Expression Analysis of a Tyrosinase Promoter Sequence in Zebrafish

Introduction

Melanin formation in pigment cells is a highly visible characteristic of animals that can be used to study key events underlying cell-specific gene expression. Tyrosinase is the core enzyme in melanogenesis and is expressed specifically in melanin-producing cells. Research on tyrosinase gene activity has concentrated on the cis- and trans-acting factors controlling its spatiotemporal pattern. The 5′ flanking regions of numerous tyrosinase genes have been analysed. This has shown that both positive and negative regulatory elements work together to modulate tyrosinase promoter activity. In particular, the consensus E-box motif, CANNTG, is important in both activation of tyrosinase gene transcription and pigment cell-specific expression. The E-box motif sequence is the binding site for a family of transcription factors with a basic helix-loop-helix (bHLH) structure (1, 2). This includes the microphthalmia protein (Mi) that has a bHLH-leucine zipper structure. The gene encoding Mi was first cloned from the mouse microphthalmia (mi) locus (3, 4). Homologues of mouse mi have been cloned from human [microphthalmia-associated transcription factor (MITF) (5)] and zebrafish [mitfa/nacre and mitfb (6, 7)].

A CATGTG type E-box motif is found at the core of three different regulatory elements in tyrosinase gene promoters, the initiator region (Inr), an 11-bp element termed the M-box (8) and the tyrosinase distal element (TDE), which is generally located farther than 1 kb upstream from the transcription initiation site. All three elements are conserved in the human, mouse, chicken, quail and turtle promoter sequences while only Inr is conserved in the frog and medaka promoters (9). In the mouse and human promoters, the CATGTG E-box motifs of the three elements can be bound by the Mi and MITF proteins (10–14).

In this study we describe the isolation of a 1041 nt genomic DNA fragment from the 5′ flanking region of the zebrafish tyrosinase gene. Within this sequence we identify a putative translation start site and a position for transcription initiation. We also identify five E-box motifs, including a putative Inr, which may be required for efficient expression of the zebrafish tyrosinase gene promoter in melanin-producing cells. We test the activity of this promoter fragment in transiently transgenic zebrafish embryos and in human melanoma cell lines. The possible mechanisms underlying melanocyte-specific expression of tyrosinase in zebrafish are discussed.

Materials and methods

RNA Isolation and cDNA Preparation

Total RNA was extracted from whole embryos at 24-h postfertilization (hpf) and 26 hpf using the Rneasy midi-kit (Qiagen GmbH, Hilden, Germany) according to the manufacturer's instructions. Isolated RNA was treated with RNase free DNase (Promega Corporation, Madison, WI, USA) at 37°C for 30 min to remove any contaminating genomic DNA and then heated to 72°C for 30 min to inactivate the DNase. One microgram of treated, total RNA was used to produce randomly primed complementary DNA (cDNA) using the SuperScript TM II kit (Life Technologies/Gibco BRL, Gaithersburg, MD, USA) as described by the manufacturer except that Oligo(dT) primer at 500 μg/ml was replaced with random hexanucleotides at 40 ng/μl.

Genomic DNA Cloning and Sequencing

The isolation of a DNA fragment consisting of 1041 nt from the upstream region of the zebrafish tyrosinase gene by non-specifically primed suppression polymerase chain reaction (NSPS-PCR) has previously been described (15). This fragment was cloned into the pCR^®– Blunt II-TOPO vector (Invitrogen Corporation, Carlsbad, CA, USA) to produce plasmid tyroTOPO4. Sequencing was performed as described in the protocol provided for the ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin Elmer Life Sciences Inc, Boston, MA, USA). The oligonucleotide primers (Genset Pacific Pty Ltd., Lismore, NSW, Australia) for sequencing are listed in Table 1.

PCR on Genomic and cDNA to Delimit the Start of Transcription

The PCRs were performed using oligonucleotide primers (Table 1) at a final concentration of 0.4 μM each, 1 μl of cDNA from 26 hpf embryos, 0.130 ng or 0.026 ng of genomic DNA, two units of Taq DNA polymerase (Stratagene, La Jolla, CA, USA), 1 × PCR buffer (10 mM Tris–HCl pH 8.8 at 25°C, 1.5 mM MgCl₂, 50 mM KCl, 0.1% Triton-X) and 0.8 mM dNTP mix in a total volume of 25 μl. Temperature cycling conditions were 97°C for 2 min, then 35 cycles of 95°C for 40 s, 52°C for 55 s, ramp step of +0.5°C/s to 72°C, 72°C for 2 min followed by 72°C for 10 min and 25°C for 10 s in a PTC-200 Thermocycler (MJ research Inc, Watertown, MA, USA).

Construction of Plasmid tyrpromCS2lacZ

PCR of the tyrosinase sequence upstream from the start codon

To amplify an 814 nt fragment from tyroTOPO4, primers PSQSal1 (5′-GTCGACTTCTTAGGTGAGAATGC-3′) and PSQHind3 (5′AAGCTTCTGCAGGTCGAGGAGAGTGAGACGCTTCAC-3′) containing restriction sites SalI and HindIII, respectively, were used (Genset Pacific) in PCR at a final concentration of 0.4 μM each with 0.0756 μM of tyroTOPO4, 2.5 units of Pfu turbo DNA polymerase, 1 × PCR PFU buffer (Stratagene), and 0.8 mM dNTP mix in a total volume of 25 μl. Temperature cycling conditions were 97°C for 2 min, then 25 cycles of 95°C for 40 s, 35°C for 40 s, ramp of +0.5°C/s to 72°C and 72°C for 3 min and 50 s, followed by 72°C for 10 min and 25°C for 10 s in a PTC-200 thermal cycler. The tyrpromCS2lacZ plasmid was formed by replacement of the CMV promoter between the SalI and HindIII sites of pCS2-lacZ with the 814 nt zebrafish tyrosinase promoter sequence amplified above.

Injection of tyrpromCS2lacZ and CSK-lacZ Plasmids into Embryos

The tyrpromCS2lacZ plasmid was linearized by restriction with SalI (New England Biolabs Inc, Berverly, MA, USA) and then purified using the Qiagen PCR direct purification kit or, after excision from a 1% TAE agarose gel, using the Qiagen gel extraction kit. CSK-lacZ control DNA was not linearized. Purified DNA was injected at a concentration of 30 ng/μl in a 0.2% solution of phenol red (Sigma Chemicals Co, St Louis, MO, USA) into zebrafish embryos at the one cell stage. These were allowed to develop at 28.5°C to 26 or 72 hpf. To inhibit melanin formation, embryos were incubated from 6 hpf in system water supplemented with 0.2 mM phenylthiocarbamide (Sigma) and replaced daily.

Staining for LacZ activity in 26 and 72 hpf Embryos

The chorions of injected embryos were removed manually and the embryos and larvae were fixed at 4°C for 40 min in a solution of 4% formaldehyde (BDH Laboratory Supplies, Dorset, England), 0.02% glutaraldehyde (Sigma) and 0.2% Igepal (Sigma) in phosphate-buffered saline (PBS). Embryos and larvae were then washed thrice with 0.02% Igepal in PBS at room temperature for 5 min. They were then stored in 1.5 ml microfuge tubes in 80% glycerol at −20°C. For staining, embryos and larvae were removed from glycerol and washed extensively in 0.02% Igepal in PBS. They were then rinsed with 2 ml of staining buffer (15.4 mM Na₂HP0₄, 4.6 mM NaH₂P0₄, 150 mM NaCl, 1 mM MgCl₂, one drop of Igepal, 1.5 mM K₄Fe₃(CH)₆, 1.5 mM K₃Fe₃(CH)₆ and water in a final volume of 2 ml prewarmed to 37°C. Embryos and larvae were then placed in 2 ml of staining buffer containing 26.5 μl/ml of an 8% solution of X-gal (Progen Industries Limited, Darra, QLD, Australia) in DMSO (BDH Laboratory Supplies) and incubated at 37°C. After sufficient staining the embryos and larvae were washed with 0.02% Igepal in PBS thrice at room temperature for 5 min, placed in 80% glycerol and stored at −20°C. Transgenic embryos and larvae were counted and scored according to the presence of β-galactosidase activity observed using light microscopy.

Luciferase Assays of Promoter Activity

Cloning of tyrpromGL3-luciferase

The 814 nt tyrosinase promoter fragment was excised from tyrpromCS2lacZ with SalI and HindIII restriction enzymes (New England Biolabs) and subcloned between the XhoI and HindIII sites of the pGL3 basic luciferase vector (Promega).

Transfection of cell lines

Cells were plated at a density of 1 × 10⁵ cells per well in six or eight well tissue culture plates in a volume of 2 ml of Dulbelco's modified Eagle's medium (DMEM) with 10% foetal calf serum (FCS) (Thermo Trace, Noble Park, VIC, Australia). After 24 h at 37°C the growth medium was replaced and cells were cotransfected with one of the three test constructs, tyrpromGL3-luciferase, SV40pGL3 promoter vector (Promega) or the empty pGL3 vector plus the reporter construct pEGFPN1 (Clontech Laboratories Inc, Palo Alto, CA, USA) which expresses EGFP fluorescent protein. The two vectors were mixed at a ratio of 2:1 (2 μg of test plasmid and 1 μg of the reporter plasmid). The DNA mix was then premixed with 50 μl of DMEM (-FCS). To this, 50 μl of lipofectamine mix containing 1 μl of Lipofectamine 2000 (Life Technologies/Gibco BRL) and 49 μl of DMEM (FCS) was added and the solutions were kept at room temperature for 20 min. This was then added to the wells of the plates and incubated for 48 h. All transfections were carried out in duplicate. Following 48 h of incubation the cells were harvested. The transfection mix was removed from all cell cultures, which were washed with PBS prior to adding 1 ml of trypsin, made up from 0.05% porcine trypsin and 0.02% ethylenediaminetetraacetic acid (EDTA), for 10 min at 37°C. Trypsin was then inactivated by addition of 0.5 ml of DMEM with 10% FCS per well and the cells were pipetted into 1.5 ml microfuge tubes. Cells were then sedimented by centrifugation and resuspended in 800 μl of PBS; 100 μl of this was mixed with 100 μl of PBS and kept for flourescence-activated cell sorter (FACS) analysis to determine transfection efficiency. The remaining 700 μl was centrifuged, washed with PBS, centrifuged again and the pellet was resuspended in 50 μl of 1 × cell culture lysis reagent (Promega). A quick spin was performed to remove any cell debris, and the supernatant was used for the luciferase assay.

Luciferase assay

To measure the amount of luciferase produced, 10 μl of cell lysate was mixed with 100 μl of luciferase assay reagent (Promega) and the luminescence was measured in a scinitillation counter. Readings started at 20 s and continued for 1 min. Luciferase assay counts were corrected for transfection efficiencies by dividing counts by transfection efficiency (see below).

Transfection efficiencies

Cell samples were analysed by FACS to determine the percentage of cells emitting fluorescence as a result of transfection with the pEGFPN1 reporter plasmid. This was used to determine the percentage of transfected cells as an estimation of transfection efficiency of the test constructs.

Results and discussion

Isolation of cDNA and Genomic Fragments Derived from the Zebrafish Tyrosinase Gene

To obtain cDNA with which to probe a zebrafish genomic DNA library for tyrosinase gene sequences we applied the reverse transcriptase-PCR protocol of Inagaki et al. (16) on randomly primed cDNA prepared from zebrafish embryos at 24 hpf. This amplified a 585 nt cDNA fragment from the zebrafish tyrosinase gene (EMBL Accession No: AJ250302). In a phylogenetic and gene expression analysis published previously (17), we demonstrated that this cDNA sequence is derived from the zebrafish tyrosinase gene orthologue rather than from either of two genes encoding structually similar enzymes which are also involved in melanogenesis, tyrosinase-related protein 1 (TRP-1) and tyrosinase-related protein 2 (TRP-2). In situ transcript hybridization analysis of zebrafish embryos showed the expected pattern of expression in the retinal pigmented epithelium (RPE) and neural crest derived melanocytes (17).

Use of the zebrafish tyrosinase cDNA fragment to probe blots of genomic DNA libraries consistently resulted in high levels of non-specific binding. Therefore, we designed oligonucleotide primers corresponding to the upstream end of our cDNA clone and used non-specifically primed suppression PCR to isolate a 1041 nt genomic DNA fragment (15). This fragment was cloned into the pCR^®– Blunt II-TOPO vector and sequenced using various primers (Table 1). Sequencing showed that the fragment contains 73 nt overlapping with the zebrafish tyrosinase cDNA clone and 968 nt of new sequence (Fig. 1). The genomic sequence includes a putative translation start site (start codon) and 848 nt of sequence upstream of this (EMBL Acession No: AJ489318).

Determination of the Location of the Transcription Initiation Site

The 848 nt sequence upstream of the start codon does not contain a canonical TATA box. To determine a possible site of transcription initiation we used the Markov Chain Promoter Finder McPromoter V3.0 program (18), a statistical system that calculates the likelihood of the presence of a transcription start site using a neural network. This indicated that the most likely position for transcription initiation lies between −49 and −149 nt upstream of the start codon (Fig. 2. This corresponds to 700–800 nt of the sequence shown in Fig. 1). The output value closest to 1.0, predicting the most probable site of transcription initiation, is at approximately −84 nt upstream from the start codon (765 nt of the sequence in Fig. 1). This is in accordance with the common position of transcription initiation sites of other characterized tyrosinase promoters (19).

The relatively few number of cells that express tyrosinase in zebrafish embryos means that the absolute level of expression of this gene, compared with most other genes at the level of the whole embryo, is extremely low. Consequently, zebrafish cDNA libraries tend to lack Tyrosinase cDNAs and techniques for amplification of cDNAs from trancript 5′ ends (i.e. RACE) are either unsuccessful or cannot amplify the entire 5′UTR (unpublished data). Therefore, to confirm that the region of genomic DNA −49 to −149 nt upstream of the start codon encompasses the actual start of transcription, we conducted PCR tests on genomic DNA and randomly primed cDNA derived from embryos at 26 hpf. For these tests, a common oligonucleotide primer, TSS, binding downstream of the start codon was paired with primers binding at various distances upstream from the start codon (Table 1). Primer TSS3 binding −10 to −28 nt upstream of the start codon was able to amplify DNA from both genomic DNA and cDNA templates when paired with TSS. However, primers TSS12 and TSS4, both of which bind farther than −149 nt upstream of the start codon, were able to amplify DNA from the genomic DNA source but not from cDNA when paired with TSS (Fig. 3). These observations are consistent with the theoretical prediction of the position of the start of transcription. This implies that our cloned putative zebrafish tyrosinase promoter fragment possesses a region of at least 700 nt upstream of the transcription initiation site. This region may contain cis-acting elements necessary for pigment cell-specific expression.

Functional Analysis of the Zebrafish Tyrosinase Promoter Clone

Functional analysis of the mouse tyrosinase promoter has shown that a region of only 270 nt upstream of, and including, the transcription start site is sufficient to drive pigment cell-specific expression in cultured mouse melanoma cells. In vivo this region confers weak, but tissue specific and developmentally regulated gene expression (20). The same region is also sufficient to confer cell-specific expression in cultured human melanoma cells (21). This region contains the M-box and Inr, together with a positive regulatory element upstream from the M-box, which does not show a consensus binding site for known transcription factors and is not conserved in the promoter sequences of other known melanocyte-specific genes (11). There is also a negative regulatory element positioned between this positive element and the M-box. Ganss et al. (11) have demonstrated that the positive regulatory element and the M-box are not sufficient to confer pigment cell-specific expression. In the ascidian, Halocynthia roretzi, the 152 nt upstream from the start codon of the tyrosinase gene, HrTyr, is essential for expression in pigment cell precursors (9). However, this 152 nt region contains no sequences corresponding to known positive or negative cis-regulatory elements (such as CATGTG E-boxes) that direct pigment cell-specific expression in vertebrate tyrosinase promoters.

To examine whether our cloned zebrafish tyrosinase promoter region could direct pigment cell-specific expression, we used PCR to amplify an 814 nt DNA fragment containing all of the available sequence upstream of the translation start codon. This was then fused to an open reading frame encoding β-galactosidase in the pCS2-lacZ vector to generate construct tyrpromCS2-lacZ. The tyrpromCS2-lacZ construct was injected into zebrafish zygotes which were allowed to develop at 28.5°C until 26 hpf before fixation and analysis of β-galactosidase activity (lacZ expression). Control injections using a fusion of a cytoskeletal actin promoter of Xenopus with lacZ, CSK-lacZ (expected to produce lacZ expression in transgenic animals) were performed in parallel. As the melanin of pigmented cells obscures the visibility of staining for lacZ, melanin synthesis was inhibited using phenylthiocarbamide. In most injected embryos, the transgene is expected to become distributed mosaically (i.e. in some, but not all, cells). In our assay, 70% of embryos injected with tyrpromCS2-lacZ (from a total of 84 embryos) were transgenic and showed lacZ expression restricted to individual cells. However, the expression was not pigment cell-specific (Fig. 4A,B). Strong β-galactosidase activity was seen predominantly in myotubes. Activity was also frequently observed in the notochord, the epidermis covering the yolk, cells within or close to the epidermis of the embryo proper and, less frequently, in the RPE (Table 2). These observations are similar to those reported by Toyoda et al. (9) in H. roretzi. These authors observed that 152 nt of the HrTyr promoter is sufficient to direct lineage-specific expression and is essential for expression in pigment cell precursors. However, they also observed increased ectopic expression in the spinal cord and in a subset of muscle precursor cells when using this minimal HrTyr promoter region compared with using longer fragments.

In 43% of embryos, lacZ expression was seen in cells within or close to the epidermis of the embryo proper. To observe more easily whether the promoter fragment was driving expression of lacZ in neural crest derived melanocytes we stained embryos for β-galactosidase activity at 72 hpf. At this developmental stage melanocytes are easily distinguished because of their characteristic large and dendritic morphology. In this assay, 25 of 28 injected larvae (89%) showed transgene expression. All of the larvae showed β-galactosidase activity in a similar pattern to that seen for embryos stained at 26 hpf. No transgenic larvae showed lacZ expression in cells resembling that expected for neural crest derived melanocytes (Fig. 4C,D). These results demonstrate that the 814 nt of putative zebrafish tyrosinase promoter sequence upstream of the start codon can function in transcription initiation. However, although lacZ expression was observed in scattered cells of the RPE, this sequence is not sufficient to drive melanocyte-specific expression in the rest of the body. Presumably, the sequence lacks enhancer and repressor elements that would drive and restrict expression to melanin-producing cells. These elements most likely lie outside the cloned region.

Tyrosinase Promoter Activity in Human Melanoma and Non-melanoma Cell Lines

As a further test of our promoter fragment clone, we analysed its ability to drive luciferase expression in melanoma and non-melanoma derived cultured human cells. The 814 nt sequence upstream of the start codon used for the transgenic study was fused to a luciferase open reading frame in the pGL3 vector to create construct tyrpromGL3-luciferase. This construct was transfected into human melanoma cell lines NM39, WMM-1175 and MM170 and non-melanoma cell lines SAOS-2 and MCF7. Cells were cotransfected with the reporter plasmid pEGFPN1 to determine the transfection efficiency. Control experiments using the empty pGL3 vector (no expression expected) or pGL3 vector containing an SV40 promoter (SV40pGL3, strong expression expected) were conducted in parallel. Luciferase activity was assayed 48 h after transfection. Promoter activities were then expressed as a percentage of the activity of the positive control SV40 promoter in the SV40pGL3 construct. All cell lines transfected with the zebrafish tyrosinase promoter construct, tyrpromGL3-luciferase, showed levels of luciferase expression similar to that seen for cells transfected with the negative control (Fig. 5). Thus, our results show that our zebrafish tyrosinase promoter fragment has low or no activity in the human cell lines tested and that any activity is not specific to melanoma cells. This might be thought to be the result of an incompatibility of zebrafish promoters with the mammalian transcriptional machinery. However, concurrent testing of a 759 nt promoter fragment from the zebrafish mitfa/nacre promoter sequence that is capable of mediating melanocyte-specific expression in transgenic fish can drive high levels of luciferase expression specifically in human melanoma cell lines (unpublished observations).

Sequence Analysis of the Putative Zebrafish Tyrosinase Promoter Clone

Our functional studies demonstrate that the 814 nt of putative zebrafish tyrosinase promoter sequence upstream of the start codon can function in transcription initiation. However, this sequence is not sufficient to drive melanocyte-specific expression. Presumably, the sequence lacks enhancer and repressor elements that would drive and restrict expression to melanocytes. These elements most likely lie outside the cloned region. Therefore, we analysed the cloned sequence for the presence or absence of potential regulatory elements necessary for melanocyte-specific regulation of tyrosinase in zebrafish.

Sequence analysis of our putative tyrosinase promoter fragment (Fig. 1) revealed five CANNTG consensus E-box motifs including one CATGTG type E-box which corresponds to the E-box motif found within the TDE, M-box and Inr motifs of other tyrosinase promoter sequences. One explanation for the activity of our zebrafish tyrosinase promoter fragment in muscle precursor cells is the presence of CANNTG E-box motifs. This consensus E-box motif is known as a binding site for transcription factors with a bHLH structure, including myogenic proteins such as MyoD (1, 22). These proteins are responsible for muscle-specific gene transcription.

The CATGTG E-box lies −183 to −188 nt upstream from the start codon. It may be present within an Inr as these are found in roughly equivalent positions in the promoters of other vertebrate tyrosinase genes (9, 10, 23). The Inr generally includes a CATGTG E-box motif and an overlapping, non-consensus octamer element, GTGATAAT. Bentley et al. (10) demonstrated in vitro that, in the human TYROSINASE promoter, the E-box of Inr is essential for promoter function and activation by the Mi protein. They also suggested that, in the absence of a canonical TATA box, the transcription start site of the human TYROSINASE gene is determined by the interaction of transcription factors with the Inr and by binding of Sp1 at a site 43 nt upstream of the Inr.

The putative zebrafish tyrosinase promoter has no associated canonical TATA sequence and we have been unable to detect a non-consensus GTGATAAT octamer element overlapping the putative Inr. This is consistent with results previously observed for the tyrosinase gene of Medaka fish (9). Studies using the human sequence show that the octamer element binds to the POU domain transcription factors, Oct-1 (10) and Brn2/N-Oct3 (24). The transcription factor Brn2/N-Oct3 appears to down-regulate the tyrosinase promoter and can negate transactivation via the Inr E-box by Mi (24). However the octamer element of Inr may not be critically important as it is absent in the mouse, dog, frog and medaka sequences (9).

Similar to other Inr motifs, the zebrafish Inr motif possesses a conserved A residue immediately 3′ to the E-box (CATGTGA, Fig. 1). Both this and/or a T residue immediately 5′ to the E-box are conserved in sites known to be targets for Mi protein binding (25). The A and/or T residues enable Mi to discriminate between different CATGTG E-box motifs in melanocytes. Either or both of these residues are essential for Mi to bind to CATGTG E-box motifs. The Mi also recognizes and binds CACGTG E boxes indiscriminately (25). Mouse Mi and human MITF proteins can transactivate mouse and human promoter sequences by binding to E-box motifs (10–14). Although not yet tested, it is possible that the homologous zebrafish proteins also discriminate E boxes in the same manner, and thus bind to the putative Inr motif and to a CACGTG E-box located at −372 to −377 nt upstream from the start codon.

The cloned region of the zebrafish tyrosinase promoter has only one CATGTG E-box. In the H. roretzi, it has been shown that E-box motifs are not required for pigment cell lineage-specific expression from a cloned tyrosinase promoter (9). In the human TYROSINASE promoter, the E-box motifs present in the Inr and the TDE are essential for the activation of tyrosinase expression by MITF and are required for the efficient expression of the human TYROSINASE gene in pigment cells (12, 14). In the mouse promoter a deletion analysis has suggested the Inr E-box motif to be important in promoter function and it is believed to be primarily responsible for pigment cell-specific promoter activity (13). The putative Inr E-box in the zebrafish tyrosinase promoter sequence is, however, not sufficient to drive melanocyte-specific expression.

A non-consensus SP1 binding site is positioned between the M-box and Inr of the human TYROSINASE promoter and functions to enhance gene expression (10). It may also play a role in the basal transcription initiation of tyrosinase in the absence of a canonical TATA box (10, 26). The SP1 binding site sequence of the human TYROSINASE promoter is not conserved in other vertebrate sequences. However, SP1 binding sites, situated in similar positions and that can compete with the human site for SP1 binding, have been found in mouse, quail and possibly, chicken, dog and frog sequences (9, 23, 27). We are unable to detect anything resembling the non-consensus SP1 binding site of the human TYROSINASE promoter in the zebrafish sequence. However, we cannot be certain that no binding site exists based only on sequence inspection.

Our expression studies in transgenic zebrafish have shown that the cloned promoter fragment can, apparently, drive expression in the RPE but not in the neural crest derived melanocytes of the body. In vitro we observed that our cloned promoter region has low activity in both melanoma and non-melanoma cell lines from humans. Thus, the 814 nt region upstream from the start codon does not appear to contain the enhancer elements necessary for high levels of expression in melanin-producing cells.

The TDEs of the human and mouse tyrosinase promoters function as pigment cell-specific enhancers. In the human promoter the TDE is responsible for tissue-specific activity. The TDE function is encoded by a 20 nt sequence situated approximately 1.86 kb upstream from the start of transcription (12). Yasumoto et al. (12, 14) have shown that the human TDE is a strong enhancer that directs high level expression in pigment cells while the proximal region containing the M-box and Inr possibly functions to drive weak pigment cell-specific expression. In the mouse promoter sequence, Ganss et al. (28) have identified a pigment cell-specific TDE enhancer element located about 12 kb upstream from the transcription start site. As for the human TDE, the mouse TDE is a strong melanocyte cell-specific enchancer and is required for faithful expression of the gene in vivo (28). Possibly, an enhancer such as a TDE is necessary for correct tyrosinase gene expression in zebrafish. However, this must lie outside the region that we have cloned.