Shades of meaning: the pigment-type switching system as a tool for discovery


Teresa M. Gunn, e-mail:


The pigment-type switching system, which controls whether melanocytes produce black/brown eumelanin or yellow/red pheomelanin, is responsible for many familiar coat coloration patterns in both domestic and wild mammals. In conjunction with the accessory proteins attractin and mahogunin ring finger 1, endogenous agonists and antagonists modulate signaling by the melanocortin 1 receptor to determine pigment type. Mutations in pigment-type switching genes can cause a variety of pleiotropic phenotypes, and these are often similar between mutants at different loci because the proteins encoded by these genes act together as part of conserved molecular pathways that are deployed in multiple biological contexts. When this is the case, pigment-type switching provides a powerful model system for elucidating the shared molecular mechanisms underlying the pigmentary and non-pigmentary phenotypes. This review outlines the current understanding of the pigment-type switching pathway and discusses the opportunities that exist for exploring the molecular basis of pleiotropic phenotypes using this model system.


From the rusty red of the Irish setter to the intense black of the Angus steer, the coat colors of mammals exhibit an arresting and beautiful diversity. Mammals can produce two types of pigment: a dark brown or black form referred to as eumelanin, and yellow/red pheomelanin. The pigment-type switching pathway controls which type of melanin is produced and allelic variation in pigment-type switching genes underlies many familiar mammalian coat color phenotypes. Although informative mutations are found in many species, the coat color biology of the laboratory mouse provides a good introduction to the system. In the dorsal hair follicles of mice, melanocytes switch production from eumelanin to pheomelanin at an early point in the hair growth cycle, producing a subapical yellow band on an otherwise eumelanic hair (Cleffmann, 1963). This gives rise to the so-called ‘agouti’ phenotype of wild-type mice (Figure 1A). Analysis of mice with deviations from this pattern has led to the identification of the genes and molecular pathways underlying pigment-type switching, which will be the focus of this review. Mutations that affect pigment-type production also underlie many of the interesting coat colors of domesticated animals such as dogs, cows, sheep, horses, pigs, and chickens (Berryere et al., 2005; Kerje et al., 2003; Kerns et al., 2004; Klungland and Vage, 2003; Marklund et al., 1996). In fact, both natural and artificial selective pressures seem to have found a rich source of exploitable variation in the pigment-type switching system, making this pathway an important model for evolutionary biology. For example, variations in genes in this pathway have been shown to underlie the evolution of locally adapted cryptic coloration patterns in several populations of wild mice (Hoekstra et al., 2006; Kingsley et al., 2009; Linnen et al., 2009; Nachman et al., 2003) and the spread of black coloration in North American wolves (Anderson et al., 2009). While humans and the other hominids do not produce agouti-banded hairs, pigment-type switching genes are highly conserved and some of them contribute to human pigmentation.

Figure 1.

 Examples of pigment-type switching phenotypes. (A) Transient production of yellow/red pheomelanin in the dorsal hair follicles results in a yellow band on an otherwise black hair, producing the agouti phenotype of wild-type mice. (B) Loss of pheomelanin production causes dark fur in mice with gain-of-function mutations in the melanocortin receptor 1 (Mc1r) or loss-of-function mutations in agouti or the accessory loci attractin (Atrn) or mahogunin ring finger 1 (Mgrn1). (C) Constitutive pheomelanin production results in yellow fur in mice with loss-of-function mutations in Mc1r or gain-of-function mutations in agouti.

A recurring theme of this review will be the usefulness of the pigment-type switching system as a model for providing insight into a wide range of pleiotropic phenotypes. Because coat color mutants often have additional phenotypes, investigation into the cellular function of the mutated genes often sheds light on biological mechanisms that regulate development and disease. These investigations are greatly facilitated by the power of the mouse coat color model system (Bennett and Lamoreux, 2003). Like coat color mutants in general, pigment-type switching mutants present a diverse range of pleiotropic phenotypes including obesity, neurodegeneration, altered pain perception, mitochondrial dysfunction, and embryonic patterning defects (Bronson et al., 2001; Castle, 1941; Cota et al., 2006; He et al., 2003; Mogil et al., 2005; Sun et al., 2007). Studying the function of these genes using pigmentation as a model system can therefore provide a useful starting point for exploring the basis of superficially dissimilar but mechanistically related phenotypes. The remainder of this review will introduce the key players of the pigment-type switching pathway and discuss the roles they play (in and out of the pigment cell) and the rich prospects for future investigation using this powerful model system.

The melanocortin-1 receptor and its ligands

The master regulator of pigment-type switching is the melanocortin 1 receptor (MC1R), a G-protein-coupled receptor (GPCR) expressed in melanocytes. In its active state, the MC1R signals through the stimulatory G-protein α subunit (Gαs) to activate adenylyl cyclase, raising levels of the second messenger cAMP (Figure 2), thereby activating a series of transcriptional events that support the production of eumelanin (discussed below). When MC1R is in its inactive state, cAMP levels are low and pheomelanin is produced. By switching from its active state to its inactive state, the MC1R directs pigment-type switching. The phenotypes of Mc1r mutants illustrate the role of this receptor in pigment-type switching: mice carrying the continuously active somber allele (Mc1rE−so) have black coats (Figure 1B), whereas mice homozygous for an inactive truncation mutation (extension, Mc1re) are yellow (Figure 1C) (Robbins et al., 1993). Similarly, loss-of-function Mc1r variants are associated with red hair in humans (Box et al., 1997; Valverde et al., 1995; for a review of pigmentation genetics in humans see Rees, 2003). The switch between the eumelanogenic and pheomelanogenic signaling states of the murine MC1R is normally mediated by the opposing effects of a pair of ligands: α-melanocyte signaling hormone (α-MSH) and agouti signaling protein (ASP) (Figure 2). Much of the fascination of the pigment-type switching system lies in the complex way these ligands modulate MC1R function.

Figure 2.

 A model of the pigment-type switching pathway. Left: α-melanocyte stimulating hormone enhances signaling through the melanocortin 1 receptor (MC1R), leading to GDP/GTP exchange by the stimulatory G-protein subunit α (Gαs) and liberation of GTP-bound Gαs from the G-protein heterotrimer. Active, GTP-bound Gαs stimulates adenylyl cyclase (AC) to promote eumelanogenesis. Center: The MC1R inverse agonist agouti signaling protein (ASP) binds to MC1R and the accessory receptor, attractin (ATRN), to inhibit Gαs activation and reduce signaling through AC, leading to pheomelanogenesis. Mahogunin ring finger 1 interacts with MC1R and is required for ASP signaling. Right: Canine β-defensin 103 (β-D) can bind MC1R to inhibit ASP binding, promoting eumelanogenesis.

The primary agonist of the MC1R, α-MSH, is produced in the skin by keratinocytes and melanocytes (Chakraborty et al., 1996; Slominski et al., 2000). It increases MC1R signaling through cAMP and PKA (Cone et al., 1993; Garcia-Borron et al., 2005) to activate genes required for pigment production (e.g., Tyr, Tyrp1, Dct) and melanosome biogenesis (Pmel117) and transport (Rab27), largely through an effect on the master melanogenic regulator, microphthalmia-associated transcription factor (MITF) (Aberdam et al., 1998; Bertolotto et al., 1998; Le Pape et al., 2009; Levy et al., 2006). This effect of α-MSH is reinforced by a positive feedback effect whereby α-MSH stimulus causes increased rates of translation of MC1R from a variant transcript (Rouzaud et al., 2003). The result is that melanocytes treated with α-MSH become more dendritic and produce more eumelanin (Le Pape et al., 2008; Sakai et al., 1997). In humans, α-MSH is a critical promoter of MC1R function in pigment-type specification: rare mutations that result in loss of the α-MSH propeptide, proopiomelanocortin (POMC), cause red hair (Krude et al., 1998). In mice, MC1R exhibits a high level of constitutive activity, although administration of α-MSH does increase MC1R signaling (Cone et al., 1993; Jackson et al., 2007; Sanchez-Mas et al., 2004). This high constitutive activity of mouse MC1R appears to be sufficient to support eumelanin production, as indicated by the fact that the coat hairs of mice lacking POMC are indistinguishable from (on a non-agouti background) or only subtly yellower than (on an agouti background) those of Pomc+ mice (Jackson et al., 2007; Slominski et al., 2005; Yaswen et al., 1999). Pigment-type switching in mice thus appears to require a reduction of MC1R signaling below baseline levels, which normally occurs when the MC1R binds its second major ligand, ASP.

In mice, ASP is secreted by cells of the dermal papilla (a structure at the base of the hair follicle) during days 2–7 of the first hair growth cycle, partly as a response to bone morphogenetic protein signaling (Bultman et al., 1992; Millar et al., 1995; Sharov et al., 2005). Agouti signaling protein acts at several levels to inhibit eumelanogenesis and to promote pheomelanogenesis. First, it is a competitive antagonist which inhibits the pro-eumelanogenic effect of α-MSH by competing with it for binding of the MC1R (Figure 2) (Blanchard et al., 1995; Lu et al., 1994). More importantly for mouse pigmentation, ASP acts as an inverse agonist of the MC1R (Hunt and Thody, 1995; Ollmann et al., 1998; Sakai et al., 1997; Siegrist et al., 1996, 1997), binding the receptor not only to competitively inhibit α-MSH-induced signaling, but also to reduce the intrinsic AC-coupled signaling of MC1R to below constitutive levels. It is not clear whether the mechanistic basis of this inverse agonism is simply the stabilization of the receptor’s inactive conformation or a more complex effect of alterations in MC1R trafficking, turnover, or post-translational modification. The Mc1r Somber mutations ESo−J and ESo−3J are especially interesting in light of this question, as the mutant MC1R proteins they encode appear to be largely resistant to the effects of ASP. It has been proposed that the glutamine (92) residue (which is mutated in the ESo−3J allele) could function as part of a salt bridge for stabilization of the inactive conformation (Robbins et al., 1993). In addition to affecting the activity of MC1R proteins it binds, ASP also causes a reduction in the rate of new MC1R biosynthesis (Rouzaud et al., 2003), further depressing signaling. Downstream of its direct effects on the receptor, ASP causes a variety of transcriptional changes in the melanocyte, which have been investigated recently using cDNA microarrays. Le Pape et al. (2009) demonstrated that ASP blocks the transcriptional effects of α-MSH and activates genes involved in morphogenesis and cell adhesion to cause a melanoblast-like appearance of the treated cells, suggesting that ASP promotes a functional de-differentiation of melanocytes. Agouti signaling protein may be more than just an inverse agonist of AC-coupled MC1R signaling as, counterintuitively, AY/A mice on a wild-type Mc1r background are slightly lighter in color than AY/A mice lacking a functional MC1R. This suggests that at least some of the effects of ASP involve a non-canonical pathway of signaling through the MC1R. In support of this possibility, it has recently been reported that some of the effects of ASP on melanocyte morphology are MC1R-dependent, but cAMP-independent (Hida et al., 2009).

Genetic analysis of ASP function has been furthered by a variety of interesting mutant alleles. The fur of homozygotes for a null mutation (ae, extreme non-agouti) is completely black (Figure 1B), reflecting continuous signaling through the MC1R. The pigmentary role of Mc1r is epistatic to that of agouti as mice with loss-of function mutations for both Mc1r and agouti exhibit the yellow Mc1re coat color phenotype. Conversely, mice heterozygous for the lethal yellow (Ay) allele of agouti express ASP ubiquitously and are yellow, similar to Mc1r loss-of-function extension mutants (Figure 1C), reflecting continuous inverse agonism of the MC1R. The A allele causes the classic agouti banding pattern on hairs across the entire pelage. The Aw allele produces dorsal agouti-banded hairs and yellow belly hairs, the latter because of the use of an additional, ventral-specific promoter that drives constitutive expression of ASP in the ventrum (Millar et al., 1995; Vrieling et al., 1994). This counter-shaded coloration pattern is widely found throughout the Mammalia, suggesting that Aw is the ancestral wild-type allele. The viable yellow agouti allele, Avy, provides an interesting model of epigenetic inheritance. The coat color of mice carrying this allele can range from agouti to completely yellow, depending on the methylation status and resulting transcriptional activity of an intracisternal A particle inserted upstream of the agouti locus. The distribution of coat color phenotypes in a litter of Avy pups is related to the phenotype of their mother (for example, yellow mothers produce more yellow pups) and the methyl donor content of the maternal diet (Cropley et al., 2006; Duhl et al., 1994a; Morgan et al., 1999). In addition to this well-characterized range of mutant agouti alleles in mice, variant agouti alleles cause distinct pigmentation phenotypes in other animals (Klungland and Vage, 2003).

In humans, the role of the ASP orthologue, ASIP, is not well understood. While the agouti banding pattern is not seen in the hair of humans or other hominids, variations in ASIP have been associated with both hair and skin pigmentation traits in human populations (Bonilla et al., 2005; Kanetsky et al., 2002; Sulem et al., 2008; Voisey et al., 2006). There seems to be a conserved role for ASP signaling in determining murine skin pigmentation as well: the glabrous (nonhairy) skin of agouti null (ae) mutant mice is darker than wild type (Hollander and Gowen,1956), whereas Mc1re mutants have a reduction of eumelanin in tail skin (Van Raamsdonk et al., 2009). Interestingly, the decrease in dermal and epidermal eumelanin in Mc1re mutant mice is not accompanied by an increase in pheomelanin content suggesting that, unlike the situation in the hair follicle, dermal and epidermal eumelanin and pheomelanin synthesis may not be tightly coupled (Van Raamsdonk et al., 2009). Because this analysis used only non-agouti mice, it is not known whether this decoupled situation also occurs in wild-type skin.

Recently, Barsh and colleagues discovered a third ligand of the MC1R by mapping the ‘K’ locus that determines dominant black coat color in dogs (Candille et al., 2007). Surprisingly, it turned out to be the gene that encodes β-defensin 103, a member of an ancient family of small antibacterial peptides best known for their role in the innate immune system. While no murine β-defensin alleles are known to cause pigmentation phenotypes, transgenic mice expressing either the black (KB) or the yellow (ky) canine allele had predominantly black coats with small patches of agouti-banded hairs. Biochemical investigations showed that β-defensin, like ASP, can bind MC1R to compete against α-MSH binding, without activating the cAMP pathway (Figure 2). This suggests that β-defensin’s mode of action in the pigment cell may involve exclusion of ASP from the binding site or interference with the activity of the ASP co-receptor, attractin (described below). Interestingly, the canine β-defensin alleles also caused reduced body size in transgenic mice. This is an opposite effect from the increased linear growth demonstrated by agouti overexpressing mice (discussed in a later section), suggesting that β-defensins may also be involved in regulation of melanocortin signaling outside the pigmentary system. Consistent with this possibility, the product of the KBβ-defensin 103 mutant allele showed an intermediate affinity for the mouse melanocortin 4 receptor (MC4R), which is involved in energy homeostasis and growth regulation.

Accessory proteins for pigment-type switching: ATRN and MGRN1

Two spontaneous mouse mutants that arose in the 1960s identified additional loci required for transduction of the ASP signal (Lane, 1960; Lane and Green, 1960; Phillips, 1963). Genetic analysis demonstrated that loss-of-function mutations at either of these loci, originally called mahogany (mg) and mahoganoid (md), resulted in dark fur in homozygotes, even in the presence of the ubiquitously overexpressed Ay allele of agouti. Atrn and Mgrn1 null mutant mice also have darkly pigmented glabrous skin, similar to that of ae mutants. The phenotypic expression of either mutation depends upon Mc1r; however, in that Mc1re/e mice are yellow regardless of their mahogany or mahoganoid genotype (Miller et al., 1997). The similar phenotype and identical epistatic placement of these genes in agouti-melanocortin receptor signaling strongly suggest that their products work together to achieve a particular step in the pigment-type switching pathway.

The mahogany locus was identified by positional cloning (Gunn et al., 1999; Nagle et al., 1999) as the mouse orthologue of the human attractin (ATRN) gene, which was named for its role in mediating clustering of immune cells (Duke-Cohan et al., 1998). Murine Atrn encodes a type I single-pass transmembrane protein with a predicted molecular weight of ∼160 kDa. Its large extracellular/lumenal domain contains a CUB domain, a C-type lectin domain, a predicted kelch propeller, several EGF-like repeats, and several plexin/semaphorin/integrin motifs. The intracellular/cytoplasmic tail of ATRN is short compared to the rest of the protein (128 amino acid residues out of 1428) and contains no named domains. Expression of full-length wild-type mouse Atrn under the control of melanocyte- or keratinocyte-specific promoters demonstrated that ATRN is required in melanocytes to rescue the pigment-type switching phenotype of Atrn null mutant mice (He et al., 2001). This, along with its transmembrane topology, suggested that ATRN might serve as an accessory receptor for ASP (Figure 2). Surface plasmon resonance demonstrated an interaction between these two proteins (He et al., 2001) and, as expected for an accessory receptor, ATRN bound a separate region of ASP from that which binds MC1R (Ollmann and Barsh, 1999; Willard et al., 1995).

While a physical association between ATRN and MC1R has not been reported to date, their paralogues, attractin-like 1 (ATRNL1) and the MC4R, have been shown to interact with each other physically (Haqq et al., 2003) and both ASP and its paralogue, agouti-related protein (AGRP), can bind MC4R (Chai et al., 2005). These observations suggest that ATRN and MC1R are likely to be interacting partners. If, however, the function of ATRN is simply to stabilize the association of MC1R and ASP (via a relatively weak interaction between ASP and ATRN), it is difficult to explain the strong dependence of pigment-type switching on the presence of ATRN and why overexpression of ASP does not compensate for loss of ATRN. The genetic interactions of Atrn, Mc1r, and agouti suggest that the effect on MC1R of binding both ASP and ATRN is qualitatively different from the effect of binding ASP alone. It is noteworthy in this respect that Atrn null mutant melanocytes have a normal cAMP response to ASP treatment but exhibit an impairment of the cAMP-independent ASP-induced morphological changes reported by Hida et al. (2009). These morphological changes also require the N-terminal portion of ASP, which binds ATRN. Taken together, these observations suggest that ATRN may be a part of a non-canonical mechanism of ASP signaling through the MC1R. The existence of such a mechanism would help explain the curious observation that agouti overexpressing mice are actually slightly darker in the absence of functional MC1R (Jackson et al., 2007; Ollmann et al., 1998).

The second major accessory locus required for pigment-type switching, mahoganoid, encodes a novel RING domain–containing protein (He et al., 2003; Phan et al., 2002). Now referred to as Mahogunin ring finger 1 (Mgrn1), its gene product belongs to the large class of E3 ubiquitin ligase proteins, which serve as specificity factors in catalyzing the transfer of the multifunctional ubiquitin tag onto target proteins, as either single units or polyubiquitin chains (reviewed by Deshaies and Joazeiro, 2009). The consequences of ubiquitination are diverse and include targeting proteins for proteasomal degradation or trafficking through the endo-lysosomal protein degradation pathway. Ubiquitination can also alter the conformation or accessibility of sequences in target molecules to mediate changes in active state, subcellular localization, and/or participation in protein complex formation (for recent reviews see Acconcia et al., 2009; Ardley and Robinson, 2005; Chen and Sun, 2009; Li and Ye, 2008; Nandi et al., 2006; Wickliffe et al., 2009). This diversity of effects makes E3 ligases important players in development, cell division, cell signaling, and disease pathways.

Given the many roles of E3 ubiquitin ligases, exactly how Mgrn1 mediates pigment-type switching has remained a fertile field for speculation. A long-standing hypothesis is that MGRN1 may ubiquitinate MC1R to direct its internalization and/or lysosomal degradation. This hypothesis is compatible with the observation of Rouzaud et al. (2003) that ASP treatment reduces MC1R protein levels in cultured melanocytes. A strong link between MGRN1 and the endo-lysosomal trafficking pathway has emerged from identification of the tumor susceptibility gene 101 product (TSG101) as a target of MGRN1-mediated ubiquitination (Jiao et al., 2009a; Kim et al., 2007). Tumor susceptibility gene 101 product is a member of the Endosomal Complex Required for Trafficking I (ESCRT-I) group of proteins, which act at the surface of incipient multivesicular bodies (MVBs) to sort monoubiquitinated transmembrane proteins into the lumenal vesicles of the MVB (reviewed by Saksena et al., 2007). The contents of the lumenal vesicles are subsequently transported to the lysosome for degradation. MGRN1-dependent ubiquitination of TSG101 appears to be important for its function: upon depletion of MGRN1 in HeLa cells, TSG101-mediated lysosomal trafficking of the epidermal growth factor receptor was impaired (Kim et al., 2007). In the mouse brain, Mgrn1 deficiency led to the accumulation of insoluble, multiubiquitinated TSG101 (Jiao et al., 2009a). Whether MGRN1 acts through TSG101 in the melanocyte to promote pigment-type switching remains an open question. Because TSG101 knockout mice are inviable as homozygotes (Wagner et al., 2003), the consequences of loss of TSG101 function on pigment production are not known. If pigment-type switching does require a MGRN1-dependent function of TSG101, this would suggest that ASP-mediated inverse agonism of MC1R requires lysosomal trafficking, most likely of the MC1R itself.

While hypotheses linking the role of MGRN1 in pigment-type switching to lysosomal trafficking are appealing, they have been challenged by a line of evidence suggesting that MGRN1 may directly disrupt MC1R signaling by uncoupling the receptor from its G proteins. Perez-Oliva et al. (2009) recently reported that, in a heterologous cell system, transient expression of MGRN1 significantly decreased signaling by MC1R without ubiquitinating it, altering its cell surface expression, or changing its rate of turnover. Further experimentation established that MGRN1 binds the MC1R and that this interaction decreases the interaction of MC1R with the Gαs, thereby uncoupling the receptor from its downstream effector, adenylyl cyclase (Figure 2). Overexpression of Gαs counteracted the effect of MGRN1 on MC1R signaling, suggesting that MGRN1 and Gαs may compete for a binding site on MC1R, although these data are also compatible with MGRN1 stabilizing the MC1R in an inactive conformation or targeting MC1R to an intracellular location less conducive to Gαs interaction. Interestingly, interaction of MC1R with MGRN did not result in the ubiquitination of the receptor. While the use of overexpressed proteins in a heterologous cell type introduces unavoidable caveats to the interpretation of their data (discussed in Walker and Gunn, 2010), these results introduce the exciting possibility that pigment-type switching may involve a novel form of ubiquitin ligase-mediated GPCR regulation at the level of Gαs binding.

Pigment-type switching: a powerful model system for exploring pleiotropy

Pigment-type switching has served as an important model of gene action and interaction for over a century. Genetic studies have established MC1R, its ligands, and the accessory proteins ATRN and MGRN1 as the basic components of a convenient model system for studying GPCR regulation. The complex pharmacology and signaling biology of this system can be manipulated to explore cell differentiation in the melanocyte and are of direct medical relevance because of the importance of MC1R signaling in the biology of melanoma. Add to this the intrinsic interest of understanding the origins of the beautiful and varied coloration of wild and domestic animals, as well as the recently demonstrated value of pigment-type switching as a foundation for field work in evolutionary biology (Anderson et al., 2009; Hoekstra et al., 2006; Linnen et al., 2009), and it becomes clear that this is a valuable model system. Furthermore, mice with defects in pigment-type switching often have additional phenotypes that promise insight into developmental and pathological pathways. The remainder of this article will focus on the valuable opportunities (past and present) the pigment-type switching system provide for investigating the molecular basis of these pleiotropic phenotypes.

From coat color to energy homeostasis

The lethal yellow (Ay) mouse was first described by Cuenot (1905). In addition to yellow fur, these mice have a metabolic phenotype that includes a longer body, juvenile-onset obesity (Castle, 1941), and hyperinsulinemia (Reviewed in Miltenberger et al., 1997). They became a well-studied model of obesity and it was thought that identification of the agouti gene would reveal a master regulator of body weight. The mutation in these mice, identified in the early 1990s, is a deletion that encompasses the agouti promoter, the Eif2s2 gene, and the coding region of the ubiquitously expressed upstream gene, Raly, placing agouti transcription under control of the Raly promoter (Duhl et al., 1994b; Heaney et al., 2009). While agouti is normally only transiently expressed during the midpoint of the hair growth cycle, the Ay mutation results in ubiquitous and continuous overexpression (Miller et al., 1993). Expression in the brain, particularly the hypothalamus, allows ASP to antagonize MC4R, which plays a central role in regulating feeding behavior and metabolism. As ASP is not normally present in the brain, this suggested that a homologous protein might normally regulate central melanocortin signaling. This led to discovery of the AGRP, an endogenous antagonist of MC4R. Agouti signaling protein and AGRP are structurally similar and transgenic mice ubiquitously expressing human Agrp develop an obesity phenotype identical to that of Ay mutant mice (Ollmann et al., 1997). Many subsequent studies have shown that AGRP, POMC and MC4R have a significant role in the hypothalamic regulation of energy homeostasis, and studies of human populations have identified haploinsufficiency for MC4R as the most common monogenic cause of severe obesity (Santini et al., 2009).

The accessory roles of Mgrn1 and Atrn appear to be important for central melanocortin signaling in the context of ectopic agouti overexpression because hypomorphic mutations for either gene protect against Ay-induced obesity (Miller et al., 1997). In the wild-type physiological situation, Atrn seems less likely to play a significant role in central melanocortin signaling. Whereas Atrn hypomorphs show increased metabolic rate and hyperactivity even in the absence of agouti (Dinulescu et al., 1998; Gunn et al., 2001), ATRN shows no biochemical affinity for AGRP and Atrn mutants are not protected against obesity induced by AGRP overexpression (He et al., 2001). Interestingly, a paralogue of ATRN, the attractin-like 1 protein (ATRNL1), has been identified as an interacting partner of MC4R (Haqq et al., 2003), although an accessory role for ATRNL1 in AGRP signaling in the central melanocortin pathway does not necessarily follow because ATRN binds the N-terminus of ASP (which is not very similar to the N-terminus of AGRP) and the N-terminal end of AGRP is actually cleaved off of the propeptide to generate the active form of AGRP (Creemers et al., 2006; Jackson et al., 2006). Mice homozygous for a null allele of Atrnl1 have no overt phenotype (Walker et al., 2007), although it is not yet known whether they are resistant to AGRP-induced obesity. Investigating whether ATRN plays a role in energy homeostasis is complicated by the fact that null mutants exhibit a resting tremor and are hyperactive. These additional sources of energy expenditure may be sufficient to explain why Atrn mutants are partially resistant to diet-induced obesity (Nagle et al., 1999). The fact that this increase in energy expenditure occurs without a concomitant increase in food intake suggests a possible role for Atrn in energy homeostasis outside of the AGRP-melanocortin system.

ATRN and MGRN1: a genetic gateway to understanding spongiform neurodegeneration

As described above, Atrn and Mgrn1 null mutant mice have indistinguishable pigmentation phenotypes and epistatic relationships to agouti and Mc1r, and hypomorphic mutants for both genes suppress Ay-induced obesity. These observations strongly suggest that ATRN and MGRN1 act together to achieve a particular step of the pigment-type switching pathway. Atrn and Mgrn1 null mutants also both develop progressive spongiform neurodegeneration (Bronson et al., 2001; He et al., 2001, 2003; Kuramoto et al., 2001). Histologically, the pathology seen in the mutants is very similar to the spongiform change observed in prion diseases such as Creutzfeld–Jacob disease and transmissible spongiform encephalopathies such as ‘mad cow disease.’ The neuropathology associated with loss of ATRN has been extensively characterized in Atrn mutant zitter and myelin vacuolation rats, which also develop spongiform neurodegeneration. Membrane-delimited vacuoles arise both in the cytoplasm of neurons and glia, and within the myelin lamellae surrounding axons (He et al., 2003; Kondo et al., 1991, 1992, 1995; Kuwamura et al., 2002). Spongiform change in Atrn and Mgrn1 null mutant mice is preceded by mitochondrial dysfunction (Sun et al., 2007), although this appears more likely to influence age-of-onset or progression of vacuolation than to be the primary cause (Jiao et al., 2009b). Given the notable phenotypic overlap between the null mutant mice, it is likely that the proteins encoded by Atrn and Mgrn1 act together in a cellular pathway that facilitates both pigment-type switching and neuronal integrity. If this is the case, then identifying the cellular role of ATRN and MGRN1 in the tractable model system of the pigment cell is likely to provide insight into why losing either protein from the brain leads to spongiform encephalopathy. It will be interesting to discover whether the mechanism underlying spongiform change in Atrn and Mgrn1 mutant mice is also involved in prion-related spongiform encephalopathies, especially in light of a recent report by Chakrabarti and Hegde (2009). They observed that MGRN1 localized to aggregates formed by a cytosolic form of prion protein, which resulted in functional depletion of MGRN1 and effects on lysosome morphology similar to those observed in cells in which Mgrn1 expression has been knocked down (Kim et al., 2007). These results suggest that prion aggregation could lead to spongiform encephalopathy at least in part as a consequence of loss of MGRN1 function.

Dark-like: another link between pigment-type switching and spongiform degeneration

The dark (da) mutation was described by Falconer (1956, 1957) as having dark dorsal hairs, reduced fertility, and small body size. The gene was never cloned and the mutant is thought to be extinct, but another spontaneous mutant with similar phenotypes was discovered at the Jackson Laboratory in 1981. As it maps to the same region of chr 7 as dark, it is likely to be a new mutation in the same locus and has been named dark-like (dal). In addition to reduced pheomelanin production, the dal mutant is small in size and has dense bones (by X-ray), mild hydrocephalus, vacuolated cells at the cortical medullary junction of the adrenal gland, and male and female reproductive degeneration (Harris et al., 2003). Genetic analysis places dal downstream of agouti transcription and upstream of Mc1r, similar to the placement of Atrn and Mgrn1 (Cota et al., 2008). Atrn and dal mutants share not only a pigment-type switching defect, but also a similar syndrome of vacuolar testicular degeneration, suggesting that a similar mechanism may cause both phenotypes. To this point, little is known about the ultrastructural characteristics of the testicular vacuoles, but at least one other similarity exists between Atrn mutant brains and dal mutant testes: loss of lipid raft domains from the affected tissue (Azouz et al., 2007; Cota et al., 2008). In addition, while neither dal/dal nor Atrn heterozygous animals develop significant spongiform neurodegeneration, dal homozygotes that are heterozygous for a null allele of Atrn do, and Atrn null mice heterozygous for the dal mutation develop testicular vacuoles at a younger age. These observations suggest that dal and Atrn may act in a shared pathway and that disrupting this pathway leads to spongiform degeneration in susceptible tissues.

Loose ends everywhere: more worlds to discover in the diverse phenotypes of pigment-type switching mutants

There are many additional phenotypes that await elucidation in mice with disrupted pigment-type switching. The Mgrn1 mutant mouse is particularly rich in examples: in addition to their coat color and spongiform neurodegeneration phenotypes, the null mutants have craniofacial abnormalities, curly hair and whiskers, and a defect in patterning the left-right body axis that manifests early in development and often leads to lethal congenital heart defects (Cota et al., 2006; Jiao et al., 2009b). None of these phenotypes are observed in Atrn, Mc1r, Pomc, or agouti mutants, suggesting that they are independent of melanocortin signaling. The diversity of phenotypes associated with loss of MGRN1 may reflect the fact that it is likely to ubiquitinate many proteins to affect their abundance, localization, and/or function. Additional modifiers of the pigment-type switching pathway continue to be discovered; for example, Enshell-Seijffers et al. (2008) recently reported that loss of the transmembrane serine protease corin in agouti mice caused expansion of the yellow band of pigment, resulting in mice with lighter (yellower) coats. Loss of corin had no discernable effect on the expression of agouti, Mc1r, a-MSH, Atrn, or Mgrn1. The effect on the duration of pheomelanin production was intermediate in agouti heterozygotes (A/a mice expressing half the normal amount of agouti ) and corin itself was normally specifically expressed in the dermal papilla during the anagen (active growth) phase of the hair growth cycle, consistent with an effect of corin on ASP activity only when ASP is expressed at low levels. It remains to be seen whether corin acts via proteolytic processing of a component of the pigment-type switching pathway or via another pathway that antagonizes the response of MC1R to ASP.

While much progress has been made in identifying genetic components of the pigment-type switching pathway, more remain to be found and the odds seem high that they will provide new insights into diverse biological systems. The darkness of the coat hairs of agouti mice and many pigment-type switching mutants can vary depending on genetic background, indicating variation in (presently) unidentified modifier loci (i.e., Cota et al., 2008; Dunn, 1920; Little, 1916; Walker et al., 2007). In mice, several mutants for which the gene has yet to be identified have ‘dark’ phenotypes similar to that of Atrn or Mgrn1 loss-of-function mutants. These include Umbrous (unmapped), dorsal dark stripe (chr 15), and dark/dark-like. Additional components are likely to be discovered through the analysis of mutant mice produced by the Knockout Mouse Project – but only if those mutants are examined on an agouti background. Outside of the murine system, additional mysteries await explanation: for example, cats and hamsters each have an X-linked locus that regulates production of black or orange pigment. Animals that are double mutant for sex-linked orange and a non-agouti allele are orange, placing sex-linked orange downstream of agouti in the pigment pathway (Alizadeh et al., 2009; Schmidt-Kuntzel et al., 2009). Surprisingly, the cat and hamster sex-linked orange mutations are likely to be in different genes as they map to non-homologous regions of the X chromosome. It is clear that more surprising discoveries of pigmentation genes and their pleiotropic functions await, and that pigment-type switching will continue to provide a powerful model system for investigating and understanding a diverse array of biological processes.