We conducted a clinical trial to compare the molecular and cellular responses of human melanocytes and keratinocytes in vivo to solar-simulated ultraviolet radiation (SSUVR) in 57 Caucasian participants grouped according to MC1R genotype. We found that, on average, the density of epidermal melanocytes 14 days after exposure to 2 minimal erythemal dose (MED) SSUVR was twofold higher than baseline (unirradiated) skin. However, the change in epidermal melanocyte counts among people carrying germline MC1R variants (97% increase) was significantly less than those with wild-type MC1R (164% increase; P = 0.01). We also found that sunscreen applied to the skin before exposure to 2 MED SSUVR completely blocked the effects of DNA damage, p53 induction, and cellular proliferation in both melanocytes and keratinocytes.
Melanoma, the most lethal form of skin cancer, arises from melanocytes. It is postulated that the proliferative response of melanocytes following sun exposure may play a role in melanoma development. Whether the magnitude of the proliferative response is the same for all people, and whether the effect is modified by topical sunscreens, is unknown. This study confirms the role of ultraviolet radiation in initiating melanocytic proliferation, implicates MC1R as a key mediator in this process, and demonstrates the effectiveness of sunscreen in preventing these molecular responses.
Melanoma is a potentially lethal cancer arising from melanocytes, the pigment-producing cells of the skin. While sunlight [presumably ultraviolet radiation (UVR)] is the principal environmental cause for this cancer, the link between sun exposure and melanoma is complex and does not accord with a simple model in which the risk of melanoma increases directly with increasing levels of exposure to the sun. The precise molecular mechanisms through which sunlight induces melanomas remain unclear, but there is evidence from epidemiologic, genetic, and molecular studies that the effects of sunlight on melanocytes are not the same for all people (Whiteman et al., 2011). It is hypothesized that the magnitude of the proliferative response of melanocytes to sun exposure may play a role in melanoma development and that this may explain, at least in part, the variation in melanoma susceptibility observed in fair-skinned populations.
Very few studies have examined the determinants of melanocyte proliferation in humans in vivo. While early studies indicated that melanocyte density in human skin increased following exposure to UVR, they also observed that the magnitude of the response varied considerably between people (Mishima, 1967; Quevedo et al., 1965). Experimental studies performed by Stierner et al. (1989) confirmed that melanocyte numbers in human skin increased following UVR exposure, and further found that people with constitutively lower melanocyte densities had proportionately greater increases in melanocyte number following UVB irradiation than those with high basal melanocyte numbers. More recent studies have shown that melanocyte density increases twofold to threefold within 3–5 weeks of repetitive UV and that pigmentation induced in skin by repeated UV irradiation protects against subsequent UV-induced DNA damage (Yamaguchi et al., 2008). Further molecular work has shown that microphthalmia-associated transcription factor (MITF) levels are initially elevated following UVR exposure and remain upregulated for 2 weeks following exposure (Coelho et al., 2009).
A number of polymorphic genes involved in pigmentation and melanocyte function have been shown to confer substantially increased risks of cutaneous melanoma in carriers. Of these, MC1R is the gene most strongly associated with melanoma risk (OR 1.4–2.4) (Palmer et al., 2000; Sturm et al., 2003). Variants of MC1R are associated with phenotypes such as red hair, fair skin, poor tanning ability, and skin sensitivity to UVR in humans; however, a recent meta-analysis by Kanetsky and colleagues has shown that MC1R variants are often carried by darkly pigmented Caucasians (Kanetsky et al., 2010), present in 15% to 33% of dark-haired and in 42% of dark-eyed individuals. Molecular studies have shown UVR induces α-MSH (alpha melanocyte stimulating hormone) and ACTH (adrenocorticotropic hormone) both of which bind to the MC1R receptor on the surface of melanocytes and activate the cyclic-AMP-dependent kinase pathway driving pigmentation and proliferation (Abdel-Malek et al., 1993; Im et al., 1998). Mouse models have shown melanocytes respond to UVR not only by producing melanin but also by proliferating, which maybe a protective response against future UVR exposure by maximizing pigment production (Walker et al., 2008). However, whether MC1R genotype exerts any effect on melanocytic proliferative responses to sunlight in vivo remains unknown. Similarly, limited data exist as to whether melanocytic proliferation is associated with host phenotypic attributes related to melanocytic function (such as nevus density or freckling propensity), nor whether the responses of melanocytes to sunlight are blocked by topical sunscreens.
Here, we present the findings of the first study in humans to assess whether molecular and proliferative responses of melanocytes at 24 h and 14 days following solar-simulated ultraviolet radiation (SSUVR) exposure are modified by MC1R genotype, host phenotype, or sunscreen.
Fifty-seven healthy volunteers were recruited, the mean age was 25 yrs, and 54% were female. Volunteers were classified according to Fitzpatrick skin type, with 9% having skin type 1, 54% skin type 2 and 37% skin type 3.
Erythema and tanning
As expected, erythema was present at test skin sites 24 h after exposure to 2 MED SSUVR, whereas unexposed sites and sites pretreated with sunscreen followed by SSUVR (sunscreen + SSUVR) were both devoid of erythema at the same time point (Figure 1A,B). Pathology analysis using eosin- and hematoxylin-stained sections revealed mild epidermal injury with acute inflammatory responses present at the SSUVR exposed site 24 h post-exposure. At 14 days, skin sites exposed to SSUVR had developed a tan and were visibly darker than the unexposed and sunscreen + SSUVR sites (Figure 1C). We observed significantly higher melanin density values at the SSUVR sites compared with both the unexposed sites and the sunscreen + SSUVR sites at 14 days (P < 0.01 paired t-test; Figure 1D).
CPD, p53 expression, and cleaved caspase-3 following SSUVR
We quantified the proportion of epidermal cells expressing cyclobutane pyrimidine dimers (CPDs, a marker of UV-specific DNA damage), p53 expression (associated with cellular stress), and cleaved caspase-3 (apoptosis) at baseline, and at 24 h and 14 days following SSUVR exposure. At 24 h, 25% of epidermal keratinocytes and 36.6% of melanocytes showed evidence of UV-induced DNA damage (CPD positive) compared with 0% at both the unexposed and sunscreen + SSUVR sites (Table 1, paired t-test P < 0.01). Less than 1% of all epidermal cells stained positively for CPD at 14 days (Table 1). P53 expression was present in 1% of keratinocytes in unexposed skin at baseline, increasing to 15.1% at 24 h following SSUVR before dropping back to 2.7% of keratinocytes at 14 days. For melanocytes, 6.7% expressed p53 at baseline, rising to 24.9% at 24 h, and returning to 6.4% at 14 days. Prior application of sunscreen to the skin completely abrogated the induction of p53 expression in both keratinocytes and melanocytes. Less than 1% of keratinocytes or melanocytes expressed cleaved caspase-3 at baseline, and there was little evidence that its expression was significantly altered following SSUVR.
Table 1. Proliferation, DNA damage, apoptosis, and P53 induction in skin 24-h and 14 days following solar-simulated ultraviolet radiation (SSUVR) exposure
Sunscreen + SSUVR
P-value – paired t-test of comparison to baseline.
% Cleaved caspase-3
% Cleaved caspase-3
Melanocyte proliferation following SSUVR
At 24 h, we observed slightly but statistically significantly lower proportions of epidermal melanocytes expressing ki-67 compared with unexposed skin at baseline (Table 1); melanocyte counts in the SSUVR site at 24 h were also slightly lower than unexposed skin at baseline (paired t-test P < 0.05; Table 1). At 14 days, however, there were significantly higher proportions of epidermal melanocytes expressing ki-67 in the SSUVR sites compared with the unexposed and sunscreen + SSUVR sites (Table 1). Epidermal melanocyte counts in the SSUVR sites at 14 days (Table 1) were about twofold higher than melanocyte counts in unexposed sites at baseline (paired t-test P < 0.05) or in the sunscreen + SSUVR sites (paired t-test P < 0.05). On average, 7% (95% CI: 6.44, 7.57) of epidermal keratinocytes expressed ki-67 in unexposed skin compared with 8.6% (95% CI: 8.00, 9.44) in skin 14 days post-exposure to SSUVR. Nests of proliferating epidermal keratinocytes were observed surrounded by hyperplasic melanocytes scattered along the basal layer 14 days after SSUVR exposure suggesting interactions between the two cell types (Figure 2). The proportion of epidermal keratinocytes expressing ki-67 was linearly correlated with the change in the number of melanocytes between baseline and 14 days after SSUVR (ρ = 0.38, P < 0.01) and with the proportion of melanocytes expressing ki-67 (ρ = 0.51, P < 0.001).
Determinants of proliferative response following SSUVR exposure
We analyzed the associations between phenotypic, clinical, and genotypic characteristics of participants and melanocyte proliferation, CPD DNA damage, p53, or cleaved caspase-3 expression levels. There was no consistent association between host phenotypes and the proportion of epidermal melanocytes expressing ki-67, CPD, p53, or cleaved caspase-3. Melanocyte proliferation was defined as the difference between the melanocyte count at baseline (Non-UVR) and melanocyte count at 14 days following exposure to SSUVR. For all phenotypic subgroups, there was an approximate doubling of melanocyte numbers at 14 days, and we found no consistent evidence that the magnitude of the melanocytic proliferative response differed according to host nevus count, freckling density, eye color, or observed tanning response (Table 2). There was a modest trend for people with auburn hair color and also skin type I to have lower melanocyte counts than people with skin type III (Table 2). The melanocytic proliferative response at 14 days following SSUVR exposure was significantly lower among people carrying variant alleles of MC1R compared with those wild-type for MC1R (Table 3). The level of UVR-induced DNA damage determined by CPD staining and induction of p53 was similar across all groups regardless of their MC1R genotype (Table 4). The MC1R genotypes represented in this cohort are listed in Tables S1 and S2, and the association between MC1R genotype and participant characteristics illustrated that all people with skin type I carried MC1R variants, and there was a trend for people with increasing freckling density to carry MC1R rising from 68% to 90%, although this trend was not statistically significant (Table S3).
Table 2. Mean melanocyte counts at 14 days grouped by phenotypic characteristics
Baseline non-UVR (SD)
SSUVR versus non-UVR mean change (SD)
Average% change SSUVR versus non-UVR
P-value – independent samples t-test or one-way anova testing whether the change in baseline to solar-simulated ultraviolet radiation (SSUVR) is equal across phenotype groups.
Blue/Green/Gray, n = 38
Brown, n = 19
Black/dark brown, n = 30
Blonde, n = 9
Auburn, n = 18
I, n = 5
II, n = 31
III, n = 21
None, n = 28
Few, n = 19
Some, n = 10
Total body nevi
<50, n = 16
50–100, n = 18
100+, n = 23
Observed skin color changes 14 days post-SSUVR
Light Tan, n = 20
Red Tan, n = 30
Dark Tan, n = 7
Table 3. Change in the mean number of epidermal melanocytes 14 days following solar-simulated ultraviolet radiation (SSUVR) exposure, by MC1R status
P-value for the paired t-test between SSUVR and baseline (non-UVR) sites.
P-value t-test of difference in CPDs or P53 at SSUVR and baseline (non-UVR) sites among MC1R variant versus MC1R wild-type participants.
% Epidermal expression CPDs 24 h
% Melanocytes expression CPDs 24 h
% Epidermal expression P53 24 h
% Melanocytes expression P53 24 h
% Epidermal expression P53 14 days
% Melanocytes expression P53 14 days
% Epidermal expression ki-67 24 h
% Melanocytes expression ki-67 24 h
% Epidermal expression ki-67 14 days
% Melanocytes expression P53 14 days
These experiments in human volunteers have yielded three principal findings. First, they have confirmed that melanocytes increase in number following exposure to SSUVR, with an approximate doubling after 14 days. Second, they suggest that the melanocytic proliferative response differs according to MC1R genotype. Third, they have shown that sunscreen completely blocks the molecular and proliferative effects of SSUVR on keratinocytes and melanocytes in vivo.
We found that people with intact MC1R had larger increases in melanocyte numbers following SSUVR than people carrying MC1R variants. One explanation for our observation is that functional MC1R modulates the proliferative response of melanocytes following SSUVR, suggesting that the α-MSH/ACTH/MC1R/cyclic-AMP pathway cascade may be involved in melanocyte proliferation and that impaired function of MC1R leads to decreased proliferative responses of melanocytes following SSUVR. The keratinocyte proliferative response did not vary across people of different MC1R genotypes suggesting this defect in cutaneous signaling due to MC1R variant alleles seems to be restricted to cell autonomous functions of melanocytes. Because melanocyte counts at baseline were similar between people carrying variant and wild-type MC1R alleles, it seems likely that the pathways governing the maintenance of constitutive melanocyte density operate independently from those that induce proliferation following SSUVR. The extent of UVR-induced DNA damage (as determined by CPDs) was similar across all groups regardless of their MC1R genotype or skin type, implying that all participants received biologically equivalent doses of SSUVR at the cellular level.
Our finding that melanocyte numbers increased to a lesser extent in people carrying MC1R variants is novel and intriguing. However, a recent study investigating in vivo melanocyte responses following UVR exposure reported no differences for melanocyte proliferation among Black, White and Asian participants (Tadokoro et al., 2005), which is potentially at odds with our finding. However, in that study, data from only three patients in each racial grouping were presented, and melanocyte counts were made only 7 days after UVR exposure. The authors noted that ‘active proliferation of melanocytes in the skin is just beginning at 1 week’, suggesting that their chosen time point maybe too early to determine whether differences between people of diverse backgrounds were evident.
We observed changes to the microenvironment of the skin following SSUVR, including changes to the organization of the epidermis with ki-67 staining keratinocytes 7% at baseline non-UVR site and decreasing to 6.3% 24 h post-SSUVR and increasing to 8.6% 14 days post-SSUVR, and these changes were linearly correlated with the change in the number of melanocytes and with the proportion of melanocytes expressing ki-67. Lin and Fisher (2007) suggest that ‘keratinocytes are the primary UV responding population’ and melanocyte regulation is largely a consequence of stimulation from the abundant epidermal cell type keratinocytes. Targeted therapeutic approaches are being designed such as potent melanocortin analogs that mimic the effects of keratinocytic cytokines like α-MSH as a strategy to prevent melanoma, particularly in individuals who express MC1R genotypes that reduce but do not abolish MC1R function (Abdel-Malek et al., 2008) even in individuals of darker pigmentation (Kanetsky et al., 2010). However, individuals carrying MC1R variants with darkly pigmented skin type (Fitzpatrick III) also exhibited a reduction in the number of epidermal melanocytes 14 days post-SSUVR, suggesting the role of MC1R maybe operating independently of pigmentation. Recent work by Chou et al. (2013) may shed light on this phenomenon. They showed that the response of melanocyte stem cells to UVR differs markedly between MC1R-defective mice and MC1R wild-type mice. Specifically, following UVB exposure, fewer melanocyte stem cells exited the hair follicle and underwent proliferation and migration to the epidermis in MC1R mutant mice. Although MC1R signaling is not involved in melanocyte stem cell maintenance, this study supports the notion that it regulates the melanocyte response to sun exposure.
We observed a decrease in ki-67 staining in melanocytes as well as the number of melanocytes 24-h post-SSUVR suggesting melanocytes perhaps overcome UVR-induced DNA damage growth arrest before cytokines released by keratinocytes drive the production of melanin pigments and hence tanning as well as melanocyte proliferation, which is present 14 days post-SSUVR. Previous studies have shown that some MC1R variants sensitize melanocytes to the cytotoxic effect of UVR and increase the burden of DNA damage and oxidative stress (Abdel-Malek et al., 2008; Kadekaro et al., 2010).We did not observe differences in the level of CPD DNA damaged or p53 expression in melanocytes between participants carrying MC1R variants 24-h post-SSUVR. The sensitivity of a person's skin to sunlight varies; as a result, we measured the minimal erythemal dose (MED) of each participant by visually inspecting a series of increasing SSUVR doses 24-h post-exposure. The 2-MED experimental dose was tailored to each participant, and on average, 65 mJ/cm2 dose was administered to participants carrying MC1R variants and 72 mJ/cm2 to participants wild-type for MC1R. The SSUVR dose administered induced the same initial levels of CPDs (24.6 and 26.1% in epidermal cells, respectively) suggesting biological equivalent bursts of cytokines released within the skin.
The proliferative response of melanocytes to UVR is of topical interest because of its postulated role in the development of melanoma. (Hacker et al., 2010; Whiteman et al., 2003). The ‘divergent pathways’ hypothesis posits that melanocyte proliferative capacity is heterogeneous in the population and that this heterogeneity manifests phenotypically as low or high nevus burden. Our data provide only limited support for this premise. We found no systematic association between the proliferative response of melanocytes to sunlight and the number of nevi on the trunk, head and neck or total body. Instead, we found evidence that the proliferative response of melanocytes to SSUVR was strongly associated with MC1R genotype. Thus, one interpretation of our data is that the pathways through which nevi are acquired are independent of those that drive proliferative response of melanocytes to SSUVR and instead dependent perhaps on melanocytes incurring some type of damage that destabilizes them and gives rise to nevi. The precise mechanism behind nevus development thus remains unclear, but epidemiologic and genetic studies indicate that both host characteristics and subsequent patterns and doses of UVR exposure are important interacting factors.
We found no evidence that any of the UVR-induced skin damage parameters measured, and subsequent responses, were not abrogated by the use of sunscreen suggesting that in accordance with the Green et al. (2011) clinical trial, the proper application of sunscreen is very likely to significantly decrease UVR-induced melanoma risk. Sunscreen applied at 2 mg/cm2 to human skin before exposure to SSUVR completely abrogated the molecular effects of SSUVR on melanocytes in vivo. While it might be assumed that such an effect could be predicted with certainty, it must be acknowledged that there remains considerable uncertainty as to the wavelengths and precise mechanisms through which sunlight causes melanoma in humans, quite apart from the debate as to whether sunscreens have a role in melanoma prevention (Diffey, 2009; Lund and Timmins, 2007). In marked contrast to squamous cells cancers of the skin, for which there is strong evidence of causality through UV-B signature mutations in key genes, no such certainty regarding UVR-associated mechanisms exists for melanoma.
Strengths of our study include the strict sampling criteria and the detailed epidemiologic data (including clinical counts of nevi blind to genotype status) accompanying the specimens. In particular, we intentionally sampled healthy people in a narrow age range (18–33 yrs) of similar ancestry and Fitzpatrick skin types 1–3. The spectral output used in the SSUVR dose closely resembled environmental solar levels with a high reproducibility. The sample size was larger than most previous studies investigating the effects of SSUVR in human skin in vivo, permitting more powerful hypothesis testing than previously. Despite the size of the study, we had sufficient statistical power to detect only large effects with precision, and so it remains possible that true but weak effects have not been detected. A further limitation is that we did not perform full sequencing of the entire MC1R gene. However, the MC1R variants genotyped in this present study comprise over 95% of the non-synonymous changes observed in the Kanetsky et al. (2006) study, which analyzed a far more ethnically diverse sample than our study (USA, Italy and Australia). Therefore, we do not believe that omitting very rare MC1R variants would markedly affect our results.
In conclusion, our work provides insight into the proliferative response of melanocytes following sun exposure. Epidermal melanocytes were found to double in number compared with unirradiated skin at 14 days after SSUVR, and sunscreen was completely effective at blocking cellular proliferation, p53 induction and generation of CPD DNA damage in both keratinocytes and melanocytes following SSUVR exposure. There was no evidence that the magnitude of the melanocytic proliferative response differed according to host phenotypic attributes such as nevus counts. Importantly, our findings identified a role for MC1R in melanocyte regulation following UVR exposure. Further explorations of the interactions between sunlight, genotype, and phenotype in melanoma development are the aim of our continuing research.
Eligible participants were healthy residents of greater Brisbane, Qld, Australia (latitude 27°S), aged >18 yrs with fair skin (Fitzpatrick Classification I–III), with no history of skin disease or photosensitivity disorder. Ineligibility criteria included as follows: bleeding diathesis; anticoagulant therapy; known or suspected hypersensitivity to local anesthetic agents; skin infection (including acne) at the site of biopsies; currently taking medications known or suspected to reduce inflammation, immunity, or healing (e.g., corticosteroids); any skin disease or use of medications known to induce photosensitivity; recent use of sunbeds; current outdoor employment; excessive recreational exposures; any chronic disease requiring ongoing medical attention; and personal history of skin cancer. Phenotypic data and relevant exposure data (including sun exposure history and skin sensitivity) were collected from study volunteers through a structured questionnaire as described previously (Whiteman et al., 2003).
The study was approved by the Human Research Ethics Committee of the Queensland Institute of Medical Research and the Queensland University of Technology. The study complied with the Declaration of Helsinki, and all participants gave their written, informed consent to take part.
We defined three skin sites on the lower back of each participant: (i) unexposed baseline site (non-UVR site), (ii) a site exposed to 2 MED solar-simulated ultraviolet radiation (SSUVR site), and (iii) a site pretreated with sunscreen and then exposed to 2 MED SSUVR (sunscreen + SSUVR site). All selected sites were representative skin of the back; nevi, freckles, other melanocytic lesions, and areas of inflammation or infection were specifically avoided. Skin biopsies were collected at 24-h and 14 days post-SSUVR. Sunscreen (Hamilton broad spectrum, SPF30+) was applied to skin site 3 by a nurse 20 min before solar exposure; it was rubbed evenly into the skin using a non-absorbable glove at a dose of 2 mg/cm2. Ultrastructural characterization of the sunscreen used in this study was performed using transmission electron micrographs and X-ray diffraction. The formulation contained titanium dioxide and zinc oxide with tubular/spindle particles up to 100 nm in length and larger rectangle shapes up to 200 nm.
Solar exposure and skin color measurements
A solar simulator Model 601 fitted with 300-watt zenon arc lamp with UV filters was used to administer the SSUVR dose (Solar Light Co, Philadelphia, PA, USA). This model is routinely used in the sunscreen industry for testing sun protection factor (SPF) and is calibrated to the US National Institute of Standards and Technology (NIST) UV standard and spectral output detailed in Figure S1. The minimal erythemal dose (MED) of each participant was determined in accordance with the Australian Standard 2604:1998. Skin sites were visually inspected 24-h later, and the minimal erythemal dose was determined. The treatment regime dose of 2 MED ranging from 44 to 130 mJ/cm2 was applied via a fiber optic cable to the sun exposed sites. Erythema was assessed by visually comparing to color charts and spectrophotometer measurements of skin reflectance at all sites (Konica Minolta, Brisbane, Qld, Australia). Images were also recorded using a whole-body imager, and a white sticker was used as a white balance indicator on all images (Canfield Imaging Systems, Fairfield, NJ, USA). Skin color observations were conducted on images 14 days post-SSUVR using RGB color spectrum cutoffs for grouping light tan, red tan, dark tan, or no change. Melanin density was estimated from spectrophotometer values using the formula described by Dwyer et al. (2002).
Sample collection, processing, and genotyping
Skin biopsies were collected from each treatment site using a protocol adapted from Chamberlain et al. (2005). Briefly, the biopsy was taken under local anesthesia from suitably prepared skin using a 2-mm biopsy punch. The tissue was immediately immersed in 10% buffered formalin solution for 24 h and processed for histology. Saliva samples were also collected, and DNA was extracted using Oragene saliva kits (DNA genotek, Ottawa, ON, Canada) following manufacturer's instructions. Genotyping was performed using the MassArray platform (Sequenom Inc, San Diego, CA, USA). An optimized multiplex assay of all nine common variants of MC1R (I155T, R142H, D84E, R160W, D294H, V92M, R163Q, V60L, and R151C) was used as previously described (Duffy et al., 2004). Participants with none of the MC1R variants listed above were classified as wild-type (WT) for these analyses. People carrying 1 or more of the ‘R’ alleles (R142H, D84E, R160W, D294H, R151C) were classified as ‘R’ variants, and people carrying 1 or more ‘r’ alleles (I155T, V92M, R163Q, V60L) were classified as ‘r’ variants (Table S1). People carrying both R and r alleles were classified as ‘R’ variants; the frequencies of MC1R heterozygotes are listed in Table S2.
Tissue samples for immunohistochemistry analysis were sectioned 4 μm thick and underwent antigen heat retrieval using EDTA 105°C for 10 min. Proliferating cells were detected using anti-Ki-67 (Dako, Glostrup, Denmark) diluted 1/200, DNA-damaged cells were detected using anti-CPD antibody (Kamiya Biomedical company, Japan) diluted 1/200, apoptotic cells were detected using anti-Cleaved caspase-3 antibody (Cell Signalling, Beverly, MA) diluted 1/500, cells expressing p53 were detected using anti-p53 antibody (Dako) diluted 1/50, and melanocytes were identified using anti-Tyrosinase (Dako) diluted 1/50. Dual staining of mouse monoclonal antibodies Ki-67 or CPD or p53 was conducted with O/N incubation of the tyrosinase antibody and detected using the biotinylated donkey anti-mouse (Jackson Laboratory, Maine, ME, USA) secondary antibody with streptavidin linked Alexa-Fluor 555 (Invitrogen, Carlsbad, CA, USA). Sections were then blocked in mouse IgG protein (Jackson Laboratory) followed by donkey anti-mouse-FAB fragment (Jackson Laboratory). The ki-67 or CPD or p53 antibodies were incubated 1 h at RT and were detected using anti-mouse Alexa-Fluor 488 (Invitrogen). The rabbit monoclonal cleaved caspase-3 antibody was incubated 1 h at RT and was detected using anti-rabbit Alexa-Fluor 488 (Invitrogen). Negative controls were processed in parallel using an identical protocol, but with the omission of the primary antibody. All slides were cover-slipped using Vectashield-DAPI mounting media (Vector laboratories, Burlingame, CA, USA). Four images per sample were captured using the DeltaVision Microscope system (Applied Precision, Issaquah, WA, USA) and each image contained on average a 300 μm length of epidermis. The deltaVision tracking software and stage technology allowed non-overlapping images to be collected along each skin section. Sections were counted using Volocity image software (PerkinElmer, Waltham, MA, USA), and cells positively stained were counted only if a visible DAPI staining nucleus was observed. All staining and quantification procedures were performed blind to the sample's ID. Morphology analysis was conducted on adjacent sections, which were stained for eosin and hematoxylin using the Varistain Gemini (Thermo Electron Corp, Waltham, MA, USA).
Paired t-tests were used to detect differences between treatments sites. To assess the effect of different categorical variables on the differences between treatment groups, anova and appropriate post hoc comparisons, along with the Bonferroni multiple hypothesis testing (MHT) correction, were used. Pearson's correlation was used to assess linear associations between continuous variables, and Pearson's chi-squared and/or Fisher's exact test (when more than 20% of cells had expected count less than 5) was used to assess associations between appropriate categorical variables. All analyses were performed using spss Statistics v.19 (IBM, Armonk, NY, USA).
The authors would like to thank all participants for their time and generosity. This work was funded by the Cancer Council Queensland and Atlantic Philanthropies. Elke Hacker and Nicholas Hayward are supported by fellowships from the National Health and Medical Research Council (NHMRC) of Australia. David Whiteman is supported by a Future Fellowship from the Australian Research Council, and Michael Kimlin is supported by a Cancer Council Queensland fellowship. The authors would like to thank Brett Chapman and Grant Montgomery for their assistance with the Sequenom genotyping platform and H. Konrad Muller for pathology analysis as well as Lambert Bekessy and Tony Raftery for their help with structural characterization of the sunscreen formulation and Graeme Walker for helpful discussions.