Fibroblasts expressing mouse c locus tyrosinase produce an authentic enzyme and synthesise phaeomelanin

More than one hundred mutations that affect pigmentation have been described in the mouse, mapping to over fifty different loci (Silvers, 1979). The study of the molecular biology of mouse pigment mutations has advanced rapidly during the last five years, as the functions of the products of several pigment loci have been defined (Bennett, 1991; Hearing and Jiménez, 1989; Hearing and Tsukamoto, 1991; Jackson, 1991). For many years it has been believed that the product of the c (albino) locus is tyrosinase (EC 1.14.18.1), the key enzyme involved in melanin synthesis. It was only relatively recently, however, that cDNAs mapping to the c locus were cloned (Kwon et al., 1989; Ruppert et al., 1988) and shown to encode a functional tyrosinase (Müller et al., 1988; Yamomoto et al., 1989).

It is now known that there exists a family of tyrosinaserelated proteins that are melanocyte-specific and share physical and antigenic properties (Hearing and Jiménez, 1989; Hearing and Tsukamoto, 1991). One tyrosinaserelated protein, TRP-1, is the product of the mouse brown locus (Jackson, 1988) and is thought to have tyrosinase activity in addition to the c locus protein (Jiménez et al., 1989, 1991), although it has also been suggested that this protein is a melanocyte-specific catalase (Halaban and Moellmann, 1990). A second tyrosinase-related protein, TRP-2, is the melanogenic enzyme dopachrome tautomerase and is the product of the mouse slaty locus (Jackson et al., 1992; Tsukamoto et al., 1992).

The presence of proteins related to c locus tyrosinase within melanogenic cells has caused confusion in the study of this enzyme (Hearing and Jiménez, 1989). In order to study the c locus tyrosinase in the absence of other proteins involved in melanin synthesis, we have established lines of mouse fibroblasts expressing mouse c locus tyrosinase (Winder, 1991). We have confirmed that this one enzyme catalyses the first two steps of melanin synthesis, both the hydroxylation of ^L-tyrosine to ^L-dopa and the oxidation of ^L-dopa to dopaquinone, and shown that tyrosinase-expressing fibroblasts synthesise a brown pigment. Fibroblasts do not contain melanosomes, the specialised organelles within which melanin is synthesised in melanocytes. Nevertheless, pigment is found within membrane-bound cytoplasmic vesicles in transfected cells (Winder, 1991). In this report, we describe further characterisation of the expressed tyrosinase at the levels of transcription, translation and post-translational modification, and demonstrate striking similarities between the expressed tyrosinase and the native enzyme. The apparent fidelity of this fibroblast expression system makes it a useful model for studying tyrosinase function and regulation.

We have also investigated the nature of the brown pigment synthesised by tyrosinase-expressing fibroblasts. Dopaquinone, the product of the oxidation of tyrosine by tyrosinase, is the common precursor of both eumelanin and phaeomelanin, the two main classes of melanin pigment found in mammalian skin and hair. Eumelanin pigments, which are black, arise from the oxidation of dopaquinone to indoles, whereas phaeomelanins arise from the oxidation of cysteinyldopa adducts and are brown (Prota, 1980). Cysteinyldopas arise either directly from the spontaneous reaction of cysteine and dopaquinone, or from the reaction of dopaquinone with glutathione followed by enzymatic hydrolysis; the major product being 5-S-cysteinyldopa in both cases (Rorsman et al., 1988). We have analysed tyrosinase-expressing fibroblasts for the presence of catechols, eumelanin and phaeomelanin, and show that expression of the single enzyme tyrosinase is sufficient for the synthesis of a pigment of phaeomelanic nature. Many additional proteins are involved in the synthesis of melanin polymers in melanocytes, and our ability to study their individual functions using expression systems should greatly improve our understanding of pigmentation.

Agarose (ultra-pure DNA grade) was purchased from Bio-Rad (Hemel Hempstead, UK). Ammonium acetate and gold-label ethanol were from Aldrich (Gillingham, UK). Concanavalin A-Sepharose and Ficoll were from Pharmacia (Uppsala, Sweden). Restriction enzymes, proteinase K and RNA molecular mass markers (II) were from Boehringer Mannheim (Lewes, UK). DNA kb ladder molecular mass markers were from BRL (Uxbridge, UK). All cell culture reagents were purchased from Gibco (Paisley, UK). Phosphate-buffered saline A (PBSA) tablets were from Oxoid (Basingstoke, UK) and trypan blue was from Flow Labs. (Irvine, UK). [α-³²P]dATP (3,000 Ci mmol⁻¹) was obtained from Amersham International (Amersham, UK). Bis-Tris, bovine serum albumin, bovine catalase (EC 1.11.1.6), ethidium bromide, MOPS, Nonidet P-40 (NP-40), polyvinylpyrrolidone, salmon testes DNA, sodium dodecyl sulphate (SDS), Triton X-100, Trizma base and ^L-tyrosine were from Sigma (Poole, UK). ^D-Dopa and ^L-dopa were obtained from Fluka AG (Buchs, Switzerland). Perchloric acid and hydriodic acid were from BDH (Poole, UK). All other reagents were of analytical grade.

3T3 Swiss mouse fibroblasts expressing mouse tyrosinase have been described previously (Winder, 1991). In brief, they were established by co-transfection with the tyrosinase expression vector pHDmcTyr1 (Müller et al., 1988) and a plasmid conferring G418 resistance. Clone c was derived from the original line 13.4 by two rounds of subcloning. Clone pKG4 was derived from 3T3 Swiss fibroblasts transfected with the G418-resistance plasmid alone. All lines were grown routinely in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% (v/v) foetal calf serum and 2 mM ^L-glutamine. Cells were passaged by incubation (37°C for 5 min) in PBSA / EDTA / trypsin (PBSA containing 0.02% (w/v) EDTA and 0.125% (w/v) trypsin). Mycoplasma tests were done routinely and were always negative.

Cells were harvested using trypsin and washed twice in PBSA then resuspended in four volumes of buffer A (20 mM KH₂PO₄ pH 7.5, 0.5% (v/v) Triton X-100). Lysates were sonicated for 10 s (MSE Ultrasonic Disintegrator model 7100, setting 5 μm) then centrifuged (9,000 g for 5 min at 4°C). The centrifugation step was repeated using the supernatant, and the final supernatant was stored at −20°C.

Cells were harvested and washed as for method A then cell pellets were frozen in solid CO₂ / ethanol. Approximately ten volumes of Millipore-purified water were added to each pellet, the suspensions were centrifuged (2,000 g for 10 min), and the water poured off. Pellets were resuspended in approximately ten volumes of buffer B (50 mM KH₂PO₄ pH 7.2) and homogenized by ten strokes of an all-glass Potter-Elvehjem homogenizer.

Cells were harvested and washed as for method A, then cell pellets were frozen in solid CO₂ / ethanol. Extracts were prepared by resuspending pellets in approximately ten volumes of Millipore-purified water for melanin determination, and ten volumes of 0.4 M perchloric acid for catechol determination.

The dopa oxidase activity of tyrosinase was determined using two different methods. The MBTH assay (Winder and Harris, 1991) uses 3-methyl-2-benzothiazoninone hydrazone (MBTH) to trap dopaquinone formed on the oxidation of _L-dopa. The product is a pink pigment with a sharp absorbance maximum at 505 nm and its formation is measured spectrophotometrically. In addition, dopa oxidase activity was assayed by measuring the formation of _L-5-S-cysteinyldopa from _L-dopa and _L-cysteine as previously described (Agrup et al., 1983; Wittbjer et al., 1989).

Catalase activity was determined by measuring the release of oxygen when a sample (200 μl) was added to an incubation chamber (3 ml) filled with 10 mM hydrogen peroxide in PBSA, pH 7.0 at 37°C. The mixture was stirred constantly and a Clark oxygen electrode (YS1 5331 Oxygen Probe, Ohio, USA) was used to record the slope of the oxygen release curve during the initial rapid release of oxygen (first 5 min) and this was compared with a standard curve constructed using different concentrations of bovine catalase.

Dopachrome tautomerase (EC 5.3.2.3) activity was determined by measuring the formation of indoles from _L-or _D-dopachrome. The procedure is a modification of the methods described by Palumbo et al. (1987) and Pawelek (1990). Dopachrome was prepared by mixing 0.5 ml of 2.5 mM _L- or _D-dopa in 0.01 M NaH₂PO₄, pH 6.0, with 0.5 ml of 5 mM NaIO₄ in the same buffer, and this solution was diluted 10-fold in 0.01 M NaH₂PO₄, pH 6.0. A sample (200 μl) was incubated with 0.5 ml of dopachrome for 5 min at 20°C. To stop the reaction 100 μl of the assay mixture was added to 30 μl of 0.4 M perchloric acid then transferred directly to 0.9 ml 0.1 M NH₄COOH, pH 6.0. The amounts of dopachrome and indoles in the assay mixture were determined by HPLC and fluorescence detection combined with u.v. detection.

Protein concentrations were determined either by using a protein assay kit (Bio-Rad) based on the method of Bradford (1976) or by the method of Lowry et al. (1951). Bovine serum albumin was used as a reference standard.

Cells were harvested using trypsin and counted in a haemocytometer (Weber, UK), then seeded in 5 ml DMEM plus penicillin (50 μg ml₋₁) and streptomycin (50 μg ml₋₁) in 25 cm² Nunc tissue culture flasks (Gibco) at a density of 8 × 10⁴ cells per cm². At 24-48 h intervals, cells were harvested from representative flasks using trypsin, the culture medium being pooled with the cells. Flasks were rinsed with PBSA and the washings also pooled. Cells were pelleted by centrifugation (300 g for 3 min at room temperature), and the colour of the pellet noted. Pigmentation was scored subjectively on an arbitrary scale from 1 to 8, reflecting the range of pellet colour from white (1) to brown (4) to black (8). Cells were resuspended in ice-cold PBSA, and a sample was mixed with an equal volume of trypan blue (0.5% (w/v) in 0.85% (v/v) saline) and counted: viable cells exclude the dye. An extract was prepared from the remaining cells by method A and assayed for tyrosinase activity.

Genomic DNA prepared using proteinase K and phenol extraction (Sambrook et al., 1989) was digested with restriction enzymes according to the manufacturer’s directions, and samples (25 μg) were electrophoresed through 0.8% (w/v) agarose gels containing 0.2 μg ml⁻¹ ethidium bromide, using 1 × TAE (40 mM Tris-HCl, 40 mM sodium acetate, 1 mM EDTA) as running buffer. Gels were blotted to nitrocellulose (Hybond-C extra, Amersham International) in 15 × SSC, 1 M ammonium acetate overnight (1 × SSC is 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0). Blots were baked for 2 h then pre-hybridised at 42°C for 2–4 h in 3 × SSPE, 5 × Denhardt’s solution, 50% (v/v) formamide, 1% (w/v) SDS, and 500 μg ml⁻¹ denatured salmon testes DNA (1 × Denhardt’s solution is 0.02% (v/v) Ficoll, 0.02% (w/v) bovine serum albumin, 0.02% (w/v) polyvinylpyrrolidone; 1 × SSPE is 0.17 M NaCl, 15 mM sodium citrate, 12.5 mM KH₂PO₄, 2 mM EDTA, pH 7.2). Blots were hybridised in the same solution at 42°C with ³²P-labelled cDNA probes, then washed: the final wash was in 0.5 × SSC + 0.1% (w/v) SDS at 65°C.

Total cytoplasmic RNA was prepared according to the method of Perbal (1984). Samples were resuspended in RNase-free water, denatured with formaldehyde and formamide, then electrophoresed through 1.4% (w/v) agarose gels containing 2.3 M formaldehyde using 40 mM MOPS, pH 7.0, 10 mM sodium acetate, 1 mM EDTA as running buffer. Gels were blotted to nylon membranes (Hybond-N+, Amersham International) in 0.05 M NaOH for 2-3 h. Blots were pre-hybridised in 6 × SSPE, 5 × Denhardt’s solution, 0.5% (w/v) SDS, 50% (v/v) formamide and 100 μg ml⁻¹ denatured salmon testes DNA. For hybridisation this solution was replaced by the same buffer omitting Denhardt’s solution and washing was as for Southern blots. To check for equal loading, blots were stripped by washing twice in boiling 0.1% (w/v) SDS, shaking gently for 15 min at room temperature after addition of each wash solution. Pre-hybridisation and hybridisation were as above except that the final wash was in 0.2 × SSC + 0.1% (w/v) SDS at 65°C.

cDNA probes were labelled with [α-³²P]dATP using a randomprimed DNA labelling kit (Boehringer Mannheim). The 2.0 kb tyrosinase cDNA probe was isolated from plasmid pHDmcTyr1 (gift from Dr. S. Ruppert) by digestion with EcoRI, and contains the complete coding sequence. The mouse glyceraldehyde phosphate dehydrogenase (GAPDH) probe was BamHI-linearised plasmid pGSM47.MGAP (gift from Dr. E. Whitelaw) containing a 1.4 kb GAPDH cDNA.

A homogenate was prepared by method B and centrifuged (100,000 g for 1 h). The pellet (membrane-bound fraction) was solubilised in 3% (v/v) Triton X-100, and tyrosinase activity was determined in this fraction and the supernatant (soluble fraction).

A Concanavalin A-Sepharose column (0.5 cm × 3.5 cm) was washed once with buffer C (1 M KCl, 20 mM KH₂PO₄ pH 7.2) and once with buffer D (20 mM KH₂PO₄ pH 7.2), then a sample of soluble fraction tyrosinase (2.2 ml) was applied. The column was washed with 2 ml buffer C followed by 2 ml buffer D, then eluted with 0.5 M methyl α-D-mannopyranoside in buffer D in 2 ml fractions, which were assayed for tyrosinase activity.

Tyrosinase was analysed by PAGE on 10% slab gels by the method of Laemmli (1970). Samples (50 μl) were mixed with 50 μl of solubilising buffer (20% (w/v) glycerol, 4% (w/v) SDS, and a small amount of bromophenol blue in 0.125 M Tris-HCl, pH 6.8). Running buffer (upper and lower) was 0.083 M Tris-HCl (one third of the concentration used by Laemmli (1970)), 0.192 M glycine, 0.1% (w/v) SDS, pH 8.3, and the gel was run at 7 mA for 30 min then 25 mA for 5.5 h. Tyrosinase activity was visualised by staining with dopa (2 mM L-dopa, 1.4 mM L-tyrosine in 0.1 M bis-Tris, pH 6.5). Trypsin-digested membrane-bound human tyrosinase (Wittbjer et al., 1989) was run on the same gel for comparison. Markers were phosphorylase b (94 kDa) and bovine serum albumin (67 kDa).

The amount of tyrosinase protein in cell extracts prepared by method B was measured by a competitive radioimmunoassay (Wittbjer et al., 1991). This assay measures both the membranebound and soluble forms of tyrosinase (Wittbjer et al., 1989, 1990a,b) and has a sensitivity range of 5-500 pmol tyrosinase per litre (Wittbjer et al., 1991).

Cell extracts were prepared by method C and the amounts of L-dopa, 5-S-L-cysteinyl-L-dopa and 5-S-glutathionyldopa were determined by HPLC and electrochemical detection (Eriksson and Persson, 1982). The detection limit for all three catechols is 1 pmol.

Cell extracts were prepared by method C and the insoluble pellet was subjected to reductive degradation in hydriodic acid as described by Ito and Fujita (1984). The yield of 4-amino-3-hydroxyphenylalanine is approximately 20%.

Cell extracts were prepared as for phaeomelanin determination and oxidative degradation in potassium permanganate was performed according to Ito and Fujita (1984). The yield of pyrrole-2,3,5-tri-carboxylic acid is approximately 2%.

We have previously described the isolation of four pigmented lines of mouse fibroblasts that express mouse tyrosinase (lines 11.7, 12.7, 13.4 and 13.5) (Winder, 1991). During long-term culture (2–3 months) both the tyrosinase activity and pigmentation of these lines decreases considerably, probably reflecting heterogeneity of the original lines. In an attempt to isolate clones of higher stability, two rounds of subcloning were undertaken using the two lines with the highest tyrosinase activity (12.7 and 13.4) and selecting the most highly pigmented clones. Several of these clones have higher tyrosinase activities than the original lines, and the maximum recorded activities of two clones derived from line 13.4 are 3-to 4-fold higher than the maximum for the parent line (Table 1). Clones derived form line 13.4 pigment most stably, and one particularly stable clone has been denoted clone c, since these cells express the mouse c locus tyrosinase.

In all four original lines of tyrosinase-expressing fibroblasts, and in clones derived from them, pigmentation is found to be more marked in some cells than others (Fig. 1). Qualitative observations suggest that pigmentation increases when the pH of the culture medium is raised, but again the degree of pigmentation varies from cell to cell (data not shown).

In all lines of tyrosinase-expressing fibroblasts, pigmentation increases with increasing cell density. To investigate whether this variation in pigmentation correlates with tyrosinase activity, clone c cells were seeded at a fixed density, and pigmentation, cell number and tyrosinase activity were measured over a time-course (Fig. 2). Tyrosinase activity (Fig. 2B) decreases over the first 24 h, increases gradually during the phase of exponential cell growth, then rises sharply in the stationary phase. Changes in tyrosinase specific activity parallel changes in pigmentation (Fig. 2C). The marked rise in tyrosinase activity in the stationary phase is not observed if the culture medium is changed every 48 h (data not shown).

The original lines 12.7 and 13.4 were analysed by Southern blotting to confirm the integration of the tyrosinase expression vector pHDmcTyr1 into the fibroblast genome. Three bands (approximately 12, 8 and 5 kb) hybridise to the tyrosinase cDNA mcTyr1 in EcoRI-digested genomic DNA from the 3T3 Swiss fibroblast parent (Fig. 3, lane 1). These bands correspond to the endogenous tyrosinase gene, which is not transcribed (Ruppert et al., 1988). DNA from line 13.4 contains one additional hybridising band of 2 kb (Fig. 3, lane 4), which is the full-length tyrosinase cDNA. All clones derived from line 13.4 show the same pattern of hybridising bands (data not shown), and Southern blots using DNA digested with HindIII and PstI demonstrate that there is one copy of the mcTyr1 sequence per genome integrated into these lines, including clone c (data not shown). In contrast, DNA from line 12.7 contains multiple additional hybridising bands compared to 3T3 Swiss (Fig. 3, lane 2) and derived clones show minor differences, confirming that the original line is heterogeneous (Fig. 3, lane 3, and data not shown). The additional bands of high molecular mass in line 12.7 may represent rearrangements or concatamers of the expression vector. Integration of the co-transfected G418-resistance marker into the genomes of lines 12.7 and 13.4 was also confirmed by Southern blotting (data not shown).

Clones derived from line 13.4, including clone c, produce two major tyrosinase transcripts of approximately 2.9 and 3.6 kb, and three less-abundant transcripts of approximately 1.9, 4.4 and 5.6 kb (Fig. 4A, lanes 3 and 4). For comparison, clones derived from line 12.7 contain transcripts equivalent to those found in clones derived from line 13.4 and an additional transcript of approximately 2.3 kb (Fig. 4A, lanes 5 and 6). In clones derived from line 12.7, the 4.4 kb transcript is most abundant. No tyrosinase transcripts are detectable in the parent 3T3 Swiss fibroblast line (Fig. 4A, lane 1) or in fibroblasts transfected with the G418-resistance plasmid alone (Fig. 4A, lane 2). Reprobing the same blot with a glyceraldehyde phosphate dehydrogenase cDNA (Fig. 4B) shows that this lack of hybridisation is not due to RNA degradation.

We have shown previously that the tyrosinase encoded by the cDNA mcTyr1 has both tyrosine hydroxylase and dopa oxidase activities when expressed in mouse fibroblasts (Winder, 1991). We also tested clone c for the activities of the tyrosinase-related proteins TRP-1 (a putative catalase) and TRP-2 (dopachrome tautomerase). There was no significant difference in either catalase or dopachrome tautomerase activities between the untransfected 3T3 Swiss fibroblasts and clone c cells (data not shown), confirming that the tyrosinase encoded at the c locus does not have either of these activities.

In melanocytes and melanoma cells tyrosinase exists in both membrane-bound and soluble forms. Proteolytic cleavage of the membrane-bound enzyme near the C terminus produces a soluble form with equivalent activity (Wittbjer et al., 1989, 1990a). We tested whether both forms are present in clone c by measuring tyrosinase activity in membrane-bound and soluble fractions prepared from a crude homogenate: 60-70% of tyrosinase activity is soluble, the remaining 30-40% being membrane-bound.

Tyrosinase is normally highly glycosylated, containing both N-linked and O-linked carbohydrates (Ferrini et al., 1987; Halaban et al., 1983; Ohkura et al., 1984). To determine whether the enzyme in transfected fibroblasts is glycosylated, soluble tyrosinase from clone c was analysed by Concanavalin A-Sepharose chromatography. The enzyme from transfected cells has strong affinity for the column: 85% binds and can be eluted only with a concentrated mannose solution. Hence, the expressed tyrosinase is a highmannose glycoprotein and contains N-linked carbohydrate.

The molecular masses of the membrane-bound and soluble forms of mouse tyrosinase from clone c are similar to those of the native mouse and human enzymes (Jiménez et al., 1991; Wittbjer et al., 1989, 1990a). For the membranebound form, the majority of tyrosinase activity is concentrated in a band of approximately 72 kDa (Fig. 5, lane 2). It is slightly larger than the purified membrane-bound human tyrosinase (approximately 69 kDa), but the latter was digested with trypsin during its isolation (Wittbjer et al., 1989). The higher molecular mass bands represent aggregates of the enzyme formed under non-reducing conditions, and the broadness of the bands is probably due to heterogeneity in glycosylation (Ferrini et al., 1987; Ohkura et al., 1984). Soluble tyrosinase from transfected fibroblasts is identical in size to the soluble human enzyme (Wittbjer et al., 1990a), having a molecular mass of approximately 53 kDa (data not shown).

Having further characterised the expressed tyrosinase, we estimated the amount of tyrosinase as a percentage of total cellular protein in transfected fibroblasts. A rabbit antityrosinase antibody that recognises both the membranebound and soluble forms of the enzyme (Wittbjer et al., 1990b, 1991) shows strong affinity for the expressed tyrosinase, and was used in a radioimmunoassay for the enzyme. It is estimated that tyrosinase constitutes approximately 0.008% of the total protein in cells from clone c.

To determine whether the brown pigment synthesised by tyrosinase-expressing fibroblasts is melanin, we tested a highly pigmented sample of clone c for the presence of the melanin synthesis intermediates L-dopa, 5 -S-cysteinyldopa, and 5-S-glutathionyldopa (Table 2). These compounds are only found at significant levels in cells that express an enzyme capable of hydroxylating tyrosine: L-dopa is the product of this reaction, and the two conjugates are intermediates in the synthesis of phaeomelanin. Clone c contains L-dopa at a level just above background and significant amounts of both 5-S-cysteinyldopa and 5-S-glutathionyl-dopa (Table 2). Untransfected fibroblasts do not contain significant amounts of any of these compounds (Table 2). We analysed the same preparation of cells from clone c for the presence of eumelanin and phaeomelanin. We could not detect pyrrole-2,3,5-tricarboxylic acid, the hydrolysis product of eumelanin, but the low yield of this compound limits assay sensitivity. The hydrolysis product of phaeomelanin, aminohydroxyphenylalanine, was readily detectable at a concentration of 0.19 μg per mg of protein (corresponding to 0.95 μg phaeomelanin per mg of protein), showing that clone c contains a phaeomelanin-like pigment.

Neither eumelanin nor phaeomelanin could be detected in untransfected fibroblasts.

Fibroblasts expressing the mouse c locus protein produce an enzyme that is post-translationally processed and has both the tyrosine hydroxylase and dopa oxidase activities of tyrosinase (Winder, 1991). It does not share the dopachrome tautomerase activity of the tyrosinase-related protein TRP-2, or the proposed catalase activity of TRP-1, confirming that the expressed c locus protein is an authentic tyrosinase.

In melanocytes and melanoma cells tyrosinase is synthesised as a membrane-bound enzyme, and a proportion undergoes proteolytic cleavage near the C terminus, releasing an enzymatically active soluble form (Wittbjer et al., 1989, 1990a). Tyrosinase in clone c is proteolytically processed, and 60-70% of the enzyme is soluble, a value comparable to that for tyrosinase in melanocytes from bovine eye (Wittbjer et al., 1990b). The expressed tyrosinase is rich in N-linked carbohydrate, indicating that it is processed in the Golgi apparatus in a similar manner to the native enzyme (Ferrini et al., 1987; Halaban et al., 1983; Ohkura et al., 1984). Additional evidence that the enzyme is post-translationally processed comes from the finding that mouse tyrosinase from transfected fibroblasts has a similar apparent molecular mass to the native enzyme. Hence, melanocytic differentiation does not appear to be a prerequisite for post-translational modification of tyrosinase.

Fibroblasts containing a single integrated copy of the tyrosinase expression vector pHDmcTyr1 (line 13.4) synthesise five different tyrosinase transcripts, and equivalent transcripts are found in line 12.7. The only integrated tyrosinase sequence that the two lines have in common is the uncorrupted 2 kb cDNA (Fig. 3), suggesting that all five mRNAs arise directly from this sequence. It is unlikely that they arise from rearrangement of the vector or by transcription of adjacent genomic sequences, since this would require identical genetic changes to have occurred in both lines. In the tyrosinase expression vector pHDmcTyr1 the tyrosinase cDNA is followed by the SV40 splice and polyadenylation sequences (Müller et al., 1988), and transcripts are expected to be approximately 3.5 kb in size if the SV40 signals are used. The polyadenylation signal AATAAA is present in the mcTyr1 cDNA, and alternative usage of this site and the SV40 signal may account for the two major transcripts in line 13.4. Furthermore, twelve different alternatively spliced transcripts of the mouse tyrosinase gene have been reported (Porter and Mintz, 1991; Ruppert et al., 1988; Terao et al., 1989; Yamamoto et al., 1989), including three that result from splicing reactions within the first exon (Porter and Mintz, 1991). It is therefore possible that the multiplicity of transcripts is at least in part due to alternative splicing. Further experiments are needed to clarify the nature of the multiple tyrosinase mRNAs in transfected fibroblasts.

Clone c has significantly higher tyrosinase activity than its parent line 13.4, but this level of activity is still an order of magnitude lower than that of highly active melanoma cell lines, such as IGR 1 (Karg et al., 1991a) and RVH 421 (Winder and Harris, 1992). We have attempted to increase the tyrosinase activity of transfected fibroblasts further by transfecting clone c with a second tyrosinase expression vector also containing the mcTyr1 sequence (unpublished observations). None of the fifty-two clones screened so far has significantly higher tyrosinase activity than clone c, suggesting that there may be a maximum level of enzyme activity compatible with survival of cells that do not normally synthesise tyrosinase. Enzyme activity probably approaches this limit in clone c, since the enzyme constitutes approximately 0.008% of total cell protein, which is close to the lower end of the range of 0.01 to 0.1% reported for melanocytes and melanoma cells (Hearing and Jiménez, 1987, 1989).

Tyrosinase catalyses the oxidation of tyrosine through to dopaquinone, which rapidly either undergoes spontaneous oxidation or reacts with thiol groups. The level of tyrosinase activity that fibroblasts can sustain is probably related to the production of cytotoxic quinones and peroxides in these reactions (Graham et al., 1978; Ito et al., 1983; Parsons, 1985). Tyrosinase-expressing fibroblasts synthesise pigment within membrane-bound vesicles as do melanocytes (Bouchard et al., 1989; Winder, 1991) but lack melanosomal proteins that are probably essential for protection against cytotoxic compounds generated by tyrosinase. Even when melanosomes are present, there appears to be a limit to the level of sustainable tyrosinase activity: when mouse tyrosinase was transfected into albino mouse melanocytes, pigment was synthesised within melanosomes but the transformants tended to die due to over-expression of the enzyme (Yamamoto et al., 1989).

Clone c has several physiological characteristics in common with melanoma cell lines, including non-uniform pigmentation, increased pigmentation with higher cell density, and increased pigmentation when the pH of the culture medium is raised (Bennett, 1983; Laskin et al., 1982; Winder and Harris, 1992). Increased tyrosinase activity and pigmentation in the stationary phase of growth would be expected if enzyme and pigment are synthesised at constant rates. However, such a mechanism cannot account for the decrease in enzyme specific activity over the first 24 h of the experiment when no significant cell growth occurs. Since the enzyme in clone c is expressed under a heterologous promoter (Müller et al., 1988; Winder, 1991) it is likely that activity is regulated at a translational or posttranslational level. The similarities in pigmentation between clone c and melanoma lines suggest that certain control mechanisms operating on tyrosinase are not specific to melanocytes. Induction of tyrosinase in the stationary phase may be related to oxidative stress, in particular to production of hydrogen peroxide in the medium by extracellular oxidation of melanin synthesis intermediates (Karg et al., 1991b, 1992). This is consistent with the observation that frequent changing of the culture medium suppresses the increase in tyrosinase activity.

It has been shown previously that expression of tyrosinase in fibroblasts is sufficient for the synthesis of a brown pigment (Bouchard et al., 1989; Winder, 1991). In this report we present evidence that the pigment in clone c is of a phaeomelanic nature. The catechols L-dopa, 5-S-cysteinyldopa and 5-S-glutathionyldopa are found in cells synthesising melanin (Rorsman et al., 1988), and are detectable in clone c cells but not in untransfected fibroblasts. In IGR 1 human melanoma cells the levels and ratios of these compounds vary with growth phase, the amounts of L-dopa and 5-S-cysteinyldopa increasing considerably in the stationary phase (Karg et al., 1991a; Rorsman et al., 1988). In stationary phase clone c cells the levels of L-dopa and 5-S-cysteinyldopa are approximately 35-40% of those in IGR 1 cells, but the level of 5-S-glutathionyldopa is 45% of that in IGR 1. Since the lifetime of dopaquinone is short, the formation of thiol adducts requires thiols to be present at the site of dopaquinone synthesis, and in human melanoma cells it is thought that dopaquinone is generated in a sub-cellular compartment enriched in cysteine (Karg et al., 1991a; Rorsman et al., 1988). The higher concentration of 5-S-cysteinyldopa relative to 5-S-glutathionyldopa in clone c cells suggests that cysteine is also present at an elevated level at the site of dopaquinone production in transfected fibroblasts, presumably the membrane-bound vesicles in which pigment is located (Winder, 1991). Formation of 5-S-glutathionyldopa is enhanced in melanoma cells growing under non-optimal conditions, and may arise from non-specific oxidation of L-dopa outside melanosomes. In clone c cells, the relatively high level of 5-S-glutathionyldopa probably reflects stress caused by the production of cytotoxic melanin synthesis intermediates, and the lack of true melanosomes.

Detection of a phaeomelanin-like pigment in clone c is consistent with the presence of 5-S-cysteinyldopa and 5-S-glutathionyldopa, and the concentration of the pigment is comparable to that of phaeomelanin in IGR 1 human melanoma cells (unpublished observations). The fact that no eumelanin could be detected may be due to the limited sensitivity of the assay, but even if present at a concentration below the assay detection limit, it is of low abundance relative to the pigment of phaeomelanic nature. This implies that all, or almost all, the dopaquinone produced by tyrosinase is “mopped up” by reaction with thiols and that virtually none is left to oxidise spontaneously to eumelanin. In turn, this suggests that in melanocytes and melanoma cells eumelanin is only produced when the level of tyrosinase activity is high enough to saturate the supply of thiols, and/or that other melanocyte-specific proteins are necessary for the formation of eumelanin. Many other proteins besides tyrosinase are indeed involved in the synthesis of melanin in vivo, including the related proteins TRP-1 and TRP-2. Analysis of the functions and interactions of these proteins using expression systems such as the fibroblast system we have described should help greatly to improve our understanding of the complex process of melanin synthesis.

Thanks are due to Mrs K.W. Brownsill for technical assistance, and to Prof. H. Harris for many helpful discussions. We also thank Dr. S. Ruppert for the generous gift of plasmid pHDmcTyr1 and Dr. E. Whitelaw for plasmid pGSM47.MGAP. A. J. W. was supported by the Cancer Research Campaign (UK) and a Junior Research Fellowship from Brasenose College, Oxford. This work was also funded by the Swedish Cancer Foundation and the Swedish Cancer Society (Project 626-B91-20XAB).