Melanocortin-1 receptor structure and functional regulation

Authors


* Address correspondence to José C. García-Borrón,
e-mail: gborron@um.es

Summary

The melanogenic actions of the melanocortins are mediated by the melanocortin-1 receptor (MC1R). MC1R is a member of the G-protein-coupled receptors (GPCR) superfamily expressed in cutaneous and hair follicle melanocytes. Activation of MC1R by adrenocorticotrophin or α-melanocyte stimulating hormone is positively coupled to the cAMP signaling pathway and leads to a stimulation of melanogenesis and a switch from the synthesis of pheomelanins to the production of eumelanic pigments. The functional behavior of the MC1R agrees with emerging concepts in GPCR signaling including dimerization, coupling to more than one signaling pathway and a high agonist-independent constitutive activity accounting for inverse agonism phenomena. In addition, MC1R displays unique properties such as an unusually high number of natural variants often associated with clearly visible phenotypes and the occurrence of endogenous peptide antagonists. Therefore MC1R is an ideal model to study GPCR function. Here we review our current knowledge of MC1R structure and function, with emphasis on information gathered from the analysis of natural variants. We also discuss recent data on the regulation of MC1R function by paracrine and endocrine factors and by external stimuli such as ultraviolet light.

Introduction

The communication of living cells with their environment is so critical that several protein families are exclusively dedicated to receive external chemical messengers or physical stimuli and trigger an adaptative response. The largest of these families is the G-protein-coupled receptor (GPCR) superfamily, formed by over 1000 members and accounting for more than 1% of mammalian genomes. GPCRs mediate the responses to a surprising variety of stimuli, including light, odorants, taste molecules, ions, neurotransmitters and hormones. In keeping with this variety of signals, the GPCRs regulate the activity of metabolic enzymes and pathways, ion channels and membrane transporters and the function of the transcriptional, motility and secretory machineries.

The melanocortin-1 receptor (MC1R) is a GPCR expressed in melanocytes and melanoma cells, as originally demonstrated by ligand-binding studies (Donatien et al., 1992; Ghanem et al., 1988; Siegrist et al., 1989, 1994). MC1R regulates the amount and type of pigment production and is a major determinant of skin phototype and sensitivity to ultraviolet (UV) light induced damage. The MC1R physiologic agonists belong to a group of small peptide hormones collectively called melanocortins (MCs), derived from the precursor proopiomelanocortin (POMC). The MCs are expressed in the pituitary and peripheral locations including the skin (Slominski, 1998; Slominski et al., 1993). They regulate important physiologic processes such as food intake and energy homeostasis, cortisol production, sexual behavior and exocrine gland secretion (Gantz and Fong, 2003; Raffin-Sanson et al., 2003; Starowicz and Przewlocka, 2003). The pigmentary actions of pituitary hormones were first recognized about a century ago, when it was demonstrated that the skin of tadpoles turns black on incubation with pituitary extracts. The rapid responses of amphibian melanophores provided sensitive assays for the purification of the melanophore-stimulating activities. The 39 amino acids adrenocorticotrophin (ACTH) was first characterized, followed by the smaller α-melanocyte-stimulating hormone (α-MSH), β-MSH and γ-MSH. Later it was shown that purified α-MSH induces skin darkening in man (Lerner and McGuire, 1961) and hair color changes in mice (Geschwind, 1966).

Melanocortin-1 receptor belongs to a five-member subfamily of GPCRs, the melanocortin receptors (MCRs), which mediate the physiologic actions of MCs by a Gs-protein-dependent activation of the cyclic AMP (cAMP) signaling pathway. The cloning of MC1R (Chhajlani and Wikberg, 1992; Mountjoy et al., 1992) and its mapping to the extension locus in the mouse (Robbins et al., 1993), a locus known to influence pigmentation (Searle, 1968), opened new perspectives for the study of its role as a key regulator of skin biology. Detailed genetic analysis, extensive phenotypic association and structure–function studies, as well as dissection of signaling and regulatory mechanisms have been performed. A number of excellent and recent reviews deal with the relationship of mammalian MC1R with melanocyte function, skin pigmentation and skin cancer risk (Bataille, 2003; Healy, 2004; Rees, 2003; Rouzaud et al., 2005; Sturm, 2002). These topics will not be discussed in depth here. Rather, we will focus on MC1R structure, functional coupling and its regulation. These aspects will be treated within the framework of the MCR family and of GPCR biology.

Structure of the MC1R

The human MC1R gene was cloned independently by Chhajlani and Wikberg (1992) and by Mountjoy et al. (1992), who also reported the cloning of mouse Mc1r. The cloning of other mammalian (Lu et al., 1998b) and non-mammalian (Schioth et al., 2005) MC1Rs followed rapidly. Based on the sequence similarity analysis, MC1R and the other MCRs belong to the Class A of GPCRs, whose prototype is rhodopsin (Gether, 2000). GPCRs are resistant to crystallization, and the direct information on their secondary and tertiary structures is basically limited to a low-resolution electron microscopy structure of bacteriorhodopsin (Subramaniam and Henderson, 1999) and the crystal structure of rhodopsin (Palczewski et al., 2000). These structures have been used as templates for computer-assisted modeling of MCR structure (Holst et al., 2002; Prusis et al., 1995, 1997; Ringholm et al., 2004). MC1R is an integral membrane protein of 317 amino acids in man and 315 in mouse, with the structural hallmarks of the GPCRs, i.e. an extracellular N-terminus, seven transmembrane (TM) fragments and an intracellular C-terminal extension (Figure 1).

Figure 1.

Structure of the human melanocortin-1 receptor. The positions for TM helices are drawn according to the two-dimensional model of Ringholm et al. (2004). The amino acid sequence corresponds to the wild type consensus (GenBank accession number AF326275). Positions for natural non-synonymous mutations are shown in red (for RHC alleles) or orange, and the variant residues are listed in Table 1. Potential post-translational modification sites are also highlighted.

The N-terminus

Most plasma membrane proteins contain an N-terminal signal peptide that targets the nascent protein to the endoplasmic reticulum (ER) membrane and is rapidly removed by proteolytic cleavage. However, the N-terminus of MC1R poorly matches the signal peptide consensus, and removal of the first 27 amino acids has no effect on ligand binding or expression levels in COS cells (Schioth et al., 1997). Moreover, a short Flag epitope fused to the MC1R N-terminus is not cleaved during the processing of the protein in HEK 293T (Sanchez-Mas et al., 2005a) or Mc1r-null B16G4F melanoma cells (Robinson and Healy, 2002). This suggests that MC1R belongs to a class of membrane proteins that use the first TM domain as a non-cleavable signal anchor directing traffic of the protein (Wallin and von Heijne, 1995). Consistent with the retention of the complete N-terminus in the mature protein, mutation of Ser6 to Ala was reported to cause a significant decrease of affinity for radiolabeled agonists (Chhajlani et al., 1996).

Melanocortin-1 receptor contains two N-terminal glycosylation sequons, 15NSTP18 and 29NQTG32, and is glycosylated as shown by electrophoretic mobility shifts following glycosidase treatment (Sanchez et al., 2002; Sanchez-Laorden et al., 2005a; Solca et al., 1989). Whether both sequons are occupied is uncertain, but the presence of a Pro on the carboxylic side of the first one strongly decreases its N-glycosylation potential. The possible role of N-glycosylation has not yet been analyzed. A Cys residue near the boundary of the N-terminus and TM1 is conserved in all MCRs and is critical for its function, as mutation to Gly (Frandberg et al., 2001) or Ala (Sanchez-Laorden et al., 2005b) inactivates the receptor. A likely explanation is its involvement in an intramolecular disulfide bond with a partner located in the third extracellular loop or, alternatively, in an intermolecular disulfide contributing to receptor dimerization (see below).

The extracellular loops (els)

The els in MC1R are small (Holst et al., 2002; Prusis et al., 1997; Ringholm et al., 2004), particularly the short el2 characteristic of all MCRs. This feature has been related with the high constitutive activity of several members of the MCR subfamily (Holst and Schwartz, 2003). The third el of the MCRs is unusually conserved compared with most GPCRs. It is rich in invariant Pro and Cys residues, suggesting a highly specialized function (Holst and Schwartz, 2003). It has been reported that mutation of residues Glu269 and Thr272 to Ala in the el3 of human MC1R lowers the binding affinity for agonists, thus suggesting the involvement of these residues in ligand recognition (Chhajlani et al., 1996). However, as discussed below, agonist binding is mostly accounted for by a network of charged and aromatic residues located in several TM fragments, including TM6. As TM6 is connected to el3, it is difficult to conclude whether Glu269 and Thr272 provide actual contacts for ligand recognition or their mutation in the context of a small, highly conserved and conformationally constrained loop induces a secondary change in the position of TM6 and alters the conformation of the binding crevice. Mutation of Cys267 and Cys275 to Gly or Ala abolishes function (Frandberg et al., 2001; Holst et al., 2002; our unpublished observations), suggesting that these residues form a structurally important intramolecular disulfide bond linking TM6 and TM7. Interestingly, Cys271 in MC4R (equivalent to Cys267 in MC1R) is disulfide bonded with Cys279 (equivalent to Cys275 in MC1R), and a C271R mutation associated with obesity has been described (Tarnow et al., 2003). These findings, together with the perfect conservation of the el3 Cys residues in all MCRs, suggest a general role of this disulfide bond in MCR function. Mutation of Cys273 to Gly causes loss-of-function (LOF) (Frandberg et al., 2001), but a C273A mutant was reported to retain some signaling potential and free Cys273 may provide a ligand for Zn(II), a metal ion that binds to MC1R and MC4R and acts as a partial agonist and as an enhancer of full agonists (Holst and Schwartz, 2003; Holst et al., 2002). However, the perfect conservation of Cys35 and Cys273 in all MCRs, and the mutagenesis data suggest that they may form additional disulfide bond(s), at least under some physiologic conditions.

The intracellular loops (ils)

The ils of GPCRs provide the binding interfaces for the heterotrimeric G proteins and contain phosphorylation targets involved in the regulation of signaling, internalization and cycling (Strader et al., 1994). In the mouse, the natural tobacco mutation S69L leads to Mc1r hyperactivity (Robbins et al., 1993). Ser69 and the homologous Ser71 in human MC1R are located within il1, suggesting that this loop is important for normal receptor activity. Six naturally occurring mutations in human MC1R cluster in the il2 (Figure 1). At least four of these variants are total or partial LOF forms, underscoring the importance of this domain for MC1R function. A 141DRY143 tripeptide, characteristic of class A receptors, is located at the interface between TM3 and il2. This is a critical element, as shown by the partial LOF of the R142H human variant (Schioth et al., 1999). Mc1r and MC1R share conserved protein kinase A (PKA) phosphorylation sites in il2 (142RYIS145 and 151RYHS154, numeration corresponding to the human sequence) and a protein kinase C (PKC) target (157TLPR160). Whether these sites are phosphorylated under some physiologic conditions remains uncertain. Il3 is poorly conserved in the MCRs. It is rich in basic residues. A target sequence for PKA and PKC (225RRRSIR230, numeration corresponding to mouse Mc1r) is present in the il3 of Mc1r, but absent in human MC1R. Therefore, the mouse receptor is likely a better substrate for the second messenger activated protein kinases.

The TM fragments and the agonist-binding crevice

Transmembrane fragments are placed approximately perpendicular to the plane of the membrane, in a counter-clockwise disposition. For Class A GPCRs, the ligand-binding site is a pocket located below the plasma membrane–extracellular medium interface, formed with the contribution of several TM fragments. Three-dimensional models of ligand–receptor complexes have been developed (Haskell-Luevano et al., 1996; Prusis et al., 1995, 1997), which agree reasonably well with site-directed mutagenesis studies (Yang et al., 1997a). These models suggest that a highly charged region containing Glu94 (TM2), Asp117 and Asp121 (TM3) interacts with an Arg residue in the His-Phe-Arg-Trp (HFRW) pharmacophore core shared by the natural MCs. A network of aromatic residues located near the extracellular side of TM4, 5 and 6 would also contribute to agonist binding by interacting with the aromatic residues of the pharmacophore (Yang et al., 1997a). Interestingly, 11 natural mutations cluster in TM2 of the MC1R (Figure 1) and several of them have important functional consequences.

The cytosolic C-terminal extension

The cytosolic tail of the MC1R is short, with some 19 amino acids of which six are invariant in the MCRs. Within GPCRs, this is a functionally relevant domain involved in: (i) interaction of the ligand–receptor complex with G protein(s) (Strader et al., 1994), (ii) correct disposition of the receptor within the plasma membrane by acylation (palmitoylation or myristoylation) of Cys residues and integration of the acyl chain within the lipid bilayer (Qanbar and Bouvier, 2003) and (iii) providing signals for intracellular trafficking of the protein (Schulein et al., 1998). Moreover, this region often contains Ser and Thr residues whose phosphorylation mediates receptor desensitization and internalization (Luttrell and Lefkowitz, 2002; Pitcher et al., 1998).

A natural R306ter mutation with premature truncation of the C-terminal tail was found in domestic dogs, with homozygous animals showing pheomelanic coats (Newton et al., 2000). A smaller deletion of the last five amino acids in MC1R, including the CSW terminal tripeptide invariant in all MC1Rs sequenced so far, is sufficient to abolish function (Sanchez-Mas et al., 2005b). Impaired signaling in the deletion mutants is due to the decreased cell surface expression. Removal of the last Trp residue or the CSW tripeptide in MC1R decreases the density of binding sites to 50% and 3% of controls, respectively. Loss of binding sites is complete when the last five amino acids are deleted (Sanchez-Mas et al., 2005b). This shows that MC1R cell surface expression is dependent on the integrity of its C-terminus. Consistent with this possibility, a dileucine-like motif (312VL313) is disrupted by deletion of the 313LTCSW317 terminal pentapeptide and is absent in the dog R306ter mutation. All the MCRs contain similar dihydrophobic signals preceded by an acidic residue. Moreover, putative acidic/dihydrophobic sequence motifs are present in the carboxyl tails of many GPCRs (Schulein et al., 1998). For MC4R, mutational analysis of this motif has confirmed its importance for surface expression (VanLeeuwen et al., 2003).

A C315A mutation greatly impairs MC1R function in HEK 293T cells transfected with the variant (Sanchez-Mas et al., 2005b). This partial LOF may be related to lack of acylation of the protein (Frandberg et al., 2001). Based on the less number of receptor-binding sites in the C315A mutant, acylation may be an important determinant of receptor density, either by improving receptor traffic to the plasma membrane or by stabilizing the protein in this location, as previously shown for other GPCRs (Qanbar and Bouvier, 2003).

Although the coding region of the MC1R gene is apparently intronless, as it is the case for most GPCR genes (Minneman, 2001), the occurrence of an alternative splice form has been reported (Tan et al., 1999). This form would encode an additional 65 amino acids at the C-terminal tail and appeared pharmacologically similar to the non-spliced MC1R. Therefore, increasing the length of the C-terminal tail would be less deleterious for signaling activity than shortening it. However, other attempts to identify alternative transcripts by rapid amplification of 3′ cDNA ends (3′ RACE) have been unsuccessful (Smith et al., 2001).

The quaternary structure of MC1R

Many GPCRs form dimers or oligomers in living cells (Angers et al., 2002; Breitwieser, 2004; Milligan, 2001). Dimerization can modulate key properties of the receptors, including ligand binding, coupling efficiency, desensitization and intracellular traffic from the ER to the plasma membrane, and through endocytic pathways. A recent bioluminescence resonance energy transfer study suggested that MC1R undergoes constitutive dimerization in heterologous cells overexpressing the receptor (Mandrika et al., 2005). Dimer formation has been demonstrated for the related MC4R and accounts for the dominant negative effect of a mutant associated with severe early onset obesity (Biebermann et al., 2003). We have confirmed that MC1R is able to form SDS-resistant dimers and higher order oligomers using a different approach involving co-immunoprecipitation of differentially epitope-tagged MC1R constructs and Western blot (Sanchez-Laorden et al., 2005a). Dimerization appeared constitutive, as it was seen in the absence of MC agonists, and is an early event in MC1R biosynthesis, taking place in the ER. A number of natural MC1R variants were found to homodimerize, and to heterodimerize with the wild-type (WT) protein. Although MC1R was not found to exhibit cooperativity in agonist binding, heterodimerization of WT and mutant forms had a wide range of functional consequences including dominant negative effects and a modulation of key pharmacologic properties such as affinity for agonists or coupling efficacy.

MC1R signaling and the induction of eumelanogenesis

Pharmacology and receptor activation

The MCRs display overlapping specificities and, given the presence of a common HFRW pharmacophore core, they usually bind several MCs. Only MC2R shows a strong selectivity for ACTH (Mountjoy et al., 1992). MC1R (Sanchez-Mas et al., 2004) and MC4R (Gantz et al., 1993b) discriminate poorly between ACTH and α-MSH, and also bind β- and γ-MSH with lower affinities. Concerning MC3R and MC5R, their preferential ligand is α-MSH, but MC3R is the least selective MCR and binds all four natural MCs with similar affinities (Gantz et al., 1993a; Labbe et al., 1994; Roselli-Rehfuss et al., 1993).

The identification of Glu94 in MC1R (Glu92 in Mc1r) as part of the agonist-binding site prompted a model for receptor activation and a ligand mimetic interpretation of the constitutive activity of the mouse Sombre-3J mutation, E92K. The highly charged domain containing Glu94, Asp117 and Asp121 would interact with the positive Arg residue in the HFRW pharmacophore. The presence of the positive charge would modify the packing of TM2 and TM3, and the movement of TM3 would transmit a conformational change to the second il loop. The new conformation of this region containing the conserved DRY peptide and other residues, known to be important for functional coupling, would allow for interaction with the Gs protein (Lu et al., 1998a). In the Sombre-3J mutation, the switch from a negative to positive charge would mimic the presence of the Arg residue of the ligand, thus inducing a conformation similar to the ligand–receptor complex. Homologous activatory mutations have been found in chicken (Ling et al., 2003), and an E94K artificial MC1R variant is a constitutively active form of similar pharmacologic profile (Lu et al., 1998a; Sanchez et al., 2002). Interestingly, the C125R mutation in fox also shows constitutive activity (Vage et al., 1997). It has been pointed out that Arg125 is located near Asp117 and Asp121 in the 3D structure of the receptor and can therefore establish ionic interactions with these residues or with Glu94, again mimicking the positive charge of the agonist. The key role of the packing of TM2 and TM3 in receptor activation is further highlighted by the equivalent L98P and L99P activatory mutations in the mice and bovine, where the introduction of a Pro residue is supposed to modify the position of TMs in this region (Robbins et al., 1993). However, the interpretation of receptor activation based on electrostatic interactions has been recently challenged by the discovery of synthetic non-peptide agonists missing a basic, positively charged group (Holder and Haskell-Luevano, 2004). These synthetic agonists bear aromatic structures, expected to mimic the hydrophobic interactions established between the aromatic residues of the MC pharmacophore and a network of aromatic residues in TM4, 5 and 6. Therefore, hydrophobic interactions may be sufficient for activation in the absence of a basic group.

The MCR family is unique because of the presence of endogenous peptide antagonists, the agouti protein (ASP) and the agouti-related protein (AGRP), in addition to agonists (reviewed by Voisey and Van Daal, 2002). The mouse agouti gene encodes for a 131 amino acid protein containing a 22 residue putative secretion signal, an internal basic region and a C-terminal domain containing 10 cysteines (Bultman et al., 1992; Miller et al., 1993). Mouse ASP is a potent antagonist of Mc1r and Mc4r, a weaker antagonist of Mc3r, but has no effect on Mc5r (Kiefer et al., 1997; Lu et al., 1994; Willard et al., 1995). Agouti signal protein (ASIP), the human homolog of ASP, contains 132 amino acids with an 85% homology with murine ASP. The secreted protein is expressed in testis, ovary, heart and at lower levels in liver, kidney and foreskin (Wilson et al., 1995). The inhibitory effect of ASIP is stronger for MC1R and MC4R than for the other MCRs (Yang et al., 1997b). ASP has pleiotropic effects and its deregulation is associated with embryonic lethality, obesity, diabetes and susceptibility to certain tumors (Bultman et al., 1992; Siracusa, 1994). Concerning melanogenesis, expression of ASP promotes the synthesis of pheomelanic pigments within murine melanocytes (Miller et al., 1993) and accounts for the banding pigmentation pattern in the hair of mice. Follicular melanocytes stop synthesizing dark eumelanin after 4 d of growth, when a pulse of ASP expression switches pigment production from eu- to pheomelanogenesis. This situation persists until eumelanogenesis resumes when ASP expression is turned off, around day six. The blockade of eumelanogenesis by ASP is at least partially accounted for by a strong inhibition of the activities of the enzymatic tyrosine hydroxylase, dopa oxidase and dopachrome tautomerase (Sakai et al., 1997). In addition, an antiproliferative action of ASP on B16 cells has been reported (Siegrist et al., 1996). The effects of ASIP on human melanocytes are similar in culture (Suzuki et al., 1997), but its role in human pigmentation has been less extensively investigated and is yet to be shown in situ.

The finding that ASP modifies the pigmentation status of melanocytes in the absence of MCs (Hunt and Thody, 1995) led to the proposal that at least part of the effects of ASP may be mediated by a specific and unknown receptor (Conklin and Bourne, 1993). However, Mc1r-deficient mouse melanocytes and melanoma cells do not respond to ASP (Abdel-Malek et al., 2001; Siegrist et al., 1996). Accordingly, it is now generally accepted that the Mc1r is the principal, if not the only, mediator of the effects of ASP on melanocytes. As discussed below, the MC-independent actions of ASP can be adequately explained in terms of inverse agonism without the need to invoke a specific receptor for ASP.

Functional coupling

The MCRs signal primarily by the activation of the heterotrimeric Gs protein and stimulation of adenylyl cyclase. The resulting increase in intracellular cAMP levels triggers the activation of PKA. cAMP is responsible for most, if not all, melanogenic actions of α-MSH (Busca and Ballotti, 2000). These include the activation of tyrosinase, the rate-limiting enzyme in melanin biosynthesis, and a switch from the biosynthesis of pheomelanins, the default pathway under basal conditions of low tyrosinase activity, to production of the darker and more photoprotective eumelanins (Ito, 2003). Tyrosinase activation occurs mainly at the transcriptional level, and is mediated by induction of Microphthalmia (Mitf), a transcription factor of the helix–loop–helix family (Vance and Goding, 2004). The superpotent agonist NDP-MSH also increases the number and size of melanosomes in mouse melanoma cells (Martinez-Esparza et al., 2001), suggesting that MC1R signaling could stimulate melanosome biogenesis.

The number of specific α-MSH binding sites in human melanocytes is less. Normal human melanocytes normally express a few hundreds of MC1R molecules per cell (Donatien et al., 1992) and melanoma cells overexpressing the MC1R at the mRNA and protein levels (Loir et al., 1999; Mountjoy et al., 1992; Salazar-Onfray et al., 2002) hardly display more than a few thousands binding sites per cell (Ghanem et al., 1988; Siegrist et al., 1989). Under these conditions MC1R density, instead of the concentration of other early components of the signaling pathway limits the cellular production of cAMP in MC-stimulated melanocytes. Accordingly, increasing the density of MC1R-binding sites in human melanoma cells by stable transfection with the MC1R gene results in higher cAMP levels following agonist stimulation (Sanchez Mas et al., 2003). Genetic analysis supports this view and indicates that there is little or no MC1R reserve in human melanocytes.

In addition to melanocytes, expression of MC1R from a surprisingly wide variety of cell types has been reported, including keratinocytes, fibroblasts, endothelial cells, sebocytes, Langerhans cells, monocytes, dendritic cells, neutrophils, granulocytes, natural killer cells, osteoclasts, Leydig and lutein cells and others (D. W. Roberts, R. A. Newton, K. A. Beaumont, J. H. Leonard & R. A. Sturm, submitted, and references therein). This suggests a large spectrum of possible MC1R physiologic roles. However, MC1R expression in non-melanocytic cells has often been demonstrated by polymerase chain reaction, using a high number of amplification cycles, and has not been confirmed at the protein level by reliable methods such as immunochemical techniques with the appropriate controls or, preferably, radioligand-binding analysis. Moreover, careful quantitative comparisons show that MC1R mRNA is much lower in non-melanocytic cells than in melanocytes, with only the higher amounts found in these latter reflected in clearly detectable protein (D. W. Roberts, R. A. Newton, K. A. Beaumont, J. H. Leonard & R. A. Sturm, submitted). Therefore, it is not yet clear what cell types other than melanocytes express the MC1R protein at physiologically relevant levels compatible with productive functional coupling.

G-protein-coupled receptors often activate more than one signaling pathway. In the MCR family, human MC3R was first shown to couple to both cAMP and the inositol triphosphate (IP3) and Ca2+ signaling systems (Kim et al., 2002; Konda et al., 1994; Wachira et al., 2003). Increases in intracellular calcium could only be observed after inhibition of PKA, and the IP3 response to agonist is biphasic, with a dose-dependent increase at low agonist concentration followed by a PKA-dependent inhibition (Konda et al., 1994). This shows that these signaling systems are regulated by PKA activity, and hence by the cAMP pathway. Coupling to Ca2+ responses has also been reported for MC5R (Hoogduijn et al., 2002), mouse Mc1r, Mc3r, Mc4r, and Mc5r (Mountjoy et al., 2001), MC4R (Sabatier et al., 2003), and MC1R expressed in keratinocytes (Elliott et al., 2004) and human melanoma cells (Eves et al., 2003). At least in HEK 293 cells, the mechanism of calcium mobilization is not perfectly understood. It appears that the Ca2+ spike is IP3- and cAMP-independent, and originates from intracellular stores (Hoogduijn et al., 2002; Mountjoy et al., 2001). Therefore, Ca2+ signaling is a common feature of the MCR family, although its role in MC1R-mediated regulation of melanogenesis is unclear.

Constitutive activity of the MCRs and inverse agonism

Agonist-independent constitutive activity is a relatively frequent feature among GPCRs (Chen et al., 2000; Kenakin, 2001; Parnot et al., 2002; Seifert and Wenzel-Seifert, 2002). Constitutive signaling has been shown for MC3R and MC4R, Mc5r (Nijenhuis et al., 2001) and human and mouse MC1R (Sanchez-Mas et al., 2004). In HEK cells expressing MC1R, basal cAMP concentrations raised with receptor expression. Maximal constitutive signalling yields about 40% the maximal intracellular cAMP concentrations achieved upon agoinst binding. This value compares well with the basal agonist-independent activity of the E92K Eso−3J natural Mc1r allele and other artificial constitutively active MC1R and Mc1r mutants (Lu et al., 1998a; Robbins et al., 1993; Sanchez et al., 2002). Mc1r expressed at a similar density also displayed significant but lower constitutive activity. Inverse agonists compete with activatory ligands for the same binding site in constitutively active receptors and, in addition, stabilize the inactive receptor conformation, thus decreasing agonist-independent signaling (Kenakin, 2001). The high constitutive activity of MC1R and Mc1r accounts for the inverse agonist behavior of ASP (Siegrist et al., 1997) and for its direct actions in the absence of agonists (Graham et al., 1997; Ollmann et al., 1998; Sakai et al., 1997; Suzuki et al., 1997). Interestingly, it has been pointed out that, whereas dominant gain-of-function mutations of the agouti gene produce completely yellow mice, Pomc1 knockout mice lacking endogenous MCs are brownish, with reduced but detectable eumelanin (Bennett and Lamoreux, 2003), consistent with constitutive Mc1r signaling. Constitutive activity of MC1R is also consistent with the finding that expression of MC1R confers constitutive pigmentation to a receptor-null amelanotic mouse melanoma cell line (Chluba-de Tapia et al., 1996) and rescues the yellow phenotype of homozygous Mc1re mice (Healy et al., 2001). However, the interpretation of these observations is complicated by the presence of natural MCs in Mc1re mice and by the fact that melanocytes and melanoma cells express and process the POMC gene to biologically active MC-like products, leading to an MC1R-based autocrine loop (Loir et al., 1997; Wakamatsu et al., 1997). Thus, MC1R can signal independently of pituitary MCs by different mechanisms, including constitutive activity and para- or autocrine stimulation.

Regulation of MC1R function

The MC1R gene is unusually polymorphic, and many of the natural variants are functionally relevant (Ringholm et al., 2004; Wong and Rees, 2005). Moreover, MC1R dimerization may conceivably give rise to dominant-negative or partial transcomplementation effects. Thus, within human melanocytes, the MC1R genotype provides a first level of regulation of MC1R signaling. In addition MC1R activity can be modulated by several mechanisms including changes in gene expression, mRNA stability and/or translation efficiency or the rate of post-translational processing of the receptor protein and its traffic through the secretory pathway. Once in the plasma membrane, MC1R activity is primarily controlled by the interaction of the receptor with the activatory MCs and, at least for mouse Mc1r, by the inhibitory ASP. Ligand binding triggers other regulatory events including homologous desensitization of signaling and receptor internalization.

Functional analysis of natural MC1R variants

The first reports on the functional consequences of MC1R receptors dealt with natural extension alleles found in the mouse. The recessive yellow (e) allele associated with a yellow pheomelanic phenotype, is a one nucleotide deletion at position 549 of the open reading frame, which causes a frame shift and a premature stop after 12 additional codons. The resulting protein lacks TM5 to TM7, and not surprisingly, is a complete LOF form (Robbins et al., 1993). Conversely, the sombre (Eso and Eso−3J) and tobacco (Etob) alleles are associated with darker eumelanic coats. The sombre Eso and Eso−3J alleles correspond to two point mutations, E92K and L98P, respectively, located in the outer portion of TM2 (Robbins et al., 1993). In heterologous expression systems, the sombre alleles behave as constitutively active receptors, able to couple with adenylyl cyclase in the absence of agonists with cAMP levels around 30–50% the maximal intracellular concentrations achieved upon agonist stimulation. Sombre Mc1r does not respond to α-MSH although it is further stimulated by the superpotent analog NDP-MSH (Robbins et al., 1993). The tobacco mutation (S69L) yields a hyperactive receptor, producing higher cAMP concentrations than WT upon agonist stimulation, rather than a constitutively active form (Robbins et al., 1993). These pioneer studies demonstrated the causal relationship of Mc1r functional status and coat color phenotypes. Moreover, they showed that eumelanin synthesis requires Mc1r activity, whereas pheomelanogenesis is a default pathway proceeding in the absence of Mc1r signaling (Ito, 2003). These conclusions have been supported by the finding of similar mutants in cattle, fox, guinea pig, panther and dog (Adalsteinsson et al., 1987; Cone et al., 1996; Klungland et al., 1995; Lu et al., 1998b).

In mankind, more than 60 non-conservative natural variants of MC1R have been reported (Table 1 and Figure 1). Population studies demonstrate that several alleles are associated with red hair and fair skin (the RHC phenotype) (Box et al., 1997; Healy et al., 2000; Smith et al., 1998; Valverde et al., 1995). Sturm and coworkers (Duffy et al., 2004; Sturm et al., 2003) compared the penetrance of a number of these variants. The strong RHC alleles (designated R) show odds ratios for red hair ranging from 50 to 120. These are the frequent R151C, R160W and D294H variants and the rare D84E and R142H alleles. The weaker RHC alleles V60L, V92M, R163Q, designated r, have odd ratios for red hair ranging roughly from 2 to 6. The R variants R142H, R151C, R160W and D294H together with the r allele V60L are present in 30 % of the North European population, and account collectively for over 60% of all cases of red hair (Healy et al., 2001).

Table 1.  Natural variants of the MC1R
VariantCodon substitutionReference
P18ACCC→GCCBastiaens et al. (2001)
N29insnt86_87insAHarding et al. (2000)
C35YTGC→TACFargnoli et al. (2003)
V38MGTG→ATGFargnoli et al. (2003)
I40TATC→ACCJimenez-Cervantes et al. (2001a)
L44VCTC→GTCFargnoli et al. (2003)
F45LTTC→CTCPastorino et al. (2004)
S47IAGC→ATCJohn et al. (2003)
V60LGTG→TTGBox et al. (1997)
A64SGCC→TCCValverde et al. (1995)
K65NAAG→AATBox et al. (1997)
R67QCGG→CAGRana et al. (1999)
F76YTTC→TACValverde et al. (1995)
A81PGCC→CCCBastiaens et al. (2001)
S83PTCG→CCGJohn et al. (2003)
S83LTCG→TTGPastorino et al. (2004)
D84EGAC→GAAValverde et al. (1995)
S90TAGC→ACGChhajlani and Wikberg (1992)
N91DAAC→GACValverde et al. (1995)
V92LGTG→TTGBox et al. (1997)
V92MGTG→ATGValverde et al. (1995)
L93RCTG→CGGSanchez et al. (2002)
T95MACG→ATGValverde et al. (1995)
V97IGTC→ATCValverde et al. (1995)
L99ICTC→ATCJohn et al. (2003)
A103VGCC→GTCValverde et al. (1995)
G104SGGT→AGTBastiaens et al. (2001)
L106QCTG→CAGValverde et al. (1995)
A111VGCG→GTGPastorino et al. (2004)
I120TATT→ACTFargnoli et al. (2003)
V122MGTG→ATGJimenez-Cervantes et al. (2001a)
R142HCGC→CACBox et al. (1997)
R142CCGC→TGCMatichard et al. (2004)
R151CCGC→TGCBox et al. (1997)
Y152stopTAC→TAAJohn and Ramsay (2002)
I155TATC→ACCBox et al. (1997)
V156LGTG→CTG(SNPs linked to gene, GenBank)
R160WCGG→TGGBox et al. (1997)
R160QCGG→CAGPastorino et al. (2004)
R162PCGG→CCGJimenez-Cervantes et al. (2001c)
R163QCGA→CAABox et al. (1997)
A164RGCC→CGCChhajlani and Wikberg (1992)
A171DGCC→GACJohn and Ramsay (2002)
V173del520_523delGTCBastiaens et al. (2001)
V174IGTC→ATCBastiaens et al. (2001)
F179insnt537_538insCPalmer et al. (2000)
F196LTTC→CTCJohn et al. (2003)
R213WCGG→TGGPastorino et al. (2004)
P230LCCG→CTGBastiaens et al. (2001)
P256SCCC→TCCJohn and Ramsay (2002)
H260PCAT→CCTBastiaens et al. (2001)
T272MACG→ATG(SNPs linked to gene, GenBank)
K278EAAG→GAGBastiaens et al. (2001)
N279SAAC→GGCBastiaens et al. (2001)
N279KAAC→AAABastiaens et al. (2001)
I287MATC→ATGHarding et al. (2000)
D294HGAC→CACValverde et al. (1995)
Y298HTAC→CACMatichard et al. (2004)
A299TGCC→ACCBox et al. (1997)
A299VGCC→GTCPeng et al. (2001)
T308MACG→ATGPastorino et al. (2004)
C315RTGC→CGCPastorino et al. (2004)

Not surprisingly, a number of studies have addressed the functional properties of the RHC alleles. The D84E variant expressed in HEK 293 cells was first reported to couple to adenylyl cyclase with dose–response curves comparable with WT (Koppula et al., 1997). However, a recent study using the same expression system found that the ability of the D84E receptor to couple to cAMP stimulation was dramatically lower than WT, with comparatively small effects on binding affinity (Ringholm et al., 2004). There is agreement in that the R alleles R142H, R151C, R160W, and D294H correspond to diminished function forms, but the degree of functional impairment and its molecular basis are not yet completely clear. Schioth et al. (1999) reported that the total binding capacity and agonist affinity in cells transfected with the mutant forms were similar to WT. Conversely, coupling studies showed a decreased but detectable ability to stimulate cAMP production upon agonist binding in the order WT ≫ R160W ≈ R151C > R142H ≈ D294H. Similar results were previously reported for R151C (Frandberg et al., 1998). This suggested that functional impairment of these RHC alleles was mainly related to a defect in activation of the Gs protein. However, recent studies demonstrate intracellular retention with markedly reduced cell surface expression for D84E, R151C, I155T and R160W, but not for D294H, thus showing that aberrant processing may contribute to the partial LOF associated with certain RHC alleles (Beaumont et al., 2005; Sanchez-Laorden et al., 2005a).

The function of the RHC alleles has also been analyzed under more physiologic conditions, using melanocyte cultures of defined MC1R haplotype or transgenic mice models. Cultures of melanocytes homozygous for the R160W allele, and compound heterozygotes for R160W and D294H, or R151C and D294H, but not homozygotes for V92M also yielded significantly reduced responses to α-MSH in terms of cAMP production following agonist stimulation (Scott et al., 2002b). Sturm and co-workers reported a study of over 300 primary melanocyte cultures of known MC1R genotype (Leonard et al., 2003). Strains homozygous for the consensus sequence had dark pigmentation and contained heavily melanized mature melanosomes, whereas homozygous R151C and R160W were lighter, with only immature melanosomes. Heterozygote cells harboring one of the RHC alleles generally showed pigmentation characteristics intermediate between homozygote WT or variant cultures, indicative of gene dosage effects. The role of the RHC alleles has also been investigated in transgenic mice, by expression of the mutants on a recessive yellow background. Consistent with retention of sizeable function, R151C, R160W and D294H partially rescued the pigmentation phenotype, although not as completely as WT (Healy et al., 2001). In these experiments, transgenic mice were gray for the R151C and R160W variants and dark yellow for D294H, at comparable transgene expression levels. Overall, these data are consistent with results obtained in heterologous cells, and support their use as convenient and reliable systems to study variant MC1R function.

A number of low penetrance r alleles have also been shown to encode partial LOF forms. V92M expressed in heterologous HEK 293 cells was first reported to activate adenylyl cyclase with dose–response curves comparable with WT (Koppula et al., 1997). However, recent data indicate that the affinity of V92M for the MCs is approximately 100-fold lower than WT, and the maximal cAMP response at a saturating dose is also significantly lower (Ringholm et al., 2004). The V60L r allele also achieves significantly lower cAMP levels after agonist stimulation, consistent with impaired functional coupling (Schioth et al., 1999). One frequent variant in East and Southeast Asians, R163Q, is a weak RHC allele (Sturm et al., 2003), with lower affinity than WT but a similar cAMP response (Ringholm et al., 2004).

For a limited number of natural variants, functional analysis has been performed but convincing phenotypic association studies are not available, due to their low frequency. Among these, the rare R162P form was of particular interest since the first MC1R sequence reported as WT had a Pro residue at this position, but other studies found an Arg residue (Chhajlani and Wikberg, 1992; Tan et al., 1999). Proline distorts helical structures and therefore, an R/P substitution at the interface between il2 and TM3 may be functionally relevant. Expression in heterologous cells demonstrated that R162P is a complete LOF allele, unable to couple to the cAMP pathway (Jimenez-Cervantes et al., 2001c). Recent data show that the R162P receptor is retained in intracellular compartments with a virtually complete absence of cell surface expression (Sanchez-Laorden et al., 2005b) (Figure 2). As partial intracellular retention has also been reported for the RHC variants R151C and R160W located in the same protein domain (Beaumont et al., 2005; Sanchez-Laorden et al., 2005a), il2 seems important not only for proper coupling but also for normal processing and intracellular traffic. Another rare natural variant, L93R, was found in a human melanoma cell line unresponsive to NDP-MSH (Sanchez et al., 2002). In heterologous systems, the L93R form failed to bind agonists and couple with cAMP production, in spite of mRNA and protein levels similar to WT. Further studies have shown that the L93R protein is also retained in an intracellular compartment (Figure 2), again pointing to an aberrant processing as the likely cause of LOF (Sanchez-Laorden et al., 2005b).

Figure 2.

Intracellular retention of the L93R and R162P variants of MC1R. HEK 293 cells were trasfected with Flag epitope-tagged WT or variant MC1R. Cells were fixed and stained for confocal laser scanning microscopy with an antiFlag monoclonal and an Alexa 568-labeled secondary antibody, with or without previous permeabilization with 1% Igepal. Only cells expressing the WT protein show an intense labeling without permeabilization, corresponding to receptor molecules located on the plasma membrane. Cells expressing the variants show a very faint external labeling, but a strong and diffuse signal is detected after permeabilization, indicative of intracellular retention. White bar, 20 μm.

Two natural alleles, I40T and V122M, were described and functionally characterized in a survey of the MC1R haplotype in the Spanish population (Jimenez-Cervantes et al., 2001a). I40 lies in the extracellular side of the TM1 and V122 is located within TM3 (Figure 1). Both residues are highly conserved in mammals and are present in mouse and human MC3R and MC5R. The affinity for NDP-MSH was one and two orders of magnitude lower than WT for I40T and V122M, respectively. In addition, the coupling efficacy of V122M was also diminished compared with WT. Therefore, the functional properties of the V122M allele are reminiscent of the r V92M form, suggesting that V122M should also behave as a low penetrance RHC variant.

Other natural MC1R variants described thus far have not been checked for function, although in some cases a reasonable prediction can be made. Four variants, S83P, Y152Stop, A171D and P256S, were reported recently (John and Ramsay, 2002) in a survey of red haired South African individuals. Another four variants, C35Y, V38M, L44V and I120T were found in a Mediterranean population (Fargnoli et al., 2003). At least the Y152Stop variant is certainly a LOF form as it should lack the last four TM fragments, a situation similar to the recessive yellow mouse allele. The C35Y allele is also very likely non-functional, as two related mutants obtained by site directed mutagenesis, C35A and C35G, are total LOF forms. The S83P and P256S alleles may also be LOF forms, as these amino acid positions are predicted intolerant to substitution (Kanetsky et al., 2004), using the Sorting Intolerant From Tolerant program that identifies the SNPs most likely to be functionally important (Ng and Henikoff, 2001, 2002). Finally, eight novel substitutions have been reported: F45L, S83L, A111V, R160Q, R213W, V265I, T308M and C315R (Pastorino et al., 2004). R160Q and C315R should be partial LOF forms, given the behavior of the related RHC allele R160W and the artificial mutant C315A described above.

In short, it would appear that the RHC alleles correspond to partial LOF alleles with varying degrees of functional impairment, ranging from moderate decreases in agonist affinity or functional coupling efficacy for weak alleles such as V92M, to an almost complete loss of functional coupling for alleles such as D84E and D294H. Moreover, recent data show that aberrant traffic with intracellular retention is a common cause of MC1R functional impairment associated with the RHC phenotype. In addition, it has recently been shown that activation of MC1R protects human melanocytes from UV-induced apoptosis, and reduces the accumulation of DNA photoproducts (Bohm et al., 2005; Kadekaro et al., 2005). This effect is accounted for by reduced levels of hydrogen peroxide and an enhanced rate of repair of cyclobutane pyrimidine dimers formed upon exposure to UV radiation. LOF mutations in the MC1R were found to abolish the antiapoptotic effect of α-MSH (Kadekaro et al., 2005). These findings strongly suggest that the increased risk of skin cancer in individuals harboring the RHC alleles would result not only from a reduced filtering effect associated with lower amounts of eumelanin, but also from a decreased ability to activate DNA repair and antiapoptotic programs.

Regulation of MC1R gene expression

The effects on MC1R expression of a variety of endo- and paracrine factors known to influence pigmentation have been analyzed. The inducibility of MC1R mRNA in human melanocytes stimulated with NDP-MSH was described by Funasaka et al. (1998), who also reported a stronger activation by treatment with TPA (24 h). A similar increase in MC1R transcript levels was observed in human melanocyte cultures stimulated with α-MSH (Scott et al., 2002a). The induction was mimicked by the adenylyl cyclase activator forskolin, suggesting that it is cAMP dependent. Conversely, incubation of these cultures in the presence of ASIP produced a clear downregulation of MC1R transcript (Scott et al., 2002a). The transcription factor(s) responsible for this effect have not been fully identified. An obvious candidate is MITF, as it is a downstream target of MC1R signaling (Vance and Goding, 2004). The minimal MC1R promoter contains an E-box (CATGTG) immediately upstream of the transcriptional initiation site (Moro et al., 1999). The E-box matches the core sequence of the M-box (AGTCATGTGCT) which is the target sequence for MITF, and MC1R promoter activity was stimulated fivefold in the presence of MITF (Aoki and Moro, 2002). However, this increase is modest compared with the stronger stimulation of the Tyrosinase promoter activity, a well-defined target of MITF. Moreover, another report failed to detect significant MITF-dependent activation of the MC1R promoter mediated by the E-box (Smith et al., 2001). Therefore, the involvement of MITF in the upregulation of MC1R expression by the MCs is uncertain.

MSH also increases MC1R mRNA levels in immortalized mouse melanocytes (Rouzaud et al., 2003). In addition, ligand-dependent modulations of the 5′ untranslated region of Mc1r mRNA were reported. RACE analysis revealed two different Mc1r transcripts in MC-stimulated melanocytes (T1 and T2) and a third one specifically associated with ASP-treated cells (T3), with untreated cells expressing the T1 form. The translation efficiency was found higher for the α-MSH-inducible T2 form, intermediate for the T1 basal transcript, and severely repressed in the ASP-associated T3. Therefore, α-MSH may increase Mc1r gene expression in mouse melanocytes by transcriptional and translational activation, whereas ASP might repress Mc1r mainly at the translational level (Rouzaud et al., 2003). In mouse hair follicles, Pomc expression undergoes cyclic changes during the hair cycle (Ermak and Slominski, 1997). Moreover, as discussed above a pulse of ASP expression during the mouse hair cycle causes a switch from eu- to pheomelanogenesis. Thus, it will be interesting to see whether those changes are associated with a differential expression of MC1R transcripts.

MC1R gene expression is regulated by UV light. Early work from John Pawelek's group demonstrated that exposure to UVB resulted in increased binding of 125I-MSH to Cloudman S91 mouse melanoma cells (Bolognia et al., 1989). Moreover, UV radiation apparently resulted in a redistribution of binding sites between the plasma membrane and an internal pool (Chakraborty et al., 1991). The same group demonstrated that the stimulatory effect is at least partially due to increased mRNA levels, suggesting a transcriptional activation (Funasaka et al., 1998), and extended these findings to normal human keratinocytes (Chakraborty et al., 1999). The inducibility of MC1R by UV radiation has been confirmed in cultured cells (Corre et al., 2004) and in human epidermis in vivo (Schiller et al., 2004). Interestingly, ectopic expression of POMC in melanocytes or other skin cell types is also increased by UV (Chakraborty et al., 1996; Corre et al., 2004; Im et al., 1998). This suggests that the tanning effect of UV light is largely dependent on the α-MSH/MC1R system and is exerted at the level of both the receptor and its activatory ligand(s) (Im et al., 1998). Transcriptional activation of POMC and MC1R is mediated by p38 stress-activated kinase and seems dependent on the phosphorylation of the ubiquitous upstream stimulating factor 1 (USF1) transcription factor, as it is absent in USF1 knockout melanocytes and is rescued upon transfection of the cells with the WT gene. The effect is most likely due to the binding of p38-phosphorylated USF1 to the conserved E-box element at −461 in the MC1R promoter (Corre et al., 2004). Interestingly, basal Pomc and Mc1r expression levels may be modulated by Mitf but seem USF1 independent, as they are similar in WT or USF1-null melanocytes, and are not upregulated by transfection of USF1 null melanocytes with the WT transcription factor (Corre et al., 2004). Studies performed with cultured mast cells further support a role of Mitf in the basal transcription of Mc1r. Mast cells derived from mice homozygous for a Mitf-inactivating mutation do not express Mc1r, and expression can be rescued by transfection with a WT Mitf construct (Adachi et al., 2000). Therefore, it appears that basal MC1R expression is regulated by MITF. Stimulation of this transcription factor by the MCs may lead to some degree of transcriptional activation as the E-box present in the MC1R proximal promoter is most likely a low affinity target for MITF.

A number of paracrine factors produced by keratinocytes such as endothelin 1 (EDN1) and basic fibroblast growth factor (bFGF), act through their corresponding receptors on the plasma membrane of melanocytes to increase proliferation and/or differentiation. EDN1 mediates a dose-dependent upregulation of MC1R mRNA in normal human melanocytes (Scott et al., 2002a). The effects of bFGF are less clear-cut, although a consistent and moderate upregulation has been reported for one melanocyte cell culture (Scott et al., 2002a). The potential regulatory effects of a number of cytokines known to contribute to cutaneous homeostasis have also been considered (Funasaka et al., 1998). Interleukin-1α and interleukin-1β were found as effective as EDN1 in upregulating MC1R mRNA in normal human melanocytes. Two cytokines that potently repress melanogenesis in melanoma cells, TNF-α (Martinez-Esparza et al., 1998) and TGF-β (Martinez-Esparza et al., 1997), downregulate moderately MC1R expression in normal melanocytes (Funasaka et al., 1998) and mouse melanoma cells (Martinez-Esparza et al., 1999). Interestingly, non-cytotoxic concentrations of hydrogen peroxide (H2O2) potently and reversibly repress Mitf and Mc1r expression in mouse melanoma cells (Jimenez-Cervantes et al., 2001b), and certain effects of TNF-α and TGF-β are mediated by reactive oxygen species (Garg and Aggarwal, 2002). As H2O2 is a byproduct of the melanogenic pathway (Nappi and Vass, 1996) and reactive oxygen species are formed in the skin following UV irradiation (Nishigori et al., 2004), it has been speculated that an H2O2-based negative feedback mechanism acting at the MITF and MC1R level may contribute to the fine tuning of the melanogenic rate under conditions of hormonal or solar stimulation (Jimenez-Cervantes et al., 2001b).

MC1R receptor desensitization and internalization

Many GPCRs undergo homologous desensitization, i.e. a strong attenuation of signaling within minutes of agonist exposure (Luttrell and Lefkowitz, 2002; Pitcher et al., 1998). Homologous desensitization is normally dependent on phosphorylation of the agonist-occupied receptor by members of a family of specific kinases termed GPCR kinases (GRKs) (Luttrell and Lefkowitz, 2002; Pitcher et al., 1998). Phosphorylated receptors become unable to couple to their signaling pathways. Within the MCR family, homologous desensitization has been reported for Mc4r in mouse hypothalamic GT1-7 cells (Shinyama et al., 2003), Mc2r in mouse adrenocortical Y1 cells (Baig et al., 2001), and Mc1r and MC1R in mouse and human melanoma cells, and in heterologous expression systems (Sanchez-Mas et al., 2005a). Adaptation of MC1R and Mc1r is likely mediated by GRK2 or GRK6. These ubiquitous and promiscuous members of the GRK family are expressed in melanoma cells and normal melanocytes. Cotransfection of either one of the kinases and receptor genes in HEK 293T cells strongly impairs signaling. Moreover, expression of a GRK2 dominant negative mutant in human melanoma cells stimulates agonist-dependent cAMP production, whereas overexpression of GRK6 in those cells reduces the cAMP response (Sanchez-Mas et al., 2005a).

Phosphorylation of GPCRs by GRKs is most often followed by internalization of the receptor-agonist complexes in endocytic vesicles, a process that may trigger new signaling events (Luttrell and Lefkowitz, 2002; Pitcher et al., 1998). Internalization may target the receptor for lysosomal degradation or be followed by recycling back to the plasma membrane. Phosphorylation-dependent internalization has been demonstrated for MC4R (Shinyama et al., 2003) and it can be speculated that agonist binding and the resulting phosphorylation by the GRKs also triggers the sequestration of the MC–MC1R complex. This would account for the presence of internal MCs in melanoma cells following their addition to the external medium, and for the time-dependent appearance of acid-resistant binding (Adams et al., 1993; Siegrist and Eberle, 1993; Varga et al., 1976; Wong and Minchin, 1996). Moreover, the possibility that constitutively active receptors are subject to constitutive phosphorylation and cycling between internal compartments and the plasma membrane has been recently considered and demonstrated for certain GPCRs such as active variants of the angiotensin II AT1A receptor (Miserey-Lenkei et al., 2002). Based on the high constitutive activity of MC1R (Sanchez-Mas et al., 2004) and on its inhibition by GRK6 (Sanchez-Mas et al., 2005a), it can be speculated that a GRK-dependent cycling mechanism might be operative for MC1R.

Although there is little doubt as to the agonist-dependent internalization of MC1R, the fate of the internalized complex remains uncertain. In a study of binding and internalization of a radiolabeled agonist in B16 mouse melanoma cells, it was shown that the ligand is rapidly internalized and translocated to the lysosomes, where it is degraded (Wong and Minchin, 1996). No evidence of receptor recycling back to the plasma membrane was found, suggesting that the protein may also be degraded. Careful binding studies have shown that long exposures to MCs decrease the number of α-MSH binding sites in certain melanoma cell lines (Siegrist et al., 1994). However, the response of human melanoma cells to long treatments with various MCs is heterogeneous: certain melanoma cell lines display an upregulation of the number of binding sites after 2 d of treatment, whereas others show a moderate decrease in receptor density, or are unresponsive (Siegrist et al., 1994). As the MCs stimulate MC1R gene expression, the number of binding sites in the plasma membrane of melanocytes continuously exposed to the agonist may depend on the balance between two opposing effects, receptor sequestration (and probably degradation) on one hand, and activation of gene expression on the other. The relative magnitude of these opposing effects may be variable from one cell line to another, thus accounting for the observed heterogeneity in the final response.

Conclusions and perspectives

Since the cloning of the MC1R cDNA in 1992, it has been established that this gene is a major determinant of skin and hair pigmentation, sun sensitivity and susceptibility to skin cancer. Major progress has also been done in our understanding of MC1R structure and functional coupling. The steps of an MC1R cycle have been delineated (Figure 3). MC1R mRNA is translated in the rough ER, and several transcripts with different 5′ untranslated regions and translation efficiency have been described. The nascent protein retains its putative signal peptide, and is post-translationally modified by glycosylation, acylation, disulfide bond formation and dimerization. C-terminal structural elements including a dileucine-like motif and the conserved terminal tripeptide CSW are important for its traffic to the plasma membrane, via the Golgi apparatus. Intracellular traffic is impaired for several natural mutants such as L93R, R162P and the well-know RHC alleles R151C and R160W, resulting in decreased receptor density on the plasma membrane and diminished or abolished function. Other RHC alleles reach efficiently the plasma membrane, but display impaired agonist binding or decreased coupling efficiency. Functional coupling to adenylyl cyclase via the Gs protein is regulated positively by the MC ligands, and negatively by ASP, at least in the mouse. In addition, MC1R is a constitutively active receptor, displaying significant functional coupling in the absence of activatory ligands. Activated MC1R is uncoupled from the cAMP pathway by GRK2 and GRK6, in a process that likely involves phosphorylation of Ser and/or Thr residues located in the C-terminal tail. The ligand–receptor complex is sequestered within endocytic vesicles. On the contrary, MC1R gene expression is upregulated by UV light, the MCs and by paracrine factors such as END1, interleukin-1α and -1β. Stimulation by UV radiation involves p38 stress-activated signaling to USF-1 and is a key step in the tanning response. Conversely, MC1R gene expression is repressed by TNF-α, TGF-β and H2O2. This complex regulation demonstrates an elaborate network of relationships between the different types of skin cells contributing to skin homeostasis.

Figure 3.

A model for the regulation of melanocortin-1 receptor (MC1R) expression and signaling. MC1R gene expression is upregulated by ultraviolet light, the melanocortin (MC) agonists, END1, and interleukin-1α and -1β, and repressed by TNF-α, TGF-β, and H2O2 (1). In mouse melanocytes, the MCs and agouti protein may induce specific transcripts with different 5′ untranslated regions and translation efficiencies. MC1R mRNA leaves the nucleus (2) and is translated by endoplasmic reticulum associated ribosomes (3). The protein is exported to the Golgi (4), and then to the plasma membrane (5), likely as a dimeric species. This intracellular traffic is blocked for certain variant alleles that do not acquire a transport-competent conformation and are retained in a still undetermined intracellular compartment. Receptor signaling is regulated by the MC agonists (6). Coupling to adenylyl cyclase via the Gs protein is inhibited by GRK2- or GRK6-dependent phosphorylation, likely in one or more of the Ser or Thr residues located in the C-terminal tail of the protein (7). The phosphorylated receptor is sequestered by internalization in endocytic vesicles (8). The fate of the internalized receptors remains unknown.

Several aspects of this model will be refined in the near future. The occurrence, origin and function of alternative transcripts will be assessed. The mechanisms of internalization and the fate of the receptor will be worked out, and the possible relationships of receptor sequestration with downstream signaling events will be clarified. The effects of MC1R dimerization on intracellular traffic, on the interactions of the receptor with specific components of the plasma membrane and the internalization machinery and on pharmacologic properties are also actively investigated. Given the high frequency of heterozygotes carrying at least one mutant allele, knowledge of the functional consequences of dimerization will allow for a better understanding of the relationship of MC1R genotype with cutaneous phenotypes.

Acknowledgements

Work in the author's laboratory is supported by grant SAF2003-03411 from the MCyT and FEDER. B.L. Sánchez-Laorden holds a FPI fellowship from the Ministry of Education, Spain. The authors are grateful to the following graduate students and visiting scientists who performed work on MC1R in their laboratory: J. Sánchez Más, C. Olivares Sánchez, M. C. Turpín, P. González, P. Zanna, I. Gerritsen, S. Germer, C. Hahmann and L.A. Guillo. The authors apologize to colleagues whose relevant work could not be cited for space limitations.

Ancillary