Transcriptional Repression of the Microphthalmia Gene in Melanoma Cells Correlates with the Unresponsiveness of Target Genes to Ectopic Microphthalmia-Associated Transcription Factor

Transcriptional Repression of the Microphthalmia Gene in Melanoma Cells Correlates with the Unresponsiveness of Target Genes to Ectopic Microphthalmia-Associated Transcription Factor J. Vachtenheim, H. Novotna, and G. Ghanem*

Laboratory of Molecular Biology, University Hospital, IIIrd Medical Faculty, Charles University, Prague ± Bulovka, Czech Republic; *Laboratory of Oncology and Experimental Surgery, Universite Libre de Bruxelles, Institut Jules Bordet, Brussels, Belgium

MITF fusion protein, remained fully functional in these cells, however, and ectopic microphthalmia transcription factor localized normally to the nucleus and bound to the tyrosinase initiator E-box in gel retardation assays. Thus, the block of differentiation in microphthalmia-transcription-factor-negative melanomas extended the transcriptional repression of the microphthalmia transcription factor gene alone, and endogenous promoters in these melanoma cells became no longer responsive to microphthalmia transcription factor when this was substituted exogenously. The data presented suggest that a speci®c nuclear context is required for the transcriptional activation of the melanocyte markers by the microphthalmia transcription factor in malignant melanocytes and this speci®city is lost concomitantly with the transcriptional repression of microphthalmia transcription factor. Key words: melanocyte/repression/ TRP-1/tyrosinase. J Invest Dermatol 117:1505±1511, 2001

In the melanocyte, expression of genes required for pigment formation is mediated by the microphthalmia transcription factor, which is also critical for the development and survival of normal melanocytes during embryogenesis. Here we show that the expression of the melanocyte-speci®c isoform of microphthalmia transcription factor is lost in a subset of human melanoma cell lines, accompanied by the repression of tyrosinase and tyrosinase-related proteins 1 and 2, the three transcriptional target genes for microphthalmia. After the forced expression of microphthalmia transcription factor in melanoma cells where the expression of endogenous microphthalmia gene was found to be extinguished, no restoration of the melanogenic phenotype occurred and the transcription of the three microphthalmia transcription factor target genes remained silent. The transcription activation domain of microphthalmia transcription factor, tested as a GAL±

L

other related transcriptional regulators, TFE3, TFEB, and TFEC. Mitf can form heterodimers with these factors that bind, as do Mitf homodimers, to the E-box consensus sequence (Hemesath et al, 1994). Tyrosinase, TRP-1, and TRP-2 are required for normal melanogenesis in melanocytes. It has been shown that Mitf activates promoters of all these three genes (Bentley et al, 1994; Hemesath et al, 1994; Yasumoto et al, 1997; Bertolotto et al, 1998a; reviewed by Goding, 2000) and also of the QNR-71 gene isolated from pigmented cells of the retina (Turque et al, 1996). The conserved sequence GTCATGTGCT (M-box) present in tyrosinase, TRP-1, and TRP-2 promoters is an essential element for binding of Mitf (Bentley et al, 1994; Hemesath et al, 1994). In mice, mutations at the Mi locus result in defects or loss of mast cells, osteoblasts, and retinal, cochlear, and skin melanocytes (reviewed by Moore, 1995). Ectopically delivered MITF has been shown to induce melanocyte-like features when introduced into cells that normally do not express microphthalmia protein; a dendritic morphology and induction of transcription of tyrosinase and TRP-1 was demonstrated in MITF-expressing NIH3T3 cells (Tachibana et al, 1996). No such effects have been observed in 10T1/2 ®broblasts, however, suggesting that only some cell types are susceptible to the melanogenic activity of MITF (Tachibana et al, 1996).

oss or defects of cellular differentiation are common characteristics frequently found in cancer cells. The crucial gene regulators important for differentiation in many lineages are the cell-fate-determining genes of the basic helix±loop±helix (bHLH) family of transcriptional activators. In the melanocytes, a unique biochemical pathway results in the formation of a visible pigment melanin, and the enzymes involved in this pathway are considered as markers for both normal and malignant melanocytes. The transcription of these marker genes [tyrosinase, tyrosinase-related proteins 1 and 2 (TRP-1 and TRP-2)] is believed to be regulated by the microphthalmia transcription factor (reviewed by Goding, 2000), a member of the basic helix±loop±helix leucine zipper (bHLH-LZ) family. Human (MITF) and mouse (Mitf) microphthalmia protein sequences are almost identical and show similarities with three Manuscript received April 20, 2001; revised July 5, 2001; accepted for publication August 10, 2001. Reprint requests to: Dr. Jiri Vachtenheim, Laboratory Mol. Biology, University Hospital, IIIrd Medical Faculty, Charles University, 18000 Prague 8 ± Bulovka, Czech Republic. Email: [email protected] Abbreviations: MITF, microphthalmia-associated transcription factor; TRP-1, TRP-2, tyrosinase-related proteins 1 and 2; bHLH-LZ, basic helix±loop±helix leucine zipper. 0022-202X/01/$15.00

´ Copyright # 2001 by The Society for Investigative Dermatology, Inc. 1505

1506

VACHTENHEIM ET AL

Human melanoma cells either retain the capacity to undergo the normal differentiation program (i.e., to synthesize the pigment melanin) or are visibly amelanotic, which does not necessarily imply that they have lost the expression of enzymes responsible for the completion of the melanogenic pathway because additional, post-transcriptional mechanisms regulating the melanin synthesis exist in melanoma cells (Kameyama et al, 1993, 1995; Eberle et al, 1998). Transformation or immortalization of melanocytes in culture by several oncogenes such as E1a, ras, neu, or basic ®broblast growth factor also causes depigmentation (Dotto et al, 1989; Melber et al, 1989; Wilson et al, 1989; Halaban et al, 1996) and downregulation of Mitf expression (Yavuzer et al, 1995; Halaban et al, 1996). Thus, in mouse models of melanoma, downregulation of differentiation is an important feature accompanying the onset or progression of malignancy. MITF exists in several isoforms that differ in the ®rst exon and are transcribed from separate promoters (Amae et al, 1998). Promoter for the melanocyte-speci®c isoform, MITF-M, is regulated by several transcriptional factors that have been recognized to bind to their motifs in this promoter and activate it in reporter assays. Wnt signaling activates the LEF-1/TCF transcription factor, which then binds to and stimulates the MITF promoter (Takeda et al, 2000). Several recent reports have shown that the transcriptional activators Pax3 and Sox10 are also able to stimulate the MITF promoter (Watanabe et al, 1998; Bondurand et al, 2000; Lee et al, 2000; Potterf et al, 2000; Verastegui et al, 2000a). Sox10 mutant lacking the transcription activation domain has been shown to inhibit the activation of the MITF promoter by the wild-type Sox10 and downregulate the levels of endogenous MITF protein, thus implicating the Sox10 protein in the regulation of MITF gene transcription in melanoma cells (Lee et al, 2000). Moreover, the level of Mitf protein and the activity of the microphthalmia promoter is upregulated after the treatment of B16 mouse melanoma cells by cAMP elevating agents, and the DNA motif to which CREB binds is required for this stimulation (Bertolotto et al, 1996; 1998b). Due to this, transcription of tyrosinase family members is also increased, and this pathway thus can mediate acute transient elevation of melanin synthesis upon hormonal stimuli. In this study, we have found that, in human melanoma cells, the absence of expression of the three MITF target genes (tyrosinase, TRP-1, and TRP-2) correlates with the extinction of transcription of the melanocyte-speci®c isoform of the MITF gene (MITF-M). We further show that the reintroduction of MITF into melanoma cells that have lost expression of the MITF gene does not lead to the restoration of the differentiated phenotype. The inability of MITF to induce melanogenesis in the MITF-negative cells is not due to the functional inhibition of the transcription activation domain or the impaired nuclear localization. These results show that when MITF is transcriptionally repressed in a subset of human melanomas, promoters of MITF-regulated marker genes, in the chromatin context, become nonresponsive to otherwise functional exogenous MITF. MATERIALS AND METHODS Cell lines Human melanoma cell lines used in this study were purchased from the American Type Culture Collection (Manassas, VA) except for cell lines IgR3, Beu, Dor, Dau, LND1, Hbl, and Scl, which have been described previously (DelMarmol et al, 1993). All melanoma cell lines were maintained in Eagle's minimal essential medium, supplemented with 10% fetal bovine serum (FBS) (Gibco), 50 units per ml penicillin, 50 cËg per ml streptomycin, 2 mM L-glutamine, and 1 mM sodium pyruvate. Other cell lines used here were cultivated in Dulbecco's modi®ed Eagle's medium containing 10% FBS, antibiotics, and L-glutamine. Plasmids and constructs Human MITF cDNA was recovered from normal human melanocytes by using reverse transcription polymerase chain reaction (RT-PCR) with primers 5¢-CGC GAA GCT TGT TAT GCT GGA AAT GCT A and 5¢-ACG CGA ATT CGC ACA ACA AGT GTG CTC CGT CTC (stop codon abolished) containing HindIII

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

and EcoRI sites, respectively, for subsequent cloning. Double-stranded oligonucleotide with the sequence encoding the hemaglutinin (HA) epitope followed by a stop codon (5¢-AATTGTTACCCCTACGACGTGCCCGACTACGCCTCCCTCGCTTAGAG) was then cloned into the EcoRI site resulting in the C-terminal-tagged MITF (MITF-HA) and the insert was sequenced on both strands. The sequence corresponded exactly to the reported human MITF cDNA (Tachibana et al, 1994). For use in transient transfections, the MITF-HA sequence was cloned into the pcDNA3 expression vector (Invitrogen, Carlsbad, CA) to obtain pCMV-MITF-HA. The HA epitope was added in order to discriminate ectopic MITF from possible low levels of the two ubiquitous isoforms of MITF, MITF-A and MITF-H (Amae et al, 1998; Vachtenheim and Novotna, 1999), which could react with the antiMITF antibody in indirect immuno¯uorescence. Also, the internal BamHI±BamHI fragment that comprises the entire bHLH-LZ region of MITF was cut off and the shortened plasmid was religated (in frame, after ®lling) to obtain pCMV-MITF-(DbHLH-LZ)-HA, which served as a negative control in transfections. The human tyrosinase promoter sequence that contains bases ±300 to +80 relative to the transcription start (Bentley et al, 1994) was cloned into the pGL3 basic reporter vector (Promega, Madison, WI) to yield pTyr-prom-Luc and used in transient transfections. The plasmid pCMV-GAL4-BD-MITF(105±199) contains sequences encoding amino acids 105±199 of MITF fused in frame to the GAL4 DNA-binding domain. The reporter plasmid pG5-E1b-TATALuc was cloned by placing the ®ve GAL-binding sites with the minimal E1b promoter region from pG5CAT (Clontech, Palo Alto, CA) upstream of the luciferase gene in the pGL3 enhancer (Promega). RT-PCR and southern blot The expression of the endogenous genes was determined by the RT-PCR method. Total RNA from melanoma cell lines or transfected cells was isolated by the guanidium thiocyanate procedure and 1 cËg of RNA was reverse transcribed with Superscript II (Life Technologies, Gaithersburg, MD) according to the instructions of the supplier. The primer pairs employed in PCR and the length of the corresponding PCR product were as follows: melanocytespeci®c isoform of MITF (MITF-M), 5¢-AGG GAG GGA TAG TCT ACC GTC TCT C and 5¢-GCA GGG AGG ATT CGC TAA CAA GTG T (1328 bp product); tyrosinase, 5¢-TTG GCA GAT TGT CTG TAG CC and 5¢-AGG CAT TGT GCA TGC TGC TT (283 bp product) (Eberle et al, 1995); TRP-1, 5¢-TCT TTC TCC CTT CCT TAC TGG and 5¢-CAT TTC TGT GTT GGT GAC TGG (630 bp product); TRP-2, 5¢-CCC GAA TTC GAG CTC TCC CAA TTA TAA AGC CAT and 5¢-CCC GAA TTC AGG TAA GGC ATG AGC ACC CTA G (1618 bp product). Primers for glyceraldehyde-3phosphate dehydrogenase (GAPDH) were from Clontech (Palo Alto, CA). No ampli®cation was seen when the reverse transcription was omitted. Identity of the PCR products was con®rmed by sequencing. Southern blotting of DNA samples was performed on Hybond N+ membranes (Amersham Pharmacia Biotech, U.K.) according to the manufacturer's instructions and probed with the MITF 32P-labeled cDNA probe encompassing the whole protein coding region. Retroviral transfer of MITF To prepare amphotrophic retroviral stocks, MITF-HA insert was cloned in the retroviral vector pBABEpuro (Morgenstern and Land, 1990) and transfected into y-2 cells, and the transient virus was used to infect amphotropic PA317 packaging cell line. 0.45-mm-®ltered medium from transiently infected PA317 cells was used to infect human melanoma cells IgR3 and RPMI7951 in the presence of 4 mg per ml Polybrene (Sigma) for 2 h as described previously (Yavuzer et al, 1995). After 2 d, infected cells were selected in 0.3 mg per ml puromycin and resistant clones or pools were expanded 1 wk later and analyzed. Cell transfections and luciferase assays For transient transfections, cells were seeded in six-well plates a day prior to transfection. U2OS, C33A, and H1299 cells were transfected by the calcium phosphate method (Chen and Okayama, 1991) and 0.25 mg of the plasmid pCMVb was included in the plasmid mix to normalize the variability in transfection ef®ciency. The reporter plasmid pG5-E1b-TATA-Luc (0.5 mg) and various amounts of pCMV-GAL4-BD-MITF(105±199) were cotransfected and the cells were harvested 48 h later and assayed for luciferase and b-galactosidase activities. IgR3, RPMI7951, and SK-MEL5 cells were transiently transfected with lipofectamine (Life Technologies) in 400 ml of Opti-Mem medium in 12-well plates overnight, and harvested 48 h later. In lipofections, 20 ng of the expression plasmid encoding Renilla luciferase was used to normalize the luciferase values by determining the dual luciferase activity (Promega).

VOL. 117, NO. 6 DECEMBER 2001

Figure 1. Expression of MITF and the three target genes analyzed by RT-PCR in human melanoma cell lines. The analysis was carried out with two independent isolations of RNA and the results of one experiment are shown. No bands were seen when the reverse transcriptase step was omitted. PCR proceeded for 30±35 cycles, except for GAPDH, which was stopped after 25 cycles. For the lengths of the PCR products see Materials and Methods. HeLa S3 cells did not express any melanocyte marker and normal human melanocytes were positive for all mRNAs (not shown).

Western blotting, immuno¯uorescence, and gel mobility shift assays The expression of MITF-HA protein was determined by Western blotting of nuclear extracts and probing with the D5 monoclonal antibody to MITF (Hemesath et al, 1998) (obtained from LabVision, Fremont, CA). Proteins were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech). For immuno¯uorescence, the cells were grown on coverslips, ®xed with 3% paraformaldehyde in phosphate-buffered saline (PBS), permeabilized for 10 min in 0.2% Triton X-100 in PBS, blocked in 5% sheep serum, 2% bovine serum albumin, and 0.05% Nonidet P-40 in PBS, and stained with anti-HA antibody 12CA5 (Roche, Mannheim, Germany) diluted to a ®nal concentration of 1 mg per ml in the blocking solution. The second antibody was a 1:50 diluted ¯uorescein isothiocyanate (FITC) labeled antimouse IgG (Roche). Gel mobility shift assays were carried out as described previously (Bertolotto et al, 1998a) except that poly(dIdC) was omitted from the binding mix and the polyacrylamide gel contained 10% glycerol. Primers derived from the human tyrosinase promoter sequence (GTCTCAGCCAAGACATGTGATAATCA and ACAGTGATTATC ACATGTCTTGGCTG) were annealed and this double-stranded probe containing the tyrosinase initiator E-box was labeled with a-32P-dCTP and Klenow ®lling and used in gel mobility shifts.

RESULTS Three MITF target genes, tyrosinase, TRP-1, and TRP-2, are expressed only in MITF-M-positive melanoma cells We have previously reported that the MITF-M isoform constitutes the vast majority of the MITF-related messenger RNA in human melanoma cell lines and have observed the loss of this melanocytespeci®c MITF-M isoform in a few cell lines (Vachtenheim and Novotna, 1999). If MITF is necessary for the expression of the target genes in melanocytes, we asked whether the expression of mRNA for MITF-M and MITF-regulated melanogenic markers correlates in human melanoma cell lines. The analysis by RT-PCR revealed that indeed the loss of MITF transcript in the three cell lines RPMI7951, IgR3, and Dau correlated with the absence of tyrosinase, TRP-1, and TRP-2 mRNAs (Fig 1). In RPMI7951 melanoma cells, however, a very low level of tyrosinase message was still detectable even in the absence of MITF, and A375 cells expressed only a low level of MITF but no marker gene (Fig 1). Although the pattern of expression of tyrosinase and TRP-2 mRNA almost faithfully copied that of MITF, the TRP-1 transcript was selectively repressed in several melanomas (SK-MEL-2, Dor, LND1, Beu, SK-MEL-1, and SK-MEL-31). Northern blot analyses con®rmed the RT-PCR results (not shown). To verify that the loss of MITF transcription was due to transcriptional repression and not a consequence of gene rearrangement, southern blot was performed on MITF-message-negative and selected MITF-

MITF REPRESSION IN MELANOMAS

1507

Figure 2. Analysis of the genomic microphthalmia locus in MITF-message-negative melanoma cell lines. 10 mg of genomic DNA was cleaved with either EcoRI (left) or BamHI (right) and subjected to electrophoresis and southern blotting. MITF cDNA was used as a probe.

message-positive melanomas and the pattern of fragments was found to be essentially the same (Fig 2). Thus, no gross gene abnormality or a chromosomal translocation involving the MITF gene underlies the loss of MITF mRNA in the three melanoma cell lines. Ectopic MITF induces transcription of endogenous tyrosinase gene in U2OS cells Transfected MITF has been demonstrated to activate promoter±reporter constructs of the three melanogenic genes tyrosinase, TRP-1, and TRP-2 (Bentley et al, 1994; Yasumoto et al, 1997; Bertolotto et al, 1998a) and to induce transcription of the endogenous tyrosinase and TRP-1 in NIH3T3 cells (Tachibana et al, 1996). To further explore the role of MITF in melanocytic differentiation, we wished to determine the effect of ectopically expressed MITF in melanoma cells in which the endogenous protein was found to be extinguished. It was essential to con®rm ®rst that the HA-tagged MITF construct was functional. When transfected into COS7 cells, the pCMV-MITF-HA vector activated the tyrosinase promoter±reporter about 5-fold whereas the deleted form of MITF missing the DNA-binding and dimerization domains [pCMV-MITF-(DbHLH-LZ)-HA] was inactive (Fig 3A). Similar levels of activation were found in several other cell types unrelated to melanocytes (not shown). The in vitro translated product bound to the M-box and initiator E-box derived from the tyrosinase promoter, as determined by gel shifts (not shown). Importantly, as shown in Fig 3(B) (lane 2), MITF provoked the expression of the endogenous tyrosinase gene in U2OS cells (human osteosarcoma), whereas the transfection of an empty vector or the pCMV-MITF-(DbHLH-LZ)-HA plasmid did not elicit any response (Fig 3B, lanes 1 and 3). This induction of the tyrosinase gene in U2OS cells, which are normally negative for melanogenic markers, was consistently observed in all transfection experiments (repeated many times); however, the two other endogenous genes, TRP-1 and TRP-2, remained silent in these cells (not shown). Nevertheless, these results demonstrate that the MITF-HA expression construct was fully functional. Consistent with the functional assays, ectopic MITF-HA localized predominantly to the nucleus in transfected U2OS cells (Fig 3C). Exogenous MITF does not reactivate target genes in MITFmessage-negative melanoma cells We then asked whether the forced expression of MITF can restore the melanocytic phenotype in melanoma cell lines where MITF transcription was repressed. To this end, we ®rst tested whether the pCMV-MITFHA construct stimulates the transiently transfected tyrosinase

1508

VACHTENHEIM ET AL

Figure 3. Testing the expression construct for MITF. (A)COS7 cells were transiently transfected with the vector alone (lane 1), pCMVMITF-HA (lane 2), or pCMV-MITF(DbHLH-LZ)-HA (lane 3) and the reporter plasmid pTyr-prom-Luc. pCMV-b was cotransfected to normalize transfection ef®ciency. (B)U2OS cells were transfected with the expression plasmids as in (A) and the expression of endogenous tyrosinase was detected by RT-PCR (lane 4 represents the positive control from the melanoma cell line DOR). (C) Cellular localization of the MITF protein expressed from the pCMV-MITF-HA plasmid revealed by staining with the anti-HA antibody 12CA5 in U2OS cells. (D) MITF stimulated the tyrosinase promoter±reporter in IgR3 and RPMI7951 cell lines. The transfected plasmids were the same as in (A).

promoter±reporter in these cell lines. As in COS7 cells (Fig 3A) and in several other cell lines (not shown), pCMV-MITF-HA activated the transiently delivered tyrosinase promoter in two human melanoma cell lines IgR3 and RPMI7951 (Fig 3D), which were devoid of any MITF-M mRNA (Fig 1). Unlike U2OS cells, however, these two cell lines did not support the induction of endogenous tyrosinase by transiently transfected MITF (data not shown). We therefore introduced MITF-HA via retroviral infection into these two melanoma cell lines and analyzed the selected clones (the third MITF-negative cell line Dau was not used because of its slow proliferation rate and very low transfection ef®ciency). Figure 4 (lanes 1, 2) shows an example of this experiment: the two puromycin-resistant clones of the MITFHA-transduced IgR3 cells that express MITF mRNA and protein levels comparable to those seen in differentiated melanoma lines expressing endogenous MITF, SK-MEL-5 and SK-MEL-28 (Fig 4, lanes 4 and 5, respectively). Importantly, the ectopic MITF-HA was not suf®cient to induce the transcription of the endogenous tyrosinase gene, as detected by northern blot or RTPCR (Fig 4, lanes 1, 2). As a positive control, a strong signal of the tyrosinase message was observable in the pigmented melanoma cells containing comparable levels of MITF protein (Fig 4, lanes 4, 5). U2OS cells transiently transfected with the MITF-HA vector showed the tyrosinase signal on RT-PCR gel, whereas the MITF protein was still below the detection limit of the D5 antibody (Fig 4, lane 7). Further, several clones or pools of transduced RPMI7951 cell line and several additional clones of infected IgR3

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 4. Retrovirally transduced MITF does not restore the transcription of melanogenic markers in clones of IgR3 cells. Puromycin-resistant clones of IgR3 cells (clones 8 and 9) with retrovirally transduced MITF-HA (lanes 1, 2), control IgR3 cells (lane 3), two differentiated melanoma cell lines SK-MEL-5 and SK-MEL-28 (lanes 4, 5) as positive controls, HeLa cell line (lane 6) as a negative control, and transiently transfected U2OS cells (lanes 7, 8) are shown. (A) Northern blot was probed with the MITF cDNA probe; (B)Western blot was performed with the anti-MITF D5 monoclonal antibody; (C) the same northern blot as in (A) reprobed with the human tyrosinase cDNA; (D) RT-PCR analysis of the tyrosinase mRNA; (E) ethidium bromide staining of 28S RNA of a gel used for northern blot; (F) control RT-PCR analysis (GAPDH). IgR3 cells transduced with an empty retroviral vector gave the same results as native IgR3 (not shown). The HA-tagged MITF protein is retarded in migration (B, lanes 1 and 2, left arrow) compared to the endogenous one due to the epitope tagging. Endogenous protein is marked by an arrow in lane 4.

cells where MITF protein was detectable by Western blot likewise did not reveal any onset of transcription of the target genes (results not shown). These ®ndings suggest that, whereas the transiently transfected promoter can be activated (Fig 3D), an unknown mechanism prevents the activation of the endogenous MITFregulated genes by forced expression of MITF. We also did not observe any signi®cant effects of MITF on cell morphology or growth rate in any of the puromycin-resistant clones derived from IgR3 or RPMI7951 (not shown). The MITF transcription activation domain is functional in all cells tested independently of their melanogenic potential, and ectopic MITF binds to DNA in vitro The N-terminal part of Mitf is able to activate transcription and the activation domain can be further narrowed to a core motif of amino acids 114±132 (Sato et al, 1997). Therefore, to elucidate the function of the MITF transactivation domain, we evaluated the ability of the sequences involved in activation (amino acids 105±199 fused to the GAL4DB domain) to stimulate transcription from the GAL reporter in cell lines with different melanogenic potential. A strong, dosedependent activation of the luciferase activity was found in all cell types tested (Fig 5A). When the highest amount of the GAL± MITF construct was transfected, the stimulation ranged from about 22-fold in RPMI7951 cells to 130-fold in IgR3 cells (Fig 5A, left and right panels). These results indicate that the strength of the activation domain does not depend on whether the cells have a

VOL. 117, NO. 6 DECEMBER 2001

MITF REPRESSION IN MELANOMAS

1509

gel retardation assays. Pigmented melanoma cell lines SK-MEL-5 and SK-MEL-28 used as controls showed strong supershifted bands upon addition of anti-MITF antibody (Fig 5C). The IgR3-derived MITF-expressing clones 8 and 9 likewise revealed this supershift, and although the supershifted band was attenuated compared to pigmented melanomas, these data clearly demonstrated that exogenous MITF could bind its consensus DNA sequence. No binding was seen in IgR3 cells (Fig 5C). Thus, as the endogenous targets were not activated, these experiments suggest that the binding of exogenous MITF to DNA may be nonproductive in IgR3-derived clones. DISCUSSION

Figure 5. Stimulation of GAL-responsive reporter by the MITF activation domain in various cells, subcellular localization of MITF in IgR3 and RPMI7951 cells, and DNA-binding activity of MITF. (A) The indicated amounts of the GAL±MITF fusion plasmid were transfected together with the reporter gene and the stimulation of luciferase activity was measured after 48 h. Each experiment was performed three to ®ve times (in duplicate samples) and one representative set of results is shown with standard deviation. (B) Cells grown on coverslips were transfected with the pCMV-MITF-HA plasmid and analyzed 48 h later. Immuno¯uorescence was performed by staining cells with the anti-HA antibody 12CA5. Shown are IgR3 melanoma cells (left) and RPMI7951 melanoma cells (right). Selected clones of cells with retrovirally transduced MITF-HA gave the same pattern of localization (not shown). (C) Gel shift assays with nuclear extracts prepared from the indicated cell lines. E-box derived from the tyrosinase promoter initiator sequence was used as a probe. Arrow indicates the supershift produced by the anti-MITF antibody (D5). Nonimmune IgG did not cause any supershift (not shown).

melanogenic potential or not. Several other cell lines that did not support the activation of endogenous tyrosinase by MITF in transient transfections (NIH3T3, A549, ras-transformed mouse melanocytes, and COS7) likewise revealed a similar pattern of activation by the GAL±MITF(105±199) fusion construct (data not shown). The data together indicate that the domain of MITF responsible for transcriptional activation is not inhibited in cells where the full-length functional MITF could not activate the endogenous tyrosinase gene. Furthermore, subcellular localization of MITF-HA protein was nuclear in IgR3 and RPMI7951 cells (Fig 5B) as well as in other cell types tested (not shown). To test whether ectopic MITF present in nuclear extracts of IgR3-derived clones (Fig 4) could bind to DNA in vitro, we carried out gel mobility shift assays. As the tyrosinase initiator E-box sequence is essential for the activation of the tyrosinase promoter by MITF (Bentley et al, 1994), we used it as a probe in

We have determined the expression of several genes related to differentiation in human melanomas, including the bHLH-LZ cellfate-determining gene MITF. Three of 17 melanoma cell lines showed total repression of MITF-M gene transcription, accompanied by the repression of all three MITF transcriptional targets. Furthermore, TRP-1 gene was selectively repressed in several MITF-positive cell lines, suggesting the existence of a speci®c repressor for the TRP-1 gene. Similar repression of TRP-1 has been reported earlier on several other melanoma cell lines; however, the MITF expression has not been analyzed (Eberle et al, 1995). A candidate gene for this repression function could be the Tbx2 protein, a T-box family member implicated in limb development, because Tbx2 was shown recently to repress the mouse TRP-1 promoter by binding to the speci®c Tbx2 binding site (Carreira et al, 1998). Our results thus support the idea that TRP-1 gene expression, besides its requirement for MITF, is regulated by additional mechanisms. It is interesting that MITF has also been proposed to transactivate the promoter of the mouse Tbx2 gene and the Tbx2 protein has been found expressed in pigmented mouse melanocytes and melanomas (Carreira et al, 2000). We have not determined here the expression pattern of the human Tbx2 gene in a panel of human melanomas. Clearly, further studies are needed to establish if Tbx2 is an in vivo target of MITF and the possible physiologic role of Tbx2 in melanocytes. In studies demonstrating the extinction of pigmentation in microcell hybrids, it was reported that human and mouse ®broblasts contain an extinguisher locus that suppresses the transcription of melanocyte marker genes (Powers et al, 1994; Powers and Davidson, 1996). It has been suggested that this extinction, in hybrids of mouse ®broblasts and hamster melanoma cells, could be due to the repression of microphthalmia (Powers and Davidson, 1996). Therefore, it is possible that an extinguisher locus has been activated in a few melanomas in the process of dedifferentiation. In this paper we further report that melanoma cells de®cient in microphthalmia protein expression are unable to reexpress endogenous MITF targets after the forced expression of MITF. These dedifferentiated melanoma cells thus resemble nonmelanocyte lineages rather than the original melanocytes. The transcriptional activity and degradation of the microphthalmia protein is post-translationally regulated by phosphorylation on serines S73 and S409, and the S73 MAPK-phosphorylated species binds the p300 coactivator and is more potent in transcription from the tyrosinase promoter±reporter (Hemesath et al, 1998; Wu et al, 2000). This phosphorylated MITF protein appears as a retarded band on sodium dodecyl sulfate polyacrylamide gel electrophoresis (Hemesath et al, 1998) and we have also observed this slowly migrating band as a major MITF form on Western blot (Fig 4B). It is therefore unlikely that the inability of ectopic MITF to induce endogenous targets in IgR3 and RPMI7951 cells would be a result of insuf®cient phosphorylation by the MAPK pathway (Erk-2 kinase) in these cells. Another reason underlying the inability of MITF to reactivate the differentiation program in MITF-negative melanomas could be that an essential heterodimeric partner of MITF required for the transcriptional activation was repressed concomitantly with MITF. MITF homodimers as well as heterodimers with TFEB, TFEC, and

1510

VACHTENHEIM ET AL

TFE3 bind ef®ciently to the E-box (Hemesath et al, 1994). Evolutionarily closest to MITF is TFE3, which seems to be widely expressed (Rehli et al, 1999), but its expression in melanomas has not been studied. Only recently, although TFE3 was demonstrated to stimulate tyrosinase and TRP-1 promoters in reporter assays, endogenous TFE3 homodimers did not bind to the M-box and MITF±TFE3 heterodimers were not found in B16 mouse melanoma cells (Verastegui et al, 2000b). It is thus uncertain what complexes operate in vivo or whether additional factors are required to activate MITF recruited to the tyrosinase promoter in a speci®c cellular context. It is believed, however, that MITF can activate transcription in the absence of any other member of the MiT subfamily. We have also observed here that exogenous MITF in IgR3-derived clones could bind the tyrosinase initiator E-box in gel retardation assays. This would indicate that nuclear extracts do not contain inhibitors preventing the MITF±DNA interaction in vitro. We cannot, however, exclude the possibility that in vivo this binding does not occur, for instance due to the nucleosome repositioning or acquiring a state of remodeled nucleosomes around the tyrosinase promoter that does not favor the in vivo binding. Several recent reports have suggested that microphthalmia protein can be a speci®c and useful marker for malignant melanoma, being present in almost all specimens of pigmented melanomas (King et al, 1999; 2001). In contrast, only a subset of desmoplastic melanomas was stained by the anti-MITF antibody (King et al, 2001; Busam et al, 2001; Koch et al, 2001; Miettinen et al, 2001). Although we could not verify whether the IgR3 and RPMI7951 melanoma cells used in this work were originally diagnosed as desmoplastic melanomas, it is notable that RPMI7951 cells originally formed pigmented melanomas in nude mice (as mentioned in the ATCC cell data sheet), and thus MITF expression might have been lost during the propagation. Further, given that microphthalmia plays a fundamental role in the survival of embryonal melanocytes and is expressed in virtually all melanoma specimens (King et al, 1999), it can be speculated that MITF also plays a role in the proliferation of malignant melanocytes. If so, when the expression of MITF is lost in malignant melanocytes, as in desmoplastic melanomas, they must develop a program independent of MITF to sustain proliferation. If MITF is then no longer required for proliferation or survival of these dedifferentiated tumor cells, the target promoters are not activated and ultimately may become nonresponsive to MITF, if this is delivered later ectopically. This nonresponsiveness of the natural MITF targets to forced expression of MITF, as described in this paper, would be compatible with this hypothesis. Our results thus warrant further search for yet unknown downstream genes transcriptionally regulated by MITF that might prove to be important for the proliferation of malignant melanocytes. We thank Dr. C. Goding for his generous contribution of plasmids. The technical assistance of I. Kohoutova is also appreciated. This work was supported by grants 3955-3 and 5428-3 from Internal Granting Agency, Ministry of Health, Czech Republic, by an institutional research project (No. MZ00000064211), and partially by the League Against Cancer, Prague.

REFERENCES Amae S, Fuse N, Yasumoto K, et al: Identi®cation of a novel isoform of microphthalmia-associated transcription factor that is enriched in retinal pigment epithelium. Biochem Biophys Res Commun 247:710±715, 1998 Bentley NJ, Eisen T, Goding CR: Melanocyte-speci®c expression of the human tyrosinase promoter: activation by the microphthalmia gene product and role of the initiator. Mol Cell Biol 14:7996±8006, 1994 Bertolotto C, Bille K, Ortonne JP, Ballotti R: Regulation of tyrosinase gene expression by cAMP in B16 melanoma cells involves two CATGTG motifs surrounding the TATA box: implication of the microphthalmia gene product. J Cell Biol 134:747±755, 1996 Bertolotto C, Busca R, Abbe P, Bille K, Aberdam E, Ortonne J-P, Ballotti R: Different cis-acting elements are involved in the regulation of TRP-1 and TRP-2 promoter activities by cyclic AMP. Pivotal role of M boxes (GTCATGTGCT) and of microphthalmia. Mol Cell Biol 18:694±702, 1998a

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Bertolotto C, Abbe P, Hemesath TJ, Bille K, Fisher DE, Ortonne J-P, Ballotti R: Microphthalmia gene product as a signal transducer in cAMP-induced differentiation in melanocytes. J Cell Biol 142:827±835, 1998b Bondurand N, Pingault V, Goerich DE, et al: Interaction among SOX10, PAX3 and MITF, three genes altered in Waardenburg syndrome. Hum Mol Genet 9:1907± 1917, 2000 Busam KJ, Iversen K, Coplan KC, Jungbluth AA: Analysis of microphthalmia transcription factor expression in normal tissues and tumors, and comparison of its expression with S-100 protein, gp100, and tyrosinase in desmoplastic malignant melanoma. Am J Surg Pathol 25:197±204, 2001 Carreira S, Dexter TJ, Yavuzer U, Easty DJ, Goding CR: Brachyury-related transcription factor Tbx2 and repression of the melanocyte-speci®c TRP-1 promoter. Mol Cell Biol 18:5099±5108, 1998 Carreira S, Liu B, Goding CR: The gene encoding the T-box factor Tbx2 is a target for the microphthalmia-associated transcription factor in melanocytes. J Biol Chem 275:21920±21927, 2000 Chen C, Okayama H: Calcium phosphate mediated gene transfer into established cell lines. In: EJ Murray, eds. Gene Transfer and Expression Protocols. Clifton: Humana Press, 1991:pp 15±21 DelMarmol V, Ito S, Jackson I, et al: TRP-1 expression correlates with eumelanogenesis in human pigment cells in culture. FEBS Lett 327:307±310, 1993 Dotto GP, Moellmann G, Ghosh S, Edwards M, Halaban R: Transformation of murine melanocytes by basic ®broblast growth factor cDNA and oncogenes and selective suppression of the transformed phenotype in a reconstituted cutaneous environment. J Cell Biol 109:3115±3128, 1989 Eberle J, Garbe C, Wang N, Orfanos CE: Incomplete expression of the tyrosinase gene family (tyrosinase, TRP-1, and TRP-2) in human malignant melanoma cells in vitro. Pigment Cell Res 8:307±313, 1995 Eberle J, Wagner M, MacNeil S: Human melanoma cell lines show little relationship between expression of pigmentation genes and pigmentary behaviour in vitro. Pigment Cell Res 11:134±142, 1998 Goding CR: Mitf from neural crest to melanoma: signal transduction and transcription in the melanocyte lineage. Genes Dev 14:1712±1728, 2000 Halaban R, Bohm M, Dotto P, Moellmann G, Cheng E, Zhang Y: Growth regulatory proteins that repress differentiation markers in melanocytes also downregulate the transcription factor microphthalmia. J Invest Derm 106:1266± 1272, 1996 Hemesath TJ, SteigrõÂmsson E, McGill G, et al: Microphthalmia, a critical factor in melanocyte development, de®nes a discrete transcription factor family. Genes Dev 8:2770±2780, 1994 Hemesath TJ, Price ER, Takemoto C, Badalian T, Fisher DE: MAP kinase links the transcription factor microphthalmia to c-Kit signalling in melanocytes. Nature 391:298±301, 1998 Kameyama K, Takemura T, Hamada Y, et al: Pigment production in murine melanoma cells is regulated by tyrosinase, tyrosinase-related protein 1 (TRP-1), DOPAchrome tautomerase (TRP-2), and a melanogenic inhibitor. J Invest Dermatol 100:126±131, 1993 Kameyama K, Sakai C, Kuge S, et al: The expression of tyrosinase, tyrosinase-related proteins 1 and 2 (TRP1 and TRP2), the silver protein, and a melanogenic inhibitor in human melanoma cells of differing melanogenic activities. Pigment Cell Res 8:97±104, 1995 King R, Weilbaecher KN, McGill G, Cooley E, Mihm M, Fisher DE: Microphthalmia transcription factor. A sensitive and speci®c melanocyte marker for melanoma diagnosis. Am J Pathol 155:731±738, 1999 King R, Googe PB, Weilbaecher KN, Mihm MC, Fisher DE: Microphthalmia transcription factor expression in cutaneous benign, malignant melanocytic, and nonmelanocytic tumors. Am J Surg Pathol 25:51±57, 2001 Koch MB, Shih IM, Weiss SW, Folpe AL: Microphthalmia transcription factor and melanoma cell adhesion molecule expression distinguish desmoplastic/spindle cell melanoma from morphologic mimics. Am J Surg Pathol 25:58±64, 2001 Lee M, Goodall J, Verastegui C, Ballotti R, Goding CR: Direct regulation of the microphthalmia promoter by Sox10 links Waardenburg±Shah syndrome (WS4)-associated hypopigmentation and deafness to WS2. J Biol Chem 275:37978±37983, 2000 Melber K, Zhu G, Diamond L: SV40-transfected human melanocyte sensitivity to growth inhibition by the phorbol ester 12-O-tetradecanoyl-13-acetate. Cancer Res 49:3650±3655, 1989 Miettinen M, Fernandez M, Franssila K, Gatalica Z, Lasota J, Sarlomo-Rikala M: Microphthalmia transcription factor in the immunohistochemical diagnosis of metastatic melanoma: comparison with four other melanoma markers. Am J Surg Pathol 25:205±211, 2001 Moore KJ: Insight into the microphthalmia gene. Trends Genet 11:442±448, 1995 Morgenstern JP, Land H: Advanced mammalian gene transfer: high titre retroviral vectors with multiple drug selection markers and a complementary helper-free packaging cell line. Nucl Acid Res 18:3587±3596, 1990 Potterf SB, Furumura M, Dunn KJ, Arnheiter H, Pavan WJ: Transcription factor hierarchy in Waardenburg syndrome: regulation of MITF expression by SOX10 and PAX3. Hum Genet 107:1±6, 2000 Powers TS, Davidson RL: Coordinate extinction of melanocyte-speci®c gene expression in hybrid cells. Somatic Cell Mol Genet 22:41±56, 1996 Powers TS, Shows TB, Davidson RL: Pigment-cell-speci®c genes from ®broblasts are transactivated after chromosomal transfer into melanoma cells. Mol Cell Biol 14:1179±1190, 1994 Rehli M, Den Elzen N, Cassady AI, Ostrowski MC, Hume DA: Cloning and characterization of the murine genes for bHLH-ZIP transcription factors TFEC

VOL. 117, NO. 6 DECEMBER 2001

and TFEB reveal a common gene organization for all MiT subfamily members. Genomics 56:111±120, 1999 Sato S, Roberts K, Gambino G, Cook A, Kouzarides T, Goding CR: CBP/p300 as a cofactor for the microphthalmia transcription factor. Oncogene 14:3083±3092, 1997 Tachibana M, Perez-Jurado LA, Nakayama A, et al: Cloning of MITF, the human homolog of the mouse microphthalmia gene and assignment to chromosome 3p14.1-p12.3. Hum Mol Genet 3:553±557, 1994 Tachibana M, Takeda K, Nobukuni Y, et al: Ectopic expression of MITF, a gene for Waardenburg syndrome type 2, converts ®broblasts to cells with melanocyte characteristics. Nature Genet 14:50±54, 1996 Takeda K, Yasumoto K, Takada R, et al: Induction of melanocyte-speci®c microphthalmia-associated transcription factor by Wnt-3a. J Biol Chem 275:14013±14016, 2000 Turque N, Denhez F, Martin P, et al: Characterization of a new melanocyte-speci®c gene (QNR-71) expressed in v-myc-transformed quail neuroretina. EMBO J 15:3338±3350, 1996 Vachtenheim J, Novotna H: Expression of genes for microphthalmia isoforms, Pax3 and MSG1, in human melanomas. Cell Mol Biol 45:1075±1082, 1999 Verastegui C, Bille K, Ortonne JP, Ballotti R: Regulation of the microphthalmiaassociated transcription factor gene by the Waardenburg syndrome type 4 gene, SOX10. J Biol Chem 275:30757±30760, 2000a

MITF REPRESSION IN MELANOMAS

1511

Verastegui C, Bertolotto C, Bille K, Abbe P, Ortonne JP, Ballotti R: TFE3, a transcription factor homologous to microphthalmia, is a potential transcriptional activator of tyrosinase and TyrpI genes. Mol Endocrinol 14:449±456, 2000b Watanabe A, Takeda K, Ploplis B, Tachibana M: Epistatic relationship between Waardenburg syndrome genes MITF and PAX3. Nature Genet 18:283±286, 1998 Wilson RE, Dooley TP, Hart IR: Induction of tumorigenicity and lack of in vitro growth requirement for 12-O-tetradecanoylphorbol-13-acetate by transfection of murine melanocytes with v-Ha-ras. Cancer Res 49:711±716, 1989 Wu M, Hemesath TJ, Takemoto CM, et al: c-Kit triggers dual phosphorylations, which couple activation and degradation of the essential melanocyte factor Mi. Genes Dev 14:301±312, 2000 Yasumoto K-I, Yokoyama K, Takahashi K, Tomita Y, Shibahara S: Functional analysis of microphthalmia-associated transcription factor in pigment cellspeci®c transcription of the human tyrosinase gene family. J Biol Chem 272:503±509, 1997 Yavuzer U, Keenan E, Lowings P, Vachtenheim J, Currie G, Goding CR: The microphthalmia gene product interacts with the retinoblastoma protein in vitro and is a target for deregulation of melanocyte-speci®c transcription. Oncogene 10:123±134, 1995