Spontaneous and UV Radiation–Induced Multiple Metastatic Melanomas in Cdk4^R24C/R24C/TPras Mice Free

In some transgenic strains of mice [Mt-Hgf/Sf (1) and TPras (2, 3)], a single erythemal UV radiation (UVR) dose of 9 kJ/m² to 2- to 4-day-old neonates induces melanoma development with greatly increased penetrance compared with chronic UVR treatments to adult mice of the same genotype. In humans, melanoma susceptibility is often characterized by mutations of CDKN2A (which encodes p16^INK4A and p14^ARF) and CDK4. INK4A is a CDK inhibitor that binds to and inhibits CDK4 (and CDK6), which otherwise phosphorylates pRb and induces G₁-S phase progression. ARF acts through a distinct pathway involving stabilization of Trp53 through abrogation of murine double minute-2–induced Trp53 degradation (4). Underlining the importance of CDK4 in melanoma susceptibility are germ-line mutations (R24C or R24H) in some melanoma kindreds (5, 6) that render the protein resistant to INK4A inhibition. It is unclear if such mutations are functionally equivalent to deletion of INK4A as other inhibitors (p15^INK4B, p18^INK4C, and p19^INK4D) are also capable of binding and inhibiting CDK4 and their interaction is also thought to be abrogated by the R24C mutation.

Ras/Raf/mitogen-activated protein kinase (MAPK) pathway activation is also critical for melanoma development. The MAPK pathway is activated through mutation of BRAF in 62% to 72% of sporadic melanomas and by RAS mutations in about 10% (7, 8). The relationship between MAPK activation and UVR exposure is complex. BRAF mutations are rarely found in mucosal and acral melanomas (i.e., those not associated with sun exposure; refs. 9, 10). Interestingly, they are also not commonly found in areas of chronic UVR exposure (e.g., the face, where NRAS mutations are more likely) but are often found in melanomas from areas of intermittent sun exposure (e.g., the back and trunk; ref. 10). Despite differences in specificity between Ras family members, mice carrying mutations either in Hras (2, 3) or Nras (11) develop melanoma. Further in vivo evidence for the role of Braf in melanocyte neoplasia comes from zebrafish in which Braf mutations induce nevus formation and cooperate with Trp53 deficiency to induce melanoma (12).

Mice carrying specific ablation of either Ink4a (13, 14) or Arf (15) rarely develop melanoma spontaneously (reviewed in ref. 16). However, melanocyte-specific activation of Hras on both Ink4 and Arf-null backgrounds leads to spontaneous melanoma development (17) but with only locally invasive, nonmetastatic tumors. As an additional model for human melanoma, Cdk4 has also been targeted in mice by homologous recombination to “knock in” the activating R24C mutation (18). These animals are prone to cancer as they age but do not develop melanomas unless exposed to potent carcinogens (19). Interesting inroads have been made into the relative roles of the pRb and p53 pathways in UVR-induced melanoma development in mice (17). UVR treatment increases melanoma penetrance in Arf^−/−/Tyr-Hras but not in Ink4a^−/−/Tyr-Hras animals. Thus, in mice at least, germ-line mutations resulting in pRb pathway dysregulation may eliminate the melanoma-inducing effects of UVR. Spontaneous melanomas in Arf^−/−/Tyr-Hras animals have a high degree of chromosomal instability whereas UVR-induced lesions are cytogenetically intact except for frequent genomic amplification targeting Cdk6 (17). To further assess the role of the pRb pathway in murine melanoma development, we have assessed spontaneous and UVR-induced melanoma in Cdk4-R24C mutant mice that also carry melanocyte-specific activated Hras.

All experiments were undertaken with the approval of the Queensland Institute of Medical Research Animal Ethics Committee (approval no. A98004M).

Mouse strains. Cdk4^R24C/R24C mice have been previously described (18). Tyr-Hras (G12V) transgenic mice (TPras) express mutant (G12V) Hras in a melanocyte-specific manner via a tyrosinase gene promoter/enhancer cassette (2). All mice were on a mixed SV129/C3H background.

UVR treatments. At days 2 to 4, animals were irradiated with a single UVR dose of 8.15 kJ/m² as previously described (3).

Cell culture and morphology. Tumors were digested in 1.2 units/mL Dispase (Invitrogen, Carlsbad, CA) and cultured in RPMI (Life Technologies, Inc., Carlsbad, CA) supplemented with 10% fetal bovine serum (CSL Ltd., Parkville, Australia). Cell lines were grown on coverslips and DNA and microtubules were examined by fluorescence microscopy as previously described (20).

Fluorescence-activated cell sorting analysis. Cells were fixed in 70% ethanol and stained with propidium iodide (Molecular Probes, Carlsbad, CA). The ploidy status was determined using the FACCalibur (BD Bioscience, Franklin Lakes, NJ) cell sorter and data were analyzed with Modfit Software (Verity, Topsham, ME). Ploidy status was determined by comparing the profile of diploid lymphocytes to that of the melanoma cell lines.

Immunohistochemistry. Paraffin-embedded sections of melanomas were dewaxed and antigen retrieval done using citrate buffer (pH 6.0). Endogenous peroxidase activity was quenched in 3% H₂O₂ for 10 minutes and sections were then washed and blocked with 10% goat serum. The primary antibodies were a p16 polyclonal antibody (sc-1207, Santa Cruz Biotechnology, Santa Cruz, CA) applied at 1:300 dilution, a p53 polyclonal antibody (CM5, Novocastra, Newcastle upon Tyne, United Kingdom) applied at 1:500, a tyrosinase-related protein 1 (Tyrp1, αPEP1) polyclonal antibody diluted 1:500 (a gift from Dr. V.J. Hearing, Laboratory of Cell Biology, National Cancer Institute, Bethesda, MD), a rabbit anti-caspase-3 (Biocare Medical, Concord, CA) diluted 1:500, and a biotin-labeled anti-MCM7 mouse monoclonal diluted 1:250 (NeoMarkers, Freemont, CA). Secondary detection was via an Envision plus detection kit (DAKO, Ely, United Kingdom), which was visualized using AEC plus (DAKO), and sections were counterstained with hematoxylin.

PCR, real-time PCR, and sequencing analysis. DNA and RNA were extracted using Qiagen kits (Qiagen, Hilden, Germany). Cdkn2a copy number was assessed by multiplex PCR with exon-specific primers using Gapdh as a control (Supplementary Fig. S1). Levels of Ink4a, Arf, Cdk6, and Myc expression were determined by SYBR Green (Qiagen) real-time quantitative PCR. cDNA was made using Superscript II reverse transcriptase (Invitrogen) and subsequent PCR reactions were carried out on a Rotorgene 3000 cycler (Corbett Research, Mortlake, New South Wales, Australia). Data were analyzed using Rotorgene 6 software as previously described (21). β-Actin was used as a PCR control and expression in melanomas and cell lines was compared with that in wild-type skin. Sequencing was done using BigDye Terminator 3 (ABI Prism, Foster City, CA) on a 3100 Genetic Analyzer (ABI Prism). Data were analyzed using SeqDoC software (22). Primers were designed to amplify the entire 1,709-bp Trp53 cDNA (open reading frame) in three overlapping fragments and were located in sequences that are not shared with the Trp53 pseudogene on chromosome 17. All primer sequences are available in Supplementary Table S1.

Statistical analysis. The survival of mice in each treatment group was estimated using Kaplan-Meier analysis and pairwise differences between the groups were assessed with the log-rank test. The associations between treatment and the proportion of mice that developed disease characteristics were tested using pairwise Fisher exact test or Mann-Whitney U test. The statistical significance levels for all tests were adjusted for the multiple pairwise comparisons using the Bonferroni method.

Cdk4^R24C/R24C/TPras mice develop multiple spontaneous and UVR-induced melanomas. We compared penetrance of melanoma at 1 year in Cdk4^R24C/R24C, Cdk4^R24C/R24C/TPras, and Cdk4^R24C/+/TPras mice after a single neonatal UVR dose of 8.15 kJ/m², with untreated mice of the same genotype as controls. Figure 1D shows the Kaplan-Meier curve for tumor-free mice at 1 year of age. Cdk4^R24C/R24C mice did not develop melanoma spontaneously or after neonatal UVR. As previously reported (3), TPras mice developed neonatal UVR-induced, but not spontaneous, melanomas (Fig. 1D). However, 58% of mice homozygous for the Cdk4-R24C mutation and also carrying the melanocyte-specific activated Hras (Cdk4^R24C/R24C/TPras) developed melanoma spontaneously. UVR treatments increased the penetrance of tumor development to 83% (and from 0% to 40% in Cdk4^R24C/+/TPras mice) and decreased the estimated age of onset compared with untreated animals (Fig. 1D). Lesions were mainly dermal melanomas. They were often multicentric, had aberrant nuclear features, and were usually accompanied by epidermal hyperplasia in UVR-treated animals (Fig. 2A). The increased melanoma susceptibility in mice carrying both activated Cdk4 and Hras is underlined by their increased propensity to develop multiple primary melanomas. All melanoma-bearing UVR-treated Cdk4^R24C/R24C/TPras animals developed more than one primary lesion, significantly more than untreated Cdk4^R24C/R24C/TPras mice (40%, P = 0.012, Fisher exact test, Bonferroni adjustment) or UVR-treated TPras mice (16%, P = 0.001; Fig. 1A).

Mice carrying activated Cdk4 and Hras develop aggressive metastatic melanomas. Cohorts carrying both mutant Cdk4 and Hras were also more likely than TPras animals to develop lymph node metastases (Fig. 1B). Metastases, defined as enlarged draining lymph nodes that stained for Tyrp1, were seen in 92% (12 of 13) of tumor-bearing UVR-treated Cdk4^R24C/R24C/TPras, in 60% (3 of 5) of spontaneous Cdk4^R24C/R24C/TPras, and in 16% (1 of 6) of UVR-treated TPras animals, respectively. All metastases involved only the draining lymph node with no evidence of visceral dissemination. Small numbers of scattered Tyrp1-positive cells were sometimes observed in the lymph nodes of melanoma-free TPras mice. However, mice with enlarged nodes generally had a single macrometastasis; these ranged in size from ∼400 to 1,000 cells. The Cdk4^R24C/R24C/TPras (+UVR) cohort had a higher rate of metastatic melanoma than the TPras (+UVR) cohort (P = 0.001, Fisher exact test, Bonferroni adjustment). In addition to being more prone to development of metastases, the primary lesions on the UVR-treated Cdk4^R24C/R24C/TPras were on average larger than those carried by the TPras animals (Fig. 1C) although the difference fell just short of significance (P = 0.057).

We hypothesized that tumors in the mice carrying the Cdk4 defect may be more aggressive because of a cell cycle control defect introduced by pRb pathway deregulation; hence, we examined cultured melanoma cells from various genotypes for evidence of mitotic defects (Fig. 3A). Nuclei in cultured TPras melanoma cells ranged from 7 to 18 μm in diameter whereas nuclei from Cdk4^R24C/R24C/TPras tumors were 20 to 70 μm in diameter. Additionally, the latter cultures had a higher proportion of multinuclear cells and generally larger cytoplasmic and nuclear volume than the TPras cells. We also assessed DNA content in two cell lines where low melanin content permitted fluorescence-activated cell sorting analysis (Fig. 3B). The TPras cell line was diploid whereas the Cdk4^R24C/R24C/TPras cell line was aneuploid. To confirm that this difference in nuclear size was not simply an artifact of culture, we also measured nuclear size in tumor sections. The nuclei in tumors from TPras animals were smaller than those in Cdk4^R24C/R24C/TPras melanomas (Mann-Whitney U test, P < 0.01; Fig. 3C and D). Interestingly, nuclei in the TPras primary tumor cells (range, 2.5-15 μm) were within the same size range as those in corresponding cultured TPras melanoma cells (7-18 μm) whereas nuclei in cultured Cdk4^R24C/R24C/TPras melanoma cells (20-70 μm) were considerably larger than their in vivo tumor counterparts (4-20 μm). The additional increase in nuclear size of Cdk4^R24C/R24C/TPras melanoma cells in culture presumably reflects their transition to aneuploidy, which did not occur in TPras cultures.

To further assess phenotypic variables that may explain the differences in aggressiveness between Cdk4^R24C/R24C/TPras and TPras tumors, we scored proliferation, apoptosis, and vascularity in melanoma sections (Fig. 2B). Tumors from UVR-treated Cdk4^R24C/R24C/TPras mice had significantly greater blood vessel density than TPras (+UVR) tumors (Mann-Whitney U test, P < 0.01), in keeping with the fact that TPras melanomas are small in situ lesions (Fig. 4A). Moreover, there were more proliferating cells in the UVR-induced Cdk4^R24C/R24C/TPras melanomas than in the UVR-induced TPras tumors although this fell just short of significance (Mann-Whitney U test, P = 0.078; Fig. 4B). There was no significant difference in the level of apoptosis between any of the cohorts (Fig. 4C).

Molecular changes in Cdk4^R24C/R24C/TPras melanomas. Genomic PCR and quantitative real-time PCR were used to assess the relative copy number of Ink4a and Arf genes, respectively. In addition to testing the melanomas, we also analyzed cell lines generated from them to confirm that stromal contamination did not mask loss of Ink4a or Arf. No loss of either gene or their transcripts was seen in tumors (n = 17) or cell lines (n = 12) from mice of any cohort carrying the Cdk4-R24C mutation. We also stained tumor sections for Ink4a by immunohistochemistry (Supplementary Fig. S2). Ink4a immunoreactivity was observed in 31 of 35 (89%) samples (Table 1).

We also stained for Trp53 protein in tumors and found that 36 of 56 (64%) of the melanomas were positive (Table 1). However, we rarely observed Trp53 staining in spontaneous Cdk4^R24C/R24C/TPras melanomas (2 of 8) whereas it was present in the majority of UVR-induced Cdk4^R24C/R24C/TPras tumors (25 of 35). Trp53 reactivity was generally sparse, moderately intense nuclear staining, such as observed in the UVR-treated skin used as a positive control (Fig. 2C). Thus, this Trp53 staining in our tumors probably represents up-regulation of the protein in some tumor cells in response to cellular deregulation or damage. In some Trp53-positive tumors, staining was more intense and occasionally cytoplasmic. This staining seemed to be specific to the UVR-induced tumors as it was observed in 11 of 35 (31%) Cdk4^R24C/R24C/TPras (+UVR) compared with 0 of 8 spontaneous Cdk4^R24C/R24C/TPras melanomas. We hypothesized that such high expression may have been indicative of a mutant Trp53, as has been shown in other studies of mouse melanomas (23). However, sequencing of the entire Trp53 open reading frame in 7 of 11 of these high Trp53 expressing tumors revealed no mutations; thus, the mode of Trp53 dysregulation in these lesions is unknown. It should be noted that the lack of staining for Ink4a or Trp53 does not necessarily mean loss of expression as neither protein can be detected by immunohistochemistry in unchallenged wild-type mouse skin.

We also examined expression of Myc and Cdk6, which have been previously shown to be up-regulated by gene amplification (>2-fold relative to melanocytes) in some mouse melanomas (17, 24). This was assessed by quantitative real-time PCR with results expressed relative to wild-type skin (normalized to a value of 1). Myc expression was significantly higher (P = 0.041, Mann-Whitney U test) in spontaneous Cdk4^R24C/R24C/TPras lesions (mean = 1.9, n = 3) than in UVR-induced melanomas of the same genotype (mean = 0.892, n = 8; Table 1). Neither cohort showed significant amplification of Cdk6 over wild-type levels (UVR-induced Cdk4^R24C/R24C/TPras: mean = 0.786, n = 8; spontaneous Cdk4^R24C/R24C/TPras: mean = 1.13, n = 3).

It has been previously found (17) that following UVR, mice carrying a pRb pathway defect, together with melanocyte-specific activation of Hras (Ink4a^−/−/Tyr-Hras), do not show an increase in melanoma development over spontaneous levels, indicating that pRb pathway defects may not be required for UVR-induced melanoma. Our data for Cdk4^R24C/+/TPras heterozygous mice add some support for this conclusion because Cdk4^R24C/+/TPras animals developed UVR-induced melanoma with the same penetrance as mice carrying only activated Hras (TPras). This indicates that their melanoma development is being driven by activated Ras alone. However, when in a homozygous state, the Cdk4 mutation did increase melanoma penetrance in mice following UVR. Cdk4^R24C/R24C/TPras mice developed melanomas with significantly greater incidence and earlier onset than the TPras animals (Fig. 1D), suggesting that involvement of the pRb pathway in UVR-induced melanoma may depend on a particular threshold of pRb pathway deregulation. Involvement of pRb pathway abrogation in UVR-induced melanoma development is not unexpected as some studies have indicated a role for pRb in DNA repair. In vitro studies with Ink4a and Arf-null murine embryonic fibroblasts (25) have shown that, at least in that cell type, pRb pathway abrogation results in a significant diminution of DNA repair capacity (interestingly, there was little difference between Ink4^−/− and Arf^−/− cells).

The penetrance of UVR-induced melanoma in TPras transgenics on an Arf-null or Trp53-null background is unknown. However, Arf^−/−/Tyr-Hras mice (17) developed UVR-induced melanomas with greater penetrance than our similarly treated Cdk4^R24C/R24C/TPras mice (80% compared with 40% at 150 days). We predict that Arf^−/−/TPras (or Trp53^−/−/TPras) mice may have an even earlier age of onset than Arf^−/−/Tyr-Hras because of the higher expression of Hras in our TPras transgenics compared with the Tyr-Hras transgenics of others (17). This would suggest that Arf^−/−/TPras (or Trp53^−/−/TPras) mice would develop melanomas with a greater penetrance and earlier onset than our Cdk4^R24C/R24C/TPras animals, and thus a Trp53 pathway defect would still be more effective at inducing melanoma in this model.

Another possible explanation for the difference between our study and that of Kannan et al. (17) is that the use of activated Cdk4 (at least in the homozygous state) ultimately has a different functional effect than Ink4a ablation. In Ink4a knockout mice, other members of the same family of cyclin-dependent kinase inhibitors (i.e., Ink4b, Ink4c, and Ink4d) may be able to compensate to some degree for Ink4a loss. In contrast, homozygosity for the activated Cdk4 would override such inhibitor redundancy. Evidence for such redundancy comes from studies with mice null for another Ink4 inhibitor, Ink4c (19). These animals develop significantly more carcinogen-induced melanocytic lesions than controls, indicating that, in addition to Ink4a, Ink4c also plays a role in suppressing melanocyte transformation via regulation of Cdk4. This argument for inhibitory redundancy is supported by our results showing that the wild-type copy of Cdk4 carried by our Cdk4^R24C/+/TPras cohort seems to play a role in protecting against initiation of UVR-induced melanoma. We found no evidence of loss of the wild-type allele in the Cdk4^R24C/+/TPras melanomas (data not shown). This is to be expected if, as we suggest, the tumor development is primarily being driven by the activated Ras.

Further highlighting the cooperativity between activated Cdk4 and UVR, the UVR-treated animals more often presented with multiple primaries and generally developed larger lesions than those in TPras animals although there were no obvious histopathologic differences between spontaneous and UVR-induced melanomas from Cdk4^R24C/R24C/TPras mice. The differences in aggressiveness seem to be due to a higher degree of cell proliferation in tumors carrying the mutant Cdk4 as there was no significant difference in tumor cell apoptosis between genotypes. In addition, the Cdk4^R24C/R24C/TPras melanomas had a significantly greater density of blood vessels than TPras lesions although we suspect that this is primarily due to the fact that the tumors carrying mutant Cdk4 are on average much larger than the in situ lesions which predominate in the TPras mice. In keeping with their aggressive phenotype, UVR-induced Cdk4^R24C/R24C/TPras tumors had larger nuclei than those from TPras animals and their corresponding cell lines exhibited other evidence of mitotic defects including aneuploidy, increased nuclear size heterogeneity, and the appearance of multinuclear cells.

Concomitant abrogation of the pRb and Trp53 pathways is a hallmark of mouse melanomas (17, 23). We saw no evidence of loss of Ink4a in the Cdk4^R24C/R24C/TPras tumors as may be expected because the pRb pathway is already deregulated by the constitutively active Cdk4. Gain of chromosome 15, with resultant overexpression of Myc, has previously been observed in spontaneous Trp53^−/−/Tyr-Hras melanomas (24). This was not observed in Ink4a^−/−/Tyr-Hras tumors (17), indicating that it may be an alternative method of abrogating the pRb axis, given that Myc overexpression alone can cause loss of the G₁ cell cycle transition via activation of either Cdk2 or Cdk4 complexes (26). We observed Myc up-regulation in some spontaneous melanomas, consistent with the results of Bardeesy et al. (24), indicating that high levels of Myc expression are sometimes observed in melanomas regardless of whether they are induced on a background of pRb or p53 pathway deficiency. Notably, this was not observed in UVR-treated melanomas in our study nor by Kannan et al. (17), indicating that Myc deregulation may be limited to spontaneous melanoma.

Cdk6 amplification is another avenue of pRb pathway deregulation in murine melanoma and is a marker for UVR-induced melanoma in Arf^−/−/Tyr-Hras mice, with Cdk6-amplified tumors showing >2-fold increase in expression compared with mouse melanocytes (17). None of our melanomas attained this level of overexpression. Thus, Cdk6 up-regulation does not seem to be a marker of UVR-induced melanoma in our strains and its amplification may be specific to UVR-induced melanomas generated on Arf-null or Trp53-null backgrounds.

We also looked for evidence of Trp53 pathway inactivation but found no loss of Arf at the genomic or transcriptional level in any melanomas generated on the Cdk4-R24C background. This might be expected as Arf is usually codeleted with Ink4a at the Cdkn2a locus, and in our models there is no selective pressure for such loss as the pRb pathway is already inactivated by mutant Cdk4. However, we often observed Trp53 expression in tumors that was never observed in nontreated wild-type skin and is probably due to cellular stress or Trp53 pathway deregulation. Trp53 expression was observed more often in UVR-induced than in spontaneous tumors (Table 1). High overexpression, not associated with Trp53 mutation but possibly indicative of further Trp53 dysregulation or resistance to degradation, was invariably found only in UVR-induced tumors. This is consistent with other studies (17) indicating that in mice carrying a pRb pathway defect, deregulation of the Trp53 pathway may be involved in melanoma development, further underlining the significant role for Trp53 pathway deregulation in UVR-induced melanoma.

In summary, we have shown that inactivation of the pRb pathway by mutant Cdk4 in mice also carrying melanocyte-specific Hras activation leads to the development of multiple metastatic melanomas. The exposure of these mice to a single neonatal UVR treatment significantly increases the penetrance and decreases the age of onset of tumors, indicating that in this model, abrogation of the pRb pathway affects response to UVR. Our results are suggestive of a model where pRb pathway defects increase susceptibility to UVR-induced melanoma.

Grant support: Queensland Cancer Fund.

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.