Identification of a Distal Enhancer for the Melanocyte-Specific Promoter of the MITF Gene

Plasmid Construction

An MITF genomic clone P1-5293 (13), carrying exon 1B and exon 1M (10), was digested with SacI, and the resulting SacI-SacI DNA fragments were subjected to Southern blot analysis. A hybridization probe was the XbaI-BamHI fragment (positions −2457 to −2147), upstream from the M promoter. One hybridizable SacI-SacI fragment of about 21 kb was identified and subcloned into pBluescript II SK+. The SacI site at its 5′ end is located 15 kb upstream from the M promoter and the 3′SacI site may be located in the downstream intron between exon 4 and exon 5. The 5′-flanking region of 15 kb was inserted into luciferase reporter plasmid pGL3-basic (Promega, Madison, USA), yielding pGL3-M-15k. Similarly, deletion constructs, pGL3-M-14k, pGL3-M-10k, pGL3-M-7k, and pGL3-M-3k, were generated. A reporter construct pGL3-MITF/M (21) and pL1, a promoterless construct, were constructed as previously described (22). A series of plasmids, containing the internal deletion of 15 kb, were constructed with the PCR-amplified DNA fragments. The PCR products were first subcloned into pGEM-T vector (Promega), and then sequenced. The target fragment, containing a putative enhancer region, was ligated to the fusion gene containing the proximal M promoter region (−120 to +95) upstream of the firefly luciferase gene.

Base changes were introduced into the potential SOX10-binding sites S1–S4 in the putative enhancer region of pGL3-bp-cis15 by the PCR-based strategy (23), yielding pGL3-bp-cis15 S1mut, pGL3-bp-cis15 S2mut, pGL3-bp-cis15 S3mut, pGL3-bp-cis15 S4mut, pGL3-bp-cis15 S1/2mut, pGL3-bp-cis15 S1/2/3mut, and pGL3-bp-cis15 S1/2/4mut. These mutant plasmids carry a single or a combination of the following base changes. S1 (CATTGAA) was changed to S1mut (CAggaAA); S2 (AACAAAA) to S2mut (AAgggAA); S3 (TTTTGTT) to S3mut (TTcccTT); and S4 (AACAAAA) to S4mut (AAgggAA).

A SOX10 expression vector was constructed by placing human SOX10 cDNA under control of the CMV promoter in pcDNA3 (Invitrogen, Carlsbad, USA), yielding pcDNA3.SOX10. The full-length SOX10 cDNA was prepared by reverse transcriptase–polymerase chain reaction (RT–PCR) from 624-mel human melanoma cells using a sense primer (5′-CATGGCGGAGGAGCAGGATCTATCGGAGGTG-3′) and an antisense primer (5′-TGAGGTGGGCAAGGAACAGGGCACACAGGCT-3′) (16). The identity of the cloned SOX10 cDNA was confirmed by determining its nucleotide sequence.

Cell Culture

HeLa human uterine cervical cancer cells were cultured in Minimum Essential Medium (Sigma, St. Louis, USA) containing 10% fetal bovine serum (FBS) and antibiotics at 37 °C under 5% CO₂/95% room air. HMV-II human melanotic melanoma cells were obtained from RIKEN Cell Bank and cultured in Ham's F-12 (Gibco, Rockville, USA) medium containing 10% FBS and antibiotics at 37 °C under 5% CO₂/95% room air (24, 25).

Transient Transfection and Reporter Analysis

HeLa cells and HMV-II cells, seeded to about 70% confluent in a 6-cm dish, were transfected by the calcium phosphate precipitation method as described previously (10, 26). The amounts of DNA used for transfection were 7 μg of test plasmid and 1.4 μg of β-galactosidase expression vector as an internal control. The duration of glycerol shock was 3 min for both cell lines. Transfected cells were then incubated for 24 h at 37°C, and lysed with LCβ (Toyo Ink, Tokyo, Japan). Cellular extracts were assayed for luciferase activity by using a PicaGene luciferase assay system (Toyo Ink, Tokyo, Japan) and Lumat LB9507 (Berthold, Bad Wildbad, Germany). Luciferase activities were normalized for transfection efficiency by dividing with β-galactosidase activity. For cotransfection assay, 2 μg of a reporter plasmid and SOX10 cDNA and 0.5 μg of β-galactosidase expression vector were introduced into HeLa cells, which lack endogenous MITF-M (8, 10) and SOX10 (27).

Gel Mobility Shift Assay

COS-7 monkey kidney cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% FBS and were transfected with pcDNA3 (vector) or pcDNA3.SOX10 (SOX10 cDNA), as described above. Nuclear extracts were prepared from transfected COS-7 cells by the method of Schreiber et al. (28). Gel mobility shift assay was performed as described previously (29). A 31-mer oligonucleotide, 5′-GTTAACAAAAATTGCCATTGAAGTCTGGAAG-3′, representing the 3′-region of the MITF distal enhancer, was end-labeled with [γ-³²P]ATP and T4 polynucleotide kinase, electrophoretically purified and used as a binding probe. The ³²P-labeled probe (approximately 10 fmol, 5 × 10⁴ cpm) was incubated with nuclear extracts (0.5 μg of protein) for 25 min on ice in a 20-μl reaction mixture containing 5% glycerol, 100 mM KCl, 10 mM (pH 7.4), 1 mM DTT, 2 μg poly(dI-dC) as a unspecific competitor. Competitor oligonucleotides were added at 20-fold molar excess. The competitors with the base changes (underlined) were mutS1(5′-GTTAACAAAAATTGCCAGGAAAGTCTGGAAG-3′), mutS2(5′-GTTAAGGGAAATTGCCATTGAAGTCTGGAAG-3′), and mutS1/S2(5′-GTTAAGGGAAATTGCCAGGAAAGTCTGGAAG-3′). After incubation, the samples were loaded onto 6% polyacrylamide gels and electrophoresed in 0.25 × TBE buffer at 150 V for 2 h. Gels were dried and exposed to X-ray films for autoradiography.

In situ Hybridization

BALB/c albino mouse embryos at various embryonic days were collected from pregnant mice, anesthetized with diethylether, and quickly fixed with 4% paraformaldehyde in 10 mM phosphate-buffered saline (PBS). The specimens were immersed in graded series of sucrose in PBS, embedded in OCT compound (Miles, Elkhart, IN, USA) and frozen. Sections were cut at 15 μm thickness with cryostat and mounted on silane-coated slides. Sox10 antisense and sense RNA probes were prepared, as previously described (30). Two types of MITF RNA probes correspond to a downstream region common to all MITF isoforms (30) and the B1b domain that is not present in MITF-M (8, 10). Digoxigenin-labelled antisense RNA probes were generated with digoxigenin-RNA labelling kit (Roche Diagnostics, Mannheim, Germany). Non-radioactive in situ hybridization was performed according to the standard procedures (31).

Accession Number

The nucleotide sequence data shown have been deposited in the GSDB/DDBJ/EMBL/NCBI DNA databases with the following accession number AB067452.

Identification of the Distal Enhancer for the M Promoter of the MITF Gene

To search for the cis-regulatory elements that are involved in the pigment cell-specific transcription of the M promoter, we analysed the function of the 5′-flanking region of exon 1M by transient expression assays (Fig. 1). Consistent with our previous report (13), a construct, pGL3-MITF/M, containing the 5′-flanking region of 2.2 kb, showed high promoter activity in melanoma cells but only marginal activity in HeLa cervical cancer cells that lack the expression of endogenous MITF-M (8). Interestingly, a construct pGL3-M-15k, containing the 5′-flanking region of 15.2 kb, gave rise to the highest luciferase activity in melanoma cells compared with pGL3-MITF/M or pGL3-M-14k that contains the 13.8-kb 5′-flanking region. Such an enhancing function was undetectable in HeLa cells. These results suggest that a cell-specific enhancer element may be present between −15.2 k and −3.8 k. In addition, there are several silencer elements that are distributed between −13.8 k and −2.2 k, as judged by the variability in the expression levels of fusion genes in melanoma cells.

To assess the role of the distal region for the M promoter activity, we constructed three types of internally truncated constructs (Fig. 2A). The enhancing activity was detected with the constructs carrying the proximal promoter region of 2.2 kb or 120 bp but not with a construct carrying the promoter region of 46 bp. Thus, the proximal promoter region (−120 to −46) is required for the enhancer function of the distal cis-acting region, but is not sufficient to direct expression of the luciferase gene in a pigment cell-specific manner (Fig. 2A), as already reported (13). In addition, the function of the distal region was further enhanced by the downstream region (−2.2 k to −120), which contains several cis-acting elements, such as SOX10 and PAX3 (14–18).

Accordingly, for simplicity, we analysed the function of the distal region by using the internally truncated constructs carrying the proximal promoter region (−120 to +95) (Fig. 2B). A construct pGL3-bp-cis1 contains the distal region (−15.2 to −13.8 k) but lacks the SOX10-binding sites that are clustered in the proximal region. Expression level of pGL3-bp-cis1 was about fivefold higher in melanoma cells than that of pGL3-bp (−120), thereby confirming the enhancer function of the distal region. The 298-bp fragment (−14.8 to −14.5 k) carried by pGL3-bp-cis15 gave rise to the highest luciferase activity (about 12-fold). In addition, both the 5′- and 3′-regions of this segment are required for the enhancer function, because deletion of its 5′- or 3′-portion abolished the enhancer function, as evident from the functional analysis of pGL3-bp-cis18 and pGL3-bp-cis16. Therefore, the distal segment (−14.8 to −14.5 k) may function as an enhancer for transcription from the M promoter in melanoma cells and is referred to as MITF-M distal enhancer (MDE) (Fig. 3). The nucleotide sequence of MDE was also found in the database of human genome projects (accession number AC010959), but the distance between exon 1B and exon 1M of the MITF gene was unknown because of many gaps in the published database. On the other hand, the 3′-flanking sequence of MDE appears to reduce the enhancer activity of MDE, as judged by the lower expression level of pGL3-bp-cis9.

Features of MDE and its Deletion in Mitf^mi–rw Mice

The 5′- and 3′-regions of MDE are required for its enhancer function (Fig. 2B) and contain potential SOX10-binding sites (termed S1–S4) (Fig. 3). Only S1 exactly matches with the (A/T)(A/T)CAA(A/T)G core sequence (32), whereas other three sites contain one mismatch at the 5′ or 3′ end of this consensus sequence. The S1 and S2 sites are located near the 3′-terminal region of MDE and are deleted in pGL3-bp-cis16. The S3 and S4 sites near the 5′-terminal region of MDE are deleted in pGL3-bp-cis18. Incidentally, there are four copies of the AACAAAA motif within MDE, and three of them (S2-4) are located in the regions required for the enhancer function.

The identification of MDE has prompted us to search for the mouse MDE counterpart, because the molecular defect of the recessive Mitf mutant, Mitf^mi–rw, is a deletion of the genomic DNA segment that starts about 6 kb upstream from exon 1M and includes exon 1H (20). The putative mouse counterpart of MDE was thus identified in the genome database of the Mus musculus chromosome 6 (accession number AC021060), and is shown below the sequence of MDE (Fig. 3). The putative mouse MDE shares about 67% identity with MDE (positions 35–288). It is noteworthy that the putative mouse MDE is located 12.3 kb upstream from exon 1M of the mouse Mitf gene (accession number AC021060), but is located downstream from exon 1B, because the exon 1B sequence is not present in the same genomic clone. Therefore, the MDE counterpart must be deleted in the Mitf gene of Mitf^mi–rw mice, which could explain white coat color of this mutant (3).

Effect of SOX10 on the M Promoter Activity

We then analysed the effect of SOX10 on the M promoter activity in HeLa cells, which do not express SOX10 endogenously (27). The basal promoter activity of each fusion gene was marginally detected in HeLa cells (see Fig. 1). Co-expression of SOX10 increased the expression of the fusion genes by more than 100-fold (Fig. 4), except for pGL3-bp (−120) that was activated by about 10-fold under the conditions used. The activation of pGL3-bp (−120) by SOX10 was unexpected, because the functional SOX10 sites are present in the promoter region between −260 and −244 (15, 17, 18). The SOX10-mediated activation of pGL3-MI-15k was slightly higher than that of pGL3-MI-14k lacking MDE, but was indistinguishable from that of pGL3-MITF/M. Using the suboptimal concentrations of SOX10, we also obtained the similar results, although the degree of transactivation decreased with lower concentrations of SOX10 (data not shown). In contrast, SOX10 remarkably increased the expression of an internally truncated construct pGL3-bp-cis15 that contains MDE at position −120, suggesting that SOX10 potentially acts on MDE. Thus, SOX10 alone is not sufficient to mediate the functional cooperation of MDE with the downstream region (−13.8 to −120 k) under the conditions used, but may ensure the efficient activation of the M promoter by SOX10.

Characterization of Potential SOX10 Sites in MDE

To explore the functional significance of the putative SOX10-binding sites in MDE, we introduced base changes into these sites (S1–S4) of pGL3-bp-cis15, in such a way that each sequence is entirely deviated from the consensus sequence. The function of MDE was undetectable in melanoma cells when S1 or S2 was disrupted (Fig. 5A). In contrast, the mutation at S3 or S4 showed no noticeable reduction in the enhancer activity, despite that both S3 and S4 share the same sequence with S2. These results suggest the crucial role of S1 and S2 for the enhancer function of MDE in melanoma cells, whereas S3 and S4 are dispensable for the function of MDE.

We next analysed the effect of mutations at the putative SOX10-binding sites on the SOX10-mediated transactivation of pGL3-bp-cis15 in HeLa cells (Fig. 5B). The degree of transactivation was decreased twofold when the S1 or S2 site was mutated. The mutations at S1 and S2 (s1/2 mut) caused a small additive reduction. In contrast, the mutation at S3 exerted no noticeable effects on the SOX10-mediated activation, whereas the mutation at S4 exhibited the marginal reducing effect. The degree of transactivation of the construct (S1/2/4mut), carrying mutations at S1, S2, and S4, was lower than that of the construct (s1/2 mut), but no further reduction was seen with the construct carrying the mutations at S1, S2, and S3. These results suggest that the juxtaposed S1 and S2 sites are involved in the SOX10-mediated activation, which is consistent with the notion that SOX10 preferentially binds to the two adjacent binding sites that are spaced by 4–6 bp (33). In fact, only S1 and S2 are well conserved at the equivalent positions of the putative mouse MDE (Fig. 3). It is also likely that the sequence containing S4 is involved in the transactivation by SOX10. Thus, the enhancer function of MDE depends on certain factors other than SOX10 that may recognize the cis-acting elements located in the 5′ portion of MDE (see Fig. 2B).

We performed a gel shift analysis using a 31-mer oligonucleotide probe containing both S1 and S2. This oligonucleotide probe was incubated with nuclear extracts from COS-7 cells expressing SOX10. SOX10 did bind to the probe, as detected as the specific complex (Fig. 6). The formation of the DNA-protein complex was inhibited by a wild-type competitor or a competitor carrying mutated S2, but was not affected by a competitor containing mutated S1 or mutated S1/2. These results suggest that SOX10 may bind S1.

Expression of Mitf-M and Sox10 mRNA in Melanoblasts

In an effort to establish the in vivo evidence for the role of Sox10 in the regulation of Mitf-M expression, we analysed the expression patterns of Mitf-M and Sox10 mRNA in migrating melanoblasts of mouse embryos by in situ hybridization. At E11.5, Sox10 mRNA is expressed in scattered melanoblasts migrating towards the otic vesicle or the epidermis (Fig. 7A,E). These cells were also stained with a probe for dopachrome tautomerase, a melanoblast marker (34) (data not shown). In situ hybridization analyses with exon 1M-specific probes were unsuccessful, probably because of the short length of exon 1M (13). We therefore used the two types of Mitf probes, an upstream B1b probe that is not present in Mitf-M and a downstream probe common to all Mitf isoforms. It is most likely that Mitf-M mRNA is expressed in melanoblasts that migrate towards the otic vesicle and epidermis (Fig. 7B,C,F), because these melanoblasts were not stained with the B1b probe (Fig. 7C,G). In contrast, Mitf isoform mRNA species containing the B1b region are exclusively expressed in RPE (Fig. 7D). All sense probes did not show any background staining in the relevant regions or cells (30) (data not shown). Thus, Mitf-M mRNA is predominantly expressed in migrating melanoblasts and is co-expressed with Sox10 mRNA in melanoblasts during early embryonic stages.

Separate Expression of Mitf-M and Sox10 mRNA in the Cochlea

In the cochlea of newborn mice (P0), Mitf-M mRNA is expressed in melanocytes of the stria vascularis, whereas Sox10 expression is restricted to supporting cells in the organ of Corti and prospective marginal cells in the stria vascularis (Fig. 8). It is noteworthy that Sox10 expression is undetectable in melanocytes of the stria vascularis. Thus, Mitf-M and Sox10 mRNA are expressed in different cell types of the cochlea, suggesting that Sox10 may regulate transcription from the M promoter in a stage-specific manner during development of the inner ear.