MECHANISMS OF INDUCTION OF SKIN CANCER BY UV RADIATION

Holly Soehnge, Allal Ouhtit and Honnavara N. Ananthaswamy

Department of Immunology, The University of Texas M D Anderson Cancer Center, 1515 Holcombe Blvd., Box 178, Houston, TX 77030

Received 10/20/97 Accepted 10/24/97

4. TUMOR SUPPRESSOR GENES AND ONCOGENES IN SKIN CANCER

Carcinogenesis by UV radiation often involves the inactivation of one or more tumor suppressor genes or the overactivation of growth-stimulatory proto-oncogenes. Tumor suppressor genes are negative growth regulators and usually are recessive in that they require both copies of the gene to be inactivated before loss of control of cell growth occurs. Accumulation of proteins that bind to and sequester tumor suppressor proteins can also make the cell more susceptible to further mutations (31). Activation of oncogenes is dominant in that a change in only one copy of the gene is required to have an effect. Protooncogenes, the normal versions of oncogenes, act to control cell proliferation and differentiation, and are divided into three groups: growth factors and growth factor receptors, signal transduction proteins, and nuclear factors (3). Carcinogenesis can result either from overexpression of the normal gene product or from expression of a mutant or altered gene product. Several genes have been extensively studied that have important roles in skin carcinogenesis. These include the tumor suppressor gene p53 and the ras oncogenes. Other candidates are currently being investigated; one such gene is patched, which may act as a tumor suppressor. Information about these genes is summarized in table 1 (see also figure 5).

4.1. The p53 tumor suppressor gene

The p53 tumor suppressor gene codes for a DNA-binding protein and mutations or loss of p53 plays a key role in the process of carcinogenesis. It is the most frequently altered gene in human cancers (>50%). The human p53 gene is localized on the chromosome 17 (17p13) and contains 11 exons spanning 20 kilobases. The mouse p53 gene is localized on chromosome 11 and also contains 11 exons (32). Recent work indicates that the p53 protein is a central element in fundamental cellular processes, including gene transcription, repair of DNA damage, control of the cell cycle, genomic stability, chromosomal segregation, senescence and apoptosis (33).

4.1.1. p53 and the cell cycle

The cell cycle is under a positive and direct control of the cyclin-dependent kinase family (Cdk) and their regulatory subunits. Cdks control the cell cycle in part by hyperphosphorylation and inactivation of negative regulators of the cell cycle, such as the Rb protein responsible for susceptibility to retinoblastoma, and associated proteins p107 and p130. It has been proposed that after genotoxic stress, the accumulation of p53 protein induces a cell cycle arrest at the G1 phase; this arrest allows the repair of DNA damage before its replication in the S phase (34, 35). However, it has been found that DNA damage might induce an irreversible arrest of normal human fibroblast mitosis (36). The function of p53 in control of the cell cycle is shown diagrammed in figure 2. Accumulation of active p53 induces the expression of different proteins that regulate the cell cycle. p21 (encoded by theWaf1 gene, called also Cip1, Sdi1 or Pic1) inactivates the Cdk-Cyclin complex by forming a Cdk2/A or E Cyclin/Proliferating Cell Nuclear Antigen/Waf1 complex.

Figure 2. The role of p53 in control of the cell cycle

Formation of this complex leads to the accumulation of hypophosphorylated pRb, causing the release of E2F, which is necessary for the induction of DNA synthesis (37, 38). p21 can also be induced by a p53-independent pathway (39). Although it is known that Gadd45, IGF-BP3 and G or D1 Cyclin G, and likely other proteins, can induce cell cycle arrest, the molecular mechanisms are not yet fully understood.

4.1.2. p53 and apoptosis

p53 can induce apoptosis (programmed cell death) by two independent mechanisms, as shown in figure 3. One pathway depends on the function of p53 as a transactivator of transcription by upregulating the expression of Bax, IGF-BP3 and Fas proteins, and by downregulating the expression of Bcl2, IGF-1R and IGFII. The up- and down-regulation of these proteins, respectively, has been correlated with the induction of programmed cell death processes (reviewed in 33). The second pathway is independent of the p53 transcriptional function but is dependent on p53 protein-protein interactions: p53 protein can bind to cellular proteins involved in DNA synthesis such as replicating protein antigen (RPA) (40), and in DNA repair such as TFIIH, including xeroderma pigmentosum group B (XPB) and D (XPD) DNA helicases, p62 and topoisomerase I (41-42).

Figure 3. The role of p53 in apoptosis

4. 1. 3. Inactivation of the p53 gene

Some mutations in the p53 gene frequently lead to the production of a protein that cannot bind specifically to DNA, and therefore loses its transactivation activity (43). However, it has been found that some mutations such as 175Pro, resulted in the loss of the ability of the p53 protein to induce apoptosis in certain cells, but these cells are still able to induce growth arrest (44). Three types of genetic alterations may affect the function of the p53 protein: (i) Partial or total deletion of the gene, which is sufficient to abrogate the tumor suppressor function, contributing to the development of tumors, as has been shown in the p53 knockout mice (45); (ii) A dominant negative effect of some critical mutations which can suppress the function of the wild type p53 protein through inhibition of its DNA binding activity and thus its transactivation function (46). It has been reported that a transgene carrying one mutated allele, which codes for mutant p53 protein 135Val, results in loss of tumor suppressor function and contributes to increased tumorigenicity by a dominant negative effect, probably by forming inactive hetero-oligomers with wild type protein, generated from the normal allele (47). It has been shown recently that all hotspot mutations in the p53 gene generate mutant proteins capable of inhibiting the transcriptional activity of the wild type protein (48). (iii) Certain "gain of function" mutants of p53 acquire an oncogenic potential. For example, the 175His p53 protein was characterized by the loss of tumor suppressor function, a dominant negative effect gain of function (47) and the ability to increase the tumorigenic and metastatic potential of cells missing the wild type p53 protein (49).

Mutational analysis of the p53 gene provides a unique opportunity to investigate the etiology, epidemiology, and pathogenesis of human cancer (50-52). It is the most commonly mutated gene in human cancers (50). To date, more than 5000 mutations have been identified and classified that are available electronically through the network databases (50). Interestingly, the sites of the point mutations are nonrandom, with more than 90% occurring in highly conserved regions of the middle third of the gene (52).

4.1.4. p53 mutations in UV-induced skin cancers

Analysis of human skin cancers and UV-induced mouse skin cancers for p53 mutation have provided new insights into the molecular mechanisms by which UV radiation induces skin cancer. The p53 gene has been found to be mutated at a high frequency in human (53-55) and mouse UV-induced skin cancers (56-58). Most hotspot mutations detected in human and mouse UV-induced skin cancers inactivate critical p53 functions. Over 90% of human cutaneous SCCs and about 56% of human BCCs contain unique mutations at dipyrimidine sites. A number of studies have found p53 mutations in precancerous actinic keratosis and sun-damaged skin (59, 61). An interesting aspect of UV-induced mutations in the p53 gene in human skin cancers is that they are frequently CT or CCTT base substitutions at dipyrimidines sites. These types of p53 mutations are not commonly present in human internal cancers or in mouse skin cancers induced by chemical carcinogens. In XP tumors, including BCC, SCC and sarcoma, 60% of the p53 mutations detected were tandem CCTT transitions and occurred at sites previously identified as hotspots (62, 63). Because of the unique nature of CCTT tandem mutations, they are termed UV "signature" mutations (64). Most of the mutations detected in XP skin tumors occurred at the non-transcribed strand implying a preferential repair of UV-lesions on the transcribed strand in human tissues.

UV-induced mouse skin cancer provides an ideal model to investigate the molecular mechanisms involved in the multistep process of carcinogenesis. Analogous to human skin cancers, UV-induced mouse skin cancers also display p53 mutations (56-58), although the frequency of mutations and the exons in which they occur differ among mouse strains, for reasons that are not yet clear. For example, in our study, p53 mutations were detected at 70-100% frequency in UV-induced SKH-hr1 and C3H mouse skin tumors, respectively (56, 57). In contrast, 20% of SCC from SKH-1/hr hairless mice and 50% of SCC from BALB/c mice exhibited p53 mutations in another study (58). Nonetheless, most of the mutations detected in UV-induced mouse skin tumors were CT and CCTT transitions at dipyrimidine sites, like those found in human skin cancers. Another important finding from these two studies is the fact that most of the p53 mutations were located on the non-transcribed strand, which is in accordance with the principles of preferential DNA repair. In other words, transcribed strands are repaired preferentially over non-transcribed strands (65, 66). Several of the UV-induced C3H mouse skin cancers contained multiple mutant p53 alleles (56) suggesting that mutant p53 alleles with single base changes were targets of secondary mutation events, perhaps due to the continued exposure of UV radiation during tumor progression.

4.1.5. p53 mutations arise early during UV skin carcinogenesis

The progression of cancer from its initial benign stages to malignancy is generally through stages exemplified by the adenoma-carcinoma sequence in colon cancer (67). These morphologically defined stages provide milestones for recording the timing of p53 mutations. In colon cancer, for example, p53 mutations are generally a late event marking the progression from the late adenoma to carcinoma stages (67). However, in the case of skin cancers where UV is the etiological agent, mutations in the p53 gene appear to have an earlier onset (57, 59, 60, 68). Using sensitive PCR- and ligase -chain reaction based methods, hotspot tandem mutations at codons 245 and 247/248 were detected in UV-exposed normal human skin cell cultures and normal human skin biopsies (69, 70). In addition, in a case-control study conducted recently in Australia, Ouhtit et al. (71) found a correlation between mutation frequency in normal skin biopsies (tandem CCTT transitions) at codons 247/248 of the p53 gene and the risk of BCC. A study by Ziegler et. al. (60) revealed p53 mutations in 60% of all AK samples examined, and a high proportion (89%) of them were of the UV "signature" type. More importantly, when normal skin flanking the AKs was examined, the frequency of p53 mutations was exceedingly small (less than 10-3). Whole mount preparations of sun-exposed and sun-shielded normal human skin revealed clonal population of p53-mutated cells in the sun-exposed skin, arising from the dermal-epidermal junction and from hair follicles (59). Recently, a combination of immunohistochemistry and sequencing analyses has revealed the presence of p53 immunopositive clonal patches that contain predominantly "UV signature" mutations in normal human sun-exposed skin (59, 75). Clones were both more frequent and larger than in sun-shielded skin, indicating that UV radiation acts as an initiator and a promoter by favoring the clonal expansion of p53-mutated keratinocytes (76).

Berg et al. (72) analyzed UVB-irradiated mouse skin for the presence of cells expressing mutant p53 protein and found several clusters of cells in the epidermis that reacted with an antibody specific for mutant p53 protein after 17 or 30 daily UVB exposures, which would cause skin tumors around 80 or 30 weeks, respectively. Such clusters expressing the mutant p53 protein persisted in the skin for at least 56 days after UVB irradiation. p53 mutations have also been detected in UV-irradiated mouse skin months before the gross appearance of skin tumors. The p53 mutations in mouse skin arose as early as the 4th wk of UV-irradiation, and the frequency of p53 mutations increased progressively and reached 50% at 12 wk of chronic UV exposure (57). In addition, around week 16, when the p53 mutation frequency reached maximum, the first mouse developed a skin tumor, and 50% of the mice developed skin cancer by wk 25. These results suggest that p53 mutations arise well before skin cancer development and that they can serve as a surrogate early biological endpoint in skin cancer prevention studies.

4.2. The patched (ptc) gene is a tumor suppressor in humans

4.2.1. ptc gene activity is conserved in Drosophila and vertebrates

The ptc gene was recently cloned in Drosophila by two different groups (77, 78). The predicted novel protein has 1286 amino acids and is 143 kilodaltons (kDa) in molecular weight; it is also predicted to be an integral membrane protein. In Drosophila, ptc is a segment polarity gene; these genes act to control the number and identity of the body segments in a developing Drosophila embryo. Development of each segment is controlled by a system of carefully regulated intercellular signals for proper growth and differentiation; the expression patterns of ptc and other genes act to specify the cell identities that eventually determine the segmental pattern.

The ptc gene controls the activities of genes that drive cell growth and differentiation by repressing their activities in cells where ptc is expressed. This is accomplished by opposing the function of the hedgehog (hh) gene, which encodes a secreted signaling protein that induces cell growth and differentiation. Cells receiving the hh signal activate transcription of target genes that result in a growth response; the transcription of ptc is also induced in these cells. Ptc responds to hh signals from adjacent cells by blocking transcription of hh target genes; these include ptc itself, in a negative autoregulatory function. Transduction of the ptc signal pathway involves the smoothened (smo) gene, another membrane protein that has characteristics of G-protein receptors; it acts downstream of or parallel to ptc (79). An additional gene further downstream in the pathway is cubitus interruptis (ci), which encodes a zinc finger protein that activates expression of hh target genes and is part of a negative feedback loop with ptc (80). Normally, there exists a balance between expression of the opposing activities of hh and ptc that determines the level of target gene expression. The hh target genes are ectopically activated in ptc mutants (81), so that abnormal cell growth may be expected to occur when hh signaling goes unchecked. Current research is underway in order to elucidate the molecular details of this important developmental pathway.

Murine (81) and human (82, 83) homologs of the ptc gene have recently been cloned. The mouse and human ptc genes are very similar, with 96% amino acid sequence identity, compared to 40% similarity to Drosophila ptc. The gene structure spans over 32 kb of genomic DNA and contains at least 23 exons; the open reading frame has 1447 amino acids. The N- and C-termini of the mouse and human genes correspond. Murine ptc was found to be transcribed in many tissues near cells producing either Sonic or Indian hedgehog, signaling proteins analogous to hedgehog in Drosophila. Ectopic Sonic hedgehog (Shh) expression in the mouse central nervous system also induced ptc transcription (81). Homologs of hh have been found in other vertebrates, including humans (84). These results suggest that the ptc signaling pathway is conserved in Drosophila and vertebrates (81), which could greatly aid in understanding of the cellular pathways involved in the function of ptc as a tumor suppressor in humans. The Ptc protein was recently shown to act as a receptor for Shh, in the mouse (85). Hh was demonstrated to bind to Ptc with high affinity, but not to the recently cloned vertebrate homolog of Drosophila Smoothened (vSmo); this was contrary to previous research suggesting that Smo was the receptor for Hh in Drosophila (79). Also, Ptc was found to form a physical complex with vSmo, suggesting that vSmo could be linked to Ptc in the signaling pathway, and that Shh, Ptc, and vSmo form a physical complex (85). A diagram depicting these findings is shown in Figure 4.

Figure 4. Model for Ptc signaling and a possible role for ptc mutations in BCC

These results were important in the elucidation of the Ptc signaling pathway in the mouse; the same concepts may apply in humans as well. The possible application of information gained from the Drosophila and mouse systems could be especially important in light of the recent discovery that ptc acts as a tumor suppressor in basal cell carcinoma in humans.

4.2.2. ptc mutations and nevoid basal cell carcinoma syndrome

The human ptc gene was found to colocalize with the map location of nevoid basal cell carcinoma syndrome (NBCCS) on chromosome 9, at 9q22.3 (82). NBCCS, also called basal cell nevus syndrome or Gorlin's syndrome, is a rare autosomal dominant disorder characterized by multiple BCCs that appear at a young age. NBCCS patients are very susceptible to the development of these tumors; in the second decade of life, large numbers appear, mainly on sun-exposed areas of the skin. This disease also causes a number of developmental abnormalities, including rib, head and face alterations, and sometimes polydactyly, syndactyly, and spina bifida. They also develop a number of tumor types in addition to BCCs: fibromas of the ovaries and heart, cysts of the skin and jaws, and in the central nervous system, medulloblastomas and meningiomas. Studies of NBCCS patients show that they have both genomic and sporadic mutations in the ptc gene, suggesting that these mutations are the ultimate cause of this disease (86). The number, distribution, and early onset of the BCCs suggested that ptc may be a tumor suppressor gene (87).

The deregulation of the ptc signaling pathway may be a general feature of basal cell carcinomas caused by ptc mutations. Consistent overexpression of human ptc mRNA has been described in tumors of familial and sporadic BCCs, determined by in situ hybridization (88). Mutations that inactivate ptc may be expected to result in overexpression of mutant Ptc, because ptc displays negative autoregulation. It is also plausible that mutation of ptc might lead to tumorigenesis through overexpression of Hedgehog proteins, given the role of the hh gene described in Drosophila. That SHH has a role in tumorigenesis in the mouse has been suggested by recent research in which transgenic mice overexpressing SHH in the skin developed features of NBCCS, including multiple BCC-like epidermal proliferations over the entire skin surface, after only a few days of skin development (89, 90). A mutation in the Shh human gene from a BCC was also described (89); it was suggested that Shh or other Hh genes in humans could act as dominant oncogenes in humans.

4.2.3. UV-induced ptc mutations play a role in BCC development

NBCCS patients develop many BCCs early in life, mostly on sun-exposed sites; the degree of sun exposure and skin pigmentation also influence the age of onset and number of skin tumors (91). These observations suggest that UV exposure is involved in the development of the BCCs in these patients, who may be more susceptible due to the ptc mutations. Most genomic ptc mutations found in NBCCS patients lead to a truncated protein (92). A mutation in a sporadic BCC from an NBCCS patient, a GCAT transition characteristic for UV-induced mutations, caused a stop codon also leading to premature termination of the protein (86). Skin fibroblasts from NBCCS patients have been shown to have increased sensitivity to killing by UVB radiation, which was not due to defects in DNA repair (93). These results suggested a connection between the genetic susceptibility of NBCCS patients to skin cancer (due to ptc mutations) and sensitivity to UV radiation.

Sporadic ptc mutations have been described in BCCs from otherwise normal individuals, some of which are UV-signature mutations. In one recent study of sporadic BCCs, five UV-signature type mutations, either CT or CCTT changes, were found out of fifteen tumors determined to contain ptc mutations (94). Another recent analysis of sporadic ptc mutations in BCCs and neuroectodermal tumors revealed one CT change in one of three ptc mutations found in the BCCs (95).

4.3. The role of ras oncogenes in skin cancer

The protooncogenes H-ras, K-ras, and N-ras encode 21-kDa proteins that share about 70% sequence homology (96). Located at the inner cell surface, ras proteins participate in signal transduction by binding to GTP. Signals involved in growth control, such as binding of activators to cell surface receptors, are transferred from the cell surface to the nucleus. Recent study has added to the understanding of this signalling pathway (97, 98). Mutations in the ras genes that cause continuous activation have been found mainly in codons 12, 13 and 61 for all members of the ras family (99). The ras genes are active when bound to GTP; mutations may cause activation by reducing the rate of hydrolysis of GTP to GDP, as has been found for a mutation to valine in codon 12 (100).

Mutations that activate ras genes occur less frequently in human skin cancers than mutations in the p53 gene. ras mutations have been reported to occur at 10-40% frequency in human skin cancers (reviewed in 101). However, skin cancers from XP patients harbor mutation in ras genes at a high frequency (53%) as compared to skin cancers from the general population which occured at 22% frequency (102). In contrast to this report, Ishizaki et al., (103) and Sato et al. (104) found a relatively low frequency of ras mutation in skin cancers from XP patients. This discrepancy could be due to differences in XP complementation groups between the Japanese and European patients. Analogous to human skin skin cancers, mutations in N-ras codon 61 of the N-ras gene have been reported in UV-induced mouse skin cancers (105).