Vitamin D receptor mediates DNA repair and is UV inducible in intact epidermis but not in cultured keratinocytes.

To the Editor

While providing a powerful approach for studying epidermal biology, cultured keratinocytes may imperfectly model a three-dimensional epidermis in which cells are architecturally ordered. We report two important examples of the limitations of cultured keratinocytes in understanding vitamin D receptor (VDR) photobiology in murine skin. Recently, the vitamin D signaling pathway has been implicated in skin cancer prevention through its role in cellular responses to ultraviolet B radiation (UVB)-induced DNA damage, and demonstrations that VDR−/− mice are susceptible to UVB-induced epidermal tumors (Ellison ; Mason ; Quigley ; Teichert ). VDR's transactivation of certain genes is also mediated by a subunit of the nucleotide excision repair (NER)/transcription factor, TFIIH (Drané ), further suggesting a potential interaction between VDR and DNA repair.

We examined the dependence of NER on VDR in detail in several model systems. First, wild-type and VDR−/− mice (Teichert ) were irradiated with UVB, and removal of the most common UVB photoproduct, the cyclobutane pyrimidine dimer (CPD), was monitored by immunofluorescence. 1 hour post-UVB, both wild-type and VDR−/− mice exhibited significant CPD levels in epidermal keratinocytes (Fig. 1a). In wild-type epidermis, CPDs were markedly diminished by 24 hours and undetectable by 48 hours post-UVB. In contrast, in VDR−/− epidermis, CPDs persisted at 24 hours, and were still clearly detectable at 48 hours, indicating impaired NER. CPD quantification indicated that even as early as 1 hour post-UV wild-type epidermis had fewer CPDs than VDR−/− epidermis (Fig. 1b).

Figure 1

VDR supports NER in intact epidermis but not cultured keratinocytes

(a,b) After 4000 J/m2 UVB, 2 day old wild-type (C57BL/6) and VDR−/− mouse skin was harvested and stained at varying times with primary anti-CPD antibodies (1:1000, Cosmo Biosciences) and fluorescein-labeled secondary antibodies (a). Mean anti-CPD signal 1 hour post-UV (b). C,D) VDR expression in floxed VDR keratinocytes expressing cre recombinase (cre) or luciferase (con) under K14 promoter control assayed by RT-PCR normalized to L19 expression (c) or Western blotting (d) (Teichert ). (eg) Floxed VDR keratinocytes expressing cre or control cells were irradiated with 177 J/m2 UVB, and assayed immediately (e) or over time for CPDs (e,f) and 6–4PPs (e,g). (h,i) Epidermal explants from2 day old mice with floxed VDR expressed control luciferase (con) or cre under K14 control. Following 354 J/m2 UVB, explants were assayed for CPDs (h) and 6–4PPs (i) at 0 or 46 hours. A subset of control (con+HU) and VDR knockout (cre+HU) explants were treated with hydroxyurea before and following UVB (h). Higher UVB dose for explants were needed to generate CPD levels comparable to those in cultured cells, likely due to scattering/reflection in whole epidermis.

To facilitate quantitative analysis, we also explored the role of VDR in DNA repair in vitro. Keratinocytes cultured from mice bearing floxed VDR and expressing cre recombinase did not significantly express VDR relative to control cells (Fig. 1c,d). UVB-irradiated cells were assayed for CPDs and the pyrimidine(6,4)pyrimidone photoproducts (6–4PPs) using a standard immunoblot assay (Yeh and Oh, 2002). In vitro, where it was possible to harvest cells within seconds following irradiation, no differences in initial CPD or 6–4PP levels were discernible between wild-type and VDR-negative keratinocytes (Fig. 1e), and both cell types were completely deficient in global genomic NER of CPDs over 48 hours, though equally proficient in repair of 6–4PP (Fig. 1f,g). These results agree with previous observations that cultured rodent cells possess poor global genomic NER of CPDs (Tang ).

We then studied explanted epidermal sheets that better preserve skin architecture than do cultured cells while providing a more easily manipulable model system for quantitatively assessing VDR effects than whole animals. Following harvest (Teichert ), explants from mice bearing floxed VDR and cre were irradiated through the stratum corneum and incubated for 46 hours before measurement of photoproducts by immunoblot assay. Floxed VDR explants expressing cre were significantly deficient in both CPD and 6–4PP repair (Fig. 1h,i). Hydroxyurea treatment did not significantly affect CPD levels at 46 hours, indicating that the observed CPD loss reflected true repair rather than dilution of DNA damage through replication, consistent with a lack of PCNA staining in explants (data not shown).

VDR has also been reported to be UVB inducible in vivo (Hong ; Lesiak ; Mallbris ), while others have reported that VDR is down regulated by UVB in vitro (Courtois ). We studied this difference in behavior in our mouse model systems. Irradiation of wild-type mice induced epidermal VDR mRNA levels 2–3.5-fold by 24 hours (Fig. 2a). In epidermal explants derived from wild-type mice, UVB induced VDR mRNA >6-fold by 24 hours, and sustained that level over 48 hours (Fig. 2b). As anticipated, VDR-negative explants exhibited undetectable VDR expression. These results indicate that UVB strongly induces VDR expression, consistent with in vivo mouse and human data. In contrast, cultured mouse keratinocytes only weakly induced VDR mRNA expression following UVB (Fig. 2c). Consistent with prior results from cultured human keratinocytes, VDR protein levels increased only slightly at low UVB doses and then actually appeared to decrease at moderate doses, though the observed differences at all UVB doses were not statistically significant (Fig. 2d).

Figure 2

UV induction of VDR in intact epidermis but not cultured keratinocytes

(a) Wild-type mice were irradiated with increasing UVB doses, and epidermis was harvested at 24 hours for VDR mRNA levels by RT-PCR, normalized to L19 expression. (b) Epidermal explants from wild-type (WT, •) and VDR−/− (■) mice were irradiated with 354 J/m2 UVB, and assayed at 0, 24 and 48 hours for VDR mRNA levels, normalized to GAPDH expression. (c,d) Cultured keratinocytes from floxed-VDR mice expressing control (•) or cre (■) were irradiated with increasing UVB doses, and assayed for VDR expression by RT-PCR, normalized to GAPDH expression (C) and by Western blotting, with results quantified and normalized to actin (d).

These results confirm that VDR is strongly induced by UVB in intact epidermis, consistent with its role in promoting NER (Ellison ). These activities, however, are not reflected in cultured keratinocytes monolayers. Similarly, sonic hedgehog is expressed and repressed by vitamin D in epidermal explants but not in cultured keratinocytes (Teichert ). Cultured rodent cells have been commonly regarded as being defective in CPD repair; this phenotype has been ascribed to rodent cells' inability to express DDB2 which encodes a DNA damage recognition protein (Hanawalt, 2001; Tan and Chu, 2002; Tang ). However, qualitative studies of DNA repair in intact epidermis typically demonstrate that repair of epidermal CPDs appears to proceed efficiently (Fig. 1 and e.g., ref. (Mitchell )), suggesting that DDB2 expression is actually not limiting in intact mouse epidermis. It has also been reported that cultured mouse keratinocytes but not fibroblasts are actually repair proficient and express DDB2 (Pines ). Interestingly, this prior study utilized keratinocytes grown on a layer of fibronectin and collagen, and it is possible that extracellular matrix or intercellular interactions or another as yet undefined tissue-related factor may specifically modulate NER activity in epidermal keratinocytes.

In summary, VDR is a UVB-inducible gene that critically supports NER activity in intact murine epidermis, but these activities are poorly recapitulated in keratinocytes cultured from the same animals. The use of epidermal explants may represent an approach that preserves the biological behavior of epidermis while providing a facile substrate for detailed molecular studies more commonly associated with cultured cells.