WIF1 is expressed by stem cells of the human interfollicular epidermis and acts to suppress keratinocyte proliferation.

TO THE EDITOR Although it is possible to isolate keratinocyte stem cells (KSCs) of the human interfollicular epithelium by virtue of their composite cell surface phenotype (α6 integrin and CD71; Li et al., 1998; Schluter et al., 2011), it remains difficult to visualize them in situ within the skin. Although murine interfollicular KSCs can be visualized in situ by DNA-labeling protocols combined with long chase periods (termed DNA label-retaining cells or LRCs; Bickenbach, 1981), this approach is not possible in humans for ethical reasons. Our aim was to identify single markers for human interfollicular KSCs at the protein level, by mining microarray data previously generated by us to identify genes preferentially expressed by KSCs compared with their more committed progeny (Schluter et al., 2011). Interfollicular epidermal cells from neonatal human skin tissue can be resolved into quiescent KSCs, cycling committed progenitor (CP) or transient-amplifying (TA) cells, and early differentiating (D) cells on the basis of their cell surface phenotype, i.e., α6briCD71dim, α6briCD71bri, and α6dim, respectively (Li et al., 1998). Importantly, functional long-term tissue reconstitution assays confirmed the identity of α6briCD71dim cells as a bona fide stem cell population, given their potency in reconstituting human interfollicular epidermis for up to 10 weeks from as little as 100 cells (Schluter et al., 2011). Further, molecular characterization by microarray validated their quiescent nature (Schluter et al., 2011). Notably, KSCs overexpressed negative regulators of key signaling pathways including Wnt, insulin, sonic hedgehog, and transforming growth factor beta, shown in Figure 1a (taken from Schluter et al., 2011). Among the Wnt signaling pathway inhibitors expressed by KSCs were Sfrp-1, Dkk-3, and (Wnt inhibitory factor-1) WIF1 (Figure 1a). Notably, WIF1 was the single most overexpressed gene in KSCs (Schluter et al., 2011)—its differential expression in KSCs was confirmed by reverse transcription–PCR (Figure 1b). These data suggested that WIF1 may be an excellent marker for interfollicular KSCs of human skin. We performed dual immunofluorescence staining in human neonatal and adult breast skin for WIF1, and keratin (K)14 or K15—markers of basal keratinocytes and α6 integrin—a basement membrane marker. Single basal keratinocytes coexpressing WIF1 and K14/K15 were found scattered throughout the basal layer in neonatal foreskin, but were more widely expressed in adult skin (Figure 1c–h; arrows). In addition, almost all dermal cells, an active site for Wnt signaling, were also positive for WIF1 (Figure 1c–h). WIF1 immunostaining on cytospins of α6briCD71dim/KSC, α6briCD71bri/TA or CP, and α6dim/D cells confirmed highest WIF1 protein expression in KSCs but not the cycling TA/CPs qualitatively (Figure 1k–m), and quantitatively (Figure 1n). Although the α6dim differentiating fraction has a higher level of WIF1 than TA/CP cells in cytospins, this population is somewhat heterogeneous, containing many suprabasal cells that are clearly WIF1 positive. Notably, cells positive for both K10 and WIF1 were consistently located in the suprabasal layers in both neonatal and adult skin in situ (Figure 1g and h). Further, dual staining for Ki67 and WIF1 in adult skin revealed that Ki67-positive cells do not express WIF1 (Figure 1c). Thus, WIF1 is highly expressed in non-cycling cells, indicating a role in negatively regulating keratinocyte proliferation. To further demonstrate that WIF1 is a bona fide interfollicular KSC marker, we sought to determine whether slow-cycling LRCs coexpressed WIF1 in human skin. LRCs can be visualized in human skin in situ in grafts on immunodeficient mice (Lyle et al., 1998), or alternately in long-term organotypic cultures of human keratinocytes, with an incidence of <1% after 6 weeks (Muffler et al., 2008). Using the latter approach validated by us previously, we stained histological sections of iododeoxyuridine-labeled long-term organotypic cultures for iododeoxyuridine and WIF1. All iododeoxyuridine-positive epidermal cells were also WIF1 positive (Figure 2f–k), whereas cycling Ki67-positive cells were not (Figure 2a and b), providing supportive evidence that WIF1 protein is an excellent marker for quiescent stem cells of the interfollicular epidermis. Notably, there was an absence of WIF1-positive cells in the hyperproliferative basal layer of organotypic cultures in the early 2-week post-labeling time points, as evidenced by the presence of many Ki67-positive cells (Figure 2a and c). WIF1-positive basal cells were detectable by 6 weeks (Figure 2b and d), a time point corresponding with the establishment of single LRCs (Muffler et al., 2008), although none of these were Ki67 positive. Notably, none of the WIF1-positive cells in the basal layer of organotypic cultures were K10 positive at all time points analyzed, and double-positive cells were only localized suprabasally (Figure 2c–e), confirming our observations in skin tissue. The functional significance of WIF1 overexpression in interfollicular KSCs is of great interest given the current evidence for an increase in Wnt signaling in wound healing, psoriasis, and basal cell carcinomas (Gudjonsson et al., 2010; Arwert et al., 2012). Indeed, activating mutations in Wnt pathway family members contribute to the progression of several major human cancers (Moon et al., 2004). Surprisingly, Wnt/beta-catenin signaling is reportedly crucial for the development of epidermal appendages, but not the interfollicular epidermis (Huelsken et al., 2001; Nguyen et al., 2009). However, abundant expression of wnt 5a, one of several wnt ligands, has been reported in normal adult skin (Pourreyron et al., 2012). Immunostaining of both neonatal and adult skin also revealed abundant expression of the canonical wnt ligand activating Wnt signal (Wnt3A) throughout the epithelium and in some dermal cells (Figure 1i and j). Given the observation that WIF1 is expressed in quiescent LRCs of adult human skin and differentiating keratinocytes, we next addressed the possibility that it may negatively regulate keratinocyte proliferation in the presence of Wnt3A. Immunostaining cultured keratinocytes for WIF1 revealed that its expression was lost in proliferating cultured keratinocytes (data not shown), consistent with the mRNA data shown in Figure 1a and b (TA/CP fraction). Therefore, we looked at the effects of exogenous WIF1 on keratinocytes. We synchronized the cell cycle activity of neonatal human foreskin keratinocytes in culture using the double thymidine block method, arresting them in late G1 phase (Figure 2l). Subsequent treatment of arrested keratinocytes with 20 ng of Wnt3A for 2 hours at 37 °C released them from the G1 arrest, i.e., a decrease in the number of cells in G1 of 26% and an increase in cells in % S+G2M of 19% (Figure 2m). Whereas treating cells with 3 μg WIF1 alone did not maintain the cells in G1, simultaneous treatment with WIF1 and the Wnt ligand Wnt3A blocked the re-entry of cells into S-phase (16%, Figure 2n, compared with 25% with Wnt3A alone, Figure 2m). Concomitantly, the combined Wnt3A and WIF1 treatment resulted in the accumulation of cells in G1 (44 vs. 24% in Wnt3A control) and a decrease of cells in G2M (16 vs. 31% in Wnt3A control; Figure 2m and n). These data show that WIF1 negatively regulates cell cycle progression in human keratinocytes in the presence of Wnt ligands, indicating a potential role in keratinocyte quiescence and differentiation. It has been reported that WIF1 achieves G1 cell cycle arrest in cancer cells, by the transcriptional downregulation of cell cycle regulatory Wnt target genes such as c-myc and Skp2 (S-phase kinase associated protein 2) (Tang et al., 2009). c-myc is a repressor of p21/WAF1 (van de Wetering et al., 2002), whereas Skp2 regulates p27/Kip1 degradation (Gstaiger et al., 2001)—thus, downregulation of c-myc and Skp2 results in accumulation of p21 and p27 (Tang et al., 2009)—two major regulators of CDK2, the main cyclin-dependent kinase controlling G1- to S-phase progression (Pfeuty et al., 2008). Consistent with this, we observed that the downstream effectors of the Wnt pathway, i.e., c-myc and Skp2 were not expressed in KSCs by reverse transcription–PCR (Figure 1b). Further, probing cell lysates of WIF1-arrested keratinocytes in western blots revealed that Wnt3A/WIF1 treatment resulted in increased p21 protein levels (Figure 2o) demonstrable quantitatively (Figure 2p). Thus, WIF1 may achieve its cell cycle arrest in keratinocytes at least in part through derepression of p21 transcription. In conclusion, we report that WIF1 is, to our knowledge, previously unreported as a marker of interfollicular KSCs, and that it inhibits cell cycle progression in human keratinocytes even in the presence of activating Wnt signals (Wnt3A). Although canonical Wnt signaling appears to be dispensable during development in the interfollicular epidermis (Huelsken et al., 2001; Nguyen et al., 2009), our data suggest that inhibition of Wnt signaling may be required for keeping interfollicular stem cells quiescent and differentiating cells from proliferating during homeostasis. The authors state no conflict of interest. This work was supported by National Institutes of Health grant RO1 AR050013-01A2 to PK and, in part, by contract research ‘Adulte Stammzellen II’ of the Baden-Württemberg Stiftung to PB. We thank Dr Sarah Ellis for her skilled assistance with confocal microscopy.