SEARCH

SEARCH BY CITATION

Keywords:

  • Pigmentation;
  • HERC2;
  • OCA2;
  • MC1R;
  • Gene–gene interactions;
  • Phenotype prediction

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Human pigmentation is a polygenic trait which may be shaped by different kinds of gene–gene interactions. Recent studies have revealed that interactive effects between HERC2 and OCA2 may be responsible for blue eye colour determination in humans. Here we performed a population association study, examining important polymorphisms within the HERC2 and OCA2 genes. Furthermore, pooling these results with genotyping data for MC1R, ASIP and SLC45A2 obtained for the same population sample we also analysed potential genetic interactions affecting variation in eye, hair and skin colour. Our results confirmed the association of HERC2 rs12913832 with eye colour and showed that this SNP is also significantly associated with skin and hair colouration. It is also concluded that OCA2 rs1800407 is independently associated with eye colour. Finally, using various approaches we were able to show that there is an interaction between MC1R and HERC2 in determination of skin and hair colour in the studied population sample.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

There are significant intrapopulation differences in human pigmentation among individuals of European ancestry. This diversity is remarkable for iris colour as well as hair and skin colouration. Variation of pigmentation in humans is mainly due to differences in amount, type and distribution of melanin, a high molecular weight polymer produced in melanocytes in a complex biochemical process called melanogenesis (Frudakis, 2008). It is understood that this multistep process is under genetic control with multiple genes involved. Inheritance of some serious pigment disorders like albinism can be caused by mutations in single genes and may follow simple Mendelian rules. Four types of albinisms have been linked to different human genes, with OCA2 (MIM 611409) being involved in the most common form of oculocutaneous albinism in Europe (Oetting et al., 2005). The red hair colour phenotype has also been found to be predominantly controlled by the single MC1R gene. Some MC1R (MIM 155555) polymorphisms significantly affect receptor performance and this leads to overproduction of pheomelanin – the red/yellow form of pigment. Carriers of such variants reveal red hair colour and fair skin (Rees, 2003). Beside red hair colour, there has also been substantial progress in understanding of the genetics of eye colour in humans. Until quite recently, OCA2 was considered the major eye colour gene. Three SNPs in intron 1 of the OCA2 gene were found to be the best predictors of blue/brown eye colour in humans and one non-synonymous SNP in exon 13 has been associated with green eye colour (e.g. Rebbeck et al., 2002; Duffy et al., 2007; Branicki et al., 2008a). However, most recent studies have shown that HERC2 (MIM 605837) located upstream of OCA2 has the predominant role in iris colour determination (Kayser et al., 2008; Sturm et al., 2008), indicating that rs916977 and rs12913832 located in intron 12 and 86, respectively are better predictors than the three intron 1 OCA2 SNPs.

Gene–gene interactions have for a long time been postulated to make an important contribution to the determination of human complex traits (Moore, 2003; Carlborg & Haley, 2004). Epistasis may be simply defined as a masking effect of one gene on another gene. Thus the full potential of a gene on phenotypic features cannot be observed due to the action of other gene variants. In statistical terms, epistasis can be described as a deviation from an additive model explaining a connection between various gene variants and a particular complex trait (Cordell, 2002; Moore, 2003). Although a variety of methods are now available to test for gene–gene interactions, it is still difficult to indicate the most suitable and effective tool. Logistic or linear regression have for a long time been commonly used to test for epistatic effects. However, it is well known that sparseness of data and multidimensionality may lead to increased type I errors when regression methods are applied. Ritchie et al. (2001) have shown that their multifactor dimensionality reduction (MDR) method, which is nonparametric and does not assume any genetic model of inheritance, can be a more effective tool for detection of gene–gene interactions. The method enables reduction of multilocus genotype data into one dimension and further evaluation of such a model using cross-validation and permutation testing. The MDR method already has an important place in the field of research on epistasis (e.g. Musani et al., 2007; Vermeulen et al., 2007).

Needless to say, epistasis must also be considered when studying the genetics of human pigmentation. Indeed, some epistatic effects between OCA2 and MC1R have been found to influence skin pigmentation in a Tibetan population (Akey et al., 2001). Other authors have also suggested epistatic effects affecting the human pigmentation phenotype (Pastorino et al., 2004; Sulem et al., 2007). Recent studies on eye colour inheritance have provided new evidence for various interactions between HERC2 and OCA2. The HERC2 gene has been reported to influence OCA2 expression (Eiberg et al., 2008; Kayser et al., 2008; Sturm et al., 2008) and some particular SNPs have also been found to interplay in determination of eye colour (Sturm et al., 2008).

We further investigated these issues and using multifactor dimensionality reduction and regression methods performed a study of epistatic effects between several most important polymorphisms within two known major eye colour genes –HERC2 and OCA2– as well as three other genes involved in pigmentation processes –MC1R, ASIP and SLC45A2, which may influence eye, skin and hair colour.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Samples

A total number of 388 specimens collected in years 2005–2007 from unrelated Europeans living in southern Poland was included in this study. The study was approved by the Ethics Committee of the Jagiellonian University, number KBET/17/B/2005, and participants gave informed consent. Samples were collected from inpatients and outpatients (with minor or undiagnosed conditions) attending dermatological consultations at the Department of Dermatology of the Jagiellonian University Hospital. The subjects were interviewed and assessed by a single dermatologist for eye, hair colour and skin phototype. Thus, phenotyping was done by a combination of self-assessment and professional observation. Eye colour was defined using the following scale: blue/grey, green, hazel, brown/black; hair colour was defined according to the scale: red, blond-red, blond, dark-blond, auburn and black. Skin phototype was assessed according to the Fitzpatrick scale: i.e. skin type I: always burns, never tans; type II: always burns, then tans; type III: always tans, burns sometimes; type IV: always tans, never burns (Fitzpatrick, 1988).

SNP Genotyping

Three SNPs located in intron 1 of the OCA2 gene: rs7495174, rs4778241 (formerly rs6497268), rs4778138 (formerly rs11855019) and two SNPs located in the nearby HERC2 gene: rs916977, rs12913832 were analysed in two multiplex assays. Multiplex 1 included three OCA2 SNPs and multiplex 2 included the remaining two HERC2 polymorphisms. The PCR reaction for both multiplexes was composed of 5 μl of Qiagen multiplex PCR kit (Qiagen, Hilden, Germany), 1 μl of primer premix (details in supplementary Table 1), 2 μl of Q solution and 2 μl (approximately 5 ng) of template DNA. Thirty two PCR cycles were applied under conditions recommended by the manufacturer with annealing temperature set at 58°C for multiplex 1 and 61°C for multiplex 2. The PCR products were purified with a mixture of ExoI and SAP enzymes (Fermentas, Vilnius, Lithuania) and subjected to minisequencing reactions with a SNaPshot multiplex kit (Applied Biosystems, Foster City, CA). Two microlitres of SNaPshot kit were combined with 1 μl of extension primers premix, 1 μl of the purified PCR product and nuclease-free water up to 10 μl. Details concerning extension primers are given in supplementary Table 2. The extension primer for rs12913832 produced an artefact in the blue channel (possibly due to self amplification). This, however, did not influence correct interpretation of the true peaks displayed in the yellow and red channels (polymorphism C/T), as we confirmed by sequencing reactions. The products of extension reactions were purified with SAP enzyme and analysed on ABI 3100 Avant genetic analyser (Applied Biosystems, Foster City, CA).

Statistical Calculations

Arlequin computer software ver 3.1 was used to test for deviations from Hardy-Weinberg equilibrium (exact test) as well as for evaluating linkage disequilibrium between pairs of analysed SNPs (expectation maximization algorithm) (Excoffier et al., 2005). SPSS computer software was used for most of the statistical calculations. Odds ratios (OR) with 95% CIs and respective P values associated with particular genotypes were calculated using logistic regression analysis and additive (where the genotypes were categorized according to the number of a particular allele carried, i.e. 0, 1, 2) as well as dominant and recessive models of allele classification (where the genotypes were categorized according to the presence or absence of a particular allele under assumption of its dominance or recessiveness, i.e. 1, 1, 0 or 1, 0, 0, respectively). Multivariate logistic regression was used to simultaneously test for the effect of five analysed SNP positions (independent variables) on dichotomous dependent variable (eye colour classified as blue vs. non-blue, brown vs. non-brown, hair colour classified as black vs. non-black, and skin colour classified as light vs. dark). PHASE ver. 2.1 was used for haplotype reconstruction (Stephens et al., 2001). The inferred haplotypes were then counted in the analysed population sample assuming a total number of 2N haplotypes. Nine major haplotypes were examined to ascertain their potential association with eye colour. A simple χ2 test was used to compare the frequencies of haplotypes between blue eyed and non-blue eyed individuals. Then logistic regression was used to calculate OR and P values by assuming an additive model unless complete separation was observed. For haplotype 15, additional tests were performed assuming dominant and recessive models.

Gene–gene interactions between five genes, namely HERC2 (genotypic data from this study), OCA2 (this study and Branicki et al., 2008a), MC1R, ASIP (extended data from Brudnik et al., 2008), and SLC45A2 (Branicki et al., 2008b) were tested in this population sample using both the multifactor dimensionality reduction (MDR) method (MDR software ver. 1.2) (Hahn et al., 2003) and logistic regression analysis. Bonferroni correction was applied for results obtained with logistic regression in order to reduce the influence of type I error (Vermeulen et al., 2007). Separate models were built for brown and blue eye colour. Hair colour was classified as black and non-black. Skin type was always classified as light (I and II according to the Fitzpatrick scale) and dark (III and IV). MDR is a nonparametric method which is a known alternative to regression methods. MDR relies on data reduction and has been applied to detection of multi-locus genotype combinations that predict disease risk. By pooling genotypes into two groups, the initial multidimensional model is reduced and in consequence we obtain one variable with two multifactor classes defined as high risk and low risk. Initial division of the data into a training set and a testing set enables the cross-validation procedure to be run (CV). CV consistency is one of the crucial measures and means the number of cross-validation intervals that a particular model was chosen by MDR, averaged across 10 runs (10-fold cross-validation is repeated 10 times). As the subjects are randomly divided into 10 equal parts, the maximum number of intervals is 10. The second important measure, weighing the classification accuracy using the MDR model, is balanced accuracy (BA). BA is calculated as (sensitivity + specificity)/2, where sensitivity measures how likely a positive classification is to be correct and specificity measures how likely a negative classification is to be correct. BA is especially useful when the dataset is not balanced, i.e. two classes contain different numbers of subjects (this was the case here, as, for example, only 45 individuals had brown eyes vs. 343 non-brown eyes). The models with the highest BA and CV consistency were considered the best for the purposes of this research (Hahn et al., 2003; Moore, 2004). The genotypes for HERC2 (rs916977, rs12913832), OCA2 (rs7495174, rs4778241, rs4778138, rs1800401 and rs1800407), ASIP (rs6058017) and SLC45A2 (rs26722, rs16891982) were categorized in an additive manner. In the case of MC1R genotypes, the same approach was applied assuming existence of major function mutations R (N29insA, D84E, R142H, R151C, Y152OCH, R160W and D294H) which significantly affect receptor performance. Thus, although we have sequence data for the complete MC1R exon, only three states were considered for this gene, i.e. 0 = no R variant carriers, 1 = one R variant carrier and 2 = two R variant carriers. Owing to the fact that our results indicated some dominance relationships in the analysed loci we also checked a variant with the five SNPs coded by assuming a recessive model of inheritance.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Population Analyses

The studied population sample consisted of 151 males and 237 females. Details concerning frequencies of eye, hair and skin pigmentation types among the studied individuals are given in Table 1. There were no differences in frequencies of eye and skin colouration between males and females. Differences in hair colour were found to be slightly significant (p = 0.023) and this was due to an imbalance in the group of red haired individuals, who were overrepresented in this population sample. We examined five single nucleotide polymorphisms located in two neighbouring genes, OCA2 and HERC2. All the analysed SNPs were found to be in Hardy-Weinberg equilibrium. The studied polymorphisms were dispersed at a distance below 200 kb. The LD testing confirmed that they are in strong linkage disequilibrium (p < 0.000001). Details of the analysed polymorphisms and their allele frequencies are given in Table 2.

Table 1.  Characteristics of the study population
Characteristicsn (%)
Female 237 (61.1)Male 151 (38.9)
Eye colour
 Blue/grey124 (52.3)92 (60.9)
 Green29 (12.2)17 (11.3)
 Hazel53 (22.4)28 (18.5)
 Brown/black31 (13.1)14 (9.3)
Hair colour
 Red/blond-red41 (17.3)43 (28.5)
 Blond38 (16.0)25 (16.6)
 Dark blond124 (52.3)58 (38.4)
 Auburn9 (3.8)3 (2.0)
 Black25 (10.5)22 (14.6)
Skin type
 Light skin (I or II)101 (42.6)71 (47.0)
 Dark skin (III or IV)136 (57.4)80 (53.0)
Table 2.  Data for the analysed polymorphisms
SNPGeneSequence positionPolymorphism and MAFHWE P valueFirst associated with eye colour
  1. a– formerly rs11855019; b– formerly rs6497268; MAF – minor allele frequency; HWE – Hardy-Weinberg equilibrium.

rs4778138aOCA226009415CT0.18640Duffy et al., 2007
0.1624 
rs4778241bOCA226012308TG1.00000Duffy et al., 2007
0.1843 
rs7495174OCA226017833CT1.00000Duffy et al., 2007
0.0528 
rs12913832HERC226039213TC0.88406Sturm et al., 2008
0.2281 
rs916977HERC226186959TC0.84149Kayser et al., 2008
0.1482 

Association Between OCA2, HERC2 and Pigmentation Characteristics

Table 3 shows odds ratios and associated P values obtained for the five analysed SNP positions and eye, hair and skin colour for additive and recessive models. It can be seen that with one exception (rs7495174 and skin colour), higher values of odds ratios were obtained under assumption of a recessive model. The lowest OR values were noted for a dominant model (data not shown). All the SNPs were found to be significantly associated with eye colour in separate tests, rs12913832 in HERC2 being the most significantly associated (additive model: OR = 21.17, CI = 12.37 – 36.24, p < 0.0000; recessive: OR = 28.21, CI = 16.17 – 49.22, p < 0.0000) and rs4778138 in OCA2 being less significant (additive: OR = 2.01, CI = 1.37 – 2.95, p = 0003; recessive: OR = 2.56, CI = 1.63 – 4.01, p = 0.0000). rs12913832 was also significantly associated with hair colour (additive: OR = 3.16, CI = 1.94 − 5.14, p = 0.0000; recessive: OR = 7.78, CI = 3.64 – 16.62, p = 0.0000) and skin type (additive: OR = 2.13, CI = 1.47 – 3.08; recessive: OR = 2.68, CI = 1.75 – 4.11, p = 0.0000). The significance level was also achieved for rs916977 and hair and skin colour as well as rs7495174 and skin colouration.

Table 3.  Associations between analysed SNPs and various phenotypic features for additive and recessive models
Characteristicrs7495174rs4778241rs4778138rs916977rs12913832
OR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)P
  1. OR – odds ratio; CI – confidence interval. Upper values are for additive model and lower values are for recessive model. Values significant after Bonferroni adjustment for multiple testing marked with bold.

Blue eye6.95 (3.00–16.09)0.00002.96 (1.98−4.41)0.00002.01 (1.37−2.95)0.00039.09 (5.29–15.61)0.000021.17 (12.37–36.24)<0.0000
7.09 (3.05–16.47)0.00003.91 (2.50–6.11)0.00002.56 (1.63–4.01)0.000010.6 (6.08–18.54)0.000028.21 (16.17–49.22)<0.0000
Black hair2.17 (0.99–4.75)0.05281.63 (0.98–2.71)0.06701.24 (0.73−2.11)0.43952.08 (1.23−3.52)0.00843.16 (1.94−5.14)0.0000
2.37 (1.05–5.35)0.03811.90 (1.02–3.52)0.04161.47 (0.77–2.78)0.24052.98 (1.60–5.55)0.00067.78 (3.64–16.62)0.0000
Light skin2.28 (1.12–4.64)0.01721.21 (0.84–1.75)0.31281.33 (0.91–1.95)0.13461.59 (1.05−2.41)0.02472.13 (1.47−3.08)0.0000
2.27 (1.10–4.69)0.02681.30 (0.85–2.00)0.22341.41 (0.90–2.20)0.13451.81 (1.14–2.88)0.01232.68 (1.75–4.11)0.0000

Multivariate logistic regression applied to all the five SNPs included as predictor variables showed that the only SNP that remained significant for eye colour was rs12913832 (additive: OR = 35.97, CI = 14.17 – 91.34, p = 0.00001; recessive: OR = 47.02, CI = 17.29 – 127.87, p = 0.0000). After excluding rs12913832, two SNPs remained significant in the additive model, namely rs916977 (OR = 7.2604, CI = 3.9527 – 13.3360, p = 0.00001) and rs7495174 (OR = 3.0536, CI = 1.1168 – 8.3489, p = 0.0296). In the recessive model only rs916977 was significant (OR = 7.82, CI = 4.18 – 14.61, p = 0.0000). Exclusion of both HERC2 SNPs showed that two OCA2 SNPs – rs7495174 (OR = 4.42, CI = 1.76 – 11.08, p = 0.0015) and rs4778241 (OR = 2.57, CI = 1.52 – 4.32, p = 0.0004) gave the best explanation for eye colour variation present in this population sample in the additive model. The same situation was found when the recessive model was tested: rs7495174 (OR = 3.72, CI = 1.46 – 9.47, p = 0.0059) and rs4778241 (OR = 3.11, CI = 1.76 – 5.51, p = 0.0001). We also checked that, in the regression model with five SNPs as variables, rs12913832 was still significantly associated with hair colour (additive: OR = 4.61, CI = 1.39 – 15.27, p = 0.0124; recessive: OR = 10.21, CI = 4.10 – 25.41, p = 0.0000), and skin type (additive: OR = 3.07, CI = 1.63–5.75, p = 0.0005; recessive: OR = 3.92, CI = 1.93 – 7.98, p = 0.0002). Detailed information on frequency distribution of rs12913832 genotypes in groups of differently pigmented individuals is presented in Table 4.

Table 4. HERC2 rs12913832 variant and pigmentation
Characteristic rs12913832
C/CC/TT/T
Hair colourRed67 (79.8%)14 (16.7%)3 (3.6%)
Blond49 (77.8)13 (20.6%)1 (1.6%)
Dark blond100 (54.9%)69 (37.9%)13 (7.1%)
Auburn5 (41.7%)7 (58.3%)0 (0%)
Black9 (19.1%)36 (76.6%)2 (4.3%)
Eye colourBlue or grey192 (88.9%)22 (10.2%)2 (0.9%)
Green31 (67.4%)14 (30.4%)1 (2.2%)
Hazel7 (8.6%)65 (80.2%)9 (11.1%)
Brown or black0 (0%)38 (84.4%)7 (15.6%)
Skin colourFair skin (I/II)124 (72.1%)41 (23.8%)7 (4.1%)
Dark skin (III/IV)106 (49.1%)98 (45.4%)12 (5.6%)

PHASE inferred fifteen haplotypes with one occurring most frequently, haplotype 15, eight haplotypes with average frequency and six rare haplotypes. Table 5 shows all the inferred haplotypes, haplotype counts in groups of blue eyed and non-blue eyed individuals, P values from χ2 testing as well as odds ratios and P values associated with haplotypes as assessed by logistic regression. The simple counting procedure indicated that two haplotypes i.e. the most frequent, haplotype 15, and less common, haplotype 8, are significantly associated with blue eye colour. Six other haplotypes were found to be associated with non-blue eye colouration. These results were confirmed by logistic regression analysis unless complete separation prevented the test. Dominance/recessiveness testing performed for haplotype 15 indicated that, as expected from single SNP testing results, the highest OR value was obtained for the recessive model (OR = 11.1168, CI = 6.84 – 18.06, p = 0.00001). OR obtained under assumption of an additive model was found to be lower (OR = 6.04, CI = 4.04 – 9.04, p = 0.00001). The obtained results indicated, however, that the predictive value in terms of iris colour does not increase when haplotypes are analysed instead of rs12913832 SNP position. Notably, the T allele of rs12913832 was present in all six haplotypes significantly associated with non-blue eye colour, while variant C was present in both haplotypes associated with blue eye colour. Non-association of haplotype 6 (rs12913832 in C state) can be explained in the light of allele C recessiveness as was found by examination of genotypes, including haplotype 6.

Table 5.  Inferred haplotypes and their association with eye colour
Haplotype no.State of SNPsNon-blue eyesBlue eyesTotalAssociation testing
χ2 P-value, traitLogistic regression OR (95% CI), P-value
  1. SNP order as in Table 2, i.e. rs4778138, rs4778241, rs7495174, rs12913832, rs916977; cs – complete separation; na – no association; nt – not tested.

1CTCTT245290.0001, non-blue0.15 (0.05 – 0.39), 0.0001
2CTCTC9090.001, non-bluecs
3CTCCC011ntnt
4CTTTT516ntnt
5CTTTC112ntnt
6CTTCC1529440.159, na1.56 (0.83 – 2.93), 0.1705
7CGTTT213240.0001, non-blue0.10 (0.03 – 0.35), 0.0003
8CGTCC011110.003, bluecs
9TTTTT348420.0001, non-blue0.16 (0.07–0.35), 0.00001
10TTTTC314ntnt
11TTTCC246ntnt
12TGCTT112ntnt
13TGTTT102120.006, non-blue0.18 (0.04 – 0.82), 0.0266
14TGTTC434470.0001, non-blue0.06 (0.02 – 0.16), 0.00001
15TGTCC1763615370.0001, blue6.04 (4.04 – 9.04), 0.00001

Examination of Gene–gene Interactions Using MDR and Logistic Regression Analysis

By pooling the genotypic data obtained here for the 5 SNPs located in HERC2 and postulated regulatory region of OCA2 with previously determined data for additional polymorphisms within OCA2 and 3 other genes (MC1R, ASIP and SLC45A2), we were able to test for potential gene–gene interactions explaining variation in pigmentation present in this population sample. Table 6 presents multilocus interaction models for hair, eye and skin colour predicted by the MDR method and summarizes some measures reflecting their values. Table 7 summarizes results of multivariate logistic regression applied for the same set of markers. Only results significant after Bonferroni adjustment for multiple comparisons (p < 0.0125) are presented for this method. Both tables show results obtained with the standard additive model of allele coding. Although testing under assumption of a recessive model did not significantly alter the predicted pattern of gene–gene interactions, important statistical values were often found to be slightly better.

Table 6.  Multilocus interaction models for hair, eye and skin colour predicted by MDR
CharacteristicModelBalanced accuracyCV consistencyP value
Eye colour brown vs. non-brownrs129138320.835810/100.0075
rs12913832; rs18004070.81575/100.0095
rs12913832; rs4778241; rs18004070.79778/100.0118
rs916977; rs12913832; rs4778138; MC1R0.71415/100.0680
Eye colour blue vs. non-bluers129138320.831810/10<0.0001
rs12913832; rs168919820.82548/10<0.0001
rs12913832; MC1R; rs168919820.82927/10<0.0001
rs12913832; rs4778241; MC1R; rs168919820.82094/10<0.0001
Hair colour black vs. non-blackrs129138320.714710/100.0791
rs12913832; MC1R0.738210/100.0452
rs12913832; rs4778138; MC1R0.759810/100.0245
rs12913832; rs4778138; MC1R; rs168919820.73326/100.0557
Skin colour light vs. darkMC1R0.705410/100.0110
rs12913832; MC1R0.70449/100.0081
rs12913832; rs4778138; MC1R0.69827/100.0136
rs12913832; rs4778241; rs4778138; MC1R0.67396/100.0333
Table 7.  Multilocus interaction models for hair, eye and skin colour predicted by logistic regression analysis. Only results significant after Bonferroni correction for multiple testing in this regression model are presented (p values < 0.0125)
 Overall model fit P valueSignificant results
Marker (P value)OR (95% CI)
Eye colour brown vs. non-brown0.0000rs12913832 (0.0000)22.6075 (8.3949–60.8819)
rs1800407 (0.0016)0.0981 (0.0232–0.4145)
Eye colour blue vs. non-blue0.0000MC1R (0.0102)0.4289 (0.2249–0.8180)
rs12913832 (0.0000)0.0226 (0.0085–0.0602)
Hair colour black vs. non-black0.0000rs12913832 (0.00001)6.1180 (2.7432–13.6446)
MC1R (0.0002)0.2312 (0.1084–0.4932)
Skin colour light vs. dark0.0000rs12913832 (0.0003)0.2683 (0.1312–0.5487)
MC1R (0.0000)4.7607 (3.3009–6.8660)

Eye colour was found by MDR to be best predicted by HERC2 rs12913832 alone (BA = 0.8318, CV = 10/10, p < 0.0001 for blue eyes as a separate class). It is noteworthy that models assuming interaction between rs12913832 and OCA2 rs1800407 or SLC45A2 rs16891982 also achieved levels of significance. The model assuming interaction between rs12913832 and rs1800407 had better values under assumption of a recessive model (BA = 0.8230, CV = 8/10, p = 0.0074) (compare with Table 6). Remarkably better values were also obtained for the model assuming interactions between rs12913832, MC1R and rs16891982 (BA = 0.8387, CV = 9/10, p < 0.0001). Also, logistic regression indicated rs12913832 as the best eye colour predictor in the blue vs. non-blue model (OR = 0.0226; CI = 0.0085 – 0.0602; p = 0.00001). In the model with brown vs. non-brown iris colour, apart from rs12913832, OCA2 rs1800407 and MC1R were also independently associated with eye colour (Table 7). The recessive model of allele coding resulted in complete separation.

Hair colour is according to the MDR method best explained assuming interaction between polymorphisms HERC2 rs12913832, OCA2 rs4778138 and the state of MC1R (BA = 0.7598, CV = 10/10, p = 0.0245). These values were even better when the recessive model was tested (BA = 0.7771, CV = 10/10, p = 0.0175). This result is partially consistent with logistic regression, which indicated HERC2 rs12913832 (OR = 6.12, CI = 2.74 – 13.64, p = 0.0000) and MC1R (OR = 0.23, CI = 0.11–0.49, p = 0.0002) as important predictors of hair colour in the studied model. In the case of the recessive model of allele coding apart from rs12913832 (OR = 13.44, CI = 4.72 – 38.33, p = 0.0000) and MC1R (OR = 0.27, CI = 0.13 – 0.56, p = 0.0005), SLC45A2 rs16891982 was also found to be involved (OR = 10.69, CI = 1.15 – 99.07, p = 0.0370).

Skin colour is according to MDR mostly under the control of MC1R (BA = 0.7054, CV = 10/10, p = 0.0110). However, the model of interaction between MC1R and HERC2 rs12913832 should also be taken into account as it has only slightly worse measure values. The interaction between MC1R and HERC2 is even more strongly supported when the recessive model of allele coding is applied (BA = 0.7138, CV = 9/10, p = 0.0053). In particular, the interaction between rs12913832 and MC1R was confirmed by logistic regression (OR = 0.27, CI = 0.13 – 0.55, p = 0.0003 and OR = 4.76, CI = 3.30–6.87, p < 0.00001, respectively). Similar values were calculated with the recessive model of allele coding (OR = 0.20, CI = 0.09 – 0.46, p = 0.0002 and OR = 4.68, CI = 3.25 – 6.72, p = 0.0000, respectively). A graphical representation of the interaction between HERC2 and MC1R as predicted by MDR is shown in Figure 1.

image

Figure 1. Summary of two-locus genotype combinations associated with “high risk” (light skin) and “low risk” (dark skin). The difference in patterns of “high risk” (dark grey) and “low risk” (light grey) cells across each of the different multilocus dimensions can be seen. C/C, C/T, T/T –HERC2 genotypes. 0/0, 0/R, R/R –MC1R state according to the number of major function variants.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Pigmentation phenotype is, in conjunction with a high level of UV exposure, considered a significant contributor to melanoma and non-melanoma skin cancers and thus is widely studied in medical genetics (Rees, 2004). Since human pigmentation might have been affected during the course of human evolution by various selective pressures, it is also a subject of research conducted in evolutionary biology (Jablonski, 2004). Moreover, eye, hair and skin colour prediction has been proposed to have potential use in forensic genetics (Tully, 2007). Variation in human pigmentation is particularly high in populations of European descent and some characteristics like blue irides and red hair colour are present almost exclusively in people of European origin. Recent research conducted on human pigmentation has enabled the discovery of many new genes and gene variants involved in eye, skin and hair colouration (Sulem et al., 2007; Han et al., 2008; Kayser et al., 2008; Sturm et al., 2008; Sulem et al., 2008). Pigmentation is presumed to be under the control of 120 genes (Bennett & Lamoreux, 2003), which may act at different stages of the complex process of melanin formation, transportation and packaging. Hence, it seems that the influence of gene–gene interactions should be considered a remarkable factor contributing to the high variability of pigmentation among humans. Here we report additional evidence that gene variants of HERC2 and OCA2 are independently associated with iris colour. Moreover, MDR and logistic regression were consistent that interactions between HERC2 and MC1R may significantly contribute to variation in hair colour and skin type in the studied population sample.

We examined three SNPs from OCA2 intron 1 which have been associated with eye, hair and skin colour (Duffy et al., 2007) and two polymorphisms within HERC2 which have been found by two independent groups as the single best eye colour predictors (Kayser et al., 2008; Sturm et al., 2008). In this study, all these SNPs were in strong linkage disequilibrium and significantly associated with eye colour in separate tests. The most significant result was obtained for HERC2 rs12913832. This confirms the findings of both Sturm et al. (2008), who selected this position from 92 SNPs as a causative site involved in eye colour determination and Eiberg et al. (2008), who also suggested its involvement in eye colour determination. Moreover, in our population, this SNP was also associated with hair colour and skin type. Han et al. (2008) also reported that rs12913832 shows significant, independent association with hair colour and tanning ability. Therefore, the ancestral T allele may be responsible for darker pigmentary phenotypes in humans. OR values obtained here for different models of inheritance tested indicate that variant T of rs12913832 polymorphism is dominant over allele C, which is strongly associated with blue iris colour. The same trend was found for the remaining analysed loci. Higher ORs and often lower P values were obtained under assumption of dominance relationships within SNP loci.

rs12913832 was the only SNP which remained significantly associated with eye, hair and skin colour in the regression model with all the five analysed polymorphisms as covariates. The strong LD among the five SNPs tested in the OCA2-HERC2 region and serious differences in minor allele frequencies may cause some difficulties in reliable inference of their actual effect on pigmentation status, e.g. whether their action is independent or interactive. The strongest association with eye colour was found for rs12913832, which, however, at the same time has the largest minor allele frequency and thus its analysis provides the highest statistical power. However, it is also worth noting that two frequent haplotypes associated with opposing effects on human pigmentation, namely haplotype 15, associated with blue eye colour and haplotype 14, associated with non-blue eye colour, differ in only one nucleotide position: rs12913832. We also checked that the predictive value did not increase when haplotypes were used instead of the SNP genotype. Even after clustering two blue eye colour associated haplotypes (15 and 8), the OR was 13.8 (data not shown) compared to OR = 28.2 associated with rs12913832. The discovery that HERC2 may be implicated in eye colour determination (Sulem et al., 2007) has drawn more attention to this gene, which is located on chromosome 15q13.1, near the known pigmentation gene OCA2. The OCA2 gene (15q11.2-12) was previously considered a major eye colour gene (Frudakis et al., 2003; Zhu et al., 2004). It consists of 24 exons and encodes an 838 amino acid protein which is an integral part of the melanosomal membrane. The product of the OCA2 gene is thought to be involved in transportation of tyrosine – the crucial substrate in the process of melanogenesis (Lee et al., 1995). Although HERC2 is not a pigmentary gene, it has been linked to Prader-Willi and Angelman syndromes, which are very often associated with hypopigmentation (Nicholls & Knepper, 2001). Due to a significant sequence homology across mammalian and invertebrate species, HERC2 is considered a highly conserved gene (Ji et al., 2000). Interactive effects between HERC2 and OCA2 were postulated years ago based on results obtained from studies on the mouse species model (Walkowicz et al., 1999). Sulem et al. (2007) hypothesised that sequences located in introns of HERC2 may affect OCA2 expression in humans. The first evidence for such an interaction was soon reported by Kayser et al. (2008).

HERC2 rs916977 was here the second best predictor of iris colour. In the model without rs12913832, rs916977 was found to be most significantly associated with eye colour. Interestingly, from the three intron 1 SNPs, only two – rs7495174 and rs4778241 – were still important when all three polymorphisms were included together in the regression analysis. In the study by Duffy et al. (2007), all three SNPs were significant. Contrarily, Kayser et al. (2008) found that two SNPs (rs7495174 and rs4778138) have an independent effect when all three are regressed together, but that was a different combination compared to our SNPs. This partial inconsistency may be caused by interpopulation differences, but this indicates that all three intron 1 SNPs should be considered as potentially valuable for prediction of pigmentation features. Our results confirmed that position rs12913832 is highly associated with eye colour and should be considered as a significant marker of blue/brown iris colour. Homozygotes C/C have more than an 80% chance of having blue irides and such a genotype was not present among individuals with brown eyes (Table 4). In the study by Sturm et al. (2008), this genotype was present in merely 1% of individuals with brown eyes, thus it can be concluded that this SNP may be applied to blue/brown eye colour prediction.

Apparently, discovery of functional variants is not possible using genetic association testing alone, especially when we need to choose, as in this case, from several closely linked polymorphisms. However, it has additionally been established that the most relevant position, rs12913832, is located in a conservative area of the HERC2 intron 86. Detailed analysis of this region revealed transcription factor binding sites, which supported the hypothesis that it may be involved in transcription regulation of the OCA2 gene (Sturm et al., 2008). The first evidence based on direct experimental genetic methods supporting this idea was shown recently by Eiberg et al. (2008). These results indicated that the conservative sequence in the intron 86 may indeed act as a transcriptional silencer. Using mobility shift assay they were also able to show that two variants of rs12913832 reveal a different protein binding pattern and thus possibly differ in affinity for nuclear factors (Eiberg et al., 2008). Additional experiments are needed to confirm these initial results and to finally establish a molecular mechanism of action of HERC2 rs12913832.

In the next step, we combined results obtained for the five SNPs from the HERC2OCA2 regulatory region with our previous genotyping data obtained for OCA2, MC1R, ASIP and SLC45A2 (MATP) for the same population sample. Data for the total number of 10 known important pigmentary SNPs and MC1R state were implemented in the model, which was subjected to analysis using the nonparametric multifactor dimensionality reduction and logistic regression. In this complete dataset, HERC2 rs12913832 remained the best SNP explaining blue/brown eye colour according to both − MDR and logistic regression (blue/non-blue model). However, it should be noted that other significant models indicated by MDR assumed interaction between this SNP and OCA2 rs1800407 or MC1R and SLC45A2 rs16891982. rs1800407 refers to a non-synonymous alteration, Arg419Gln, which was one of the first OCA2 polymorphisms associated with eye colour (Rebbeck et al., 2002) and this finding has been replicated in other studies (Frudakis et al., 2003; Duffy et al., 2007; Branicki et al., 2008a). It has been found that this SNP has an independent effect on eye colour when simultaneously tested with HERC2 rs12913832 in a single regression model and thus may be involved in modification of the HERC2 effect (Sturm et al., 2008). In our study, rs1800407 was also significant for eye colour when regressed together with other polymorphisms in the brown/non-brown model. It is interesting that also MC1R was found to be significant in this model (Table 7). rs16891982 also refers to a non-synonymous alteration previously implicated in human pigmentation. The variant Leu374 was associated with darker pigmentation in populations of European origin (Graf et al., 2005; Fernandez et al., 2008; Han et al., 2008).

MDR and logistic regression were consistent about interactive effects between HERC2 rs12913832 and MC1R affecting human hair colour and skin colour. The best model for hair colour predicted by MDR includes, apart from rs12913832 and MC1R, OCA2 rs4778138 as well. This latter SNP has been previously shown to be independently associated with freckle and mole number (Duffy et al., 2007), thus its involvement in pigmentation seems to be possible. In the variant with the recessive model of allele coding, logistic regression also implicated the above mentioned SLC45A2 rs16891982. Indeed, our previous study showed that in this population sample, the rs16891982 variant was significantly associated with black hair (Branicki et al., 2008b). The epistatic effects between rs12913832 and MC1R are even more evident in the case of skin colouration. Such an interaction was predicted by both MDR and logistic regression. MDR found two almost equivalent models, the first assuming action of MC1R alone (this is not a surprise as the role of MC1R in skin colour determination is very significant) and the second indicating interaction between HERC2 and MC1R. The latter was more significant under assumption of dominance relationships within the HERC2 locus. It can be seen from Figure 1 that major function MC1R variants R strongly predispose to light skin colour. This observation has been previously confirmed in many studies (Rees, 2003). However, the T allele of the HERC2 rs12913832 is able to mask this effect in individuals who have one MC1R R variant. This masking effect is not visible for two MC1R variant carriers, which provides additional evidence for an important role of MC1R in pigmentation. MC1R controls the crucial step of pheomelanin/eumelanin production and therefore is considered the major pigmentation gene. The MC1R gene is located on chromosome 16q24.3 and encodes the G protein-coupled receptor associated with the melanocyte membrane. Its activation leads to cAMP increase and efficient eumelanin production. Some non-synonymous mutations within MC1R, known as major function mutations (R) have been proved to decrease eumelanin production and thus lead to the typical phenotype of red hair colour and fair skin (Rees, 2003). However, it has been found that in the population of Liguria, that some two-R variant carriers did not reveal typical red hair and light skin phenotype. Therefore, it is possible that despite controlling the key point of melanogenesis, MC1R may be masked by other genes at least in populations with physiologically darker pigmentation (Pastorino et al., 2004). Akey et al. (2001) reported an interactive effect of MC1R and OCA2 influencing skin features. Our results indicate that HERC2 rs12913832 may be one of the genes responsible for masking the effect of MC1R polymorphisms which significantly affect performance of the melanocortin 1 receptor. Further studies are required to confirm this preliminary result. In particular, a more objective and accurate method of quantifying skin colour, such as reflectance spectrophotometry, would be advisable, since the Fitzpatrick scale applied in our research may be inadequate, confounding skin response with constitutive skin pigmentation level.

In conclusion, our data provided additional evidence that HERC2 rs12913832 is highly associated with eye colour and also has an important effect on hair colour and skin type. The ancestral allele T that is linked with darker pigmentation phenotypes seems to be dominant over allele C. Gene variants HERC2 rs12913832 and OCA2 rs1800407 are independently associated with iris colour and thus both may be considered useful for eye colour prediction. Interactions between HERC2 and MC1R may contribute to variation in other pigmentation traits – hair colour and skin type and the allele T of the HERC2 gene may be epistatic to major function MC1R alleles.

Acknowledgments

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

We would like to express our gratitude to all sample donors. We would also like to thank all our colleagues who helped at different stages of this project, amongst whom are Ashley Whelan, Magda Łącka and Olga Rydzyk, to mention just a few. This research was supported by grants from the Ministry of Science and Higher Education no 0T00C01829 and the Institute of Forensic Research no I/G/2008.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information
  • Akey, J. M., Wang, H., Xiong, M., Wu, H., Liu, W., Shriver, M. D. & Jin, L. (2001) Interaction between the melanocortin-1 receptor and P genes contributes to inter-individual variation in skin pigmentation phenotypes in a Tibetan population. Hum Genet 108, 516520.
  • Bennett, D. C. & Lamoreux, M. L. (2003) The color loci of mice – a genetic century. Pigment Cell Res 16, 333344.
  • Branicki, W., Brudnik, U., Kupiec, T., Wolańska-Nowak, P., Szczerbińska, A. & Wojas-Pelc, A. (2008a) Association of polymorphic sites in the OCA2 gene with eye colour using the tree scanning method. Ann Hum Genet 72, 184192.
  • Branicki, W., Brudnik, U., Draus-Barini, J., Kupiec, T. & Wojas-Pelc, A. (2008b) Association of the SLC45A2 gene with physiological human hair colour variation. J Hum Genet, DOI:DOI: 10.1007/s10038--008-0338--3.
  • Brudnik, U., Branicki, W., Wojas-Pelc, A. & Kanas, P. (2008) The contribution of melanocortin 1 receptor gene polymorphisms and the agouti signalling protein gene 8818A>G polymorphism to cutaneous melanoma and basal cell carcinoma in a Polish population. Exp Dermatol, DOI:DOI: 10.1111/j.1600--0625.2008.00760.x.
  • Carlborg, O. & Haley, C. S. (2004) Epistasis: too often neglected in complex trait studies? Nat Rev Genet 5, 618625.
  • Cordell, H. J. (2002) Epistasis: what it means, what it doesn't mean, and statistical methods to detect it in humans. Hum Mol Genet 11, 24632468.
  • Duffy, D. L., Montgomery, G. W., Chen, W., Zhao, Z. Z., Le, L., James, M. R., Hayward, N. K., Martin, N. G. & Sturm, R. A. (2007) A three-single-nucleotide polymorphism haplotype in intron 1 of OCA2 explains most human eye-color variation. Am J Hum Genet 80, 241252.
  • Eiberg, H., Troelsen, J., Nielsen, M., Mikkelsen, A., Mengel-From, J., Kjaer, K. W. & Hansen, L. (2008) Blue eye color in humans may be caused by a perfectly associated founder mutation in a regulatory element located within the HERC2 gene inhibiting OCA2 expression. Hum Genet 123, 177187.
  • Excoffier, L., Laval, G. & Schneider, S. (2005) Arlequin ver 30: An integrated software package for population genetics data analysis. Evolutionary Bioinformatics Online 1, 4750.
  • Fernandez, L. P., Milne, R. L., Pita, G., Avilés, J. A., Lázaro, P., Benítez, J. & Ribas, G. (2008) SLC45A2: a novel malignant melanoma-associated gene. Hum Mutat 29, 11611167.
  • Fitzpatrick, T. B. (1988) The validity and practicability of sun-reactive skin types I through VI. Arch Dermatol 124, 869871.
  • Frudakis, T., Thomas, M., Gaskin, Z., Venkateswarlu, K., Chandra, K. S., Ginjupalli, S., Gunturi, S., Natrajan, S., Ponnuswamy, V. K. & Ponnuswamy, K. N. (2003) Sequences associated with human iris pigmentation. Genetics 165, 20712083.
  • Frudakis, T. (2008) Molecular photofitting. Predicting ancestry and phenotype using DNA. Academic Press.
  • Graf, J., Hodgson, R. & Van Daal, A. (2005) Single nucleotide polymorphisms in the MATP gene are associated with normal human pigmentation variation. Hum Mutat 25, 278284.
  • Hahn, L. W., Ritchie, M. D. & Moore, J. H. (2003) Multifactor dimensionality reduction software for detecting gene-gene and gene-environment interactions. Bioinformatics 19, 376382.
  • Han, J., Kraft, P., Nan, H., Guo, Q., Chen, C., Qureshi, A., Hankinson, S. E., Hu, F. B., Duffy, D. L., Zhao, Z. Z., Martin, N. G., Montgomery, G. W., Hayward, N. K., Thomas, G., Hoover, R. N., Chanock, S. & Hunter, D. J. (2008) A genome-wide association study identifies novel alleles associated with hair color and skin pigmentation. PLoS Genet 4, e1000074.
  • Jablonski, N. (2004) The evolution of human skin and skin color. Annu Rev Anthropol 33, 585623.
  • Ji, Y., Rebert, N. A., Joslin, J. M., Higgins, M. J., Schultz, R. A. & Nicholls, R. D. (2000) Structure of the highly conserved HERC2 gene and of multiple partially duplicated paralogs in human. Genome Res 10, 319329.
  • Kayser, M., Liu, F., Janssens, A. C., Rivadeneira, F., Lao, O., Van Duijn, K., Vermeulen, M., Arp, P., Jhamai, M. M., Van Ijcken, W. F., Den Dunnen, J. T., Heath, S., Zelenika, D., Despriet, D. D., Klaver, C. C., Vingerling, J. R., De Jong, P. T., Hofman, A., Aulchenko, Y. S., Uitterlinden, A. G., Oostra, B. A. & Van Duijn, C. M. (2008) Three genome-wide association studies and a linkage analysis identify HERC2 as a human iris color gene. Am J Hum Genet 82, 411423.
  • Lee, S. T., Nicholls, R. D., Jong, M. T., Fukai, K. & Spritz, R. A. (1995) Organization and sequence of the human P gene and identification of a new family of transport proteins. Genomics 26, 354363.
  • Moore, J. H. (2003) The ubiquitous nature of epistasis in determining susceptibility to common human diseases. Hum Hered 56, 7382.
  • Moore, J. H. (2004) Computational analysis of gene-gene interactions using multifactor dimensionality reduction. Expert Rev Mol Diagn 4, 795803.
  • Musani, S. K., Shriner, D., Liu, N., Feng, R., Coffey, C. S., Yi, N., Tiwari, H. K. & Allison, D. B. (2007) Detection of gene x gene interactions in genome-wide association studies of human population data. Hum Hered 63, 6784.
  • Nicholls, R. D. & Knepper, J. L. (2001) Genome organization, function, and imprinting in Prader-Willi and Angelman syndromes. Annu Rev Genomics Hum Genet 2, 153175.
  • Oetting, W. S., Garrett, S. S., Brott, M. & King, R. A. (2005) P gene mutations associated with oculocutaneous albinism type II (OCA2). Hum Mutat 25, 323329.
    Direct Link:
  • Pastorino, L., Cusano, R., Bruno, W., Lantieri, F., Origone, P., Barile, M., Gliori, S., Shepherd, G. A., Sturm, R. A. & Bianchi-Scarra, G. (2004) Novel MC1R variants in Ligurian melanoma patients and controls. Hum Mutat 24, 103.
  • Rebbeck, T. R., Kanetsky, P. A., Walker, A. H., Holmes, R., Halpern, A. C., Schuchter, L. M., Elder, D. E. & Guerry, D. (2002) P gene as an inherited biomarker of human eye color. Cancer Epidemiol Biomarkers Prev 11, 782784.
  • Rees, J. L. (2003) Genetics of hair and skin color. Annu Rev Genet 37, 6790.
  • Rees, J. L. (2004) The genetics of sun sensitivity in humans. Am J Hum Genet 75, 739751.
  • Ritchie, M. D., Hahn, L. W., Roodi, N., Bailey, L. R., Dupont, W. D., Parl, F. F. & Moore, J. H. (2001) Multifactor-dimensionality reduction reveals high-order interactions among estrogen-metabolism genes in sporadic breast cancer. Am J Hum Genet 69, 138147.
  • Stephens, M., Smith, N. J. & Donnelly, P. (2001) A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68, 978989.
  • Sulem, P., Gudbjartsson, D. F., Stacey, S. N., Helgason, A., Rafnar, T., Magnusson, K. P., Manolescu, A., Karason, A., Palsson, A., Thorleifsson, G., Jakobsdottir, M., Steinberg, S., Pálsson, S., Jonasson, F., Sigurgeirsson, B., Thorisdottir, K., Ragnarsson, R., Benediktsdottir, K. R., Aben, K. K., Kiemeney, L. A., Olafsson, J. H., Gulcher, J., Kong, A., Thorsteinsdottir, U., Stefansson, K. (2007) Genetic determinants of hair, eye and skin pigmentation in Europeans. Nat Genet 39, 14431452.
  • Sulem, P., Gudbjartsson, D. F., Stacey, S. N., Helgason, A., Rafnar, T., Jakobsdottir, M., Steinberg, S., Gudjonsson, S. A., Palsson, A., Thorleifsson, G., Pálsson, S., Sigurgeirsson, B., Thorisdottir, K., Ragnarsson, R., Benediktsdottir, K. R., Aben, K. K., Vermeulen, S. H., Goldstein, A. M., Tucker, M. A., Kiemeney, L. A., Olafsson, J. H., Gulcher, J., Kong, A., Thorsteinsdottir, U. & Stefansson, K. (2008) Two newly identified genetic determinants of pigmentation in Europeans. Nat Genet 40, 835837.
  • Sturm, R. A., Duffy, D. L., Zhao, Z. Z., Leite, F. P., Stark, M. S., Hayward, N. K., Martin, N. G. & Montgomery, G. W. (2008) A single SNP in an evolutionary conserved region within intron 86 of the HERC2 gene determines human blue-brown eye color. Am J Hum Genet 82, 424431.
  • Tully, G. (2007) Genotype versus phenotype. Forensic Sci Int: Genetics 1, 105110.
  • Vermeulen, S. H., Den Heijer, M., Sham, P. & Knight, J. (2007) Application of multi-locus analytical methods to identify interacting loci in case-control studies. Ann Hum Genet 71, 689700.
  • Walkowicz, M., Ji, Y., Ren, X., Horsthemke, B., Russell, L. B., Johnson, D., Rinchik, E. M., Nicholls, R. D. & Stubbs, L. (1999) Molecular characterization of radiation- and chemically induced mutations associated with neuromuscular tremors, runting, juvenile lethality, and sperm defects in jdf2 mice. Mamm Genome 10, 870878.
  • Zhu, G., Evans, D. M., Duffy, D. L., Montgomery, G. W., Medland, S. E., Gillespie, N. A., Ewen, K. R., Jewell, M., Liew, Y. W., Hayward, N. K., Sturm, R. A., Trent, J. M. & Martin, N. G. (2004) A genome scan for eye color in 502 twin families: most variation is due to a QTL on chromosome 15q. Twin Res 7, 197210.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  9. Supporting Information

Table S1 PCR primers for two multiplex reactions

Table S2 Extension primers used for two SNaPshot reactions

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
AHG_504_sm_TablesS1-S2.doc25KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.