Modulation of Patched-Associated Susceptibility to Radiation Induced Tumorigenesis by Genetic Background

Although only a small proportion of cancer in adults is attributable to the effects of mutations in known cancer-susceptibility genes, these infrequent “cancer families” have been extremely informative. An increasing body of literature suggests that, in the presence of high penetrance alleles, genetics can play a major role in determining individual risk. In this regard, the study of modifier genes interacting with high penetrance genes may point to the identification of alleles contributing to cancer risk in the general population (1). Nevoid basal cell carcinoma syndrome (NBCCS) is a highly penetrant autosomal dominant trait, characterized by developmental abnormalities and predisposition to tumor formation (2). Inherited mutations of the Patched (Ptch1) gene have been identified as responsible for NBCCS (3). Affected individuals show extensive inter- as well as intrafamilial variability of the phenotype (4). Particularly, patients with the same molecular lesion can exhibit very different symptoms, suggesting that even in the presence of highly expressing familial genetic disorders low penetrance alleles, which are common in the population, are likely to contribute to the risk of developing cancer.

One of the features of NBCCS is an increased risk of skin cancer, particularly basal cell carcinoma (BCC), induced by UV and ionizing radiation (5). Ionizing radiation is a well-characterized carcinogen, capable of inducing tumors in both humans and animals (6); because radiation is used for diagnostic and therapeutic purposes, identification of specific genetic polymorphisms associated with an increased risk of radiation-induced malignancy may be of great importance.

The existence of inherited genetic susceptibility to sporadic BCC development in the general population has also been suggested (7), but the identification of genetic modifiers of tumorigenesis in humans is complicated by the interactions among numerous genes, lifestyle, and exposure to environmental carcinogens. Mouse models offer opportunities to overcome these problems and to control the genetic component through the use of engineered mutations and selective breeding (8).

Two different strains of mice in which the Ptch1 gene is inactivated through targeted disruption of exons 6–7 (9) or exons 1–2 (10) have been generated recently. Heterozygous Ptch1 knockout mice display the typical symptoms of NBCCS, including an increased propensity to develop spontaneous tumors in the brain (medulloblastomas) and soft tissues (rhabdomyosarcomas), and BCC-like tumors in the skin in response to UV or ionizing radiation exposure (11, 12). Genetic background has been shown to influence rhabdomyosarcoma development in Ptch1 heterozygotes (13). Analogously, variable incidences of spontaneous medulloblastoma (MB) have been reported depending on background (14, 15). These data show that tumorigenesis is modified by mouse strain-specific alleles interacting with Ptch1 haploinsufficiency.

As a first step toward identification of genes that control susceptibility to BCC tumorigenesis, Ptch1^neo67/+ mice on CD1 background were crossed with carcinogenesis-resistant (Car-R) and carcinogenesis-susceptible (Car-S) lines of mice, generated in our laboratory by phenotypic selection for high susceptibility or resistance to inducers of skin cancers starting from a cross of eight parental inbred strains (16, 17). Sixteen consecutive generations of selective breeding produced an interline difference in susceptibility >100-fold. The selected characters are regulated by interaction of several independently segregating genes/loci (Quantitative trait loci). Car-S and Car-R mice represent a genetically heterogeneous model, intermediate between standard inbred strains and the general population of mammals (18), and, analogously to other rodent models of genetic modification of tumorigenic response (19), their use for genetic analysis allows mapping of quantitative trait loci with higher resolution when compared with classical crosses between inbred strains (20). We have shown recently that the allele combinations obtained by phenotypic selection in the Car-S and Car-R model result in a similar pattern of susceptibility/resistance to skin tumorigenesis after radiation exposure (21).

In this study we investigate whether a tissue-specific or a stage-specific interaction between Ptch1 deficiency and genetic backgrounds with different susceptibility to skin cancer (Car lines) affects the spectrum of neoplasms in mice. We show that F1SPtch1^neo67/+ mice are highly susceptible to BCC induction after ionizing radiation exposure, whereas F1RPtch1^neo67/+ are completely resistant. The additional increase in BCC susceptibility observed in N2Car-SPtch1^neo67/+ mice (F1SPtch1^neo67/+ × Car-S) suggests a contribution from recessive-acting Car-S cancer modifiers and demonstrates a clear interaction of a rare highly penetrant mutant gene (Ptch1) and of low penetrance, potentially more common variant genes (Car-S and Car-R alleles) in radiation-induced cancer.

Mice lacking one Ptch1 allele (Ptch1^neo67/+) were derived by gene targeting of 129/Sv ES cells and maintained on the CD1 strain background (9). A colony was established in the animal facility at ENEA Casaccia (Rome, Italy) by crossing Ptch1^neo67/+ heterozygous males with CD1-wild-type (wt) females. Generation of Car-R and Car-S lines of mice has been described (17). For study of genetic background effects, Ptch1^neo67/+ heterozygous males on CD1 background were crossed to Car-S and Car-R females to generate F1SPtch1^neo67/+ and F1RPtch1^neo67/+ heterozygotes, respectively, and wt F1S and F1R mice. To generate N2 backcross mice, F1SPtch1^neo67/+ heterozygous mice were backcrossed onto the Car-S strain. About 460 mice of both sexes have been enrolled in this study. Mice were treated in accordance with the Institutional Animal Care and Use Committee. Animals were genotyped by the procedure described in Hahn et al. (9).

Mice were exposed to X-ray irradiation using a Gilardoni CHF 320 G X-ray generator (Gilardoni S.p.A., Mandello del Lario, Lecco, Italy) operated at 250 kVp, 15 mA, with filters of 2.0 mm Al and 0.5 mm Cu (HVL = 1.6 mm Cu).

In a first experiment F1SPtch1^neo67/+ and F1RPtch1^neo67/+ and respective wt littermates (F1S-wt and F1R-wt) of 60 days of age were subjected to local irradiation of the dorsal skin with a single dose of 4 Gy X-rays. Anesthetized mice were positioned on a lead plate, and shaped lead shields (4-mm thickness) were placed on each mouse to provide protection to the body parts to be spared. A dorsal skin area of 3 cm² was irradiated through a trapezoidal window of a lead shield.

In a second experimental series F1SPtch1^neo67/+ and N2Ptch1^neo67/+ and respective wt littermates (F1S-wt and N2-wt) of 4–8 days of age were whole-body irradiated with a single dose of 3 Gy X-rays. Control groups were left untreated and observed for the whole experimental length to determine the spontaneous rate of BCC development.

There was no increase in acute mortality associated with radiation treatments. The summary of treatment schemes and size of experimental groups are reported in Table 1.

Mice were observed daily for their whole life span. Upon decline of health (i.e., severe weight loss, paralysis, ruffling of fur, or inactivity), or when tumors were visible, they were sacrificed and autopsied. Grossly normal appearing dorsal skin samples (5 cm²), brain, and any other visible mass were fixed in 4% buffered formalin. Samples were then embedded in paraffin wax according to standard techniques. Serial sections of skin tissues were cut at 7-μm thickness and stained with H&E.

Kaplan-Meier survival curves of Ptch1 heterozygous and wt mice were compared and the log-rank test Ps were calculated. The significance of differences in frequencies between tumor types was evaluated by Fisher’s exact test. P < 0.05 was considered significant.

Samples from BCCs >0.5 cm in diameter and noncancerous tissue were snap frozen. DNA was extracted using the NucleoSpin Tissue kit (Macherey-Nagel, Duren, Germany). Amplification of the Ptch1 wt and targeted alleles was performed as described previously (14).

PCR primers for chromosome 13 markers were purchased from Research Genetics (Huntsville, AL). Because the mouse crosses used in this study were obtained from genetically heterogeneous lines of mice, we could not rely on microsatellite PCR product size available for inbred strains,⁵ and allelic polymorphisms had to be tested for each microsatellite marker. Approximately 50 microsatellites spanning the length of chromosome 13 were used to examine 6 BCCs from F1SPtch1^neo67/+ irradiated mice. PCR experiments were performed as described previously (22) and reactions conducted on a Hybaid PCR Express thermal cycler. A typical PCR experiment involved analysis of DNAs from tumors and corresponding genomic DNA from normal tissue as control.

To simplify representation of Ptch1 deletions, whenever a noninformative marker was localized between two retained markers it was considered as retained. The same criterion was applied to noninformative markers localized between two lost markers. Only when a noninformative marker was localized between one retained and one lost marker was it represented as noninformative.

For skin tumor induction, 8–10-week-old F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice and wt littermates were initiated with a single 9,10-dimethyl-1,2-benzanthracene (Sigma Chemical Co., St. Louis, MO) application on the dorsal skin that had been shaved 3 days before. Promotion was started 7 days after initiation and continued with two weekly applications of 12-O-tetradecanoylphorbol-13-acetate (Sigma Chemical Co.) dissolved in acetone for topical application to the shaved area. In view of the very large difference in susceptibility between Car-R and Car-S, 9,10-dimethyl-1,2-benzanthracene and 12-O-tetradecanoylphorbol-13-acetate doses applied were higher in the F1R (25 μg and 5 μg, respectively) than in the F1S strain (2.5 μg and 1 μg, respectively). Tumor parameters were expressed as the percentage of mice bearing one or more papillomas (tumor incidence) and the mean number of papillomas in the total mouse population (tumor multiplicity ± SE). Four months after the end of promotion mice were inspected for recording of papilloma progression to carcinoma.

We examined whether the Car-S and Car-R genetic backgrounds influence the survival and tumorigenesis of unirradiated and irradiated Ptch1^neo67/+ populations, monitored for their entire life span for the presence of tumors. Fig. 1 shows the survival of unirradiated (Fig. 1,A) and irradiated (Fig. 1,B) Ptch1^neo67/+ and wt mice on F1S and F1R backgrounds. In the unirradiated groups, median survival times were 76 and 70 weeks in F1SPtch1^neo67/+ and F1RPtch1^neo67/+ heterozygotes, respectively, and 97 weeks in both F1S and F1R wt mice. Differences in survival between heterozygous and wt mice were highly statistically significant for both F1S (P = 0.0013) and F1R (P < 0.0001) mice, indicating that Ptch1 haploinsufficiency strongly affects mouse life span. In contrast, differences between unirradiated F1S and F1R heterozygotes and between unirradiated F1S and F1R wt mice were not statistically significant, indicating that alleles present in Car-S and Car-R lines did not modify survival. Accordingly, analysis of tumorigenesis in unirradiated F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice showed a marked predisposition to tumor development that was independent from genetic background, with a cumulative tumor incidence of 51% and 56% in F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice, respectively. The tumor spectrum observed in these two groups was also similar (Table 2).

After local irradiation of adult mice with 4 Gy of X-rays, F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice showed significantly decreased survival compared with their unirradiated counterparts, with median survival times of 60 and 47 weeks, respectively (P = 0.0038; P = 0.0005). In contrast, no significant decrease in survival was observed in irradiated F1S and F1R wt mice compared with controls. Differences in survival between irradiated heterozygotes and their wt counterparts were highly significant both in F1S (P < 0.0001) and F1R mice (P < 0.0001). Thus, in addition to life span, Ptch1 haploinsufficiency also affects the response to ionizing radiation. However, the decrease in survival after irradiation was not associated with a significant increase in tumor incidence in F1S and F1R heterozygotes, despite a trend toward an increased tumorigenesis (Table 2). Additionally, although the survival of irradiated F1RPtch1^neo67/+ was significantly shortened compared with irradiated F1SPtch1^neo67/+ (P = 0.0003), the difference in tumor incidence between the two groups (61% and 68%, respectively) was not statistically significant. The tumor spectrum observed was also similar. One notable and significant exception was represented by BCC (see below) that was induced in F1SPtch1^neo67/+ but not in F1RPtch1^neo67/+ mice (Fig. 2, A and B; Table 2). This underlines the importance of genetic background interactions in the development of Ptch1-associated tumors.

Consistent with our previous results with CD1Ptch1^neo67/+ mice (12), examination of the unirradiated dorsal skin of F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice showed the presence of microscopic hyperproliferations of basaloid cells at interfollicular locations or in association with hair follicles (Fig. 2,C). However, the number of these microscopic lesions varied greatly in the two mouse strains (Table 3). Only 9% of F1RPtch1^neo67/+ presented these lesions, compared with 23% of F1SPtch1^neo67/+ mice. Moreover, 3 of 31 (10%) unirradiated F1SPtch1^neo67/+ mice had microscopically detectable nodular BCCs, consisting of nests of basaloid cells with peripheral palizading organized in entirely intradermic nodules (Fig. 2,D), and 1 of 39 (3%) had a macroscopic BCC-like tumor occurring at old age (2 years), showing histological features of infiltrating human BCC (Fig. 2,E). In contrast, neither nodular nor macroscopic BCCs were found in F1RPtch1^neo67/+ mice (Table 3). These observations suggest that F1SPtch1^neo67/+ mice have both a higher frequency of initiating lesions and a higher propensity to neoplastic progression of BCC lesions compared with F1RPtch1^neo67/+ mice. Neither proliferation areas, nor nodular or macroscopic BCCs were observed in wt mice of either background.

After localized radiation exposure of F1SPtch1^neo67/+ mice, we observed an increase in the percentage of mice with basaloid hyperproliferation areas, nodular and infiltrative BCCs compared with unirradiated mice (Table 3). In this strain, the incidence of basaloid hyperproliferations increased from 23% to 36%, whereas the incidence of nodular BCC increased from 10% to 29%. Infiltrative BCCs were also induced by radiation in 17% F1SPtch1^neo67/+ compared with a spontaneous incidence of 3% (P = 0.0287). By contrast, F1RPtch1^neo67/+ were completely resistant to BCC induction. In fact, in this strain ionizing radiation exposure did not enhance formation of hyperproliferation areas nor induce development of nodular or infiltrative BCCs. Differences in nodular BCC incidence between F1RPtch1^neo67/+ and F1SPtch1^neo67/+ mice were statistically significant (0% versus 29%; P = 0.0018). Macroscopic BCC incidences were also statistically significant (0% versus 17%; P = 0.03). Overall, these data show a significant modulation of BCC induction due to genetic background.

Irradiated F1S and F1R wt mice never developed hyperproliferation areas, nodular or infiltrative BCCs.

To examine additionally the effects of susceptibility alleles carried by Car-S mice on survival and tumorigenesis in Ptch1-deficient mice, we studied N2Ptch1^neo67/+ mice generated by backcrossing F1SPtch1^neo67/+ heterozygous mice onto the Car-S strain. Due to the very rapid rate of spontaneous tumor development in Ptch1^neo67/+ mice, this experimental series was performed by irradiating mice as newborns.

Fig. 3 shows that the exposure of newborn F1SPtch1^neo67/+ mice to 3 Gy of X-rays significantly reduced the survival compared with control mice. The median survival time decreased from 76 to 40 weeks (P = 0.0001). A similar median survival time (37 weeks) was observed in N2Ptch1^neo67/+ mice after neonatal irradiation.

No significant difference in tumor incidence and tumor spectrum was observed between neonatally irradiated F1SPtch1^neo67/+ and N2Ptch1^neo67/+. Neonatal irradiation induced a high incidence of thymic lymphoma in both strains (Table 2), but no correlation with Ptch1 genotype was observed (data not shown).

Of particular interest were the results of MB induction in Car-S-derived lines. Compared with our previous report of neonatally irradiated CD1Ptch1^neo67/+ mice, in which neonatal irradiation dramatically increased the incidence of MB (51%) over the spontaneous rate (8%; P = 0.0001; Ref. 14), neonatal irradiation of F1SPtch1^neo67/+ mice did not induce a significant increase in MB incidence (13%) compared with unirradiated mice of the same background (3%; Table 2). The difference in MB incidence between irradiated F1SPtch1^neo67/+ or N2Ptch1^neo67/+ mice and CD1Ptch1^neo67/+ mice was highly significant (13% versus 51%, P = 0.0004; 0% versus 51%, P < 0.0001), suggesting that irradiation of Ptch1 heterozygotes elicits a different tumorigenic response in the cerebellum depending on genetic background.

Table 2 also shows that the BCC phenotype was enhanced in N2 backcross mice, as shown by a 2-fold incidence of BCC as compared with F1SPtch1^neo67/+ mice (see below).

The comparison of data of BCC tumorigenesis in F1S and F1R mice provides only limited information on the influence of Car-S-derived alleles on susceptibility to BCC induction, because of the complete genetic suppression of the BCC phenotype in F1RPtch1^neo67/+ mice. Instead, comparison of Car-S-derived lines (F1SPtch1^neo67/+ and N2Ptch1^neo67/+) with CD1Ptch1^neo67/+ mice, carrying a random association of susceptibility and resistance alleles, can better clarify the contribution of low-penetrance Car-S alleles to BCC risk. In Table 4, data of BCC tumorigenesis in F1SPtch1^neo67/+ and N2Ptch1^neo67/+ mice irradiated at 4–8 days of age are compared with recent data from our laboratory on similarly treated CD1Ptch1^neo67/+ mice (12). Comparison of F1SPtch1^neo67/+ with CD1Ptch1^neo67/+ mice shows increased incidence of both nodular (1.8-fold) and infiltrative (4-fold) BCCs in F1SPtch1^neo67/+. These differences, however, were not statistically significant. N2Ptch1^neo67/+ mice, in contrast, were significantly more susceptible to BCC induction than CD1Ptch1^neo67/+ mice, with a 3-fold incidence of nodular BCCs (67% versus 23%; P = 0.0044) and an 8-fold incidence of infiltrative BCCs (31% versus 4%; P = 0.0064). These results suggest that BCC development was enhanced by Car-S-derived alleles and indicate the existence of both dominant and recessive Car-S alleles contributing to BCC susceptibility. Interestingly, the incidence of BCC precursor lesions (i.e., basaloid cell hyperproliferations) was lower in F1SPtch1^neo67/+ and N2Ptch1^neo67/+ compared with CD1Ptch1^neo67/+ mice (33–37% versus 63%; Table 4). This observation suggests that a faster rate of progression of BCC precursor lesions can be determined by a susceptible genetic background.

The kinetics of BCC induction in irradiated F1SPtch1^neo67/+ and N2Ptch1^neo67/+ mice is shown in Fig. 4. For comparison the BCC response induced by neonatal irradiation in CD1Ptch1^neo67/+ is also reported (12). Although N2Ptch1^neo67/+ mice showed a steeper BCC induction curve compared with F1SPtch1^neo67/+ and CD1 Ptch1^neo67/+ mice, differences in tumor latency were not statistically significant because of the few N2 mice (n = 16). Nevertheless, these results suggest that alleles present in the Car-S genetic background decrease the latency to BCC formation after exposure to ionizing radiation.

BCCs from F1SPtch1^neo67/+ and N2Ptch1^neo67/+ mice were grossly and histologically indistinguishable from BCCs of CD1Ptch1^neo67/+ (Fig. 2, D and E; Ref. 12). Similar to previous results with CD1Ptch1^neo67/+ mice, the majority of BCCs from F1SPtch1^neo67/+ (15 of 17) and N2Ptch1^neo67/+ (4 of 5) developed in male mice. As described for locally irradiated mice, wt mice irradiated as newborns never developed hyperproliferation areas, nodular or macroscopic BCCs.

To determine whether the development of macroscopic BCCs in the F1SPtch1^neo67/+ line was dependent on the complete loss of Ptch1 function, allelic imbalance at the Ptch1 locus was examined in infiltrative BCCs from irradiated F1SPtch1^neo67/+ mice. Overall, altered Ptch1 allelic patterns were observed in 14 of 15 (93%) BCCs from F1SPtch1^neo67/+ mice irradiated with 3 or 4 Gy of X-rays, as well as in a single spontaneous BCC (data not shown). To additionally elucidate the molecular events affecting the Ptch1 gene in BCCs, microsatellites spanning the length of chromosome 13 were used to examine 6 BCCs from irradiated F1SPtch1^neo67/+ mice. We never observed loss of the whole chromosome 13, as indicated by retention of the most proximal and distal microsatellite markers. All of the BCCs, instead, had interstitial losses of chromosome 13. Deletions had variable extensions, ranging from 6 to 16 cM. Although the position of breakpoints tended to be heterogeneous, there was not a continuum of breaks along the chromosome (Fig. 5). The absence of highly stringent breakpoint clustering is concordant with a genetic mechanism based on loss of a critical gene within the chromosome 13 minimal deleted region where we localized Ptch1 according to the Mouse Chromosome 13 Linkage Map.⁶

To test whether Ptch1 haploinsufficiency influenced chemical skin carcinogenesis on either background, we treated F1SPtch1^neo67/+ and F1RPtch1^neo67/+ mice and littermate controls with a single initiating dose of 9,10-dimethyl-1,2-benzanthracene, followed by twice-weekly promotion with 12-O-tetradecanoylphorbol-13-acetate. Mice of all groups developed papillomas or SCCs but not BCCs. F1S mice were very susceptible to papilloma induction by low 9,10-dimethyl-1,2-benzanthracene and 12-O-tetradecanoylphorbol-13-acetate doses (2.5 μg and 1 μg, respectively), whereas F1R mice were resistant to high doses (25 μg and 5 μg; Fig. 6). In fact, after a promotion period of 52 days 96% of F1S mice developed tumors, compared with only 24% of F1R mice. An average of 10.2 ± 0.4 and 12.2 ± 0.1 tumors developed in wt and heterozygous F1S mice at the end of promotion compared with 3.2 ± 1.0 and 2.8 ± 1.0 tumors in wt and heterozygous F1R mice, respectively. However, in the two genetic backgrounds tumor incidence and multiplicity were not substantially different in Ptch1 heterozygous versus wt mice.

Conversion frequency of papillomas to SCC, evaluated 4 months after the end of promotion, was also not influenced by the Ptch1 allelic status. In F1S mice we observed a carcinoma incidence and multiplicity of 8% and 0.11 in wt mice compared with 12% and 0.2 in Ptch1 heterozygous mice. In F1R mice the carcinoma incidence and multiplicity were 19% and 0.22 in wt mice compared with 7% and 0.1 in Ptch1 heterozygotes.

Overall, these data indicate that, although there was a strong effect of genetic background on susceptibility to squamous cell papilloma and carcinoma development, this was not affected by Ptch1 genotype.

It is well known that genetic background on which a highly penetrant cancer gene is studied can dramatically alter the individual risk for developing a specific tumor (23, 24). The genetic component may also limit or enhance the effect of exposure to environmental cancer-causing agents (25, 26). To examine the effect of genetic background together with exposure to ionizing radiation on the development of tumors, particularly BCCs, in Ptch1^neo67/+ mice we took advantage of the outbred Car-S and Car-R mouse lines that have been enriched for the specific combinations of alleles that control the skin tumor-susceptible or -resistant phenotypes (17). We report here that modifier alleles present in the Car-S and Car-R lines strongly influenced skin tumor development in Ptch1 heterozygous mice.

Consistent with published work from our laboratory (14), Ptch1 deficiency significantly shortened the life span of unirradiated mice compared with wt mice (Fig. 1,A). This decrease was independent from genetic background, because it was observed both in F1SPtch1^neo67/+ and F1RPtch1^neo67/+ populations. In addition, coherently with the role suggested for Ptch1 in the response to ionizing radiation (9), Ptch1 deficiency increased radiation susceptibility: after local irradiation with 4 Gy X-rays a significant life shortening was observed in F1SPtch1^neo67/+ and F1RPtch1^neo67/+ heterozygous but not in wt mice (Fig. 1 B).

In this study we show that although the genetic background did not have a strong impact on survival and cumulative tumor incidence of Ptch1^neo67/+ mice, significant effects were seen in the tumor spectrum arising in these animals. Alleles present in the Car-S genetic background increased the susceptibility to BCC and decreased the incidence of MB in Ptch1^neo67/+ mice (12, 14), particularly after ionizing radiation exposure. Moreover, the interaction of Car-R alleles with Ptch1 haploinsufficiency resulted in a nearly complete suppression of the BCC phenotype in F1RPtch1^neo67/+ mice. Overall, a modulation of 0–67% of BCC development was observed by introducing Car-R and Car-S skin cancer modifier loci into the CD1 background of Ptch1^neo67/+ mice

In previous work, we have proposed a multistep model of BCC tumorigenesis in CD1Ptch1^neo67/+ mice, in which accumulation of sequential genetic alterations leads to step-wise progression of BCC lesions (12). In this mouse model, Ptch1 deficiency is sufficient to initiate BCC precursor lesions, i.e., areas of hyperproliferation of basaloid cells that progress to nodular and then infiltrative BCCs (Fig. 2) after additional genetic damage due to radiation. Here, we show that genetic background strongly influenced all stages of this process. In addition to a higher frequency of preneoplastic BCC lesions (areas of basal cell hyperproliferation) in F1SPtch1^neo67/+ (23%) compared with F1RPtch1^neo67/+ mice (9%), the susceptible genetic background of Car-S mice led to spontaneous occurrence of later stages of BCC development, i.e., nodular (10%) and infiltrative BCCs (3%) in the skin of unirradiated F1SPtch1^neo67/+ mice (Table 3). These BCC subtypes were not detected in unirradiated F1RPtch1^neo67/+ (this study), nor were they observed in unirradiated CD1Ptch1^neo67/+ mice in our previous report (12). In addition, radiation exposure increased both the incidence of basal cell hyperproliferations (36%) and of nodular (29%) and infiltrative (17%) BCC only in F1SPtch1^neo67/+ mice. In contrast, neither enhancement of basaloid hyperproliferations, nor development of nodular or infiltrative BCCs was detected in the skin of F1RPtch1^neo67/+ mice after irradiation. This suggests that skin cancer modifier loci carried by the Car-S and Car-R mice act on the entire BCC tumorigenic process rather than on specific stages. Such modifying alleles may act directly on the Sonic hedgehog (Shh)-Ptch1 signaling pathway or on other aspects of the transformation of basal cells and progression to BCC.

Susceptibility to BCC development segregated as a dominant phenotype in neonatally irradiated F1SPtch1^neo67/+ mice, as shown by the 4-fold incidence of infiltrative BCC compared with CD1Ptch1^neo67/+ mice (Table 4). A similar difference in infiltrative BCC incidence (3.5-fold) is shown by the comparison of locally irradiated F1SPtch1^neo67/+ mice (this study) with similarly treated CD1Ptch1^neo67/+ mice (P = 0.0457; Ref. 12). However, an additional gain in susceptibility was observed in neonatally irradiated N2 mice, with a 3-fold incidence of nodular BCC (67% versus 24%; P = 0.004) and an 8-fold incidence of infiltrative BCC (31% versus 4%; P = 0.0064) compared with CD1Ptch1^neo67/+. These observations suggest that both dominant and recessive alleles contribute to susceptibility of Car-S-derived lines. ANOVA suggested the interaction of 8–10 independent quantitative trait loci in the regulation of susceptibility to papilloma development in the Car-S/Car-R mouse model (17), and the background effects observed suggest a role for some of them also in BCC tumorigenesis. In this respect, we have shown previously the existence of a Thr/Pro polymorphism in the coding region of the parathyroid hormone-related protein (PTHrP) gene (Pthlh) that is linked with susceptibility (Pro) or resistance (Thr) to papilloma and SCC induction by chemical agents in Car lines (27). Interestingly, a role for PTHrP as a mediator of the Hedgehog signaling pathway effects in the regulation of chondrocyte differentiation has been demonstrated (28, 29). Furthermore, overexpression of PTHrP in the skin of transgenic mice interferes with hair follicle development, causing ventral hairlessness and hyperkeratosis (30). Given the follicular origin of BCC (31), this suggests that Pro/Thr allelic variants of PTHrP from Car-S and Car-R mice may be among the host factors enhancing or inhibiting BCC tumorigenesis in F1SPtch1^neo67/+ or F1RPtch1^neo67/+ heterozygous mice. Determination of whether Pthlh influences BCC tumorigenesis in addition to papilloma development awaits analysis of allele-specific changes in BCCs.

The occurrence of spontaneous BCC development only in unirradiated F1SPtch1^neo67/+ mice and its lack in CD1Ptch1^neo67/+ mice suggest an influence of Car-S-derived alleles on tumor progression. Cancer modifier loci, in fact, may influence specific stages of carcinogenesis. For example, in lung tumorigenesis Par1 affects both tumor multiplicity and size, whereas Par2 acts specifically on tumor multiplicity and Papg1 and Sluc only on tumor size (19, 32). In the liver, Hcs loci affect the clonal expansion of hepatocellular tumors rather than susceptibility to tumor initiation (33). In this study, we show that Car-S-derived lines (F1SPtch1^neo67/+ and N2Ptch1^neo67/+), despite increased susceptibility to spontaneous and radiation-induced BCC tumorigenesis, show lower frequencies of hyperproliferation areas compared with CD1Ptch1^neo67/+ mice (33–36% versus 63%; Table 4). This suggests that Car-S-derived cancer modifiers may facilitate BCC progression rather than act on the frequency of initiating basal cell lesions.

Compared with BCC development, susceptibility to radiation-induced MB was reversed on the Car-S and CD1 genetic backgrounds, and F1SPtch1^neo67/+ and N2Ptch1^neo67/+ mice had a lower MB incidence than CD1Ptch1^neo67/+ mice (12.5% versus 51%, P = 0.0004; 0% versus 51%, P < 0.0001; Ref. 14). This result indicates that there are modifier loci that act on specific pathways to tumor development in different tissues, in agreement with the lack of apparent correlation between susceptibility to carcinogenesis in different organs of mouse strains (1). Consistent with this, tumor susceptibility and resistance in the Car lines were initially considered tissue-specific, as shown by the lack of interline difference in sarcoma induction after s.c. injection of chemical carcinogens (34). However, subsequent experiments have shown that Car-R mice carry cancer-resistance loci that inhibit susceptibility to both skin and lung tumorigenesis (35). The resistance of Ptch1 heterozygous Car-S-derived lines to MB development might reflect the presence of unlinked cancer modifier alleles in Car-S mice, with a subset providing susceptibility to skin cancer and an independent subset providing resistance to MB. Alternatively, among the skin cancer modifiers segregated in the Car-S line, one or a few might influence both types of tumorigenesis. Additional investigation is required to discriminate these possibilities. These results underline the importance of the genetic background when studying the effect of a specific germ-line mutation in the mouse. To study the phenotype associated with a defect in a specific gene, the effect of any germ-line mutation should therefore be examined on several different genetic backgrounds.

Determination of the Ptch1 allelic status in ionizing radiation-induced BCCs showed that 93% of BCCs from F1SPtch1^neo67/+ mice had loss of heterozygosity at the Ptch1 locus. This result is in close agreement with the 100% loss of heterozygosity observed previously in grossly visible BCCs from CD1Ptch1^neo67/+ mice (12). Although genetic background has been reported to affect the pattern of genetic changes in mouse tumors (36), our results suggest that the genetic changes at the Ptch1 locus required for BCC tumorigenesis are not influenced by genetic background and that Ptch1 must be inactivated to give rise to BCC. Consistent with this notion is the finding of wt Ptch1 loss also in the spontaneous, late-developing BCC (Table 2), confirming that complete loss of Ptch1 function is the rate-limiting event in BCC tumorigenesis. This event may only rarely occur without radiation insult. These observations strongly support a role for Ptch1 as BCC gatekeeper (37).

Data on microsatellite-based loss of heterozygosity analysis of chromosome 13 provide information on radiation-induced loss of the wt Ptch1 allele. The mechanism involved is not whole chromosome loss, but induction of interstitial losses, as shown by retention of the proximal and distal regions of chromosome 13 in all of the BCCs examined.

The involvement of deregulation of the Shh pathway in spinocellular tumors has been debated. The observation that tumors such as SCCs do not express high levels of Shh target genes suggests that the Shh-Ptch pathway may not have a major role in the growth of these tumors (11). On the other hand, SCCs were found at slightly higher frequency in irradiated Ptch1 heterozygous mice, and mutations in the human Ptch1 gene were detected at low frequency in sporadic SCCs (38, 39). Our results of chemical skin carcinogenesis on F1S and F1R Ptch1 heterozygous and wt mice indicate that the development of papillomas and SCCs was not affected by the Ptch1 genotype. In fact, the interaction of Ptch1 heterozygosity with Car-S- or Car-R-derived skin cancer modifiers did not influence the quality (induction of papillomas and SCCs but not of BCCs) or quantity (similar tumor burden in heterozygous and wt mice) of tumors in response to chemical skin carcinogenesis. This shows that deregulation of the Shh pathway is not involved in development of papillomas and SCCs.

The results of this study provide evidence for a significant effect of genetic background on Ptch1-associated tumors in a mouse model of NBCCS. A genetic predisposition to sporadic BCC development has also been shown in the general population, and efforts have been made to identify the loci associated with susceptible phenotypes (7, 40). However, detection of interacting effects of genes on complex phenotypes, such as cancer susceptibility in humans, has proven difficult. An alternative strategy to identify these genes involves mouse models susceptible to development of BCC, but inherent resistance of the mouse has hampered this approach. Here, we describe an original model, in which a combination of Ptch1 deficiency with a skin tumor-susceptible genetic background (Car-S) facilitates formation of nodular and infiltrative BCC. Because these BCC lesions can be easily classified and quantified, this model may allow for the identification of genes that modify BCC tumorigenesis and, thus, could contribute significantly to our understanding of factors that put individuals at risk of BCC development. Finally, the ability to induce BCC at high rate after a single radiation exposure provides a useful model to study its pathogenesis and will facilitate development of therapies against the most common human tumor.