Skint-1 is a highly specific, unique selecting component for epidermal T cells

Abstract

αβ T-cell repertoire selection is mediated by peptide–MHC complexes presented by thymic epithelial or myeloid cells, and by lipid–CD1 complexes expressed by thymocytes. γδ T-cell repertoire selection, by contrast, is largely unresolved. Mice mutant for Skint-1, a unique Ig superfamily gene, do not develop canonical Vγ5Vδ1⁺ dendritic epidermal T cells. This study shows that transgenic Skint-1, across a broad range of expression levels, precisely and selectively determines the Vγ5Vδ1⁺ dendritic epidermal T-cell compartment. Skint-1 is expressed by medullary thymic epithelial cells, and unlike lipid–CD1 complexes, must be expressed by stromal cells to function efficiently. Its unusual transmembrane–cytoplasmic regions severely limit cell surface expression, yet increasing this or, conversely, retaining Skint1 intracellularly markedly compromises function. Each Skint1 domain appears nonredundant, including a unique decamer specifying IgV-domain processing. This investigation of Skint-1 biology points to complex events underpinning the positive selection of an intraepithelial γδ repertoire.

Vast numbers of “unconventional” T cells constitutively associate with body surfaces such as the gut, reproductive tract, and lung, existing as intraepithelial lymphocytes (IELs) that differ from conventional systemic T cells by several criteria. For example, many do not undergo circulation through secondary lymphoid tissue, do not express CD4 or CD8, and are not restricted to peptide–MHC antigens. They are often highly oligoclonal, with disproportionate enrichment in γδTC receptor (TCR)⁺ cells, display an “activated-yet-resting” phenotype, and may interact directly with epithelial cells (1). Many such properties were elucidated via study of murine Vγ5Vδ1⁺ dendritic epidermal T cells (DETCs) (2, 3), and they have fueled the view that IELs form lymphoid stress-surveillance compartments that respond rapidly and oligoclonally to a limited set of “stress antigens” on dysregulated epithelial cells (4, 5). Consistent with this, mice with DETC deficiencies may display increased susceptibility to carcinogenesis and tissue inflammation and show wound healing defects (3, 6–8). Likewise, deficits in human skin γδTCR⁺ cells have been associated with chronic wound healing deficiencies (9). Such results underpin the need to identify and characterize molecules mediating interactions between epithelial cells and their associated T cells (4).

DETCs may be activated when the receptor NKG2D engages MHC-I–related ligands, such as Rae-1, expressed by dysregulated epithelial cells (5, 10, 11). Indeed, NKG2D defines a major axis of immunosurveillance and is commonly targeted by tumors and viruses (5, 12, 13). This notwithstanding, the significance of TCR-mediated interactions is suggested by the increased susceptibility to inflammation-associated squamous cell carcinoma shown by Vγ5Vδ1^−/− mice that harbor “replacement” NKG2D⁺ TCRγδ⁺ DETCs expressing other TCRs (11). Hence, there is much interest in how IEL compartments expressing particular oligoclonal TCRs form and are maintained.

The DETC Vγ5Vδ1 TCR is the product of direct germline gene segment recombination, which is favored by the limited scope of RAG-mediated V[D]J-recombination during the narrow period [from embryonic day 14.5 (E14.5) to E18] when DETCs develop (2, 14). Nonetheless, IEL repertoires also appear to be shaped in part by thymic selection, as is true for all other T cells (15–18). For conventional αβ T cells, engagement of the newly generated TCR by “medium-affinity” peptide–MHC complexes (pMHCs) positively selects cells for further differentiation, provided that they avoid high affinity self-pMHC engagements that promote apoptosis or other forms of effector inactivation (19). By this means, emerging T cells are educated about the molecular form of normal self, yet purged of those cells that could be activated by peripheral autoantigens. By contrast, IEL selection appears to be different, as the purging of systemic T cells that occurs in mice expressing transgenic (Tg), autoantigen-reactive TCR does not extend to IEL (20). This is consistent with progenitors of conventional versus unconventional T cells differing in key signaling pathway (21, 22). Thus, the so-called agonist selection theory posits that the same autoantigens select IELs and activate them in the periphery, thereby promoting the formation and function of focused immunosurveillance repertoires (22, 23). Unfortunately, exploration of this important concept is hampered by ignorance of natural antigens for IEL, and of the components of unconventional T-cell selection machinery.

In this vein, we recently found that, although FVB/N mice supplied by Taconic Farms (FVB.Tac) harbor DETCs in comparable numbers to other strains, they lack the canonical Vγ5Vδ1 TCR, and spontaneously develop dermatitis. The FVB.Tac defect reflected impaired thymic maturation selectively of Vγ5Vδ1⁺ DETC progenitors that, for example, remain CD45RB^lo (17), and was attributed to a premature stop codon in Skint-1, the prototypic member of a unique Ig supergene family closely related to the B7 and butyrophilin gene families (24). Skint-1 is rapidly evolving and is expressed by keratinocytes and thymic stroma, and its deficiency can be complemented by engaging thymocytes with agonist anti-TCR antibodies (17). These data suggested Skint-1 to be the first identified natural component of the selection machinery for a specific IEL compartment. This biological characterization of Skint-1 validates that view, offering insight into the complexity of mechanisms selectively regulating IEL maturation.

Results

Ubiquitous Skint-1 Restores Vγ5Vδ1⁺ DETC Development.

We previously reported that introducing Skint-1 into embryos of Skint-1–mutant FVB.Tac mice variably rescued normal phenotypic maturation of Vγ5Vδ1⁺ DETC progenitors (24). To determine whether it would rescue a mature canonical DETC repertoire, we created lines of FVB.Tac mice Tg for Skint-1 expressed from a β-actin promoter, with the mature protein tagged with an aminoterminal FLAG epitope (NF) to permit its detection. Tg.NF-Skint1 fully rescued Vγ5Vδ1⁺ DETC progenitor maturation, as evidenced by normal ratios and numbers of CD45RB^hi/CD45RB^lo Vγ5⁺ thymocytes at E17 (Fig. 1A). The transgene also restored normal DETCs in all founders (lines 1, 25, 39) and their progeny (Fig. 1 B and C). γδTCR⁺ (GL3⁺) T cells colonized the skin soon after birth (compared with the delay in FVB.Tac non-Tg littermate controls [NLC]) and showed characteristically high surface γδTCR expression, and approximately 94% stained with 17D1, an antibody that defines the Vγ5Vδ1 clonotype (16).

Fig. 1.

Ubiquitous Tg.NF-Skint1 expression restores Vγ5Vδ1⁺ DETC development. (A) Flow cytometry of thymocyte suspensions prepared from FVB.Jax, FVB.Tac, and FVB.Tac Tg.NF-Skint1^−/+ line 39 fetuses at E17. Data are gated to show only Vγ5Vδ1⁺ (17D1⁺) thymocytes. Similar results were observed in all Tg.NF-Skint1 founder lines. (B) Flow cytometry of epidermal suspensions prepared from body wall skin of 3-wk-old mice. The bottom panels show γδTCR⁺ cells only. (C) Staining of DETC in FVB.Jax and Tg.NF-Skint1 epidermal sheets. (Scale bar: 20 μm.) (D and E) Quantitative RT-PCR analysis of Skint-1 expression in pooled E17 thymi and in skin from individual animals from Tg.NF-Skint1 (D) and untagged Skint1 (E) Tg lines, calculated relative to GAPDH. (F and G) Flow cytometry analysis of Skint1 Tg animals versus B6xSJL NLCs, performed as in A and B.

A key prediction of agonist selection is that overexpression does not negatively select. Consistent with this, rescued DETC compartments were indistinguishable from WT, even in mice in which E17 thymus and adult skin expressed approximately 30- to 100-fold more Skint-1 mRNA than NLC (Fig. 1D). Moreover, DETC progenitor development and mature DETC repertoire formation was normal in Skint1-sufficient C57BL/6 × SJL mice rendered Tg for an additional Skint-1 allele (line 17) expressed at approximately 50-fold higher levels than normal (Fig. 1 E–G).

Skint-1 Selectively Regulates Vγ5Vδ1⁺ DETCs.

Consistent with its expression from a β-actin promoter, Tg.NF-Skint1 protein was detected in all tissues examined, running between 33 kD and 40 kD (depending on gel composition), and as larger complexes (>75 kD) partially resistant to reduction (Fig. 2A and subsequent sections here). Despite expressing Tg.NF-Skint1, no tissues beyond skin showed alterations to their T-cell repertoires: Vγ5⁺ cells did not accumulate in thymus, spleen, or gut (Fig. 2B), and other TCRs (e.g., Vγ4 and Vγ1 in spleen and Vγ7 in gut IEL) were expressed normally (Fig. 2C and Fig. S1). Even the highly related, reproductive IEL repertoire that also develops from fetal thymic precursors, and which uses a Vγ6Vδ1 TCR (in which Vδ1 and the Vγ6 CDR3 are each identical to the Vγ5Vδ1 DETC TCR) was normal in Tg.NF-Skint1 females (Fig. 2D). Thymic and peripheral αβ T-cell subsets were likewise unaltered in FVB.Tac and B6xSJL Tg mice (Fig. 2E and Fig. S1). Whereas confinement of the effects of Skint-1 to DETCs in normal mice might reflect its very restricted gene expression, its highly selective effects in Tg mice expressing it ubiquitously presumably reflect its biology.

Fig. 2.

Tg NF-Skint1 does not regulate γδTCR repertoires outside the skin. (A) Western blot of Tg.NF-Skint1 protein in tissues (T, thymus; S, spleen; G, gut) from founder line 39. Protein was immunoprecipitated from Tg.NF-Skint1⁻ and Tg.NF-Skint1⁺ tissue lysates with mouse anti-FLAG M2 antibody, and immunoblotted with rabbit anti-FLAG antibody. Putative monomer (◀) and dimer species (◁) are observed in all Tg tissues. (B–E) Flow cytometry of cell suspensions prepared from the thymus, spleen, gut, and uterus of Tg⁻ and Tg⁺ animals of Tg.NF-Skint1 line 39. In B and C, all cells shown are gated on γδTCR⁺ and numbers reflect the percentage of γδ T cells, whereas CD3⁺ cell and total live cells are shown in D and E, respectively. (D) 17D1 binds to γδTCR⁺ cells that are Vγ5⁻; this corresponds to a demonstrated GL3-dependent cross-reactivity with the Vγ6Vδ1⁺ TCR used by the uterine IEL compartment (33).

Skint-1 Is Functionally Expressed by Selecting Cells.

Skint-1 mRNA was expressed, albeit across a 10-fold range, in nonhematopoietic (i.e., CD45⁻) E15 thymic stromal cells from all mouse strains tested (Fig. 3A). As would be predicted for a thymic selecting component, Skint-1 expression is independent of thymocytes that respond to it, being expressed comparably in WT and TCRδ^−/−.FVB mice (Fig. 3A). Skint-1 and Skint-2 expression increase as the thymus expands in size in neonates and young adults (Fig. 3 A and B). During this time, there is a major expansion of medullary thymic epithelial cells (mTECs), key mediators of αβ T-cell selection. Consistent with this, Skint-1 (and Skint-2) RNA is substantially enriched in mTEC versus cortical thymic epithelial cell (cTEC) preparations, and maps in situ to the medulla (Fig. 3C).

Fig. 3.

Skint-1 is expressed in thymic medullary epithelium in a TCR-independent fashion and effects function from a stromal cell context. (A) Quantitative RT-PCR analysis of Skint-1 expression in CD45⁻ thymic stroma (Left and Center) or FACS-purified TEC (Right). CD45⁻EpCam⁺ TEC from pooled FVB.Jax E15-E18 thymi were divided into Ly51⁺ cTEC and Ly51⁻ mTEC subsets, and compared with total E17 thymus. Skint-1 expression is relative to GAPDH. (B) Quantitative RT-PCR analysis of Skint-2 expression as in A. (C) In situ hybridization analysis of Skint-1 expression in adult FVB.Jax thymus. Right: Close-up of medullary area indicated (Left). (D) Analysis of Vγ5Vδ1⁺ DETC precursor maturation (17D1⁺CD45RB^hi) in mixed RTOC of purified CD45⁺ thymocyte progenitors and CD45⁻ stroma from FVB.Jax, FVB.Tac, or Tg.NF-Skint1 E15 thymi.

The significance of Skint-1 expression by TEC was functionally tested in reaggregate fetal thymic organ culture (RTOC). We previously reported (17) that Vγ5Vδ1⁺ DETC progenitors will mature into CD45RB^hi cells when reaggregated with WT FVB.Jax stroma, but not FVB.Tac stroma, with an example of the latter shown in Fig. 3D (∼1% Vγ5Vδ1⁺CD45RB⁺ cells). FVB.Jax stroma supports maturation of thymocytes from Tg.NF-Skint1 mice, and consistent with the phenotypic rescue of Tg mice, stroma of FVB.Tac Tg.NF-Skint1 mice was comparably effective (∼4% vs. ∼6% Vγ5Vδ1⁺CD45RB⁺ cells; Fig. 3D). However, when Tg.NF-Skint1 was provided on thymocytes, but not on stroma, maturation was not supported (∼1% Vγ5Vδ1⁺CD45RB⁺ cells; Fig. 3D). The importance of the cell type expressing Skint-1 was further tested by seeding FVB.Tac RTOC with Skint-1 transductants of either human embryonic kidney 293 cells or murine bone marrow stromal OP9-DL1 cells. 293.Skint-1 cells promoted no maturation whereas OP9-DL1. Skint-1 promoted some (Fig. S2A). However, whereas OP9-DL1 cells can support both αβ and γδ T progenitor development (25, 26), OP9-DL1 cells and OP9-DL1.Skint-1 cells consistently failed to support DETC maturation outside of RTOC.

Structure–Function Analysis of Skint-1 Protein.

Given its predicted structure, the simplest model of Skint1 is as an mTEC ligand for a thymocyte counter receptor, e.g., the TCR. However, substantial Skint1 surface expression was not seen following NF-Skint-1 transduction of 293 cells (Fig. 4 A and B), MK murine keratinocytes, or murine TEP1.1 TEC (Fig. S3), as judged by anti-FLAG antibody or a polyclonal antibody raised against KYVERTELL, a conserved Skint1 IgV domain peptide. Conversely, there was ample surface expression of a FLAG-tagged form of the Skint1-related intestinal epithelial molecule, Btnl1, when transduced into gut epithelial cells (Fig. S3). The FVB.Tac-derived mutant allele of Skint-1 (Skint1_TacTAA) actually showed slightly higher surface expression than WT, as detected by KYVERTELL, although the inverse was shown by anti-FLAG (Fig. 4 A and B and Discussion). In sum, neither WT nor mutant Skint1 is readily expressed at the cell surface, consistent with which surface expression of Tg.NF-Skint1 on primary cells from Tg mice was barely detectable, despite the detection of protein by immunoprecipitation and Western blotting (Fig. 2A).

Fig. 4.

Biological and biochemical properties of Skint1. (A) Graphic representation of the Skint-1 constructs tested. (B) Surface flow cytometry of 293 cells transfected with the constructs in bicistronic vectors. All constructs include a FLAG tag at the N terminus and the KYVERTELL motif in the IgV domain; hCD2 is the marker of transduction efficiency. (C) Analysis of Vγ5Vδ1⁺ DETC precursor maturation in FVB.Tac RTOC 12 d after transduction with the constructs from A. (D) The multimerization of WT and Tac alleles is analyzed by Western blot analysis of 293 transfectants. Protein was immunoprecipitated with mouse anti-FLAG M2 or mouse anti-HA antibody, and immunoblotted with rabbit anti-FLAG antibody.

A signatory feature of Skint1 is the putative multipass transmembrane (TM) region (24). To test if this inhibited cell surface expression, a construct was created that terminated Skint1 after a single TM region (Skint1_1TM). Surface expression increased but was low relative to the fusion of Skint1 IgV–IgC to the human CD4 TM (Skint1_VChCD4_TM; Fig. 4 A and B). Thus, poor surface expression of Skint1 is largely attributable to sequences carboxy-terminal to the IgV–IgC ectodomains. Furthermore, Skint1 surface expression was not markedly improved by treatment of cells with any of several cytokines or “cell dysregulators” such as heat shock, epigallocatechin-3-gallate (a caspase 14 activator), or azacytidine, which alters DNA demethylation (Fig. S3). The efficacy of these agents was controlled for, e.g., by monitoring cytokine-induced increases in surface MHCII expression.

The functional potential of various Skint-1 alleles was determined by transduction of FVB.Tac RTOC and quantitating Vγ5Vδ1⁺ CD45RB^hi cells 12 d later as a marker of DETC maturation. Each transduced RTOC displayed comparable numbers of GFP⁺ transductants. As expected, WT Skint1 consistently rescued maturation, whereas Skint1_TacTAA did not; however, nor did Skint1_1TM or Skint1_VChCD4_TM (Fig. 4C and Figs. S2B and S4A). This was not because increased cell-surface expression of Skint1 drove negative selection, because, relative to controls, these alleles did not inhibit DETC progenitor maturation in transduced WT FVB.Jax RTOC (Fig. S4B). Thus, surface display of the Skint1 IgV-IgC ectodomains was insufficient for function. Intriguingly, the same was true for either of two forms of Skint1 retained in the endoplasmic reticulum via KKDQ or KDEL motifs (Fig. S4A).

More insight was sought by analyzing the Skint1 proteins encoded by the various alleles. On nonreducing gels (i.e., -DTT), there were many high molecular mass (hMM) forms (Fig. S4A, Left), with a putative dimer at approximately 70 kD (predicted monomeric MW of 38 kD). Reduction (i.e., +DTT) did not substantially affect the hMM forms of Skint1, but did evoke a putative monomer (Fig. S5A, Right, and Fig. 4D, lane 1). Comparable patterns were shown by Skint1_VChCD4_TM, which predictably migrated slightly faster, whereas Skint1_TacTAA was consistently expressed at slightly lower levels and was more sensitive to reduction, suggestive of instability (Fig. S5A). To test whether the hMM forms reflected Skint1 homomultimerization, 293 cells were single- or double-transfected with Skint-1 fused to aminoterminal FLAG- or HA-epitope tags. FLAG-Skint1 could be immunoprecipitated from double- but not single-transductants by using anti-HA antibody, followed by Western blotting with anti-FLAG (lanes 9–16, Fig. 4D). This was equally true for Skint1 or Skint1_TacTAA. Hence, Skint1 forms homomeric complexes in cells. These are not artifacts of lysis, because anti-HA did not retrieve FLAG-Skint1 from mixtures of single transductant lysates.

Nonredundancy of Skint-1.

Several Skint genes share very high structural similarity and expression patterns (Fig. S6) (24). To test whether Skint-1 is truly nonredundant or might be substituted for by overexpressing other Skint genes, FVB.Tac RTOC was transduced with GFP⁺ viruses expressing Skint-2 (the gene most similar to Skint-1) and Skint-7, respectively (Fig. S5B). Neither rescued DETC progenitor maturation. Skint1/Skint2 domain-swap chimeras (e.g., SkintV1C1T2; SkintV2C2T1, where, e.g., V1 refers to Skint1-derived IgV; V2 refers to Skint2-derived IgV) were then created to identify which regions of Skint1 were essential for activity (Fig. S5B). Most constructs were comparably expressed as protein (Fig. S5C), yet none recapitulated the activity of Skint-1 when expressed in RTOC (Fig. S5B). Hence, each discrete domain of Skint1 was required for its biological activity.

For all chimeras containing Skint1 IgC, lysates contained low molecular mass forms approximately the size of IgV domains, which are the putative products of proteolysis (Fig. S5C, arrowheads). To map the sequences responsible for this, further Skint1/Skint2 chimeras were made, starting with SkintV1C1T2 and incrementally substituting 10-aa blocks of C1 with C2 (Fig. 5A). This showed that cleavage was specified within the aminoterminal 10 aa of Skint1 IgC. When these 10 aa were specifically replaced in Skint1 by the counterpart Skint2 sequence [Skint1_{Sk2(140–149)}], the lysates of the corresponding transfectants showed no low molecular mass forms, as well as conspicuously fewer hMM forms, despite comparable monomer expression (Fig. 5B). When this form was functionally assessed in FVB.Tac RTOC, it did not rescue DETC progenitor maturation. Likewise, the 10-aa fragment was unable to confer biological activity on Skint2 (Fig. 5C). Hence, the nonredundancy of Skint1 extends to a 10-aa motif in IgC that is unique in the mouse genome, yet 90% conserved in rat Skint-1.

Fig. 5.

Proteolytic processing specified by IgC is required for function. (A and B) Western blot analysis of 293 cells transfected with N-terminal FLAG-tagged Skint-1/Skint-2 chimeric constructs. Protein was immunoprecipitated with mouse anti-FLAG M2 and immunoblotted with rabbit anti-FLAG antibody. In A, the recombination between Skint1 and Skint2 was shifted in 10-aa increments in the IgC, whereas in B, only the first 10 aa of the Skint1 IgC was replaced with the analogous sequence from Skint2. (C) Analysis of Vγ5Vδ1⁺ DETC precursor maturation in FVB.Tac RTOC 12 d after transduction with the indicated chimeric constructs.

Discussion

Although positive selection of cells with “useful” receptors is well documented for conventional, MHC-restricted αβ T cells and for CD1-restricted NKT cells, its relevance to γδ cell biology has remained controversial (27). This article provides functional validation that Skint-1 drives the selective development of DETC progenitors, and offers information about how it operates. Provision of Skint-1 to Skint1-deficient FVB.Tac mice rescued the DETC repertoire with almost surgical precision in several independent Tg mouse lines. Its major effects were limited to DETCs, even when it was broadly expressed, consistent with the prospect that Skint1 engages the Vγ5Vδ1 TCR. Nonetheless, no direct evidence yet exists for thymocyte or DETC binding by Skint1-FcR fusions and other recombinants, and such proteins seeded into FVB.Jax or FVB.Tac RTOC did not affect DETC development. Moreover, it is premature to interpret studies of Skint1 binding to soluble γδTCRs until the latter reagents are themselves validated.

Skint1 counter-receptor identification has been confounded by properties of the Skint1 protein. Applying the methods of Munro and colleagues (28), among others, Skint1 most likely localizes to the plasma membrane, and its intracellular retention prevents its function. Nonetheless, we show that its tail severely limits surface expression, and that this is not obviously relieved by agents that stress cells or alter cell growth. MR1, which selects mucosa-associated invariant αβ T cells, is also difficult to detect at cell surfaces (29). Skint1 also seems to form detergent- and reduction-resistant hMM species, suggestive of complexes that presumably include other proteins, at least one of which may cleave Skint1 as specified by 10 aa in Skint1-IgC. Strict requirements for complex formation might explain why no Skint1 domain is redundant, and why efficient function requires Skint expression from stromal cells.

Although cortical/medullary segregation is incomplete by E16.5, when DETC selection is in progress, the findings that a bipotent progenitor in the thymus differentiates into cTECs and mTECs (30, 31) accommodate the possibility that DETC selection is determined by the very earliest mTECs. Hitherto, the inability of OP9-DL1 stromal cells to support the differentiation of DETC progenitors has been striking, but this is not relieved simply by the enforced expression of Skint-1. When OP9-DL1.Skint-1 cells are seeded into RTOC that includes TECs, some progenitor maturation was rescued. However, the failure of Skint1-mutant TECs and Skint1⁺ thymocytes to jointly support efficient DETC differentiation argues that the TEC are not simply sources of critical cytokines. Rather, they may provide other epithelial-specific moieties, possibly including other Skint proteins, that may collectively regulate expression of ligands for the TCR and/or coactivating receptors that selectively promote DETC progenitor maturation. Such a complex processing pathway evokes MHC- and CD1-restricted selection that involve highly regulated expression of the antigen that engages developing thymocytes. Note that, although thymocytes can select NKT cells by using lipid–CD1 complexes, they can effect CD4 T-cell selection only when engineered to express CIITA, a transcription factor that activates a battery of antigen-presentation molecules (32). A similar situation may hold true for Skint-1 selection.

A complex cell biology of Skint1 also offers an explanation for the severe impairment of canonical DETC development in FVB.Tac mice, as we show here that the Skint1_TacTAA allele can be expressed as a protein that is as readily detected on the cell surface as WT Skint1 and that can multimerize. Possibly significant is the reciprocal efficacy of anti-FLAG and anti-KYVERTELL to detect surface expression of WT Skint1 and Skint1_TacTAA, which might reflect different, biologically distinct conformations of the ectodomain adopted by the WT and mutant proteins, respectively. Ongoing studies seek to define the means by which DETC progenitors receive the Skint1-dependent signal, and the molecular pathway they then activate to promote selection. If the TCR is not the direct target of Skint1 and/or its associated complex, the target may be an uncharacterized molecule unique to DETC progenitors, or a shared molecule which Skint1 and/or its complex engages with differential affinity on different progenitors. Either explanation would provide novel insight. Likewise, Skint-1 is constitutively expressed by primary keratinocytes, offering the chance to compare how its peripheral function compares to its function as a selecting component.

Materials and Methods

Cloning and Tg Mice.

Overlap PCR introduced the FLAG epitope (plus a silent PsiI site for genotyping) into Skint-1 amplified from cDNA of C57BL/6 skin, which was then cloned into bicistronic vectors. Untagged and FLAG-tagged Skint-1 inserted into the pCAGGS vector were injected into blastocyst pronuclei after purification of SalI and PstI fragments. All animal husbandry followed locally approved protocols.

Phenotyping.

Phenotypic analyses were performed as described previously (17) and are described in detail in SI Materials and Methods.

Expression Analysis.

RNA and protein expression analyses were performed according to conventional methods, as described in detail in SI Materials and Methods.

Cell Sorting and Reaggregate Thymic Organ Cultures.

For a full description of procedures and reagents for cell sorting and RTOC, see SI Materials and Methods.