Identification of the Elements Regulating the Expression of the Cell Adhesion Molecule MCAM/MUC18

The cell adhesion molecule melonoma cell adhesion molecule (MCAM)/MUC18/CD146 is specifically up-regulated on tumors of neuroectodermal origin and in animal models confers metastatic capacity to human melanoma cells. To identify critical regions regulating MCAM expression in melanomas, 1 kilobase of the MCAM 5′ region was analyzed for promoter activity and transcription factor binding in 1 glioma, 1 carcinoma, and 4 melanoma cell lines. The minimal MCAM promoter (−106/+22 base pair (bp)) consists of 4 Sp-1 sites, two AP-2 elements, one cAMP responsive element, and the initiator surrounding the transcriptional start site. Analysis of mutated constructs indicated that the cAMP-responsive element is a major transcriptional activator in the majority of cell lines. Site-directed mutagenesis revealed that, in AP-2 expressing cells, the AP-2 site within the core promoter (−23 bp) has an inhibitory influence on MCAM expression while the AP-2 sites at −131 and −302 bp are activating. Functional AP-2 was observed in both MCAM positive and MCAM negative melanoma cell lines indicating that expression of MCAM does not require loss of this transcription factor. Furthermore, all MCAM constructs were strongly expressed in MCAM negative as well as MCAM positive cells, indicating that the expression of this gene is not controlled solely by the presence of transactivating factors binding to the investigated region.

Adhesion molecules mediate cell-cell and cell-matrix interactions and also function as signal transducers thereby playing fundamental roles in many physiological and pathological processes (1)(2)(3). Alterations in the adhesive properties of tumor cells have been implicated in tumorigenesis and in the biological behavior of many tumors (2,3). In human malignant melanoma the development of advanced and metastatic disease is associated with the induction or up-regulation of the 113-kDa highly glycosylated integral membrane protein, MCAM 1 / MUC18 (CD146). MCAM is predominantly expressed by smooth muscle and vascular endothelium in normal adult tissues (4,5) and has been shown to function as a cell adhesion molecule, mediating cation independent adhesion through interaction with an unidentified ligand (6,7).
MCAM was originally identified as a melanoma-associated antigen and is only rarely observed on carcinomas (8,9). In melanoma MCAM is a progression antigen, being strongly expressed only in malignant melanocytic lesions and increasing in frequency and strength with increasing vertical thickness of primary tumors. The strongest expression is observed in metastatic lesions (4,10). Studies in animal models suggest that MCAM may in fact play a role in tumor growth or metastasis formation in vivo. De novo expression of MCAM cDNA conferred metastatic potential, as assessed in experimental metastasis assays, to two independent MCAM negative human melanoma cell lines (11,12). In unrelated experiments, the generation of highly tumorigenic variants by insertional mutagenesis in the slow growing melanoma cell line WM35 was also associated with an induction or up-regulation of MCAM (13).
Inasmuch as MCAM expression by melanomas appears to be associated with tumor growth and/or metastatic capacity, it is important to understand how its expression in these cells is regulated. The MCAM 5Ј region contains multiple potential binding sites for Sp1 and AP-2 and a CRE (14) and two recent studies have examined the role of these elements in regulating expression in melanomas. Jean et al. (15) reported that constitutive MCAM expression in melanoma cells is due to the loss of the transcription factor AP-2 which represses MCAM promoter activity. Karlen and Braathen (16) were unable to identify a role for AP-2 and concluded that Sp1 is necessary and sufficient to account for constitutive MCAM expression in melanoma cell lines. In the study presented here, approximately 1 kilobase of the 5Ј region of the human MCAM gene has been analyzed for promoter activity and transcription factor binding in 3 MCAM positive and 3 MCAM negative cells lines. Deletion and sitespecific mutational analyses indicate that Sp1, CRE, and AP-2 are all important regulators of MCAM transcription in melanoma cell lines. Functional AP-2 was observed in both MCAM expressing and nonexpressing cells indicating that MCAM expression does not require the loss of this transcription factor. Furthermore, in this larger panel of cells, 5Ј region activity as assessed in transient transfection assays did not correlate with MCAM expression. These results indicate that additional elements are involved in regulating MCAM expression in melanoma cells.
Transient Transfection-For transfection of the reporter constructs, the FuGENE 6 transfection reagent (Roche Molecular Biochemicals, Mannheim, Germany) was used according to the manufacturer's recommendations. 2.5 ϫ 10 5 cells were transfected with 3 l of FuGENE 6 reagent and 1 g of DNA and assayed 48 h later using luciferase lysis and assay reagents from Promega. 20 l of cell lysate was measured with 100 l of luciferase assay reagent. Expression of luciferase in cells transfected with the promoterless pXP-2 vector and the pXP-2 plasmid containing the Rous sarcoma virus promoter, were used as negative and positive controls, respectively.
To standardize the measured relative light units, the protein content of the lysates was determined using the Bio-Rad protein assay reagent (Bio-Rad) and luciferase activity was presented as relative light units/g of protein. Transfections were carried out in triplicate and at least three independent experiments were performed, using different batches of DNA. DNA was prepared using the JETSTAR 2.0 Plasmid kit (Genomed Inc., Bad Oeyenhausen, Germany).
Preparation of Nuclear Extracts-Nuclear extracts were prepared according to a modified method of Schreiber et al. (18). Approximately 5 ϫ 10 7 cells were washed 3 times in PBS and lysed in 2 ml of low salt buffer A (10 mM Hepes-KOH, pH 7.6, 15 mM KCl, 2 mM MgCl 2 , 0.1 mM EDTA, pH 8.0, 1 mM dithiothreitol, complete Mini-EDTA free protease inhibitor mixture tablet (Roche Molecular Biochemicals)) ϩ 1% Nonidet P-40. Nuclei were pelleted by centrifugation for 5 min at 2200 rpm, 4°C, and washed once with 2 ml of buffer A ϩ Nonidet P-40. 300 l of buffer C (25 mM Hepes-KOH, pH 7.6, 50 mM KCl, 0.1 mM EDTA, pH 8.0, 10% glycerol, 1 mM dithiothreitol, Complete Mini-EDTA free protease inhibitor mixture) was added to 500 l of pellet volume. The salt concentration of the nuclear suspension was adjusted to 0.4 mM NaCl and nuclear protein was extracted on ice for 45-60 min. Debris was pelleted by centrifugation and the supernatants aliquoted and stored at Ϫ80°C. Protein concentrations were determined using the Bio-Rad protein assay reagent.
AP-2 mobility shift reactions contained 17 l of the premixed incubation buffer (Stratagene, La Jolla, CA), 50,000 cpm oligonucleotide (5Ј-GAACTGACAGCCCCCGGCAGCCCCCG-3Ј, MCAM-specific sequence is underlined), and 10 g of nuclear extract in a reaction volume of 25 l. The reaction was incubated for 20 min on ice. For supershift reactions, antibodies (4 l of anti-AP-2 or 4 l of control serum) were added for the last 20 min of a 40-min incubation of protein with DNA. Double-stranded MCAM-specific oligonucleotides containing the CRE, Sp1, and AP-2 sites in the core promoter were end labeled with ␥-32 P using T4 polynucleotide kinase (New England Biolabs, Beverly, MA). Nonlabeled competitor oligonucleotides, when used, were added at the same time as the labeled oligomers. Antibodies to transcription factors and the oligonucleotides containing the consensus recognition motifs used in EMSA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Human recombinant AP-2 was purchased from Promega. The contents of the binding reactions were electrophoresed at room temperature on a 4% nondenaturing polyacrylamide gel in 0.25 ϫ Tris borate/EDTA running buffer for 2-3 h. Gels were dried and exposed to x-ray film.
Immunoprecipitation and Western Blotting-For MCAM Western blotting, cells were lysed in PBS containing 1% Nonidet P-40 (Calbiochem, San Diego, CA) and protease inhibitors. The protein content of the samples was measured using the Lowry protein assay (Bio-Rad). 0.5 mg of protein lysate was precleared by overnight incubation with horse serum-saturated Sepharose. After preclearing, the samples were exposed to anti-MCAM monoclonal antibody MUC18BA.3, bound to Protein G-agarose (Sigma), overnight at 4°C. The bound material was eluted from the washed immunosorbents (95°C, 5 min), separated on a 7.5% SDS-PAGE gel under reducing conditions and electrophoretically transferred to nitrocellulose membranes. The filters were blocked for 1 h in 5% skim milk in PBS and then incubated for 1 h with monoclonal antibody MUC18BA.3 followed by horseradish peroxidase-linked rabbit anti-mouse immunoglobulin (anti-IgM ϩ IgG) 1:5000 (Dako, Hamburg, Germany). The membranes were washed in PBS ϩ 0.1% Tween and the bound antibody was visualized by the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Uppsala, Sweden) according to the manufacturer's instructions.
For the detection of AP-2 protein, 30 g of nuclear extract was separated on a 7.5% SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was blocked in 1% casein in PBS and incubated with anti-AP-2 serum (C-18, 1:500, Santa Cruz Biotechnology) followed by horseradish peroxidase-linked goat anti-rabbit immunoglobulin 1:8000 (Dako).
RT-PCR for mRNA Detection-Total RNA was isolated using the RNeasy extraction kit from Qiagen. The RNA preparation was subjected to DNase I digestion and phenol extraction followed by RNA precipitation with 100% ethanol overnight at Ϫ20°C. 5 g of total RNA was used for reverse transcription (RT)-PCR. First strand cDNA synthesis was performed in a reaction volume of 20 l using Superscript II Reverse Transcriptase (Life Technologies, Paisley, United Kingdom) according to the manufacturer's instructions. 5% of the first strand reaction was used for PCR. Samples were subjected to 35 rounds of temperature cycling (94°C for 35 s, 58°C for 55 s, 70°C for 50 s).

RESULTS
The MCAM 5Ј region contains no TATA or CAAT boxes, but instead an initiator sequence surrounding the transcriptional start 29 bp upstream of the first ATG (14). The 150-bp region immediately upstream of the initiator is highly GC-rich and contains five potential recognition sequences for the transcription factor Sp1 (Ϫ41/Ϫ36 bp, Ϫ72/Ϫ67 bp, Ϫ77/Ϫ72 bp, Ϫ100/ Ϫ95 bp, and Ϫ125/Ϫ119 bp), two for AP-2 (Ϫ23 bp, Ϫ131 bp) and a consensus sequence motif for the cAMP response element at position Ϫ32. Additional AP-2 elements are found further upstream at Ϫ302 and Ϫ505 and downstream at position ϩ14. At positions Ϫ208 and Ϫ591 CArG box motifs (5Ј-CC(A or T) 6 GG) are present and at position Ϫ236 is a binding element for the transcription factor MyoD. In order to examine the role of elements in the 5Ј region of the MCAM gene in regulating tissue restricted expression, 934 bp (Ϫ912/ϩ22 bp) of this region were fused to the promoterless firefly luciferase reporter gene. This construct as well as serially 5Ј truncated fragments and constructs carrying deletions and site directed mutations The luciferase activity driven by the 934-bp MCAM 5Ј region (mcluc-912) in each of these cell lines is shown in Fig. 1C. The promoterless plasmid pXP-2 was used as negative control and luciferase activity is presented as relative light units/g of protein.
Expression ranged from 800-fold (SB-2) to 10 4 -fold (Mel888) over that observed for pXP-2. Luciferase activity driven by the Rous sarcoma virus promoter was higher but showed a similar expression pattern (data not shown), suggesting that the differences observed between the cells reflect differences in transfection efficiencies. The MCAM full-length construct is strongly expressed in both MCAM expressing and MCAM nonexpressing cells indicating that MCAM expression is not solely determined by the transactivating factors which bind to this region.
The activity of a series of deletion constructs was similar in all cells and results are presented for LN215, Mel888, and WM2664 (Fig. 2). In these studies the luciferase activity of the various constructs is presented in relation to that of the full-length mcluc-912 which was set at 100%. The expression of construct mcluc-del bearing an internal deletion of 151 bp directly upstream of the transcriptional start, was less then 10% of that observed with the intact 5Ј region (mcluc-912), indicating a critical role for this region in MCAM transcription. The luciferase activity driven by constructs mcluc-173, consisting solely of the deleted region plus the initiator, and mcluc-106 were similar to (LN215) or higher than (Mel888, WM2664) that observed with the full-length 912-bp region. These observations indicate that all of the information required for optimal promoter activity is present in the first 128 base pairs of the MCAM 5Ј region. Further deletion of Sp1 sites (mcluc-62 and mcluc-37) reduced promoter activity to less than 20% of the full-length clone indicating an important role of Sp1 as a transcriptional activator of MCAM expression. The additional deletion of the putative CRE (Ϫ32 bp) and the AP2 site at Ϫ23 bp (mcluc-7) abolished the remaining activity in all examined cell lines.
The minimal region of the MCAM promoter required to drive transcription is therefore included in the 128-bp region from Ϫ106 to ϩ22 bp. This region contains 4 Sp1 sites, a CRE site, two AP2-binding sites, and the initiator surrounding the tran-scription start.
DNA mobility shift assays, using an MCAM-specific DNA fragment (Ϫ47 bp/Ϫ33 bp) containing the proximal Sp1 site produced a double band DNA-protein complex with nuclear extracts from all 6 cell lines. As shown for the cell line MelJuSo (Fig. 3A), this complex was abolished by an excess of unlabeled probe (lanes 2 and 3) and by addition of unlabeled Sp1 consensus oligonucleotide (lane 4), but not by mutated Sp1 sequence (lane 5) or by an oligonucleotide containing a binding site for the CREB/ATF family (lane 6). An antibody directed against Sp1 specifically blocked formation of the DNA-protein complex (Fig. 3B, lane 2) indicating that the protein binding to this site is immunologically related to Sp1.
The cAMP-responsive Site Is an Important Element of the MCAM Core Promoter-To assess the role of the cAMP responsive element at Ϫ32 bp in the regulation of basal MCAM expression, this site was mutated in the context of the fulllength construct and tested in transient transfection assays. As shown in Fig. 4A, mutation of the CRE site reduced reporter activity by approximately 70% compared with mcluc-912 in five of the six tested cell lines indicating that this site plays an important role in the MCAM core promoter. Interestingly no reduction of luciferase expression with the CRE mutant could be observed in the cell line WM2664. DNA shift analyses were performed with a MCAM-specific oligonucleotide including this motif (Ϫ36 bp/Ϫ22 bp). A complex consisting of a double band was detected in nuclear extracts from all 6 cell lines (Fig. 4B). Binding was abrogated by an excess of unlabeled oligonucleotide (lane 2) and by an excess of consensus CRE oligonucleotide (lane 3), but not by addition of a mutated CRE oligonucleotide (lane 4) or an AP-2 binding motif (lane 5). Identical results were obtained with nuclear extracts from WM2664 and MelJuSo. The observed DNA-protein complex could be partially supershifted with anti-CREB-1 antibody (Fig. 4C, lane 2), indicating that one of the proteins binding to this sequence is CREB1. Although specific protein binding to the CRE site was also shown for the cell line WM2664 this regulatory element seems not to be as important in mediating transcriptional activation as in the other cell lines examined.
The Core Promoter AP-2 Site Acts as a Negative Element While Upstream AP-2 Sites Act as Positive Elements-A perfect consensus AP-2-binding site is located at Ϫ23 bp within the MCAM core promoter. To test the role of this site in MCAM transcription it was mutated in the construct mcluc-106, which still shows optimal promoter activity. Luciferase activity driven by mcluc-106 and by the AP-2 mutant (mcluc-mAP2-23) were compared in the six cell lines. As seen in Fig. 5A, mutation of this AP-2 site increased luciferase expression 2.5-5-fold in all four melanoma cell lines indicating that this AP-2 site acts as a negative element in basal expression. The AP-2 mutation FIG. 2. Definition of the MCAM core promoter region The indicated reporter constructs were transiently transfected as described under "Experimental Procedures." Results are shown for LN215, Mel888, and WM664 cells. Results are presented as percent luciferase activity relative to the activity of the full-length construct mcluc-912 which was set at 100%. The results are presented as the mean and standard deviation of three independent experiments. had no effect on reporter activity in the carcinoma cell line Colo320 or in the glioma cell line LN215.
Gel retardation assays were performed using an oligonucleotide containing the MCAM specific AP-2-23 binding site. This probe reacted with nuclear extracts from all the melanoma cell lines (Fig. 5, B, lanes 1, 6, 7 and 10). This binding is specific as it was abrogated by an excess of unlabeled oligonucleotide (lane In light of the observation that mutation of the AP-2-binding site at Ϫ23 bp led to an increase in reporter gene activity, the putative AP-2-binding sites at ϩ14, Ϫ131, and Ϫ302 were also individually mutated and examined for activity in Mel JuSo. As shown in Fig. 5D, mutation of the AP-2-binding site at ϩ14 did not significantly affect the activity of the MCAM promoter, while mutation of the sites at Ϫ131 or Ϫ302 reduced promoter activity (by 70 and 44%, respectively).
Mutations in the AP-2-binding sites led to changes in transcriptional activity only in the melanoma cells. The fact that AP-2 binding was also only observed in these cells raises the question of whether or not AP-2 is expressed in the glioma and carcinoma cell lines. In order to determine this, all of the cell lines were examined for AP-2 expression using RT-PCR and Western blotting.
As shown in Fig. 6A, AP-2 ␣ mRNA expression was observed in all cells except the carcinoma cell line Colo320 (lane 6). Compared with the melanoma cells, the glioma cells appear to express a lower level of AP-2 mRNA. RT-PCR analyses for AP-2B mRNA, a dominant-negative acting splice variant, indicated that it is expressed in all cells except the carcinoma cell line Colo320 (Fig. 6D). Expression of AP-2 protein was analyzed by Western blot. 30 g of nuclear protein were analyzed from each cell and the position of the AP-2 band was determined from the migration of recombinant AP-2 (Fig. 6B, lane  1). Incubation of an identical blot with control serum indicated that the smaller band is not AP-2 related (data not shown). AP-2 protein was observed in all four melanoma cell lines (lanes 3-6) but was undetectable in the glioma cell line LN215 (lane 7) and in the carcinoma cell line Colo320 (lane 2). The observed lack of AP-2 binding in LN215 and Colo320 can therefore be explained by the absence of detectable AP-2 protein. gression to metastatic disease. In light of the evidence that MCAM may actually play a role in tumor growth or metastasis formation in vivo (11)(12)(13), an understanding of how expression of this molecule is regulated is of great importance. In this study transient transfection, site-directed mutagenesis, and transcription factor binding were used to define the major regulatory elements present in the 5Ј region of the human MCAM gene. The MCAM core promoter was shown to comprise the region from Ϫ106 bp to ϩ22 bp containing 4 Sp1 sites, 1 cAMP response element, and two AP-2 sites near the initiator motif containing the transcriptional start (14). Using oligonucleotide competition experiments Karlen and Braathen (16) could demonstrate only Sp1 binding to the MCAM promoter and concluded that this transcription factor was sufficient to drive constitutive MCAM expression in melanomas. Confirmation that Sp1 is an important regulator of MCAM expression is seen here in the results of deletion analysis where removal of all putative Sp1 sites reduced the promoter activity by 80%. However, site-directed mutagenesis also revealed important regulatory roles for the CRE element at Ϫ32 and for several of the putative AP-2-binding sites in the majority of cells studied, and CREB-1 and AP-2 were shown to specifically bind to the MCAM promoter in these cell lines. Mutation of the CRE element which was shown to bind CREB-1, reduced promoter activity by 70% in 5 of the 6 cells examined. A role for CRE in regulating constitutive MCAM expression was suggested by Xie et al. (19) who observed that transfection of the melanoma cell line MeWo with a dominant negative CREB mutant led to down-regulation of MCAM expression. The CRE element does not play a crucial role in MCAM expression in all cells inasmuch as mutation of this element did not effect basal MCAM transcriptional activity in the melanoma cell line WM2664. MCAM expression can be up-regulated by exposure to elevated cAMP levels (9), suggesting that this CRE element may also mediate inducible MCAM expression. A dual role of the CRE in promoting basal and inducible transcription has been described for other genes (20,21).
Site-directed mutation of the AP-2 sites at ϩ14, Ϫ23, Ϫ131, and Ϫ302 bp suggested that this transcription factor influences MCAM promoter activity in a complex manner. While mutation of the site at ϩ14 bp had no effect on promoter activity, mutation at Ϫ23 bp increased reporter activity up to 5-fold and mutations at Ϫ131 and Ϫ302 bp decreased reporter activity by 70 and 55%, respectively, in melanoma cell lines. The activating role of the AP-2 site at Ϫ302 bp is also apparent in the analysis of the deletion constructs as deletion of the region between Ϫ341 and Ϫ258 bp leads to a reduction in transcriptional activity only in AP-2 positive cells (data not shown). The reduction in reporter activity between mcluc-173 and mcluc-106 is observed in all cells suggesting that positively and negatively acting elements are present in this region. Functional AP-2-binding sites mediating transcriptional activation are frequently found upstream of the core promoter region (22,23) while AP-2 sites localized within the core promoter have been shown to mediate repression in a variety of genes (24,25). EMSAs and supershift analyses indicated specific binding of AP-2 complexes (predominantly AP-2 ␣ homodimers) to the element at Ϫ23 bp. The AP-2 mutations had no influence on reporter activity assessed in the glioma or carcinoma cells lines and no AP-2 binding complexes were observed in either cell. Both cells were shown by Western blot and RT-PCR analysis to lack AP-2 expression The use of a larger panel of MCAM positive and MCAM negative cell lines in the present study has led to conclusions which differ in several respects from previously reported studies. Although Karlan and Braathan (16) did not show that the condi-  tions used for their EMSAs were permissive for AP2 and CREB binding, their failure to observe binding of these factors could also be due to the cell line (Sk Mel 2) studied. Our examination indicates that the CRE-and AP2-binding sites do not contribute to basal MCAM expression in all cell lines. Previous reports also indicated that the activity of the 900-bp MCAM 5Ј region as assessed by transient transfection of CAT reporter constructs correlates with MCAM expression (15,16). In both of these studies, the melanoma cell line SB-2 was the sole MCAM negative cell examined. In the present study of 3 MCAM positive and 3 MCAM negative cell lines, no correlation could be observed between reporter activity driven by the 5Ј region (Ϫ912 bp or deletion constructs) and the MCAM expression status of the cell lines. The reporter activity was similar (10 3 -10 4 over vector alone) in all lines tested. Although the activity of this region was in fact lowest in the SB-2 cell line, it was highest in the melanoma cell line Me888 which is also MCAM negative (Fig. 1B). These results suggest that the expression of MCAM is not solely dependent on the constellation of trans-acting factors binding to the 900-bp MCAM 5Ј region studied. The elements relevant for repression of MCAM expression in MCAM negative cells may be located elsewhere or need to be embedded in chromatin to accurately reflect MCAM expression.
Jean et al. (15) reported an inverse correlation between AP-2 expression and MCAM expression in melanoma cell lines and have proposed that the presence of this transcription factor represses MCAM expression. While the results presented here support a negative effect of AP-2 binding at Ϫ23 bp on MCAM promoter activity, all AP-2 expressing cells examined (Mel JuSo, WM-2664, SB2, and Me888) did nevertheless show strong activity of the full-length MCAM 5Ј region. This may reflect a counterbalance of the more upstream AP-2 sites, two of which have been shown to have an activating effect on MCAM expression. Furthermore, Mel JuSo and WM-2664 express high levels of MCAM mRNA and protein and are tumorigenic in nude mice (15) 2 ; despite the presence of significant levels of AP-2 protein. That the AP-2 present in the MCAM positive cells is functional is evident from gel retardation experiments and from the altered reporter gene expression observed when the AP-2 sites at Ϫ23, Ϫ131, or Ϫ302 bp are mutated. These results do not rule out a role for AP-2 in the regulation of MCAM expression but clearly indicate that AP-2 by itself cannot explain the constitutive MCAM expression in melanoma cell lines. Furthermore, they indicate that expression of MCAM by melanoma cells does not require the loss or down-regulation of the transcription factor AP-2.