DNA polymerase ζ deficiency causes impaired wound healing and stress-induced skin pigmentation

Mice harboring DNA polymerase ζ–defective keratinocytes are shown to have a defect in wound healing and a striking p53-dependent migration of melanocytes to the skin following UV radiation or wounding.


Introduction
Many types of DNA damage are impediments to DNA replication (1). One mode of DNA damage tolerance uses specialized DNA polymerases to bypass DNA damage (2). Arguably, the most critical of these specialized enzymes is DNA polymerase ζ (pol ζ ). Evidence is accumulating that pol ζ is frequently used in response to circumstances that stall DNA replication forks. Such circumstances include encounters with template DNA damage or with DNA sequences that interrupt replication, such as chromosomal fragile sites (3,4) and stable non-B DNA structures (5,6). In the absence of Rev3l, which encodes the catalytic subunit of pol ζ , cells accumulate DNA double-strand breaks and chromosome aberrations (7,8). Inactivation of Rev3l results in embryonic lethality in mice (9,10).
In cells from adult animals, elimination of pol ζ function has varying consequences, depending on the target tissue. Disruption of Rev3l in the hematopoietic system is not well tolerated (11,12,13). Mice are viable, however, after ablation of Rev3l in epithelial tissues using Cre recombinase driven by a keratin-5 promoter (14). These mice have substantially increased baseline levels of DNA damage stress, and their keratinocytes are extremely sensitive to killing by UV radiation (14).
Cells can proliferate in the absence of pol ζ in some circumstances. For example, T antigen-immortalized MEFs have a loss of checkpoint controls and can proliferate without functional Rev3l, but at the expense of an order of magnitude increase in frequency of DNA double-strand breaks and chromosome rearrangements (9,14). Similarly, some human cancer cell lines can propagate following inactivation of REV3L, but with increased sensitivity to many DNA-damaging agents (15). One consequence of the genome protective function of pol ζ is that Rev3l is a suppressor of spontaneous tumor formation. An increased incidence and reduced latency of lymphomas was observed in Tp53-defective mice with mosaic deletion of Rev3l in the lymphocyte lineage (11). Conditional partial deletion of Rev3l in the mammary gland enhances the incidence of mammary gland tumorigenesis (11). Deletion of Rev3l in keratin-5-expressing cells results in spontaneous squamous cell carcinomas in~90% of mice (14).
These experiments suggest that pol ζ may be necessary for continued proliferation of primary cells, even in the absence of exogenous DNA damage. We, therefore, examined wound healing, a process dependent on rapid proliferation responses. We report here the discovery that Rev3l-defective epidermis exhibits a proliferative defect in wound healing. Furthermore, we found that both wound healing and UV irradiation in adult mice engender striking epidermal skin pigmentation, caused by stress-induced migration of Rev3l-proficient melanocytes to the Rev3l-defective epidermis. This proliferation-associated pigmentation is p53 dependent and is absent in p53-defective areas of the epidermis. This study extends the function of pol ζ to the process of proliferation during wound healing. The results suggest Rev3l-deleting mice as a tool for dissecting communication between damaged keratinocytes and melanocytes, including signaling relevant to human disease.

Results
Abnormal wound architecture and proliferation defects in Rev3ldeficient skin To determine whether a defect in pol ζ affects adult cellular proliferation in the absence of externally applied DNA damage, we carried out in vivo and ex vivo wound healing experiments. We used adult Rev3l −/lox mice carrying one disrupted Rev3l allele and one functional allele with essential residues flanked by loxP sites for targeting by Cre recombinase (Fig 1A). Control mice were Rev3l +/lox , so that one normal Rev3l copy remained after disruption of the floxed allele. As an indicator of Cre recombinase action, the mice also harbored the mT/mG transgene (16). Constitutively expressed RFP is deleted upon expression of Cre, allowing expression of GFP ( Fig 1B). Cre recombinase was driven by the bovine keratin-5 (BK5) promoter which is active in epithelia (17) but leaves Rev3l unaffected in the dermis (Fig 1C), melanocytes, and other non-epithelial tissues (14).
Excisional wounds were introduced into the dorsal skin of experimental and control mice, and wound re-epithelialization was followed through the steps of cell migration, proliferation, and differentiation. Defects in migration, but not proliferation, are associated with chronic nonhealing wounds (18). GFP fluorescence of the dorsal skin was monitored noninvasively daily ( Fig 1D) to measure closure of the wound, which took place at the same rate in Rev3l deleting and proficient epidermis (Fig 1E). To examine proliferation and migration more closely, the skin of BK5.Cre; Rev3l −/lox and Rev3l +/lox mice was examined at day 3 of wound healing (Fig 1F). Migration of keratinocytes in the skin adjacent to the wound occurred in animals from both genotypes, as shown by staining with cytokeratin-6 (K6), a marker of migrating epithelial cells (19). In the epidermis immediately adjacent to the 3-d-old wound, DNA synthesis as measured by incorporation of bromodeoxyuridine (BrdU) was similar in the two genotypes ( Fig 1G). However, epidermal cells that had migrated into the wound area showed markedly reduced proliferation in Rev3l-deficient skin, with 50% of the control level of BrdU incorporation (Fig 1G).
The architecture of the closed wound (day 10) was markedly abnormal in BK5.Cre; Rev3l −/lox mice. The basal epidermis was discontinuous and disorganized, contrasting with the organized stratification seen in BK5.Cre; Rev3l +/lox skin (Fig 2A). Staining with cytokeratin-10 (K10), which marks the suprabasal but not the basal Cre show loxP-mediated recombination (marked by GFP expression) in epidermal cells, but not dermal or panniculus carnosus cells. (D) IVIS spectrum pictures of GFP fluorescence (shown in red) and the wound site (shown in gray) over time after wounding (from day 0 to day 10) from one mT/mG transgene-expressing mouse. (E) Quantification of wound re-epithelialization of excisional wounds in BK5.Cre; Rev3l +/lox and BK5.Cre; Rev3l −/lox mice (n = 5 mice of each genotype) over a 10-d period after wounding. (F) Keratin-6 staining of skin sections 3 d after wounding. Distance bar: 100 μm. (G) Quantification of BrdU incorporation into epidermal cells 3 d after wounding, within 1 mm of the wound edge and in epidermal cells invading into the wound area (n = 5 mice of each genotype). Data points are mean ± SEM; *P < 0.05, **P < 0.01. cells of the epidermis showed that there were fewer, larger cells in Rev3l-defective epidermis (Fig 2B). We extended the analysis by analyzing skin explants from BK5.Cre; Rev3l −/lox and BK5.Cre; Rev3l +/lox mice. GFP-positive (Rev3l-deficient) keratinocytes arising from isolated explants were monitored and measured over time after plating into culture (Fig 2C). Cell numbers were quantified by staining nuclei with Hoechst 33342 dye ( Fig 2D). Emergence of cells from the Rev3ldeficient explants stopped at day 5, whereas outgrowth from the Rev3l-proficient explants continued through day 8 ( Fig 2E). The average outgrowth area for all the skin explants was equal for the two genotypes up to day 5, but Rev3l-deficient outgrowths contained only about half the number of keratinocytes as controls (Fig 2E). In summary, wound healing in Rev3l-deficient epidermis is aberrant because of a marked decrease in proliferation of keratinocytes.
Abnormal epidermal pigmentation associated with stress stimuli in Rev3l-deleting skin Melanocytes normally reside in the hair follicles of adult mice and export melanin to hair-producing keratinocytes (20). Unlike (E) Quantification of the total outgrowth area from skin explants at 5-8 d after plating and cell density at 5 d after plating (n = 3 mice per genotype). Data points are mean ± SEM; *P < 0.05, **P < 0.01. human skin, normal-haired mouse skin remains unpigmented (constitutively or in response to UV radiation) because melanocytes are absent from the interfollicular epidermis (21,22). Unexpectedly, we found abundant melanin pigment in the basal epidermis of Rev3l-deleting skin after wounding (Figs 3A and S1). To analyze the stimuli leading to pigmentation, BK5.Cre; Rev3l −/lox and Rev3l +/lox C57BL/6 mice were exposed to UVB radiation. Mice with intact Rev3l in the epidermis showed no UV-induced pigmentation (as expected), but melanin was deposited abundantly in the basal epidermis of Rev3l-deficient UV-irradiated mice (Figs 3B and S1). We also treated mice with 12-O-tetradecanoylphorbol-13-acetate (TPA), a phorbol ester that promotes cell proliferation and inflammation via the protein kinase C pathway (23). TPA promoted proliferation in Rev3l-proficient mouse skin but had only a minor effect on proliferation in Rev3l-deficient epidermis (14). We found that treatment of mice with TPA for 2 wk also caused pronounced pigmentation of Rev3l-deleted skin, with melanin observed in the epidermis of these mice (Figs 3C and S1).
Analysis of the time course of UV radiation-induced pigmentation showed that pigmentation appeared by 4 d after exposure (Fig 4). Maximal pigmentation was seen at 6 d and persisted for at least 10 d (Fig 4). Occasional patches of non-pigmented skin (Fig 4, arrow) appeared over time following UV irradiation in the Rev3ldeficient mice, as discussed below. Therefore, multiple stressors cause an abnormal epidermal pigmentation response in Rev3ldeficient skin.

Melanocytes migrate to the epidermis of stressed Rev3l-deficient skin
To determine whether melanocytes are present in the epidermis of the stressed pigmented skin of BK5.Cre; Rev3l −/lox mice, we stained for TRP2 (product of the gene Dct), an enzyme involved in melanin synthesis that is specific for detection of differentiated melanocytes and melanocyte stem cells (20,24,25). To avoid interference from melanin, we examined skin from previously described nonpigmented albino mice that had been treated with TPA or with UV radiation (14). TRP2-positive cells were very rare in unchallenged Rev3l-proficient or Rev3l-deficient epidermis. TRP2-positive cells were present in the basal layer of Rev3l-deficient epidermis 3 d after UV irradiation (Fig 5A and D), after 2 or 40 wk of TPA application ( Fig 5B and E), or 3 d after wounding (Fig 5C and F). These melanocytes apparently originated from hair follicles, where they are normally resident in mice ( Fig S2).

P53 stabilization in Rev3l-deficient keratinocytes
Proliferation in the absence of Rev3l causes accumulation of breaks in genomic DNA, which activate p53 and other DNA damage response factors (7,14). In the BK5.Cre-driven model used here, this activation of DNA damage signaling occurs in wounded, UVirradiated, or TPA-treated keratinocytes. The melanocytes are Rev3lproficient in these mice (they do not express BK5.Cre). Therefore, it is important to gain insight into the signaling from Rev3l-defective keratinocytes that triggers an unusual response in the Rev3lproficient melanocytes.
Because melanocytes are normally resident in the hair follicles, we measured the proportion of hair follicles in undamaged skin that had p53-positive cells in the bulge region (which contains keratinocyte and melanocyte stem cells) or in the hair bulb. In Rev3l-proficient hair follicles, only weak p53 staining was observed in fewer than 30% of the bulge area cells and less than 20% of the hair bulbs. In contrast, hair follicles in Rev3l-deficient skin had very strong p53 staining in almost 100% of the bulge and hair bulb regions (Fig 6A and B). In addition to the stabilization of p53, the increased DNA double-strand breaks in proliferating pol ζdefective cells were detected by γ-H2AX formation (14). We found γ-H2AX-positive cells in similar regions of the Rev3l-deficient hair follicles (Fig 6C and D).
Adult mice have epidermal melanocytes only in the feet, the tail, and the ears. Several mouse models have been described with constitutively activated p53 arising from endogenous stress, and these animals have more darkly pigmented feet, referred to as the "sooty foot" pattern (26,27,28). We, therefore, examined foot pigmentation in the BK5.Cre-deleting mice. Indeed, darker pigmentation was observed in the footpad of animals with both Rev3l alleles inactivated in epidermal keratinocytes ( Fig S3). This increased melanin production in a location that naturally harbors melanocytes in the epidermis is consistent with constitutive activation of p53 in Rev3l-defective epidermis.
Keratinocytes secrete signaling molecules that affect the proliferation, migration, and melanin production of melanocytes (25). These include KITL, pro-opiomelanocortin-alpha (POMC, also known as alpha-melanocyte stimulating hormone), and endothelin 1 (EDN1), all of which have gene transcripts directly regulated by p53 (27,29,30); transcription of human transforming growth factor, beta 1 (TGFB1) and WNT signaling pathway genes can be indirectly affected by p53 (31,32). We isolated RNA from the epidermis and hair follicle keratinocytes of untreated, Rev3l-deficient, and Rev3lproficient skin from pigmented mice and measured the expression levels of these genes. The RT-PCR confirmed the absence of Rev3l expression in keratinocytes from the BK5.Cre; Rev3l −/lox mice (Fig 6E). We found an increase in Kitl expression in the Rev3l-deficient compared with proficient keratinocytes (Fig 6F) and a possible increase in Edn1 (Fig 6G). No change was observed in Tgfb1 levels ( Fig 6H), whereas little Wnt1 or Pomc signal was detected. The enhanced expression of the melanocyte-regulating gene Kitl in the untreated keratinocytes of Rev3l-deficient mice is likely to contribute to the increased signaling to melanocytes.
P53 and γ-H2AX accumulate in stressed Rev3l-deficient epidermis and hair follicles When present in the epidermis, melanocytes deposit pigment in response to signals from epidermal keratinocytes. The pigmentation response that follows UV radiation of the skin of Rev3ldeficient mice is preceded by increased γ-H2AX staining (reflecting DNA damage) and highly activated p53 in epidermal and hair follicle keratinocytes (14). By immunohistochemical staining, we found that wounding, like UV radiation, increased the number of γ-H2AXand p53-positive cells in Rev3l-deficient mice compared with Rev3lproficient mice in both the interfollicular epidermis (Fig 7) and the hair follicles ( Fig S4).
The data are consistent with a mechanism whereby persistent strong signaling from keratinocytes causes the activation and migration of melanocytes from hair follicles to epidermis and deposition of melanin. When melanocytes are already resident in the epidermis, as in the mouse footpad, inactivation of Rev3l in keratinocytes is sufficient to promote melanization. The baseline level of DNA damage stress without additional radiation or wounding is not sufficient to cause melanocyte migration and melanin deposition in the interfollicular epidermis. However, in mice with Rev3ldefective epidermal keratinocytes, a very strong signal is caused by damage and proliferation induction following UV radiation that drives migration of melanocytes from hair follicles and deposition of melanin in the epidermis. An intermediate but pronounced signal is provoked by wound healing of the interfollicular epidermis.
TPA caused an increase in γ-H2AX formation, proliferation, DNA breaks, and p53 stabilization in the hair follicles, but not in the epidermis of Rev3l-deficient skin (Figs 7 and S4). Rev3l-deficient epidermis has innately poor proliferative capacity, which is likely why TPA does not stimulate significant proliferation (14) or increase DNA breaks and p53 stabilization. The mechanism of TPA as a pro-pigmentation signal in Rev3l-deficient skin may involve known activities such as enhancement of cleavage of KITL to its soluble, signaling form (33) and activation of the protein kinase C pathway (34). Time course of pigmentation following irradiation of mice with one functional copy of Rev3l (top row) or with inactivation of both copies in K5-expressing keratinocytes (bottom row). The white arrow points to an example of a non-pigmented area. At day 10, Rev3lproficient mice began to regrow hair, resulting in the mildly darkened appearance. The right-most panel shows evidence of epidermal pigmentation after 2 wk of TPA application.
Melanocytes are not recruited to areas that lack p53 activation in Rev3l-deficient skin The experiments summarized above suggest that the signaling cascade causing melanocytes to proliferate, migrate, and produce melanin is initiated by p53 stabilization in response to DNA breaks. If so, cell clones without p53 activation would not signal to melanocytes.
We observed areas of Rev3l-deficient epidermis that showed no p53 stabilization after UV irradiation. These p53-unresponsive areas provided an opportunity to test the role of p53 in melanocyte activation. Patches of skin with and without p53 accumulation after UV radiation were analyzed (Fig 8A). The p53-negative areas proliferated following UV irradiation (Fig 8B), even while having high levels of DNA breaks as demonstrated by γ-H2AX staining (Fig 8C). This lack of p53 activation allows damaged cells to escape cell death and proliferate.
Rev3l-deficient basal epidermis with low proliferation and high p53 activation was populated with melanocytes (TRP2-positive cells) (Fig 8D). However, areas containing epidermal keratinocytes with high proliferation (BrdU incorporation) and low or no stabilized p53 had very few melanocytes (Fig 8D). This may account for the non-pigmented areas in the Rev3l-deficient skin that become more apparent with time after UV irradiation (Fig 4).

DNA pol ζ is required for normal wound healing
This study tested whether wound healing, a response that requires epidermal cell proliferation, is dependent on pol ζ function. We found that wound healing was defective in Rev3l-deficient mouse epidermis, even in the absence of externally applied DNA damage. Rev3l-defective keratinocytes could migrate, allowing an initial epidermal wound closure step. However, their pronounced proliferative defect once resident in the wound bed led to architecturally aberrant, disordered wound healing. This included deficits in the basal layer of the epidermis, hypocellularity, and the presence of enlarged suprabasal cells. Consistent with this, skin explants of Rev3l-deficient keratinocytes were not able to sustain significant outgrowth. Although migration of keratinocytes from explants did occur, the cells within the outgrowth were unable to proliferate for more than a few days.
The results support the interpretation that pol ζ inactivation is incompatible with the rapid proliferation of normal mammalian cells with intact cell cycle checkpoints. For example, the rapidly dividing and developing mouse embryo loses viability during gestation in the absence of pol ζ . A broad inactivation of Rev3l in hematopoietic progenitors in the adult also does not give rise to propagating B or T cells (11), although disruption is tolerated in mature resting B cells expressing CD19 or CD21 (12,13). The experiments described here were possible because mice can survive elimination of Rev3l function in epidermal keratinocytes. However, these animals display baseline skin abnormalities, including reduced epidermal cellularity and irregular hair cycling, with gradual hair loss (14).
Hyperactivation of a p53-dependent pigmentation pathway in pol ζ-defective epithelium We discovered an unexpected consequence of proliferation in the absence of pol ζ : stress-induced epidermal pigmentation. Pigmentation after UV radiation of human skin protects the cell nucleus against additional DNA damage by absorbing further radiation (20). In mice, UV radiation-induced tanning does not normally occur in the bulk of the epidermis. One major reason for this is that, in contrast to human skin, melanocytes are not resident throughout the epidermis (20,21). Instead, melanocytes in the mouse are largely localized in the hair follicles. Exceptions are the feet, the tail, and the pinnae of the ear, which do harbor epidermal melanocytes that pigment the epidermis upon DNA damage stimulation in a p53-dependent manner (29). Several genetic models or developmental stages can alter melanocyte distribution and the propensity for stress-induced pigmentation. The skin of murine neonates (<10 d old) contains widespread epidermal melanocytes that respond to UV radiation or wounding by proliferating and pigmenting the epidermis (22,35). Activation of KITL, for example, by expression in keratinocytes with a K14 promoter, leads to localization of melanocytes throughout the epidermis (36,37). Pigmentation by melanin deposition further requires signaling to melanocytes by POMC, which activates melanocytes via cAMP (37). Both the KITL and POMC pathways appear necessary for skin pigmentation in adult mice.
Several mouse models with increased p53 stabilization have darker pigmentation in the non-haired skin of mice, particularly the feet. These include models where p53 is stabilized by a decrease in the p53-negative regulator MDM2 or where cellular stress is elevated by impairing ribosomal function (26,27,28,29,30). Mice with a catalytically-dead DNA protein kinase subunit have impaired DNA break repair and also show p53-dependent pigmentation of non-haired skin (38). In these cases, the stabilization of p53 in epidermal areas that already have resident melanocytes is enough to stimulate those melanocytes to produce pigment. In these models, however, there is no migration of melanocytes to the epidermis of unchallenged haired skin. We find here that mice with Rev3l-deficient epidermal keratinocytes have a baseline level of DNA damage stress that is sufficient to activate p53 and initiate pigmentation in non-haired skin by resident epidermal melanocytes (Fig 9).
Rev3l-deficient hair follicles contain melanocytes that respond rapidly to epidermal stress. Persistent DNA damage in the hair follicle in Rev3l-deficient mice appears to promote a response of melanocytes to stress signals from epidermal keratinocytes. The nature of this stress signal is not clear but may involve the activation of the protein kinase C pathway (as is seen with TPA treatment) or the creation of a cytokine gradient caused by UV radiation-, wounding-, or TPA-induced epidermal inflammation. The process of epidermal pigmentation in the haired skin of Rev3l-deficient mice appears to be p53 dependent. This is supported by the lack of melanocytes or pigmentation in p53-negative areas of UV-irradiated epidermis. The origin of these p53-negative areas is not known, but they are unlikely to be p53-mutant patches because the pAb240 antibody used in this study recognizes both wild-type and mutant conformations of p53. P53 is likely to be a key player in epidermal pigmentation because in keratinocytes P53 directly up-regulates the transcription of genes that encode proteins that are secreted and bind to the melanocortin 1 receptor on melanocytes. These include KITL (27), POMC (29), and EDN1 (30). Activation of melanocortin I receptor stimulates production of eumelanin, increases pigmentation, and may increase repair of common UV-induced DNA photoproducts (39). The activation occurs via palmitoylation by ZDHHC13, an enzyme with multiple substrates and pleiotropic effects in mouse skin (40).
We presented some evidence here that elements required for the pigmentation pathway are activated in the absence of pol ζ function. The KITL axis is up-regulated in Rev3l-deficient keratinocytes compared with Rev3l-proficient keratinocytes. The significant twofold difference is inclusive of the entire population of Rev3l-deficient skin cells and so probably underestimates the activation seen in the target proliferating cells that have an activated p53 pathway.

Relevance to human pathologies
It is important to understand pathways affecting skin pigmentation, given the roles of melanocytes in protecting against UV radiation damage and in the pathogenesis of melanoma in humans. The strong signaling invoked by a pol ζ defect is expected to be useful in models studying melanomagenesis, wound healing problems, and pigmentation abnormalities (including genetic syndromes with overactive KITL).
A summary of the consequences of pol ζ deficiency in keratinocytes is outlined in Fig 9. The most remarkable feature is the damage-induced migration of melanocytes to the epidermis in mice that have a disruption of Rev3l in the keratinocytes. The precise mechanism is unknown, but apparently occurs via signaling to Rev3l-disrupted hair follicles (possibly through a cytokine gradient). Keratinocytes in both the epidermis and the follicles are under baseline DNA damage stress as shown by DNA breaks and p53 activation. Epidermal wounding, or UV radiation, causes attempted proliferation of Rev3l-defective epidermal keratinocytes, which greatly increases DNA breaks and p53 activation. This generates a signal strong enough to cause the migration of melanocytes to the epidermis. Stabilized p53 activates the expression of propigment genes, such as Kitl and Pomc, which produce proteins that signal to melanocytes. Once in the epidermis, melanocytes pigment the epidermal keratinocytes in response to DNA damage/stress signals.

Mouse strains and breeding
All animal work in this study were completed according to the University of Texas MD Anderson Cancer Center Institutional Animal Care and Use Committee guidelines. Experiments done on pigmented mice used animals on a C57BL6/J background with residual strain 129 background. Mice expressed the BK5.Cre transgene (17) and the mT/mG Cre reporter transgene (16), with either the Rev3l wild-type (+), knockout (−), or floxed (lox) alleles. All other experiments (non-pigmented UV radiation and TPA experiments) were carried out using tissues from the animals and experiments described (14). These mice were on an albino strain background (tyrosinase mutation) that was at least 95% FVB/N. The FVB strain mice that originate from the Taconic laboratory (FVB/T) have an additional mutation in Skint1, which leads to a defect in skin γ-δ T cells (41). We genotyped for this mutation and found that our FVB/ N mice carried the wild-type allele of Skint1 as expected.

Ex vivo wound healing experiments
Ex vivo wound healing experiments were conducted as described (42,43). 7-10-wk-old pigmented mT/mG mice with the genotypes BK5.Cre; Rev3l +/lox and BK5.Cre; Rev3l −/lox were euthanized by CO 2 asphyxiation followed by cervical dislocation, and their dorsal skin was removed. Twelve 4-mm punch biopsies (Accupunch; Xemax Surgical Products) from each mouse skin were taken and placed into a 24-well tissue culture plate. 200 μl explant growth medium/ well was added to each well and the explants were placed into a 37°C, 5% CO 2 -humidified incubator for 16 h. An additional 1.5 ml of explant growth medium was then added to each well. Over the next 4 d, the keratinocytes began to migrate out of the explants. Daily starting at day 5, the expanding explants were photographed using a Nikon TS-100 fluorescent microscope equipped with a DS-L2 camera at 40× magnification, detecting GFP (keratinocytes) and RFP (fibroblasts) fluorescence. Each picture had a scale bar added to it, and all the pictures from one explant (up to 18 pictures per explant) were merged using the Adobe Photoshop CS4 Photomerge function. The area of cell dispersion was measured from these merged pictures. In addition, at 5 d after plating, Hoechst 33342 dye (10 μg/ml) was added to six explants to label live cell nuclei, and these were incubated for 1 h at 37°C. Pictures were then taken each of the GFP (keratinocytes) and Hoechst 33342 (blue) fluorescence on the Nikon fluorescent microscope at 100× magnification. These pictures were merged using the Photoshop Photomerge function, the area of outgrowth was measured, and then the number of Hoechst-stained nuclei were counted to obtain the number of cells per area of outgrowth. The average number of cells per area of outgrowth for each explant was multiplied by total outgrowth area to obtain the total number of cells per explant.
In vivo wound healing experiments 7-10-wk-old pigmented mT/mG mice with the genotypes BK5.Cre; Rev3l +/lox and BK5.Cre; Rev3l −/lox were used. The dorsal skin of the mice was shaved and 24 h later the skin was depilated with Nair (Church & Dwight). The following day, the mice were anesthetized with isoflurane and a strip of siliconized tape (3M) was placed across their shoulder blades (this prevents wound retraction caused by contraction of the muscles under the skin). 5-mm wounds were made using an Accupunch through the tape. While under anesthesia, GFP fluorescence in the epidermis of these mice was imaged on an IVIS spectrum (using epifluorescence with filter set 491/501), which allowed us to examine the wound area and its changing size during the process of re-epithelialization. Pictures were taken of the GFP fluorescence on the IVIS spectrum every 24 h for 10 d. The wound area of these mice was measured over time using Adobe Photoshop. After the fluorescent pictures were taken on the 10 th day, the mice were injected i.p. with BrdU (100 μg/g bodyweight) and 30 min later were killed by CO 2 asphyxiation followed by cervical dislocation. The dorsal skin was removed and fixed with 10% (wt/vol) formalin in PBS. After 24 h, the skin was moved to 70% (vol/vol) ethanol, and then was paraffin-embedded, sectioned, and stained. Another cohort of mice was wounded, as above, and then allowed to heal for 3 d before they were injected with BrdU and killed (no IVIS spectrum pictures were taken). The dorsal skin of these mice was also removed, and fixed, sectioned, and stained as above.

Exposure of mice to UV radiation
The 7-10-wk-old pigmented mT/mG mice were shaved and their skin was depilated with Nair 48 h before UV radiation. The mice were put into single-animal Plexiglass containers in a carousel, so that each animal received an equivalent dose of UV radiation. They were irradiated with 1,200 J/m 2 UVB using Phillips UVB TL 20W/12 RS lights at a fluence of~5 J/m 2 ⋅s; immediately afterward, their dorsal skin was photographed with a Canon EOS Rebel XSi digital SLR camera; and the mice were returned to their cages. Every 24 h for 10 d, the dorsal skin of the mice was photographed. On the 10 th day, the mice were injected with BrdU (100 μg/g bodyweight) and 30 min later were killed by CO 2 asphyxiation, followed by cervical dislocation.

Quantification of immunohistochemical staining of mouse tissues
Using the digital images of the slides, skin sections were chosen and antibody-positive cells were quantified from BrdU-, p53-, or γ-H2AX-stained slides (14). For the keratin-10 stain, the healed wound areas in the 10-d samples were selected in the digital slide. From there, the total epidermal cell number was counted and then was compared with the total cells that were epidermal but not staining for K10 (basal epidermal cells). Cell size was quantified by measuring total epidermal cell area (using the ImageScope program) and then dividing by the total number of epidermal cells in that area. TRP2-positive cells were quantified by selecting 11 scattered 1-mm sections of skin from the digital images, then counting the number of TRP2-positive epidermal cells in that 1-mm section. In the experiments quantifying the hyper-proliferative and control areas of skin, TRP2 was quantified by using the ImageScopepositive pixel counter and then comparing this with the total epidermal area. This was done because the epidermal area could be different between the hyper-proliferative and control skin. Hair follicles with p53-or γ-H2AX-positive cells in their bulge or hair bulb regions were also quantified. The bulge region was defined as the area of hair follicle a short distance below the sebaceous gland. The hair bulb is the region of the hair follicle immediately adjacent to the dermal papillae.

Isolation and gene expression of keratinocytes
Pigmented 7-10-wk-old mT/mG mice were shaved and their skin was depilated 48 h before sacrifice. The animals were killed by CO 2 asphyxiation followed by cervical dislocation, and then the dorsal skin was removed. It was processed into a single-cell suspension by digestion with collagenase (Sigma-Aldrich), dispase (Sigma-Aldrich), and trypsin (Thermo Fisher Scientific) as in reference 44. Briefly, the skin was minced, washed with PBS, and incubated for 12-13 h in a collagenase/dispase solution. The cells were further incubated for 5 min with trypsin, followed by mechanical disruption using 18-, 20-, and 22-gage needles. The cells were incubated for 5 min with red blood cell lysis buffer (BioVision) and then washed and flow-sorted for RFP-positive (dermal) cells or GFP-positive (epidermal and hair follicle keratinocyte) cells. RNA was extracted from these cells and gene expression was assayed using an Applied Biosystems 7900HT Fast Real-Time PCR System. We measured expression of murine Rev3l

Statistical analysis
Analysis of the outgrowth area or cell number of the explants, or the TRP2-, p53-, or γ-H2AX-positive cells in UV-irradiated or TPA-treated skin by immunohistochemistry was performed using a two-way ANOVA (P < 0.05). Statistical analysis of the immunohistochemical staining of the wounds or undamaged hair follicles or the gene expression analysis was performed by unpaired t test. Comparison of immunohistochemical staining in the hyper-proliferative, p53positive and p53-negative epidermal areas was performed using one-way ANOVA with Tukey's multiple comparisons test (P < 0.05).