Abstract
Treatment with 8-methoxypsoralen (8-MOP) and ultraviolet radiation (primarily UVA), called PUVA therapy, has been used to treat different chronic skin diseases but led to a significant increased risk for skin cancer. The National Toxicology Program (NTP) performed a study in mice treated with PUVA that showed a significant increase in squamous cell carcinomas of the skin. In the present study, we evaluated the protein expression of p53 and PCNA and DNA mutations of p53 and H-ras genes in both hyperplastic and neoplastic squamous cell lesions from the NTP study. By immunohistochemical staining, protein expression of both p53 and PCNA was detected in 3/16 (19%) of hyperplastic lesions and 14/17 (82%) of SCCs in groups treated with both 8-MOP and UVA. The mutation frequency of p53 in SCCs from mice administered 8-MOP plus UVA was 15/17 (88%) with a predominant distribution of mutations in exon 6 (14/15 – 93%). No H-ras mutations were detected in the hyperplastic lesions/tumors. The mutagenic effect of PUVA on the p53 tumor suppressor gene may lead to a conformational modification and inactivation of the p53 protein, which are considered critical steps in PUVA-induced skin carcinogenesis. The p53 mutational frequency and patterns from our study were different from those reported in human PUVA-type tumors.
Introduction
Treatment of the skin diseases psoriasis and vitiligo with psoralen-containing compounds such as 8-methoxypsoralen (8-MOP), followed by exposure to UV light, has been commonly used for several decades. Application of 8-MOP and UV radiation (primary UVA, 320–400 nm), called PUVA therapy, is also used to treat cutaneous T-cell lymphomas (Stern et al., 1984; Swinyard and Pathak, 1985; Edelson et al., 1987; Lindelof and Sigurgeirsson, 1991; Horio, 2000; McGinnis et al., 2003). This therapy acts via a photosensitization reaction by psoralen, which, through its planar aromatic structure, first intercalates the DNA double helix and then covalently binds pyrimidine bases through a photocycloaddition reaction driven by the UVA exposure (Gasparro et al., 1997, 1998).
Epidemiological studies and case reports in patients treated with PUVA therapy have documented a significant increase in skin tumors, mainly squamous cell carcinomas, with a relative risk ranging from 10 to 20 (Stern et al., 1984; Lindelof et al., 1991; Lever and Farr, 1994; Maier et al., 1996; Stern et al., 1998; Stern and Lunder, 1998). Other tumors of lower incidences, including basal cell carcinomas and melanomas, have been reported (Stern et al., 1984; Bruynzeel et al., 1991; Stern et al., 1997, 1998; Stern and Lunder, 1998; McGinnis et al., 2003). The International Agency for Research on Cancer (IARC) has classified PUVA as a group I carcinogen (IARC, 1980, 1987).
Mutational analysis performed on squamous cell carcinomas from PUVA-treated patients showed a pattern of p53 mutations related to coupled exposure to 8-MOP and UVA, in which the base substitution event was attributed to the direct photo-damaging action of UV light on DNA, rather than the photo-activation of the psoraslen administered orally (Nataraj et al., 1997; Wang et al., 1997). In vitro molecular assays of mouse cell lines (Santamaria et al., 2002) and yeast cells transfected with the human p53 wild-type gene (Inga et al., 1998; Monti et al., 2000), and administered PUVA therapy showed a consistent mutational pattern as well. These data showed a direct involvement of the p53 gene product in skin tumorigenesis.
The p53 protein has been widely studied in the last decade for the pivotal role it plays in cell cycle regulation, DNA repair, and cell differentiation (Ko and Prives, 1996). That approximately 50% of human cancers contain mutations in the p53 gene has been estimated and investigators have predicted that, in the majority of the remaining tumors the p53 signaling pathway is inactivated by up-regulation of p53 inhibitors or down-regulation of p53 cooperators (Attardi and Jacks, 1999).
p53 is a highly conserved gene with a mutational pattern that allows for comparison between animal models and humans. The mouse p53 occurs as a tetramer, and, for each monomer, four main areas can be identified: (1) the N-terminal activation domain that represents the binding site for the Mdm2 protein, the principal p53 regulator; (2) the DNA-binding domain; (3) the tetramerization domain; and (4) the C-terminal regulatory domain, target of the ubiquitination process that leads to protein degradation (Klein et al., 2001; Zhao et al., 2001). The distribution of mutations in human p53 is mainly located in the DNA-binding domain of the protein across 4 of the 5 most conserved regions, while the one remaining, in the N-terminal domain, is involved in the Mdm2 binding-site arrangement (Pavletich et al., 1993).
The National Toxicology Program (NTP) performed a study in mice exposed to PUVA therapy (Dunnick et al., 1991) to test its carcinogenic potential. To evaluate better the effect of UV radiation, the hairless phenotype mouse strain HRA/Skh was chosen. This strain has a homozygous genotype for hairless (hr/hr) and albino (c/c). A dose-related increase in squamous cell hyperplasias was described in mice, and an increase in squamous cell carcinomas was detected in mice treated with high doses of 8-methoxypsoralen (8-MOP) (Dunnick et al., 1991). This study showed that the combined treatment of 8-MOP and UVA, at levels comparable to those used for PUVA therapy in humans, produced hyperplastic lesions and tumors of the skin in mice.
Our study was designed to examine the p53 mutational pattern in skin tumors of mice administerd PUVA therapy. Skin tumors were evaluated for mutations in p53 exons 5, 6, 7, and 8 that encompass almost the entire DNA-binding domain of the mouse p53 gene (Tam et al., 1999; Hong et al., 2000, 2003). These tumors as well as hyperplastic skin lesions were examined by immunohistochemical methods to determine the expression and distribution pattern of p53 protein and proliferating cell nuclear antigen (PCNA), which consists of 2 different protein complexes, the first involved in DNA replication and the second associated with DNA repair activity (Maga and Hubscher, 2003).
H-ras proto-oncogene was also examined for mutations in the skin tumors following the combined treatment with 8-MOP/UVA. H-ras is known to play a key role in chemically induced skin carcinogenesis in rodents (Mangues and Pellicer, 1992), and, moreover, it is a known mutational target in many human tumors, including squamous cell carcinomas from PUVA-treated psoriasis patients (Nishigori et al., 1994; Portella et al., 1994; Sills et. al., 1999; de Gruijl et al., 2001; Kreimer-Erlacher et al., 2001).
Materials and Methods
Experimental Design
The hairless phenotype mouse strain HRA/Skh was chosen, since skin tumors develop in a short time without any significant increase in tumors in other organs (Mann, 1971a, 1971b; Forbes, 1981). The experimental animals were randomized and treated for 52 weeks.
Mice were administered 8-MOP in the feed at a dose of 0 and 625 ppm (0–80 mg/kg of body weight) and then exposed to UV for 5 minutes with an intensity of 2 J/cm2 for a total of 150 pulse feedings and 144 UV radiations over the treatment period (Table 1). Mice were exposed to UV radiation either unfiltered or filtered to remove the short UV wavelength (280–320 nm) emitted from the UV lamp used. During the UV treatment, racks of animal cages were placed in front of the radiation system, and mice were allowed to move freely.
Groups of mice treated with 625 ppm 8-MOP and exposed to either filtered or unfiltered UV radiation exhibited significant increases in skin neoplastic lesions (Table 1).
Tissues Retrieval and Selection
Paraffin-embedded blocks containing proliferative skin lesions (Table 1) were obtained from the following 5 groups: (1) group C, untreated mice; (2) group F, mice treated with filtered UV; (3) group U, mice treated with unfiltered UV; (4) group PUVA-F, mice treated with 8-MOP and filtered UV; and (5) group PUVA-U, mice treated with 8-MOP and unfiltered UV.
Immunohistochemistry
The streptavidinbiotin-peroxidase system (Rao et al., 1996; Williams et al., 2001) was used to detect p53 protein and PCNA. The p53 oncoprotein polyclonal antibody (Signet Lab. Dedham, MA) and the PCNA monoclonal antibody (Coulter Immunology, Hialeah, FL) were used. Positive and negative controls were run with the two staining procedures.
Immunohistochemical analysis of p53 and PCNA was performed on 3 normal skin specimens from group C, 16 squamous cell hyperplasias (SCH) and 10 SCCs from group PUVA-F, and 7 squamous cell carcinomas (1 in situ) from group PUVA-U. Positive staining for p53 and PCNA proteins were determined when greater than 5 positive cells were identified in the skin samples evaluated.
DNA Isolation, Amplification, and Cycle Sequencing
Mutational analysis was performed on 1 SCC (in situ) and 1 squamous cell papilloma (SCP) from group C, 4 SCPs from group F, 5 SCPs from group U, and, for immunohisochemical analysis, on the same 10 SCCs from group PUVA-F, and 7 SCCs (1 in situ) from group PUVA-U, which were used for immunohistochemical analysis.
Neoplastic skin lesions from representative areas of 5–8 unstained slides containing 10 μm thick paraffin-embedded tissue slices were scraped into a microcentrifuge tube. After deparaffination, tissues were rehydrated and the DNA was isolated using the Pico Pure™ DNA extraction Kit (Arcturus Engineering Inc., Mountain View, CA), according to the manufacturer’s instructions.
The extracted DNA was amplified by the PCR method (Saiki et al., 1988) using the nested amplification approach. The primer sets for p53 exons 5, 6, 7, and 8 were specifically designed, and their characteristics are outlined in Table 2. The details of the primers used for the amplification of H-ras fragment, including codon 61, have been described (Devereux et al., 1991). Positive and no-DNA controls for both H-ras and p53 were run with all sets of reactions.
The PCR reaction mixtures were separated on a 2% agarose gel, and the amplification bands were cut from the gel and then purified using the QIAquick gel extraction kit (QIAGEN Inc, Valencia, CA).
Samples were sequenced utilizing a cycle sequencing kit (U.S. Biochemical, Cleveland, OH), which incorporates α-33P dideoxynucleotide (ddNTP) terminators (A, C, G, T) into the sequencing products.
The sequencing reaction products were analyzed by electrophoretic separation on an 8% acrylamide gel containing 8 M urea. Gels were dried and exposed to X-ray films overnight.
Results
We examined p53 protein and PCNA expression in 3 normal skin samples from group C, 16 SCHs from group PUVA-F, and 17 SCCs of which 10 were from group PUVA-F and 7 from group PUVA-U. Positive staining was found for both p53 and PCNA in 3/16 (19%) of hyperplasias and 10/10 (100%) of carcinomas from group PUVA-F and 4/7 (57%) from group PUVA-U. None of the 3 normal skin samples was positive for p53 and PCNA (Table 3). Nuclear p53 staining was prominent in anaplastic cells; p53 protein and PCNA displayed a positive association in the SCCs (Figure 1). The p53 protein expression in hyperplastic lesions was predominantly in the basal epithelial cells, while in the carcinomas it was widespread and consisted of intense nuclear localization. In the SCCs, p53 protein and PCNA were also closely associated in areas of anaplasia (Figure 1).
Exons 5, 6, 7, and 8 of the p53 gene were sequenced. There was no significant difference in mutational frequency and profiles between PUVA-F and PUVA-U groups. In the PUVA-F group 10/10 (100%) of the SCCs had p53 mutations, and in the PUVA-U group mutations were detected in 5/7 (71%) of SCCs. Overall 15/17 (88%) of SCCs displayed p53 mutations. All of the mutations found were base substitutions; no deletions or insertions were detected (Tables 4 and 5).
Comparing the immunohistochemical staining with the mutational analysis for 17 SCCs (10 from group PUVA-F and 7 from group PUVA-U), we found that positive immunohistochemical staining was strongly correlated with the missense mutations in the p53 gene, with few exceptions (Table 5).
The p53 mutation data are summarized in Tables 4 and 5. In SCCs from group PUVA-F and PUVA-U, the majority of p53 mutations were detected in exon 6. A total of 4 SCCs from the PUVA-F group had mutations in at least 2 different codons; 2 in exons 5 and 6, one in exons 6 and 7, and one with both mutations in exon 6. In the PUVA-U group, 1 SCC had 4 p53 mutations, 1 in exon 5 and 3 in exon 6. Of the 14 base substitutions identified in the PUVA-F group, 3 were silent mutations, while 11 were missense mutations leading to amino acid substitutions in the p53 protein. The 8 base substitutions identified in the PUVA-U group were missense mutations. Overall, 13 codons displayed transitions, while 9 showed transversions. The transition/transversion rate was 9/6 and 3/5 for group PUVA-F and group PUVA-U, respectively. In only 1 case, in a SCC of the PUVA-F group, a silent mutation was not accompanied by a missense mutation. Seven codons displayed only the mutated allele: 2/14 (14%) in the PUVA-F group and 5/8 (63%) in the PUVA-U group (Figure 2).
In group C no p53 mutations were detected in the SCC in situ or in the SCP group. Three of 9 (33%) SCPs in the F and U groups had mutations: 2/4 (50%) in the F group and 1/5 (20%) in the U group. Each mutated sample displayed 2 different mutations in 2 different codons of different exons. All of the mutations described in these groups were missense mutations. Overall, the mutations found in the groups other than PUVA-F and PUVA-U appeared randomly distributed without any specific pattern.
None of the samples tested exhibited p53 mutations which corresponded to the typical solar-like thymine dimer mutation (e.g., CC→TT).
The significant percentages of mutations showing only the mutated p53 allele, both in the group treated with UVA and in those treated with 8-MOP and UVA, are consistent with loss of the wild-type allele and the p53 mutations contributing to loss of function of the p53 tumor suppressor gene and to the development of skin tumors.
No mutations were detected in H-ras codon 61 in any of the SCCs of the PUVA-F and PUVA-U groups examined (data not shown), suggesting that the ras signal transduction pathway does not play a role in the development of skin tumors following the combined 8-MOP/UVA treatment.
Discussion
Our study showed that photoactivated 8-MOP induced a high frequency of p53 mutations in SCCs. The nuclear p53 and PCNA staining in SCCs of groups PUVA-F and PUVA-U (Figure 1) was consistent with an increased half-life of these proteins in the nucleus, and good correlation occurred between p53 and PCNA expression.
In our study, the route of exposure was similar to that of humans (Table 1); 8-MOP was administered orally, and the UV intensity was close to that used for patients (Dunnick et al., 1991; Nataraj et al., 1996). The mutational pattern in SCCs in the present study partially matched that previously described for mice, where 8-MOP was topically applied and a different level of UV light intensity was used (Nataraj et al., 1996; Monti et al., 2000; Santamaria et al., 2002). The thymine nucleotide was a target of the 8-MOP photocycloaddition (Gasparro et al., 1998) (Figure 3, Table 5). Similarly, in our study, 8/9 (89%) of the mutations occurred on a thymine nucleotide in the PUVA-F group, revealing the specific interaction between 8-MOP and primary UVA light, with a wavelength between 320 and 400 nm. The high incidence of mutations detected in exon 6 may also reflect the higher T content in exon 6 of the 4 exons analyzed (Table 4).
Previous experimental studies reported that thymine (T) is the preferential base for adduct formation while interacting with photoactivated 8-MOP (Gasparro et al., 1997; Gasparro, 1998) (Figure 3); moreover, the mutations at 5′-TpA/ApT or 5′ TpT were described as a target of the coupled 8-MOP/UVA action in mice (Nataraj et al., 1996). The p53 mutagenic effect of PUVA therapy has been widely studied, but the mutational profile described in rodent studies is not consistent with the human data (Peritz and Gasparro, 1999). The 5′-TpA-3′ dinucleotide sequence, described in mice as the preferable target for PUVA-induced mutations in SCCs (Nataraj et al., 1996), has not been found in SCCs from patients treated with PUVA (Wang et al., 1997; Gasparro, 1998). Human SCCs from PUVA-treated patients display a high incidence of solar-like p53 mutations, such as C→T or, more commonly CC→TT. PUVA therapy is therefore considered to be a potent enhancer of UV mutations in humans exposed to sunlight (Wang et al., 1997; Gasparro, 1998). However, recent studies showed that squamous cell carcinomas from PUVA-treated humans had a high frequency of PUVA type mutations in p53 genes (Wolf et al., 2004; Stern et al., 2002). Humans treated for chronic skin diseases, such as psoriasis and vitiligo, are sometimes advised to be exposed to sunlight to alleviate clinical symptoms (Stern et al., 1984; Cather and Menter, 2002). In humans, UV exposure is a possible cause of p53 mutations (Giglia-Mari and Sarasin, 2003), which is further enhanced by PUVA treatment. The SCC mutational pattern seen in our study, however does not completely agree with these types of classification. Of the 22 p53 mutations identified in SCCs of group PUVA-F and PUVA-U, 9 (41%) involved a thymine base; 8/9 occurred in the PUVA-F group. The dinucleotide sequence 5′-TpA-3′ was involved in 3/22 (14%) mutations: 2 in group PUVA-F and 1 in group PUVA-U.
The p53 mutational frequency in tumors from our study was different from the distribution of p53 mutations in human skin cancers. Most of the mutations found in this study were detected in exon 6, while most of the hot spots in the human p53 gene are grouped in exons 5, 7, and 8, near the center and the end of the DNA-binding domain (Figure 4). The findings are not surprising, since p53 exons 5, 7, and 8 have a high frequency of 5′-CpG-3′ methylated cytosines carrying dinucleotides, which are known to deaminate spontaneously to thymine inducing a transiton mutation C→T (Giglia-Mari and Sarasin, 2003; Szymanska and Hainaut, 2003). Most of the human p53 hot spots code for arginine that mostly results from a 5′-CpG-3′ site in the codon sequence. In contrast, 8-MOP has a remarkable tendency to form adducts with thymine nucleotides; thus, the identification of p53 mutations in thymine-rich areas is consistent with the predominant findings of mutations in exon 6 in our study.
Several studies have evaluated UV-light mutagenic effects on the mouse genome. The mutational spectra obtained from these studies demonstrated dipyrimidine mutational effects of UV light on both p53 and H-ras genes (Kress et al., 1992; Kanjilal et al., 1993; Nishigori et al., 1994; van Kranen et al., 1995; Berg et al., 1996; Brash et al., 1996; Dumaz et al., 1997; de Gruijl et al., 2001; Ichihashi et al., 2003). In SCCs and SCPs from all the groups analyzed in our study, however no dipyrimidine mutations were found for H-ras codon 61 and for p53. UV-associated mutations (C→T or CC→TT) were not detected in our study, because mice were exposed to therapeutic UV light with a power that is much lower than the UV band in sunlight. Furthermore, the UV wavelength chosen was the one that best activates the photo-damaging action of 8-MOP, thus increasing mainly the mutational frequency detected following PUVA therapy. Our findings are consistent with the low incidence of SCCs and low and random p53 mutational pattern found in the SCCs from groups treated with either filtered or unfiltered UV light, with the same intensity used in both 8-MOP and UVA groups.
The low spontaneous tumor rate, together with the low mutational frequency found in mice of group C and in groups treated only with UV, filtered or unfiltered (F and U), when compared to the high rate of tumors and p53 mutations in mice treated with 8-MOP and UVA (groups PUVA-F and PUVA-U), further confirm the carcinogenic and mutagenic potential of PUVA therapy in mice. PUVA remains a powerful treatment strategy for chronic skin diseases, such as psoriasis, vitiligo, and cutaneous T-cell lymphoma; however, the carcinogenic risk from such treatment should not be underestimated. Preventive measures should be adopted when this clinical approach is used. Doses, schedules, and methods of treatment should minimize the 8-MOP intake and UVA exposure to the effective doses that reduce the risk of skin cancer development. Concurrently, a detailed screening plan should be followed to allow early diagnosis of SCCs in humans treated with PUVA.
Footnotes
Acknowledgments
We thank Norris Flagler for his photographic expertise.