Accelerated elimination of ultraviolet-induced DNA damage through apoptosis in CDC25A-deficient skin

Abstract

Cell division cycle 25A (CDC25A) is a dual-specificity phosphatase that removes inhibitory phosphates from cyclin-dependent kinases, allowing cell-cycle progression. Activation of cell-cycle checkpoints following DNA damage results in the degradation of CDC25A, leading to cell-cycle arrest. Ultraviolet (UV) irradiation, which causes most skin cancer, results in both DNA damage and CDC25A degradation. We hypothesized that ablation of CDC25A in the skin would increase cell-cycle arrest following UV irradiation, allowing for improved repair of DNA damage and decreased tumorigenesis. Cdc25a ^fl/fl /Krt14-Cre recombinase mice, with decreased CDC25A in the epithelium of the skin, were generated and exposed to UV. UV-induced DNA damage, in the form of cyclopyrimidine dimers and 8-oxo-deoxyguanosine adducts, was eliminated earlier from CDC25A-deficient epidermis. Surprisingly, loss of CDC25A did not alter epidermal proliferation or cell cycle after UV exposure. However, the UV-induced apoptotic response was prolonged in CDC25A-deficient skin. Double labeling of cleaved caspase-3 and the DNA damage marker γH2A.X revealed many of the apoptotic cells in UV-exposed Cdc25a mutant skin had high levels of DNA damage. Induction of skin tumors by UV irradiation of Cdc25a mutant and control mice on a skin tumor susceptible to v-ras ^Ha Tg.AC mouse background revealed UV-induced papillomas in Cdc25a mutants were significantly smaller than in controls in the first 6 weeks following UV exposure, although there was no difference in tumor multiplicity or incidence. Thus, deletion of Cdc25a increased apoptosis and accelerated the elimination of DNA damage following UV but did not substantially alter cell-cycle regulation or tumorigenesis.

Abbreviations:

8-oxo-dG

8-oxo-deoxyguanosine

ATR

ataxia telangiectasia and Rad3-related

BrdU

bromodeoxyuridine

CDC25A

cell division cycle 25A

CDK2

cyclin-dependent kinase 2

CPD

cyclopyrimidine

DAPI

4ʹ,6-diamidino-2 phenylindole

TBS-T

Tris-buffered saline Tween 20

TUNEL

terminal dUTP nucleotide end-labeling

UV

ultraviolet

Introduction

One of the most serious assaults on the skin is from overexposure to UV in the form of sunlight. UV exposure causes most non-melanoma skin cancer, which is the most common type of cancer in the USA (1). Chronic exposure to UV, including UVB and to a lesser extent UVA, is the major cause of non-melanoma skin cancer because of its DNA-damaging capabilities. Exposure to UVB causes damage to DNA primarily in the form of cyclobutane pyrimidine dimers (CPDs) (2), which causes most UVB-induced mutations (3). Another type of UV-induced DNA damage occurs due to the production of reactive oxygen species. Increasing evidence suggests that the immunosuppressive effects of reactive oxygen species cause conversion of solar keratoses to squamous cell carcinoma (4). A well-known biomarker of oxidative stress is the oxidized guanosine moiety, 8-oxo-deoxyguanosine (8-oxo-dG (5,6)). Point mutations at both the oxidized guanine nucleotide itself and at the base adjacent to the 8-oxo-dG have been found in the Ras oncogene in skin and other cancers in response to repair failure (reviewed in ref. 7).

Coordinated regulation of DNA repair and cell-cycle progression is crucial in preventing mutations and cancer. The cell division cycle 25 (CDC25) family of phosphatases, which include CDC25A, CDC25B and CDC25C, regulate progression through the cell cycle and are important for maintaining genomic stability in response to DNA damage (reviewed in ref. 8). Functional overlap in CDC25 family members is probable, since Cdc25b and Cdc25c null mice develop normally. In contrast, Cdc25a null mice are embryonic lethal, consistent with an essential role for CDC25A in cell-cycle progression. These data also suggest that CDC25A may be an important target for cancer therapy (8).

In the normal cell, the removal of the inhibitory phosphates on CDK2 (cyclin-dependent kinase 2) by the protein tyrosine phosphatase CDC25A allows for the activation of CDK2-Cyclin E and CDK2-Cyclin A, enabling advancement into and progression through S-phase (9). In response to UV-induced DNA damage, ataxia telangiectasia and Rad3-related is activated, which leads to degradation of CDC25A (9). Without CDC25A to activate cyclin-CDK complexes, which can result in cell-cycle arrest. Cell-cycle arrest allows time for DNA repair prior to DNA replication, thus preventing mutations (9). Aside from bypassing a mechanism for cell-cycle arrest, overexpression of CDC25A suppresses the cell’s response to oxidative DNA damage and stress, which is known to increase cancer risk (10). Further evidence of a role for CDC25A in cancer has been provided by immunohistochemical studies in human cancer. Increased CDC25A protein is found in breast, hepatocellular, ovarian, thyroid, esophageal and colorectal cancers, with its overexpression frequently correlating with increased malignancy and poor prognosis (8). Inhibitors of CDC25 phosphatase activity inhibit growth of transplantable rat hematomas in vivo (11), an indication of their promise for clinical use.

Because of the role of the CDC25A phosphatase in DNA damage checkpoints, we hypothesized that deletion of Cdc25a in the skin would increase cell-cycle arrest following UV irradiation, allowing for improved repair of DNA damage and decreasing skin tumorigenesis. In order to investigate this hypothesis, the response of mice with a skin-targeted deletion of Cdc25a to UV irradiation was examined. This investigation revealed that deletion of Cdc25a accelerated elimination of UV-induced DNA damage, increased apoptotic elimination of cells with DNA damage, reduced skin tumor growth at early time points, but surprisingly did not alter cell-cycle regulation or skin tumor multiplicity.

Materials and methods

Animals

Cdc25a ^fl/fl mice with loxP sites inserted flanking exons 1–3 of the Cdc25a gene were developed on a C57BL/6 × C3HF1 background as described in (12). Cdc25a ^fl/fl mice were crossed with Krt14 promotor-driven Cre recombinase mice on an FVB/N background (The Jackson Laboratory, Bar Harbor, ME) and progeny backcrossed to generate skin-targeted Cdc25a knockouts (Cdc25a ^fl/fl/Cre ⁺) and controls (Cdc25a ^fl/fl/Cre ^-or Cdc25a ^wt/wt/Cre ⁺ ). v-ras ^Ha transgenic Tg.AC mice on an FVB/N background were crossed with the Cdc25a mutants to produce Tg.AC⁺ Cdc25a mutants (Cdc25a ^fl/fl/Cre ⁺/Tg.AC⁺ and Cdc25a ^fl/fl/Cre ⁺/Tg.AC⁻) and controls (Cdc25a ^fl/fl/Cre ⁻/Tg.AC⁺ and Cdc25a ^fl/fl/Cre ⁻/Tg.AC⁻). Genotyping was performed by PCR using the following primers: 5ʹ-CAG AGC CTG AAG TCC TGT GAA GG-3ʹ and 5ʹ-CTG CCT AGT GTA GTT CCT ACA GCG-3ʹ (Cdc25a (12)); 5ʹ-ACC AGC CAG CTA TCA ACT CG-3ʹ and 5ʹ-TTA CAT TGG TCC AGC CAC C-3ʹ and 5ʹ-CTA GGC CAC AGA ATT GAA AGA TCT-3ʹ and 5ʹ-GTA GGT GGA AAT TCT AGC ATC ATC C-3ʹ (Krt14-Cre, the last two primers produce a positive control band in all samples (13)); and 5ʹ-ATT CTG AAG GAA AGT CC-3ʹand 5ʹ-TGG ACA AAC TAC CTA CAG-3ʹ (Tg.AC (14)).

UV irradiation and tumor measurements

Eight-week-old Cdc25a mutant and control Tg.AC⁻ mice (for short-term experiments) or 20-week-old Cdc25a control and mutant Tg.AC⁺ mice (for the tumor experiment) were sham or UV irradiated once a week over a period of 4 weeks for a cumulative exposure of 4.3 kJ/m². Cdc25a mutant and control Tg.AC⁻ mice exposed to 1 kJ/m² or sham irradiated were euthanized 18h later. UV irradiation was generated using Ultraviolet B TL 40W/12 RS bulbs (The Richmond Light Company, Richmond, VA) covered with Kodacel triacetate film to block UVC (Eastman Kodak, Rochester, NY), resulting in approximately 30% UVA and 70% UVB measured using Oriel Goldilux radiometric photodetector probes (Newport Corporation, Irvine, CA). Dorsal hair was clipped the day before UV irradiation using a Wahl clipper (Wahl Clipper Corporation, Sterling, IL) and close shaved with a Braun Series 3 shaver (Proctor and Gamble, Cincinatti, OH) just before exposure. Mice were injected with approximately 0.075mg per gram body weight bromodeoxyuridine (BrdU) (Sigma-Aldrich, St.Louis, MO) 1h before euthanasia. Tumor diameter in two dimensions was measured using calipers and tumor size calculated as length times width. All procedures were approved by the Creighton University Institutional Animal Care and Use Committee.

Immunoblotting

Epidermis was separated from the skin using the heat shock method as described in (15). Protein was extracted from tumors or epidermis in lysis buffer containing 10mM Tris (pH 7.4), 150mM sodium chloride, 10% glycerol, 1% Triton X-100, 1mM ethylenediaminetetraacetic acid, 1X Complete Protease Inhibitor Cocktail (Roche, Germany), 1mM sodium orthovanadate (ICN Biomedicals, Aurora, OH), 1.5 µM ethylene glycol tetraacetic acid (Sigma, St.Louis, MO) and 10 µM sodium fluoride using a tissue homogenizer (Tissue Master 125, Omni International LLC, Bedford, NH). Protein was quantified using Coomassie Brilliant Blue G-250 (Bio-Rad, Hercules, CA). Immunoblotting was performed using standard techniques with antibodies recognizing CDC25A (1:200 Santa Cruz, Santa Cruz, CA), actin (Sigma-Aldrich, St.Louis, MO) or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Cell Signaling, Boston, MA). Horseradish peroxidase–conjugated secondary antibodies (Cell Signaling, Beverly, MA), chemiluminescent reagents (Pierce, Rockford, IL) and autoradiography were used to visualize results. Ponceau S staining (Sigma-Aldrich, St.Louis, MO) and actin immunoblotting confirmed the evenness of loading and transfer. Densitometry was performed using ImageLab software (Bio Rad Life Science Research, Rochester, NY).

Immunostaining, TUNEL and morphometric analyses

Hematoxylin and eosin (Sigma-Aldrich, St.Louis, MO) staining was performed using standard techniques. Ethanol-fixed tissue sections were incubated in BrdU (1:3, Becton Dickinson, Franklin Lakes, NJ) or CPD (1:1000, Kamiya Biomedical, Seattle, WA) antibody and formalin-fixed sections were incubated in cleaved caspase-3 (Cell Signaling, Beverly, MA) or phospho-histone H2A.X (Millipore, Billerica, MA) antibody. Sections were then incubated in Alexa-Fluor 488 (Invitrogen, Carlsbad, CA) or biotinylated F(Ab)₂ fragment goat antimouse secondary (Jackson ImmunoResearch, West Grove, PA), followed by Texas-Red Streptavidin (Vector Laboratories, Burlingame, CA) and 4ʹ,6-diamidino-2-phenylindole (DAPI) in mounting media (Molecular Probes, Invitrogen, Carlsbad, CA). Terminal transferase dUTP nick-end labeling (TUNEL) (Promega, Madison, WI) was performed following the manufacturer’s protocol. Immunohistochemistry for CDC25A was performed using anti-CDC25A antibody (Santa Cruz, CA), horseradish peroxidase–conjugated secondary antibody (Vector Labs, Burlingame, CA), diaminobenzidine substrate (Sigma Chemical Corp.) and hematoxylin counterstain (Sigma Chemical Corp). For some experiments, cleaved caspase-3 and γH2A.X immunofluorescence were performed as described above and results photographed. Coverslips were removed and slides were washed overnight with agitation with several changes of phosphate-buffered saline and incubated in stripping solution containing 200 µM glycine and 3.5 µM sodium dodecyl sulfate (Fisher Scientific), before immunohistochemistry for CDC25A. Controls for the immunostaining and TUNEL included replacing enzyme with buffer for TUNEL and replacing primary antibody with buffer for immunostaining. BrdU- and TUNEL-positive cells were quantified by counting positively stained cells and the DAPI-positive basal cells in each section using a ×20 objective with the investigator blinded as to the identity of the sample.

Flow cytometric analysis

Cell-cycle analysis was performed using flow cytometry of 50 µm sections from formalin-fixed paraffin-embedded blocks as described in (16). The sections were deparaffinized, rehydrated, digested in 0.05% pepsin (Sigma, St.Louis, MO) in phosphate-buffered saline with pH 1.5 and then incubated in Tris Buffered Saline (pH 7.6), Ribonuclease-A, propidium iodide and IGEPAL CA-630 (Sigma, St.Louis, MO (17)) before analysis using a FACSCalibur flow cytometer (Becton Dickinson, Franklin Lakes, NJ) and CellQuest analysis (BD Biosciences, San Jose, CA).

Southwestern assays

Epidermis was collected by liquid nitrogen scraping of flash-frozen skin as in (18). DNA was extracted from the epidermis by incubating in buffer containing 200 µg/ml Proteinase K and 100mM NaCl in 10% SDS, 50mM EDTA and 100mM Tris (pH 8.0). DNA was precipitated in 4.21M NaCl, 0.63M KCl and 10mM Tris (pH 8.0) with two volumes 100% ethanol. DNA was resuspended in 0.4mM NaOH and 0.1mM EDTA (Sigma, St.Louis, MO), heated to 95°C and neutralized with 2M ammonium acetate. DNA was blotted onto a nitrocellulose membrane, dried, blocked in 5% bovine serum albumin in Tris-buffered saline with 0.1% Tween-20 (TBS-T), incubated in primary antibody (Anti-8-oxo-dG, 1:500, Japan Institute for Control on Aging, Shizuoka, Japan or Anti-Thymine Dimer, 1:500, Kamiya Biomedical Company, Seattle, WA), washed in TBS-T, incubated in horse antimouse IgG horseradish peroxidase–linked secondary (1:10 000, Cell Signaling Technologies, Beverly, MA) in 5% bovine serum albumin /TBS-T. Blots were washed in TBS-T and signal was detected by incubation with LumiGLO chemiluminescent Reagent (1:1, Cell Signaling Technologies) using a BioRad Chemidoc XRS+ (Bio Rad Life Science Research, Hercules, CA) or by exposure of Kodak BioMax Film (Carestream Health, Rochester, NY). Densitometry was performed using ImageLab software (Bio Rad Life Science Research).

Statistical analyses

A Student’s two-tailed t-test or two-way analysis of variance (ANOVA) were used to determine statistical significance. P ≤ 0.05 was considered to be significant.

Results

Deletion of Cdc25a-accelerated elimination of DNA damage from UV-exposed skin

To test our hypothesis that deficiency of CDC25A allows for more efficient repair of DNA damage following UV irradiation, skin-targeted Cdc25a mutant mice were developed by breeding mice homozygous for loxP sites inserted flanking exons 1 through 3 of Cdc25a (Cdc25a ^fl/fl) with Krt14 promoter-driven Cre recombinase (Cre ⁺) mice. Cdc25a mutant mice (Cdc25a ^fl/fl /Cre ⁺) were healthy and viable, with no obvious phenotype. Immunoblotting revealed reduced CDC25A protein in mutant epidermis (Figure 1A). CDC25A protein was localized primarily in the epidermis of control mice, with both nuclear and cytoplasmic staining apparent (Figure 1B), as has been previously described in other tissues (19). Cdc25a mutant skin, in contrast, had patches of CDC25A positivity intermixed with CDC25A negative cells in the epidermis (Figure 1B). Groups of Cdc25a control and mutant mice were exposed to UV once a week for 4 weeks and DNA damage in the form of cyclopyrimidine dimers was detected using a CPD-specific antibody in immunofluorescence assays. CPDs were not detected in sham-irradiated skin (Figure 2A). After 12h of UV irradiation, CPDs were detected in the basal and suprabasal keratinocytes of the epidermis, as well as fibroblasts of the superficial dermis (Figure 2A). In order to quantify CPDs, Southwestern assays were performed using isolated epidermis, revealing levels of cyclopyrimidine dimers that peaked 2h following UV irradiation in both mutant and control mice (Figure 2B and 2C), corresponding to the rapid formation of cyclopyrimidine dimers in both mutant and control mouse skin. Consistent with our hypothesis, mutant Cdc25a skin eliminated cyclopyrimidine dimers from the epidermis more rapidly as evidenced by the return to background, sham-irradiated levels by 36h following UV, whereas controls still had elevated CPDs at 72h (Figure 2A and 2C).

Fig. 1.

Decreased CDC25A protein in Cdc25a mutant epidermis. (A) Epidermis from Cdc25a mutant and littermate control mice was immunoblotted as indicated. Each lane is protein from a separate mouse. 
(B) Immunohistochemistry for CDC25A (brown) in untreated Cdc25a mutant and littermate control mice with a hematoxylin counterstain (blue). Scale bar indicates 50 µm.

Fig. 2.

Improved elimination of UV-induced DNA damage in CDC25A-deficient epidermis. (A–D) Groups of 8-week-old Cdc25a mutant and littermate control mice were exposed four times for a cumulative dose of 4.3 kJ/m² UVA/B or were sham irradiated and euthanized at the indicated time points following the last exposure (N ≥ 3 mice). Representative images for immunofluorescence for CPD shown, with CPD in red and DAPI in blue (A). Southwestern blots using CPD (B) or 8-oxo-dG (D) antibodies are shown, where each spot is DNA from a separate mouse. DNA (0.25 µg) from each mouse was blotted in duplicate on each membrane (not shown). Quantification of multiple Southwestern assays for CPD (C) or 8-oxo-dG (E) was averaged. *Indicates a significant difference compared to the corresponding UV-exposed control group using ANOVA, where P ≤ 0.05. Scale bar indicates 60 µm.

Another form of DNA damage resulting from the formation of reactive oxygen species in response to UV irradiation is the 8-oxo-dG adduct (6). Southwestern assays were performed using an 8-oxo-dG antibody in order to quantify relative levels of 8-oxo-dG in Cdc25a control and mutant epidermis. Surprisingly, sham-irradiated mouse skin had more 8-oxo-dG than at 2h after UV irradiation for both Cdc25a controls and mutants (Figure 2D and 2E). Thereafter, levels peaked at 12h post-UV before declining. The 8-oxo-dG adducts were removed from Cdc25a mutant epidermis earlier than from controls, with significantly less in mutants compared with controls at 72h following UV (Figure 2E). Consistent with our hypothesis, CPDs and 8-oxo-dG adducts were removed from Cdc25a mutant skin more quickly than from controls.

Deletion of Cdc25a did not reduce cutaneous proliferation

In order to determine whether accelerated removal of UV-induced DNA damage upon deletion of Cdc25a was due to cell-cycle arrest in Cdc25a mutant epidermis, both the cell proliferation and the cell cycle were examined. Quantification of BrdU-positive cells as a measure of cell proliferation revealed no difference in the number of BrdU-positive cells in sham-irradiated Cdc25a mutant and control epidermis (Figure 3A and 3B). There was little change in BrdU labeling after UV irradiation in either genotype until 18h (Figure 3B). By 18h following UV irradiation, proliferation began to increase and then peaked at 36h post-UV irradiation in both genotypes before declining (Figure 3A and 3B). Surprisingly, there was no significant difference in BrdU incorporation between the Cdc25a mutants and controls at any time point following UV irradiation (Figure 3A and 3B).

Fig. 3.

Unaltered proliferative response to UV irradiation upon deletion of Cdc25a. Groups of 8-week-old Cdc25a mutant and littermate control mice were exposed four times for a cumulative dose of 4.3 kJ/m² UVA/B or were sham irradiated and euthanized at the indicated time points following the last exposure. Representative images for BrdU immunofluorescence (A) and quantification of BrdU labeling index (B) are shown. Red is BrdU and blue is DAPI (A). Bar indicates 60 µm (A). DNA content flow cytometry was performed on 50 µm skin sections and the percentage of G₀/G₁ (C), S-phase (D) and G₂/M cells (E) are shown. *Indicates a significant difference when compared with the corresponding UV-exposed control group using two-way ANOVA, where P ≤ 0.05 (E).

To further examine the effect of CDC25A on the cell cycle, DNA content flow cytometry was performed. Similar to the BrdU incorporation results, the percentage of cells in S-phase increased significantly between 2h and 36h post-UV irradiation for Cdc25a mutants (P = 0.01), with a corresponding decrease in G₀/G₁-phase cells, and then S-phase cells declined (Figure 3C and 3D). Although there were no significant differences in a direct comparison of the genotypes, S-phase cells did not increase significantly between 2h and 36h post-UV irradiation in the control genotype (P = 0.60). Interestingly, mean G₂/M-phase cells were significantly reduced in UV-exposed Cdc25a mutant compared with control mice (Figure 3E). Taken together, these results reveal surprisingly little effect of CDC25A on the cell cycle following UV exposure.

UV-induced apoptosis was prolonged in Cdc25a mutants compared with controls

Given that elimination of DNA damage was improved in the absence of cell-cycle arrest in Cdc25a mutant skin, we hypothesized that ablation of Cdc25a might increase apoptosis in response to UV irradiation. The number of TUNEL-positive epidermal cells increased in Cdc25a controls following UV exposure, peaking at 12h and then decreasing to near baseline levels by 18h (Figure 4A, arrowheads, Figure 4B). An increase in TUNEL positivity in the superficial cells of the dermis was also detected after UV irradiation (Figure 4A). Although the numbers of Cdc25a mutant TUNEL-positive epidermal cells were similar 12h after UV exposure, mutant skin had significantly more TUNEL-positive cells at 18h following UV in comparison with control skin (Figure 4A, arrowheads, Figure 4B). Staining of cleaved caspase-3 positive cells also demonstrated maximal apoptosis in Cdc25a control and mutant epidermis 12h post-UV irradiation (Figure 4C, green nuclei). In addition, increased apoptosis in mutant epidermis compared with control was confirmed at 18h post-UV irradiation (Figure 4C, green nuclei).

Fig. 4.

Increased UV-induced apoptosis of cells with DNA damage in Cdc25a mutant epidermis. (A–D) Groups of 8-week-old Cdc25a mutant and littermate control mice were exposed four times for a cumulative dose of 4.3 kJ/m² UVA/B or were sham irradiated and euthanized at the indicated time points following the last exposure. Representative images for TUNEL (A) and quantification of TUNEL labeling index in the epidermis (B) are shown. Blue is DAPI and green is TUNEL-positive cells (A). (A) Arrowheads indicate TUNEL-positive epidermal cells and scale bar indicates 60 µm. (B) *Indicates a significant difference when compared with the corresponding UV-exposed control group using two-way ANOVA, where P ≤ 0.05 (B). (C) Representative images of double labeling for cleaved caspase-3 (green) as a measure of apoptosis, γH2A.X (red) as a measure of DNA damage, and DAPI (blue) to identify nuclei are shown. Photomicrograph showing red fluorescence only is shown below the multicolor panels at 12 and 18h for better visualization of γH2A.X immunofluorescence. Arrows indicate cleaved caspase-3- and γH2A.X-double positive cells. Scale bar indicates 60 µm. (D) CDC25A immunohistochemistry (top), cleaved caspase-3 immunofluorescence (middle) and γH2A.X immunofluorescence (bottom) is shown in Cdc25a mutant skin 18h post-UV. The arrow indicates a CDC25A-negative keratinocyte that is positive for caspase-3 (green) and γH2A.X (red).

To determine whether increased apoptosis in mutant skin was a mechanism for the elimination of cells with DNA damage, double labeling of cleaved caspase-3, an apoptotic marker, and γH2A.X, a DNA damage marker, was performed. Many cleaved caspase-3 positive cells were also positive for phosphorylated histone γH2A.X (Figure 4C, arrows). Triple labeling for CDC25A, cleaved caspase 3 and γH2A.X revealed some cells lacking CDC25A that were positive for both cleaved caspase-3 and γH2A.X (Figure 4D, arrows). Surprisingly, apoptotic cells were also detected adjacent to, as well as distant from, cells with DNA damage (Figure 4C). Thus, genetic deletion of Cdc25a in murine epidermis increased apoptotic death of cells with DNA damage following UV irradiation.

Decreased tumor growth in Cdc25a-deficient mice without altered tumor incidence or tumor multiplicity

In order to examine the influence of CDC25A on UV-induced skin tumorigenesis, Cdc25a mutants and controls were crossed with v-ras ^Ha transgenic Tg.AC mice, which were genetically initiated to develop skin papillomas after minimal exposure to UV (20,21). These transgenic mice have markedly increased susceptibility to the development of UV-induced skin tumors, allowing for easy quantification of tumors in response to UV. Mutations in c-ras ^Ha occur in up to 46% of human non-melanoma skin cancers (22–25), supporting the relevance of the Tg.AC model for skin cancer research. Tumors were first observed in UV-exposed Cdc25a control mice at 3 weeks after the start of UV irradiation and at 4 weeks in the mutants (Figure 5A). By 5 weeks, Tg.AC⁺/Cdc25a mutants had an average of 1 tumor per mouse and controls had about 2.5 tumors per mouse. Tumor number in the controls peaked at 6 weeks with a mean of 4.3 tumors per mouse. In the mutants, tumor number peaked at 7 weeks with nearly 4 tumors per mouse (Figure 5A). Tumor multiplicity decreased after this primarily due to euthanasia of moribund mice or of mice with a heavy tumor burden. There were no statistical differences in tumor multiplicity or tumor incidence between genotypes at any time point (Figure 5A-B). Between 4 and 6 weeks following the start of the experiment, Cdc25a mutants had significantly smaller tumors as compared with controls (Figure 5C). However, between 7 and 9 weeks post-UV irradiation, a large decrease in the mean tumor area for the Cdc25a control group upon euthanasia of one mouse with a high tumor burden eliminated this difference (Figure 5C). At later time points, tumor areas were similar in the two genotypes, as some mice were euthanized for health reasons, and tumor area did not significantly differ between the two groups for the duration of the experiment. Histopathological characterization revealed that most lesions were well-circumscribed, exophytic tumors with marked papillomatosis, acanthosis and variable hyperkeratosis (Supplementary Figure 1 is available at Carcinogenesis Online). Mitotic activity was minimal and no epithelial atypia or invasion was noted. A single squamous cell carcinoma in situ was identified in each of the UV-exposed Cdc25a mutant and control groups.

Fig. 5.

Smaller tumors in UV-exposed Cdc25a mutant skin but no difference in tumor incidence or multiplicity. (A–F) Groups of 20-week-old Cdc25a mutant and littermate control mice were exposed four times for a cumulative dose of 4.3 kJ/m² UVA/B or were sham irradiated and tumor development monitored (N = 10 mice). Tumor multiplicity (A), tumor incidence (B) and mean tumor area (C) are shown. *Indicates a significant difference when compared to the corresponding Cdc25a control group. (D) Immunoblotting of papillomas using the indicated antibodies. Numbers under the panel indicate the mean for densitometry after normalization to GAPDH. Each lane is protein from a separate tumor. (E) Immunohistochemistry for CDC25A in Cdc25a mutant and control papillomas. Brown is CDC25A immunostaining and blue is hematoxylin counterstain. Scale bar indicates 50 µm.

To determine whether decreased CDC25A was maintained in the tumors, immunoblotting for CDC25A was performed with tumors obtained within the last 2 weeks of the experiment. As shown in Figure 5D, CDC25A was not decreased in the Cdc25a mutant tumors as it was in Cdc25a mutant skin (Figure 1A and 1B). Real-time RT–PCR demonstrated similar Cdc25a transcripts in tumors from Cdc25a mutant and control mice (data not shown). Because stromal CDC25A could mask decreased CDC25A protein or transcripts in the mutants in these experiments, CDC25A localization was also examined. Although CDC25A immunostaining of mutant skin revealed patches of cells with strong CDC25A staining next to regions lacking CDC25A immunopositivity (Figure 1B), CDC25A was detected uniformly in mutant tumor epithelium although with fainter signal than in control tumor epithelium (Figure 5E). CDC25B and CDC25C were also detected in control and mutant tumors but were not increased in mutants compared with controls, indicating no compensatory upregulation of other family members in the mutant tumors (data not shown). Real-time RT–PCR for Cdc25b and Cdc25c also did not reveal any compensatory increase in Cdc25a family member expression that may have been detected non-specifically by the CDC25A antibody (data not shown). Taken together, these experiments suggest a negative selection pressure for cells lacking CDC25A during UV-induced skin tumorigenesis.

Discussion

Although CDC25A is overexpressed in a number of internal cancers including breast, colorectal and esophageal cancer and associated with a poor prognosis (8), little is known about the role of CDC25A in skin carcinogenesis. The CDC25A phosphatase plays an important role in cell-cycle regulation through its deactivation of cyclin-dependent kinases. Previous experiments in our laboratory have shown that Erbb2 activation in response to UV irradiation maintains CDC25A levels, allowing for cell-cycle progression (16). Inhibition of Erbb2 leads to decreased CDC25A, increased S-phase arrest and reduced skin tumorigenesis following UV irradiation. For these reasons, the role of CDC25A in UV-induced skin carcinogenesis and the response of skin to UV were investigated.

In order to investigate the role of CDC25A in non-melanoma skin tumorigenesis, a skin-targeted Cdc25a mutant mouse was created by utilizing a Krt14 promoter-driven Cre recombinase gene. Krt14 is expressed in the basal layer of the epidermis by embryonic day 14 (26), before development of hair follicles as outgrowths of the epidermis or differentiation of the basal layer into multilayered squamous epidermis (27–29). We expected that Cdc25a would be targeted to keratinocytes of the skin, both in the epidermis and in the hair follicles. Our results are consistent with incomplete excision of the floxed allele in the cre/loxP system (30), as our immunoblotting and immunostaining results revealed some CDC25A protein in Cdc25a ^fl/fl /Cre ⁺ epidermis. Moreover, CDC25A protein was detected uniformly in mutant tumor epithelium, indicating a selection bias for Cdc25a intact cells during tumorigenesis.

The skin-targeted Cdc25a knockout model was used to investigate the hypothesis that deletion of Cdc25a increases cell-cycle arrest following UV irradiation, resulting in improved repair of DNA damage and decreased skin tumorigenesis. Consistent with our hypothesis, the UV signature CPD lesion and the oxidized base 8-oxo-dG were removed from the epidermis more rapidly in UV-exposed Cdc25a mutants compared with UV-exposed controls. Removal of DNA damage may have been due to accelerated repair, altered cell-cycle kinetics or increased apoptosis of cells with DNA damage. Although nucleotide excision repair can be used for oxidative damage (reviewed in ref. 31), CPDs and 8-oxo-dGs would normally be repaired through distinct mechanisms, nucleotide excision repair for CPDs and base excision repair for 8-oxo-dG as reviewed in ref. 5. Thus, our results suggest CDC25A modulation of cell-cycle arrest or cell death facilitates elimination of the damage rather than a direct effect on DNA damage repair pathways. Consequently, we anticipated that ablation of Cdc25a would increase S-phase arrest due to its established role in cell-cycle regulation. Surprisingly, however, deletion of Cdc25a had minimal effects on the cell cycle after UV irradiation. In addition to the analyses of proliferation following several exposures to UV, a similar analysis of BrdU incorporation was performed using Cdc25a mutant and control skin 18h after 1 kJ/m² UV irradiation, revealing no significant differences between genotypes using a Student’s t-test (P = 0.29).

In the absence of cell-cycle changes to account for improved elimination of UV-induced DNA damage, apoptotic cell death was examined. In control mice, UV-induced apoptosis peaked at 12h and then declined precipitously by 18h. In the mutant skin, in contrast, apoptosis increased at 12h and was maintained at that elevated level until 18h, as observed both through TUNEL and cleaved caspase-3 immunofluorescence. Because many cleaved caspase-3 positive cells were also positive for phosphorylated histone H2A.X, and some of these were lacking CDC25A, the elevated apoptosis would result in the elimination of cells with DNA damage. Thus, accelerated removal of CPDs and 8-oxo-dG adducts in Cdc25a mutants was likely due to increased apoptosis of damaged cells. A role for CDC25A in suppression of apoptotic cell death has previously been documented by Zou, et al. (10) who found that CDC25A interacts directly with and inhibits apoptosis signal-regulating kinase 1, which is activated in response to reactive oxygen species and regulates multiple proteins involved in the cellular stress-response pathway leading to apoptosis. Thus, CDC25A may inhibit the apoptotic response to oxidative stress. In our hands, CDC25A appeared to limit the apoptotic response to UV, although the mechanism for this effect awaits further investigation.

The influence of CDC25A on UV-induced tumor development was evaluated in v-ras ^Ha Tg.AC mice, a genetically initiated model that rapidly and uniformly develops skin tumors in response to carcinogens such as UV irradiation. The mutant transgene in this model is relevant to human skin carcinogenesis, because up to 46% of human cutaneous squamous-cell carcinomas have Ras mutations (22–25). There was no difference in tumor number or incidence between Cdc25a mutant and control genotypes. However, Cdc25a mutant mice had slower growing skin tumors, as tumor area in the mutant group was statistically less than Cdc25a controls in the first 6 weeks of the experiment. This difference disappeared at later time points, as some mice with high tumor burdens were euthanized for humane reasons. Thus, in these experiments, deletion of Cdc25a reduced tumor growth only early on but did not affect tumor number. The increased apoptosis in Cdc25a mutant keratinocytes following UV exposure may have resulted in decreased growth of tumors at early time points. Because the Cdc25a deletion is inefficient in this model, an increase in cell death in Cdc25a mutant cells may have led to the overgrowth of CDC25A-expressing cells over time. This could result in the differences between Cdc25a mutants and controls disappearing at later time points, consistent with our data. The improved elimination of DNA damage in Cdc25a mutant mice predicted decreased tumorigenesis in the mutants. The lack of decrease in tumor number upon deletion of Cdc25a may be a consequence of the Tg.AC model used, because in this model tumor development correlates strongly with proliferative and hyperplastic responses (32). Because only four exposures to UV are adequate to promote tumors in this model, UV-induced DNA damage may have had less influence on tumorigenesis than UV-induced cell proliferation. Alternatives to the Tg.AC model, such as UV irradiation of Cdc25a mutant mice on a hairless SKH-1 background, might reveal a more pronounced effect of CDC25A. Alternatively, the lack of deletion of Cdc25a in many epidermal cells (Figure 1B) may have allowed for the development of CDC25A-expressing tumors in the mutant mice, a hypothesis consistent with our immunoblotting, immunostaining and real-time RT–PCR results using mutant and control tumors. It is also possible that CDC25B and CDC25C could compensate for the loss of CDC25A in the skin, as has been shown previously in adult mice (12). However, real-time RT–PCR and immunostaining did not show increased CDC25B or CDC25C in mutant tumors (data not shown).

To summarize, decreased levels of CDC25A in UV-irradiated skin resulted in accelerated removal of DNA damage in the epidermis and increased apoptotic cell death, which were accompanied by decreased tumor growth but no change in tumor number. Inhibitors of CDC25A have been developed such as the vitamin K dual-specificity phosphatase inhibitor Cpd5 and the CDC25A-specific phenyl maleimide 20 (PM-20, (33,34)). PM-20 reduces the tumor burden in rat liver cancer in vivo (34). Thus, in other organs, CDC25A has been shown to be a promising target for anticancer therapy. The data presented here, however, document improved elimination of DNA damage, increased UV-induced apoptosis and delayed tumor growth in Cdc25a mutant skin, but no differences in the number of UV-induced skin tumors in Cdc25a mutant mice. Further experiments are required to fully evaluate the influence of CDC25A and its potential importance as a therapeutic target in skin cancer.