- Email: [email protected]
Tyrosinase-Related Protein-2
Cell-Density-Dependent Regulation of Expression and Glycosylation of Dopachrome Tautomerase/TyrosinaseRelated Protein-2 Thomas J. Hornyak, Daniel J. Hayes, and Edward B. Ziff*
The Ronald O. Perelman Department of Dermatology and *Department of Biochemistry and Howard Hughes Medical Institute, New York University Medical Center, New York, U.S.A.
The expression of the dopachrome tautomerase gene (Dct) and its protein product, tyrosinase-related protein-2, was studied in the cultured, phorbol-esterdependent murine melanocyte cell line melan-a. Increased cell density was found to stimulate Dct expression both in cells stably transfected with a Dct promoter-lacZ construct and endogenously in nontransfected cells. Increased Dct expression under these conditions corresponds to increased tyrosinaserelated protein-2 production. Tyrosinase-related protein-2 was found to exist in two distinct glycoforms with different endoglycosidase sensitivities. Density-dependent expression of tyrosinase-related
protein-2 was independent of time of cell growth, cell proliferation, and soluble factors, implying that cell-cell contact is the important determinant governing increased Dct expression under these conditions. Tyrp1 gene expression and tyrosinase-related protein-1 production were also induced under similar conditions. The results show that cell-cell contact between melanocytes induces a coordinated response at both transcriptional and nontranscriptional levels that induces production of the tyrosinase-related proteins that have a signi®cant role in melanization. Key words: cell-cell contact/melanization/transcription. J Invest Dermatol 115:106±112, 2000
T
embryo apart from melanocytes and their precursors, including the pigmented retinal epithelium, the inner ear, the telencephalon, and the endolymphatic duct (Steel et al, 1992). The distinct difference in the onset of expression of Dct during development suggests that, despite overall sequence similarity with other members of the tyrosinase family, the mechanism of its regulation of expression may differ substantially from that of either Tyrp1 or Tyr. In this paper, we report that both Dct and Tyrp1 are induced by increases in cell density in cultured mouse melanocytes. Induction of Dct and Tyrp1 resulting in increased cellular TRP-2 and TRP-1 levels, respectively, appears to be predominantly regulated at the transcriptional level, with an additional nontranscriptional level of control identi®ed for modifying TRP-2 levels. Two distinct glycoforms of TRP-2 are identi®ed, the levels of which diminish differently after high cell density melanocytes are trypsinized and grown at low cell density. Together with previous results from density-dependent studies of tyrosinase expression (Mahalingam et al, 1997), these ®ndings show that cell-cell contact between melanocytes coordinately increases the expression of the tyrosinase gene family in melanocytes to induce pigmentation and affect the processing of TRP-2 glycoforms. Differences between the patterns of Dct and Tyrp1 induction, however, suggest that these genes are regulated differently under similar cellular conditions, and may be relevant to the differential expression pattern noted in the developing murine embryo.
hree genes possessing signi®cant sequence similarity (Budd and Jackson, 1995) characteristically expressed by melanocytes encode enzymes crucial to the type of eumelanin synthesized. These genes, the members of the tyrosinase gene family, are tyrosinase (Tyr), tyrosinase-related protein-1 (Tyrp1), and dopachrome tautomerase (Dct). The Dct gene encodes the enzyme dopachrome tautomerase, also known as tyrosinase-related protein-2 (TRP-2). Dopachrome tautomerase catalyzes the conversion of DOPAchrome to 5,6dihydroxyindole-2-carboxylic acid (DHICA) in the melanin biosynthetic pathway (Tsukamoto et al, 1992). Tyrp1 encodes the protein tyrosinase-related protein-1 (TRP-1), which has been shown to possess DHICA oxidase activity (Kobayashi et al, 1994). Tyr encodes the enzyme tyrosinase, the rate-limiting enzyme in the melanin biosynthetic pathway, which possesses both a tyrosine hydroxylase activity, converting tyrosine to dihydroxyphenylalanine (DOPA), and a second, DOPA oxidase, activity, converting DOPA to DOPAquinone, which rapidly rearranges to DOPAchrome (Korner and Pawelek, 1982). All three enzymes are contained within the melanosome (Orlow et al, 1994). Developmentally, the expression of Dct occurs in migratory melanocytes and in melanocyte precursors earlier than the expression of either Tyrp1 or Tyr. Expression of Dct is observed in areas of the murine
Manuscript received September 3, 1999; revised March 31, 2000; accepted for publication April 25, 2000. Reprint requests to: Dr. Thomas J. Hornyak, Department of Dermatology, Henry Ford Health Science Center, One Ford Place 5D, Detroit, Michigan 48202. Email: [email protected] Abbreviations: CMV, cytomegalovirus; HMW, high molecular weight; LMW, low molecular weight; TRP-1, tyrosinase-related protein-1; TRP2, tyrosinase-related protein-2 0022-202X/00/$15.00
MATERIALS AND METHODS Plasmid construction Plasmids pPDct/luc and pPDct/lacZ were generated from the expression plasmid pPDct. Plasmid pPDct was constructed by ligating pcDNA3DCMV, the 4.5 kb fragment lacking the cytomegalovirus (CMV) promoter obtained from a BglII±EcoRV digest of the eukaryotic expression vector pcDNA3 (Invitrogen, Carlsbad, CA),
´ Copyright # 2000 by The Society for Investigative Dermatology, Inc. 106
VOL. 115, NO. 1 JULY 2000
with the 3.4 kb BamHI±Eco47III fragment from the clone lB plasmid (Budd and Jackson, 1995) containing nucleotides ±3240 to +148 relative to the transcriptional start site in the EMBL/GenBank version of the sequence (X85126 M. Musculus Tyrp2 gene) published previously (Budd and Jackson, 1995). (A comparison of the electronic version of the sequence and the published version reveals that the second and third nucleotides of the hexanucleotide recognition sequence for Eco47III (GCGCTC) are missing in the published version. Nonetheless, the enzyme still cleaves at this location.) Plasmid pPDct/luc was generated by subcloning the 2.7 kb XhoI±SalI fragment (containing the luciferase coding region linked to an SV40 polyadenylation sequence) from the plasmid pGL2-Basic (Promega, Madison, WI) into the XhoI site 3¢ to the EcoRV/Eco47III linkage in pPDct. Plasmid pPDct/lacZ was generated by subcloning the SalI fragment from the plasmid pNL-bgal (gift of Dr. Bob Schneider, Department of Biochemistry, NYU Medical Center), containing the lacZ fragment from the plasmid pCH110 (Pharmacia, Stockholm, Sweden), into the XhoI site of pPDct. Plasmids CMV/lacZ and RSV/luc were gifts from Dr. Theodore Lee, NYU Medical Center. Cell culture, transfections, and expression assays Melan-a cells, obtained by courtesy of Dr. Dorothy Bennett, were plated at the indicated densities and cultured in Eagle's minimal essential medium supplemented with 25 mM NaHCO3, 0.1 mM 2-mercaptoethanol, 5% fetal bovine serum, 1% penicillin-streptomycin, 1 mM sodium pyruvate, 200 mM tetradecanoyl phorbol acetate (TPA; Sigma, St. Louis, MO), pH 6.9 in 5% CO2, essentially as described by Bennett et al (1987). Melan-a cells were transfected with calcium phosphate precipitates as described by Chen and Okayama (1987). With careful attention to the pH of the BES-buffered saline transfection solution, with an optimal pH of 6.95, transfection ef®ciencies of up to 10% could be obtained using this method, based upon counts of lacZ-stained melan-a cells in several microscopic ®elds following transfection with the plasmid CMV/lacZ. Deviations as low as 0.03 pH units, however, signi®cantly diminished this level of transfection ef®ciency. For transient transfections, transfected cells were incubated at 37°C, 3.5% CO2, 95% humidity for 24 h prior to changing of growth media and subsequent incubation at 5% CO2. Cell pellets were lysed in 100 ml of 100 mM potassium phosphate, pH 7.8, 0.2% Triton X-100, and 1 mM dithiothreitol. Samples were assayed by mixing 50 ml of clari®ed cell extract, 100 ml of 1 mM D-luciferin, and 300 ml luciferase reaction mixture consisting of 25 mM glycylglycine, pH 7.8, 15 mM MgSO4, 1 mM adenosine triphosphate, 0.1 mg per ml bovine serum albumin (BSA), and 1 mM dithiothreitol and monitoring in a Berthold Lumat LB 9501 luminometer. b-Galactosidase assays of transiently transfected cells utilized the o-nitrophenyl-b-D-galactopyranoside assay (Sambrook et al, 1989). For stable transfections, 1.8 3 105 melan-a cells were plated in 6 cm tissue culture plates and transfected with 7.2 mg pPDct/ lacZ 36 h later. Cells were placed in 3.5% CO2 for 16 h prior to changing of the media and continued growth for 72 h. Cells were then split 1:62.5 into 25 cm plates and selection was begun with 500 mg per ml G418. b-Galactosidase expression was detected by ®xing cells in 0.5% glutaraldehyde in phosphate-buffered saline (PBS) for 15 min, followed by a 30 min incubation at 37°C in 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mM MgSO4, 1 mg per ml 5-bromo-4-chloro-3indolyl-b-D-galactoside [X-gal (Sigma) prepared as a 20 mg per ml solution in dimethylformamide], all prepared in PBS. Northern blotting and autoradiography Total RNA puri®cation from melan-a cell pellets was performed by a previously described method (Chomczynski and Sacchi, 1987). Formaldehyde gel electrophoresis and the transfer of total RNA to Duralon-UV membranes (Stratagene, La Jolla, CA) was performed using standard methods (Ausubel et al, 1993). A 1.7 kb probe for Dct (Fig 1) was generated by cleaving plasmid pTRP2a (Jackson et al, 1992) with EcoRI, and a 0.5 kb probe for Dct (see Fig 6) was generated by cleaving the same plasmid with Bsn36I and EcoRI. A 0.3 kb probe for Tyrp1 was generated by cleaving the plasmid pMT4 (Shibahara et al, 1986) with StuI and HpaI. Random primed labeling was performed with a-32P-dCTP according to the manufacturer's instructions (Boehringer Mannheim, Basel, Switzerland), as was membrane hybridization with Quik-Hyb (Stratagene). Blots were exposed to X-Omat AR ®lm (Kodak, Rochester, NY). Western blotting Melan-a cells were grown to the indicated densities, either scraped from the plate or trypsinized followed by subsequent neutralization with growth media, washed twice with PBS, counted, and lysed with RIPA buffer with enzyme inhibitors (1 3 PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 0.1 mg per ml phenylmethylsulfonyl ¯uoride, 0.1 mg per ml aprotinin,
CELL DENSITY AND TRP-2 EXPRESSION
107
1 mM sodium orthovanadate). Lysate supernatants were collected following centrifugation, and the total protein concentration of each lysate was determined against a series of BSA standards. Equal amounts of total cellular protein were loaded on lanes of an 8% SDS-polyacrylamide gel. Proteins were transferred after electrophoresis to Protran nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) overnight. Blots were blocked with 5% nonfat dry milk (NFDM; Carnation, Glendale, CA) in 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.05% Tween-20 (TBST). Primary antibody solution was 1:5000 aPEP8 or aPEP7 antiserum in 0.25% NFDM in TBST, or 1:500 mouse b-actin monoclonal antibody (Chemicon). Secondary antibody solution was 1:5000 peroxidase-labeled antirabbit (Amersham, Piscataway, NJ) or, for the monoclonal antibody, 1:5000 horseradish-peroxidase-labeled antimouse (Amersham), each in 0.25% NFDM in TBST. Antibody washes were followed with three washes of TBST, 10 min each. Signal detection was with SuperSignal Chemiluminescent substrate following the manufacturer's instructions (Pierce, Rockford, IL). Endoglycosidase digestions Eighty micrograms of total cellular protein from a con¯uent melan-a cell lysate were each treated with either N-glycosidase F or endoglycosidase H (both Boehringer Mannheim). Protein samples were digested with 1 U of N-glycosidase F in a volume of 35 ml for 4 h and 8 h at 25°C in a buffer of 50 mM sodium phosphate, 1% Triton X-100, pH 7.4, or with 5 mU of endoglycosidase H under the same conditions of volume, temperature, and time in a buffer of 0.5 M sodium citrate, pH 5.5. Samples were heat quenched and mixed with SDS sample buffer prior to electrophoresis.
RESULTS Cell density and Dct expression To characterize the transcriptional activity of the 3.2 kb upstream regulatory sequence of the murine Dct gene (Jackson et al, 1992), plasmids containing Dct regulatory sequence linked to the reporter genes luciferase (plasmid pPDct/luc) and b-galactosidase (plasmid pPDct/lacZ) were transfected into melan-a cells both stably and transiently. Cells transfected with pPDct/lacZ were selected with G418 for stable integration, and resulting colonies were stained with X-gal, an indicator of b-galactosidase activity (Fig 1a). The staining pattern of colonies suggested that the Dct promoter was more active in cells at the center of the colony, in the region of highest cellular density. To ensure that this did not represent progressive loss of integrated DNA after cell division in G418-selected clones, we derived cell lines from selected colonies and showed that these expressed higher relative levels of b-galactosidase activity at increased cell density (data not shown). In contrast, transient transfections of pPDct/luc into melan-a cells under standard low cell density conditions (Chen and Okayama, 1987) of 1 3 106 cells per 10 cm plate (0.05±0.1 3 the cell density of melan-a cells at con¯uence) showed minimal transcriptional activity relative to a promoterless luciferase plasmid (Fig 1b). To test whether the density-dependent induction of Dct promoter activity in stable transfectants mimicked a densitydependent induction of the endogenous gene, con¯uent melan-a cells were plated at low density and permitted to grow to high density over the course of 7 d. Total RNA was extracted from the starting population of con¯uent cells and from populations of cells on successive days after plating and was subjected to denaturing gel electrophoresis and northern transfer. The blot was probed for Dct, gapdh, and b-actin mRNA to assess relative amounts of expression of these genes (Fig 1c). Quantitation of autoradiograms by scanning laser densitometry, normalized to gapdh, showed a 16-fold increase in the relative expression of Dct under these conditions (Fig 1d). We conclude that Dct expression in cultured melan-a cells is strongly dependent upon cell density, and that the regulatory elements required for cell-density-induced expression are contained within the proximal 3.2 kb of the Dct upstream regulatory sequence, the same sequence that confers tissue-speci®c expression of Dct in transgenic mice (Mackenzie et al, 1997). Effect of cell density upon TRP-2 production and glycosylation and TRP-1 production To test whether the density-dependent induction of Dct mRNA expression was accompanied by an induction of TRP-2 protein, con¯uent
108
HORNYAK ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 1. Cell-density-induced expression of Dct. (a) Photomicrograph of a colony of stably transfected cells following selection in G418. Colonies were identi®ed and stained for b-galactosidase expression 18 d after selection was begun. (b) Basal transcriptional activity of Dct promoter. Melan-a cells were plated at 1 3 106 cells per 10 cm plate and transfected 16 h later with 5 mg of either pPDct/luc, RSV/luc, or pGL2-Basic with cotransfection of 5 mg CMV/ lacZ as internal control with total plasmid DNA brought to 20 mg by addition of pcDNA3. Transfected cells were then incubated for 53 h after removal of calcium phosphate precipitate prior to collection of cell lysate and assay for luciferase and b-galactosidase activities. Data represent the mean corrected ratios of luciferase to b-galactosidase activities from four replicates, with error bars signifying mean 6 SD. (c) Autoradiograms of northern blots. Total RNA was isolated from melan-a cells at con¯uence (*) and on days 1±7 after plating at low density. Total RNA (6.7 mg) from melan-a cells as described was loaded onto a denaturing formaldehyde gel and transferred to nitrocellulose membrane following electrophoresis. Autoradiograms show results after exposure following hybridization with cDNA probes from Dct (Jackson et al, 1992), gapdh, and b-actin. (d) Quanti®cation of autoradiography signals from northern blot. A scanning laser densitometer was used to quantify the signals from the Dct probes and the gapdh probes. The ratio of Dct/gapdh = 1 was assigned to cells at con¯uence (*) and used to normalize all other ratios. The ratio of Dct/gapdh is designated as relative Dct RNA expression.
melan-a cells were plated at low density and grown over the course of 7 d to high density. Cell lysate was prepared from con¯uent cells and from cells on successive days after plating and analyzed by Western blotting with aPEP8 rabbit polyclonal antisera (Tsukamoto et al, 1992) to TRP-2. The concentration of total protein in each sample was determined, and equal amounts of total cellular protein were loaded in each lane of the gel. Additionally, the blot was probed with an antibody to b-actin to con®rm that similar amounts of protein were loaded in each lane (Fig 2c). Analysis of the blot with aPEP8 antisera revealed two sets of bands (Fig 2a). We found that, after reaching a minimum 4 d after plating, the relative expression of each of these bands increased with continued growth in culture. Cells were counted prior to lysis (Fig 2d), which con®rmed the density dependence of TRP-2 production in melan-a cells under these conditions. The slowest migrating set of bands, a doublet, migrated with an apparent molecular weight of 72 kDa. The faster migrating band migrated at a lower apparent molecular weight of 60 kDa. The calculated molecular weight of TRP-2, however, based upon the amino acid sequence lacking the putative signal sequence (Jackson et al, 1992), is 56 kDa, suggesting that the two forms of the protein observed upon Western blot are in fact distinct forms of TRP-2 that differ in their post-translational modi®cation, rather than the glycosylated and nascent unglycosylated forms suggested previously (Tsukamoto et al, 1992). To investigate this, cell lysates from con¯uent cells were treated with N-glycosidase F to remove completely N-linked oligosaccharides from the polypeptide chain or with endoglycosidase
H to distinguish between high-mannose glycosylated forms and more completely processed forms containing complex oligosaccharides. Western immunoblot (Fig 3) showed that N-glycosidase F treatment reduced both bands to a single band of 56 kDa on this gel. Endoglycosidase H treatment reduced only the lower band to a comigrating band, suggesting that the high molecular weight (HMW), endo H-resistant glycoform is the terminally processed form of TRP-2, whereas the low molecular weight (LMW) form is an incompletely processed, high-mannose intermediate. Closer examination of the Western blot in Fig 2(a) shows that the relative amounts of the HMW and LMW glycoforms of TRP-2 decrease at different rates when cells are trypsinized at con¯uence (lane 1, * in Fig 2a) and subsequently plated at low density (lane 2, day 1 in Fig 2a), with the HMW glycoform signi®cantly decreased at day 1 relative to the LMW glycoform, which does not decrease signi®cantly until day 3. Hence, the trypsinization of con¯uent cells is associated not only with an overall decrease in TRP-2 production, but also with selectively rapid disappearance of the HMW glycoform. To determine whether the effect observed with TRP-2 protein was more general, the blot was probed with aPEP1 antisera to TRP-1. This resulted in the detection of three distinct bands (Fig 2b), with the highest molecular weight band corresponding to the major glycosylated species described previously (Kobayashi et al, 1994). Some decrease in TRP-1 level was observed on days 1±3 following trypsinization, with subsequent increases at higher cell densities, but the changes were not of the magnitude observed with
VOL. 115, NO. 1 JULY 2000
CELL DENSITY AND TRP-2 EXPRESSION
109
Figure 3. Endoglycosidase sensitivities of TRP-2 glycoforms. Melan-a cells at con¯uence (2.0 3 107 cells per 10 cm plate or 2.6 3 105 cells per cm2) were scraped from the plate, counted, suspended in growth media, washed twice with ice-cold PBS, and lysed with RIPA buffer. Total protein concentration determination was performed against BSA standards. Eighty micrograms of total protein were loaded onto individual lanes of an 8% SDS-polyacrylamide gel before (lane 1) and after (lanes 2±5) digestion with N-glycosidase F and endoglycosidase H. TRP-2 glycoforms were detected by Western immunoblotting with aPEP8 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate.
TRP-2. Additionally, in contrast to TRP-2, selectively rapid disappearance of the highest molecular weight TRP-1 band was not observed. We conclude that both TRP-2 and TRP-1 protein levels are induced by increases in cell density, although the effect observed with TRP-2 is greater than that observed with TRP-1.
Figure 2. Cell-density-induced production of TRP-2 and TRP-1. Con¯uent melan-a cells were trypsinized, plated at 5 3 105 cells per 10 cm plate, and grown for 7 d. Media and TPA were changed and replenished daily. Total protein was obtained from cells at con¯uence (*) and on days 1±7 after plating at low density. One hundred micrograms of total protein from each set of cells were loaded into individual lanes of an 8% SDSpolyacrylamide gel and electrophoresed. (a) TRP-2 was detected by Western immunoblotting with aPEP8 antisera, horseradish-peroxidaseconjugated antirabbit, and chemiluminescent substrate. (b) TRP-1 was detected by Western immunoblotting with aPEP7 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate. (c) bactin was detected by Western immunoblotting with mouse antiactin monoclonal antibody, horseradish-peroxidase-conjugated antimouse, and chemiluminescent substrate. (d) Determination of cell densities. Cells at con¯uence prior to low density plating (*) and on days 1±7 after low density plating were counted following trypsinization and cell densities at trypsinization calculated. Error bars represent the mean 6 SD of four independent cell counts.
Time and TPA independence of cell-density-induced TRP-2 production The magnitude of the cell density effect observed with TRP-2 prompted further experiments to examine determinants of the effect. To determine whether the density dependence observed was associated with the length of time cells were grown in culture or the proliferation state of the cells, cells were plated at increasing densities, instead of the same low density, cultured for the same length of time, and harvested to determine relative levels of TRP-2 production. This was done both in the presence and in the absence of TPA, which is required for these cells to proliferate in culture (Bennett et al, 1987). The results (Fig 4a) show that, in the presence and in the absence of TPA, increased TRP-2 production is observed with increases in cell density when cells are plated at different densities and are harvested at the same time. In the presence of TPA (Fig 4a, lanes 1±4), increased TRP-2 production is observed over the full range of cell densities measured at harvest (Fig 4b, bars 1±4). In the absence of TPA (Fig 4a, lanes 5±8), an increase of TRP-2 production was observed only at the highest density of cells harvested (3.6 3 104 cells per cm2, lane 8 in Fig 4a). This density is comparable to the density in lane 2 where increased TRP-2 production was observed in the presence of TPA. The comparison of cell densities at harvest with plating density indicated that there was no cell proliferation over time (Fig 4b, bars 5±8). This result shows that cell proliferation is not required for density-dependent TRP-2 production to occur. Furthermore, the same lower level of TRP-2 production was observed for cells harvested at lower densities in the absence of TPA (lanes 5±7) as was observed for cells harvested at low density in the presence of TPA (lane 1). This demonstrates that growth arrest alone is not suf®cient for TRP-2 induction and implies that a threshold amount of cell-cell contact is required.
110
HORNYAK ET AL
Figure 4. Time and TPA independence of cell-density-induced TRP-2 production. Con¯uent melan-a cells (at 2.2 3 105 cells per cm2) were trypsinized and plated at increasing densities of 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2), 1 3 106 cells per 10 cm plate (1.3 3 104 cells per cm2), 2 3 106 cells per 10 cm plate (2.5 3 104 cells per cm2), and 5 3 106 cells per 10 cm plate (6.4 3 104 cells per cm2). Cells were grown in culture in the presence or absence of 200 nM TPA for 3 d, harvested by scraping, and cell lysate was prepared by the addition of RIPA buffer. Total protein concentrations were determined by comparison with BSA standards. (a) Total protein (7.5 mg) from cells grown in the presence (lanes 1±4) and in the absence (lanes 5±8) of TPA was loaded into individual lanes of an 8% SDS-polyacrylamide gel and electrophoresed. TRP-2 glycoforms were detected by Western immunoblotting with aPEP8 antisera, horseradish-peroxidase-conjugated antirabbit, and chemiluminescent substrate. (b) Determination of ®nal cell densities. Cells from samples initially plated at increasing densities were counted following removal from the plate by scraping and ®nal cell densities were calculated. Error bars represent mean 6 SD of four or six independent cell counts.
Independence from soluble factors of cell-density-induced TRP-2 production The medium was changed daily in cells plated at low density and cultured for extensive lengths of time to high density (Fig 2), in order to limit the potential effect of soluble factors present in higher concentrations in high density melan-a cells. Such soluble factors might induce TRP-2 production independent of intercellular contacts. An additional experiment was performed to test the effect of soluble factors upon TRP-2 production directly. Melan-a cells were plated both at high density (set A) and at low density (sets B, C, and D). After an initial period of growth, conditioned medium from high density plates was transferred to one set (set B) of low density plates. Growth was then continued and cells were subsequently harvested. As incubation with conditioned media from high density cells did not signi®cantly increase TRP-2 production (Fig 5), it is likely that soluble factors
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 5. Lack of dependence of cell-density-induced TRP-2 production on soluble factors. (a) Con¯uent melan-a cells (at 2.2 3 105 cells per cm2) were trypsinized and plated at densities of 5 3 106 cells per 10 cm plate (6.4 3 104 cells per cm2) (set A) or 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2) (sets B, C, and D). Thirtytwo hours after plating, conditioned media from sets A (high density plates) and C (low density plates) was transferred to sets B (low density plates) and D (low density plates), respectively, from which media had previously been removed. Fresh growth media and TPA were subsequently added to sets A and C. Cells were harvested by scraping 75±78 h after initial plating, counted, washed twice with ice-cold PBS, and lysed with RIPA buffer after centrifugation. Total protein concentrations were determined by comparison with a set of BSA standards, and 25 mg of total protein from each set of cells A-D were loaded into individual lanes of an 8% SDSpolyacrylamide gel and electrophoresed. Relative amounts of TRP-2 were detected by Western immunoblotting with aPEP8 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate. (b) Determination of cell densities. Cells were counted after scraping from the plates and ®nal cell densities were calculated. Error bars represent mean 6 SD of six independent cell counts.
do not signi®cantly induce TRP-2 production in melan-a cells, and that cell-cell contact is the important determinant in densitydependent TRP-2 induction. This experiment also shows that trypsinization alone does not reduce TRP-2 production in cells
VOL. 115, NO. 1 JULY 2000
CELL DENSITY AND TRP-2 EXPRESSION
111
apart from cell density, as the TRP-2 signals from cells at high density trypsinized at the same time as cells at low density is much greater. Cell-density-induced expression of tyrosinase-related protein genes To correlate the cell-density-dependent inductions of TRP-2 and TRP-1 protein with changes in gene expression, we compared Dct and Tyrp1 mRNA levels directly as a function of cell density. Restriction fragments from divergent regions of the Dct and Tyrp1 cDNAs were prepared and used as probes on a northern blot containing total RNA prepared from cultured melan-a cells at con¯uence and at various times after low density plating. The signal from the 3¢ Dct probe (Fig 6a) con®rmed the density-dependent induction of gene expression noted above (Fig 1c). Hybridization with the 3¢ Tyrp1 probe also showed density dependence of Tyrp1 gene expression, although in a different pattern from that observed with Dct. The decrease in Tyrp1 expression following trypsinization and plating at low density is more rapid and greater in magnitude than with Dct, evident from the undetectable expression of Tyrp1 on day 3 compared with the expression of Dct. Tyrp1 expression is induced at lower cell densities than Dct expression, however, as can be seen by the higher level of Tyrp1 expression on day 5. A similar pattern of Tyrp1 induction was observed when the northern blot in Fig 1(c) was probed with the 3¢ Tyrp1 probe (data not shown). These results demonstrate that both Dct and Tyrp1 expression are increased by increased cell density in cultured mouse melanocytes, with the different patterns of induction occurring most probably by different mechanisms in each case. DISCUSSION Cell density effect and TRP-2 production Densitydependent induction of protein expression may take place through transcriptional, translational, or post-translational mechanisms. The 16-fold induction of Dct (Fig 1c, d; Fig 6a) and even larger induction of Tyrp1 mRNA expression (Fig 6a) noted with increased cell density shows that control at the transcriptional level is extremely important as a determinant of TRP-2 and TRP-1 levels. Additionally, we found that TRP-2 exists as two distinct glycoforms with differential endoglycosidase sensitivities. Production of each of these glycoforms was enhanced by increased cell density. The endo H sensitivity of the 63 kDa LMW glycoform, combined with the fact that the LMW glycoform is the form ®rst observed in a pulse-chase experiment (Tsukamoto et al, 1992), suggests that it is an incompletely processed glycoform of TRP-2 predominantly localized to the endoplasmic reticulum and the cis-Golgi, whereas the 75 kDa, HMW glycoform is likely to be the terminally processed form of TRP-2 present in melanosomes. As the HMW glycoform disappears more rapidly after con¯uent cells are trypsinized and plated at low density (Fig 2), it is tempting to speculate that melanosomal TRP-2 might be selectively degraded, or that the melanosome itself is rearranged or secreted by trypsinization. It is unlikely on the basis of existing evidence, however, that gross rearrangement or signi®cant secretion of melanosomes can account for the change observed in HMW TRP-2 levels after trypsinization. TRP-1 and TRP-2 are found together in the melanosome (Orlow et al, 1994), and it is unlikely that persistence of the highest molecular weight TRP-1 glycoform observed after trypsinization (Fig 2b) in the absence of detectable Tyrp1 message at a comparable time (Fig 6a) could occur with concomitant loss or signi®cant reorganization of melanosomes. These observations also suggest that fully processed TRP-1 is quite stable, in comparison to fully glycosylated TRP-2, following trypsinization and low cell density replating. The possibility that the melanosome undergoes dramatic rearrangement following trypsinization and replating of melanocytes is further discounted by our own observations (unpublished data) and those of others (R. Boissy, personal communication) of highly melanized melanosomes, both on the light and electron microscopic level, in samples of cells that have been subjected to trypsinization and
Figure 6. Cell-density-induced upregulation of tyrosinase-related protein genes. (a) Con¯uent melan-a cells (at 1.9 3 105 cells per cm2) were trypsinized and plated at 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2). Total RNA was isolated from con¯uent cells initially (*) and from cells on days 3, 5, 7, 8, and 10 after plating. Total RNA (9.4 mg) from each sample was loaded onto individual lanes of a denaturing 1% formaldehyde agarose gel and separated by electrophoresis. RNA was transferred to nitrocellulose by northern blotting. Blot was hybridized with cDNA probes from the 3¢ end of Dct (nucleotides 1658±2182; Jackson et al, 1992), from the 3¢ end of Tyrp1 (nucleotides 2150±2479; Shibahara et al, 1986), and from gapdh (Clontech). Hybridization of the blot with both probes simultaneously (data not shown) showed two distinct bands, slightly different in size, as evidence that the probes did not cross-hybridize with each other's mRNA. (b) Determination of cell densities. Cells were counted after trypsinization and cell densities were calculated. Error bars represent mean 6 SD of six or eight independent cell counts.
replating. The selective loss of the HMW glycoform following trypsinization and replating at low cell density is an indication that a distinct, nontranscriptional mechanism can also regulate intracellular TRP-2 levels. Detection of the nascent, unglycosylated form of TRP-2 in these experiments is unlikely given the simultaneity of protein synthesis and glycosylation (Kiely et al, 1976; Bergman and Kuehl, 1978; Glabe et al, 1980) except in the presence of a glycosylation inhibitor such as tunicamycin. Our experiments agree with those of others (Tsukamoto et al, 1992) concerning the apparent molecular weights of forms of TRP-2; we do not observe
112
HORNYAK ET AL
a 110 kDa band in our experiments, as was previously reported (Halaban et al, 1996). It is possible that differences in immunoblotting techniques may account for the appearance of nonspeci®c bands. Cell density induction of gene expression ± cell contact versus growth arrest Cell-contact-induced gene expression may occur via mechanisms induced prior to or coincident with the state of growth arrest induced by nontransformed cells at con¯uence or by the withdrawal of growth media. The cell-density-dependent induction of the tyrosinase-related protein genes that we observe is probably not related to growth arrest. First, correlation of the cell growth curves (Figs 2b, 6b) with the patterns of induction of either TRP-2 production or of Dct and Tyrp1 expression, respectively (Figs 2a, 6a), suggests that induction of expression occurs at cell densities signi®cantly below con¯uence, implying that growth arrest per se is not required for induction of these genes. This argument has been made previously to explain the induction of DEP-1, a receptor-like protein-tyrosine-phosphatase, at subcon¯uent cell densities (Ostman et al, 1994). The mechanism of regulation in this case, either transcriptional, translational, or post-translational, was not determined. Additionally, we induced growth arrest by depriving melan-a cells of TPA required for their proliferation (Bennett et al, 1987). We did not observe an increase in TRP-2 production relative to cells proliferating in the presence of TPA. These factors suggest that cell-cell contact prior to any contact-dependent growth arrest induces expression of the tyrosinase-related protein genes. It would be interesting to attempt to prove this point by, for example, correlating the relative expression of Dct with the relative proportions of melan-a cells at different points of the cell cycle. Density dependence of induction of tyrosinase gene family members The fact that Dct and Tyrp1 are both induced at increased cell density in melan-a cells, together with the ®nding that Tyr is induced along with the concomitant induction of speci®c DNA-protein complexes at increased cell density in murine B16 melanoma cells (Mahalingam et al, 1997), suggests that these genes are coordinately upregulated by signals induced by cell-cell contact. As these genes each encode distinct proteins identi®ed to be enzymes in the melanin biosynthetic pathway, it is possible that cellular determinants may be extremely important in the maintenance of constitutive melanization or the induction of facultative melanization in the murine melanocyte. Although melanocytes normally reside among and in contact with keratinocytes in the follicular epithelium, they exist in contact with one another in certain cutaneous conditions in humans characterized by abnormalities in melanocyte proliferation, such as benign melanocytic nevi and malignant melanoma. Coordinate increases in the level of transcription of melanosomal enzyme genes may help explain the increased pigmentation characteristic of some of these lesions. The development of methods for studying the expression of these proteins on the single cell level by using ¯uorescent confocal microscopy or green-¯uorescent-proteintagged proteins may permit the assessment of whether heterogeneous cell±cell interactions, in addition to the melanocyte±melanocyte interactions we have described, are also capable of inducing tyrosinase-related protein expression. Results of these experiments may lead to insights regarding the nature of protein±protein interactions at the cell membrane responsible for inducing gene expression. Melanization involves not only the production of enzymes to convert tyrosine to melanin, but also the formation and maintenance of the melanosome, the subcellular organelle within which melanin is synthesized. It is possible that cell-contact-mediated mechanisms are important for the production of other melanosome-speci®c proteins whose formation is required for the transport of melanogenic enzymes to the melanosomes and for the production of other proteins, such as myosin V (Wei et al, 1997; Wu et al, 1997; Lambert et al, 1998), important for transport and appropriate localization of melanosomes within the cell.
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
We thank Ling-Yu Chiu and Jessica Gavin for technical assistance, and members of the Ziff laboratory including Ted Lee, Paul Issack, Chris Daly, and Latika Khatri for valuable advice and suggestions. We appreciate gifts of plasmids from Ted Lee, Ian Jackson, and Shigeki Shibahara, and are particularly grateful to Ian Jackson for making the Dct promoter plasmid available to us prior to publication. We thank Dorothy Bennett for permission to use the melan-a cell line and Vincent Hearing for the gift of aPEP8 and aPEP1 antisera. We thank Vijayasaradhi Setaluri for a close reading of the manuscript prior to submission. A question by Ruth Halaban was important for prompting the endoglycosidase studies. This work was supported by National Institutes of Health Grants AR01992 and AR45001 to T.J.H. E.B.Z. is an Investigator of the Howard Hughes Medical Institute.
REFERENCES Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K: Current Protocols in Molecular Biology, pp 4.9.1±4.9.13. New York: John Wiley, 1993 Bennett DC, Cooper PJ, Hart IR: A line of non-tumorigenic mouse melanocytes, syngeneic with the B16 melanoma and requiring a tumour promoter for growth. Int J Cancer 39:414±418, 1987 Bergman LW, Kuehl WM: Temporal relationship of translation and glycosylation of immunoglobulin heavy and light chains. Biochemistry 17:5174±5180, 1978 Budd PS, Jackson IJ: Structure of the mouse tyrosinase-related protein-2/ dopachrome tautomerase (Tyrp2/Dct) gene and sequence of two novel slaty alleles. Genomics 29:35±43, 1995 Chen C, Okayama H: High-ef®ciency transformation of mammalian cells by plasmid DNA. Mol Cell Biol 7:2745±2752, 1987 Chomczynski P, Sacchi N: Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156±159, 1987 Glabe CG, Hanover JA, Lennarz WJ: Glycosylation of ovalbumin nascent chains: the spatial relationship between translation and glycosylation. J Biol Chem 255:9236±9242, 1980 Halaban R, Bohm M, Dotto P, Moellmann G, Cheng E, Zhang Y: Growth regulatory proteins that repress differentiation markers in melanocytes also downregulate the transcription factor Microphthalmia. J Invest Dermatol 106:1266±1272, 1996 Jackson IJ, Chambers DM, Tsukamoto K, Copeland NG, Gilbert DJ, Jenkins NA, Hearing V: A second tyrosinase-related protein, TRP-2, maps to and is mutated at the mouse slaty locus. EMBO J 11:527±535, 1992 Kiely ML, McKnight GS, Schimke RT: Studies on the attachment of carbohydrate to ovalbumin nascent chains in hen oviduct. J Biol Chem 251:5490±5495, 1976 Kobayashi T, Urabe K, Winder A, et al: Tyrosinase related protein 1 (TRP1) functions as a DHICA oxidase in melanin biosynthesis. EMBO J 13:5818± 5825, 1994 Korner A, Pawelek J: Mammalian tyrosinase catalyses three reactions in the biosynthesis of melanin. Science 217:1163±1165, 1982 Lambert J, Onderwater J, Haeghen Y, Vancoillie G, Koerten H, Mommaas A, Naeyaert J: Myosin V colocalizes with melanosomes and subcortical actin bundles not associated with stress ®bers in human epidermal melanocytes. J Invest Dermatol 111:835±840, 1998 Mackenzie MA, Jordan SA, Budd PS, Jackson IJ: Activation of the receptor tyrosine kinase Kit is required for the proliferation of melanoblasts in the mouse embryo. Dev Biol 192:99±107, 1997 Mahalingam H, Watanabe A, Tachibana M, Niles RM: Characterization of densitydependent regulation of the tyrosinase gene promoter: role of protein kinase C. Exp Cell Res 237:83±92, 1997 Orlow S, Zhou B-K, Chakraborty A, Drucker M, Pifko-Hirst S, Pawelek J: Highmolecular weight forms of tyrosinase and the tyrosinase related proteins: evidence for a melanogenic complex. J Invest Dermatol 103:196±201, 1994 Ostman A, Yang Q, Tonks NK: Expression of DEP-1, a receptor-like proteintyrosine-phosphatase, is enhanced with increasing cell density. Proc Natl Acad Sci 91:9680±9684, 1994 Sambrook J, Fritsch EF, Maniatis T: Molecular Cloning: A Laboratory Manual. 2nd edn, p. 16 66. Plainview, New York: Cold Spring Harbor Laboratory Press, 1989 Shibahara S, Tomita Y, Sakakura T, Nager C, Chaudhuri B, Muller R: Cloning and expression of cDNA encoding mouse tyrosinase. Nucl Acids Res 14:2413±2427, 1986 Steel K, Davidson DR, Jackson IJ: TRP-2/DT, a new early melanoblast marker shows that steel growth factor (c-kit ligand) is a survival factor. Development 115:1111±1119, 1992 Tsukamoto K, Jackson IJ, Urabe K, Montague PM, Hearing VJ: A second tyrosinaserelated protein, TRP-2, is a melanogenic enzyme termed DOPAchrome tautomerase. EMBO J 11:519±526, 1992 Wei Q, Wu X, Hammer J: The predominant defect in dilute melanocytes is in melanosome distribution and not cell shape, supporting a role for myosin V in melanosome transport. J Muscle Res Cell Motil 18:517±527, 1997 Wu X, Bowers B, Wei Q, Kocher B, Hammer JI: Myosin V associates with melanosomes in mouse melanocytes: evidence that myosin V is an organelle motor. J Cell Sci 110:847±859, 1997
The Ronald O. Perelman Department of Dermatology and *Department of Biochemistry and Howard Hughes Medical Institute, New York University Medical Center, New York, U.S.A.
The expression of the dopachrome tautomerase gene (Dct) and its protein product, tyrosinase-related protein-2, was studied in the cultured, phorbol-esterdependent murine melanocyte cell line melan-a. Increased cell density was found to stimulate Dct expression both in cells stably transfected with a Dct promoter-lacZ construct and endogenously in nontransfected cells. Increased Dct expression under these conditions corresponds to increased tyrosinaserelated protein-2 production. Tyrosinase-related protein-2 was found to exist in two distinct glycoforms with different endoglycosidase sensitivities. Density-dependent expression of tyrosinase-related
protein-2 was independent of time of cell growth, cell proliferation, and soluble factors, implying that cell-cell contact is the important determinant governing increased Dct expression under these conditions. Tyrp1 gene expression and tyrosinase-related protein-1 production were also induced under similar conditions. The results show that cell-cell contact between melanocytes induces a coordinated response at both transcriptional and nontranscriptional levels that induces production of the tyrosinase-related proteins that have a signi®cant role in melanization. Key words: cell-cell contact/melanization/transcription. J Invest Dermatol 115:106±112, 2000
T
embryo apart from melanocytes and their precursors, including the pigmented retinal epithelium, the inner ear, the telencephalon, and the endolymphatic duct (Steel et al, 1992). The distinct difference in the onset of expression of Dct during development suggests that, despite overall sequence similarity with other members of the tyrosinase family, the mechanism of its regulation of expression may differ substantially from that of either Tyrp1 or Tyr. In this paper, we report that both Dct and Tyrp1 are induced by increases in cell density in cultured mouse melanocytes. Induction of Dct and Tyrp1 resulting in increased cellular TRP-2 and TRP-1 levels, respectively, appears to be predominantly regulated at the transcriptional level, with an additional nontranscriptional level of control identi®ed for modifying TRP-2 levels. Two distinct glycoforms of TRP-2 are identi®ed, the levels of which diminish differently after high cell density melanocytes are trypsinized and grown at low cell density. Together with previous results from density-dependent studies of tyrosinase expression (Mahalingam et al, 1997), these ®ndings show that cell-cell contact between melanocytes coordinately increases the expression of the tyrosinase gene family in melanocytes to induce pigmentation and affect the processing of TRP-2 glycoforms. Differences between the patterns of Dct and Tyrp1 induction, however, suggest that these genes are regulated differently under similar cellular conditions, and may be relevant to the differential expression pattern noted in the developing murine embryo.
hree genes possessing signi®cant sequence similarity (Budd and Jackson, 1995) characteristically expressed by melanocytes encode enzymes crucial to the type of eumelanin synthesized. These genes, the members of the tyrosinase gene family, are tyrosinase (Tyr), tyrosinase-related protein-1 (Tyrp1), and dopachrome tautomerase (Dct). The Dct gene encodes the enzyme dopachrome tautomerase, also known as tyrosinase-related protein-2 (TRP-2). Dopachrome tautomerase catalyzes the conversion of DOPAchrome to 5,6dihydroxyindole-2-carboxylic acid (DHICA) in the melanin biosynthetic pathway (Tsukamoto et al, 1992). Tyrp1 encodes the protein tyrosinase-related protein-1 (TRP-1), which has been shown to possess DHICA oxidase activity (Kobayashi et al, 1994). Tyr encodes the enzyme tyrosinase, the rate-limiting enzyme in the melanin biosynthetic pathway, which possesses both a tyrosine hydroxylase activity, converting tyrosine to dihydroxyphenylalanine (DOPA), and a second, DOPA oxidase, activity, converting DOPA to DOPAquinone, which rapidly rearranges to DOPAchrome (Korner and Pawelek, 1982). All three enzymes are contained within the melanosome (Orlow et al, 1994). Developmentally, the expression of Dct occurs in migratory melanocytes and in melanocyte precursors earlier than the expression of either Tyrp1 or Tyr. Expression of Dct is observed in areas of the murine
Manuscript received September 3, 1999; revised March 31, 2000; accepted for publication April 25, 2000. Reprint requests to: Dr. Thomas J. Hornyak, Department of Dermatology, Henry Ford Health Science Center, One Ford Place 5D, Detroit, Michigan 48202. Email: [email protected] Abbreviations: CMV, cytomegalovirus; HMW, high molecular weight; LMW, low molecular weight; TRP-1, tyrosinase-related protein-1; TRP2, tyrosinase-related protein-2 0022-202X/00/$15.00
MATERIALS AND METHODS Plasmid construction Plasmids pPDct/luc and pPDct/lacZ were generated from the expression plasmid pPDct. Plasmid pPDct was constructed by ligating pcDNA3DCMV, the 4.5 kb fragment lacking the cytomegalovirus (CMV) promoter obtained from a BglII±EcoRV digest of the eukaryotic expression vector pcDNA3 (Invitrogen, Carlsbad, CA),
´ Copyright # 2000 by The Society for Investigative Dermatology, Inc. 106
VOL. 115, NO. 1 JULY 2000
with the 3.4 kb BamHI±Eco47III fragment from the clone lB plasmid (Budd and Jackson, 1995) containing nucleotides ±3240 to +148 relative to the transcriptional start site in the EMBL/GenBank version of the sequence (X85126 M. Musculus Tyrp2 gene) published previously (Budd and Jackson, 1995). (A comparison of the electronic version of the sequence and the published version reveals that the second and third nucleotides of the hexanucleotide recognition sequence for Eco47III (GCGCTC) are missing in the published version. Nonetheless, the enzyme still cleaves at this location.) Plasmid pPDct/luc was generated by subcloning the 2.7 kb XhoI±SalI fragment (containing the luciferase coding region linked to an SV40 polyadenylation sequence) from the plasmid pGL2-Basic (Promega, Madison, WI) into the XhoI site 3¢ to the EcoRV/Eco47III linkage in pPDct. Plasmid pPDct/lacZ was generated by subcloning the SalI fragment from the plasmid pNL-bgal (gift of Dr. Bob Schneider, Department of Biochemistry, NYU Medical Center), containing the lacZ fragment from the plasmid pCH110 (Pharmacia, Stockholm, Sweden), into the XhoI site of pPDct. Plasmids CMV/lacZ and RSV/luc were gifts from Dr. Theodore Lee, NYU Medical Center. Cell culture, transfections, and expression assays Melan-a cells, obtained by courtesy of Dr. Dorothy Bennett, were plated at the indicated densities and cultured in Eagle's minimal essential medium supplemented with 25 mM NaHCO3, 0.1 mM 2-mercaptoethanol, 5% fetal bovine serum, 1% penicillin-streptomycin, 1 mM sodium pyruvate, 200 mM tetradecanoyl phorbol acetate (TPA; Sigma, St. Louis, MO), pH 6.9 in 5% CO2, essentially as described by Bennett et al (1987). Melan-a cells were transfected with calcium phosphate precipitates as described by Chen and Okayama (1987). With careful attention to the pH of the BES-buffered saline transfection solution, with an optimal pH of 6.95, transfection ef®ciencies of up to 10% could be obtained using this method, based upon counts of lacZ-stained melan-a cells in several microscopic ®elds following transfection with the plasmid CMV/lacZ. Deviations as low as 0.03 pH units, however, signi®cantly diminished this level of transfection ef®ciency. For transient transfections, transfected cells were incubated at 37°C, 3.5% CO2, 95% humidity for 24 h prior to changing of growth media and subsequent incubation at 5% CO2. Cell pellets were lysed in 100 ml of 100 mM potassium phosphate, pH 7.8, 0.2% Triton X-100, and 1 mM dithiothreitol. Samples were assayed by mixing 50 ml of clari®ed cell extract, 100 ml of 1 mM D-luciferin, and 300 ml luciferase reaction mixture consisting of 25 mM glycylglycine, pH 7.8, 15 mM MgSO4, 1 mM adenosine triphosphate, 0.1 mg per ml bovine serum albumin (BSA), and 1 mM dithiothreitol and monitoring in a Berthold Lumat LB 9501 luminometer. b-Galactosidase assays of transiently transfected cells utilized the o-nitrophenyl-b-D-galactopyranoside assay (Sambrook et al, 1989). For stable transfections, 1.8 3 105 melan-a cells were plated in 6 cm tissue culture plates and transfected with 7.2 mg pPDct/ lacZ 36 h later. Cells were placed in 3.5% CO2 for 16 h prior to changing of the media and continued growth for 72 h. Cells were then split 1:62.5 into 25 cm plates and selection was begun with 500 mg per ml G418. b-Galactosidase expression was detected by ®xing cells in 0.5% glutaraldehyde in phosphate-buffered saline (PBS) for 15 min, followed by a 30 min incubation at 37°C in 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, 1 mM MgSO4, 1 mg per ml 5-bromo-4-chloro-3indolyl-b-D-galactoside [X-gal (Sigma) prepared as a 20 mg per ml solution in dimethylformamide], all prepared in PBS. Northern blotting and autoradiography Total RNA puri®cation from melan-a cell pellets was performed by a previously described method (Chomczynski and Sacchi, 1987). Formaldehyde gel electrophoresis and the transfer of total RNA to Duralon-UV membranes (Stratagene, La Jolla, CA) was performed using standard methods (Ausubel et al, 1993). A 1.7 kb probe for Dct (Fig 1) was generated by cleaving plasmid pTRP2a (Jackson et al, 1992) with EcoRI, and a 0.5 kb probe for Dct (see Fig 6) was generated by cleaving the same plasmid with Bsn36I and EcoRI. A 0.3 kb probe for Tyrp1 was generated by cleaving the plasmid pMT4 (Shibahara et al, 1986) with StuI and HpaI. Random primed labeling was performed with a-32P-dCTP according to the manufacturer's instructions (Boehringer Mannheim, Basel, Switzerland), as was membrane hybridization with Quik-Hyb (Stratagene). Blots were exposed to X-Omat AR ®lm (Kodak, Rochester, NY). Western blotting Melan-a cells were grown to the indicated densities, either scraped from the plate or trypsinized followed by subsequent neutralization with growth media, washed twice with PBS, counted, and lysed with RIPA buffer with enzyme inhibitors (1 3 PBS, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 0.1 mg per ml phenylmethylsulfonyl ¯uoride, 0.1 mg per ml aprotinin,
CELL DENSITY AND TRP-2 EXPRESSION
107
1 mM sodium orthovanadate). Lysate supernatants were collected following centrifugation, and the total protein concentration of each lysate was determined against a series of BSA standards. Equal amounts of total cellular protein were loaded on lanes of an 8% SDS-polyacrylamide gel. Proteins were transferred after electrophoresis to Protran nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany) overnight. Blots were blocked with 5% nonfat dry milk (NFDM; Carnation, Glendale, CA) in 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.05% Tween-20 (TBST). Primary antibody solution was 1:5000 aPEP8 or aPEP7 antiserum in 0.25% NFDM in TBST, or 1:500 mouse b-actin monoclonal antibody (Chemicon). Secondary antibody solution was 1:5000 peroxidase-labeled antirabbit (Amersham, Piscataway, NJ) or, for the monoclonal antibody, 1:5000 horseradish-peroxidase-labeled antimouse (Amersham), each in 0.25% NFDM in TBST. Antibody washes were followed with three washes of TBST, 10 min each. Signal detection was with SuperSignal Chemiluminescent substrate following the manufacturer's instructions (Pierce, Rockford, IL). Endoglycosidase digestions Eighty micrograms of total cellular protein from a con¯uent melan-a cell lysate were each treated with either N-glycosidase F or endoglycosidase H (both Boehringer Mannheim). Protein samples were digested with 1 U of N-glycosidase F in a volume of 35 ml for 4 h and 8 h at 25°C in a buffer of 50 mM sodium phosphate, 1% Triton X-100, pH 7.4, or with 5 mU of endoglycosidase H under the same conditions of volume, temperature, and time in a buffer of 0.5 M sodium citrate, pH 5.5. Samples were heat quenched and mixed with SDS sample buffer prior to electrophoresis.
RESULTS Cell density and Dct expression To characterize the transcriptional activity of the 3.2 kb upstream regulatory sequence of the murine Dct gene (Jackson et al, 1992), plasmids containing Dct regulatory sequence linked to the reporter genes luciferase (plasmid pPDct/luc) and b-galactosidase (plasmid pPDct/lacZ) were transfected into melan-a cells both stably and transiently. Cells transfected with pPDct/lacZ were selected with G418 for stable integration, and resulting colonies were stained with X-gal, an indicator of b-galactosidase activity (Fig 1a). The staining pattern of colonies suggested that the Dct promoter was more active in cells at the center of the colony, in the region of highest cellular density. To ensure that this did not represent progressive loss of integrated DNA after cell division in G418-selected clones, we derived cell lines from selected colonies and showed that these expressed higher relative levels of b-galactosidase activity at increased cell density (data not shown). In contrast, transient transfections of pPDct/luc into melan-a cells under standard low cell density conditions (Chen and Okayama, 1987) of 1 3 106 cells per 10 cm plate (0.05±0.1 3 the cell density of melan-a cells at con¯uence) showed minimal transcriptional activity relative to a promoterless luciferase plasmid (Fig 1b). To test whether the density-dependent induction of Dct promoter activity in stable transfectants mimicked a densitydependent induction of the endogenous gene, con¯uent melan-a cells were plated at low density and permitted to grow to high density over the course of 7 d. Total RNA was extracted from the starting population of con¯uent cells and from populations of cells on successive days after plating and was subjected to denaturing gel electrophoresis and northern transfer. The blot was probed for Dct, gapdh, and b-actin mRNA to assess relative amounts of expression of these genes (Fig 1c). Quantitation of autoradiograms by scanning laser densitometry, normalized to gapdh, showed a 16-fold increase in the relative expression of Dct under these conditions (Fig 1d). We conclude that Dct expression in cultured melan-a cells is strongly dependent upon cell density, and that the regulatory elements required for cell-density-induced expression are contained within the proximal 3.2 kb of the Dct upstream regulatory sequence, the same sequence that confers tissue-speci®c expression of Dct in transgenic mice (Mackenzie et al, 1997). Effect of cell density upon TRP-2 production and glycosylation and TRP-1 production To test whether the density-dependent induction of Dct mRNA expression was accompanied by an induction of TRP-2 protein, con¯uent
108
HORNYAK ET AL
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 1. Cell-density-induced expression of Dct. (a) Photomicrograph of a colony of stably transfected cells following selection in G418. Colonies were identi®ed and stained for b-galactosidase expression 18 d after selection was begun. (b) Basal transcriptional activity of Dct promoter. Melan-a cells were plated at 1 3 106 cells per 10 cm plate and transfected 16 h later with 5 mg of either pPDct/luc, RSV/luc, or pGL2-Basic with cotransfection of 5 mg CMV/ lacZ as internal control with total plasmid DNA brought to 20 mg by addition of pcDNA3. Transfected cells were then incubated for 53 h after removal of calcium phosphate precipitate prior to collection of cell lysate and assay for luciferase and b-galactosidase activities. Data represent the mean corrected ratios of luciferase to b-galactosidase activities from four replicates, with error bars signifying mean 6 SD. (c) Autoradiograms of northern blots. Total RNA was isolated from melan-a cells at con¯uence (*) and on days 1±7 after plating at low density. Total RNA (6.7 mg) from melan-a cells as described was loaded onto a denaturing formaldehyde gel and transferred to nitrocellulose membrane following electrophoresis. Autoradiograms show results after exposure following hybridization with cDNA probes from Dct (Jackson et al, 1992), gapdh, and b-actin. (d) Quanti®cation of autoradiography signals from northern blot. A scanning laser densitometer was used to quantify the signals from the Dct probes and the gapdh probes. The ratio of Dct/gapdh = 1 was assigned to cells at con¯uence (*) and used to normalize all other ratios. The ratio of Dct/gapdh is designated as relative Dct RNA expression.
melan-a cells were plated at low density and grown over the course of 7 d to high density. Cell lysate was prepared from con¯uent cells and from cells on successive days after plating and analyzed by Western blotting with aPEP8 rabbit polyclonal antisera (Tsukamoto et al, 1992) to TRP-2. The concentration of total protein in each sample was determined, and equal amounts of total cellular protein were loaded in each lane of the gel. Additionally, the blot was probed with an antibody to b-actin to con®rm that similar amounts of protein were loaded in each lane (Fig 2c). Analysis of the blot with aPEP8 antisera revealed two sets of bands (Fig 2a). We found that, after reaching a minimum 4 d after plating, the relative expression of each of these bands increased with continued growth in culture. Cells were counted prior to lysis (Fig 2d), which con®rmed the density dependence of TRP-2 production in melan-a cells under these conditions. The slowest migrating set of bands, a doublet, migrated with an apparent molecular weight of 72 kDa. The faster migrating band migrated at a lower apparent molecular weight of 60 kDa. The calculated molecular weight of TRP-2, however, based upon the amino acid sequence lacking the putative signal sequence (Jackson et al, 1992), is 56 kDa, suggesting that the two forms of the protein observed upon Western blot are in fact distinct forms of TRP-2 that differ in their post-translational modi®cation, rather than the glycosylated and nascent unglycosylated forms suggested previously (Tsukamoto et al, 1992). To investigate this, cell lysates from con¯uent cells were treated with N-glycosidase F to remove completely N-linked oligosaccharides from the polypeptide chain or with endoglycosidase
H to distinguish between high-mannose glycosylated forms and more completely processed forms containing complex oligosaccharides. Western immunoblot (Fig 3) showed that N-glycosidase F treatment reduced both bands to a single band of 56 kDa on this gel. Endoglycosidase H treatment reduced only the lower band to a comigrating band, suggesting that the high molecular weight (HMW), endo H-resistant glycoform is the terminally processed form of TRP-2, whereas the low molecular weight (LMW) form is an incompletely processed, high-mannose intermediate. Closer examination of the Western blot in Fig 2(a) shows that the relative amounts of the HMW and LMW glycoforms of TRP-2 decrease at different rates when cells are trypsinized at con¯uence (lane 1, * in Fig 2a) and subsequently plated at low density (lane 2, day 1 in Fig 2a), with the HMW glycoform signi®cantly decreased at day 1 relative to the LMW glycoform, which does not decrease signi®cantly until day 3. Hence, the trypsinization of con¯uent cells is associated not only with an overall decrease in TRP-2 production, but also with selectively rapid disappearance of the HMW glycoform. To determine whether the effect observed with TRP-2 protein was more general, the blot was probed with aPEP1 antisera to TRP-1. This resulted in the detection of three distinct bands (Fig 2b), with the highest molecular weight band corresponding to the major glycosylated species described previously (Kobayashi et al, 1994). Some decrease in TRP-1 level was observed on days 1±3 following trypsinization, with subsequent increases at higher cell densities, but the changes were not of the magnitude observed with
VOL. 115, NO. 1 JULY 2000
CELL DENSITY AND TRP-2 EXPRESSION
109
Figure 3. Endoglycosidase sensitivities of TRP-2 glycoforms. Melan-a cells at con¯uence (2.0 3 107 cells per 10 cm plate or 2.6 3 105 cells per cm2) were scraped from the plate, counted, suspended in growth media, washed twice with ice-cold PBS, and lysed with RIPA buffer. Total protein concentration determination was performed against BSA standards. Eighty micrograms of total protein were loaded onto individual lanes of an 8% SDS-polyacrylamide gel before (lane 1) and after (lanes 2±5) digestion with N-glycosidase F and endoglycosidase H. TRP-2 glycoforms were detected by Western immunoblotting with aPEP8 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate.
TRP-2. Additionally, in contrast to TRP-2, selectively rapid disappearance of the highest molecular weight TRP-1 band was not observed. We conclude that both TRP-2 and TRP-1 protein levels are induced by increases in cell density, although the effect observed with TRP-2 is greater than that observed with TRP-1.
Figure 2. Cell-density-induced production of TRP-2 and TRP-1. Con¯uent melan-a cells were trypsinized, plated at 5 3 105 cells per 10 cm plate, and grown for 7 d. Media and TPA were changed and replenished daily. Total protein was obtained from cells at con¯uence (*) and on days 1±7 after plating at low density. One hundred micrograms of total protein from each set of cells were loaded into individual lanes of an 8% SDSpolyacrylamide gel and electrophoresed. (a) TRP-2 was detected by Western immunoblotting with aPEP8 antisera, horseradish-peroxidaseconjugated antirabbit, and chemiluminescent substrate. (b) TRP-1 was detected by Western immunoblotting with aPEP7 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate. (c) bactin was detected by Western immunoblotting with mouse antiactin monoclonal antibody, horseradish-peroxidase-conjugated antimouse, and chemiluminescent substrate. (d) Determination of cell densities. Cells at con¯uence prior to low density plating (*) and on days 1±7 after low density plating were counted following trypsinization and cell densities at trypsinization calculated. Error bars represent the mean 6 SD of four independent cell counts.
Time and TPA independence of cell-density-induced TRP-2 production The magnitude of the cell density effect observed with TRP-2 prompted further experiments to examine determinants of the effect. To determine whether the density dependence observed was associated with the length of time cells were grown in culture or the proliferation state of the cells, cells were plated at increasing densities, instead of the same low density, cultured for the same length of time, and harvested to determine relative levels of TRP-2 production. This was done both in the presence and in the absence of TPA, which is required for these cells to proliferate in culture (Bennett et al, 1987). The results (Fig 4a) show that, in the presence and in the absence of TPA, increased TRP-2 production is observed with increases in cell density when cells are plated at different densities and are harvested at the same time. In the presence of TPA (Fig 4a, lanes 1±4), increased TRP-2 production is observed over the full range of cell densities measured at harvest (Fig 4b, bars 1±4). In the absence of TPA (Fig 4a, lanes 5±8), an increase of TRP-2 production was observed only at the highest density of cells harvested (3.6 3 104 cells per cm2, lane 8 in Fig 4a). This density is comparable to the density in lane 2 where increased TRP-2 production was observed in the presence of TPA. The comparison of cell densities at harvest with plating density indicated that there was no cell proliferation over time (Fig 4b, bars 5±8). This result shows that cell proliferation is not required for density-dependent TRP-2 production to occur. Furthermore, the same lower level of TRP-2 production was observed for cells harvested at lower densities in the absence of TPA (lanes 5±7) as was observed for cells harvested at low density in the presence of TPA (lane 1). This demonstrates that growth arrest alone is not suf®cient for TRP-2 induction and implies that a threshold amount of cell-cell contact is required.
110
HORNYAK ET AL
Figure 4. Time and TPA independence of cell-density-induced TRP-2 production. Con¯uent melan-a cells (at 2.2 3 105 cells per cm2) were trypsinized and plated at increasing densities of 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2), 1 3 106 cells per 10 cm plate (1.3 3 104 cells per cm2), 2 3 106 cells per 10 cm plate (2.5 3 104 cells per cm2), and 5 3 106 cells per 10 cm plate (6.4 3 104 cells per cm2). Cells were grown in culture in the presence or absence of 200 nM TPA for 3 d, harvested by scraping, and cell lysate was prepared by the addition of RIPA buffer. Total protein concentrations were determined by comparison with BSA standards. (a) Total protein (7.5 mg) from cells grown in the presence (lanes 1±4) and in the absence (lanes 5±8) of TPA was loaded into individual lanes of an 8% SDS-polyacrylamide gel and electrophoresed. TRP-2 glycoforms were detected by Western immunoblotting with aPEP8 antisera, horseradish-peroxidase-conjugated antirabbit, and chemiluminescent substrate. (b) Determination of ®nal cell densities. Cells from samples initially plated at increasing densities were counted following removal from the plate by scraping and ®nal cell densities were calculated. Error bars represent mean 6 SD of four or six independent cell counts.
Independence from soluble factors of cell-density-induced TRP-2 production The medium was changed daily in cells plated at low density and cultured for extensive lengths of time to high density (Fig 2), in order to limit the potential effect of soluble factors present in higher concentrations in high density melan-a cells. Such soluble factors might induce TRP-2 production independent of intercellular contacts. An additional experiment was performed to test the effect of soluble factors upon TRP-2 production directly. Melan-a cells were plated both at high density (set A) and at low density (sets B, C, and D). After an initial period of growth, conditioned medium from high density plates was transferred to one set (set B) of low density plates. Growth was then continued and cells were subsequently harvested. As incubation with conditioned media from high density cells did not signi®cantly increase TRP-2 production (Fig 5), it is likely that soluble factors
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
Figure 5. Lack of dependence of cell-density-induced TRP-2 production on soluble factors. (a) Con¯uent melan-a cells (at 2.2 3 105 cells per cm2) were trypsinized and plated at densities of 5 3 106 cells per 10 cm plate (6.4 3 104 cells per cm2) (set A) or 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2) (sets B, C, and D). Thirtytwo hours after plating, conditioned media from sets A (high density plates) and C (low density plates) was transferred to sets B (low density plates) and D (low density plates), respectively, from which media had previously been removed. Fresh growth media and TPA were subsequently added to sets A and C. Cells were harvested by scraping 75±78 h after initial plating, counted, washed twice with ice-cold PBS, and lysed with RIPA buffer after centrifugation. Total protein concentrations were determined by comparison with a set of BSA standards, and 25 mg of total protein from each set of cells A-D were loaded into individual lanes of an 8% SDSpolyacrylamide gel and electrophoresed. Relative amounts of TRP-2 were detected by Western immunoblotting with aPEP8 antisera, horseradishperoxidase-conjugated antirabbit, and chemiluminescent substrate. (b) Determination of cell densities. Cells were counted after scraping from the plates and ®nal cell densities were calculated. Error bars represent mean 6 SD of six independent cell counts.
do not signi®cantly induce TRP-2 production in melan-a cells, and that cell-cell contact is the important determinant in densitydependent TRP-2 induction. This experiment also shows that trypsinization alone does not reduce TRP-2 production in cells
VOL. 115, NO. 1 JULY 2000
CELL DENSITY AND TRP-2 EXPRESSION
111
apart from cell density, as the TRP-2 signals from cells at high density trypsinized at the same time as cells at low density is much greater. Cell-density-induced expression of tyrosinase-related protein genes To correlate the cell-density-dependent inductions of TRP-2 and TRP-1 protein with changes in gene expression, we compared Dct and Tyrp1 mRNA levels directly as a function of cell density. Restriction fragments from divergent regions of the Dct and Tyrp1 cDNAs were prepared and used as probes on a northern blot containing total RNA prepared from cultured melan-a cells at con¯uence and at various times after low density plating. The signal from the 3¢ Dct probe (Fig 6a) con®rmed the density-dependent induction of gene expression noted above (Fig 1c). Hybridization with the 3¢ Tyrp1 probe also showed density dependence of Tyrp1 gene expression, although in a different pattern from that observed with Dct. The decrease in Tyrp1 expression following trypsinization and plating at low density is more rapid and greater in magnitude than with Dct, evident from the undetectable expression of Tyrp1 on day 3 compared with the expression of Dct. Tyrp1 expression is induced at lower cell densities than Dct expression, however, as can be seen by the higher level of Tyrp1 expression on day 5. A similar pattern of Tyrp1 induction was observed when the northern blot in Fig 1(c) was probed with the 3¢ Tyrp1 probe (data not shown). These results demonstrate that both Dct and Tyrp1 expression are increased by increased cell density in cultured mouse melanocytes, with the different patterns of induction occurring most probably by different mechanisms in each case. DISCUSSION Cell density effect and TRP-2 production Densitydependent induction of protein expression may take place through transcriptional, translational, or post-translational mechanisms. The 16-fold induction of Dct (Fig 1c, d; Fig 6a) and even larger induction of Tyrp1 mRNA expression (Fig 6a) noted with increased cell density shows that control at the transcriptional level is extremely important as a determinant of TRP-2 and TRP-1 levels. Additionally, we found that TRP-2 exists as two distinct glycoforms with differential endoglycosidase sensitivities. Production of each of these glycoforms was enhanced by increased cell density. The endo H sensitivity of the 63 kDa LMW glycoform, combined with the fact that the LMW glycoform is the form ®rst observed in a pulse-chase experiment (Tsukamoto et al, 1992), suggests that it is an incompletely processed glycoform of TRP-2 predominantly localized to the endoplasmic reticulum and the cis-Golgi, whereas the 75 kDa, HMW glycoform is likely to be the terminally processed form of TRP-2 present in melanosomes. As the HMW glycoform disappears more rapidly after con¯uent cells are trypsinized and plated at low density (Fig 2), it is tempting to speculate that melanosomal TRP-2 might be selectively degraded, or that the melanosome itself is rearranged or secreted by trypsinization. It is unlikely on the basis of existing evidence, however, that gross rearrangement or signi®cant secretion of melanosomes can account for the change observed in HMW TRP-2 levels after trypsinization. TRP-1 and TRP-2 are found together in the melanosome (Orlow et al, 1994), and it is unlikely that persistence of the highest molecular weight TRP-1 glycoform observed after trypsinization (Fig 2b) in the absence of detectable Tyrp1 message at a comparable time (Fig 6a) could occur with concomitant loss or signi®cant reorganization of melanosomes. These observations also suggest that fully processed TRP-1 is quite stable, in comparison to fully glycosylated TRP-2, following trypsinization and low cell density replating. The possibility that the melanosome undergoes dramatic rearrangement following trypsinization and replating of melanocytes is further discounted by our own observations (unpublished data) and those of others (R. Boissy, personal communication) of highly melanized melanosomes, both on the light and electron microscopic level, in samples of cells that have been subjected to trypsinization and
Figure 6. Cell-density-induced upregulation of tyrosinase-related protein genes. (a) Con¯uent melan-a cells (at 1.9 3 105 cells per cm2) were trypsinized and plated at 5 3 105 cells per 10 cm plate (0.6 3 104 cells per cm2). Total RNA was isolated from con¯uent cells initially (*) and from cells on days 3, 5, 7, 8, and 10 after plating. Total RNA (9.4 mg) from each sample was loaded onto individual lanes of a denaturing 1% formaldehyde agarose gel and separated by electrophoresis. RNA was transferred to nitrocellulose by northern blotting. Blot was hybridized with cDNA probes from the 3¢ end of Dct (nucleotides 1658±2182; Jackson et al, 1992), from the 3¢ end of Tyrp1 (nucleotides 2150±2479; Shibahara et al, 1986), and from gapdh (Clontech). Hybridization of the blot with both probes simultaneously (data not shown) showed two distinct bands, slightly different in size, as evidence that the probes did not cross-hybridize with each other's mRNA. (b) Determination of cell densities. Cells were counted after trypsinization and cell densities were calculated. Error bars represent mean 6 SD of six or eight independent cell counts.
replating. The selective loss of the HMW glycoform following trypsinization and replating at low cell density is an indication that a distinct, nontranscriptional mechanism can also regulate intracellular TRP-2 levels. Detection of the nascent, unglycosylated form of TRP-2 in these experiments is unlikely given the simultaneity of protein synthesis and glycosylation (Kiely et al, 1976; Bergman and Kuehl, 1978; Glabe et al, 1980) except in the presence of a glycosylation inhibitor such as tunicamycin. Our experiments agree with those of others (Tsukamoto et al, 1992) concerning the apparent molecular weights of forms of TRP-2; we do not observe
112
HORNYAK ET AL
a 110 kDa band in our experiments, as was previously reported (Halaban et al, 1996). It is possible that differences in immunoblotting techniques may account for the appearance of nonspeci®c bands. Cell density induction of gene expression ± cell contact versus growth arrest Cell-contact-induced gene expression may occur via mechanisms induced prior to or coincident with the state of growth arrest induced by nontransformed cells at con¯uence or by the withdrawal of growth media. The cell-density-dependent induction of the tyrosinase-related protein genes that we observe is probably not related to growth arrest. First, correlation of the cell growth curves (Figs 2b, 6b) with the patterns of induction of either TRP-2 production or of Dct and Tyrp1 expression, respectively (Figs 2a, 6a), suggests that induction of expression occurs at cell densities signi®cantly below con¯uence, implying that growth arrest per se is not required for induction of these genes. This argument has been made previously to explain the induction of DEP-1, a receptor-like protein-tyrosine-phosphatase, at subcon¯uent cell densities (Ostman et al, 1994). The mechanism of regulation in this case, either transcriptional, translational, or post-translational, was not determined. Additionally, we induced growth arrest by depriving melan-a cells of TPA required for their proliferation (Bennett et al, 1987). We did not observe an increase in TRP-2 production relative to cells proliferating in the presence of TPA. These factors suggest that cell-cell contact prior to any contact-dependent growth arrest induces expression of the tyrosinase-related protein genes. It would be interesting to attempt to prove this point by, for example, correlating the relative expression of Dct with the relative proportions of melan-a cells at different points of the cell cycle. Density dependence of induction of tyrosinase gene family members The fact that Dct and Tyrp1 are both induced at increased cell density in melan-a cells, together with the ®nding that Tyr is induced along with the concomitant induction of speci®c DNA-protein complexes at increased cell density in murine B16 melanoma cells (Mahalingam et al, 1997), suggests that these genes are coordinately upregulated by signals induced by cell-cell contact. As these genes each encode distinct proteins identi®ed to be enzymes in the melanin biosynthetic pathway, it is possible that cellular determinants may be extremely important in the maintenance of constitutive melanization or the induction of facultative melanization in the murine melanocyte. Although melanocytes normally reside among and in contact with keratinocytes in the follicular epithelium, they exist in contact with one another in certain cutaneous conditions in humans characterized by abnormalities in melanocyte proliferation, such as benign melanocytic nevi and malignant melanoma. Coordinate increases in the level of transcription of melanosomal enzyme genes may help explain the increased pigmentation characteristic of some of these lesions. The development of methods for studying the expression of these proteins on the single cell level by using ¯uorescent confocal microscopy or green-¯uorescent-proteintagged proteins may permit the assessment of whether heterogeneous cell±cell interactions, in addition to the melanocyte±melanocyte interactions we have described, are also capable of inducing tyrosinase-related protein expression. Results of these experiments may lead to insights regarding the nature of protein±protein interactions at the cell membrane responsible for inducing gene expression. Melanization involves not only the production of enzymes to convert tyrosine to melanin, but also the formation and maintenance of the melanosome, the subcellular organelle within which melanin is synthesized. It is possible that cell-contact-mediated mechanisms are important for the production of other melanosome-speci®c proteins whose formation is required for the transport of melanogenic enzymes to the melanosomes and for the production of other proteins, such as myosin V (Wei et al, 1997; Wu et al, 1997; Lambert et al, 1998), important for transport and appropriate localization of melanosomes within the cell.
THE JOURNAL OF INVESTIGATIVE DERMATOLOGY
We thank Ling-Yu Chiu and Jessica Gavin for technical assistance, and members of the Ziff laboratory including Ted Lee, Paul Issack, Chris Daly, and Latika Khatri for valuable advice and suggestions. We appreciate gifts of plasmids from Ted Lee, Ian Jackson, and Shigeki Shibahara, and are particularly grateful to Ian Jackson for making the Dct promoter plasmid available to us prior to publication. We thank Dorothy Bennett for permission to use the melan-a cell line and Vincent Hearing for the gift of aPEP8 and aPEP1 antisera. We thank Vijayasaradhi Setaluri for a close reading of the manuscript prior to submission. A question by Ruth Halaban was important for prompting the endoglycosidase studies. This work was supported by National Institutes of Health Grants AR01992 and AR45001 to T.J.H. E.B.Z. is an Investigator of the Howard Hughes Medical Institute.
REFERENCES Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG, Smith JA, Struhl K: Current Protocols in Molecular Biology, pp 4.9.1±4.9.13. New York: John Wiley, 1993 Bennett DC, Cooper PJ, Hart IR: A line of non-tumorigenic mouse melanocytes, syngeneic with the B16 melanoma and requiring a tumour promoter for growth. Int J Cancer 39:414±418, 1987 Bergman LW, Kuehl WM: Temporal relationship of translation and glycosylation of immunoglobulin heavy and light chains. Biochemistry 17:5174±5180, 1978 Budd PS, Jackson IJ: Structure of the mouse tyrosinase-related protein-2/ dopachrome tautomerase (Tyrp2/Dct) gene and sequence of two novel slaty alleles. Genomics 29:35±43, 1995 Chen C, Okayama H: High-ef®ciency transformation of mammalian cells by plasmid DNA. Mol Cell Biol 7:2745±2752, 1987 Chomczynski P, Sacchi N: Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162:156±159, 1987 Glabe CG, Hanover JA, Lennarz WJ: Glycosylation of ovalbumin nascent chains: the spatial relationship between translation and glycosylation. J Biol Chem 255:9236±9242, 1980 Halaban R, Bohm M, Dotto P, Moellmann G, Cheng E, Zhang Y: Growth regulatory proteins that repress differentiation markers in melanocytes also downregulate the transcription factor Microphthalmia. J Invest Dermatol 106:1266±1272, 1996 Jackson IJ, Chambers DM, Tsukamoto K, Copeland NG, Gilbert DJ, Jenkins NA, Hearing V: A second tyrosinase-related protein, TRP-2, maps to and is mutated at the mouse slaty locus. EMBO J 11:527±535, 1992 Kiely ML, McKnight GS, Schimke RT: Studies on the attachment of carbohydrate to ovalbumin nascent chains in hen oviduct. J Biol Chem 251:5490±5495, 1976 Kobayashi T, Urabe K, Winder A, et al: Tyrosinase related protein 1 (TRP1) functions as a DHICA oxidase in melanin biosynthesis. EMBO J 13:5818± 5825, 1994 Korner A, Pawelek J: Mammalian tyrosinase catalyses three reactions in the biosynthesis of melanin. Science 217:1163±1165, 1982 Lambert J, Onderwater J, Haeghen Y, Vancoillie G, Koerten H, Mommaas A, Naeyaert J: Myosin V colocalizes with melanosomes and subcortical actin bundles not associated with stress ®bers in human epidermal melanocytes. J Invest Dermatol 111:835±840, 1998 Mackenzie MA, Jordan SA, Budd PS, Jackson IJ: Activation of the receptor tyrosine kinase Kit is required for the proliferation of melanoblasts in the mouse embryo. Dev Biol 192:99±107, 1997 Mahalingam H, Watanabe A, Tachibana M, Niles RM: Characterization of densitydependent regulation of the tyrosinase gene promoter: role of protein kinase C. Exp Cell Res 237:83±92, 1997 Orlow S, Zhou B-K, Chakraborty A, Drucker M, Pifko-Hirst S, Pawelek J: Highmolecular weight forms of tyrosinase and the tyrosinase related proteins: evidence for a melanogenic complex. J Invest Dermatol 103:196±201, 1994 Ostman A, Yang Q, Tonks NK: Expression of DEP-1, a receptor-like proteintyrosine-phosphatase, is enhanced with increasing cell density. Proc Natl Acad Sci 91:9680±9684, 1994 Sambrook J, Fritsch EF, Maniatis T: Molecular Cloning: A Laboratory Manual. 2nd edn, p. 16 66. Plainview, New York: Cold Spring Harbor Laboratory Press, 1989 Shibahara S, Tomita Y, Sakakura T, Nager C, Chaudhuri B, Muller R: Cloning and expression of cDNA encoding mouse tyrosinase. Nucl Acids Res 14:2413±2427, 1986 Steel K, Davidson DR, Jackson IJ: TRP-2/DT, a new early melanoblast marker shows that steel growth factor (c-kit ligand) is a survival factor. Development 115:1111±1119, 1992 Tsukamoto K, Jackson IJ, Urabe K, Montague PM, Hearing VJ: A second tyrosinaserelated protein, TRP-2, is a melanogenic enzyme termed DOPAchrome tautomerase. EMBO J 11:519±526, 1992 Wei Q, Wu X, Hammer J: The predominant defect in dilute melanocytes is in melanosome distribution and not cell shape, supporting a role for myosin V in melanosome transport. J Muscle Res Cell Motil 18:517±527, 1997 Wu X, Bowers B, Wei Q, Kocher B, Hammer JI: Myosin V associates with melanosomes in mouse melanocytes: evidence that myosin V is an organelle motor. J Cell Sci 110:847±859, 1997