p53 alterations in human tumors often involve missense mutations that may confer dominant-negative or gain-of-function properties. Dominant-negative effects result in inactivation of wild-type p53 protein in heterozygous mutant cells and as such in a p53 null phenotype. Gain-of-function effects can directly promote tumor development or metastasis through antiapoptotic mechanisms or transcriptional activation of (onco)genes. Here, we show, using conditional mouse technology, that epithelium-specific heterozygous expression of mutant p53 (i.e., the p53.R270H mutation that is equivalent to the human hotspot R273H) results in an increased incidence of spontaneous and UVB-induced skin tumors. Expression of p53.R270H exerted dominant-negative effects on latency, multiplicity, and progression status of UVB-induced but not spontaneous tumors. Surprisingly, gain-of-function properties of p53.R270H were not detected in skin epithelium. Apparently, dominant-negative and gain-of-function effects of mutant p53 are highly tissue specific and become most manifest upon stabilization of p53 after DNA damage. [Cancer Res 2007;67(10):4648–56]
- 5p3 mutation
- conditional mouse models
- DNA damage
- Animal models of cancer
- Genotype-phenotype correlations
p53 is the most extensively studied tumor suppressor gene, encoding a transcriptional regulator that controls cell cycle progression and apoptosis ( 1). In unstressed cells, p53 is present in a latent form and is maintained at low levels through targeted degradation. In response to DNA damage or cellular stress, its activity increases to exert its function as a transcription factor ( 2), resulting in a cascade of events that eventually prevent tumor development. The p53 gene is altered in more than 50% of spontaneous tumors in humans, and germ-line p53 mutations confer cancer predisposition to individuals with Li-Fraumeni syndrome ( 3, 4).
In contrast to modifications found in other tumor suppressors, which are typically deleted, truncated, silenced, or otherwise down-regulated in many cancers, the majority of p53 alterations are missense mutations in the DNA binding domain, disrupting the ability of the protein to bind DNA and activate transcription ( 5– 7). Loss of the remaining wild-type allele frequently occurs in a later stage of tumor development. Functional consequences of missense mutations compared with deletions of p53 have been extensively studied in vitro (ref. 8 and references therein). p53 missense mutations are commonly loss-of-function mutations that mimic absence of p53, because mutant p53 is no longer able to inhibit cell cycling or induce apoptosis. In addition, these missense mutations are thought to have dominant-negative and/or gain-of-function characteristics ( 8– 10). Dominant-negative p53 mutants can inhibit the function of wild-type p53 through protein-protein interactions ( 11), whereas gain-of-function mutants have additional functions not seen in wild-type p53 ( 12). However, the majority of studies describing these mutant p53 characteristics are based on in vitro overexpression analyses.
The first in vivo studies to determine the fundamental role of p53 as a tumor suppressor were done in mice with heterozygous or homozygous deletions of p53 ( 13– 15). To more accurately mimic LFS and examine the effect of (specific) p53 missense mutations on tumorigenesis in vivo, transgenic strains were produced, which overexpress mutant p53 ( 16– 20). Collectively, studies with these transgenic models show that mutant p53 overexpression results in accelerated spontaneous and carcinogen-induced tumorigenesis, with a dominant-negative effect of the mutant protein ( 16). To more physiologically model LFS in mice and investigate the potential dominant-negative or gain-of-function effects by mutant p53 in vivo, (conditional) mutant p53 knock-in mice carrying targeted mutations in the endogenous p53 gene locus were recently generated ( 21, 22). One p53 mutant mouse strain was generated with an arginine to histidine mutation at codon 172, corresponding to the p53.R175H hotspot mutation that destroys the structural integrity of the p53 protein ( 21, 22). A second strain was engineered to carry an arginine to histidine mutation at p53 codon 270 ( 22), which corresponds to the p53.R273H hotspot mutation frequently found in human cancers and replaces an arginine that directly contacts DNA. The phenotypes of these mutant mice indicate clear gain-of-function properties of mutant p53, with a metastatic tumor phenotype as the most striking difference between missense mutant mice and p53-null mice. Furthermore, tumor spectra changed, with an overall increase in the incidence of carcinomas and B-cell lymphomas in p53R270H/+ mice and osteosarcomas in p53R172H/+ mice ( 22).
However, the above-described studies focused on effects of constitutive expression of the p53 point mutants in all tissues ( 21, 22). One of the strengths of conditional mouse technology is the potential for tissue-specific analyses of the adverse effects of gene alterations using transgenic mice with tissue- and cell-specific Cre expression. Because it is well known ( 6, 23, 24) that the tumorigenic potential of individual mutant p53 proteins can vary between cell types and tissues, it would be important to analyze the effect of such mutant proteins exclusively in a tissue or cell type of interest. Furthermore, tissue-specific analysis can be preferred to investigate the effect of carcinogen exposure. We and others have previously studied the tissue-specific effects of the p53 missense mutation at codon 270 in mammary gland and lung epithelium, respectively ( 25, 26). Both models mimic human tumor development and, as such, are suitable to study and develop better treatment strategies for breast and lung cancer patients, respectively.
Here, we analyze whether the p53.R270H mutation has dominant-negative and/or gain-of-function properties in mouse skin epithelium, either unchallenged or challenged by DNA damage–inducing chronic UVB exposure. Nonmelanoma skin cancer is currently the most common type of human cancer, and its incidence is increasing at an astonishing rate ( 27). p53 is found mutated in 50% of skin cancers overall and up to 90% of squamous cell carcinoma (SCC) specifically ( 28). Interestingly, in the vast majority of these tumors, the remaining wild-type p53 allele is intact rather than lost through loss of heterozygosity (LOH) as frequently observed in other tumor types ( 14, 29). This might indicate that the heterozygous presence of a p53 mutation is sufficient to trigger skin tumor development. Here, heterozygous p53LSL-R270H/+ mice were crossed with K14cre transgenic mice, resulting in epithelium-specific expression of the mutant p53.R270H protein. Skin tumor development and acute cellular responses after UVB exposure were subsequently compared with those in conditional heterozygous and homozygous p53 knockout mice as controls. In this way, a clean skin-specific comparison of effects can be made between p53 missense and p53 null mutations. We show here that, especially after UVB-induced DNA damage, a p53.R270H mutation shows dominant-negative but not gain-of-function activities in skin tumor development.
Materials and Methods
Mice and Genotyping
Cloning of the p53LSL-R270H targeting vector, homologous recombination experiments in embryonic stem cells, and generation of conditional p53LSL-R270H mice were described elsewhere ( 22, 30). Genotyping of p53LSL-R270H mice was done by a PCR/digestion–based assay as described previously ( 25).
K14cre mice expressing Cre-recombinase under control of a human K14 gene promoter were used to induce Cre-mediated deletion of the floxed stop-cassette specifically in dividing cells of stratified epithelia ( 31). The presence of Cre-recombinase was determined by PCR as described before ( 25).
The presence of a p53F2-10 allele in p53 conditional knockout mice and a deletion of exons 2 to 10 in p53 (p53Δ2-10 allele) were detected as described earlier ( 31).
To obtain homozygous hairless mice of all genotypes, all strains were crossed twice to hairless SKH1 mice (Charles River). Animals in experiment were obtained from crossing these hairless p53 mutant and/or p53 conditional knockout mice to hairless K14cre mice.
Analysis of Spontaneous and UVB-Induced (Skin) Tumor Development
Spontaneous tumor development was determined in groups of six female and two male mice of the following genotypes: K14cre;p53+/+, K14cre;p53LSL-R270H/+, K14cre;p53F2-10/+, K14cre;p53F2-10/F2-10, and K14cre;p53LSL-R270H/F2-10. For UVB experiments, groups of 14 mice of the same genotypes, all consisting of 50% females and 50% males, were irradiated daily with a dose of 600 J/m2 UVB using Philips TL12 lamps, starting at the age of 6 weeks. All mice were checked weekly for the development of tumors until a maximum age of 78 weeks. Mice carrying tumors ≥4 mm in size or ≥15 small tumors were sacrificed. Control tissues and a maximum of six skin tumors were collected and processed for histopathology and DNA/RNA isolation following standard procedures.
Short-Term UVB Exposure
For determination of the maximum tolerated dose (MTD), two mice of each genotype were exposed to daily doses of UVB of 0, 500, 750, 1,000, or 1,250 J/m2 during seven subsequent days and checked for erythema and other acute effects. For determination of apoptotic/sunburn cells, two mice per genotype were exposed to a single dose of 500 and 1,000 J/m2. Front and back skin were isolated together with control tissue 24 h after the UVB exposure and processed for histopathology and DNA/RNA isolation following standard procedures.
Molecular Analysis of Tumors and Tissues
Expression of the R270H point mutation or deletion of exons 2 to 10 of p53. Analysis of in vivo expression of the p53.R270H mutant allele in skins, control tissue (spleen), and tumors was done as described earlier ( 25). Deletion of the conditional p53F2-10 allele in skin (tumors) was determined by PCR as described for mouse genotyping above.
Determination of loss of the wild-type p53 allele (LOH) and additional p53 gene mutations in skin tumors. LOH was detected by reverse transcription-PCR (RT-PCR) as described previously ( 25). Frozen skin tumors of UVB-exposed mice were analyzed for acquired p53 mutations by direct sequencing as described earlier ( 32).
Histology and Immunohistochemistry
Collected tissues and tumors from short-term and long-term UV studies were preserved in a neutral aqueous phosphate-buffered 4% solution of formaldehyde (10% neutral buffered formalin). The tissues were embedded in paraffin wax, sectioned at 5 μm, and stained with H&E for histopathologic evaluation.
Immunohistochemistry of skin sections from short-term UV studies. Isolated skin samples from short-term UV studies were stained with the following antibodies: caspase-3 (Asp175, 1:200; Cell Signaling), Bax (P19, 1:50; Santa Cruz Biotechnology), Puma (Puma, 1:2,000; Cell Signaling), and Perp ( 33). All stainings (except for Perp) were done as described earlier ( 34) using a secondary goat anti-rabbit/biotin antibody (Vector Laboratories) and subsequently a streptavidin-complex peroxidase Elite kit (Vector Laboratories). For antigen retrieval, deparaffinized tissue sections were heated for 30 min in a 10 mmol/L citrate buffer (pH 6.0) at 95°C. Perp staining was done as reported previously ( 33).
Immunohistochemistry of skin tumors from long-term UV studies. Four tumors per genotype were analyzed for the expression of the oncogenes cyclin D1 and H-Ras. Antibodies used were the rabbit monoclonal antibody SP4 for cyclin D1 (1:200; GeneTex, Inc.) and F235 for H-Ras (1:2,000; Santa Cruz Biotechnology). The staining protocol for cyclin D1 was identical to the protocols described in the previous section. For H-Ras stainings, a secondary donkey anti-mouse/biotin antibody (Jackson ImmunoResearch) was used.
Statistical analyses of tumor-free survival curves included calculation of Kaplan-Meier distributions of survival of two different treatment groups and comparison by a two-sided log-rank test (SPSS, version 11). Multiplicity of tumors was statistically analyzed by an unpaired t test. P < 0.05 was taken as significant.
Skin-specific p53.R270H expression and p53.F2-10 deletion. To assess whether the introduction of K14cre leads to epithelial-specific removal of the transcriptional stop-cassette and subsequent expression of the p53.R270H mutant protein, we determined expression of the mutation in skin and control tissue of young K14cre;p53LSL-R270H/+ mice by RT-PCR. As is clear from Fig. 1A , the p53.R270H mutation is expressed in skins of young adult K14cre;p53LSL-R270H/+ mice. As expected, in non-epithelium tissue (i.e., spleen) of the same mice, only the wild-type allele was detectable. K14cre-induced recombination of the p53F2-10 allele in skin of young adult homozygous and heterozygous conditional p53 knockout mice was very efficient as all skin samples tested showed a recombined p53F2-10 allele ( Fig. 1B).
Survival and spontaneous skin tumor development in K14cre;p53LSL-R270H/+ mice. To examine a potential dominant-negative and/or gain-of-function effect of the p53.R270H mutation on spontaneous tumorigenesis of epithelial cell origin, both female and male K14cre;p53+/+, K14cre;p53LSL-R270H/+, K14cre;p53F2-10/+, K14cre;p53F2-10/F2-10, and K14cre;p53LSL-R270H/F2-10 in a hairless background were followed as they aged for the development of spontaneous tumors. All genotypes analyzed seem susceptible for spontaneous tumor development, especially in the skin ( Fig. 2A ). Large differences in tumor latency times occur when functional p53 is (partly) lacking, as significantly reduced tumor latencies are observed in both heterozygous K14cre;p53LSL-R270H/+ and K14cre;p53F2-10/+ mice when compared with K14cre;p53+/+ littermates (P = 0.0013 and 0.0011, respectively; Fig. 2B). However, no difference in tumor latency time could be observed between heterozygous K14cre;p53LSL-R270H/+ and K14cre;p53F2-10/+ mice (P = 0.8645), pointing to the absence of a dominant-negative effect of the p53.R270H mutation in epithelium cells during spontaneous (skin) tumor development. In contrast, a clear p53 gene dosage effect was present, as homozygous K14cre;p53F2-10/F2-10 mice show a significantly accelerated (skin) tumor development when compared with heterozygous K14cre;p53F2-10/+ (P = 0.0002) and K14cre;p53LSL-R270H/+ mice (P = 0.0002). Spontaneous tumor-free survival curves of K14cre;p53LSL-R270H/F2-10 and K14cre;p53F2-10/F2-10 mice seem identical, indicating that in this K14cre mouse model, a gain-of-function effect of the p53.R270H mutant protein is absent during spontaneous tumor development. The majority of tumors develop in the skin, but also epithelial mammary tumors were found as reported earlier after recombination of conditional p53 and Brca alleles with K14cre ( 31). Here, mammary tumors were only found in K14cre;p53LSL-R270H/+ mice, similar to results reported recently in a mammary gland–specific WAPCre background ( 25). Molecular analysis of skin tumors as well as mammary tumors clearly showed p53.R270H expression that was absent in a lymphoma in the jejunum ( Fig. 2C), confirming epithelium-specific expression of mutant p53. In line with this, deletion of exons 2 to 10 in K14cre;p53F2-10/+ mice was visible in all skin tumors analyzed ( Fig. 2D).
UVB-induced skin tumor development in p53-defective mice. The effect of the p53.R270H hotspot mutation on the tumor suppressive activity of p53 after DNA damage was subsequently investigated by exposure of mice to a daily dose of UVB radiation. First, the MTD was determined by exposing mice to various doses of UVB, ranging from 500 to 1,250 J/m2, during seven subsequent days. MTDs of all p53 defective mouse strains were not different from wild-type K14cre mice on a SHK1 hairless background (i.e., in between 750 and 1,000 J/m2). The presence of a R270H mutation in p53 and/or a deletion of one or two p53 alleles apparently do not result in an altered acute sensitivity of the skin to UVB irradiation. Skin tumor-free survivals were determined after a daily dose of 600 J/m2 UVB and are shown in Fig. 3A . Clearly, K14cre;p53LSL-R270H/+ mice were much more skin tumor prone than K14cre;p53+/+ mice (P < 0.0001, Fig. 3B), indicating that the p53.R270H mutant plays an important role in initiation of UVB-induced skin carcinogenesis. Interestingly, and in contrast to spontaneous skin tumor development, tumor latency times of K14cre;p53LSL-R270H/+ mice after UVB exposure were significantly reduced compared with those in K14cre;p53F2-10/+ mice (P = 0.0057), pointing to a dominant-negative effect of the p53.R270H mutation. Again, a p53 gene dose effect was evidently present, as skin tumors developed significantly faster in homozygous K14cre;p53F2-10/F2-10 mice compared with heterozygous K14cre;p53F2-10/+ mice (P < 0.0001). Consistent with the spontaneous tumor data, no gain-of-function effect of the p53.R270H mutation was detectable on tumor latency in skin epithelium after UVB irradiation.
The presence of a dominant-negative but absence of a gain-of-function effect of the p53.R270H mutation on UV-induced latency times was also reflected in multiplicity of UVB-induced skin tumors. Figure 3C shows the observed number of skin tumors after 100 and 120 days of UVB exposure. The average number of skin tumors per mouse at day 100 was 6.5 ± 1.4 in K14cre;p53LSL-R270H/+ mice, with only 0.6 ± 0.2 tumors developed in wild-type mice during the same time period. Furthermore, a lower number of tumors was observed in K14cre;p53F2-10/+ mice (2.4 ± 0.6) compared with K14cre;p53LSL-R270H/+ mice. In contrast, tumor multiplicity in homozygous K14cre;p53F2-10/F2-10 mice was similar to that in K14cre;p53LSL-R270H/F2-10 mice. The number of tumors per animal increased over time in all genotypes analyzed and, although smaller, differences in tumor multiplicity between genotypes were also detectable after 120 days of UVB exposure ( Fig. 3C).
Molecular analysis of UVB-induced skin tumors. To verify the observed dominant-negative effect of the p53.R270H mutation on skin tumor latency as well as tumor multiplicity on the molecular level, a representative nonselective subset of 11 tumors from 11 individual K14cre;p53LSL-R270H/+ mice was analyzed for LOH by RT-PCR analysis. No LOH was observed in any of the analyzed UVB-induced skin tumors ( Fig. 3D), indicating that loss of the wild-type p53 allele was not advantageous for tumor formation.
It is known that during later phases of UV-induced skin tumor development in both humans and mice, p53 function can be (further) altered by acquired gene mutations ( 32, 35– 40). We investigated therefore whether the development of UV-induced skin tumors in K14cre;p53LSL-R270H/+ mice was solely the result of the initial p53.R270H mutation, or whether it was accompanied by the formation of additional p53 mutations. For this, we analyzed at random 17 skin tumors from 16 individual K14cre;p53LSL-R270H/+ mice by direct exon sequencing of the p53 gene. The majority of tumors (14 of 17) contained one or more additional p53 mutations (Supplementary Table S1).
Histologic characterization of skin tumors. Spontaneous and UV-induced skin lesions were analyzed for histologic characteristics. Both groups of mice developed epithelial, proliferative lesions forming a continuum from nonneoplastic squamous keratosis and acanthosis to squamous cell papillomas (SCP), actinic keratosis, and keratoacanthoma with increasing degrees of atypia and dysplasia, carcinoma in situ (CIS), up to invasive SCC ( Fig. 4 ). Basal cell carcinomas were also found in unexposed mice with deleted p53 alleles, but not in mice with a p53.R270H mutation or after UV exposure. There seems to be a trend towards more spontaneous carcinomas in K14cre;p53LSL-R270H/+ mice when compared with mice with a heterozygous deletion of p53, with an incidence of carcinomas almost similar to that found in K14cre;p53F2-10/F2-10 and K14cre;p53LSL-R270H/F2-10 mice ( Fig. 4A).
The majority of proliferative lesions after UVB in all genotypes were classified as actinic keratosis, representing premalignant precursors of SCC. No gross differences in UV-induced actinic keratosis numbers were detectable between the five genotypes analyzed (data not shown). For determination of tumor progression status, the ratios between benign versus malignant skin tumors was calculated. Calculated ratios were 0.47, 0.5, and 0.45 for K14cre;p53+/+, K14cre;p53LSL-R270H/+, and K14cre;p53F2-10/+ mice, respectively, indicating that although inactivation of one p53 allele in skin epithelium has an influence on time of onset of UV-induced tumors (see Fig. 3A), no effect was observed on tumor progression to a malignant state. Furthermore, no dominant-negative effect of p53.R270H mutant protein could be observed on skin tumor progression. Interestingly, strongly accelerated tumor progression was observed in mice with homozygous p53 inactivation, as already reported previously for UV-exposed conventional p53−/− mice ( 35). The ratio benign versus malignant tumors was very low in K14cre;p53F2-10/F2-10 as well as K14cre;p53LSL-R270H/F2-10 mice (≤0.05), indicating that loss of the second p53 allele has dramatic effects on tumor progression.
A p53 gain-of-function effect has been described to be accompanied by overexpression of oncogenes ( 12). To investigate possible gain-of-function effects of the p53.R270H mutation in skin tumor development, sections of UV-induced skin tumors were stained with antibodies against cyclin D1 and H-Ras, which are both associated with skin tumor development in mice and men ( 41, 42). Figure 4B shows a representative panel of tumors stained for H&E and cyclin D1. Prominent cyclin D1 expression was visible in all UV-induced skin tumors tested, varying from SCP to SCC. However, no differences could be observed between tumors developed in the various p53-defective genotypes (data not shown), pointing to the absence of a gain-of-function effect. The same results were obtained for H-Ras: clear expression was visible in UV-induced skin tumors without genotypic differences (data not shown).
Analysis of sunburn cells after short-term UVB exposure. Inactive p53 in mouse skin has been reported to reduce the appearance of sunburn cells (apoptotic keratinocytes generated by overexposure to UV), and as such, the loss of p53 provides a survival advantage to UV-damaged cells ( 43). To identify potential dominant-negative effects of the p53.R270H mutation on apoptosis in an early stage, we analyzed skin samples of the five different p53-defective genotypes for the presence of apoptotic cells after short-term UVB exposure using H&E and caspase-3 staining. In addition, the UVB-exposed skin sections were stained for different known p53 targets involved in apoptosis, with Bax and Puma representing the mitochondrial intrinsic pathway and Perp representing the cell membrane extrinsic pathway ( 44). Figure 5A shows that the highest amount of caspase-3–positive cells are found in K14cre;p53+/+ skin exposed to a single dose of 500 J/m2 UVB, and all UV-exposed skin samples of p53-defective mice showed a significant decrease in caspase-3–positive cells. However, hardly any difference could be observed between K14cre;p53LSL-R270H/+, K14cre;p53F2-10H/+, or K14cre;p53F2-10H/F2-10 mice ( Fig. 5A). In line with this, the amount of sunburn cells after a single exposure to UVB, counted in H&E sections, was increasing with dose, with the highest amount in K14cre;p53+/+ skin and decreased numbers of sunburn cells in all p53-defective genotypes (data not shown). Figure 5B shows examples of Bax stainings of K14cre;p53+/+, K14cre;p53LSL-R270H/+, and K14cre;p53F2-10/F2-10 skin. No detectable expression of the Bax protein was observed in unexposed skin, whereas UV exposure resulted in a slight increase in Bax-positive cells in wild-type skins. The number of Bax-positive cells was decreased in p53-defective skin compared with K14cre;p53+/+ skin, with similar staining patterns in skins of K14cre;p53LSL-R270H/+ and K14cre;p53F2-10/+ mice. Homozygous inactivation of p53 in skin resulted in a slightly lower amount of Bax-positive cells than heterozygous inactivation. Similar results were obtained for staining against Perp (data not shown).
In the present study, we have investigated potential dominant-negative and/or gain-of-function effects of a p53.R270H mutation, equivalent to the human hotspot R273H mutation, on development and progression of mouse skin tumors. For this, conditional mutant p53LSL-R270H mice ( 22) were crossed with K14cre mice ( 31), resulting in physiologic expression levels of the mutant p53.R270H protein in skin epithelium. This analysis was done for skin tumors that arose spontaneously as well as after the induction of DNA damage, because one could hypothesize inhibitory or oncogenic effects of mutant p53 might be more abundant following p53 activation by DNA damage. Because we used as controls conditional K14cre;p53F2-10 knockout mice on a similar hairless background as the K14cre;p53LSL-R270H mice, a clean comparison could be made between deletion and mutation of p53 in skin without early occurrence of lymphomas or sarcomas. Lethality caused by these non-epithelial tumors has hampered analysis of UV-induced skin tumor formation in conventional p53-null mice up to now ( 35, 45).
Our data indicate absence of dominant-negative and gain-of-function effects of the R270H mutation on spontaneous skin tumor development, because (a) heterozygous K14cre;p53LSL-R270H/+ mice do not show accelerated tumor development compared with heterozygous K14cre;p53F2-10/+ mice, and because (b) homozygous K14cre;p53F2-10/F2-10 and K14cre;p53LSL-R270H/F2-10 mice develop tumors with similar latencies. These results suggest that expression of mutant p53.R270H in skin epithelium does not interfere with the function of wild-type p53 and, furthermore, does not have novel oncogenic functions, at least not with respect to tumor latency. These findings are in contrast to previous results found in mammary epithelium ( 25) and lung epithelium ( 26), where a spontaneous dominant-negative phenotype of the p53.R270H protein was found, indicating that dominant-negative properties are tissue specific and/or dependent of p53 expression levels.
Although we have not detected dominant-negative and gain-of-function effects of p53.R270H on the latency of spontaneous skin tumors, histopathologic examination revealed a small change in tumor spectrum ( Fig. 4A). K14cre;p53LSL-R270H/+ mice spontaneously develop slightly more carcinomas compared with K14cre;p53F2-10/+ mice, suggesting a possible dominant-negative effect of the p53.R270H mutation on tumor progression. Although the absolute numbers of tumors analyzed are too low to draw firm conclusions, this increase in carcinoma development is in line with previous results in p53R270H/+ mice, which showed a significantly higher incidence of carcinomas in lung, liver, kidney, intestine, and skin compared with conventional p53+/− mice ( 22). However, the additional gain-of-function effect observed in p53R270H/− mice (which develop more carcinomas than conventional p53−/− mice) is absent in our skin-specific p53 mouse mutants, because no difference in skin tumor progression could be observed between K14cre;p53LSL-R270H/− and K14cre;p53F2-10/F2-10 mice. An explanation for this could lie in the fact that the conventional p53−/− mice used in earlier studies did not develop carcinomas at all ( 22), likely due to early death from lymphomas. This clearly shows the importance of using tissue-specific mouse mutants for genotype-phenotype correlation studies.
Here, in a SKH-1 background, mammary tumors were only found in K14cre;p53LSL-R270H/+ mice, whereas in a recently conducted study in a FVB background, also K14cre;p53F2-10/F2-10 mice developed mammary tumors at high incidence. 5 Genetic background differences are probably causing this apparent discrepancy.
Interestingly, effects of the presence of one mutant p53.R270H allele after the induction of DNA damage are very different from those observed spontaneously. Chronic in vivo UV exposure leads to the development of a broad pattern of skin tumors in mice of all p53 genotypes. In contrast to the spontaneous tumor development study, a clear dominant-negative effect of the p53.R270H mutation on both latency and multiplicity of UV-induced skin tumors was detected. Apparently, the expression levels of mutant p53 are crucial for manifestation of dominant-negative activity during skin tumor formation. These findings are corroborated by previous results from these p53 point mutant mice ( 21, 22), where accumulation of mutant p53 was found following induction of DNA damage in homozygous mutant MEFs. In addition, the dominant-negative activity of mutant p53 increased after DNA damage in various tissues of heterozygous mutant mice ( 21, 22). We have found similar stabilization of mutant p53.R270H in the skin after short-term UVB exposure (data not shown). It is well known that upon induction of DNA damage, wild-type p53 protein accumulates and stabilizes, resulting in transactivation of target genes protecting the genome by a variety of cellular responses ( 2). Apparently, to uncover its dominant-negative or gain-of-function characteristics, mutant p53 also needs to accumulate and be stabilized, presumably through mechanisms similar to that of wild-type p53 ( 46). Stabilization of mutant p53 may strongly inhibit tumor suppressive functions of wild-type p53 and as such account for the observation that K14cre;p53LSL-R270H/+ mice display a more severe tumor phenotype than K14cre;p53F2-10/+ mice after exposure to UVB. However, due to the construction of the K14cre;p53LSL-R270H mouse model, cells surrounding the tumor are p53+/−, whereas these are p53+/+ in K14cre;p53F2-10 mice. A supporting effect of the surrounding cells on tumor development can therefore not be excluded.
The accelerated skin tumor development in K14cre;p53LSL-R270H/+ mice was a clear dominant-negative effect of the mutation, because no loss of the wild-type p53 allele was found in these tumors. The absence of LOH was not entirely unexpected, because also previous studies showed that LOH was a rare event in UV-induced skin tumors of p53+/− mice ( 35, 36, 43, 47, 48). However, in the majority of skin tumors analyzed, we identified one or more additional p53 mutations in codons described before for various p53-defective mice exposed to UV ( 32, 47). It remains yet to be determined whether these additional mutations inactivate the wild-type allele and are critical for skin tumor formation, because it might well be that these mutations are formed during the process of tumor development when genome instability is a well-known characteristic.
Dominant-negative effects of p53.R270H protein could not be shown in progression of skin tumors after UVB exposure, because the same fraction of malignant tumors was found in heterozygous point mutants, heterozygous conditional knockouts, and even wild-type p53 mice ( Fig. 4A). Maybe the load of UV-induced DNA damage was too high to find subtle differences in tumor progression at the time of tumor isolation. We used tumor size and multiplicity as an end point in our experiments (i.e., tumors ≥ 4 mm or ≥15 small tumors). For detection of subtle differences, it might be better to analyze tumor progression at fixed time points after start of UV exposure, as was done for determining multiplicity of tumors. Only mutation or deletion of both p53 alleles resulted in significantly more malignant CIS and SCC, most likely because total absence of p53 results in loss of cell cycle checkpoints, impaired DNA damage responses, and apoptotic resistance ( 43).
Tumor-associated p53 mutant proteins can exert gain-of-function activity via inhibition of p53-independent apoptosis as well as through activation of oncogenes ( 8). Overexpression of cyclin D1 has been found in (early stages of) many tumors, including mouse and human skin tumors ( 41, 42). Furthermore, the expression of cyclin D1 is related to sun exposure ( 41). Several studies also show a correlation between cyclin D1 expression levels and Ras activation. Cyclin D1 is a critical target of oncogenic Ras in mouse skin carcinogenesis and has a role as a downstream mediator of Ras activity during tumor development ( 42). Therefore, protein levels of these two oncogenes were determined in skin tumors from mice of all p53 genotypes, to identify potential gain-of-function effects of p53.R270H on oncogene expression in skin tumor development. Lack of gain-of-function properties was suggested by similar patterns of cyclin D1 and H-Ras staining in UV-induced skin tumors of mice of all genotypes, including p53R270H/F2-10 and p53F2-10/F2-10. Apparently, protein levels of skin tumor–related oncogenes are not directly increased by the presence of p53.R270H mutant protein, at least not the well-known examples analyzed here.
The dominant-negative effect of p53.R270H on UVB-induced tumor induction could not be clearly shown in the early apoptotic response after short-term UVB exposure. Although the number of apoptotic cells was decreased dramatically compared with wild-type responses, no differences were observed between K14cre;p53LSL-R270H/+ and K14cre;p53F2-10/+ mice in apoptotic cell numbers and expression of known p53 targets involved in different apoptotic pathways. Apparently, the selection of cells resistant against apoptosis in an early stage is not the only crucial factor in initiation and development of UVB-induced skin tumors. A decreased ability of damaged cells to undergo cell cycle arrest, or the induction of mutations in preneoplastic lesions, might also strongly influence tumor development.
In conclusion, our studies show that p53.R270H mutant protein has dominant-negative but not gain-of-function properties in skin epithelium. More specifically, dominant-negative features of mutant p53 protein in skin epithelium are exclusively apparent after the induction of DNA damage. Whether these DNA damage–related differences in mutant p53 characteristics are also found in other tissues and/or after exposure to other DNA-damaging compounds is an interesting question that remains to be addressed.
Grant support: Dutch Cancer Society grant KWF 2000-2352 and NIH/National Institute of Environmental Health Sciences, Comparative Mouse Genomics Centers Consortium grant 1UO1 ES11044-02.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank the Central Animal Facility Laboratory (NVI-GPL) for their skillful (bio)technical support.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
↵5 X. Liu et al. Somatic loss of BRCA1 and p53 in mice induces mammary tumors with pathologic and molecular features of human BRCA1-mutated basal-like breast cancer, submitted for publication.
- Received December 20, 2006.
- Revision received February 16, 2007.
- Accepted February 26, 2007.
- ©2007 American Association for Cancer Research.