containing an MFI in the top 25% of their respective channels. Colocalized CD3 and F-actin spots were selected for quantification on a per cell basis.
For γH2A.X analysis, the cell function was used to identify γ-H2A.X + nuclei by filtering for the top 20% of cells by quality. Nuclei were quantified per image and averaged over three images per mouse.
iSIM. Epidermal sheets were prepared, separated and fixed in acetone as above and blocked in 5% normal goat serum for 1 h at room temperature. Sheets were stained with anti-Skint1 for 1 h at 37 °C, washed and stained with anti-rat IgG AF568 (Invitrogen), anti-CD3-FITC, anti-V γ 5-FITC (all BD) and anti-TCRδ-AF647 (BioLegend). iSIM was performed using a VT-iSIM microscope (VisiTech International) with a ×100/1.45-NA objective. Raw images were processed with the Richardson-Lucy deconvolution algorithm and analyzed with NIS-Elements (Nikon) or Fiji (NIH) software. Skint1 staining relative to T cell bodies was quantified using Definiens Developer XD software (Definiens). Individual cells and Skint1 particles were detected using Otsu's automatic multilevel thresholding and multiresolution segmentation algorithms. The MFI of Skint1 particles within and outside of T cell body masks was quantified.
Histology. Whole ears were flattened in a histology cassette and fixed in neutral buffered formalin for 24 h and dehydrated in 70% ethanol for 24 h. Ears were embedded, sectioned and stained with H&E by the Experimental Histopathology Facility at The Francis Crick Institute. Sections were photographed using a SlideScanner (Zeiss) at 10× magnification. Epidermal thickness was determined as the average of 30 manual measurements using QuPath software.
Ear swelling measurement. Total ear thickness was calculated as the average of three measurements of the ear pinna with a micrometer (Mitutoyo) after discarding the first measurement.
Measurement of CPDs. CPDs were measured by ELISA. Epidermal sheets were separated by NH 4 SCN as above. Epidermal DNA was isolated using a DNeasy Blood and Tissue kit (Qiagen) and denatured at 100 °C for 10 min. DNA (10 ng) was dried onto protamine sulfate-coated 96-well plates (Nunc), and ELISAs were performed using anti-CPDs (Cosmo Bio), biotinylated goat anti-mouse IgG (Invitrogen), streptavidin-horseradish peroxidase (R&D) and a TMB Substrate Reagent set (BD), as per the anti-CPDs manufacturer's protocol. Absorbance was measured using an Infinite F200 Pro spectrophotometer (Tecan).
Toluidine blue permeability. Whole ears were dehydrated and rehydrated by sequential 2-min incubations in ice-cold 25%, 50%, 75%, 100%, 75%, 50% and 25% methanol in PBS and incubated in cold PBS for 5 min. Ears were stained in 0.1% (wt/vol) toluidine blue (Sigma-Aldrich) for 2 min and then destained in PBS for 1 min. Ears were photographed with an Stemi SV 11 stereomicroscope (Zeiss). Toluidine blue staining was quantified in Fiji by color deconvolution (H&E setting) and thresholding to measure total ear area (green channel) and toluidine blue area (indigo channel).
qPCR. For analysis of total epidermal gene expression, epidermal sheets were separated by NH 4 SCN as above and homogenized in buffer RLT (Qiagen) using stainless steel beads (Qiagen) and a TissueLyser II (Qiagen). RNA was isolated using the RNeasy Mini kit (Qiagen) and an RNase-free DNase set (Qiagen).
For analysis of gene expression from 4-1BB hi and 4-1BB lo DETCs, mice were shaved, and epidermal suspensions from ears and trunk skin were prepared using TrypLE Express (Gibco) as described above but with a 4-h incubation to separate the epidermis. Cells were stained with Live/Dead Fixable Near-IR dye (Molecular Probes) and anti-CD45-FITC, anti-TCRδ-BV421 and anti-4-1BB-PE (all BioLegend) and sorted. RNA was isolated as described above.
cDNA was synthesized using Superscript II Reverse Transcriptase, dNTP set and Anchored Oligo(dT)20 primer (Invitrogen). qPCRs were run with PowerUp SYBR Green (Applied Biosystems) and 0.4 μM primer sets listed in Supplementary  Table 5 on a ViiA 7 Real-Time PCR System (Thermo Fisher). Threshold cycle (C t ) values were processed using QuantStudio software, and ΔC t values were calculated as 2 (reference -gene) , where the reference was CypA (cyclophilin).
Single-cell RNA sequencing. Wil-type FVB mice were treated with 10 μg of anti-Skint1 or rat IgG2a isotype control by i.d. injection into the ear pinna, as described above. Seven days later, epidermal suspensions were made from ear skin and were stained as described above using Live/Dead Fixable Near-IR dye (Molecular Probes), anti-EpCAM-PE (eBioscience) and anti-CD45-APC and anti-I-A/I-E-FITC (all BioLegend), and live CD45 + MHCII -EpCAMlymphocytes were sorted using an Aria Fusion (BD). Samples were prepared using the Chromium Single Cell 3′ GEM, Library and Gel Bead kit v3 (10x Genomics). Cell viability was assessed using an EVE cell counter and trypan blue viability stain. Approximately 10,000 cells were loaded into the 10x Chromium, which was operated according to manufacturer's instructions. Sequencing was performed on an Illumina HiSeq 4000 with a read configuration recommended by 10x Genomics for the sequencing of these libraries (28-base pair (bp) read 1, 98-bp read 2, 8-bp index 1).
Bioinformatic analysis of single-cell RNA sequencing. Data were processed using CellRanger v3.0.2 using the prebuilt mm10 index v3.0.0, and individual Seurat objects were created. All of the following functions belong to the Seurat v3.1.1 package 56 , and their parameters are mentioned only if differing from the default. The following steps were applied to the data: 1. Cells with less than 500 features, more than 2,500 features and more than 12% mitochondrial reads were excluded. 2. Feature counts were normalized per cell using the NormalizeData function. 3. The FindVariableFeatures function was used to select the 2,000 most variable genes. 4. These genes were used to identify integration anchors (FindIntegrationAnchors, dims = 1:20, anchor.features = 2050), and the two samples (isotype control and anti-Skint1) were integrated using the IntegrateData function. 5. The expression of each gene was centered using the ScaleData function parametrized with all the available features. 6. Principal component analysis was performed with the RunPCA function (npcs = 100). 7. Clustering was performed using the FindNeighbors function (dims = 1: 40) and the FindClusters function (resolution = 1.18). 8. Dimension reduction was performed using RunUMAP (dims = 1:40). 9. Clusters were manually annotated using the following marker genes: and Trdv4, and their cell IDs were extracted.
The original Seurat objects were filtered using DETC IDs to remove other cell types from the analysis. Steps 1 to 8 were reapplied to the resulting data with the following differences: FindNeighbors (dims = 1:23), FindClusters (resolution = 0.28) and RunUMAP (dims = 1:23).
Differential expression analysis between the two experimental conditions was performed using the FindMarkers function of Seurat (logfc.threshold = 0, min. pct = 0, min.cells.feature = 0 and min.cells.group = 0). The resulting gene list was sorted by avg_logFC (decreasing order), and GSEA v3.0 was run for the Skint1 selection signature with the following parameters: -nperm 1000, -scoring_scheme classic, -set_min 3 and -set_max 50000. The Skint1 selection signature was generated from the top 200 genes overexpressed in E15 V γ 5 + thymocytes from wild-type FVB versus Tac embryos (Supplementary Table 1), as determined by Illumina microarray analysis and described previously 35 .
Expression plots and violin plots were produced using the normalized counts.
Bulk RNA sequencing. Wild-type C57BL/6 mice were irradiated with 45 mJ cm -2 UVB as described above or were left unchallenged. Dorsal ear skin was collected at 24 h after UVR, and single-cell suspensions were made and stained as described above with Live/Dead Fixable Near-IR dye (Molecular Probes) and anti-CD3-PerCPCy5.5, anti-CD45-APC, anti-I-A/I-E-BV605 and anti-TCRδ-BV421 (all BioLegend). Live CD45 + CD3 + TCRδ + DETCs were sorted directly into buffer RLT Plus (Qiagen) using an Aria Fusion (BD), and RNA was isolated using an RNeasy Micro Plus kit (Qiagen). cDNA was prepared from 10 ng of input RNA using the Ovation RNA-Seq System V2 (NuGEN). Final libraries were then constructed using the Ultralow Library System V2 (NuGEN). The resulting libraries were pooled for sequencing on an Illumina HiSeq 2500 platform with single-ended 75-bp reads.
Bioinformatic analysis of bulk RNA sequencing. Raw reads were quality and adapter trimmed using cutadapt v1.9.1 (ref. 59 ) before alignment. Reads were then aligned and quantified using RSEM v1.3.0 and STAR v2.5.2 (refs. 60,61 ) against the mouse genome GRCm38 and annotation release 86, both from Ensembl. Differential gene expression analysis was performed in R using the DESeq2 v1.24.0 package 62 . Differential genes were selected using a 0.05 false discovery rate threshold. Normalization and VST was applied on raw counts before performing principal component analysis and Euclidean distance-based clustering. For the heat maps, Z scores were computed on the VST of the raw counts. Heat maps were made in R using the ComplexHeatmap v2.0.0 package 63 . GSEA was run as described above using gene signatures created from the top 200 overexpressed genes in each DETC cluster from the single-cell RNA-sequencing experiment.
Statistics. Data were analyzed in Prism 9.3.0 (GraphPad) and R v3.6.0. Statistical tests were performed as indicated in the figure legends, and n values are also provided. All error bars represent mean ± s.d. Unless explicitly stated in the figure legends, all dots and n values refer to individual mice. Mice were randomly assigned to treatment groups where applicable. No data were excluded. Data were presumed to be normally distributed, and all tests were two tailed where applicable. Statistical significance was defined as P < 0.05, and all significant P values are shown in the figures. Welch's correction was applied to t-tests where variance was significantly different between sample groups as determined by an F-test.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
RNA-sequencing datasets generated in this paper are available under accession codes GSE160476 (single cell) and GSE160477 (bulk). Previously published datasets used in this paper are provided in the Supplementary Tables. The public gene set used in Extended Data Fig. 3e is available at GO:0050778. Source data are provided with this paper.