Epitope Mapping of the Melanosomal Matrix Protein gp100 (PMEL17)

Melanosomes, specific organelles produced only by melanocytes, undergo a unique maturation process that involves their transition form amorphous rounded vesicles to fibrillar ellipsoid organelles, during which they move from the perinuclear to the distal areas of the cells. This depends upon the trafficking and processing of gp100 (also known as Pmel17 and the silver protein), a protein of great interest, because it elicits immune responses in melanoma patients but in which specific function(s) remains elusive. In this study, we have used biochemical and immunochemical approaches to more critically assess the synthesis, processing, glycosylation, and trafficking of gp100. We now report that gp100 is processed and sorted in a manner distinct from other melanosomal proteins (such as tyrosinase, Tyrp1 and Dct) and is predominantly delivered directly to immature melanosomes following its rapid processing in the endoplasmic reticulum and cis-Golgi. Following its arrival, gp100 is cleaved at the amino and at the carboxyl termini in a series of specific steps that result in the reorganization of immature melanosomes to the fibrillar mature melanosomes. Once this structural reorganization occurs, melanogenic enzymes begin to be targeted to the melanosomes, which are then competent to synthesize melanin pigment.

Melanosomes are proving to be rich resources for several major lines of scientific research. First, because they have such distinctive and readily identifiable structural characteristics, they are ideal models for studying mechanisms involved in the biogenesis of subcellular organelles (1)(2)(3)(4)(5). Second, because a number of melanosome-specific proteins are localized in those organelles, the processing and trafficking of those proteins via cellular sorting pathways are other fruitful areas of active research (6 -8). Third, a number of inherited human pigmentary diseases, such as oculocutaneous albinism, Hermansky-Pudlak syndrome, and Griscelli syndrome, involve aberrant processing and/or sorting of proteins to melanosomes or, as in the case of Griscelli syndrome, abnormal movement of melanosomes within melanocytes (9 -11). Finally, all six known melanosomal proteins serve as specific targets of the immune system and are involved in immune responses to malignant melanoma, vitiligo, and other pigmentrelated diseases (12)(13)(14).
Taken together, the melanosome thus serves as a focal point for research from all angles of the scientific spectrum. Our group has been involved in several aspects of such research, and one recent project has begun to unveil the proteome of early melanosomes (15). The previous study revealed that although melanosomes have several unique constituent proteins, they also share common protein components with the lysosome/ endosome compartment, with the endoplasmic reticulum (ER) 1 compartment, and with secretory and synaptic vesicles, much of which had been anticipated based on biochemical and ultrastructural studies and/or upon pleiotropic effects of various pigmentary diseases on other tissues and organelles.
Melanosomes undergo perhaps the most dramatic maturation process of subcellular organelles, starting life as amorphous rounded vesicles in the perinuclear area and then transforming into fibrillar, ellipsoid organelles in which melanin is synthesized and deposited (1,2,8). The transition from the earliest vesicular form, termed a Stage I melanosome, to the fibrillar Stage II melanosome is dependent upon the trafficking and processing of gp100 (also known as Pmel17 and the silver protein) (1,16). gp100 was initially identified as a melanoma-specific antigen (17,18) and was cloned during an attempt to isolate the gene encoding tyrosinase (19), the enzyme essential to melanin biosynthesis. gp100 has proven to be an enigmatic protein of great interest because it elicits some of the strongest humoral and cellular responses in malignant melanoma patients, yet its specific function in the melanosome has remained difficult to ascertain. Several recent studies (1,15,16,20) show that gp100 undergoes a proteolytic cleavage from its full-length 100-kDa form into several smaller fragments and that this cleavage results in the release of the amino-terminal fragment of gp100 into the lumen of the melanosome, whereupon it becomes (or is associated with) the fibrillar component that serves as the solid-state matrix upon which melanin is deposited after its synthesis. An early study by our group (21) shows that, in mouse melanocytes, gp100 is rapidly sorted to Stage I melanosomes without being processed from the trans-Golgi network through the classical endosomal pathway, but the specific steps involved in that process and whether that route is similarly taken by gp100 in human melanocytes have remained unknown.
To that end, we have now characterized the processing, glyco-sylation, and trafficking of the melanosomal protein gp100 in human melanocytes. Taken together, the results showed that gp100 is processed and sorted in a manner distinct from other melanosomal proteins and is predominantly delivered directly to Stage I melanosomes following its processing in the ER and cis-Golgi. Following its delivery there, gp100 was cleaved both at the amino and at the carboxyl termini in a series of specific steps that resulted in the transition of Stage I melanosomes to mature Stage II melanosomes, whereupon melanosomal enzymes were delivered to the organelles and pigment was synthesized.

MATERIALS AND METHODS
Cells and Antibodies Used-MNT-1 pigmented human melanoma cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) as described previously (1). SK-MEL-28 amelanotic human melanoma cells were obtained from the ATCC (Manassas, VA) and were cultured in minimum essential medium (Invitrogen) containing 10% fetal bovine serum, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 2.8 g/ml sodium bicarbonate, 2 mM L-glutamine, and penicillin/streptomycin.
Antibodies to the organelle markers Bip, EEA1, and GM130 were purchased from BD Transduction Laboratories (Lexington, KY). Horseradish peroxidase-linked anti-rabbit IgG (whole antibody) and horseradish peroxidase-linked anti-mouse IgG were from Amersham Biosciences.
Electrophoresis and Western Blotting-Cell extracts were prepared using the M-PER mammalian protein extraction reagent (Pierce), and protein concentrations were measured using the BCA protein assay (Pierce). The cell extracts were mixed with 2ϫ with Tris-glycine SDS sample buffer (Invitrogen), supplemented with 0.1% 2-mercaptoethanol, and boiled for 10 min. The samples were then separated by SDS-PAGE (8 -16% Tris-glycine gels, Invitrogen) and transferred to Immobilon-P transfer membranes (Millipore, Bedford, MA). The blots were blocked in 10% (w/v) nonfat dry milk in TBS-T buffer (10 mM Tris-HCl (pH 7.4), 137 mM NaCl, 0.1% (w/v) Tween 20) for 1 h at room temperature and then incubated with the appropriate first antibodies diluted (as noted in the legends to Figs. 1, 2, 4, and 5) in 2% nonfat dry milk in TBS-T for 2 h at room temperature. After five washes with TBS-T, the blots were incubated in horseradish peroxidase-linked anti-rabbit or anti-mouse whole antibodies (1:10,000) (Amersham Biosciences) in 2% nonfat dry milk in TBS-T for 2 h at room temperature. After five further washes with TBS-T, the immunoreactivities of the antibodies were detected using an ECLϩ Western blotting detection system (Amersham Biosciences) according to the manufacturer's instructions. The Bench-Mark prestained protein ladder (Invitrogen) was used to establish the molecular weight curve for the Western blotting.
Immunoprecipitation-Extracts of MNT-1 cells (400 g of protein in 400 l, prepared with M-PER) were incubated with mouse monoclonal antibodies raised against gp100 (25 l of HMB45, 5 g of HMB50, 5 g of HMB55, or 5 g of NKI/beteb) or with normal mouse IgG (5 g) as a control. After continuous mixing for 2 h at 4°C, 35 l of protein G-Sepharose suspended in M-PER was added and was subjected to further mixing for 2 h at 4°C. The antibody-antigen complexes were precipitated by centrifugation, and the supernatants were kept as the unbound fraction. The pellets were then washed six times with 800 l of M-PER. Finally, the absorbed proteins were eluted with 45 l of Tris-glycine SDS sample buffer with 0.1% 2-mercaptoethanol by heating at 95°C for 10 min. Each supernatant (20 g of protein in 20 l) was separated using 8 -16% Tris-glycine SDS gels and transferred to Immobilon-P membranes. Rabbit polyclonal antibodies (␣PEP13h, ␣mSiN, ␣PEP1h, and normal rabbit serum as a control) were used to detect gp100 precipitated with mouse monoclonal antibodies against gp100 as detailed in the legend to Fig. 3. The detection method was the same as described above.
Metabolic Labeling-Metabolic labeling and immunoprecipitation experiments were performed as follows. MNT-1 cells were cultured in 10-cm dishes using complete medium (1). Prior to labeling, the cells were incubated in methionine-and cysteine-free Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% dialyzed fetal bovine serum (Invitrogen) for 30 min at 37°C in a humidified incubator with 5% CO 2 . The cells were labeled for 30 min with 0.5 mCi of [ 35 S]methionine and cysteine mixture (Redivue Promix, Amersham Biosciences) in 1 ml of methionine-and cysteine-free Dulbecco's modified Eagle's medium containing 10% dialyzed fetal bovine serum. Following the labeling period, the isotope-supplemented medium was removed, and complete medium containing 1 mM unlabeled methionine was added. After suitable chase periods (as detailed in the legend to Fig. 3), the medium was removed, and the cells were washed three times with phosphate-buffered saline and then solubilized by adding 1 ml of M-PER with complete protease inhibitor mixture (Roche Applied Science). The 35 S-labeled cell lysates were precleared with 50 l of protein G-Sepharose 4 fast flow (Amersham Biosciences) for 2 h at 4°C with continuous mixing. The supernatants were collected by centrifugation at 4°C and then incubated with 15 l of rabbit polyclonal antibody (␣PEP7h, ␣PEP13h, ␣mSiN, or normal rabbit serum as a control) or 15 l of mouse monoclonal antibody (HMB45, HMB50, HMB55, or NKI/beteb at the dilutions noted in the figure legends) for 2 h at 4°C with mixing. The immune complexes were separated by incubation with 30 l of protein G-Sepharose 4 fast flow for 2 h at 4°C and were further washed twice with 1 ml of radioimmune precipitation assay buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% SDS, 1.0% deoxycholic acid, 1.0% Triton X-100 supplemented with complete protease inhibitor mixture), twice with 1 ml of radioimmune precipitation assay buffer supplemented with 2 M NaCl, and then twice with 1 ml of radioimmune precipitation assay buffer. The final pellets were resuspended in Tris-glycine SDS sample buffer with 2-mercaptoethanol and heated at 95°C for 5 min. The samples were separated on 8 -16% Tris-glycine SDS-polyacrylamide gels. After the electrophoresis, the gels were dried and exposed in a storage phosphor screen and then scanned with a STORM PhosphorImager (Amersham Biosciences) and analyzed with the ImageQuant program.
Glycosidase Digestions-Endoglycosidase H (Endo H) and peptide: N-glycosidase F (PNGaseF) were purchased from New England Biolabs (Beverly, MA). MNT-1 cell lysates (5 g) extracted with M-PER were subjected to each glycosidase digestion according to the manufacturer's instructions. Samples were digested with 1250 units of Endo H or of PNGaseF for 3 h at 37°C. After the digestion reaction, 1 g of each cell lysate was subjected to SDS-PAGE, and immunoreactive bands were detected by Western blotting using antibodies as indicated in the legend to Fig, 5.
Confocal Immunohistochemistry-For confocal immunohistochemistry, cells were plated in 2-well Lab-Tek chamber slides (Nalge Nunc International, Naperville, IL) 24 or 48 h before each experiment and then fixed in 4% paraformaldehyde for 15 min at 4°C. After washing in phosphate-buffered saline, the cells were permeabilized with 0.01% Triton X-100 for 3 min at room temperature (for GM130 and HMB45) or with 100% methanol for 15 min at 4°C (for Bip and EEA1). The cells were then incubated in phosphate-buffered saline containing 5% normal goat serum and 5% normal horse serum for 1 h at room temperature as a blocking step. The cells were then incubated in phosphatebuffered saline containing 5% normal goat serum and 5% normal horse serum and a mixture of monoclonal and polyclonal antibodies overnight at 4°C at the dilutions noted in the legend to Fig. 6. This was followed by incubation with the corresponding secondary antibodies, i.e. goat anti-rabbit IgG labeled with Texas Red (dilution, 1:100, Vector, Burlingame, CA) or horse anti-mouse labeled with fluorescein isothiocyanate (dilution, 1:100, Vector). Nuclei were counterstained with 4Ј,6diamidino-2-phenylindole (Vector). All preparations were examined with a confocal microscope (model LSM510, Zeiss) equipped with helium-neon, argon-krypton, and UV laser sources. Controls included sections stained as detailed above but omitting the first antibody.

Expression of gp100 in Pigmented and in Unpigmented
Melanoma Cells-Based on its predicted amino acid sequence, gp100 is initially translated as a native protein of ϳ70 kDa; it contains a signal sequence at its amino terminus and thus is expected to be immediately taken up into the ER. gp100 has five potential N-glycosylation sites (at positions 81, 106, 111, 321, and 568), and because of its importance as a melanomaspecific antigen, a number of monoclonal antibodies have been generated in mice that recognize it at various epitopes, namely HMB45, HMB50, HMB55, and NKI/beteb (26), although the specific epitopes recognized by those monoclonals have not been adequately defined. As additional probes to study its synthesis and processing, we have produced antibodies in rabbits against peptides that correspond to the amino and carboxyl termini of gp100; these antibodies are termed ␣mSiN (21) and ␣PEP13h (27), respectively. We were initially interested in characterizing the expression, processing, and subcellular distribution of gp100 in several pigmented and unpigmented human melanoma cell lines and at the same time in mapping the various epitopes of gp100 recognized by those different monoclonal antibodies.
We initially compared the expression profile of gp100 in a pigmented melanoma cell line, MNT-1, and in a number of amelanotic melanoma cell lines, including SK-MEL-28 ( Fig. 1). SP1 keratinocytes were used as a negative control. ␣PEP13h recognized two major specific bands (100 kDa and 26 kDa) in MNT-1 cells, indicating that the carboxyl terminus of gp100 is present in those two bands. ␣PEP13h also detected a minor band at ϳ70 kDa. ␣mSiN reacted specifically with the fulllength 100-kDa protein band and also with 45-, 30-, 26-, and 10-kDa fragments, which contain the amino terminus of the protein (the 65-kDa fragment is nonspecific because it is also recognized in SP1 keratinocytes). Thus, both ␣PEP13h and ␣mSiN recognize the full-length 100-kDa form of gp100 in MNT-1 cells, and both antibodies also detect 26-kDa fragments, showing that gp100 is cleaved at a point about 26 kDa from its amino and its carboxyl termini. Interestingly, SK-MEL-28 amelanotic melanoma cells also express gp100, although their processing and PAGE banding patterns are quite distinct from those in MNT-1 cells (e.g. only faint traces of the 26-kDa bands were seen in those cells). Several other amelanotic melanoma cell lines did not express levels of gp100 detectable by Western blot (data not shown).
In contrast, the monoclonal antibody HMB45 reacted only with a low molecular weight heterogeneous fragment of gp100 that has an approximate size of 34 -38 kDa. MNT-1 cells reacted positively with HMB45 as did SK-MEL-28 cells at a reduced level, but reactivity with HMB45 was completely negative in SP1 keratinocytes (and in the other amelanotic melanoma cell lines tested). HMB50, HMB55, and NKI/beteb do not work well in the Western blot format, although they function well in immunohistochemistry and in immunoprecipitation formats, as discussed below.
For comparison, we also examined the expression of other melanosomal proteins in those same cell lines. Tyrosinase (Fig.  1), TYRP1, and MART1 (not shown) were strongly positive in the highly pigmented MNT-1 cells, as reported previously (1), and DCT was faintly positive in all melanoma cell lines tested (data not shown). The expression of tyrosinase and MART1 was positive in the amelanotic SK-MEL-28 melanoma cells, but tyrosinase levels in those cells are much reduced and are only partially glycosylated, as recently detailed (28).
To further determine the epitope map and reactivity patterns of the various antibodies to gp100, we performed immunodepletion studies in which extracts of MNT-1 cells were purified over immunoaffinity columns containing HMB45, HMB50, HMB55, or NKI/beteb (or IgG as a control). The flowthrough (unabsorbed) and the absorbed fractions were then separated by SDS-PAGE and were subsequently reacted in Western blots with ␣PEP13h (which recognizes the carboxyl terminus of gp100), ␣mSiN, ␣PEP1h, or control IgG. HMB45 has been reported previously not to work well in immunoprecipitation protocols (22), and that was indeed confirmed in this study, because all gp100 detectable by ␣PEP13h or by HMB45 was found in the flow-through (unabsorbed) fraction of the HMB45-treated sample and in the control IgG as well ( Fig. 2A). gp100 was significantly absorbed by HMB50, HMB55, and NKI/beteb, as shown by the reduced levels of gp100 present in the flow-through fraction unabsorbed by those antibodies.
␣PEP13h readily detected both the full-length 100-kDa and the 26-kDa fragment of gp100 in samples absorbed by HMB50, HMB55, or NKI/beteb (Fig. 2B), which demonstrates that each of their epitopes is located in the 26-kDa fragment at the carboxylterminal region of gp100. In contrast, ␣mSiN recognized only the full-length 100-kDa gp100 in samples absorbed by HMB50, HMB55, or NKI/beteb, which is consistent with the data reported above. As controls, no gp100-reactive bands were recognized in extracts reacted with ␣PEP1h (an antibody that recognizes TYRP1), which was used as a negative control, and reactivity was also negative if no second antibody was used (control).
Synthesis, Processing, and Stability of gp100 -To assess the synthesis and processing of gp100, we used pulse-chase metabolic labeling of MNT-1 cells and immunoprecipitation with the gp100-specific antibodies (Fig. 3). As an example of the processing of other melanosomal proteins, tyrosinase processing was also examined immediately after the 30-min pulse and at various times of chase up to 24 h. Tyrosinase (detected by ␣PEP7h) was processed from its native 65-kDa form to the fully FIG. 1. Western blot analysis of gp100 in melanoma cells. MNT-1 and SK-MEL-28 melanoma cells (and SP1 keratinocytes as negative controls) were solubilized, and 5 g of the protein of each extract were separated on 8 -16% gradient Tris-glycine SDS-polyacrylamide gels. The proteins were transferred to polyvinylidene difluoride membranes and were detected by the antibodies ␣PEP13h (at 1:5000), ␣mSiN (at 1:1250), HMB45 (at 1:500), and ␣PEP7h (at 1:5000) using ECL as detailed under "Materials and Methods." Š, major specific band; , minor specific band; 4, background band (i.e. present in SP1 controls). glycosylated 75-kDa form quantitatively within 1.5 h, and the protein was quite stable, with much of it remaining after a 24-h chase, as described previously for mouse melanocytes (29).
Using the antibody recognizing the carboxyl terminus of gp100 (␣PEP13h), a faint band is detectable at 70 kDa, which represents the native newly synthesized protein. However, even at that zero chase time, the bulk of gp100 has undergone early glycosylation in the ER (see "Glycosylation and Sorting of gp100") and is detectable as the full-length 100-kDa band by ␣PEP13h (and also by ␣mSiN). After a 1.5-h chase, the 26-kDa cleaved fragment of gp100 becomes detectable by ␣PEP13h, and interestingly, a similarly sized fragment is also detected by ␣mSiN (but that latter fragment obviously is distinct because it contains the amino terminus of gp100 rather than the carboxyl terminus). As shown previously in murine melanocytes (21), the half-life of gp100 is relatively short, and it is essentially undetectable by all antibodies within 6 h of chase. As discussed above, HMB45 does not work well in immunoprecipitation protocols, but HMB50, HMB55, and NKI/beteb work quite well. All three of those gp100-specific monoclonal antibodies show patterns virtually identical to that of ␣PEP13h. These data demonstrate that newly synthesized gp100 undergoes very rapid early glycosylation in the ER and then is quickly cleaved into three fragments, a timing that coincides with the maturation of melanosomes from Stage I to Stage II (cf. below).
We next examined the processing and stability of gp100 by blocking protein synthesis with cycloheximide in conjunction with Western blotting (Fig. 4). Because HMB45 is not useful in immunoprecipitation, this approach by Western blot to examine the stability of the gp100 epitope recognized by HMB45 was highly informative. As a positive control for this approach, tyrosinase (recognized by ␣PEP7h) can be seen as highly stable even after 12 h of cycloheximide treatment (which is consistent with the metabolic labeling studies above). Also consistent with the above results, the gp100 carboxyl terminus epitope recognized by ␣PEP13h was quickly lost, even within 1 h of cycloheximide treatment. In contrast, the HMB45-reactive fragment of gp100 was generated quickly and was completely stable throughout the 12 h of exposure to cycloheximide. These results clearly show that newly synthesized gp100 is rapidly converted into the HMB45-reactive low molecular weight form, and that fragment (shown previously to be associated with the fibrillar matrix of the melanosome) is highly stable.
Glycosylation and Sorting of gp100 -Tyrosinase is highly resistant to digestion with Endo H as described previously (30) (Fig. 5A). This reflects the fact that tyrosinase is processed through the ER and subsequently the trans-Golgi network, at which point it becomes fully glycosylated and resistant to Endo H (31). Treatment with PNGaseF, which removes all carbohydrate residues, reveals the native protein chain of tyrosinase (ϳ55 kDa).
In contrast, the mature form of gp100 recognized by ␣PEP13h remains highly sensitive to Endo H treatment, and treatment with PNGaseF has no further effect on its size. These results indicate clearly that gp100 contains N-linked carbohydrate residues and is processed in the ER but that at least the bulk of it is not further glycosylated, thus implying that it is not processed through the trans-Golgi/endosomal network and must be delivered to melanosomes via another sorting pathway. Although treatment with neuraminidase to remove sialic acid groups has a significant effect on the mobility and heterogeneity of tyrosinase, it has no discernible effect on gp100. Note that the 26-kDa fragment of gp100 is also glycosylated to some extent because it is sensitive to treatment with PNGaseF and lowers the size of that band by about 15 kDa. However, in stark contrast to those results, treatment with  Fig. 1. The 26-kDa nonspecific minor band represents the small chain of IgG that elutes from the column in most cases; in some lanes the 26-kDa band derived from gp100 is also seen at that position.
Endo H or PNGaseF had no significant effect on the mobility of the 35-kDa fragment of gp100 recognized by HMB45 (except to increase its reactivity with that antibody). A previous report has shown that two mannosidase-specific inhibitors (swainsonine and deoxymannojirimycin) had dramatic effects on the glycosylation of tyrosinase, TYRP1, and DCT but had no detectable effect on the molecular size of gp100 (32). Thus, the sum of these results shows that, although gp100 undergoes early glycosylation events in the ER, it is not further processed through the trans-Golgi/endosomal network.
One would predict that this transition might affect the sol-ubilization characteristics of those epitopes, and we have conducted experiments to examine that using different extraction buffers of increasing solubilization potential (Fig. 5B). As the extraction conditions increase from a nonionic detergent only (Fig. 5B, lanes 1, Triton X-100) to increasing concentrations of an ionic detergent (SDS) and deoxycholate (lanes 2-4) and finally with M-PER (lanes 5), the extraction of the full-length (membrane-bound) gp100 recognized by ␣PEP13 (and the 26-kDa fragment) is relatively stable, but the extraction of the fibrillar matrix-associated short fragment (ϳ35 kDa) of gp100 recognized by HMB45 is dramatically increased. We have pro- posed before that as melanin production ensues and melanin is deposited on the fibrillar matrix, the HMB45-reactive epitopes are gradually covered, and by the time melanosomes have matured to the Stage III form, reactivity with HMB45 is lost. Such a scenario is consistent with the solubilization pattern seen in Fig. 5B, because solubilization with a nonionic detergent disrupts only hydrogen bonds but does not denature proteins, and very little melanin is solubilized in this way. Treatment with the ionic detergent SDS and deoxycholate disrupts stronger bonds (including sulfur bonds) and denatures proteins, and that harsher treatment still only slightly increases melanin solubility and HMB45 reactivity. However, extraction with M-PER, a highly efficient but proprietary reagent, is quite effective at solubilizing the melanin, and this dramatically increases the reactivity of HMB45 with the re-exposed epitope of gp100.
Immunohistochemical Analysis-The subcellular distribution of melanosomal proteins in pigmented and in unpigmented melanoma cells was then examined using confocal immunohistochemistry (Fig. 6). The yellow color indicates colocalization, whereas distinct signals are red or green. gp100 detected by ␣PEP13h (Fig. 6, red in all panels) colocalizes strongly with the ER marker (Bip) in MNT-1 (top row) and in SK-MEL-28 (bottom row) melanoma cells and was also found in the early cis-Golgi (GM130). However, very little gp100 colocalized in early endosomes (Fig. 6, EEA1), and in general there was complete separation of those signals, which is consistent with the biochemical and immunochemical studies presented in Figs. 3 and 5. There is also virtually complete segregation of gp100 recognized by ␣PEP13h with HMB45 (i.e. processed gp100 in Stage II melanosomes), as previously reported in MNT-1 cells (1). Interestingly, however, there was more colocalization of those signals in SK-MEL-28 cells, suggesting that the maturation of melanosomes from Stage I to Stage II is impaired to some degree in those cells. In the pigmented MNT-1 cells, tyrosinase is delivered to early melanosomes and colocalizes with gp100 to some extent, but the disruption of tyrosinase sorting to early melanosomes is seen clearly in SK-MEL-28 cells, and this misrouting results in the amelanotic phenotype of those cells. DISCUSSION gp100 has been an enigmatic protein with respect to its function in pigmentation, although all studies to date agree on its critical importance as a melanoma marker and as a specific tumor target. In this study, we have shown that gp100 is synthesized as a 70-kDa nascent protein that undergoes immediate early glycosylation in the ER and processing in the cis-Golgi. The bulk of gp100 is then quickly sorted to Stage I early melanosomes without going through the later glycosylation events that are associated with processing in the trans-Golgi. This was demonstrated clearly using pulse-chase metabolic labeling, digestion with glycosidases, and Western immunoblotting. The 100-kDa form of gp100 remains Endo H-sensitive, and its trafficking pathway is quite distinct from the pathway(s) involved in the trafficking of tyrosinase and the other melanosomal proteins (7,33). A small amount (ϳ5%) of gp100 does seem to undergo additional late glycosylation events that increase its mass to ϳ110 kDa, a form that is Endo H-resistant (16,34). The presence of gp100 in the endosomal system has been reported previously as evidence that it is processed through that pathway to reach Stage I melanosomes (6,16). Such evidence has been obtained primarily using nonmelanocytic cells transfected with wild-type and/or mutant constructs of gp100 and also by confocal immunohistochemistry. Such a model may not be physiologically relevant, because the target organelle (and its melanocyte-specific components) are not expressed in those cells; hence the normal process would be disrupted. The colocalization of gp100 in endosomal compartments does not provide a clue as to whether they are coming or going from melanosomes. In contrast to the conclusions of those previous studies, our results suggest that the gp100 found in the endosomal system is in the process of being removed from the cell.
An interesting consideration is whether other proteins follow a pathway of processing through the Golgi but bypass the endosomal sorting pathway. As one example of such an occurrence, secretory granules do form from the Golgi apart from the endosomal pathway (35), and it is interesting to note that our recent proteomics study (1) shows that early melanosomes do FIG. 5. Glycosylation analysis and solubilization of melanosomal proteins. A, MNT-1 melanoma cells were solubilized and then digested with Endo H or PNGaseF (or buffer as a control) to determine glycosylation and maturation of proteins, as detailed under "Materials and Methods." Tyrosinase was detected by ␣PEP7h (at 1:5000), whereas gp100 was detected by ␣PEP13h (at 1:5000) and by HMB45 (at 1:500). B, MNT-1 cells were solubilized with various buffers and detergents as noted below, and epitopes were then detected by ␣PEP13h or by HMB45, as detailed under "Materials and Methods." Lanes 1, 1% Triton X-100; lanes 2, 1% Triton X-100 and 0.01% SDS; lanes 3, 1% Triton X-100, 0.01% SDS, and 1% deoxycholate; lanes 4, 1% Triton X-100, 0.1% SDS, and 1% deoxycholate (radioimmune precipitation assay buffer); lanes 5, M-PER. in fact contain several proteins thought to be specific to secretory granules. Melanosomes can be considered as secretory granules in some respects, because they are specialized vesicles in which their ultimate purpose is to be secreted from melanocytes. We cannot rule out the possibility that the final delivery of gp100 to Stage I melanosomes occurs from the trans-Golgi, because the sorting and processing machinery to intracellular organelles, including secretory granules, is located there.
Taken together with our recent proteomics analysis of early melanosomes (15), which found a number of ER markers associated with those organelles, and with confocal immunohistochemistry results showing the presence of ER proteins even in late Stage III and IV melanosomes, the biogenesis of early melanosomes as vesicles budding directly from the ER (as proposed more than 4 decades ago) (36) seems to be a reasonable mechanism for their formation. We would extend that proposed pathway to include the fact that gp100 (the existence of which was unknown until just more than a decade ago) is trafficked in those ER/early Golgi vesicles, which become de facto Stage I melanosomes. Small amounts of the other melanosomal proteins (such as TYR and TYRP1) are found in Stage I melanosomes (1), and it now seems likely that they are there by virtue of being trapped in the gp100 vesicles that become Stage I melanosomes, because those other melanosomal proteins are processed normally through the ER and cis-and trans-Golgi. However, the bulk of those other melanosomespecific proteins are delivered via sorting vesicles only following the maturation of Stage I to Stage II melanosomes (1,8), a process that depends on the subsequent proteolytic cleavage of gp100 at its carboxyl and amino termini and its subsequent incorporation into the fibrillar matrix of melanosomes. The importance of gp100 cleavage to the maturation of early melanosomes is underscored by the fact that melanoma cells (pigmented or amelanotic), wherein gp100 is expressed and cleaved correctly (e.g. MNT-1 cells), do form Stage II melanosomes, whereas melanoma cells that do not express gp100 (or are not able to process gp100 into an HMB45-recognizable form) do not produce Stage II melanosomes and thus do not produce pigment, even if they express tyrosinase and/or the other melanosome-specific proteins.
A number of monoclonal antibodies have been developed over the years that recognize different forms of gp100; Fig. 7 presents a summary of the processing events known to occur for gp100 and the various epitopes recognized by the different gp100 antibodies elucidated in this study. gp100 is synthesized as a 661-amino acid polypeptide, which has an initial mass of 70 kDa. Maresh et al. (37) report that gp100 is glycosylated at three N-linked sites at residues Asn-81, Asn-106, and Asn-111. They further report that gp100 is then cleaved into a small carboxyl-terminal fragment (at residue Val-467) and a larger amino-terminal fragment (37), a scenario confirmed in our study. Our results showed further that a second cleavage occurs, approximately at residues 95-100, which generates a 26-kDa fragment containing the amino-terminal region. PNGaseF digestion of that amino-terminal fragment reveals that ϳ15 kDa represent carbohydrate residues, the core size of that protein fragment thus being 11 kDa. ␣PEP13h recognizes the carboxyl terminus of gp100 by design, and HMB50, HMB55, and NKI-beteb recognize epitopes that are nearby based on their recognition of full-length gp100 and also on the fact that the carboxyl terminus fragment cleaved from that protein remains immunoreactive with ␣PEP13h. ␣mSiN recognizes the amino terminus of gp100 by design, whereas HMB45 recognizes an internal site on the middle fragment. The sum of those results shows that Asn-321 and Asn-568 are also fully glycosylated, each contributing about 5 kDa to the mass of the protein. The interesting and specific feature of HMB45 is that it recognizes only the processed (cleaved) form of gp100 (hence the 45-kDa surname) and thus is specific for Stage II melanosomes, whereas the other antibodies (monoclonal and polyclonal) recognize intact gp100 in early melanosomes and do not distinguish between stages of melanosomes. At the time of gp100 cleavage, the carboxyl-terminal epitope recognized by ␣PEP13h remains in the melanosomal membrane (1,16), but reactivity with that antibody is then quickly lost, suggesting that it is further degraded soon thereafter. Reactivity with HMB45 is itself lost as melanin is produced and deposited on the melanosomal matrix presumably by physical masking of the gp100 epitope by that pigment. The conformation-dependent nature of the epitope recognized by HMB45 allows us to confirm that the gp100 cleavage events seen using biochemical analyses actually occur in vivo.
An interesting and important question therefore is which protease(s) is involved in the processing of gp100 to its truncated form (steps that are critical to the maturation of melanosomes)? It will be important for future study to identify the protease(s) involved in gp100 trimming and the sites of those proteolytic cleavage events that play such important roles in determining melanosome function and pigmentation and in the generation of gp100 epitopes recognized by the immune system in human melanoma cells. Recently, the involvement of proprotein convertases in the initial cleavage and processing of gp100 near its carboxyl terminus in a post-Golgi compartment was reported (20), and the presence of a related enzyme, prohormone convertase, has also been reported in melanosomes (38), as have a number of different peptidases and proteases (15). Future study will naturally be directed toward identifying those proteolytic components apparently present in early melanosomes, and clues should become available as our proteomics analysis of Stage I melanosomes proceeds. The importance of gp100 processing goes beyond its implications for melanosome biogenesis and maturation, because it plays a role in the generation of gp100 peptides and their presentation on the surface of melanoma cells, an important criterion to the recognition of those tumor cells by the host immune system.