Protection from UV-induced skin carcinogenesis by genetic inhibition of the ataxia telangiectasia and Rad3-related (ATR) kinase

Generation of Transgenic Mice Expressing a Kinase-Inactive Form of ATR in Skin.

We generated transgenic mice expressing a kinase-dead form of human ATR (ATR-kd) under the human keratin-14 (K14) promoter (Fig. 1A) to target expression to the basal layer of the epidermis (22). The presence of the β-globin intron (associated with improved mRNA stability) (23) was required for expression of ATR-kd under the K14 promoter, perhaps because of the large size (>8 kb) of ATR. Compared with wild-type ATR, the kinase-inactive (ATR-kd) protein has a single amino-acid substitution of Asp²⁴⁷⁵ to Ala in the conserved kinase domain and thus lacks kinase activity (24). ATR-kd acts as a dominant negative: previous studies showed that the phenotype of ATR-kd expression was essentially the same as that resulting from suppression of wild-type ATR function. Specifically, ATR-kd expression (24, 25), decreased ATR expression attributable to Seckel mutation (26, 27), and ATR knockdown by siRNA (28) all showed hypersensitivity to UV and decreased phosphorylation of Chk1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Generation of K14 promoter-driven ATR-kd transgenic mice. (A) Schematic of ATR-kd transgene. FLAG-tagged, human ATR-kd was inserted into the polylinker (PL) region of the human K14 promoter/enhancer (P/E) construct containing a β-globin intron. Asterisk indicates the mutation site for inactivation of ATR kinase activity (Asp²⁴⁷⁵ → Ala²⁴⁷⁵). Solid arrows indicate the primer set used to detect the transgene in genomic DNA (B). Dashed arrows indicate the primer set used to quantify mRNA expression of the transgene (C). (B) ATR-kd transgene (ATR-kd Tg) is detected in genomic DNA from transgenic (Tg) mice, not from littermate controls (Ctrl), via PCR amplification and agarose gel analysis. (C) ATR-kd transgene mRNA is expressed in transgenic (Tg) mice, but not in transgene-negative littermate controls (Ctrl), as revealed by real-time RT-PCR analysis. (Left) Amplification plot of ATR-kd transgene. cDNA containing ATR-kd transgene sequence was used as positive control (cDNA). Samples from littermate controls (Ctrl) and negative control samples (no template) showed no amplification after 40 cycles. (Right) Amplification plot of the RNA quality control gene Gapdh. (D) FLAG-tagged ATR-kd protein is expressed in ATR-kd transgenic (Tg) mouse keratinocytes. Lysates (500 μg of total protein) of primary keratinocytes isolated from Tg mice and transgene-negative littermate controls (Ctrl) were used for immunoprecipitation (IP) with anti-FLAG antibody, followed by immunoblot (IB) analyses with the indicated antibodies. Input samples (15 μg of total protein) were from the same cell lysates but were not immunoprecipitated. The anti-ATR antibody used detects both human and mouse proteins.

Because the FVB mouse strain in which these transgenic mice were created was highly resistant to UV carcinogenesis, we crossed the ATR-kd transgenic mice into an Xpc^−/− background. Xpc^−/− mice have a global genome-repair deficiency, are sensitive to UV, and are prone to develop tumors after a relatively short period of UV treatment, recapitulating the human disease xeroderma pigmentosum complementation group C (XPC) (29). All ATR-kd transgenic mice and transgene-negative littermate controls used in this study were Xpc^−/−. After generating these transgenic mice, we extensively searched for phenotypic differences between ATR-kd transgenic mice and transgene-negative littermate controls. No obvious differences were noted in weight gain, macroscopic appearance of skin, density of hair, wound healing, or skin structures by histological examination. ATR-kd transgenic mice showed no spontaneous skin tumor development in either the FVB or Xpc^−/− background.

We first confirmed stable genomic transgene integration into the transgenic mice via detection of ATR-kd DNA by using PCR (Fig. 1B). Next, we examined the expression of ATR-kd transgene at the RNA and protein levels. Expression of transgene mRNA was quantitated by real-time RT-PCR. Transgene mRNA was detected only in ATR-kd transgenic mice, whereas expression of a control gene (Gapdh) was similar in ATR-kd transgenic mice and littermate controls (Fig. 1C). Protein expression of FLAG-tagged ATR-kd transgene was detected after enrichment by immunoprecipitation only in transgenic mice (Fig. 1D). We concluded that protein expression of the FLAG-tagged transgene was low because (i) total ATR protein bands were similar between keratinocytes derived from transgene-negative littermate controls and from ATR-kd transgenic mice (Fig. 1D) and (ii) FLAG-tagged ATR could not be detected without immunoprecipitation (Fig. 1D, Input vs. IP:αFLAG).

ATR-kd Transgene Expression in Keratinocytes Inhibits UV Responses and Augments UV-Induced Apoptosis via ATR–Chk1 Pathway Inhibition.

To investigate the effect of ATR-kd expression on UV responses in keratinocytes, we isolated primary mouse keratinocytes from ATR-kd transgenic mice and transgene-negative littermate controls. Primary keratinocytes from ATR-kd transgenic mice showed partial inhibition of UV-induced phosphorylation of Chk1, a downstream target of ATR, similar to inhibition by caffeine (nonspecific ATM/ATR inhibitor) (Fig. 2A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Primary keratinocytes from ATR-kd transgenic mice demonstrate ATR signaling inhibition and augmentation of UV-induced apoptosis. (A) ATR-kd inhibits UV-induced phosphorylation of Chk1. Primary mouse keratinocytes isolated from ATR-kd transgenic (Tg) mice and transgene-negative littermate controls (Ctrl) were treated with medium or caffeine (3 mM) for 30 min before UV irradiation. Cells were irradiated with 75 mJ/cm² of UVB light and harvested 1 h after UV exposure for immunoblot analyses with the indicated antibodies. (B) ATR-kd does not inhibit doxorubicin-induced phosphorylation of ATM. Primary mouse keratinocytes isolated from ATR-kd transgenic (Tg) mice and transgene-negative littermate controls (Ctrl) were treated with medium or doxorubicin (1 μM) for 2 h and harvested for immunoblot analyses with the indicated antibodies. Arrowheads indicate same molecular size, corresponding to the ATM protein. (C and D) ATR-kd augments UV-induced apoptosis measured by sub-2N DNA content. Primary mouse keratinocytes were isolated from ATR-kd transgenic (Tg) mice and transgene-negative littermate controls (Ctrl). Cells were treated with medium or caffeine (3 mM) for 30 min before UV irradiation. Cells were irradiated with 75 mJ/cm² of UVB light, harvested 24 h later, and stained with propidium iodide for flow cytometry analysis. (C) Representative DNA content patterns of cells without caffeine treatment with percentage of sub-2N DNA content indicated. (D) Mean of percentage of sub-2N DNA content is shown (n = 4). (Error bars = SEM.)

Given the known functional similarity between ATR and ATM, we wanted to determine whether ATR-kd nonspecifically inhibits ATM activation because such nonspecificity would complicate interpretation of these studies. We assessed ATM activation via phosphorylation at Ser¹⁹⁸⁷ in mouse ATM (which corresponds to Ser¹⁹⁸¹ in human ATM) (30). To avoid activating ATR, we used doxorubicin, a topoisomerase II inhibitor that selectively activates ATM compared with ATR. ATR-kd did not inhibit doxorubicin-induced phosphorylation of ATM (Fig. 2B). This finding demonstrates that ATR-kd selectively inhibits ATR-mediated phosphorylation, not ATM phosphorylation.

Because DNA-damaged human keratinocytes with ATR or Chk1 inhibition undergo apoptosis (28), we examined the effect of ATR-kd transgene expression on UV-induced apoptosis in keratinocytes freshly isolated from mice. Keratinocytes from transgene-positive mice showed augmentation of UV-induced apoptosis (sub-2N DNA content of 30%) compared with littermate controls (15%) (Fig. 2C). This augmentation by ATR-kd was similar to augmentation by caffeine in transgene-negative keratinocytes (32.1%) (Fig. 2D). Addition of caffeine did not further augment UV-induced apoptosis of ATR-kd keratinocytes (28.5%). These findings suggest that caffeine's ability to augment UV-induced apoptosis was mediated through ATR inhibition.

A preliminary study on apoptotic sunburn cells in epidermis was carried out in an earlier strain (FVB genetic background) of ATR-kd transgenic mice. Unlike our primary keratinocyte studies in cells derived from Xpc^−/− animals, no difference between ATR-kd transgenic mice and transgene-negative littermate controls was observed in UV-induced sunburn cell formation at 6 or 10 h after 30 mJ/cm² of UVB light.

ATR-kd Transgenic Mice Chronically Irradiated with UV Demonstrate Longer Tumor Latency and Lower Skin Tumor Incidence.

All ATR-kd transgenic mice used for chronic UV experiments were verified for transgene mRNA expression by real-time RT-PCR. The ATR-kd transgenic mouse cohort and that of transgene-negative littermate controls were carefully matched to be comparable in age, sex, and coat color between groups. All mice were verified to have Xpc^−/− genotype and were irradiated with UVB light three times a week for 40 wk (cumulative UVB dosage as shown in Fig. 3A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

ATR-kd transgene delays tumor onset and suppresses UV tumorigenesis. (A) UV irradiation timeline indicating cumulative UVB dosage. Mice were irradiated with 250–2,500 J/m² of UVB light per day, thrice weekly. (B) Percentage of tumor-free mice is shown for each group: ATR-kd transgenic (Tg) mice or transgene-negative littermate controls (Ctrl). All mice in both groups were Xpc^−/−. Kaplan–Meier curves were compared by using the logrank test for statistical significance. (C) Tumor onset is delayed in ATR-kd transgenic mice. Mean number of weeks of UV treatment required for developing an initial tumor is shown. (Error bars = SEM.) (D) ATR-kd transgene suppresses UV-induced tumor development. Mean number of tumors per mouse is shown up to 19 wk, the point when some mice with advanced tumors were killed and the cohort was no longer complete. (Error bars = SEM.) Statistical significance in mean number of tumors per mouse between the groups was as shown at the indicated time points: *P ≤ 0.05, **P < 0.01. (E) Correlation of visible tumor features with histological tumor invasiveness in a randomly selected subset of tumors at time of euthanasia. Before histologic examination, 47 tumors (22 from ATR-kd transgenic mice and 25 from transgene-negative littermate controls) were segregated into five groups as shown based on size, bleeding, and ulceration/cratering/erosion. Larger tumors (≥3 mm) with bleeding and/or ulceration/cratering/erosion were presumed to be invasive SCCs. Indeed, all 34 tumors that had visible features associated with invasive SCCs were verified pathologically to be invasive SCCs. The 14 tumors that did not meet clinical criteria for invasive SCC were found microscopically to be SCC in situ (n = 7), keratoacanthoma (n = 5), epidermal hyperplasia (n = 1), and invasive SCC (n = 1). (F) ATR-kd transgenic mice had fourfold fewer clinically defined invasive SCCs than transgene-negative littermate controls (Ctrl) did, although this result did not meet statistical significance (P = 0.071). Mean number of invasive SCCs per mouse after 23 wk of UV treatment is shown. (Error bars = SEM.)

Transgene-negative littermate controls started developing tumors after 12 wk of UV treatment (Fig. 3B). Compared with transgene-negative littermates, ATR-kd transgene-positive mice showed significantly delayed tumor development as measured by the fraction of tumor-free mice (P = 0.0029; Fig. 3B) as well as a 3-wk delay in time of onset of the first tumor (P = 0.0064; Fig. 3C). In addition, at any given time point, the average number of tumors in ATR-kd mice was significantly lower than that in transgene-negative littermate controls (Fig. 3D). ATR-kd mice demonstrated 69% fewer tumors after 19 wk of UV treatment (P = 0.019).

ATR-kd Mice Demonstrate a Trend Toward Decreased Number of Invasive SCCs.

Although tumor latency was extended in ATR-kd mice, continued chronic UV irradiation eventually caused tumor development in all transgene-negative littermate controls and transgenic mice by 34 wk of UV treatment. Tumors that developed in these Xpc^−/− mice in the presence or absence of the ATR-kd transgene were mainly keratoacanthomas and SCCs. These lesions were similar to those previously described (29). Tumors that developed in ATR-kd mice were histologically indistinguishable from those in transgene-negative littermate controls, suggesting that ATR-kd did not affect the development of specific types of tumors and did not change the tumor spectrum.

Continued UV irradiation promotes invasive progression of tumors. We developed criteria for clinically identifying invasive SCCs in UV-irradiated mice to monitor tumor progression by visual inspection without biopsy (which would eliminate developing tumors). The criteria were based on three visible tumor features: (i) ≥3-mm tumor diameter; (ii) bleeding from a tumor; and (iii) ulceration, cratering, or erosion of a tumor. The clinical definition of invasive SCC required that a tumor be ≥3 mm and also display bleeding, ulceration/cratering/erosion, or both. Histological analyses of tumors on UV-irradiated mice revealed that these clinical criteria for invasive SCC strongly correlated with histological diagnosis (Fig. 3E). Tumors without bleeding or ulceration/cratering/erosion were in situ tumors in 13 of 14 specimens, regardless of tumor diameter. All tumors that met the clinical criteria for invasive SCCs were in fact histologically invasive SCCs (33 of 33 tumors).

Using these validated visual criteria for invasive SCC, we monitored the emergence of these tumors during chronic UV treatment. We found that ATR-kd mice developed four times fewer invasive SCCs than transgene-negative littermate controls did (0.08 vs. 0.33 invasive SCCs per mouse; Fig. 3F), although statistical power was not sufficient to reach significance between the two groups (P = 0.071). We also measured latency from initial tumor development to clinically defined invasive SCC. Among all tumors that ultimately progressed to clinically defined invasive SCCs (n = 142 with 68 in transgenic mice and 74 in transgene-negative littermate controls), the average latency from initial tumor detection to invasive SCC development was 22.2 ± 1.8 d (mean ± SEM). The latency in transgenic mice (18.5 ± 2.3 d, n = 68) was somewhat shorter than that in the transgene-negative littermate controls (25.6 ± 2.8 d, n = 74), a difference that approached statistical significance (P = 0.052).