|Home | About | Journals | Submit | Contact Us | Français|
Tyrosinase (TYR), tyrosinase-related protein-1 (TYRP1/gp75) and dopachrome tautomerase (DCT/TYRP2) belong to a family of melanocyte-specific gene products involved in melanin pigmentation. During melanocyte development expression of tyrosinase family genes is thought to be orchestrated in part by the binding of a shared basic helix–loop–helix transcription factor MITF to the M box, a regulatory element conserved among these genes. In transformed melanocytes, expression of tyrosinase and TYRPs is highly variable. Whereas TYR expression in melanoma cells is regulated by both transcriptional and post-translational mechanisms, TYRP1/gp75 transcription is often completely extinguished during melanoma tumor progression. In this study, we investigated the mechanisms of selective repression of TYRP1 transcription. Interestingly, in early stage melanoma cells TYRP1 mRNA could be induced by inhibition of protein synthesis. Transient transfection experiments with a minimal TYRP1 promoter showed that the promoter activity correlates with expression of the endogenous TYRP1 gene. Nucleotide deletion analysis revealed novel regulatory sequences that attenuate the M box-dependent MITF activity, but which are not involved in the repression of TYRP1. Gel mobility shift analysis showed that binding of the transcription factor MITF to the TYRP1 M box is selectively inhibited in TYRP1– cells. These data suggest that protein factors that modulate the activity of MITF in melanoma cells repress TYRP1 and presumably other MITF target genes.
Tyrosinase (TYR) and tyrosinase-related proteins (TYRPs) are a family of closely related melanocyte-specific gene products involved in melanin pigmentation. Tyrosinase is the critical and rate-limiting enzyme that catalyzes the initial steps in melanin synthesis and the tyrosinase-related proteins, tyrosinase-related protein-1 (TYRP1/gp75) and dopachrome tautomerase (DCT/TYRP2) catalyze distal steps that control the type of melanin produced (reviewed in 1,2). In addition to their roles in pigmentation, tyrosinase family proteins also influence the biology of melanocytes and melanoma. There is evidence that TYRP1 is involved in the maintenance of melanosome ultrastructure and affects melanocyte proliferation and melanocyte cell death (3–5). In patients with malignant melanoma, tyrosinase family proteins are frequent targets for an immune response (6–8; reviewed in 9). Accordingly, resistance of melanoma cells to killing by tumor-reactive T cells has been reported to correlate with the loss of expression of tyrosinase and TYRPs (10). Mechanisms that regulate the expression of tyrosinase family genes in melanoma cells are not well understood.
The structural organization of the human TYR and TYRP genes and the 5′ flanking regions required for their tissue-specific expression have been characterized (11–19). Using transient transfection of reporter plasmid constructs, a 3.6 kb tyrosinase 5′ DNA fragment has been shown to direct maximal pigment cell-specific expression of the reporter gene. This 3.6 kb fragment contains multiple E boxes (with a core CANNTG motif) including a tyrosinase distal element (TDE), an 11 bp M box and an initiator E box (Inr-E) (12,14,17). For TYRP1, a 215 bp 5′ DNA fragment containing only the M box is sufficient to direct maximal expression (12). Binding of the microphthalmia-associated transcription factor MITF, a basic helix–loop–helix (bHLH) family transcription factor, to the TDE and/or M box of the TYR and the M box of the TYRP1 genes is critical for their melanocyte-specific expression (20,21).
Since the TYR and TYRP1 genes have a common enhancer element and a shared transcription factor, it is reasonable to hypothesize that these genes are coordinately regulated during melanocyte development and differentiation (12). However, it is now known that TYRP1 can be regulated independently of tyrosinase in melanocytes and melanoma (22–26). In cutaneous neoplastic melanocytes, loss of TYRP1 expression seems to coincide with the appearance of transformed melanocytes in the underlying dermis (27). Accordingly, in most melanoma specimens and cell lines, expression of TYRP1 mRNA and/or TYRP1/gp75 protein is not detectable (27–29). TYRP1 expression is frequently silenced at the level of TYRP1 gene transcription (23,25,29). In contrast, melanoma cells exhibit variable abundance of TYR mRNA and TYR protein. Expression of tyrosinase has been shown to be regulated primarily by post-translational mechanisms (30,31). This divergence in the regulation of the TYR and TYRP1 genes in melanoma provides a useful model for understanding the mechanisms involved in activation and maintenance of patterns of MITF target gene expression.
In this study, we investigated the mechanisms for selective repression of the TYRP1 gene. We show that repression of TYRP1 in melanoma cells is relieved by inhibition of protein synthesis and that binding of MITF to the TYRP1 M box, but not to the TYR M box, is inhibited in TYRP1– melanoma cells. Our data suggest that in melanoma cells the TYRP1 gene and other MITF targets presumably involved in tumor progression are selectively silenced by activation of factors that interfere with the binding of MITF to the M box.
Isolation and primary culture of human melanocytes was described previously (26). Primary (WM35, WM75 and WM98-1) and metastatic (451Lu) human melanoma cell lines were kindly provided by Dr Meenhard Herlyn (Wistar Institute, Philadelphia, PA) (27). The metastatic melanoma cell line SK-MEL-19 was described earlier (23).
The human tyrosinase and TYRP1 promoter–luciferase re porter constructs, including pHTL12, pHTRPL16, pHTRPL4, pHTRPL4M and pHTRPL18, and the promoterless plasmid pL1 were kindly provided by Dr S. Shibahara (Tohoku University School of Medicine, Sendai, Japan). Briefly, the constructs contain the firefly luciferase reporter gene linked to the 5′-flanking region of the human TYR and TYRP1 genes. The fusion gene pHTL12 contains the TYR 5′-flanking region from –3600 to +56 (17). The TYRP1 constructs pHTRPL16, pHTRPL4 and pHTRPL18 contain the human TYRP1 gene segments spanning positions –3600 to +82, –839 to +82 and –133 to +82, respectively. Plasmid pHTRPL4M contains the ACTGTG motif instead of the CATGTG motif in the M box (17).
Deletion mutants of TYRP1 promoter–luciferase reporter constructs, pHTRPL18a (–127/–29), L18b (–58/+82), L18c (–127/+6), L18d (–73/+6), L18e (–62/+6) and L18h (–62/+189) were generated by PCR using appropriate primers that generated 5′ BamHI and 3′ XhoI sites. Purified PCR products were digested with BamHI and XhoI and cloned into the same sites in the luciferase reporter plasmid. The nucleotide substitutions were introduced into the E box motif of pHTRPL18d by PCR. This mutant, L18dM, contains the ACGTTG motif instead of the CAGTTG motif in the E box (–60/–55). The sequences of all constructs were confirmed by automated DNA sequencing.
Melanoma cells in 6 cm dishes were transfected with plasmids using LipofectAmine Plus transfection reagents (Life Technologies, Rockville, MD). Seven micrograms of each promoter–reporter DNA construct and 3 µg pSV-β-galactosidase control vector (Promega, Madison, WI), containing the SV40 early promoter as an internal control, were used for transfection. Transfected cells were lysed in reporter lysis buffer (Promega). Protein concentration was estimated using the BCA protein assay (Pierce Chemical Co., Rockford, IL). Luciferase and β-galactosidase activities were assayed using a luciferase assay system and a β-galactosidase enzyme assay system, respectively (Promega). Luciferase and β-galactosidase activities per unit protein were calculated and the luciferase activity was normalized to the β-galactosidase activity to account for differences in transfection efficiency. The relative luciferase activity was the ratio of normalized luciferase activity to that obtained with promoterless construct pL1. The data shown are means ± SD from at least three independent experiments.
A 1.4 kb fragment encoding the open reading frame of human MITF was amplified by RT–PCR from melanoma mRNA using the primers 5′-ATGCTGGAAATGCTAGAATAT-3′ and 5′-CTAACAAGTGTGCTCCGTCTC-3′ and was cloned into the pCR2.1 vector using a TOPO TA cloning kit (Invitrogen, Carlsbad, CA). The MITF cDNA was excised with EcoRI and subcloned into vector pCMV5a (Sigma Chemical Co., St Louis, MO). The orientation and sequence of the MITF cDNA was verified by automated DNA sequencing. Expression of functional MITF protein was verified by co-transfection of pCMV5a-MITF with a TYRP1 promoter plasmid in HeLa cells.
Gel mobility shift assays were performed as described previously (32). Melanoma cell pellets were homogenized in buffer A (10 mM HEPES, 1.5 mM MgCl2, 10 mM KCl, 1 mM dithiothreitol and proteinase inhibitors) and nuclei were collected by centrifugation and incubated at 4°C for 30 min in 0.5–1 ml of solution B (20 mM HEPES, 1.5 mM MgCl2, 400 mM NaCl, 10% sucrose, 20% glycerol, 0.2 mM EDTA and proteinase inhibitors). Supernatant was collected by centrifugation at 16 000 g for 10 min and dialyzed against solution C (20 mM HEPES, 1.5 mM MgCl2, 100 mM KCl, 20% glycerol, 0.2 EDTA, 1 mM dithiothreitol and proteinase inhibitors) overnight at 4°C. The protein concentration was estimated using a BCA protein assay kit (Pierce Chemical). Synthetic double-stranded oligonucleotides used as probes or competitors are shown in Table Table1.1. Probes were end-labeled with [γ-32P]ATP (ICN Biomedicals, Costa Mesa, CA) using T4 polynucleotide kinase (Promega). Between 10 and 20 µg nuclear extract were incubated with 1.25 ng labeled probe in 20 µl of buffer C containing 1 µg poly(dI–dC) and 10 µg BSA. For competition experiments, a 60-fold molar excess of unlabeled oligonucleotides was added prior to addition of the labeled probe. For supershift assays, 1 µg anti-MITF mAb C5 (Lab Vision Corp., Freemont, CA) or the control anti-gp75 mAb TA99 was added to the nuclear extract 1 h prior to incubation with the labeled probe. DNA–protein complexes were separated on 5% polyacrylamide gels. The gels were fixed in 10% isopropanol, 10% acetic acid, dried and subjected to autoradiography.
Total RNA was isolated from cell pellets using an Ultraspec-II RNA isolation system (Biotecx Laboratories, Houston, TX). Northern analysis was performed as described previously using a Northern Max kit and a Strip-EZ DNA probe synthesis and removal kit (Ambion, Austin, TX) (25,27).
Western blot analysis was performed as described earlier (27). Briefly, cells were solubilized in lysis buffer containing 1% SDS, 10 mM Tris pH 7.4, and proteinase inhibitors (Roche Molecular Biochemicals, Indianapolis, IN). Protein was estimated using the BCA protein assay kit (Pierce Chemical). Total cellular protein was subjected to 9% SDS–PAGE and transferred electrophoretically to a PVDF membrane (Perkin Elmer Life Sciences, Boston, MA). The blots were incubated in blocking buffer [5% non-fat milk in Tris-buffered saline (TBS) containing 10 mM Tris, pH 7.5, 100 mM NaCl] at room temperature for 3 h and then at 4°C overnight with anti-MITF monoclonal antibody C5 and anti-γ-tubulin polyclonal antibody (Sigma Chemical Co), used at 1:500, 1:250, 1:500 and 1:5000, respectively. Blots were washed with TBST (TBS containing 0.1% Tween 20) with frequent changes of wash buffer. They were then incubated with donkey anti-mouse (for MITF) or anti-rabbit (for γ-tubulin) horseradish peroxidase antibody (Amersham Pharmacia Biotech, Piscataway, NJ) diluted in TBST at 1:2000–1:2500 for 1 h. Protein bands were detected by chemiluminescence using an ECL kit (Amersham Pharmacia Biotech).
Northern blot analysis of melanocytes and representative primary (WM35 and WM98-1) and metastatic melanoma (non-pigmented 451Lu and pigmented SK-MEL-19) cell lines showed that whereas TYRP1 mRNA is expressed abundantly in melanocytes and pigmented melanoma cell lines, its expression is undetectable in many primary and metastatic melanoma cell lines that express MITF and/or TYR (Fig. (Fig.1A).1A). These melanoma cell lines show highly variable expression of TYR mRNA. The mRNA and protein levels of the melanocyte transcription factor MITF are also variable (Fig. (Fig.1A1A and B). These data are consistent with our earlier immunohistochemical observation that in neoplastic dermal melanocytes in vivo TYRP1 expression is extinguished (27).
DNA fragments corresponding to TYRP1 exons could be successfully amplified by PCR from genomic DNA of TYRP1-deficient (referred to hereafter as TYRP1–) melanoma cell lines, demonstrating that the loss of TYRP1 gene expression in melanoma cells is not due to deletion of the TYRP1 gene (data not shown). We investigated whether histone deacetylation and/or DNA methylation, mechanisms frequently involved in silencing of genes in tumor cells, are also involved in repression of TYRP1 in melanoma cells. As shown in Figure Figure2A,2A, incubation of TYRP1– melanoma cell lines with trichostatin A (TSA), an inhibitor of histone deacetylase, or with the DNA demethylation agent 5-aza-2′-decoxycytidine (AZA), alone or in combination, did not induce TYRP1 mRNA. The effectiveness of these inhibitors is demonstrated by the induction of p16, a gene known to be down-regulated by these mechanisms (Fig. (Fig.22A).
Earlier we showed that treatment of cultured primary human melanocytes and TYRP1+ melanoma cells with the pharmacological agent hexamethylene bisacetamide (HMBA) results in selective and reversible repression of TYRP1 transcription and that HMBA-induced repression of TYRP1 requires de novo protein synthesis (25). We tested whether down-regulation of TYRP1 in melanoma cells also requires de novo protein synthesis. As shown in Figure Figure2B,2B, treatment of TYRP1 mRNA-negative melanoma cell lines with the protein synthesis inhibitor cycloheximide induced TYRP1 mRNA expression.
The time course of induction and abundance of TYRP1 mRNA in melanoma cell lines by cycloheximide was variable. In the early (radial growth phase) primary melanoma cell line WM35, abundant TYRP1 mRNA was induced as early as 4 h after treatment with the inhibitor, whereas in advanced (vertical growth phase) primary WM98-1 and metastatic melanoma cell line 451Lu, weak induction was detectable only after 12 h treatment. These data show that active protein synthesis is required for repression of TYRP1 transcription in melanoma cells and suggest that this repression may be relieved in cell lines derived from early primary melanoma.
To test whether TYRP1 mRNA expression in melanoma cells reflects the activity of the TYRP1 promoter, we measured TYRP1 promoter activity in TYRP1– melanoma cell lines WM35 and 451Lu and TYRP1+ cell line SK-MEL-19. The cells were transiently transfected with plasmid pHTRPL4, containing a 921 bp (–839/+82) TYRP1 genomic DNA fragment upstream of the luciferase reporter gene, and with plasmid pHTRPL4M, containing a mutant M box (Fig. (Fig.3).3). In TYRP1+ SK-MEL-19 cells, the pHTRPL4 plasmid containing the 921 bp (–839/+82) TYRP1 promoter produced a 6-fold higher luciferase activity than the M box mutant construct pHTRPL4M, confirming that the M box, located between –49 and –39, plays a critical role in TYRP1 promoter activity. The relative luciferase activity of pHTRPL4 in TYRP1– WM35 and 451Lu cells was ~3-fold lower compared to that in SK-MEL-19 cells, and was only slightly higher than the background activity of the M box mutant plasmid pHTRPL4M. These data show that the activity of the TYRP1 promoter in melanoma cells correlates with TYRP1 mRNA expression. Since no TYRP1 mRNA was detectable in TYRP1– melanoma cells, the small amount of TYRP1 promoter activity found in these cells may reflect differences in the sensitivity of detection by northern blotting and luciferase activity. Alternatively, whereas MITF and/or other factors can bind weakly to the transfected TYRP1 promoter DNA and activate transcription of the reporter gene, conformation of the endogenous TYRP1 promoter DNA may not be conducive for such binding.
To identify the cis-acting elements responsible for the repression of TYRP1 promoter activity, we generated deletion constructs of the TYRP1 promoter and compared their activity in TYRP1+ SK-MEL-19 and TYRP1– 451Lu melanoma cells. The luciferase reporter constructs used for this analysis are shown schematically on the left in Figure Figure44.
Analysis of sequence upstream of the M box. First, transfection with plasmid pHTRPL18 showed that the 215 bp TYRP1 5′ DNA fragment (–133/+82) was sufficient to direct reporter gene expression preferentially in TYRP1+ SK-MEL-19 melanoma cells compared to TYRP1– 451Lu cells. The luciferase activity driven by this minimal TYRP1 promoter was nearly equal to that of the 891 bp TYRP1 fragment, in that the reporter activity in SK-MEL-19 cells was ~30-fold higher than promoterless pL1 and 3-fold higher compared to 451Lu cells. Deletion of nucleotides between positions –133 and –58 upstream of the M box (pHTRPL18b) decreased the promoter activity in both SK-MEL-19 (~3-fold) and 451Lu (1.5-fold) cells, producing nearly similar activity in both cell lines (compare pHTRPL18 and pHTRPL18b in Fig. Fig.4A).4A). These results suggest that nucleotides between –133 and –58, upstream of the M box, are necessary for maximal preferential activation of the TYRP1 promoter in TYRP1+ melanoma cells.
Analysis of sequence downstream of the M box. Interestingly, deletion of 3′ sequences in the promoter region between nucleotides +6 and +82 resulted in a 2- to 3-fold increase in the luciferase activity in both SK-MEL-19 and 451Lu cells (Fig. (Fig.4B).4B). This increase in promoter activity could be seen with two different constructs, pHTRPL18a (–127/–29) and pHTRPL18c (–127/+6), indicating the presence of a negative regulatory element(s) downstream of the M box between nucleotide positions –29 and +82. Although deletion of this putative negative regulatory element(s) increased the reporter activity in both TYRP1+ and TYRP1– cells, the activity of this truncated promoter in TYRP1+ SK-MEL-19 cells was still 2- to 3-fold higher compared to its activity in 451Lu cells (pHTRL18a and pHTRPL18c in Fig. Fig.4B),4B), suggesting that the negative regulatory element(s) located downstream of the M box is not involved in repression of the TYRP1 promoter.
Furthermore, whereas deletion of nucleotides between positions –127 and –62 in the 3′-truncated (Δ+6/+82) promoter decreased TYRP1 promoter activity to the basal level in SK-MEL-19 cells, absence of the 3′-negative elements(s) highlighted its preferential activity in these cells (compare the luciferase activity of pHTRPL18b in Fig. Fig.4A4A with pHTRPL18d and pHTRPL18e in Fig. Fig.4B).4B). These data show that the M box and its flanking sequences (between –62 and –29) are sufficient for repression of TYRP1 in human melanoma cells.
In earlier studies we showed that HMBA selectively represses transcription of the TYRP1 gene in melanocytes and pigmented melanoma cells (Fig. (Fig.5A;5A; 23,25,26). Treatment of SK-MEL-19 cells with 5 mM HMBA for 48 h results in extinction of both TYRP1 mRNA and TYRP1 protein without a loss of expression of TYR and MITF (Fig. (Fig.5A5A and B). This selective repression of TYRP1 in vitro seems to mimic the repression of TYRP1 in melanoma in vivo. First, similar to the repression of TYRP1 and concomitant activation of the microtubule-associated protein-2 (MAP2) gene in invasive melanoma cells in the dermis (27), treatment of SK-MEL-19 cells with HMBA also results in repression of TYRP1 and induction of MAP2 mRNA expression (Fig. (Fig.5C).5C). Second, similar to the requirement of de novo protein synthesis for the repression of TYRP1 in TYRP1– melanomas, HMBA-mediated repression of the TYRP1 gene is abrogated when protein synthesis is inhibited by cycloheximide (Fig. (Fig.5D).5D). Third, inhibition of histone deacetylation and/or DNA methylation does not prevent HMBA-induced repression of TYRP1, similar to the inability of these agents to induce TYRP1 mRNA expression in TYRP1– melanoma cells (Figs (Figs5E5E and and2).2). These data suggest that HMBA-mediated repression of TYRP1 transcription is a suitable in vitro model to study the mechanisms of TYRP1 gene repression in melanoma. Therefore, we asked whether the same promoter region that is involved in the repression of TYRP1 in TYRP1– melanoma cells also mediates its down-regulation by HMBA.
We transfected SK-MEL-19 cells with TYRP1 promoter constructs and measured the luciferase activity in untreated and HMBA-treated cells (Fig. (Fig.6).6). As shown in Figure Figure6A,6A, in SK-MEL-19 cells the minimal promoter construct pHTRPL18 produced a nearly 10-fold higher activity compared to the M box mutant construct. Treatment of transfected melanoma cells with HMBA resulted in 5-fold lower luciferase activity (Fig. (Fig.6A).6A). In contrast, treatment with HMBA did not produce a significant change in the luciferase activity driven by TYR promoter construct pHTL12 (Fig. (Fig.6C).6C). This is consistent with the selective repression of endogenous TYRP1 by HMBA. However, HMBA treatment of SK-MEL-19 cells consistently resulted in increased cellular TYR mRNA (Fig. (Fig.5A).5A). The lack of a significant effect of HMBA on TYR promoter activity is consistent with the possibility that post-transcriptional mechanisms, such as mRNA stabilization, contribute to the HMBA-induced increase in cellular TYR mRNA (Fig. (Fig.5A;5A; 25).
Deletion of nucleotides between –133 and –58 (pHTRPL18b), which resulted in a >3-fold lower TYRP1 promoter activity (compared to that of the minimal promoter pHTRPL18) in untreated controls, also abrogated the inhibitory effect of HMBA (Fig. (Fig.6A,6A, compare the open bar for pHTRPL18b with the solid bar for pHTRPL18 and the solid bars for pHTRPL18 and pHTRPL18b). Thus, similar to its role in the repression of TYRP1 promoter activity in TYRP1– cells, the region between –133 and –58 is necessary for HMBA-mediated repression of TYRP1 promoter activity.
Deletion of a 3′ negative regulatory element(s) between nucleotides +82 and +6 in the TYRP1 promoter (constructs pHTRPL18a and pHTRPL18c) resulted in a 3-fold higher activity compared to the minimal promoter pHTRPL18 in control SK-MEL-19 transfectants (Fig. (Fig.6B).6B). Deletion of sequences upstream of the M box, between nucleotides –127 and –73, in the 3′-truncated promoter plasmid pHTRPL18c, on the other hand, decreased the TYRP1 promoter activity to the basal level. Treatment with HMBA showed that that these truncated TYRP1 promoters remained sensitive to inhibition by HMBA.
These data show that: (i) nucleotide sequences downstream of the TYRP1 M box (–29/+6) are not involved in repression of TYRP1 transcription by HMBA, similar to the lack of their involvement in repression of TYRP1 in WM35 and 451Lu cells (Fig. (Fig.4B);4B); (ii) the 79 bp region between nucleotide positions –73 and +6 is necessary and sufficient for HMBA-mediated repression of M box driven TYRP1 promoter activity.
Role of sequences flanking the M box. We searched the transcription factor databases for potential transcription factor-binding sites within this 79 bp sequence and found an additional E box motif CAGTTG, located between nucleotides –60 and –55 upstream of the M box. This motif is known to be recognized by bHLH leucine zipper transcription factor family proteins. To test whether this E box plays a role in regulation of TYRP1 promoter function, we generated a mutant 79 bp promoter with a CA→AC inversion in the CAGTTG motif. We transfected SK-MEL-19 cells with the wild-type pHTRPL18d and the mutant pHTRPL18dM plasmids and measured the luciferase activity in control and HMBA-treated transfectants. As shown in Figure Figure7A,7A, the 79 bp TYRP1 promoter with a mutation in the E box immediately upstream of the M box produced luciferase activity comparable to the wild-type promoter and also remained sensitive to inhibition by HMBA. These data allowed us to further demarcate the region between nucleotide positions –62 and +6, containing primarily the M box, as the TYRP1 promoter region responsible for its selective repression in melanoma cells.
Since the MITF-binding M box is the only recognizable regulatory element within this 79 bp promoter fragment, we tested whether overexpression of MITF, the factor that binds to the M box, prevents HMBA-induced repression of TYRP1 promoter activity. Co-transfection of SK-MEL-19 cells with the TYRP1 promoter construct pHTRPL4 and the MITF expression plasmid pCMV5a-MITF showed that while transfection of MITF induced a >2-fold increase in TYRP1 promoter activity in SK-MEL-19 cells, overexpression of MITF was unable to relieve repression caused by HMBA (light hatched bar and dark stippled bar in Fig. Fig.7B).7B). These data suggest that selective repression of TYRP1 transcription in invasive melanoma in vivo and by HMBA in vitro is mediated by mechanisms that inhibit activation of the TYRP1 promoter by MITF.
To investigate whether repression of TYRP1 is caused by the inability of MITF to bind to the TYRP1 M box, we performed mobility shift assays. End-labeled synthetic oligonucleotides corresponding to the TYR distal element (TYR TDE), TYR M box (TYR M) and TYRP1 M box (TYRP M) were incubated with nuclear extracts from control SK-MEL-19 cells (Fig. (Fig.8).8). All three oligonucleotides produced a specific protein–DNA complex (Fig. (Fig.8A,8A, lanes 1, 4 and 7). The formation of this radioactive DNA–protein complex was inhibited by excess corresponding cold wild-type oligonucleotides (Fig. (Fig.8A,8A, lanes 2, 5 and 8) but not by mutant oligonucleotides (MTYR TDE, MTYR M box and M1TYRP M box) (Fig. (Fig.8A,8A, lanes 3, 6 and 9). Sequences of the synthetic oligonucleotides used in these experiments and their ability to compete with TYRP1 M box oligonucleotide are shown in Table Table1.1. Gel mobility shift assays with nuclear extracts of TYR mRNA-positive and TYRP1 mRNA-negative melanomas showed that the intensity of the M box DNA–protein complex band correlates with the abundance of TYR and expression of TYRP1 mRNA in melanoma cell lines (Fig. (Fig.88B).
We used a monoclonal anti-MITF antibody to investigate whether the M box DNA–protein complexes contain MITF. When anti-MITF mAb was included during incubation of the labeled oligonucleotides with control SK-MEL-19 nuclear extract, a slower migrating (supershifted) band could readily be detected with the TYR TDE (Fig. (Fig.8C,8C, lane 2) and M box (data not shown). A weaker supershifted band could be seen with the TYRP1 M box (Fig. (Fig.8C,8C, lane 5). This slower migrating band was not detectable with a control isotype-matched IgG, demonstrating that the supershifted complexes contain a protein(s) recognized by the anti-MITF mAb (Fig. (Fig.8B,8B, lanes 1 and 4).
In nuclear extracts of SK-MEL-19 cells treated with HMBA, the intensity of the supershifted TYR TDE (Fig. (Fig.8C8C and D, lane 3) and M box (data not shown) bands was comparable to that of control cells (Fig. (Fig.8C8C and D, lane 2). This is consistent with the inability of HMBA to inhibit endogenous TYR transcription (Fig. (Fig.5)5) and transfected TYR promoter activity (Fig. (Fig.6).6). Although no TYRP1 mRNA or protein is detectable in HMBA-treated cells, a TYRP1 M box DNA–protein complex is readily detectable with their nuclear extracts. However, with the nuclear extracts of HMBA-treated cells, the anti-MITF antibody supershifted TYRP M box band was barely detectable (Fig. (Fig.8C8C and D, lanes 5 and 6). These data show that although an M box DNA–protein complex that co-migrates with the M box–MITF complex is formed in TYRP1 repressed cells, this complex contains little or no MITF detectable by the MITF-specific antibody.
In summary, our data show that the M box containing the 79 bp TYRP1 5′ fragment is sufficient to produce tissue-specific promoter activity and selective repression of TYRP1 in melanoma cells. Based on the requirement of new protein synthesis for TYRP1 repression and the TYRP1 M box gel shift analysis, we propose that in melanoma cells the TYRP1 gene is silenced by activation of a repressor(s) that selectively inhibits binding of MITF to the TYRP1 M box.
Binding of the melanocyte-specific transcription factor MITF to the common regulatory element M box in the tyrosinase family gene promoters is believed to regulate their expression during melanocyte development and differentiation (21). However, the mechanisms that orchestrate the regulation of tyrosinase family genes by MITF in transformed melanocytes are not clear. For example, in human melanoma cells, whereas TYR mRNA shows variable abundance, TYRP1 expression is often completely absent (23,25–29). This suggests that whereas TYR is regulated by a quantitative rheostat-like switch, which allows the rate of transcription to be continuously variable, the TYRP1 gene is regulated by a binary switch, which allows the gene to exist in only ‘on’ or ‘off’ (33). Thus, unlike tyrosinase expression, which is regulated by multiple mechanisms, including transcription, mRNA stability and post-translational mechanisms, TYRP1 gene expression seems to be regulated primarily by a binary on/off transcriptional switch.
Promoter methylation and/or histone deacetylation, which are mechanisms known to be involved in silencing of p16INK4a, a gene known to play a role in melanoma tumorigenesis, do not seem to contribute to repression of TYRP1, although the TYRP1 gene is located close to the p16INK4a gene on chromosome 9 (34–37). Recently, it was reported that the transcriptional repressor Id1 plays a role in silencing of the p16 gene in early melanomas, while additional genetic changes during tumor progression result in irreversible loss of its expression (38). It is possible that a similar mechanism may be responsible for silencing TYRP1 (29).
Although MITF can transactivate both the TYR and TYRP1 genes by binding to the M box sequences, the magnitude of activation of the TYRP1 promoter is >20-fold lower than that of the TYR promoter. Therefore, it was suggested that additional regulatory elements similar to the TDE or the initiator E box of TYR may be located in the TYRP1 genomic DNA outside the 3.6 kb fragment (pHTRPL16) (17). Nonetheless, in our experiments the 215 bp minimal promoter exhibited differential activity in TYRP1+ and TYRP1– melanoma cells, suggesting that the 215 bp promoter contains the necessary regulatory elements for differential regulation of TYRP1 in melanoma cells.
A detailed analysis of the 215 bp promoter revealed several interesting features. First, a canonical TATA box is not present in this minimal TYRP1 DNA fragment (GenBank accession no. L38952). It is known that transcription of TATA-less genes is initiated by a downstream core promoter element (DPE) in conjunction with the Inr initiator element. A DPE is found in many Drosophila and mammalian genes (39). The DPE element consisting of the consensus sequence G-A/T-C-G is typically located ~30 bp downstream of the transcription start site. Analysis of the TYRP1 sequence revealed a DPE-like sequence C-A-G-C at position +33. Surprisingly, deletion of sequences between –29 and +82 that include this motif produced a 3-fold higher promoter activity compared to the 3.6 kb TYRP1, suggesting the presence of a primarily negative regulatory element(s) within this region. It has recently been suggested that DPEs can act as bifunctional basal transcription regulatory elements (40). Therefore, it is conceivable that this sequence plays a role in the regulation of TYRP1 basal transcription but not in silencing of this gene in melanoma cells. It is important to note, however, that using a S1 nuclease protection assay Sturm et al. (19) placed the transcription start site 63 bp downstream from that of Yokoyama et al. (15) and identified a TATAAA box motif approximately 20 bases upstream of the start site. However, there is no evidence for a function of this element in TYRP1 transcription (S.Shibahara, personal communication). In the TYRP1 gene promoter region downstream of the M box, including the first intron, another enhancer element was identified. The enhancer function of this sequence is not specific to melanocytes (12). Based on the differential activity of TYRP1 promoter constructs pHTRPL18d and pHTRPL18e, which lack the above-mentioned elements, in TYRP1+ and TYRP1– cells, we propose that the nucleotide sequence between –62 and +6 is sufficient for melanocyte-specific expression of TYRP1 and its selective regulation in melanomas.
Deletion analysis of the 215 bp minimal promoter revealed additional positive regulatory elements located between nucleotides –127 and –58 upstream of the M box. Our data show that although these regulatory elements influence the activity of the M box, they do not seem to be involved in selective repression of TYRP1 in melanoma cells. In the mouse tyrp1 gene two cis-acting elements, MSEu and MSEi (melanocyte-specific elements), containing the sequence GTGTGA, have been reported to act as negative regulatory elements, and binding of brachyury-related T box family transcription factor Tbx2 to these elements appears to correlate with repression of the mouse tyrp1 gene (41,42). However, a motif homologous to MSE is not present within the 839 bp sequence upstream of the human TYRP1 gene (17). Another transcription factor reported to repress the mouse tyrosinase family genes is the bHLH family protein Mash1 (43). In transient co-expression studies, we found that both Tbx2 and Mash1 are able to repress human TYRP1 promoter activity. However, these factors also down-regulated the TYR promoter and their effect was not selective to the TYRP1 promoter (D.Fang and V.Setaluri, unpublished observation).
We searched the 215 bp minimal promoter sequence for potential transcription factor-binding sites using a transcription factor database (44) and found several sites upstream of the M box, including TGACTTG (positions –119 to –113) for AP1, TATTC (positions –112 to –108) for OCT1, CAAAT (positions –107 to –103) for BRN2, CTGGC (positions –84 to –89) for NF1, GAAAT (positions –57 to –61) for CEBPB and a second E box motif CAGTTG (positions –56 to –51) for bHLH zip family transcription factors. TYRP1 promoter constructs pHTRPL18d and pHTRPL18e, which lack all these elements except the E box motif CAGTTG, show differential activity in TYRP1+ and TYRP1– cells, and mutations in the E box sequence did not affect the magnitude of TYRP1 promoter activation or its repression by HMBA in melanoma cells. This suggests that the E box located upstream of the M box is not necessary for repression of TYRP1 in melanomas.
Although it has been established that MITF transactivates the TYRP1 gene, the strength of MITF binding to the TYRP1 M box and the magnitude of activation of the TYRP1 promoter by MITF is lower compared to that of the TYR promoter. It is worth noting here that expression of TYRP1 mRNA does not correlate with the abundance of cellular MITF. Furthermore, overexpression of MITF is unable to relieve the HMBA-mediated repression of TYRP1 transcription. Interestingly, co-transfection of MITF-negative HeLa cells with the MITF expression plasmid was sufficient to activate the TYRP1 promoter and produce significant luciferase reporter activity (data not shown). These data suggest that a repressor, present in abundant amounts in TYRP1– and HMBA-treated melanoma cells, but not in non-melanocytic cells, is able to repress transactivation of the TYRP1 promoter by MITF.
The observation that TYRP1 is repressed in both benign and malignant melanocytes located in the dermis suggests that stromal interactions in the dermis provide signals for down-regulation of TYRP1 transcription. We investigated this by measuring TYR and TYRP1 promoter activities in transfected SK-MEL-19 melanoma cells cultured with dermal fibroblast-conditioned medium and co-cultured with primary dermal fibroblasts. In contrast to selective down-regulation of TYRP1 in melanoma cells, co-culture with fibroblasts as well as addition of dermal fibroblast-conditioned medium resulted in down-regulation of both the TYR and TYRP1 promoter activities (data not shown). Selective repression of TYRP1 in neoplastic melanocytes, therefore, seems to be caused by mechanisms intrinsic to melanocytes. Inhibitors of protein kinase C (calphostin C and bis-indolylmalimide), mitogens and stress-activated protein kinases (SB 203580 and PD98059) did not relieve HMBA-mediated repression of TYRP1 transcription, suggesting that the intracellular signaling pathways mediated by these kinases are not involved in TYRP1 repression (D.Fang and V.Setaluri, unpublished observation). In this context, it is interesting to note that regulation of a subset of late muscle structural genes requires cooperation of p38 signal transduction and promoter-specific binding of MyoD (45).
Based on our data, we propose a model for regulation of TYRP1 by MITF. In normal human melanocytes, TYRP1 expression is activated by binding of the phosphorylated homodimeric transcriptional factor MITF to the M box sequence (21). Regulatory elements located upstream of the M box enhance transactivation of the TYRP1 gene (Fig. (Fig.4,4, plasmids pHTRPL18d and pHTRPL18e). The 3′ regulatory elements, including the DPE-like element, on the other hand, attenuate TYRP1 promoter activity. The requirement of protein synthesis for TYRP1 repression and the mobility shift patterns of the TYRP1 M box in melanoma cells suggest that a repressor(s) that selectively inhibits the binding of MITF to the TYRP1 M box is activated (Fig. (Fig.8).8). Thus, whereas the constitutive transcription of TYRP1 in melanocytes is controlled by positive and negative regulatory elements flanking the M box, selective repression of this gene in melanomas appears to be caused by factors that inhibit binding of MITF to the M box. It is reasonable to assume that transcription of additional MITF target genes, presumably involved in melanoma progression, is also repressed by such factors.
We thank Drs Andrew Thorburn, G. L. Prasad and Mark Lively for critical reading of the manuscript and useful suggestions, and Namrata Sangha for expert technical assistance. This work was supported in part by grant AR44617 (to V.S.) from the National Institutes of Health and by a Dermatology Foundation Dermik Laboratory Research Fellowship (to D.F.).