Genetics of Sex-linked yellow in the Syrian Hamster

Abstract

Alternating patches of black and yellow pigment are a ubiquitous feature of mammalian color variation that contributes to camouflage, species recognition, and morphologic diversity. X-linked determinants of this pattern—recognized by variegation in females but not in males—have been described in the domestic cat as Orange, and in the Syrian hamster as Sex-linked yellow (Sly), but are curiously absent from other vertebrate species. Using a comparative genomic approach, we develop molecular markers and a linkage map for the euchromatic region of the Syrian hamster X chromosome that places Sly in a region homologous to the centromere-proximal region of human Xp. Comparison to analogous work carried out for Orange in domestic cats indicates, surprisingly, that the cat and hamster mutations lie in nonhomologous regions of the X chromosome. We also identify the molecular cause of recessively inherited black coat color in hamsters (historically referred to as nonagouti) as a Cys115Tyr mutation in the Agouti gene. Animals doubly mutant for Sly and nonagouti exhibit a Sly phenotype. Our results indicate that Sly represents a melanocortin pathway component that acts similarly to, but is genetically distinct from, Mc1r and that has implications for understanding both the evolutionary history and the mutational mechanisms of pigment-type switching.

THE genetics of coat color is a longstanding and rich model system for studying gene action and cell signaling. Like classical genetic systems in invertebrate model organisms, the cell types and tissues that give rise to mammalian hair color are well characterized and experimentally accessible, and alterations in gene activity can be easily detected. Coat color mutations are especially useful for studying genes and pathways unique to vertebrate genomes and have led to a deeper understanding of diverse biological processes because much of the molecular machinery used by the pigmentary system is either shared by or homologous to molecules used in other physiological pathways (Jackson 1997; Bennett and Lamoreux 2003; Steingrimsson et al. 2006).

This approach has proven particularly useful for dissecting the molecular mechanisms of pigment-type switching, a phenomenon in which melanocytes choose between synthesizing eumelanin (a relatively insoluble black or brown pigment) or pheomelanin (a cysteine-rich red or yellow pigment that is soluble in dilute alkali) (Silvers 1979). A focal point for pigment-type switching is the Agouti–melanocortin 1 receptor (Mc1r) pathway. Mc1r is a G-protein coupled receptor expressed in melanocytes, whereas Agouti protein is a paracrine signaling molecule secreted by specialized dermal cells that inhibits Mc1r signaling (Cone et al. 1996; reviewed in Barsh 2006). In laboratory mice, gain-of-function mutations that constitutively activate the Mc1r (e.g., somber, Mc1r^so) cause exclusive production of eumelanin, whereas loss-of-function Mc1r mutations (e.g., recessive yellow, Mc1r^e) cause exclusive production of pheomelanin (Robbins et al. 1993). On the other hand, because Agouti is a Mc1r antagonist, gain-of-function Agouti mutations (e.g., lethal yellow, A^y) cause exclusive production of pheomelanin, whereas loss-of-function mutations (e.g., extreme nonagouti, a^e) cause exclusive production of eumelanin (Siracusa 1994; Hustad et al. 1995).

In the early 1900s, Sewall Wright (Wright 1917) concluded that genetic mechanisms that control pigmentary variation have been largely conserved during mammalian evolution. In particular, the Agouti phenotype—in which individual hairs display a subapical band of pheomelanin on an otherwise eumelanic background—is observed across a wide range of mammalian phyla. In the past decade, gain- and loss-of-function Mc1r mutations have been identified in many domestic animals (Andersson 2003; Klungland and Vage 2003), and Agouti and/or Mc1r variation has been shown to contribute to pigmentary variation in several natural populations (Eizirik et al. 2003; Nachman et al. 2003; Mundy et al. 2004; Hoekstra et al. 2006; Steiner et al. 2007). Most experimental work on the Agouti–Mc1r pathway has been carried out in laboratory mice. Nonetheless, comparative zoologic studies by Wright (1918) and others (Little 1957; Searle 1968) suggested that some components of mammalian pigment-type switching are not represented as coat color mutations in laboratory mice. In addition to a regular black and yellow stripe pattern found in many carnivores that is usually attributed to the Tabby gene (Searle 1968), irregular black and orange patches due to the activity of an X-linked gene are also present in the domestic cat and the Syrian hamster; neither of these phenomena have been described in laboratory mice.

In the domestic cat, female-specific variegation of black and orange coat color patches has been appreciated for more than a century and helped support the initial hypothesis that random and epigenetically heritable X inactivation is a universal feature of placental mammals (Lyon 1962). A similar phenotype thought to be caused by mutation of a homologous X-linked gene was described in Syrian hamsters >40 years ago (Robinson 1966). As in cats, hemizygous male hamsters and homozygous female hamsters are completely yellow, while heterozygous females exhibit a characteristic tortoiseshell pattern of yellow patches on a black background. (The terms tortoiseshell and calico are often used to refer to the phenotype, the gene, and/or the allele in both cats and hamsters; here we use Orange, O, to describe the gene in cats and Robinson's original designation, Sex-linked yellow, Sly, to describe the gene in hamsters.)

More than 100 mouse coat color mutations have been characterized over the past century, and multiple alleles exist for those directly implicated in pigment-type switching (Agouti, Mc1r, Atrn, Mgrn, Sox18), none of which lie on the X chromosome (Bennett and Lamoreux 2003). The apparent absence from laboratory mice of an X-linked pigment-type switching mutation is perplexing in the face of Ohno's law, which predicts conservation of synteny on the X chromosome across different mammalian species due to selection against altered gene dosage (Ohno 1969). Genetic characterization of Sex-linked yellow in Syrian hamsters could shed light on this paradox and identify additional components of the pigment-type switching pathway. However, by contrast to laboratory mice, very little of the classical work on coat color variation in Syrian hamsters (Nixon et al. 1970) has been developed at the molecular genetic level.

We have established a laboratory-based colony of Syrian hamsters to study the biology and genetics of Sex-linked yellow together with other coat color variants and confirmed that the mutation behaves similarly to what was originally reported (Robinson 1966, 1972). Here we use a comparative genomic approach to generate a molecular genetic map of the Syrian hamster X chromosome that includes Sex-linked yellow, which provides a basis for molecular genetic and epistasis studies revealing that Sly represents a component of the melanocortin pathway similar to, but independent of, the Mc1r. Comparison to analogous work carried out for Orange in domestic cats (Schmidt-Küntzel et al. 2009, accompanying article, this issue) indicates that the cat and hamster mutations likely lie in nonhomologous regions of the X chromosome, which has implications for understanding both evolutionary history and the mutational mechanisms of pigment-type switching.

MATERIALS AND METHODS

Animal husbandry:

Sly and other coat color variants in the Syrian hamster are well recognized in the hobbyist community, but are not maintained in an academic- or a research-oriented setting; none of the laboratory animal vendors we contacted were aware of a sex-linked and/or variegated phenotype. We obtained animals with a diverse set of coat color phenotypes from San Joaquin Valley Fisheries (Fresno, CA), including belted, black, cream, golden, red-eyed dilute, tortoiseshell, and tricolor (analogous to calico in the domestic cat). For phenotypic characterization and linkage mapping of Sly, we used tortoiseshell females and black males and confirmed that tortoiseshell segregated as an X-linked trait in accord with Mendelian expectations as indicated in Figures 1 and 2 and Table 2. All animal work was carried out under an Asia Pacific Laboratory Accreditation Cooperation-approved protocol.

Figure 1.—

Pigmentary phenotypes of hamsters with different Agouti and Sly genotypes. (A) Yellow, black, tortoiseshell, and golden animals as described in the text. Genotypes are inferred from X-linked inheritance of the tortoiseshell phenotype and from molecular genetic analysis of Agouti as described in Figure 4. (B) Representative zigzag hairs from golden (Sly⁺/Sly⁺; A^W/A^W) and yellow (Sly^To/Y; a/a) animals. Top panels show ∼80% of the hair; the lower 20% of a “golden” hair is black, and the lower 20% of a “yellow” hair is pale yellow with a white base. (The dimensions of the hair are such that the shaft is difficult to see in photos that contain the entire length of the hair.) Dashed box indicates region of the hairs shown at higher magnification in the bottom panels and illustrates that the tip of the yellow hair contains a mixture of yellow and black pigment, while the shaft of the yellow hair is pale relative to the corresponding region of a golden hair.

Figure 2.—

Transmission and linkage of Sly. (A) Structure of pedigrees used for segregation and linkage analysis. Shaded, hatched, and yellow symbols represent black, tortoiseshell, and yellow animals, respectively, with associated Sly genotype inferred on the basis of coat color phenotype and sex. As described in the text, 17 such pedigrees yielded 155 F₃ progeny, who were then genotyped for the molecular markers as indicated. (B and C) Haplotype segregation diagram, with gray or yellow indicating whether the chromosome of origin carried Sly⁺ or Sly^To, respectively, and haplotypes organized into single (B) and multiple (C) recombinants. Gene order is based on multiple-regression analysis of intermarker distances; minimizing the number of double crossovers yields an alternative order for the first three markers, DXBar112–DXBar52–DXBar51. Recombination frequency (RF) in centimorgans between each marker and Sly is given on the right together with the number of informative meioses.

TABLE 2.

Phenotype of F₃ progeny from black × tortoiseshell intercross

Phenotype	Sex	No. progeny
Black	Male	39
Yellow	Male	33
Tortoiseshell	Male	0
Black	Female	46
Yellow	Female	0
Tortoiseshell	Female	37

Based on cross as depicted in Figure 2A and described in the text.

Genetic markers:

We first developed simple sequence length polymorphism (SSLP) markers using a comparative genomic strategy in which aligned regions of the mouse and human X chromosomes were used to identify potential SSLP targets for PCR amplification using Primer 3 (Rozen and Skaletsky 2000). Precomputed alignments of the human and mouse genomes were obtained from ftp://hgdownload.cse.ucsc.edu/goldenPath/hg16/vsMm4/axtTight/. From 39,889 regions of human–mouse X chromosome alignments, we used a custom perl script to identify 7204 pairs of alignments in which each pair was separated by a 100- to 300-bp gap (in which no alignment was recorded and therefore may have contained an SSLP). We filtered the 7204 pairs of alignments by the extent of similarity, allowing only those alignments in which there was a perfect human–mouse match of at least 20 bp, yielding 943 alignment pairs. Among these, suitable primers (that met default criteria for Primer3) could be designed for 452. Among the 452 alignment pairs, the intervening (to be amplified) sequence contained one or more SSLP targets in both the mouse and the rat genomic sequence in ∼30 alignments, all of which were used for PCR of hamster genomic DNA. This approach yielded seven amplicons that contained polymorphic SSLPs; we then designed new internal primers directly from the hamster sequence for subsequent amplification of hamster genomic DNA.

We supplemented the SSLP framework map with SNP-based markers ascertained by resequencing introns or between pairs of conserved noncoding sequences. For resequencing introns, exon-based primer pairs were chosen using a perl script designed to target oligonucleotides corresponding to little or no codon degeneracy (and therefore likely to be conserved between mouse and hamster). For resequencing between pairs of conserved noncoding sequences, we modified the strategy described above to start with precomputed mouse–human alignments (Kent et al. 2003; Brudno et al. 2004). From ∼300 amplicons from hamster DNA, we identified nine SNPs used to supplement the SSLP-based linkage map. For seven of the nine SNP markers, primers based on mouse sequence yielded robust and specific amplification products from hamster genomic DNA; in two cases (DBarX111 and DBarX118) we developed new internal primers directly from the hamster sequence.

Identity of the amplicons for all 16 primer pairs was investigated by aligning the amplified hamster sequence to the predicted mouse amplicon. In four cases (DXBar53, DXBar 54, DXBar DXBar 57, and DXBar 119), no significant similarity was observed between mouse and hamster amplicons, probably due to internal repetitive sequence and/or lack of evolutionary constraint; therefore the identity of these amplicons relies on comparative mapping results (Figure 3).

Figure 3.—

Comparative X chromosome maps for mouse, hamster, human, dog, and rat. (A) Physical position (rounded to the nearest megabase) is given for mouse, human, and dog markers, and the location of the pseudoautosomal regions (PAR or PAR1) is indicated for mouse and human. As described in the text, the hamster map refers to Xp (the long arm is heterochromatic) with the position of the centromere based on comparison to molecular cytogenetic results (Kuroiwa et al. 2001a,b,c). A homology block on human (46–54 Mb) and dog (40–45 Mb) Xp corresponds to the position of Sly in the hamster. (B) Position of markers on the hamster X chromosome relative to the rat.

Molecular genetic and linkage analysis:

SSLP-based markers were amplified with dye-labeled primers; SNP-based markers were sequenced with dye terminators after amplification; all markers were separated on a capillary instrument equipped for fluorescence detection. Genotypes for SSLP markers were called manually; genotypes for SNPs were called with an automated commercial detection platform (CodonCode). Prior to linkage analysis, Mendelian error checking was performed; apparent instances of non-Mendelian transmission were inspected for potential genotype errors and either corrected or dropped from the analysis. An initial map order was established by minimizing the number of double crossovers and subsequently refined with a regression-based analysis for intermarker distance implemented in JoinMap (Stam 1993).

Molecular cytogenetics:

Fluorescence in situ hybridization (FISH) of bacterial artificial chromosome (BAC) probes to metaphase chromosomes from cultured mouse and hamster fibroblast cells was carried out as described previously (Raudsepp and Chowdhary 2008). We selected three mouse BACs (RP24-77C5, RP24-186P1, and RP24-96H13) on the basis of their gene content as inferred from the July 2007 (Build 37) assembly on the UCSC genome browser, http://genome.cse.ucsc.edu/ (Karolchik et al. 2003). Each BAC contains one or more genes that are found in segments whose order is physically conserved among mice, humans, dogs, and cats; the genes (and BACs) were chosen to lie close to, and possibly flanking, the cat Orange gene as depicted in Figure 5. BACs were labeled by nick translation with digoxigenin (77C5 and 196H13)- or biotin (186P1)-coupled UTP, hybridized to nuclear chromatin in pairwise combinations (77C5 + 186P1 or 196H13 + 186P1), and detected with indirect immunofluorescence (Raudsepp and Chowdhary 2008). For cross-species detection (mouse probes hybridized to hamster chromosomes), probe concentration was increased by ∼5- to 10-fold, and hybridization was carried out for 72 rather than 24 hr.

Figure 5.—

Comparative genomics of Sly and Orange. (A) Representative images of FISH for mouse BAC probes 1 and 2 hybridized to mouse or hamster chromosomes, as indicated. (BAC probes 1, 2, and 3 are, respectively, RP24-77C5, RP24-186P1, and RP24-96H13.) (B) Diagram of BAC hybridization results showing that the region homologous to the cat Orange gene lies in the centromere-proximal region of the mouse X chromosome but the centromere-distal region of the hamster X chromosome. By contrast, genetic mapping studies place the hamster Sly gene close to the centromere of Xp. (C) Summary of comparative mapping. The position of the cat Orange mutation is based on work from Schmidt-Küntzel et al. (2009).

RESULTS

Sex-linked yellow in hamsters—phenotype and genetics:

Commercial suppliers of Syrian hamsters to research laboratories carry few, if any, coat color variants. We identified a major pet store supplier in the California central valley, imported several different coat color variants, and established standardized pedigrees from seven tortoiseshell females and six black males to confirm and characterize the phenotype and inheritance pattern. In what follows, we refer to mutant and nonmutant alleles for Sly as Sly^To and Sly⁺, respectively.

As described originally (Robinson 1966), mutant animals (Sly^To/Y or Sly^To/Sly^To) are orange–yellow in color with a darker dorsum than ventrum and a sooty appearance, whereas nonmutant animals (Sly⁺/Sly⁺ or Sly⁺/Y) are “golden,” presumably due to the presence of an Agouti allele that promotes an extended band of pheomelanin. However, in the hobbyist community today, Sly^To is almost always described on a presumptive nonagouti (a/a) background such that nonmutant animals (Sly⁺/Sly⁺ or Sly⁺/Y) are black.

We examined dorsal hair pigment distribution from yellow (Sly^To/Y; a/a), black (Sly⁺/Y; a/a), and golden (Sly⁺/Y; A^W/A^W) animals. As in the mouse (Sundberg 1994), the hamster coat contains three major types of hair: long guard hairs and shorter hairs, awls and zigzags, that constitute the underfur and whose relative proportions are similar among yellow, black, and golden hamsters (Table 1). Pigment-type switching phenotypes are most evident in the zigzag hairs, which compose the majority of the underfur. Hairs from golden animals exhibit a dark tip and base separated by a wide pheomelanic band that extends over ∼80% of the hair length (thereby accounting for a golden rather than a typical brushed Agouti appearance). Hairs from yellow animals exhibit a dark tip (but less dark than in golden animals), a pale yellow shaft (less intense than the pheomelanic band of golden hairs), and a white base (Figure 1).

TABLE 1.

Hair type distribution and pigmentary phenotypes

Animal	Guard hairs (%)	Awls/auchenes (%)	Zigzags (%)
Golden (AW/AW; Sly+/Sly+)	Black (2)	Black (4)	Banded (84)a
		Banded (8)
		White (2)
Yellow (a/a; SlyTo/Y)	Dark-tippedb (2)	Dark-tippedb (10)	Dark-tippedb (88)

^aBlack tip and base, wide yellow band.

^bDarkened tip (mixed pigment types), pale yellow shaft, white base.

To characterize the inheritance of Sly, we used a three-generation breeding scheme in which phase could be inferred unambiguously for X-linked markers; tortoiseshell females were always mated to black males, such that only a single Sly allele was segregating among F₃ progeny (Figure 2A). A total of 155 F₃ progeny were obtained from 17 pedigrees. X-linked inheritance of Sly was confirmed by the absence of yellow females in each pedigree and a distribution of F₃ progeny that did not deviate significantly from that expected for X linkage (Table 2). These observations confirmed that each kindred is segregating a single X-linked allele that results in constitutive pheomelanin production and provided a sufficient number of meiotic events to carry out a linkage scan of the hamster X chromosome.

Linkage scan of the X chromosome:

We used a combination of comparative genomic and PCR-based approaches to develop molecular genetic markers for the hamster X chromosome. Among all possible combinations of markers (seven SSLP markers and nine SNP markers) with 155 F₃ individuals, 1077 assays yielded informative results and were used to establish a genetic map (Figures 2B and 3, Table 3). The hamster X chromosome map covers 46 cM with a median intermarker distance of 4.1 cM; by comparison, the mouse X-specific region is 72 cM in length (Blake et al. 2003). The hamster map includes markers that map to both ends of the human X chromosome, but lacks markers corresponding to two key regions in mouse (0–20 Mb and 85–129 Mb) and one key region in human (54–98 Mb). Thus, we cannot determine whether the shorter genetic length of the hamster X chromosome relative to the mouse represents a reduced ratio of genetic/physical distance in the hamster relative to the mouse or a terminal region on the hamster X that has not been captured by our genetic markers.

TABLE 3.

Genetic markers

Marker name	Type	Primersa	Coordinatesb
DBarX111	2 (Rbm10)	GTCCTGAAAGGCCCGTGT	20212996–20214673
		GAACCAAAGGCCAAAACTGA
DBarX 51	1	GGAGGCTTAGCTTACCCCTTT	40781605–40782207
		ACGTGAAATTGTGTGCGTGT
DBarX 52	1	CACATGAGCTCTGTTGCATCT	53759564–53760159
		AAAGTCACCCTCATGGGAAG
DBarX 112	2 (Bgn)	GACCACAACAAAATCCAGGCa	70738334–70738684
		CTCTCACCTGGAGGAGCTTGa
DBarX 113	2 (Tbl1x)	GGACGCTCACACAGGAGAAGa	74895226–74895824
		GTCCTTGGAAGGTTTTGACTGa
DBarX 114	2 (CNS)	CCAGCTTTGGTCATTTGACAGa	84947982–84948629
		AGGGACTTGGAAAGCTCACAa
DBarX 53	1	AAGGCAATGGCTTCATTGTT	129415830–12941667
		TGCTTCTGACCTTTGAGCAG
DBarX 54	1	CCACCTCTGTGAGCACATTC	131291877–131292275
		TGTCTAAATGTGTTTCTCACACTCC
DBarX 55	1	ATGCACCCCTCCAGTTCTGC	136812895–136813493
		TGGATCTGCAGTAATGTGCAGT
DBarX 56	1	GAGCGTCCCGGGAGCTCCTT	147389767–147390410
		CCACCTCTTACAGGAAGCC
DBarX 115c	2 (CNS)	ATTGCCTGCCAAATCACACTa	156649400–156649723
		TGATTAAGTGACCCCAGAAATCa
DBarX 57	1	TCCATTGTTATTCAAGGAAGGAA	158286467–158286797
		GCTTTTGTGCAGATTCCCAG
DBarX 119c	2 (CNS)	AGGTGATGTCAGCGGCTCTa	158595286–158595666
		TCTTTCCAGAACTCATCCCCa
DBarX 116	2 (CNS)	GTCGATTAGCATTGGCATTTa	162665068–162665558
		AAAGCATAGCACACACAAGGAa
DBarX 118c	2 (Prps2)	GCTCATTGGTCAGCCAATCT	163801537–163803099
		CCATGTGCTTTGTGATTCCA
DBarX 117	2 (CNS)	GAACAGCAAGGAGGAAATTGa	164867514–164867994
		CCTGAAGCCTCCCACTCCa

Marker names correspond to those in Figures 2 and 3; types “1” and “2” refer to SSLP and SNP markers, respectively, the latter followed either by the gene name or conserved noncoding sequence (CNS) depending on how the primers were derived. Oligonucleotide primers were originally designed from mouse genome sequence, but unless otherwise stated (see footnote a), new primers were generated from the amplified hamster sequence and used for subsequent genotyping.

^aAll of the type 1 (SSLP) primers represent hamster sequence. However, for seven of the nine type 2 (SNP) primers (those indicated with a superscript a), oligonucleotides based on the mouse sequence yielded robust products when amplifying hamster genomic DNA, and we did not redesign new hamster primers.

^bPhysical position of amplicons on the mouse X chromosome based on coordinates from the July 2007 (build 37) release.

^cNot included in Figures 2 and 3 because they added no new map information. Genotypes from DXBar115 and DXBar119 were identical to those obtained with DXBar57; genotypes from DXBar118 were identical to those obtained with DXBar117.

Comparing the relative position of homologous markers between the mouse, hamster, human, and dog X chromosomes reveals a considerable degree of intrachromosomal rearrangement between hamsters and mice and between hamsters and humans or dogs (Figure 3A). However, the relative position of markers in the hamster is almost identical to that in the rat (Figure 3B), consistent with previous molecular cytogenetic work (Kuroiwa et al. 2001a,b,c), suggesting that gene order on the X chromosome is largely conserved between several rodent species (including the rat and the Syrian hamster), although not in the laboratory mouse.

The Syrian hamster X chromosome is metacentric, but with the long arm composed entirely of constitutive heterochromatin (DiPaolo and Popescu 1973); thus, our molecular map applies to hamster Xp. Comparison of our results to molecular cytogenetic studies suggests an orientation in which DXBar111 and DXBar51 lie most centromere proximal and centromere distal, respectively. Sly lies in a 9.6-cM region at the centromere-proximal end, flanked by DXBar111 and DXBar56 (Figure 3). Transposed against the human and dog X chromosomes, this region defines 8- and 5-Mb intervals, respectively, that also lie close to the human and dog centromeres.

Candidate genes for Sly—Agouti and Mc1r:

As indicated above, although Sly^To was originally described on an Agouti (golden) background, most animals carrying Sly^To are presumably nonagouti, such that Sly⁺/Y and Sly⁺/Sly⁺ animals display a black coat, whereas Sly⁺/Sly^To animals display patches of yellow hair on a black background, presumably in regions where melanocyte clones have undergone epigenetic inactivation of the wild-type X chromosome (Figure 1A). However, coat color genetics in hamsters are based almost entirely on the similarity of phenotypes and genetic interactions to those observed in laboratory mice, and there are at least two hamster genes that can yield recessive inheritance of a black coat (Kuramoto et al. 2002).

To investigate the underlying molecular basis for black coat color (and, indirectly, the epistasis relationship between black coat color and Sly), we used a PCR-based strategy to determine the coding sequence for the hamster Agouti gene. Within exon 4, which encodes the majority of the Agouti protein-coding region, we detected a G to A mutation that predicts a Cys115Tyr substitution (Figure 4A) found in both black hamsters and yellow hamsters in our colony, but not in golden hamsters. This residue is conserved among all known Agouti homologs and helps stabilize the key active loop responsible for melanocortin receptor antagonism (McNulty et al. 2005). Furthermore, a Cys115Ser mutation in a mouse Agouti transgene has the same effect as a null allele (Perry et al. 1996). Taken together, these results indicate that loss-of-function in Agouti is responsible for the black coat color in our colony.

Figure 4.—

Genetics and epistasis of Agouti and Sly. (A) As described in the text, black hamsters carry a missense mutation that predicts a Cys115Tyr substitution, predicted to disrupt Agouti function. (B) Sly is epistatic to Agouti because Sly^To/Sly^To; a/a animals exhibit the same phenotype as Sly^To/Sly^To; A^W/A^W animals.

Similar to what has been described for Orange in tortoiseshell and calico cats (Searle 1968), the patches of black and yellow pigment in Sly⁺/Sly^To hamsters become larger in the presence of white spotting mutations (Robinson 1972), suggesting that Sly^To acts in a melanocyte-autonomous manner, similar to Mc1r^e in laboratory mice. To investigate the possibility that Sly^To might represent an Mc1r loss-of-function allele that had been transposed to the hamster X chromosome, we used oligonucleotide primers based on the mouse and rat genome to PCR amplify the hamster Mc1r gene and flanking sequences from yellow and black hamsters. We identified several synonymous and/or flanking sequence SNPs that were heterozygous in at least one male animal (data not shown), but did not observe any alterations expected to impair Mc1r function; thus, as in all other mammals, including cats (Eizirik et al. 2003), the hamster Mc1r appears to be autosomal. We conclude that, similar to Orange in cats, Sly is epistatic to Agouti but genetically distinct from Mc1r (Figure 4B).

Relationship to Orange in domestic cats:

The puzzling similarities between cat Orange and hamster Sly—puzzling given the evolutionary distance between phyla and the absence of a similar mutation in any other mammal—prompted us to investigate whether the two mutations lie in homologous locations. In an accompanying article, Schmidt-Küntzel et al. (2009) demonstrate that Orange lies in a 6-cM, ∼11-Mb interval that corresponds to 96–106 Mb and 120–131 Mb in the dog and human genomes, respectively. By contrast, flanking markers for hamster Sly delineate a different interval, on the short arm of the dog (40–45 Mb) and human (46–54 Mb) X chromosomes (Figure 5C).

A potential caveat to the conclusion that the hamster Sly and cat Orange mutations lie in nonhomologous locations is that the large number of evolutionary breakpoints apparent from comparing the mouse and hamster X chromosomes to those of other mammals might mask one or more chromosomal rearrangements in which segments harboring the Orange gene were transferred to what we currently recognize as the Sly region. To investigate this possibility, we carried out cytogenetic FISH experiments to evaluate the location of the cat Orange interval in mouse and hamster chromosomes. We selected three BAC clones from a mouse genomic DNA library that each carried genes found in homologous locations in humans, dogs, and cats and whose positions were likely to flank the Orange mutation (Figure 5C). Each of the three BACs was hybridized to metaphase spreads from an X/X mouse, an Sly⁺/Y hamster, and an Sly^To/Y hamster. As expected, the cytogenetic locations for the three BACs on the mouse X chromosome correspond to their positions on the physical map, at ∼38 Mb (BAC1: 77C5), 39 Mb (BAC2: 186P1), and 47 Mb (BAC3: 196H13) (Figure 5A). Hybridization of the same BAC probes to hamster chromosomes yielded weaker signals, and it was not possible to order the probes in double-labeling experiments. However, all three probes hybridized to the same region at the telomeric end of hamster Xp for both the Sly⁺- and the Sly^To-bearing chromosomes (Figure 5, A and B). These results suggest that the X-chromosomal region that contains Orange (as defined by genetic mapping experiments carried out by Schmidt-Küntzel et al. 2009) remained intact during rodent evolution and is distinct from the region where Sly maps in hamsters.

DISCUSSION

In most respects, the effects of Sly in hamsters and Orange in domestic cats are similar to those caused by Mc1r loss-of-function mutations in other mammals. Both Sly and Orange yield a uniformly pheomelanic pelage whose effects are epistatic to those of nonagouti, as is the case for Mc1r loss-of-function in mice (Silvers 1979), horses (Marklund et al. 1996; Rieder et al. 2001), and dogs (Newton et al. 2000; Kerns et al. 2004). Furthermore, both Sly and Orange interact with white-spotting mutations in a manner that suggests they are melanocyte autonomous, as is the case for mosaic Mc1r mutations in pigs (Kijas et al. 2001) and mice (Lamoreux and Mayer 1975) and presumptive Mc1r mutations in guinea pigs and rabbits (Searle 1968). However, both Sly and Orange are genetically distinct from the Mc1r and possibly from each other. In what follows, we consider two explanations that could account for these observations.

Sly and Orange may be orthologs, representing loss-of-function mutations in the same (previously unrecognized) component of melanocortin signaling that acts in a manner and cell type similar to that of the Mc1r, but lies genetically upstream, such as a transcription factor required for Mc1r expression or a receptor-associated protein required for proper cell surface expression and/or targeting of the Mc1r. Absence of a similar X-linked phenotype in mammals other than hamsters and cats could be explained by redundancy, e.g., if the Sly/Orange gene was duplicated early in mammalian evolution, and the duplicated genes were redundant with regard to pigmentary function in most mammals. According to this scenario, most mammals would then have two Sly/Orange genes, but one of the duplicated genes would have been lost, by chance, from the hamster and, independently, from the cat lineage.

If Sly and Orange are orthologs, the apparent lack of “homology” in their respective chromosomal locations (homology as assessed with the comparative mapping approach in Figure 5) likely reflects additional chromosomal rearrangements during rodent evolution, in which the Orange gene in an ancestral mammal translocated to the centromere-proximal region of hamster Xp. Although the FISH-based cytogenetics do not support such a rearrangement, the sequences tested (three BACs in an ∼10-Mb region) represent only a fraction of the interval and cannot exclude the possibility of small chromosomal rearrangements that will become apparent only from additional genome sequences.

Alternatively, Sly and Orange could represent unusual gain-of-function mutations in which a (perhaps previously known) component of melanocortin signaling has been activated in melanocytes. For example, retrotransposition of Agouti into a melanocyte-specific locus on the X chromosome or a point mutation of a melanocyte-specific X-linked Gi-coupled receptor that causes constitutive activation would be expected to yield a phenotype and set of genetic interactions similar to those observed for Sly and Orange. Both sorts of events would be very rare—much less frequent than loss-of-function mutations—and therefore might explain why X-linked pheomelanism has not been observed in species where very large numbers of animals have been subjected to a natural screen, i.e., laboratory mice and humans. This idea is also consistent with preliminary analyses of the cat and dog X chromosome sequences, in which the dog region homologous to that which carries Orange also harbors several pseudogenes whose origin is autosomal.

A consideration of Sly^To candidate genes must also account for the epistatic relationship between Sly and Agouti; for example, Sly is unlikely to represent a gene such as Sox18, which acts in the dermal papilla to promote Agouti expression (Pennisi et al. 2000; Fitch et al. 2003). From this perspective, genetic interactions between Mc1r and Sly would also be helpful in evaluating potential candidate genes. Variation at Mc1r is curiously absent from the cat, but prospects for determining whether Sly lies upstream, downstream, or parallel to Mc1r in the hamster are encouraging, since recessively inherited cream coat color in the hamster is thought to represent the action of Mc1r (Magalhaes 1954; Robinson 1964).

These questions can be resolved, of course, by molecular identification of the Sly and Orange genes and should be facilitated by availability of additional mammalian genome sequences. Defining the molecular genetics of Sly and Orange may also be useful as a tool for understanding the biology of other color variation patterns thought to affect the distribution of pheomelanin and eumelanin, such as tabby striping in domestic and wild cats, zebra striping in horse × zebra hybrids, and rostro-caudal striping seen in squirrels and/or chipmunks (Searle 1968).