Study of the MC1-R thus provides valuable insight into the biology of pigmentation. This article begins with an overview of melanocortin receptors, proopiomelanocortin (POMC)-derived ligands, and the agouti antagonist as well as their roles in regulating production of eumelanin and pheomelanin, including UV-induced melanogenesis. A brief description of mouse-coat-color genetics is then followed by a discussion of human MC1-R variants, both as determinants of pigmentation phenotype and as potential factors in the development of cutaneous melanoma and nonmelanoma skin cancer.

MELANOCYTES AND MELANIN

Visible skin and hair color also depends on the amount and chemical composition of the melanin within melanosomes. Melanin exists in 2 basic forms in human skin, brown-black eumelanin and yellow-red pheomelanin; these melanins differ in their biochemical structure and ultrastructural appearance within melanosomes. Eumelanosomes are elliptical, have a highly organized matrix with striated filaments, and contain high-molecular-weight, poorly soluble eumelanin. In contrast, pheomelanosomes are spherical, have an unstructured particulate matrix, and contain low-molecular-weight, soluble, cysteine-rich pheomelanin.^{3, 6} The essential and rate-limiting enzyme in the melanin biosynthetic pathway, tyrosinase, regulates 3 steps including the initial conversion of tyrosine to dihydroxyphenylalanine (DOPA). The pathway diverges after the formation of dopaquinone, at which point production of pheomelanin involves the addition of a cysteinyl group that accounts for its yellow-red color.^{3, 7} Whereas eumelanin synthesis is associated with increased levels of tyrosinase activity and involves additional melanogenic enzymes such as tyrosinase-related protein (TRP)-1 and TRP-2/dopachrome tautomerase (regulators of distal steps in the pathway and/or stabilizers of tyrosinase), pheomelanin synthesis in murine melanocytes is associated with a reduction in tyrosinase, a marked reduction to absence of TRP-1 and TRP-2, and an absence of the P protein.^8-9

The ratio of eumelanin to pheomelanin, as well as total melanin content, is higher in skin types V to VI than skin types I to II.^{7, 10} Pheomelanin levels are greatest in "fire" red hair, while eumelanin predominates in most human hair colors other than red.⁴ The ratio of eumelanin to pheomelanin determines not only visible characteristics, but also photoprotective and cytotoxic properties. Pheomelanin has been found to be photolabile, generating oxidative stress and resulting in photosensitivity rather than photoprotection, whereas eumelanin has some inherent cytotoxicity but confers substantial photoprotection.³ In addition, eumelanosomes form supranuclear melanin "caps" that help shield the nuclei of melanocytes and keratinocytes from UVR.¹¹

MELANOCORTIN RECEPTORS

View this table: [in this window] [in a new window] Table 1. Major Forms of Melanocortin Receptors*

The 5 melanocortin receptors represent a subfamily of G protein–coupled receptors with 7 transmembrane-spanning domains (Figure 1). They are among the smallest G protein–coupled receptors yet to be identified (297-359 amino acids; the MC1-R is 317 amino acids) and are 39% to 61% identical to each other at the amino acid level.¹² On ligand binding, the subunit of the receptor-coupled stimulatory G protein (G_s) activates adenylate cyclase, which increases production of the second messenger cyclic adenosine monophosphate (cAMP) (Figure 2^19-20). In the case of the melanocyte MC1-R, elevated intracellular cAMP results in stimulation of tyrosinase gene expression and activity, melanocyte proliferation, and melanocyte dendricity.^21-22 This mechanism could help explain the hyperpigmented café au lait macules in McCune-Albright syndrome, a disorder in which somatic activating mutations of the gene encoding G_s result in increased intracellular cAMP, thus mimicking the physiologic effect of ligand binding to MC1-R. Although some investigators have proposed alternate mechanisms for regulation of melanogenesis by -MSH, MC1-R activation of the cAMP-dependent pathway is required for optimal -MSH–induced pigmentation.^22-23 We will focus on MC1-R, the key melanocortin receptor in the skin and the primary melanocortin receptor expressed in melanocytes.

View larger version (59K): [in this window] [in a new window] Figure 1. Primary structure and consensus amino acid sequence of the human melanocortin-1 (MC1-R) receptor. Red circles indicate MC1-R variants that result in an amino acid change (nonsynonymous); light brown circles indicate variants that do not result in an amino acid change (synonymous); and arrows indicate known loss-of-function variants. In one study, proline was reported at 162 (Mountjoy et al19) and in another, arginine was reported at 164 (Chhajlani et al20).

View larger version (144K): [in this window] [in a new window] Figure 2. Schematic of a G protein–coupled receptor such as melanocortin-1. Binding of the ligand to the receptor results in activation of adenylate cyclase via the subunit of the receptor-coupled stimulatory G protein (Gs). The increase in intracellular cyclic adenosine monophosphate (cAMP) leads to an increase in tyrosinase activity. GDP indicates guanosine diphosphate; GTP, guanosine triphosphate; ATP, adenosine triphosphate; and P in a circle, phosphate group.

The human MC1-R gene was first cloned in 1992,^19-20 and has been mapped to chromosome 16q24.3.²⁴ The MC1-R is found on many cell types in the skin, including melanocytes, keratinocytes, fibroblasts, endothelial cells, and antigen-presenting cells; however, melanocytes clearly have the highest density of MC1-R.²⁵ The functions of the MC1-R on these various cell types are now unfolding and are related in part to the known anti-inflammatory and immunosuppressive effects of -MSH.^25-26 We will hereafter focus on the melanocyte MC1-R.

For the human MC1-R, in vitro studies of competitive binding and measurement of adenylate cyclase activity and/or cAMP production have shown relative potencies as follows: -MSH = ACTH>-MSH>-MSH (Table 1).^19-20,27 This differs from the murine MC1-R, which has a higher binding affinity for -MSH than for ACTH.¹² Although the murine MC1-R is expressed at a higher density, the human counterpart is more sensitive to both hormones, increases synthesis of cAMP for a longer period, and does not undergo desensitization.²⁸ Nevertheless, the human and murine MC1-R are alike in their ability to specifically induce eumelanin synthesis on activation (Figure 3).²⁹ Expression of MC1-R is stimulated by UVR, -MSH, ACTH, and a variety of UVR-induced, keratinocyte-derived cytokines and growth factors known to influence melanocyte function, including interleukin 1, endothelin-1, and basic fibroblast growth factor.^{7, 30} It has been postulated that activation of MC1-R expression is central to the paracrine regulation of melanocytes, including induction of photoprotective melanization of the skin in response to UVR exposure.⁷

View larger version (119K): [in this window] [in a new window] Figure 3. Control of mammalian melanogenesis. Binding of melanocyte–stimulating hormone (MSH) to the melanocortin-1 receptor (MC1-R) enhances eumelanogenesis while a dysfunctional MC1-R or binding of agouti signaling protein (ASP) to the MC1-R results in pheomelanogenesis. The latter process has been described in mice but not in humans.

POMC-DERIVED PEPTIDES

View larger version (50K): [in this window] [in a new window] Figure 4. Posttranslational processing of proopiomelanocortin (POMC). PC indicates prohormone convertase; MSH, melanocyte-stimulating hormone; JP, joining peptide; ACTH, corticotropin; LPH, lipotropic hormone; End, endorphin; Des, desacetyl; and Ac, acetylated.

Central production of melanocortin peptides occurs in the pituitary gland, where anterior lobe corticotrophs produce primarily ACTH and, in most mammals, intermediate lobe melanotrophs and hypothalamic cells produce -MSH. Following proteolytic cleavage, active forms of -MSH are generated on amidation, resulting in desacetyl -MSH, and subsequent acetylation (Figure 4), with the latter process thought to result in improved secretion and increased biologic potency.³² However, the human pituitary, in which the intermediate lobe is rudimentary, normally secretes very little -MSH. As a result, circulatory levels of the peptide in humans are extremely low.³³ In addition, there is controversy as to whether the major circulating form is acetylated -MSH or the less bioactive desacetyl -MSH. Nevertheless, it is clear that centrally produced melanocortins can dramatically influence cutaneous pigmentation, as evidenced by the generalized hyperpigmentation with accentuation in sun-exposed areas in patients with disorders characterized by pituitary hypersecretion of ACTH and, to a variable degree, -MSH. These include Addison disease and Nelson syndrome (enlarging pituitary tumor after bilateral adrenalectomy for Cushing disease); cutaneous hyperpigmentation also has been reported in a patient with pituitary -MSH hypersecretion, ACTH deficiency, and secondary adrenal insufficiency.³⁴ Conversely, patients with hypopituitarism have been noted to have a decreased ability to tan. Moreover, a phenotype of red hair, severe obesity, and adrenal insufficiency has been described in 2 probands with mutations in the POMC gene resulting in a deficiency of circulating melanocortins.³⁵

Peripherally, POMC is expressed by a variety of epidermal and dermal cells, including melanocytes, keratinocytes, fibroblasts, endothelial cells, and inflammatory cells.^{14, 25} Immunohistochemical staining of human skin has demonstrated that -MSH is the predominant melanocortin peptide synthesized by melanocytes and probably by Langerhans cells, whereas levels of ACTH exceed those of -MSH in differentiated keratinocytes. Production of -MSH and ACTH is accompanied by epidermal expression of PC1 and PC2, with PC2 tending to coexpress with -MSH (Figure 4). Overall, ACTH peptides are present in higher concentrations than -MSH in the human epidermis.³⁶ ACTH[1-17], one of the more abundant of these peptides, binds to the MC1-R with an affinity comparable to that of acetylated -MSH, and both activate adenylate cyclase and stimulate melanogenesis with greater potency than either desacetyl -MSH (the predominant form of -MSH in the skin) or ACTH[1-39].³⁶

Both UVR and the epidermally derived, UVR-induced cytokine interleukin 1 stimulate synthesis and enzymatic processing of POMC by melanocytes and keratinocytes³⁷; in addition, UVR has recently been shown to down-regulate the expression of neprilysine, a peptidase that cleaves and inactivates POMC-derived peptides.³⁸ This results in an increase in the local concentration of the ligands -MSH and ACTH that occurs synergistically with the aforementioned UVR-induced expression of MC1-R in melanocytes, producing biologic amplification of the UVR signal.³⁹ These findings suggest a paracrine as well as autocrine role of POMC-derived peptides in the regulation of UVR-induced cutaneous pigmentation.²⁸ However, the relative contribution of the various centrally and peripherally produced forms of -MSH and ACTH to melanogenesis in vivo has yet to be determined.^{13, 33}

AGOUTI ANTAGONISM OF MELANOCORTIN RECEPTORS AND MOUSE COAT COLOR GENETICS

View larger version (27K): [in this window] [in a new window] Figure 5. Pheomelanogenesis and eumelanogenesis in mice. Wild-type mice have temporal regulation of agouti signaling protein (ASP) production, resulting in a subterminal band of yellow pheomelanic pigment in an otherwise black eumelanic hair shaft (left), whereas mice with either a dominant mutation at the agouti locus (Ay/A) resulting in overproduction of ASP or a recessive loss-of-function melanocortin-1 receptor (MC1-R) mutation (e/e) have an entirely yellow hair shaft. MSH indicates melanocyte-stimulating hormone; A, agouti locus; and e, extension locus.

Historically, there have been 2 major strains of mice with an entirely yellow coat.¹⁸ In one, a dominant mutation at the agouti locus (A^y/A) causes excessive synthesis of ASP throughout the body, which results in a uniform yellow coat (Figure 5) and obesity; ASP antagonism of the hypothalamic MC4-R is the cause of the obesity.⁴⁰ A similar coat color phenotype but normal body weight is found in recessive yellow mice (e/e), where a loss-of-function mutation in MC1-R (encoded by the extension locus) results in animals that produce pheomelanin in their follicular melanocytes (Figure 5).⁴³ More recently, a third type of mouse has been described, which lacks melanocortin peptides owing to a POMC-null mutation. Its phenotypic features include a yellow coat, obesity, and adrenal insufficiency (analogous to the rare human POMC gene mutations described above); these mice have a somewhat darker, "dirty blond" coat color, suggesting that basal MC1-R activity in the absence of ligands may be higher than that of a nonfunctional receptor (Figure 3).⁴⁴

Further studies have shown how the genetics of mouse coat color play out on a molecular level. Melanocytes with a functional MC1-R respond to -MSH with elevated cAMP and increased expression of tyrosinase and melanogenic proteins such as TRP-1 and TRP-2, resulting in eumelanogenesis. Without a functional MC1-R or in response to ASP, there is a reduction in cAMP levels and basal tyrosinase activity as well as an abrogation of expression of other melanogenic proteins, resulting in pheomelanogenesis (Figure 3).^{8, 41}

HUMAN MC1-R VARIANTS, RED HAIR, AND CUTANEOUS PIGMENTATION

In the initial case-control study screening for MC1-R variation in humans, Valverde et al⁴⁷ sequenced the MC1-R gene in unrelated Northern European individuals with red hair and a poor tanning response (n = 30) as well as those with brown/black hair and a good tanning response (n = 30). They found 9 different variants of the MC1-R gene, with at least 1 variant allele present in 70% of the individuals with red hair and none of the controls. The authors then analyzed the involved regions of the MC1-R gene in an additional group unselected for hair color or skin type (n = 75), and overall found at least 1 variant allele in 82% of individuals with red hair, 33% of those with blond/fair hair, and less than 20% of those with brown/black hair; changes in both alleles were found only in those with red hair (29% of the red-haired group).

Subsequent work has identified at least 30 allelic variants of the human MC1-R gene, most of which result in a single amino acid substitution (Table 2,^47-59 Figure 1).² Three particular variants, Arg151Cys (ie, arginine residue at codon 151 changed to cysteine), Arg160Trp, and Asp294His, hereafter referred to as the "red-hair-color" alleles, have been shown in several studies to have highly significant associations with red hair, fair skin, and/or poor tanning ability.^48-51,60-61 The odds of having red hair increase 9- to 16-fold with the presence of 1 of these variant alleles, and individuals carrying 2 of these alleles almost always have red hair.⁴⁹ Due to the weak power of statistical testing in the relatively small sample sizes analyzed thus far, it is not possible to exclude a pigmentary role for many of the less common MC1-R variants. However, other alleles such as Val60Leu and Val92Met that are present at a relatively high frequency in white populations (Table 2) have not been shown to have an association with red hair. Some studies have found the former allele to have an association with blond/fair hair,^48-49 but the latter allele, which is common in Asian populations as well as both lightly and darkly pigmented whites, has not been linked with any particular pigmentary characteristics.

View this table: [in this window] [in a new window] Table 2. MC1-R Variants: Allele Frequency47-57 and Functional Significance52,53,58,59*

Functional studies have confirmed the impact of certain missense mutations in the human MC1-R gene on receptor signaling via the second messenger cAMP. In the first description of a nonfunctional MC1-R isolated from a human subject, Frändberg et al⁵⁸ found that cells transfected with DNA encoding the Arg151Cys receptor variant were able to bind the radiolabeled -MSH analogue ¹²⁵I-NDP([Nle⁴, D-Phe⁷])-MSH with identical affinity as did the wild-type MC1-R, but the transfected cells could not be stimulated by -MSH to produce cAMP. Subsequently, a similar inability to induce cAMP production was found with the MC1-R variants Arg142His, Arg160Trp, and Asp294His, and reduced stimulation of cAMP production compared with the wild type was noted with Val60Leu (Table 2).⁵⁹ Arg142His and Asp294His also displayed slightly decreased -MSH binding affinity compared with the wild-type MC1-R.⁵⁹ Although one study found that Val92Met bound -MSH with a 5-fold lower affinity than the wild type,⁵² this variant as well as Asp84Glu have been shown to induce cAMP production in a manner comparable with the wild type.⁵³ Thus, although some MC1-R variants may be functionally neutral polymorphisms, Arg142His and the 3 red-hair-color alleles are indeed loss-of-function mutations, accounting for the association of the latter 3 variants with a phenotype of red hair. Finally, single nucleotide insertion mutations that produce frameshifts and result in a prematurely terminated, nonfunctioning receptor (null allele) have also been described in individuals with red hair (Table 2)^{2, 50, 54}; an analogous frameshift mutation in mice produces the recessive yellow phenotype (Figure 5).⁴³

Many MC1-R variants, including those such as Asp84Glu and Val92Met that neither show loss of function nor have an association with red hair, are clustered in the region between the first cytoplasmic loop and first extracellular loop, spanning the second transmembrane domain (Figure 1). In contrast, several of the known loss-of-function mutations, including 2 red-hair-color alleles, involve the second intracellular loop (Figure 1), which in this receptor superfamily has been shown to be a critical domain for G-protein coupling.⁶² One could thus speculate that these mutations interfere with interactions between the receptor and G protein, resulting in an inability to induce cAMP production on ligand binding.^58-59 Of note, both Arg142His and Arg151Cys occur within consensus sequences for cAMP-dependent protein kinase recognition (amino acids 142-145 and 151-154),²⁰ and therefore may also potentially block receptor phosphorylation.

Recent studies have provided evidence for a dosage effect of MC1-R variants on both hair color and skin type. Most individuals with red hair are homozygous or compound heterozygous (ie, have 2 different mutant alleles at a particular locus) for MC1-R gene mutations. Flanagan et al⁵⁰ found that among a cohort of unrelated Northern European individuals with red hair (n = 99), 85% were homozygous or compound heterozygous and 13% were heterozygous for MC1-R mutations; interestingly, a significantly greater proportion of the latter group had strawberry-blond or auburn rather than "pure" red hair. A significant heterozygote effect on sun sensitivity has also been reported for loss-of-function MC1-R mutations. In a Northern European cohort (n = 111) unselected for skin type or hair color, individuals with 1 variant allele were intermediate with regard to both skin type and tanning ability between those with 2 variants and those with none.⁶³ The effect of a single variant allele on skin type was confirmed in a case-control study in which individuals of skin types I to II without red hair (n = 61) were more likely to be MC1-R heterozygotes than those with skin types greater than II (n = 60; P<.001).⁵⁰ A significant dosage effect of MC1-R variants on the number of cutaneous freckling sites has also been described,⁵⁰ but no relationship between MC1-R variants and eye color has been reported.

Classic genetic studies dating back to the mid-20th century suggest that human red hair color approximates an autosomal recessive trait,^64-65 and the results of a recent family study analyzing MC1-R sequence variation and hair color phenotype in 11 large kindreds with a preponderance of red hair are in keeping with this conclusion.⁵⁰ Highly penetrant recessive red-hair alleles include the 3 red-hair-color variants known to result in loss of receptor function. In addition, Val60Leu, which is able to induce only a slight increase in intracellular cAMP levels, may act as a less penetrant recessive allele. Of note, the ins179 (insertion of C at nucleotide 537) MC1-R variant, known to produce a truncated and completely inactive MC1-R, in one family seemed to act as a dominant allele, perhaps owing to haploinsufficiency of the remaining functional MC1-R allele.⁵⁰

Because rare individuals with red hair have homozygous wild-type MC1-R alleles, it has been postulated that unidentified changes outside the coding region may exist and alter expression of the MC1-R gene. Recent work has characterized the human MC1-R promoter⁶⁶ and identified an MC1-R isoform generated through alternative messenger RNA splicing, the functional significance of which is currently unknown.⁶⁷ However, MC1-R variants are not solely responsible for the human red-hair phenotype. For example, dizygotic twin pairs discordant for red hair have been found to have identical MC1-R genotypes by descent,⁴⁸ and red hair has also been described in individuals who were compound heterozygous for POMC gene mutations.³⁵ Although human pigmentation is clearly a complex genetic trait, the effect of the single locus encoding the MC1-R is substantial.

Extensive polymorphism of the MC1-R coding region has been observed in Northern European white populations; in population-based studies, variant MC1-R alleles were identified in approximately 50% to 75% of these individuals.^49-51,54 Recently, 2 groups of investigators examined the MC1-R sequence in individuals from other regions of the world to elucidate the role of selection on polymorphism at this locus.^54-55 As anticipated, MC1-R variants associated with red hair in Europe were rarely found in other populations. Conversely, Arg163Gln, a variant found infrequently in Europeans, was present in over 70% of East/Southeast Asians and Native Americans. In contrast, African populations completely lacked nonsynonymous MC1-R variants (ie, those coding for a different amino acid sequence), although the synonymous variant Thr314Thr (A to G at nucleotide 942) was extremely common. Despite the overall similarity of these 2 studies, the authors reached very different conclusions. Due to increased polymorphism at the MC1-R locus in Europeans compared with Africans, and the low average nucleotide diversity at other human loci, Rana et al⁵⁵ inferred that diversifying selection had a role, perhaps related to selection against dark skin in areas of low ambient UVR due to the resultant impaired cutaneous production of vitamin D and predisposition to rickets. Harding et al⁵⁴ used a more quantitative approach and found that there exists strong functional constraint to maintain a wild-type MC1-R in Africa, where eumelanin is required to protect against UVR-induced burning (thus avoiding fluid loss and secondary infection) as well as skin cancer.¹³ These authors argued that European MC1-R diversity does not result from enhanced selection, but rather reflects neutral expectations with relaxation of constraint.⁵⁴

MC1-R VARIANTS AND SKIN CANCER

The relationship between MC1-R variants and cutaneous melanoma also has been the subject of controversy. Valverde et al⁵⁶ initially found that MC1-R variants were significantly more common in cutaneous melanoma cases (n = 43) than in controls (n-44); the association was largely independent of skin type, and a particular rare variant, Asp84Glu, was found in 23% of patients with melanoma and in none of the controls. Subsequent work failed to confirm this association, and it was concluded that Asp84Glu was unlikely to be a major risk factor for cutaneous melanoma, with the results of the original small study ascribed to chance population sampling in the setting of a large number of potential variant alleles.^{57, 68} However, a recent large Australian study⁵¹ comparing patients with familial and sporadic cutaneous melanoma (n = 460) and unrelated controls (n = 399) provided strong evidence that the 3 red-hair-color variants are also associated with melanoma (P<.001; a finding independent of family history), with an approximate doubling of melanoma risk for each allele carried. Interestingly, among fair-skinned individuals alone there was no association between MC1-R variants and melanoma, but the association persisted among those with a medium or dark complexion. The authors concluded that the effect of MC1-R variant alleles on melanoma risk is partially mediated by determination of a light-skinned pigmentation phenotype, but that these alleles also increase melanoma risk in individuals whose darker skin would normally be protective.

Thus recent studies suggest that MC1-R variants may confer a risk of both melanoma and NMSC beyond that explained by visible differences in pigmentation phenotype. One caveat on the interpretation of these results relates to potential reporting bias owing to the lack of an entirely reliable, objective method to define constitutive skin type.² Although, overall, the relationship between frequency of MC1-R variation and skin type is consistent, it is conceivable that some individuals with light skin (therefore also more likely to carry a MC1-R variant) who have had more exposure to the sun (and as a result an increased risk of UV-induced skin cancer) may tend to report a darker skin color.⁵¹ One possible explanation for a true association is that MC1-R mutations influence melanoma development through modulation of melanocyte growth and differentiation⁵⁶; however, the loss-of-function mutations associated with increased melanoma risk would actually be expected to reduce stimulation of melanocyte proliferation.

A second hypothesis involves the change in pigmentary protection from UVR that occurs in the switch from eumelanin to pheomelanin production. Considering that loss-of-function MC1-R alleles result not only in an inability to produce photoprotective eumelanin, but also increase production of photosensitizing and potentially mutagenic pheomelanin, perhaps it is an increased pheomelanin-eumelanin ratio that negates the protective effect of darker cutaneous pigmentation.

In summary, the MC1-R clearly plays a key role in determining the type of melanin that is produced within melanocytes. Further studies will provide insight into how this influences the risk of developing NMSC and melanoma beyond its effect on pigmentation phenotype.

Corresponding author and reprints: Julie V. Schaffer, MD, Department of Dermatology, Yale University School of Medicine, 333 Cedar St, New Haven, CT 06520 (e-mail: votavajr{at}biomed.med.yale.edu).