As in many mammals, coat colour in cattle results from the relative presence of eumelanin (black-brown pigment) and phaeomelanin (red-yellow pigment), the two basic pigments produced by melanocyte cells 1. Pigment production takes place in the melanosomes, organelles containing the enzymes directly involved in pigment biosynthesis. Tyrosinase (TYR) is the rate-limiting enzyme in the melanogenesis pathway. High levels of this enzyme are required for the production of eumelanin, whereas low enzyme levels result in the production of phaeomelanin 23. Tyrosinase activity is regulated by the melanocortin 1 receptor (MC1R or α-MSHR), whose stimulation by α-melanocyte-stimulating hormone (α-MSH) leads to the production of eumelanin 4. Phaeomelanin is produced in absence of α-MSH stimulation, either as result of a non-functional MC1R receptor 5 or in the presence of the Agouti protein, which is secreted by cells adjacent to melanocytes and acts as an antagonist of the α-MSH action by blocking the MC1R 6. In addition to the genes coding for these proteins essential for pigmentation, work in mice has uncovered more than 120 genes involved in colour variation. These include genes involved in the biosynthesis of melanin (Tyrp1, Tyrp2), the biology of melanocytes and melanosomes (e.g. ePomc1, Mitf, Silver, Ap3, Mlph, Myo5a, Rab27a) and migration and survival of melanocytes during development (e.g. Kit, Kitl, Edn3 and Ednrb) 7. This information provides a number of candidate genes that may also affect coat colour in other species, including cattle.

In cattle, the Extension locus (MC1R), located on chromosome 18 8, plays a major role in the regulation of the synthesis of eumelanin versus phaeomelanin. The most common alleles at this locus are the dominant E^D allele and the recessive e allele, which are responsible for the black and red colour, respectively, and code for a receptor which is not affected by the Agouti protein. Breeds with a mix of red and black hairs (wild-type colour) carry Agouti-receptive alleles (E⁺, E¹ and E²) 910. In contrast to mice, the role of the bovine locus for the Agouti protein (ASIP) in colour variation seems limited as no allelic variants have been found in the coding sequence 11.

In some cattle breeds, such as Dexter, Galloway, Charolais, Highland and Simmental, a lightening or dilution of the base colour defined by the Extension locus is observed. A wide range of colours results from this dilution phenomenon (white, cream, dun, gold, yellow, pale red, grey or brown). The gene responsible of the dilution seems to vary between breeds: e.g. the pale (dun) coat colour observed in Dexter cattle is due to the gene coding for the TYRP1 (tyrosine related protein 1), but this gene has been excluded as being responsible for dilution in other breeds 12. The silver (SILV) gene, which codes for a type I integral membrane protein in the pre-melanosome matrix (PMEL17) 13, and which is essential for melanosome development 1415, has been found to be responsible for coat colour dilution in Highland cattle 16.

The Charolais breed exhibits the most extreme case of dilution, as pure-bred Charolais individuals have a uniform white coat colour, despite the e/e Extension genotype of most Charolais cattle. Therefore, the characteristic coat colour of this breed results from a phaeomelanin dilution over the entire body. Charolais cattle are thought to be homozygous for a dilution mutation (Dilution Charolais, Dc/Dc), that in heterozygotes (Dc/dc⁺) produces an intermediate phenotype: grey colour if the dilution applies to eumelanin (e.g. in a Black Angus × Charolais cross) or pale red or yellow if the background pigment is phaeomelanin 17.

Using an F2 Holstein × Charolais population, Kühn and Weikard 18 recently reported an association between dilution of black pigment (eumelanin) and a region on bovine chromosome 5 including the SILV gene. However, these authors did not include animals of red background in their analysis, and therefore it remains to be shown whether the locus affecting the dilution of phaeomelanin, typical of the Charolais breed, co-locates with the linkage association reported by these authors.

We performed a genome scan to localise the genetic locus responsible for the dilution phenotype using a Charolais × Holstein experimental population obtained through a F2 and balanced Backcross design. The linkage analysis included individuals with both black and red coat colour background, and therefore addressed the localization of the major locus involved in the dilution of the phaeomelanic pigment in the Charolais breed.

The variation observed in the coat colour of the F2 and Backcross individuals of this Charolais × Holstein population provides an opportunity to investigate effects and the mode of inheritance of the Charolais dilution locus. Based on the phenotypic data (Table 1), it is clear that the Dc and the Extension loci are mainly responsible for the variation in coat colour observed. As multiple alleles at the Extension (MC1R) locus were segregating in this population, the effect of the Dc locus on both types of backgrounds was confirmed by the observation of a complete or partial dilution affecting individuals with E^D-, E⁺e or ee MC1R genotype. The pale colour observed in individuals with E⁺e MC1R genotype demonstrates the dilution of pigments produced by both Agouti-responsive (E⁺-) and non-responsive (ee) melanocytes. This consistency of the effect across MC1R genotypes was also supported by the results of the genome scan, in which the same region on chromosome 5 showed linkage with the three dilution-related traits analyzed. In addition, the additive effects estimated for Quantitative-Black and Quantitative-Red had similar size. The dominance effect was very small relative to the additive effect; therefore a single copy of the Dc allele originating from the Charolais is sufficient to dilute either eumelanin or phaeomelanin. Heterozygous individuals, Dc/dc⁺, are generally of intermediate phenotype (light-grey or light-red) and two copies of the Dc allele are required to produce a complete dilution of the original pigment (white phenotype). These results are consistent with the inheritance of the Charolais dilution locus described in the literature 17. The data also support the assumption of alternative fixed alleles at the Dc locus in the founder lines, on which the regression analysis was based. Therefore, the power of detection of this locus was maximised, which is reflected in the high significance of the associations identified on chromosome 5.

The location of the major gene associated with the Quantitative-Dilution, Quantitative-Black and Quantitative-Red traits is coincident with the position reported in a linkage study of eumelanin dilution (black pigment) in a Holstein × Charolais F2 population 18. In the region of bovine chromosome 5 flanked by markers ETH10 and DIK5248, there are several metabolic candidate genes directly related to pigmentation pathways (ErbB3, SILV), and members of gene families where at least one member is suggested to have an effect on pigmentation (BLOC1S1, RAB5b, DCTN2 and MYO1A). Among these, the SILV gene is the only one with an established function in the melanocyte and therefore is the strongest candidate. This gene is between 56.407 and 56.415 Mb in the latest version of the bovine genome sequence assembly (Build 3.1) 21, according to which, ETH10 is located at 55.333 Mb 21. It codes for a pre-melanosomal matrix protein (PMEL17) necessary for the formation of the fibril matrix upon which melanin intermediates are deposited late in melanosome maturation 14. Mutations in the SILV gene are known to cause diluted phenotypes in mice 22, horse 23 and dog 24, although in these species the effect is to block the production of eumelanin without effects on phaeomelanin. In chicken, allelic variations in this gene also block the production of black pigment in the plumage leading to the Smoky, Dun, and Dominant white colour variants 25.

In cattle, a Charolais-specific allele has previously been reported in exon 1 of the bovine SILV gene 19. This mutation is a G>A substitution that results in a change from glycine to arginine within the N-terminal signal sequence of the PMEL17 protein. Among thirteen breeds tested, the A allele was only identified in pure-breed Charolais individuals or Charolais crosses 19. This mutation was genotyped across the individuals of the resource population scored for coat colour (F2, CB1 and HB1). The observed distribution of genotypes within the three genetic background groups supports the hypothesis of fixation of alleles in the founder lines, with the A allele only present in Charolais founders. To test the association between the SILV c.64A>G genotypes and the diluted phenotype, the genotype of this mutation was included as a fixed effect in the regression model fitted to the Quantitative-Dilution, Quantitative-Black and Quantitative-Red traits. For all traits, inclusion of this variant resulted in the disappearance of the highly significant linkage associations, suggesting that the Dc locus is either due to or in strong linkage disequilibrium with the SILV c.64A>G mutation.

Some discrepancies between SILV c.64A>G and the phenotype were observed however, which draws into question whether SILV c.64A>G is the causative mutation underlying the Dc dilution effect, as other authors have suggested 18. For these discordant animals, the CHROMPIC analysis including the SILV c.64A>G mutation and the Dc locus (presumed genotypes based on phenotypes) did not suggest genotyping errors for the tested mutation, however, most of these animals appeared as double recombinants at the Dc locus. The probability of a genuine double recombination event in such a small chromosomal interval is very low, and to detect several such double recombinants in the number of animals examined here would be very unlikely. Hence, apart from possible phenotype-genotype mismatches, these double recombination events are more likely to be the result of either phenotypic mis-scoring or the effect of other loci influencing coat colour. Possible mis-scoring may be explained by difficulties in distinguishing between the partially and the completely diluted phenotypes (especially Light-Grey/Light-Red against Off-White) or in scoring some individuals showing a non-homogenous dilution along the body (e.g. darker head than body).

Under the possibility of another locus or loci affecting the coat colour variation in this population, the results of the analysis of Grey-Intensity may help to interpret the minor gene effects revealed by the primary analysis. The proximal region of chromosome 28 was the only significant effect at the suggestive level for the diluted-related traits and for the Grey-Intensity trait, which suggests that these significant associations could result from the true quantitative nature of coat colour intensity within and between the phenotype classes. This locus could, therefore, be considered as a candidate for the genetic background effects that underlie subtle variations in colour, and that in certain cases could lead to discordance between colour score and the SILV locus genotype (e.g. this could explain discrepancies such as AG animals that were scored as White). A colour-associated gene, LYST (lysosomal trafficking regulator), maps to the proximal end of chromosome 28 26. Mutations in this gene are responsible for Chediak-Higashi syndrome 1 in human and mouse (beige mutant). This disorder has been reported in Japanese black cattle 27 and is characterized by prolonged bleeding time and, more relevantly for this paper, a light coat colour. Our results indicate that allelic variation at this gene, possibly not associated with illness, could underlie the different shades of colours observed in the partially diluted colour categories by acting as a modifier of the Dc locus. Increased marker density in this chromosomal region would be required before an epistatic analysis between this locus and the Dc locus could be conducted.

Other genetic effects may be the result of the interaction of the causal mutation of the Charolais dilution phenotype and other mutations in the SILV gene. For instance, the AG individual with Dark-Red phenotype rather than the expected Light-Red may be explained if another mutation rescued the dilution effect due to the SILV c.64A>G mutation as seems to be the case with the Smoky phenotype in chickens, which in addition to the 9-bp deletion in exon 10 of the SILV gene associated with the Dominant white phenotype, also have an additional deletion in exon 6 that partially restores pigment production 25.

Based on the CHROMPIC analysis, only four discordant animals showed a putative single recombination event between the SILV c.64A>G and the Dc locus, however, these discrepancies could not be conclusively confirmed as the phenotype of these individuals was intermediate between pale (Light-Grey/Light-Red) and Off-White. Based on the lack of convincing recombinants between the SILV c.64A>G mutation and the Dc locus, this allelic variant of the SILV gene cannot be ruled out as the causal mutation of the Charolais dilution phenotype. The effect of this locus on the phenotype is supported by the loss of significance in the regression analysis when this mutation is included as a fixed effect in the model. However, this does not exclude the possibility of a different mutation tightly linked to SILV c.64A>G being the Dc causal mutation, although we and others 18 have not found other mutations associated with coat colour in the coding region of the SILV gene.

The interaction between the SILV gene and pigment type appears to be complicated. The pigment-specificity of mutations in the SILV gene observed in other species 22232425 is in agreement with the critical role reported for this protein in eumelanosomes but not in phaeomelanosomes 28 and the suppression of PMEL17 expression seen in murine phaeomelanosomes 2930. However, recent work in Highland cattle reported a 3-bp deletion in exon 1 of the bovine SILV gene associated with the dilution of both red and black pigments 16. This finding, and the likely association of the SILV gene and the Dc locus, which affects both pigments, are intriguing and may suggest that the role of PMEL17 differs between species. This is plausible as the genuine function of the SILV gene product in pigmentation is not completely understood 14 and the biological basis of pigmentation may vary with species. Mutations in the SILV gene that have only been shown to affect eumelanin background are located in the c-terminal sequence of the SILV gene and affect the transmembrane or cytoplasmatic domains of the protein 22232425. It is possible that mutations closer to the N-terminal end (such as exon 1, where both cattle mutations are found) could lead to more general interference with pigment production. Exon 1 codes for the signal peptide sequence of the protein 14, which is thought to determine the entry of PMEL17 into the secretory pathway prior to its processing and cleavage 31.

The work reported in this paper localizes the locus responsible for the dilution effect of the Charolais breed, Dc, to bovine chromosome 5, and demonstrates that this locus acts on both black (eumelanin) and red (phaeomelanin) pigment backgrounds. The bovine SILV gene was assessed as a candidate for this linkage association. Although the non-synonymous SILV c.64A>G mutation, previously described 19, does not explain all the phenotypes in the population studied, no convincing evidence was found to exclude it as the causative mutation for the Charolais dilution phenotype. Other genetic effects, such as those observed on chromosome 28 for pigment intensity, may be influencing the coat colour variation of this population. A candidate gene for this effect, LYST, has been identified.

BG-G carried out part of the genotyping experiments, performed the statistical analyses, performed error-checking on phenotype and genotype data, and drafted the manuscript. PW participated in the design and coordination of the study, compiled the phenotype data and helped to draft the manuscript. JLW conceived of the study, participated in its design and coordination, selected the initial marker panel and helped to draft the manuscript. All authors read and approved the final manuscript.