Ocular albinism type 1 is an X-linked disorder characterized by severe reduction of visual acuity, retinal hypopigmentation, foveal hypoplasia, optic misrouting and the presence of giant melanosomes (macromelanosomes) in skin melanocytes and retinal pigment epithelium. The protein product of the OA1 gene is a pigment cell specific membrane glycoprotein, displaying structural and functional features of G protein-coupled receptors (GPCRs). However, in contrast to all other previously characterized GPCRs, OA1 is not localized to the plasma membrane, but is targeted to intracellular organelles, namely late endosomes/lysosomes and melanosomes. These unique characteristics suggest that OA1 represents the first example described so far of an exclusively intracellular GPCR and regulates melanosome biogenesis by transducing signals from the organelle lumen to the cytosol. These findings support previous hypotheses that GPCR-mediated signaling might also operate at the internal membranes in mammalian cells.
Ocular albinism type 1 (OA1, Nettleship-Falls type; MIM 300500) represents the most common form of ocular albinism, with an estimated prevalence of 1:50 000, and is transmitted as an X-linked trait, with carrier females showing only minor signs of the disease. Affected males exhibit optic abnormalities typical of all forms of albinism, including severe reduction of visual acuity because of foveal hypoplasia, nystagmus, strabismus, photophobia, iris translucency, hypopigmentation of the retina, and misrouting of the optic tracts, resulting in loss of stereoscopic vision (King et al., 1995). Histological analysis reveals the presence of giant melanosomes (macromelanosomes) in skin melanocytes and retinal pigment epithelium (RPE), suggesting an underlying defect in melanosome biogenesis (Garner and Jay, 1980; O'Donnell et al., 1976; Wong et al., 1983).
The gene responsible for ocular albinism was isolated by a classical positional cloning strategy from the distal short arm of the X chromosome (Bassi et al., 1995). OA1 is expressed exclusively in the pigment cells of the skin and eyes, consistent with the clinical phenotype of the disease, and encodes a predicted protein of 404 amino acids (Bassi et al., 1995). At the time of gene cloning, no obvious homologies were detected with other proteins in the database by conventional Blast searches (Bassi et al., 1995). Subsequently, the mouse ortholog was identified, displaying a 78% identity at the protein level with the human gene and the same expression pattern (Bassi et al., 1996; Newton et al., 1996). OA1 expression is regulated similarly to that of melanogenic enzymes, such as tyrosinase (TYR). In fact, expression of OA1 parallels temporally and spatially that of TYR during development, is regulated by the transcription factor Mitf, is upregulated by the α-Melanocyte Stimulating Hormone (α-MSH) and is inhibited by its antagonist Agouti Signal Protein (ASP) (Samaraweera et al., 2001; Surace et al., 2000; Vetrini et al., 2004).
The biology of ocular albinism and the function of the OA1 protein have been only partially elucidated by a number of dedicated studies. In this review, we will focus on the human OA1 and the mouse Oa1 proteins, trying to summarize our present knowledge on the unique structural and functional features of these enigmatic molecules and to withdraw more general implications for cell biology.
Sequence, structure and homologies of OA1
As more sophisticated bioinformatic tools became available, important similarities were identified between OA1 and the superfamily of G protein-coupled receptors (GPCRs) (Schiaffino et al., 1999). First, OA1 showed the presence of seven putative transmembrane alpha helices, as predicted by the PHDhtm program (EMBL, Heidelberg). Second, the Blast2 and FastA algorithms revealed appreciable homologies between OA1 and members of families A, B, and E of the GPCR superfamily. Most of conservation is restricted within the transmembrane domains (TM). In addition, OA1 contains several residues highly conserved in most GPCRs, including cysteines in the first and second putative lumenal/extracellular loops (e1 and e2); a DRY-like motif at the end of TM3; aspartic acid, tryptophan and proline in the middle of TM2, 4 and 5, respectively; and sequential cysteine tryptophan in the middle of TM6 (Figure 1) (Schiaffino et al., 1999). Importantly, the critical role of some of these highly conserved residues is demonstrated by the presence of mutations in patients with ocular albinism (Figure 1) (d'Addio et al., 2000; Schiaffino et al., 1999).
Considering that GPCRs belonging to different families do not share major primary sequence similarities, but rather a common three-dimensional structure that requires the presence of critical residues at precise positions (Gether, 2000; Pierce et al., 2002), the sequence homologies between OA1 and GPCRs viewed in the context of the seven spanning membrane topology are highly significant. Alignment between OA1 and these receptors indicates that it is equally distant from all three families, possibly defining a new GPCR subfamily (Schiaffino et al., 1999). In addition, orthologs of OA1 have now been identified in several species. OA1 is significantly conserved in mammals, birds, amphibians and fishes, particularly at the N-terminus and from the third to the seventh transmembrane domains, while the C-terminal tail and the first cytosolic and lumenal/extracellular loops (i1 and e1) display higher divergence (Figure 1).
Biochemical features of OA1
The endogenous OA1 protein was identified by Western immunoblotting in normal human melanocyte extracts as a doublet of 45 and 48 kD and a broad band of 60 kD (Schiaffino et al., 1996). Demonstration that these represent the relevant protein species was confirmed by their detection in COS7 cells transfected with expression vectors for OA1 (d'Addio et al., 2000; Schiaffino et al., 1996) and by their absence in melanocyte extracts from patients with ocular albinism (d'Addio et al., 2000; Schiaffino et al., 2002). Studies in both melanocytes and transfected COS7 cells established that the 45 kD band corresponds to the OA1 protein backbone, the 48 kD polypeptide to a yet unknown post-translational modification, and the 60 kD band to the N-glycosylated form of both the former protein species (d'Addio et al., 2000; Schiaffino et al., 1996).
Glycosylation of OA1 initiates in the endoplasmic reticulum (ER), where it can be inhibited by tunicamycin, and is completed in the Golgi apparatus, where it can be prevented by deoxymannojirimycin (d'Addio et al., 2000). Despite a unique putative N-glycosylation site located in the first lumenal/extracellular loop (N106, in e1; a second predicted site, N263, is embedded within TM VI), glycosylation results in a relevant molecular shift compared with the protein backbone (about 15 kD). Finally, upon solubilization in Triton X-114, OA1 segregates in the detergent phase, behaving as a typical integral membrane protein (Schiaffino et al., 1996). Similar results were obtained with the murine Oa1 protein. Oa1 was found to migrate predominantly as a 48 kD glycosylated form, which could be reduced to a 44–46 kD doublet by treating melanocytes with tunicamycin or the protein extracts with glycopeptidase F. Phase separation in Triton X-114 showed that Oa1 fractionates in the detergent phase, together with integral membrane proteins (Samaraweera et al., 2001), and specific mutagenesis studies determined that glycosylation occurs at a unique conserved site, N106, as postulated for the human protein (Shen and Orlow, 2001a).
Together, these results indicate that OA1 is an integral membrane glycoprotein of the exocytic pathway, consistent with its homologies to GPCRs. The most typical feature of these receptors is represented by their ability to transduce signals by coupling with heterotrimeric G proteins in response to agonist binding. Accordingly, the endogenous OA1 was found to co-precipitate with the Gαi, Gαo and Gβ subunits of heterotrimeric G proteins in melanocyte extracts (Schiaffino et al., 1999). Similar results were obtained in COS7 cells transfected with the recombinant wild-type OA1, but not with mutants carrying amino acid substitutions identified in patients with ocular albinism (Schiaffino et al., 1999). In addition, the third cytosolic loop (GST-i3), and the C-terminal tail of OA1 (6xHis-CT) were able to bind Gαi, Gαq/11, and Gβ by in vitro binding assays, while no interaction was observed with other Gα subunits or unrelated intracellular proteins (Schiaffino et al., 1999). These findings suggest that OA1 functionally behaves as a canonical GPCR.
Subcellular localization of OA1
In contrast to canonical GPCRs, OA1 is not localized to the cell surface, but is exclusively detectable on the membrane of intracellular organelles. In fact, immunofluorescence and immunogold analyses, as well as in vivo cell surface biotinylation performed in normal human melanocytes showed that OA1 is excluded from the plasma membrane (Schiaffino et al., 1999). However, some controversy exists on the precise subcellular localization of the protein. By double immunofluorescence analysis in normal human melanocytes, the endogenous OA1 was found to co-localize with the melanosomal matrix glycoprotein gp100/Pmel17 (Schiaffino et al., 1999, 2002), which labels melanosomes at early developmental stages (Raposo et al., 2001). The proteomic analysis of a purified early melanosome fraction obtained from the human melanoma line MNT1 confirmed that OA1 is associated to these organelles (Basrur et al., 2003). In addition, immunogold and immunoperoxidase analyses at the ultrastructural level in normal human melanocytes demonstrated that OA1 is also present on melanosomes at later stages of maturation, where gold particles were observed both on the cytosolic side and in the interior of the organelles, suggesting that part of the protein might be internalized in intralumenal vesicles (Schiaffino et al., 1996, 1999).
By contrast, studies on the mouse ortholog showed that the endogenous Oa1 and the recombinant OA1-green fluorescent protein (GFP) fusion protein in melan-c and melan-a mouse melanocytes, respectively, co-localize widely with LAMP1, but only modestly with TRP1, a marker of mature melanosomes (Samaraweera et al., 2001). In addition, upon density gradient fractionation of subcellular organelles, Oa1 was not found to peak at the dense fractions as tyrosinase and TPR1, suggesting that it is mostly an endo-lysosomal protein, with minimal or no co-localization with melanosomes (Samaraweera et al., 2001). The reason for these discrepancies might partially lie in the different techniques, markers, and cell models utilized. In order to solve this issue, we recently performed quantitative immunofluorescence and immunogold analyses of the OA1 distribution in MNT1 cells. Our results indicate that OA1 is similarly distributed along both the late endosomal/lysosomal and the melanosomal compartments in melanocytic cells (M.V. Schiaffino and C. Tacchetti, unpublished results).
However, it should be underlined that, in agreement with previous observations (Schiaffino et al., 1996), morphometric analysis of immunogold-labeled pigmented melanosomes revealed that most of OA1 is associated to intralumenal vesicles at late melanosome maturation stages (K. Cortese, personal communication). As suggested for Pmel17 (Donatien and Orlow, 1995; Raposo et al., 2001), the presence of melanin at this location might prevent the accessibility of epitopes to antibodies. Furthermore, in consideration of the absorption spectra of melanin, annihilation of fluorescence by the pigment may occur. Thus, the low co-localization between Oa1 and markers of mature melanosomes by immunofluorescence analysis (Samaraweera et al., 2001) may be because of technical biases. In addition, melanin has been shown to reduce the detergent solubility of melanosomal matrix proteins, such as Pmel17, but not of tyrosinase and TRPs (Donatien and Orlow, 1995). Given the lumenal localization of OA1 in mature melanosomes, its apparent failure to co-sediment with tyrosinase and TRP1 in the Percoll gradient dense fractions (Samaraweera et al., 2001) could result from its poor extractability in the presence of high pigment concentrations.
Nothing is known about the sorting mechanism driving OA1 to its compartment, except that glycosylation does not seem to be required (Shen and Orlow, 2001a). Furthermore, as typically observed with other melanosomal proteins including tyrosinase and TRP1 (Winder et al., 1993), the sorting signals of OA1 are also recognized in non-melanocytic cells, where they function to deliver the heterologously expressed human and mouse proteins to late endosomes and lysosomes (Schiaffino et al., 1999; Shen et al., 2001b). The topological orientation of OA1 on the organelle membrane has not been directly investigated. However, the homologies with GPCRs (Schiaffino et al., 1999), the presence of a glycosylated asparagine in the predicted first lumenal loop (Schiaffino et al., 1996; Shen and Orlow, 2001a) and the localization of the C-terminal tail to the cytoplasmic side by immunoperoxidase staining at the ultrastructural level (Schiaffino et al., 1999) suggest that OA1 might be inserted in the melanosomal-lysosomal membrane with the N-terminus toward the lumen and with the C-terminus toward the cytoplasm. Thus, by comparison with receptors at the plasma membrane, a putative ligand should bind OA1 on the lumenal side of the organelles.
Disease-causing mutations of OA1
Several types of mutations have been identified in patients with ocular albinism, including large and small deletions, frameshifts and stop codons (Bassi et al., 2001; Rosenberg and Schwartz, 1998; Schiaffino et al., 1995; Schnur et al., 1998), suggesting that the disease is because of a loss-of-function mechanism. Many missense mutations were also identified of unknown functional significance. To shed light into the molecular pathogenesis of the disorder and possibly define critical functional domains within the OA1 protein, 19 missense mutations were introduced within the human cDNA and expressed in COS7 cells (Figure 1 and Table 1) (d'Addio et al., 2000). When the different OA1 mutants were compared with the wild-type protein by Western and immunofluorescence analyses, two main behaviors were observed. About three of five of mutants were retained in the ER, showing defective glycosylation and, when expressed at low levels or endogenously in melanocytes, reduced yield (d'Addio et al., 2000). These results imply a major sensitivity of OA1 to ER-mediated retention and degradation and indicate that protein misfolding plays a major role in the pathogenesis of ocular albinism.
The remaining two of five of mutants displayed processing and sorting behaviors indistinguishable from the wild-type protein (d'Addio et al., 2000). Significantly, most of these mutations are located within or close to the putative second and third cytosolic loops of OA1, two regions that in canonical GPCRs are known to play a critical role in G protein-coupling (Bourne, 1997). Thus, it is conceivable that these mutations interfere with the functional activity of OA1. Similar results were obtained by introducing 13 missense mutations within the GFP-tagged mouse protein, given that the majority of mutated amino acids are conserved in the mouse (Figure 1 and Table 1) (Shen et al., 2001b). In this case, some mutations causing ER retention of the human OA1 induced a mixed ER-lysosomal distribution when inserted in the mouse Oa1 (Table 1), suggesting a delayed exit from the ER. The only relevant discrepancy between the human and mouse studies regards the W133R mutation, which determines ER retention of the human protein, but normal lysosomal sorting of its mouse counterpart (Table 1). Despite the high homology between the human and mouse proteins, it is possible that some mutations exert their complete effect only on the human background. Alternatively, the GFP tag might modify the consequences of some mutations, even though the sorting behavior of Oa1-GFP appears identical to the wild-type protein when expressed in melanocytes (Samaraweera et al., 2001).
Function of OA1: activation mechanism and downstream pathway
The exclusive intracellular localization of OA1 and its accumulation into specialized organelles such as the melanosomes are unprecedented among GPCRs and suggest that OA1 might transduce information from the organelle lumen to the cytosol to regulate proper melanosome biogenesis. OA1 could function as a ‘sensor’ of melanosome maturation and inhibit organelle overgrowth, triggering a signal transduction cascade through activation of heterotrimeric G proteins on the cytoplasmic side of the melanosomal membrane (Figure 2) (Schiaffino et al., 1999). Consistent with this idea, the G protein subunits Gαi and Gβ were found associated to melanosomes by double immunofluorescence staining with melanosomal markers and to co-localize with OA1 by double immunogold analysis (Schiaffino et al., 1999). Based on the predicted membrane topology of OA1 (see above), its ligand should come from the melanosomal lumen and might be represented by melanin itself or by melanin precursors and intermediate products. Alternatively, OA1 might be activated by less conventional mechanisms, such as organelle enlargement, which in the absence of OA1 signaling could lead to macromelanosome formation. Finally, we cannot exclude that OA1 behaves as a constitutively active receptor and, therefore, that its ligand actually acts as an inverse agonist or inhibitor.
The downstream pathway triggered by OA1 remains mysterious and at present can only be hypothesized based on the consequences arising from loss-of-function of the receptor in humans and mice. The histological hallmark of OA1 is the presence of macromelanosomes in the skin and eyes of patients with ocular albinism, suggesting a defect in melanosome biogenesis (Garner and Jay, 1980; O'Donnell et al., 1976; Wong et al., 1983). Despite the cytological abnormalities, the skin typically shows normal pigmentation. In contrast, the RPE appears hypopigmented, possibly because melanin concentrates into few macromelanosomes instead of being evenly dispersed by means of many normal size organelles. In fact, cells containing giant melanosomes also show a decreased number of normal melanosomes (Garner and Jay, 1980; Wong et al., 1983). In order to study the pathogenesis of ocular albinism, an Oa1 mouse knockout (KO) has been generated (Incerti et al., 2000). Analysis of the KO mice revealed that they represent a good model of the human disease, displaying hypopigmentation of the fundus, misrouting of the optic fibers at the chiasm and the presence in the RPE of giant melanosomes comparable with those described in patients with ocular albinism (Incerti et al., 2000).
In the RPE of KO mice macromelanosomes are first detectable after birth (P1) and their number increases progressively until adulthood. The observation of single core melanosomes within the giant pigmented organelles and the absence of intermediates of melanosome–melanosome fusion claim against the idea that macromelanosomes derive from fusion between individual organelles. In contrast, they suggest an origin by abnormal overgrowth of apparently normal mature melanosomes, which are unable to control their final size and progressively grow by addition of new melanin and membranes (Incerti et al., 2000). Based on these results, Oa1 appears to play an inhibitory role restricted to the final stages of melanosome maturation. Together, the human and mouse OA1-null phenotypes indicate that loss of the receptor's activity perturbs the size of melanosomes and perhaps also their rate of biogenesis. As a possible scenario to explain these effects, Oa1 could modulate the resources (enzymes, membranes, melanin precursors) allocated to developing melanosomes via membrane traffic. A progressive reduction of these resources to mature melanosomes would maintain their normal size and favor the development of immature melanosomes. When this control is lacking, mature melanosomes would continue to increase, subtracting resources to and consequently reducing the number of new melanosomes.
A role of Oa1 in the regulation of membrane traffic toward the melanosomes has been previously proposed by S. Orlow and coworkers, based on the finding that expression of Oa1-GFP in COS7 cells induces perinuclear clustering of the mannose-6-phosphate receptor (M6PR) compartment, while Oa1 mutants do not show this capability (Shen et al., 2001b). Oa1 was hypothesized to direct vesicular traffic from late endosomes to mature melanosomes, so that in its absence either the late endosomal compartment continues to enlarge to acquire the characteristics of macromelanosomes, or it promotes continued traffic to mature melanosomes to eventually generate the giant organelles (Shen et al., 2001b,c). A responsibility of OA1 in supervising membrane traffic between lysosomes and melanosomes would explain why the macromelanosomal phenotype is more severe in the RPE than in skin melanocytes. In fact, the former is involved in continuous phagocytosis and degradation of photoreceptor outer segments and therefore needs efficacious mechanisms to segregate lysosomes from adjacent organelles.
As its downstream pathway, the role of OA1 in the development of the visual system is poorly understood. The misrouting of the optic tracts, at least in the mouse, develops much earlier in embryonic life (between E12.5 and E16) than the appearance of macromelanosomes (Incerti et al., 2000). Therefore, the giant organelles might represent only an epiphenomenon of the disease and cannot be the direct cause of the optic defect, which must be caused by some other alteration of the melanogenic process. In contrast to other forms of albinism, melanin is not dramatically reduced in ocular albinism, which is characterized by the unusual coexistence of the typical albino visual defects and the presence of a substantial amount of melanin in the eyes (O'Donnell et al., 1976). The possibility that OA1 functions as a melanin receptor suggests that this protein might represent the missing link between melanin deficiency in oculocutaneus albinism (OCA) type 1 and 2, and the severe developmental abnormalities of the visual system observed in all types of albinism. Therefore, identification of the ligand/s of OA1 and unraveling of its downstream pathway could lead to crucial insights on both melanocyte biology and retinal development.
OA1 as a model for GPCR-mediated signaling at internal membranes
A number of studies have revealed that heterotrimeric G proteins, best known as plasma membrane transducers, are also widely distributed in the internal membranes of mammalian cells (Nurnberg and Ahnert-Hilger, 1996) and are involved in various membrane trafficking processes along the secretory and endocytic pathways. These include formation of constitutive secretory vesicles and immature secretory granules from the trans Golgi network (TGN) (Leyte et al., 1992); regulated secretion in exocrine (Ohnishi et al., 1997), endocrine (Konrad et al., 1995) and neuroendocrine cell types (Vitale et al., 2000); and transmitter storage in neurons and platelets (Ahnert-Hilger et al., 2003). At the plasma membrane, heterotrimeric G proteins are part of the most common signal transduction system by which cells respond to extracellular stimuli (Pierce et al., 2002). In the presence of a specific ligand, they become activated by GPCRs and in turn bind to and modulate the activity of a wide array of effectors.
Consequently, the presence of heterotrimeric G proteins associated with subcellular organelles has suggested the existence, at intracellular locations, of GPCR-mediated signal transduction systems similar to those observed at the plasma membrane (Jamora et al., 1997; Leyte et al., 1992; Nurnberg and Ahnert-Hilger, 1996). However, no GPCR that is primarily localized to internal membranes has been described to date. Based on the structural, biochemical and morphological evidence summarized above, OA1 might represent the first example identified so far of an exclusively intracellular GPCR (Schiaffino et al., 1999). Therefore, at present the study of the OA1 pathway represents a unique model for unraveling not only the biological bases of ocular albinism, but also the mechanisms of signal transduction possibly mediated by GPCR-G protein systems at the internal membranes.
We thank Drs Paola Bagnato, Valeria Marigo and Rosanna Piccirillo for critical reading of the manuscript and Dr Katia Cortese for personal communication on OA1 intralumenal localization in highly pigmented melanosomes. This work was supported in part by the generous donation of the Vision of Children Foundation-San Diego, the National Institutes of Health (grant no. 5R01EY014540 from the National Eye Institute), and Telethon-Italy (grant no. F3) to M.V.S. and by Telethon-Italy (grant no. GTF03001) and MIUR (Minister of University and Research) PRIN program to C.T.