To further investigate the use of DNA repair-enhancing agents for skin cancer prevention, we treated Cdk4R24C/R24C/NrasQ61K mice topically with the T4 endonuclease V DNA repair enzyme (known as Dimericine) immediately prior to neonatal ultraviolet radiation (UVR) exposure, which has a powerful effect in exacerbating melanoma development in the mouse model. Dimericine has been shown to reduce the incidence of basal-cell and squamous cell carcinoma. Unexpectedly, we saw no difference in penetrance or age of onset of melanoma after neonatal UVR between Dimericine-treated and control animals, although the drug reduced DNA damage and cellular proliferation in the skin. Interestingly, epidermal melanocytes removed cyclobutane pyrimidine dimers (CPDs) more efficiently than surrounding keratinocytes. Our study indicates that neonatal UVR-initiated melanomas may be driven by mechanisms other than solely that of a large CPD load and/or their inefficient repair. This is further suggestive of different mechanisms by which UVR may enhance the transformation of keratinocytes and melanocytes.
The results of Dimericine treatment in basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) have led to much discussion about the potential use of DNA repair-enhancing agents in skin cancer prevention. The development of ‘super-sunscreens’ that in addition to shielding the skin from ultraviolet radiation (UVR) also contain active compounds that function to enhance DNA repair of damaged ‘sunburnt’ cells, is underway. We have used the Cdk4R24C/R24C/NrasQ61K mouse model to test how enhanced repair of DNA following UVR will impact on the development of melanoma and found that it had no effect. Our findings argue that DNA repair enhancement may not be helpful, at least for some types of melanoma, and are suggestive of potentially different mechanisms of UVR carcinogenesis between melanocytes and keratinocytes.
Exposure to solar ultraviolet radiation (UVR) is the main environmental risk factor for melanoma development. The precise mechanism by which sunlight induces melanoma is still unclear, however sun exposure can cause the generation of various types of cellular and DNA damage, immune suppression and inflammation (Abdel-Malek et al., 2008). The importance of effective DNA repair is highlighted in patients with the rare genetic disease xeroderma pigmentosum (XP), in which a defect in DNA repair is associated with the development of melanoma and other skin cancers much earlier in life and with increased risk compared to the general population (Kraemer et al., 1994). Tanaka et al. (1975) first showed that treating XP cells in culture with the bacterial DNA excision repair enzyme T4 endonuclease V reduced DNA damage induced by UVR. Further clinical studies have demonstrated that the repair enzyme encapsulated in a liposomal delivery vehicle can be transported to intracellular compartments of human and mouse skin (Wolf et al., 2000; Yarosh et al., 1992). The hydrophobicity afforded by the liposomal complex allows for the penetration of the enzyme through the stratum corneum, delivering T4 endonuclease V to epidermal keratinocytes (Yarosh et al., 1994). The T4 endonuclease V protein lacks a nuclear localization sequence (NLS), entering the nucleus through passive diffusion after initially accumulating in the cytoplasm of cells (Yarosh et al., 1994).
Once inside the nucleus, T4 endonuclease V, a T4 pyrimidine dimer glycosylase, acts on cyclobutane pyrimidine dimers (CPDs). The enzyme scans through DNA, specifically binding to any CPDs. Upon recognition of a CPD, it bends the duplex DNA and rotates the nucleotide base complementary to the 5′ residue of the CPD, resulting in a cleavage event and base excision of the dimer (Morikawa and Shirakawa, 2000). It is likely that an exonuclease degrades the area surrounding the single-strand break, and a polymerase fills in the gap, however the precise mechanism downstream of T4 endonuclease V is unclear.
Topical application of the T4 endonuclease V liposome lotion (known as Dimericine) accelerated the removal of UVR-induced DNA damage in XP patients and reduced the rate of actinic keratoses and basal-cell carcinoma (BCC) (Yarosh and Klein, 1996; Yarosh et al., 2001). Dimericine has also been shown to reduce the incidence of SCC in UVR exposed SKH-1 hairless mice (Yarosh et al., 1992). We established a mouse model of melanoma using transgenic animals carrying a melanocyte-specific Nras activating mutation and the cell cycle-deregulating Cdk4 R24C mutation (Hacker et al., 2006). Exposing these mice to a single neonatal dose of UVR greatly increased melanoma penetrance in this model. The aim of this study was to determine whether treatment with Dimericine would prevent, or reduce, UVR-induced melanoma development in this model.
Dimericine treatment reduced the level of DNA damage following UVR exposure
Topical Dimericine treatment 30 min prior to UVR exposure reduced the level of DNA damage present in mouse epidermal keratinocytes after a single neonatal UVB dose of 5.9 kJ/m2. Evidence of significant erythema (peeling) was observed 72 h after UVR in placebo-treated, but not in Dimericine-treated animals (Figure S1). We observed only a relatively small amount of 6–4 photoproduct damage in the epidermis at 3 h post-UVR, with an average of only 7.5 and 3.4% of cells staining in placebo-treated (n = 3) and Dimericine-treated (n = 3) animals respectively (Mann Whitney P = 0.014) (Figures 1A and 2A). We observed no evidence of 6–4 photoproducts in non-UVR treated animals (Figure 1A). The majority of 6–4 photoproducts were removed by 24 h in both placebo and Dimericine treated animals (data not shown). We rarely observed any melanocytes harbouring 6–4 photoproducts and found no significant difference between Dimericine +UVR treated animals and placebo treated animals.
Dimericine treatment significantly reduced the percentage of epidermal cells harbouring CPDs post-UVR, the greatest difference between Dimericine and placebo treated animals was observed at the 3 h post-UVR timepoint (Mann Whitney P = 0.004) (Figure 1B). No CPDs were observed in the epidermis of non-UVR treated animals (data not shown). We analysed 128 epidermal melanocytes from three placebo +UVR treated animals and found that at 3 h after UVR 92% carried CPDs while in the Dimericine +UVR cohort (n = 3) we analysed 87 epidermal melanocytes and found that only 59% carried CPDs (Pearson Chi-squared P < 0.001)(Figure 2B). Dimericine treatment also reduced the number of extra-follicular dermal melanocytes carrying CPDs at 3 h post-UVR, with 36% of 330 melanocyte harbouring CPD damaged compared with 75% of 260 extra-follicular melanocytes in placebo treated animals (Pearson Chi-squared P < 0.001). These results suggest that Dimericine penetrates the skin and reduces DNA damage in melanocytes in the epidermis and dermis. Interestingly, even without Dimericine CPDs were undetectable in both epidermal and extra-follicular melanocytes by 24 h post-UVR. In contrast, epidermal keratinocytes (both basal and suprabasal) still exhibited CPD damage at 24 h post-UVR (Figure 1B). Dimericine sped up the initial removal of CPDs from these cells (at the 3 h timepoint), but by 24 h post-UVR there was no significant difference in the level of CPDs between placebo and Dimericine treated animals.
Proliferation of melanocytes inhibited by Dimericine treatment
At 72 h post-UVR there was an increase in proliferating epidermal keratinocytes as measured by Ki67 positivity. Animals treated with placebo lotion (n = 3) had on average 16% of epidermal cells (both basal and suprabasal keratinocytes) proliferating, while Dimericine-treated animals (n = 4) had on average 9% and non-UVR-treated animals (n = 3) only 4% of epidermal cells proliferating (Figures 3A and 2C). These differences are statistically significant (anova, P < 0.001), and post hoc analyses (Tukey’s test) revealed that all three means were statistically significantly different from each other. There was a higher percentage of epidermal melanocytes proliferating at 72 h post-UVR in placebo-treated animals (55%, 133/241) compared with Dimericine-treated (36%, 45/126) and non-UVR treated animals (24%, 8/25). Dimericine treatment significantly reduced the proliferation of epidermal melanocytes (Pearson Chi-squared P < 0.01). We also analysed Ki67 positivity of extra-follicular dermal melanocytes at 72 h post-UVR and found no difference between animals treated with placebo +UVR (26%, 63/247) and Dimericine +UVR (23%, 61/263) (Pearson Chi-squared P = 0.6). There were a higher percentage of hair follicle outer root sheath (ORS) melanocytes proliferating at 72 h post-UVR in placebo-treated animals (68%, 141/206) compared with Dimericine-treated (43%, 54/125) and non-UVR treated animals (31%, 35/112) (Pearson Chi-squared P < 0.001).
Migration of melanocytes to the epidermis following UVR exposure
The numbers of epidermal melanocytes increased following UVR exposure and peaked at 72 h post-UVR (Figure 3B). Dimericine treatment prior to UVR resulted in a slight decrease in epidermal melanocyte number at 72 h post-UVR (Figures 2D and 2E). The average number of epidermal melanocytes per field reduced from 14.4 in placebo treated animals (n = 3) to 9.3 in animals treated with Dimericine (n = 4) (anova contrast t-test P < 0.01) (Figure 3B). In comparison the non-UVR cohort did not exhibit an increase in the number of epidermal melanocytes over the period of the experiment (Figure 3B).
Dermal melanocytosis is part of the phenotype of animals carrying the melanocyte-specific Nras mutation (Ackermann et al., 2005). There was a much greater number of dermal extrafollicular melanocytes in Cdk4R24C/R24C/NrasQ61K mice compared to homozygous Cdk4R24C/R24C animals regardless of treatment regimen (Placebo +UVR, Dimericine +UVR or non-UVR) (Figure 3C). These differences were statistically significant between Cdk4R24C/R24C/NrasQ61K and homozygous Cdk4R24C/R24C animals (anova P < 0.01) (Figure 3C).
Dimericine did not prevent melanoma development in Cdk4R24C/R24C/NrasQ61K mice
Thirty-three percent of mice homozygous for the Cdk4-R24C mutation, and also carrying the melanocyte-specific activated Nras mutation (Cdk4R24C/R24C/NrasQ61K), developed melanoma spontaneously by 250 days (Table 1). A single neonatal UVR treatment increased the tumour penetrance and significantly decreased the age of onset (Log rank test Bonferroni correction, P > 0.001) (Table 1). We compared the penetrance of melanoma in Cdk4R24C/R24C/NrasQ61K mice, both with and without Dimericine treatment after a single neonatal UVR dose. We observed no statistically significant difference in the onset of melanoma development between the placebo +UVR and the Dimericine +UVR treated cohorts (Log rank test Bonferroni correction, Table 1).
Table 1. Melanoma penetrance in Cdk4R24C/R24C/NrasQ61K mice
Melanoma penetrance at 250 days (%)
Mean age-of-onset (days)
Log rank test P-value**
Log rank test P-value**
Cdk4R24C/R24C/NrasQ61K animals on a FVB background were used in this study.
*Statistically significant comparisons.
**Corrected for multiply pair wise comparisons Bonferroni method P < 0.004 for significance.
All melanoma cases were confirmed by pathologist (HKM).
We observed no phenotypic or histopathological differences between lesions that developed after Dimericine and placebo treatment (Figure 4). Lesions were mainly dermal melanomas (Figures 4B and C). They were often multicentric, had aberrant nuclear features, and were positive for Tyrp-1 (Figures 4C and D). We did observe a number of melanocytes scattered along the basal layer near the tumour site and superficial dermal nests in adult Cdk4R24C/R24C/NrasQ61K animals regardless of UVR treatment (Figures 4D and E). We also assessed indicators of tumour aggressiveness. We saw no significant difference in the average number of lesions per animal between the non-UVR, +UVR, placebo +UVR and Dimericine +UVR treated cohorts (Table 2, Pearson’s Chi-squared P = 0.09). Similarly, there was no difference in the average size of lesions among animals treated with either +UVR, placebo +UVR and Dimericine +UVR (Figure 4F) (anova P = 0.37).
Table 2. Number of melanoma cases per animal
1 melanoma per animal (%)
>1 melanoma per animal (%)
Total number of animals
To further investigate the use of nucleotide excision repair (NER)-enhancing agents for skin cancer prevention after UVR exposure we treated mice topically with Dimericine immediately prior to neonatal UVR exposure, which has a powerful effect in exacerbating melanoma development in mouse melanoma models. Unexpectedly, we saw no difference in penetrance or age of onset of melanoma after neonatal UVR between Dimericine-treated and control animals. This is not the case for non-melanoma skin cancer induction in mice – which is significantly reduced after Dimericine treatments (Yarosh et al., 1992). Notably, chronic UVR is necessary for UVR-induced non-melanoma skin cancer induction in mice. This is normally ineffective in inducing melanoma, even in transgenic animals that are highly melanoma-prone after neonatal UVR (Larue and Beermann, 2007). Our results are further suggestive of different mechanisms of UVR tumorigenesis between the respective cell types (melanocytes versus keratinocytes) and possibly a different role for CPDs in melanoma induction after neonatal UVR than in SCC induction after repeated UVR exposures in mice. It is not known how well neonatal UVR in mice corresponds to settings whereby UVR may induce melanoma in humans, but it is suspected to be somewhat analogous to childhood sunburns.
One caveat of this study is that we do not know the dose-responsiveness for melanoma induction after neonatal UVR in the model we have used (Cdk4R24C/R24C/NrasQ61K mice). For instance, melanoma may be induced by relatively low exposures that result in significantly lower levels of CPDs than the dose we delivered. However the work of De Fabo et al. (2004), the only study so far to assess the role of different total UVB doses in neonatal UVR-induced melanoma, suggests that the relationship between UVB dose and melanoma induction is close to linear in the albino Hgf model. We suspect that non-linearity of response does not explain the lack of reduction in melanoma incidence in our study due to enhancement of DNA repair. Notably, in both this and our previous study, we do not see any evidence that the neonatal melanocytes in Cdk4R24C/R24C/NrasQ61K mice do not efficiently remove CPDs. In fact epidermal melanocytes remove CPDs more efficiently than surrounding keratinocytes. This would fit with the notion that up-regulation by UVR of keratinocyte-derived cytokines (e.g. αMSH, EDN1, KITLG) results not only in stimulation of melanocyte melanin production, differentiation, and proliferation, but also in enhancement of their DNA repair capacity (Abdel-Malek et al., 2008). Interestingly, we also noted that melanocytes were often present in the epidermal basal layer adjacent to tumours. It is unclear whether this effect is driven by changes in the keratinocytes adjacent the tumour, melanoma cells, or both.
It was also interesting to note that treatment with Dimericine, which does not have any direct action on 6–4 photoproducts, did however reduce the amount of 6–4 damage. It could be speculated that Dimericine shifts repair of CPDs into base excision repair from NER by making a glycosylic bond break and releasing the 5′ pyrimidine base. As NER complexes have to repair both CPDs and 6–4 photoproducts. When the NER is freed from the task of repairing CPDs, there is more available to focus on the 6–4 photoproducts, and the expected result would be somewhat faster repair. Additionally, this study has only examined global genomic repair in the skin following treatment with Dimericine and there may be effects related to transcription coupled repair that may exacerbate tumorigenesis.
Further complicating the interpretation of our results, other studies show that a defective NER response does enhance melanoma development in animals defective in both Ink4a and Arf. Ink4a/Arf−/−/Xpa−/− (Van Schanke et al., 2006) and Ink4a/Arf−/−/Xpc−/− (Yang et al., 2007) mice develop melanoma with higher rates after neonatal UVR than for Ink4a/Arf−/− alone. The reason for the apparently contrasting results is not obvious. It could involve the germline mutations carried by the mice themselves. Our animals carried a germline cell cycle defect (Cdk4-R24C) and a melanocyte-specific Nras mutation at the time of UVR exposure – whereas in the other studies the animals did not have a prior Ras pathway mutation. Interestingly, in the Yang et al. (2007) study Kras mutations were induced during tumorigenesis in many of the melanomas developing in the Ink4a/Arf−/−/Xpc−/− mice. Thus it is possible that one action of CPD photoproducts in melanomagenesis is to induce mutations in the genes that were already defective in the mice in our study (Cdk4 and Ras). Alternatively, defective NER signalling may exacerbate tumorigenesis in other ways in addition to simply resulting in defective removal of CPDs. These other mechanisms would not manifest in our study. Furthermore, as dimericine and placebo treated animals have the same initial levels of CPDs immediately post-UVR, the initial levels of CPDs and resultant burst of cytokine release within the skin, may be more important for melanomagenesis, than the length of time dimers are present.
Another interpretation of our findings is that melanomagenesis after neonatal UVR is not simply due to the presence of CPDs generated by UVR exposure. Wolnicka-Glubisz et al. (2007) have previously shown that neonatal mice show a defective inflammatory response to UVR suggesting that damaged cells may not be properly recognized and targeted by the immune system and removed when necessary. In addition, other types of DNA (or other) damage such as oxidative lesions have been suggested to play a role in melanomagenesis after neonatal UVR. Notably, treatment of Hgf transgenic mice with antioxidants after neonatal UVR delayed the age of onset of melanoma (Cotter et al., 2007), which enhancement of NER did not do in our study. Another possible contributing factor to melanoma genesis after neonatal UVR is the proliferative and migratory response of neonatal melanocytes (Walker et al., 2008). This may contribute to their destabilisation and increase their risk of transformation. Our results reveal the importance of this response in melanocyte transformation as it was reduced by enhancement of NER using Dimericine, but still elevated compared to non-UVR treated animals. In connection with this melanocyte response, Cdk4R24C/R24C/NrasQ61K mice have dermal melanocytosis, which is not present in wild type mice. While epidermal melanocytes greatly increase in number after UVR exposure, interfollicular dermal melanocyte numbers in the melanoma-prone mice did not change. All of the proliferative response is centred on the hair follicle, with melanocytes migrating upwards into the basal layer from the ORS after UVR exposure (Walker et al., 2008). This opens up the question as to from which population are the melanomas emanating, is it the proliferating epidermal/ORS melanocytes or the non-responding dermal population? In our model and most other mouse models melanomas appear to originate from the dermis. This would suggest that the proliferating melanocyte population may not be one from which the melanomas are initiated.
In summary, treatment with Dimericine prior to UVR exposure reduced CPDs in mouse skin but did not protect against melanoma development. This finding indicates that UVR-induced melanomas are perhaps driven by mechanisms other than solely that due to persistence of CPD-associated damage.
All experiments were undertaken with the approval of the QIMR animal ethics committee (approval number A98004M).
Cdk4R24C/R24C mice have been previously described (Rane et al., 1999). Tyr-Nras (Q61K) transgenic mice express mutant (Q16K) Nras in a melanocyte-specific manner via a tyrosinase gene promoter/enhancer cassette previously described (Ackermann et al., 2005). All mice were on a FVB background. Diagnosis of melanoma was confirmed by a pathologist (HKM) and Tyrp-1 staining preformed to confirm melanocytic origin of the tumour cells, as previously described (Hacker et al., 2006).
UVR and dimericine treatments
At day 2 postnatal, animals were irradiated with a single UVR dose of 5.9 kJ/m2 as described previously (Walker et al., 2008). Animal received either the Dimericine liposome lotion or placebo lotion 30 min prior to irradiation as previously described (Yarosh et al., 2001). Briefly, the Dimericine liposome lotion consisted of 1 mg/l T4 endonuclease V, encapsulated in liposomes, in a 1% hydrogel lotion and the placebo lotion consisted of equivalent empty liposomes, without enzyme, in a 1% hydrogel lotion. Animals used in the proliferation and DNA damage assays, were monitored and photos taken of each animal prior to euthanasia. The same region of dorsal skin was taken from each animal (Figure S2).
Proliferation and DNA damage immunofluorescent assays
Paraffin-embedded sections of mice skins were dewaxed, and antigen retrieval performed using Dako retrieval solution, pH 6.0 (Dako, Glostrup, Denmark). Proliferating cells were detected using rat anti-Ki67 (Dako, M7249) diluted 1/100 and melanocytes were identified using tyrosinase-related protein 1 (Tyrp1, αPEP1) rabbit polyclonal antibody diluted 1:400 (a gift from Dr V.J. Hearing, NIH, Bethesda). Biotinylated goat anti-rat secondary (Jackson ImmunoResearch laboratories, West Grove, PA, USA) was used for Ki-67 detection with streptavidin linked Alexa-Fluor 488 (Invitrogen, Carlsbad, CA, USA) and Tyrp1 was detected using donkey anti-rabbit Alexa-Fluor 555 (Invitrogen).
Damaged cells were detected using mouse anti-thymine dimer (KTM53) (Kamiya biomedical company, Seattle, WA, USA) diluted 1/200 and melanocytes were identified using Tyrp1 antibody diluted 1:400. A mouse-on-mouse block (Biocare medical, Concord, MA, USA) was conducted for 30 min prior to O/N incubation with primary antibodies. Biotinylated donkey anti-mouse secondary (Jackson) were used for thymine dimmer (CPD) detection with streptavidin linked Alexa-Fluor 488 (Invitrogen) and Tyrp1 detected using donkey anti-rabbit Alexa-Fluor 555 (Invitrogen). Cells with 6–4 photoproducts were detected using mouse anti-6–4 photoproducts (64M-2) (Medical and biological laboratories, Nagoya, Japan) diluted 1/100 and melanocytes were identified using Tyrp1 antibody diluted 1:400. A mouse-on-mouse block (Biocare medical) was conducted for 30 min prior to O/N incubation with primary antibodies. The same secondary detection system as the thymine dimer assay was used.
All slides were cover-slipped using vectorshield-DAPI mounting media (Vector laboratories, Burlingame, CA, USA). Images were captured using the DeltaVision Microscope system (Applied Precision, Issaquah, WA, USA). The deltaVision tracking software and stage technology allowed non-overlapping images to be collected along each skin section. Ten fields were counted using an ×20 objective along a skin section from each animal for all assays but the ki67 staining which used five fields per animal. The sampled fields were 300 μm in length spaced one field apart and we did not select areas for specific cell types or staining patterns. Sections were counted using Image J (available at: http://rsbweb.nih.gov/ij/index.html). All staining and quantification procedures were performed blind to the samples ID.
The Prism software package (GraphPad Software, La Jolla, CA, USA) was used to generate graphs and the spss version 15 software package (SPSS, Chicago, IL, USA) was used to perform statistical analyses. Statistical significance was P < 0.05 and denoted by * in subsequent graphs. Mann Whitney tests were performed to compare the average percentage of epidermal cells carrying DNA damaged amongst placebo and Dimericine treated groups. Pearson’s Chi-squared tests were performed to compare the percentage of melanocyte populations harbouring CPD damage and proliferating amongst placebo and Dimericine treated cohorts. One-way anova were used to test for evidence of difference in: mean proliferative response 72 h post-UVR; mean tumour size across cohorts; across the non-UVR, +UVR, Placebo +UVR and Dimericine +UVR cohorts. Statistically significant results were investigated further using Tukey’s post hoc test. The incidence of melanoma development was modelled using Kaplan-Meier analysis, and the log-rank test was used to test for pairwise differences between the groups. The statistical significance levels for all tests were adjusted for the multiple pairwise comparisons using the Bonferroni method P < 0.004. A three-way anova was used to compare the increase in epidermal or extrafollicular melanocytes following UVR. The hypothesized difference between treatments was expressed as a contrast of means and tested with the associated t-test. Pearson’s Chi-squared test was used to compare the number of melanoma cases per animal between cohorts.
We thank David Yarosh and AGI Dermatics for the T4 endonuclease V DNA repair enzyme (Dimericine) lotion. This work was funded by the Cancer Council of Queensland. Elke Hacker and Nicholas Hayward are supported by fellowships from the National Health and Medical Research Council (NHMRC) of Australia and Graeme Walker is supported by a fellowship from the Cancer Council Queensland. We are grateful to Dr Vince Hearing for the Tyrp1 antibody, and Dr Friedrich Beermann for the Nras mice.