Skin pigmentation and DNA repair constitute intrinsic mechanisms evolved to prevent skin carcinogenesis.1, 2 Skin pigmentation, which inhibits DNA damage, manifests a preventive and DNA repair a corrective mechanism. A number of genes involved both in DNA repair and pigmentation, contingent to environmental and other historical selection constraints, are highly polymorphic.3–5 Many of the polymorphisms in the genes involved in these processes, as a corollary, modulate the risk and contribute to the inter-individual differences in susceptibility to the skin cancers including basal cell carcinoma of skin (BCC).6–9
The G-protein coupled Melanocortin receptor 1 (MC1R) is a pivotal component in the pathway involved in the production of melanin in melanocytes.10 Activation of MC1R by α-MSH (α-melanocyte-stimulating hormone) leads to increased cellular cAMP. The levels of cAMP, through tyrosinase, control the switch over of synthesis from pheomelanin to photo protective eumelanin.11 The gene encoding MC1R is highly polymorphic and many variants are associated with an increased risk of skin cancers.12, 13 More than 60 nonconservative variants of the receptor are known and at least 3 frequent variants are strongly associated with high risk phenotypes of red hair and fair skin (RHC alleles; R151C, R160W and D294H) in Caucasians.14–16
Several studies have reported association between variants in the MC1R gene and risk of BCC including genotype-host factor interactions.7, 17, 18 None of the investigations so far has explored the interaction between MC1R variants and risk genotypes in DNA repair genes. In this report, based on 529 BCC cases and 533 controls, we have investigated the effect of variants in the entire MC1R gene and their interactions with host factors like skin complexion and skin response to sun-exposure. In our previous study on the same study population, we found that the T241M polymorphism in the XRCC3 gene was associated with modulation of BCC risk.8 Therefore, in the present study we also analyzed the effect of the interaction between the MC1R variants and the T241M XRCC3 polymorphism on the disease risk.
Material and methods
Cases and controls were recruited as part of a large study on risk of various cancers because of environmental arsenic exposure in Hungary, Romania and Slovakia between 2002 and 2004.19 The recruitment was carried out in the counties of Bacs, Csongrad and Jasz-Nagykun-Szolnok in Hungary; Bihor and Arad in Romania and Nitra in Slovakia. Five hundred twenty-nine skin cancer cases were invited on the basis of histopathological examinations by pathologists. About 533 hospital-based controls were included in the study, subject to fulfillment of a set of criteria. All general hospitals in the study area were involved in the process of control recruitment and a rotation scheme was used in order to achieve appropriate geographical distribution. The controls in general included surgery, orthopaedic and trauma patients with conditions such as appendicitis, abdominal hernias, duodenal ulcers, cholelithiasis and fractures; patients with malignant tumors, diabetes and cardiovascular diseases were excluded. They were also broadly matched with cases for age, gender, country of residence and ethnicity.8 Skin types in both cases and controls were classified on the basis of complexion and the effect of sun-exposure; the Fitzpatrick classification was not used because of nonavailability of facilities uniformly across all recruiting centers of the participating countries.
Subsequent to signing of consent forms, clinicians took venous blood and other biological samples from cases and controls. The blood samples were kept deep frozen at −80°C until analysis. A general questionnaire was completed by trained personnel after an interview of the recruited cases and controls. The questionnaire was designed to include information on individual cumulative sun exposure in summer, sun-tanning, skin-complexion, effects of sun-exposure on skin and age/s at diagnosis of BCC. In addition, the interviews included items on demographic, life-style, socioeconomic, medical history, occupational exposures, drinking and nutritional habits, as well as detailed residential history. Ethnic background for the cases and controls was recorded along with other characteristics of the study population. Local ethical board approved the study plan and design.
Genotyping by sequencing
DNA was extracted from blood samples of cases and controls using Qiagen mini-preparation kits. A 1,107 bp fragment that included the entire coding sequence of the MC1R gene was amplified using primers MC1R-F (5′-GCA GCA CCA TGA ACT AAG CA-3′) and MC1R-R (5′-CAG GGT CAC ACA GGA ACC A-3′). PCR was carried out in a total volume of 13 μl which contained 10 ng of template DNA, 1.5 mM MgCl2, 0.15 μM of each primer, 1.3 μl 10× reaction buffer, 0.11 mM of each dNTP, 0.4 U DNA polymerase (Platinum Taq; Invitrogen, Paisley, UK) and 5% DMSO. The reaction was carried out using the temperature conditions with an initial denaturation step of 94°C for 4 min was followed by 3 cycles of denaturation at 94°C for 25 sec, annealing at 56°C for 20 sec and amplification at 72°C for 30 sec, 3 cycles with annealing at 55°C, 4 cycles with annealing at 54°C, 5 cycles with annealing at 53°C and lastly 32 cycles with an annealing temperature of 52°C. The final amplification was at 72°C for 4 min and amplified product was verified on 1% agarose gels. PCR products were purified by incubation at 37°C for 40 min with 0.75 μl ExoSapIT (USB Corporation, Cleveland, OH) followed by an enzyme inactivating step at 85°C for 15 min. Purified PCR products were subjected to direct sequencing with primers SEQ-1R (5′-GGC CAT GAG CAC CAG CAT A-3′), SEQ-2F (5′-GAC GTG ATC ACC TGC AGC TC-3′) and SEQ-3F (5′-CTG GGC ATT TTC TTC CTC TG-3′). The sequencing reactions were performed using BigDyeR Terminator Cycle Kit (Applied Biosystems, Foster City, CA) under thermal conditions as 96°C for 2 min followed by 27 cycles at 96°C for 30 s, 54°C for 10 s and 60°C for 4 min. The reaction products were precipitated with 2-propanol, washed with 75% ethanol, diluted in 25 μl water and loaded on an ABI prism 3100 Genetic analyzer (Applied Biosystems). Primary sequencing data were analyzed using a sequencing analysis program (Applied Biosystems).
The association between BCC and genotype/haplotype of MC1R polymorphisms was estimated as odds ratios (OR), 95% confidence intervals (CI) and p-values using SAS version 9.1. Logistic regression included age, sex and nationality as covariates. The possible effects of complexion, sun-exposure and sun effect were also assessed by p-values. Gene–host factor and gene–gene interactions were explored after a model selection based on likelihood ratio tests (LRT) and estimated by testing departure from risk-ratio multiplicativity or additivity. The host factors, skin complexion and skin response to sun-exposure were categorized into high (light complexion or burns/blisters, H), medium (medium complexion or mild burns, M) and low (dark complexion or tan/no change, L) risk groups. For each factor, noncarriers of the MC1R variants in the low risk group were used as references and relative risks were determined for 5 other groups. The relative risks were measured as RMRH for carriers of MC1R variants in high risk group, RMRM for carriers in medium risk group, RMRL for carriers in low risk group, RORH for noncarriers in high risk group, RORM for noncarriers in medium risk group. Multiplicative interaction index (MII) for MC1R variants and high risk host factor was calculated as (MIIhigh = RMRH/RMRL × RORH) and for MC1R variants and medium risk host factor as (MIImedium = RMRM/RMRL × RORM). Interaction contrast ratio (ICR) for genotype and host factors was calculated as (ICRhigh = RMRH − RMRL − RORH + 1 and ICRmedium = RMRM − RMRL − RORM + 1). For interaction between the genotypes, using the group that was noncarrier for the MC1R variants and carrier for XRCC3 T241M variant (RORC) as reference, 3 relative risks were estimated. RMRC was the relative risk for the carriers of MC1R and XRCC3 variants; RMRO for the carriers of MC1R and noncarriers of XRCC3 variants; and RORO for noncarriers of MC1R and XRCC3 variants. MII was calculated as [MII = RMRO/(RMRC × RORO)] to test departure from multiplicativity (> or < 1). ICR was calculated (ICR = RMRO − RMRC − RORO + 1) to test departure from additivity (> or < 0). Confidence intervals and p-values for MII and ICR were determined using bootstrap method with 10,000 simulations.
The haplotype frequencies in cases and controls, and the haplotypes carried by each individual (diplotype) were inferred with the SAS/Genetics software module. The analysis was carried out to examine the phase of MC1R polymorphisms, detected by direct DNA sequencing, using the expectation–maximization algorithm to generate maximum likelihood estimates of haplotype frequencies. Linkage disequilibrium was calculated with Haploview software (www.broad.mit.edu/mpg/haploview/documentation.php). The association of MC1R genotypes with the age of BCC onset was assessed by Kruskal-Wallis tests. To quantify the contribution of MC1R polymorphisms in the genesis of BCC in general population we calculated population attributable fraction (PAF).20, 21
In this study 529 cases with BCC and 533 matched controls were recruited from Hungary, Romania and Slovakia. The mean age of cases (237 men and 292 women) was 64.8 (±10.3) years (median 67; range 30–85) and that of controls (274 men and 259 women) was 60.0 (±11.8) years (median 60; range 28–83). While the complexion and nature of skin response to sun exposure showed association with BCC risk, the average cumulative sun-exposure was not associated with the risk (Table I).
Table I. Distribution of BCC Cases and Controls for Different Characteristics
Sun exposure estimated by taking a mean of eight categorical variables measuring average daily exposure to the sun in summer over the respondents lifetimes. For two cases and six controls exposure information was not available.
Skin response to sun-exposure
Average cumulative sun-exposure (hr per day during summer)2
The genotyping of the MC1R gene by direct sequencing in the cases and controls showed presence of 9 common nonsynonymous and 1 synonymous single nucleotide polymorphisms (Table II). The frequencies of variant alleles for the nonsynonymous polymorphisms detected in the cases and controls ranged between 0.003 and 0.12 (Table II). Minor allele frequency for the synonymous T314T polymorphism in both cases and controls was 0.13. One or more MC1R variant alleles that caused amino acid changes were present in 363 (69%) cases compared to 299 (56%) controls. The presence of one or more variants was associated with statistically significant increased risk of BCC (OR 1.66, 95% CI 1.28–2.14; Table II). The analysis of data after adjustment for arsenic exposure (measured as time weighted average concentration in μg/l over the life time of an individual) did not show any change in the risk of BCC associated with the MC1R variants. On the basis of the frequency of variants in the MC1R gene in controls (56%) and the observed association in this study indicated a population attributable fraction of 27.0%. The data analysis did not show any dependence between the onset age of BCC and the MC1R genotype. The median age of BCC onset was 67 years among both carriers and noncarriers of MC1R variants (5th and 95th percentile 45–78 years).
OR for risk associated with the presence of 2 variants in the gene was 2.69 (95% CI 1.77–4.08) compared to 1.48 (95% CI 1.13–1.94) for only 1 variant (Table II). With the exception of 2 controls no other subjects included in this study carried 3 variants simultaneously. The so-called red hair color (RHC) variant alleles that included R151C, R160W and D294H were related to higher risk of BCC (OR 2.04; 95% CI 1.45–2.83) than the non-RHC variants (OR 1.48; 95% CI 1.11–1.96). The OR for the risk associated with the RHC category that also included the D84E and I155T variants was 2.04 (95% CI 1.47–2.83) compared to non-RHC alleles (OR 1.45, 95% CI 1.08–1.93). The analysis of individual polymorphisms showed statistically significant association with the BCC risk only for R160W and R163Q polymorphisms. The presence of R160W (RHC variant) showed significant increased risk of BCC (OR 1.69; 95% CI 1.18–2.42). OR associated with the risk in the carriers of the R163Q (non-RHC) variant was 1.76 (95% CI 1.06–2.95). Combined OR for the risk due to R160W or R163Q polymorphisms was 2.21 (95% CI 1.56–3.12) and for other variants in gene was 1.44 (95% CI 1.09–1.91). With the exception of 3 cases and 2 controls the variant alleles for R160W and R163Q polymorphisms were mutually exclusive.
In addition, we detected 18 nonsynonymous and 5 synonymous rare polymorphisms in the gene distributed in 18 cases and 18 controls (Table III). The most common among rare polymorphisms included a G > A synonymous base change in codon 233, present in 2 cases and 5 controls. Other notable changes included 2 nonsense mutations, 1 each in cases and controls.
In the absence of allele-specific determination of variants, the individual haplotypes (diplotypes) could only be inferred by maximum likelihood estimates. The phase-analysis of 10 single nucleotide polymorphisms (including the synonymous T314T) in the MC1R gene revealed 7 haplotypes that accounted for over 95% cases and controls (Table IV). The most common haplotype comprised the consensus MC1R sequence and was present in 55.4% cases and 65.4% controls. All haplotypes other than 1 and 5 carried 1 variant each. Haplotype 5 (Table IV) carried 2 variant alleles, A of V92M (G > A) and G of T314T (A > G) polymorphisms, which as shown in evolutionary studies, were in linkage disequilibrium (D′ 0.96, 95% CI 0.92–0.99; LOD 146.11 and r̂2 0.64).22, 23 The 2 haplotypes associated with significantly increased risk of the BCC contained variants individually associated with risk, namely R160W (OR 1.78; 95% CI 1.11–2.86) and R163Q (OR: 2.08; 95% CI: 1.01–4.30).
Table IV. Effect of Haplotypes Due to Common MC1R Variants on the Risk of BCC
We also investigated the interaction between MC1R genotype and individual complexions on BCC risk, and the interaction between MC1R genotype and skin response to sun-exposure. The presence of MC1R variant alleles was associated with significantly increased risk of BCC for light and medium complexions (Fig. 1). The fair-complexion individual carriers of MC1R variants were at the highest risk (OR 9.68; 95% CI 3.82–24.52). The OR for the risk of BCC due to fair complexion in the absence of any MC1R variants was 5.17 (95% CI 1.98–13.49). The interaction between MC1R genotypes and both light and medium complexions was close to multiplicativity (MIIhigh = 1.24, 95% CI 0.83–1.79, p = 0.17; MIImedium = 0.87, 95% CI 0.57–1.24, p = 0.24) and greater than additivity (ICRhigh = 3.95, 95% CI 2.99–5.48, p < 0.001; ICRmedium = 0.67, 95% CI 0.03–1.28, p = 0.04).
An analogous model was used to explore interaction of the R160W and R163Q variants with complexion on BCC risk (Fig. 1). The odds ratio for the risk due to the R160W variant in fair complexion individuals was 13.39 (95% CI 4.80–37.33) and in people with medium complexion 5.63 (95% CI 2.00–15.79). When compared with fair complexion noncarriers, the OR for the risk of the disease in corresponding carriers was 2.61 (95% CI 1.47–4.65). In medium complexioned carriers increase in risk was nonsignificant (OR 1.37, 95% CI 0.76–2.48) when compared with their noncarrier counterparts. Calculation of risk for the carriers of R163Q variant over the fair-complexion background gave an OR of 10.25 (95% CI 3.32–31.67); the OR for the risk in carriers with a medium complexion was 12.57 (95% CI 3.52–44.86) compared to dark complexion individuals without any MC1R variants. Risk associated with the R163Q polymorphism in fair-complexion was only marginally significant (OR 1.95, 95% CI 0.93–4.08) compared to fair-complexion noncarriers. The medium complexion carriers were at a significantly increased risk (OR 3.17, 95% CI 1.19–8.40) compared to the corresponding noncarriers. The interaction between R160W polymorphism and light complexion was nonsignificantly greater than additive (ICRhigh = 3.72, 95% CI −0.65 to 7.15, p = 0.08) and less than additive with medium complexion [ICRmedium = −2.60, 95% CI −7.30 to (−0.11), p = 0.04]. On the other hand, the interaction between R163Q polymorphism and light complexion was multiplicative (MIIhigh = 1.24, 95% CI 0.62–3.12, p = 0.26); and a similar interaction with medium complexion was marginally greater than multiplicative (MIImedium = 1.92, 95% CI 0.94–4.90, p = 0.07).
Similar analyses of interaction between MC1R variants and individuals categorized on the basis of skin response to sun exposure showed that the carriers of MC1R variants who were affected by severe sunburns or blisters were at significantly increased risk (OR 2.38, 95% CI 1.58–3.59, Table V). Noncarriers with severe sunburns or blisters showed a nonsignificant increased risk of BCC (OR 1.52, 95% CI 0.90–2.55). The interaction between genotypes and 2 skin response categories was significantly greater than multiplicative (MIIhigh = 1.29, 95% CI 1.07–1.56, p = 0.01; MIImedium = 1.65, 95% CI 1.38–1.96, p < 0.001).
Table V. Effect of Interaction Between MC1R Variants and Skin Response to Sun-Exposure On BCC Risk
OR and 95% CI for the risk associated with combined effect of the MC1R variants and the sun effects on skin was adjusted for age, sex and nationality. Global p-value for the combined effect was <0.0001.
MII is multiplicative interaction index and ICR, interaction contrast ratio.
In a previous study on the same population we observed that carriers of variant allele for the T241M polymorphism in the XRCC3 gene were at a decreased risk of BCC (OR 0.73 95% CI 0.61–0.88); conversely noncarriers were at an increased risk (OR 1.51, 95% CI 1.17–1.95).8 In the present study, we investigated the combined effect of the T241M polymorphism in XRCC3 and the variants in the MC1R gene (Table VI). Individuals simultaneously carrying any MC1R variant and common allele for the T241M XRCC3 polymorphism showed an OR of 2.51 (95% CI 1.73–3.63) for the risk of BCC. In the carriers of variant allele for the XRCC3 polymorphism risk of MC1R variants was similar to the overall risk due to any MC1R variant. The combined genotype effect was multiplicative (MII = 1.00, 95% CI 0.86–1.16, p < 0.51) and not additive (ICR = 0.34, 95% CI 0.08–0.58, p < 0.001).
Table VI. Effect of Interaction Between the MC1R Variants and T241M XRCC3 Polymorphism on the Risk of BCC
The genotype data for the T241M (C>T) polymorphism in the XRCC3 gene was from a previous study on the same set of BCC cases and controls (Ref.8).
OR and 95% CI for the risk associated with the combination of MC1R and XRCC3 genotypes were adjusted for age, sex and nationality. Adjusted global p-value for the combined effect was <0.0001.
Sun-exposure is the major etiological factor in the genesis of BCC; however, the ultimate risk involves an interplay between genetic, host and environmental factors.9, 24 In the present study we assessed the effect of variants in the entire MC1R gene on population susceptibility to BCC, which is the most common skin cancer. Through direct DNA sequencing we detected 10 common and 23 rare variants in the gene. Our data analysis, besides confirming the presence of the MC1R variants as an independent risk factor, showed interaction of effect with other genetic and host factors. The genetic association studies are in general linked with poor reproducibility, however, our investigation was based on a prior hypothesis and results are in accord with an earlier study.7 The proportion of BCC attributable to MC1R variants as per our data was fairly high when compared with population based contribution of disease-segregating germ-line mutations in high penetrance genes like BCRA1 and 2 in breast cancer, CDKN2A in melanoma and patched in Gorlin syndrome.21, 25 The MC1R variants within the gene reportedly alter the penetrance of the germ-line CDKN2A mutations and influence the frequency of somatic B-RAF mutations in melanoma.26–29
Hypomorphic mutants encoded by the RHC variants within human MC1R are either unable to bind ligand or activate adenylyl cyclase due to intracellular retention of misfolded receptor molecules.30–32 In conformity we observed that besides a dose-dependent increase in BCC risk, the variant alleles linked to red hair color phenotypes (RHC) were associated with higher risk of BCC than other variants. At the level of individual variants, in contrast to earlier studies, in the present investigation the R151C variant was not associated with a statistically significant increased risk of BCC.7, 17 However, our data showed a significant association of the R163Q, a non-RHC variant along with the R160W, with increased risk of the disease as reported previously in a study on female nurses.7 An estimate of haplotype frequency and relative linkage of different polymorphisms within the gene showed that the risk of BCC emanated from individual variants and not the haplotypes per se. While the R160W has been shown as a major risk variant in melanoma, the risk associated with the R163Q variant seems to be specific to BCC, which implied different operational mechanisms through which MC1R variants influence risk of 2 skin cancers.7, 17 However, the common occurrence of the R163Q variants and concurrent low risk of BCC in Asian populations point to the involvement of additional host and genetic factors in Caucasian populations.14, 23
A 2-fold increased risk in the MC1R variant carriers with a fair complexion when compared with the corresponding noncarriers, indicated a nonpigmentary receptor function. The notion was further augmented by the observation of more than multiplicative interaction between the effect of the R163Q variant and medium skin complexion. The carriers of the R163Q variant with a medium complexion were at 3-fold increased risk compared the noncarriers with a similar complexion. The R163Q variant in earlier studies was associated with lower affinity for α-MSH but normal level of cAMP response than wild type MC1R.31 However, more recent studies have associated this variant with decreased surface expression, reduced affinity for α-MSH as well as cAMP signaling but to a lower extent than the RHC alleles.30, 33 Nonpigmentary functions of MC1R include regulation of cytokines and their receptors involved in immune and inflammatory responses through modulation of NF-κB.34, 35 The effect of MC1R variants on the immune regulation and anti-inflammatory actions of α-MSH could also be the cause of observed greater than multiplicative interaction between the variants and the type of skin response to the sun-exposure.34 The expression of MC1R in cells other than melanocytes is reported to be much lower with unclear functional relevance.36, 37 However, the reported association of the variants in the gene with μ-opioid analgesia in humans and mice provided evidence for its expression and function in cells other than melanocytes.38
The synthesis of melanocortins in skin is reported to enhance DNA repair, preserve melanocyte survival and stimulate melanogenesis to provide optimal photoprotection mediated via MC1R.2, 39, 40 One of the plausible explanations for the effect on a tumor of keratinocytes by MC1R expressed primarily in melanocytes could be due to reduced protection from UV induced DNA damage because of perturbed eumelanin/phaeomelanin balance. Melanin produced in melanosomes inside melanocytes is delivered to keratinocytes.2 UV radiation has been reported to trigger over 30-fold increased induction of α-MSH in keratinocytes of mice and humans without inducing a tanning response in the absence of functional MC1R.41 The loss of function variants within the MC1R gene are associated with increased UV induced DNA damage and apoptosis; reduced repair and photoprotection due to formation of free radical generating phaeomelanin that can lead to DNA strand breaks.42, 43 According to the multistage models of cancer, multiplicative interactions have been interpreted as 2 exposures (the MC1R variants and common allele of the XRCC3 polymorphism in the present study) affecting different stages.44, 45 In the context of the observed results we hypothesize the probable perturbation of apoptosis by the XRCC3 variant allele, which in turn could be triggered by UV-phaeomelanin generated DNA strand breaks. XRCC3, one of the RAD51 like gene paralogs, is involved in homologous recombination repair of DNA double strand breaks.46 Functional evaluations have predicted possible damaging consequences because of T241M XRCC3 polymorphism.47 The T241M variant allele in the gene has been associated with increased levels of bulky DNA adducts, with plausible explanation for the paradoxical protective effect based on the notion of increased apoptosis because of accumulated DNA damage.48
In conclusion, our data in this study confirmed that the presence of nonsynonymous variants within the MC1R gene constitutes an independent risk factor for BCC. The estimates from this study suggested a large proportion of BCC cases could be attributed to the polymorphisms within the gene. The mechanism through which the MC1R variants influence the risk likely involves complex interactions with other genetic and host risk factors. Such interactions probably involve skin pigmentation, DNA repair, apoptosis, immune, and other biological processes. This study reiterates the concept that while the individual genotypes could be indicative of susceptibility, the overall estimation of individual risk would be ultimately dependent on the complex interactions between known and unknown factors.