Genomic Cloning and Promoter Analysis of Aortic Preferentially Expressed Gene-1

Aortic preferentially expressed gene-1 (APEG-1) was originally identified as a 1.4-kilobase (kb) transcript preferentially expressed in differentiated vascular smooth muscle cells (VSMC). Its expression is markedly down-regulated in de-differentiated VSMC, suggesting a role for APEG-1 in VSMC differentiation. We have now determined that APEG-1 is a single-copy gene in the human, rat, and mouse genomes and have mapped human APEG-1 to chromosome 2q34. To study the molecular mechanisms regulating its expression, we characterized the genomic organization and promoter of mouse APEG-1. APEG-1 spans 4.5 kb in the mouse genome and is composed of five exons. Using reporter gene transfection analysis, we found that a 2.7-kb APEG-15′-flanking sequence directed a high level of promoter activity only in VSMC. Its activity was minimal in five other cell types. A repressor region located within an upstream 685-base pair sequence suppressed the activity of this 2.7-kb promoter. Further deletion and mutation analyses identified an E box motif as a positive regulatory element, which was bound by nuclear protein prepared from VSMC. In conjunction with its flanking sequence, this E box motif confers VSMC-specific enhancer activity to a heterologous SV40 promoter. To our knowledge, this is the first demonstration of an E box motif that mediates gene expression restricted to VSMC.

The de-differentiation of vascular smooth muscle cells (VSMC) 1 from a quiescent and contractile phenotype to a proliferative and synthetic phenotype is one of the most prominent features of arteriosclerosis, the leading cause of death in developed countries (1). These phenotypic changes may result from vascular injuries caused by smoking, hypercholesterolemia, hy-perhomocysteinemia, hypertension, or trauma (1)(2)(3)(4)(5)(6). Owing to, at least in part, a lack of markers specific for differentiated VSMC, the molecular mechanisms regulating VSMC differentiation are largely unknown (2).
Although several gene products have been used as specific markers for differentiated smooth muscle cells (SMC), such as smooth muscle ␣-actin, smooth muscle myosin heavy chain, calponin, SM22␣, and caldesmon (7)(8)(9)(10)(11)(12), their expression in vivo is not restricted to VSMC. However, a 0.4-kb segment of the SM22␣ promoter, which contains two CArG elements, has been shown recently to confer expression only in the arterial SMC of transgenic mice (13,14). Because mutation of the proximal CArG element eliminates all SM22␣ promoter activity in transgenic animals, this element appears to be necessary and sufficient for high-level expression restricted to the SMC lineage (13). The CArG element binds to nuclear proteins such as serum response factor and YY1. Serum response factor and YY1 are both expressed ubiquitously; thus it is unclear how the CArG element regulates arterial SMC-specific expression conferred by the 0.4-kb segment of the SM22␣ promoter. This puzzle could be explained by the presence of arterial SMCspecific transacting factors or co-factors that have yet to be identified.
Aortic preferentially expressed gene-1 (APEG-1) was cloned in our laboratory by virtue of its preferential expression in VSMC (15). It is a 1.4-kb message and encodes a 12.7-kDa protein (15). APEG-1 is expressed in differentiated VSMC in vivo and is down-regulated rapidly in de-differentiated VSMC in vitro and in injured arteries in vivo (15). These data suggest that APEG-1 may serve as a sensitive marker for VSMC differentiation, and that it may play a role in regulating growth and differentiation in this cell type.
In this report we show that APEG-1 is a single-copy gene. The APEG-1 transcription unit spans 4.5 kb in the mouse genome and contains five exons. Using reporter gene transfection analysis, we found that 2.7 kb of the APEG-1 5Ј-flanking sequence contains potent VSMC-specific promoter activity. Further deletion and mutation analyses identified an E box motif as a positive regulatory element, which was bound by nuclear protein prepared from VSMC. In conjunction with its flanking sequence, this E box motif confers VSMC-specific enhancer activity to a heterologous promoter.

EXPERIMENTAL PROCEDURES
Cell Culture and Reagents-Rat aortic SMC (RASMC) were harvested from the thoracic aortas of adult male Sprague-Dawley rats (200 -250 g) by enzymatic digestion (16). Bovine aortic endothelial cells were harvested from bovine aortic endothelium as described (17). U-2 OS cells were kindly provided by Dr. T.-P. Yao (Dana-Farber Cancer * This work was supported by the Bristol-Myers Squibb Pharmaceutical Research Institute and National Institutes of Health Grants F32-HL10113 (to M. D. L.), KO8-HL03194 (to M. A. P.), and RO1-GM53249 (to M.-E. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We dedicate this work to the late Edgar Haber in gratitude for his enthusiasm and support of our research.
Southern Blot Analysis and Chromosomal Localization of APEG-1-Genomic DNA was extracted from cultured mouse embryonic stem cells, RASMC, and human umbilical vein endothelial cells. Genomic DNA (10 g) was digested with 50 units of the restriction enzyme XbaI, EcoRI, BamHI, HindIII, or SacI, separated on a 0.8% agarose gel, denatured, and transferred to NitroPure filters (MSI, Westboro, MA) by using a standard protocol (19). The filters were hybridized with [␣-32 P]dCTP-labeled APEG-1 cDNA probes from the appropriate species (the cDNA probes contain exons 2 to 5). After hybridization, filters were washed at 55°C in 0.2 ϫ SSC (30 mM sodium chloride, 3 mM sodium citrate), 0.1% sodium dodecyl sulfate and exposed to Kodak BioMax films with intensifying screens. To locate APEG-1 on a specific human chromosome, we hybridized a genomic Southern blot membrane containing an EcoRI-digested human monochromosomal somatic cell hybrid panel (BIOS Laboratories, New Haven, CT) with a human APEG-1 cDNA probe under the same hybridization and washing conditions.
Genomic Library Screening and Sequence Analysis-A 129/SvJ mouse genomic library in FIX II vector (Stratagene, La Jolla, CA) was screened with a mouse APEG-1 cDNA probe. Four positive clones were purified and subcloned into pUC 18 (Promega, Madison, WI) or pBlue-Script II SK (Stratagene) plasmid vectors. The genomic DNA and exonintron junctions were sequenced by the dideoxy chain termination method with primers designed from the mouse APEG-1 cDNA. Intron sizes were determined by polymerase chain reaction (PCR) with flanking exon primers, and by direct sequencing. The DNA sequences were assembled and analyzed with Sequencher 3.0 software (Gene Codes Co., Ann Arbor, MI) running on a PowerMacintosh (Apple Computer, Cupertino, CA).
RNase Protection Assay-A 305-bp mouse APEG-1 fragment (Ϫ122 to ϩ183) was amplified by PCR from genomic DNA using the forward primer 5Ј-CGTTCgagcTCCACCACTCCAGGG-3Ј (the lowercase sequence indicates a SacI site introduced for cloning purposes) and the reverse primer 5Ј-GAGGCTTTGCACACGGAC-3Ј and subcloned into the pCR-Script vector (Stratagene). After linearization this construct was used to produce an [␣-32 P]UTP-labeled antisense riboprobe with T3 RNA polymerase. Gel-purified antisense riboprobe was hybridized overnight with 20 g of total RNA from mouse aorta, heart, or skeletal muscle. Yeast tRNA was also hybridized as a negative control. RNase A (0.1 unit) and RNase T1 (4 units) were added subsequently to digest single-strand RNA, and the remaining riboprobe was separated on an 8% sequencing gel along with a sequencing ladder using the linearized template and reverse primer. The gel was exposed to Kodak X-OMAT film overnight with an intensifying screen at Ϫ80°C.
5Ј RACE (Rapid Amplification of cDNA Ends) Experiment-A 5Ј RACE system (Life Technologies, Inc., Manassas, VA) was used according to the protocol provided by the manufacturer. The nested primers were R592, 5Ј-CCATATTCGTTGACCGCC-3Ј for reverse transcription; R354, 5Ј-TTGGGGGTGCCTTGGAAGAAGAGTC-3Ј for the initial PCR; and R309, 5Ј-TGGGACTGAGCTTCATGGTAGGGGTTCGG-3Ј for the nested PCR. One microgram of total RNA from mouse aorta was used. Chloramphenicol acetyltransferase (CAT) RNA and primers provided by the manufacturer were used as positive controls. The 5Ј RACE product was cloned into the pCR-II vector (Invitrogen, Carlsbad, CA) for sequence analysis.
Transient Transfection and Reporter Activity Assays-p(Ϫ2663/ϩ76) and pGL3-Control (equal moles) were used to transfect growing cells on 60-mm dishes with DEAE-dextran for RASMC (20) or LipofectAMINE (Life Technologies, Inc.) for other cell types. To correct for variability in transfection efficiency, we co-transfected 1 g of pCAT3-Control in all experiments. Luciferase and CAT activities were measured 48 -60 h after transfection as described (20,21). To compare APEG-1 promoter activity in different cell types, we expressed the normalized activity of p(Ϫ2663/ϩ76) as a percentage of that of pGL3-Control. In the promoter deletion and mutation analyses, the normalized activity of each construct was expressed as a percentage of the activity of p(Ϫ2663/ϩ76). All data are presented as the mean Ϯ S.E.
Nuclear Protein Extraction and Electrophoretic Mobility Shift Assay-RASMC and NIH-3T3 nuclear extracts were prepared from cells grown on 150-mm dishes essentially as described (22). C2C12 cells were grown in 10% fetal bovine serum then kept in 2% horse serum for 3 days to induce differentiation into myotubes. Protein concentration was measured by the Bradford dye-binding method (23) with the Bio-Rad protein assay system (Bio-Rad). For the electrophoretic mobility shift assay, oligonucleotide probes were synthesized according to the mouse APEG-1 exon 1 sequences E (5Ј-GGGCCTCAGCTGGGTCAG-3Ј) and Emut (5Ј-GGGCCTCAGCacGGTCAG-3Ј). An 18-bp E box-containing probe was also synthesized according to the mouse muscle creatine kinase (MCK) enhancer sequence (5Ј-CCCCACACCTGCTGCCT-3Ј). An unrelated Oct-1 sequence (5Ј-TTATGCAAAATAATAAAACGTATT-3Ј) was made as a nonspecific competitor. Double-stranded oligonucleotide (50 pmol) was end-labeled with [␥-32 P]ATP by using polynucleotide kinase (New England Biolabs, Beverly, MA) and purified on a Sephadex G-25 column (Roche Molecular Biochemicals, Indianapolis, IN). A typical binding reaction consisted of 8 g of nuclear extract, DNA probe (20,000 cpm), 250 ng of poly(dI-dC)⅐poly(dI-dC), 25 mM HEPES (pH 7.9), 40 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol, and 10% glycerol. A molar excess (100-fold) of unlabeled, double-stranded oligonucleotide was added to the reaction to compete for DNA binding. For mobility supershift experiments, 4 g of an anti-E2A protein (E12 and E47) monoclonal antibody (Yae antibody, Santa Cruz Biotechnology, Santa Cruz, CA) was included in the binding reaction and incubated on ice for 20 min before the probe was added. Binding reactions were incubated on ice for 15 min and resolved by 5% nondenaturing polyacrylamide gel electrophoresis in 0.5 ϫ TBE buffer (44.5 mM Tris base, 44.5 mM boric acid, 1 mM EDTA) at 4°C.

APEG-1 Is a Single-copy Gene and Is Located on Human
Chromosome 2q-We performed Southern analysis with a fulllength APEG-1 cDNA probe to determine whether APEG-1 is a single-copy gene. Hybridization of EcoRI-, BamHI-, HindIII-, or SacI-digested genomic DNA from human, rat, and mouse revealed a simple pattern that suggests that APEG-1 is a singlecopy gene in all three species (Fig. 1A).
To test the possibility that APEG-1 associates with certain genetic disorders that have been mapped, we determined the chromosomal location of the human APEG-1 gene. Southern hybridization to an EcoRI-digested human/rodent monochromosomal somatic cell hybrid panel revealed that APEG-1 is located on human chromosome 2 (Fig. 1B). This result was confirmed by using an independent human/rodent somatic cell hybrid panel (data not shown). To determine the subchromosomal localization of human APEG-1, we carried out a genomic PCR analysis by using the GeneBridge 4 radiation hybrid panel with specific primers from the human APEG-1 cDNA sequence (24). The PCR results were then screened against a sequencetagged site data base for the human genome (25). APEG-1 mapped to human chromosome 2q34 between markers D2S360 and D2S353 (data not shown). Despite the preferential expression of APEG-1 in VSMC, a search of the human genome data base did not reveal an association with known inherited vascular diseases or with a particular gene cluster (26).
Genomic Cloning and Organization of APEG-1-Using a fulllength mouse APEG-1 cDNA probe, we identified several clones from a 129/SvJ mouse genomic library. These APEG-1 genomic sequences assembled into a 4.5-kb organization that contained five exons and four introns (Fig. 2, A and B). The open reading frame for the APEG-1 protein began in the second exon and terminated at the 5Ј-end of the fifth exon (Fig. 2B). The identified exon-intron junctions (Fig. 2B) were all in agreement with the consensus 5Ј GT and 3Ј AG sequences (27).
Identification of the APEG-1 Transcription Start Site-To determine the APEG-1 transcription start site, we performed RNase protection assays. A 305-bp (Ϫ122 to ϩ183) APEG-1 antisense riboprobe encompassing the 5Ј-untranscribed region and most of the first exon was hybridized with RNA from mouse aorta, brain, heart, and undifferentiated and differentiated Monc-1 cells (Monc-1 cells are transformed neural crest cells that can be differentiated into VSMC) (18). After RNase digestion (Fig. 3A), three partially protected riboprobe signals of between 100 and 200 nucleotides were observed primarily in the mouse aorta and to a lesser extent in the heart and the differentiated Monc-1 cells (28 and 57% compared with aorta, respectively). Very little signal was seen in the brain (12%) and the undifferentiated Monc-1 cells (11%). The most 5Ј transcription start site (Fig. 3A, long arrow) for mouse APEG-1 mapped to a CA nucleotide pair (Fig. 3A, bent arrow), which is found at many eukaryotic transcription start sites (28). Two shorter fragments (Fig. 3A, arrowheads) corresponded to downstream CA pairs and may also be transcription start sites. In addition, a fully protected riboprobe (minus sequence from the vector) was found in the aorta, brain, heart, and differentiated Monc-1 cells (Fig. 3A, asterisk). This fully protected probe was generated by a 4-kb APEG-1 isoform. 2 To confirm the transcription start site of APEG-1, we performed a 5Ј RACE experiment. We amplified and cloned a 363-bp cDNA fragment from mouse aortic RNA (Fig. 3B). After we had removed sequences introduced by the 5Ј RACE procedure, this 363-bp product contained a 309-bp APEG-1 cDNA sequence whose 5Ј end (Fig. 3C, bent arrow) was the same as the primary transcription start site (Fig. 3A, long arrow) identified by the RNase protection assays.

APEG-1 Promoter Region Contains a CArG Motif Specific to Smooth Muscle Gene Expression-To identify potential cis-act-
ing elements that may be important in the regulation of APEG-1 promoter activity, we cloned and sequenced the 3.3-kb 5Ј-flanking region (Fig. 4). There are no upstream TATA-like sequences in the APEG-1 5Ј-flanking sequence; however, there is a consensus initiator motif (5Ј-YYCAYYYYY-3Ј) at the APEG-1 transcription start site (Figs. 3A and 4, bent arrow) (29,30). We identified several cis-acting elements (by their consensus sequences) that have the potential to regulate APEG-1 promoter activity (Fig. 4). Among them are two CArG (or CArG-like) elements known to regulate expression of SMCspecific genes (13,(31)(32)(33). In addition, we identified two Sp1 sites in the proximal promoter region (bp Ϫ52 and bp Ϫ143 relative to the transcription start site) that may potentiate transcription from the TATA-less APEG-1 promoter (34).
The 2.7-kb APEG-1 Promoter Is Inhibited by a 5Ј Repressor-We reported elsewhere that expression of APEG-1 is down-regulated in de-differentiated VSMC both in vivo and in vitro (15). Thus although we had expected VSMC-specific APEG-1 promoter activity, we were surprised that 2.7 kb of the APEG-1 5Ј-flanking region directed high levels of promoter activity in cultured, and therefore de-differentiated, RASMC.

FIG. 3. Analysis of mouse APEG-1 transcription start site by RNase protection assay and 5 RACE.
A, 25 g of total RNA extracted from mouse aorta, brain, heart, and undifferentiated and differentiated (5 days) Monc-1 cells was hybridized to an [␣-32 P]UTP-labeled antisense riboprobe as described under "Experimental Procedures." The transcription start site was identified by running a sequencing ladder with the reverse primer (ϩ182) on the same gel. Long arrow marks the largest fragment protected by APEG-1 mRNA, which indicates the transcription start site. It corresponds to an A nucleotide preceded by a C nucleotide (right). Two arrowheads indicate shorter fragments that correspond to downstream CA nucleotide pairs. Asterisk indicates a fully protected probe that does not contain the vector sequence. It is the result of a riboprobe hybridizing with a 4-kb APEG-1 isoform whose transcript shares genomic sequence with APEG-1 (see text). An RNA size marker is included to indicate the length of the nucleotide sequence. Riboprobe hybridized with yeast tRNA was used as a negative control, and the lane without RNase treatment (no RNase) shows the input of riboprobe in the hybridization (2.5 ϫ 10 5 cpm/reaction). One additional probe lane (Probe) (2.5 ϫ 10 4 cpm) was loaded to indicate the size of the riboprobe in the absence of hybridization and RNase treatment. The signal intensity of the primary transcription start site is shown as a percentage of the intensity of the aorta signal. B, 5Ј RACE was used to locate the 5Ј-end of mouse APEG-1. A 363-bp fragment was obtained from mouse aortic RNA and cloned for sequencing. C, 5Ј-end sequence of the 363-bp product is shown on the right. The poly(G) sequence was introduced during the 5Ј RACE experiment. The 5Ј-end of mouse APEG-1 is indicated by an arrow.
One explanation for this discrepancy would be the presence of negative DNA regulatory elements outside the 2.7-kb APEG-1 5Ј-flanking sequence. To test this possibility, we constructed p(Ϫ3336/ϩ76) and p(Ϫ3336/ϩ76)Rev by cloning an additional 685 bp of APEG-1 5Ј-flanking sequence into p(Ϫ2663/ϩ76) in both orientations. In comparison with that of p(Ϫ2663/ϩ76), the promoter activity of p(Ϫ3336/ϩ76) and p(Ϫ3336/ϩ76)Rev was reduced markedly (Fig. 6). An additional (upstream) 4-kb DNA sequence did not decrease promoter activity further (data not shown). Taken together, these results indicate that an orientation-independent transcription repressor is located between bp Ϫ3336 and Ϫ2663 5Ј of the APEG-1 transcription start site. This APEG-1 transcriptional repressor may explain, at least in part, the decrease in expression of APEG-1 in dedifferentiated VSMC (15).
An E Box Motif Mediates High Levels of APEG-1 Promoter Activity in RASMC-To identify the cis-acting element in the APEG-1 promoter responsible for its potent activity in RASMC, we made three deletion constructs from the 5Ј end of p(Ϫ2663/ ϩ76). In transient transfection experiments, the three deletion constructs, p(Ϫ1073/ϩ76), p(Ϫ479/ϩ76), and p(Ϫ355/ϩ76), had promoter activity similar to that of p(Ϫ2663/ϩ76). A fourth deletion construct, p(Ϫ122/ϩ76), showed a 20% reduction in promoter activity (Fig. 7). These data suggest that most of the APEG-1 promoter activity is contained between bp Ϫ122 and ϩ76.
To further localize the positive cis-acting element, we generated a series of 3Ј deletion constructs based on p(Ϫ479/ϩ76) and p(Ϫ122/ϩ76). These allowed us to determine whether the presence of the 76-bp exon 1 sequence was important for promoter activity. In comparison with p(Ϫ479/ϩ76), p(Ϫ479/ϩ38), and p(Ϫ122/ϩ38) both had much lower promoter activity (16 and 4%, respectively) (Fig. 7). These results indicated that the sequence between bp ϩ38 and ϩ76 in exon 1 was essential for APEG-1 promoter activity. When we inspected the sequence between bp ϩ38 and ϩ76 for potential transcription factorbinding sites, we identified an E box motif (CAGCTG) at bp ϩ39 to ϩ44 (Fig. 4). To determine the importance of this E box motif, we mutated its sequence from CAGCTG to CAGCAC in p(Ϫ479/ϩ76) and p(Ϫ122/ϩ76). As demonstrated by transfection experiments with p(Ϫ479/ϩ76)Emut and p(Ϫ122/ ϩ76)Emut, mutation of the exon 1 E box motif caused a dramatic reduction in APEG-1 promoter activity (Fig. 7). These data indicate that this E box motif located at the 5Ј-untrans- lated region (5Ј-UTR) is essential for high level APEG-1 promoter activity in RASMC.
Although not commonly found, transcription regulatory elements have been documented to locate to the 5Ј-UTR of a few other genes (35)(36)(37). For instance, the 5Ј-UTR of the herpes simplex virus type 1 ICP22 gene and the human integrin ␤3 gene contain cis-acting elements that mediate high level expression of these genes (36,37). Furthermore, the human A ␥globin gene also has regulatory elements in the 5Ј-UTR. One of these elements binds to the erythroid transcription factor GATA-1 and may regulate transcription of the human A ␥globin gene during development (35).
It is noteworthy that one CArG box and one CArG-like box are located at bp Ϫ1531 to Ϫ1522 and bp Ϫ443 to Ϫ434 of the APEG-1 5Ј-flanking region, respectively (Fig. 4). The CArG box is crucial to the expression of several other SMC-specific genes (13,(31)(32)(33), although there is no known SMC-specific, CArG box-binding protein. In the case of the APEG-1 promoter, however, deletion of the CArG and CArG-like boxes did not alter its activity (Fig. 7), indicating that the two boxes are dispensable. This dispensability distinguishes APEG-1 from other SMCspecific genes and suggests the existence of CArG-independent mechanisms of SMC-specific gene expression. Indeed, we have shown that the CArG-less promoter of mouse CRP2/SmLIM directs a high level of VSMC-specific reporter gene expression in transgenic mice (38).
The E Box in APEG-1 Exon 1 Is Not Bound by E12 and E47 Proteins-Transcriptional regulation via the E box motif CANNTG is important for the regulation of myogenesis (39), immunoglobulin gene expression (40), cell development and differentiation (41)(42)(43)(44), and cell proliferation and apoptosis (45). It has been shown that the ubiquitously expressed E2A gene products E12 and E47 (46) heterodimerize with tissuespecific transcription factors to regulate gene expression via the E box motif (47), although an E47 or E12 heterodimerization partner has not been identified in VSMC. We therefore wanted to determine by electrophoretic mobility shift assay whether the APEG-1 exon-1 E box was bound by nuclear protein prepared from VSMC, and whether E12 and E47 were present in the binding complex. Incubation of a probe containing the APEG-1 E box (E) but not one containing a mutated E box (Emut) with RASMC nuclear extract resulted in a DNAprotein complex (Fig. 8A). This DNA-protein complex was specific because only an excessive amount (100-fold) of unlabeled E box oligonucleotide could compete with it for binding (Fig. 8A); the mutated E box oligonucleotide and an unrelated Oct-1 oligonucleotide could not compete with it.
To see if E47 or E12 was present in the DNA-protein complex, we performed mobility shift assays with the Yae monoclonal antibody specific to E47 and E12 (48). Addition of the Yae antibody to the binding reaction did not, however, cause a mobility supershift or a disappearance of the DNA-protein complex (Fig. 8B). As a positive control, we used another E box-containing mouse enhancer sequence, MCK, and nuclear proteins extracted from differentiated C2C12 myotubes (48). The Yae antibody was able to supershift a distinct binding complex in C2C12 myotubes which is presumably composed of heterodimeric E2A proteins and MyoD or myogenin (Fig. 8B). These results indicate that the E12 and E47 proteins are not present in the APEG-1 E box-binding protein complex.
We next tested the specificity of the APEG-1 E box-binding protein in two additional cell types and found that the same binding complex also existed in NIH-3T3 fibroblasts and C2C12 myotubes (Fig. 8C). This result and the deletion/mutation analyses of the APEG-1 promoter by transfection experiments would suggest that the E box is essential for APEG-1 FIG. 5. Cell type-specificity of mouse APEG-1 promoter activity. Mouse APEG-1 promoter construct p(Ϫ2663/ϩ76) and pGL3-Control were each transfected into the indicated cell types, and reporter luciferase activity was measured as a representation of promoter activity. The difference in transfection efficiency was corrected by the CAT activity from a co-transfected pCAT3-Control vector. For each cell type, the corrected p(Ϫ2663/ϩ76) promoter activity is shown as a percentage of that of pGL3-Control.
FIG. 6. A repressor region is located between bp ؊3336 and ؊2663 5 of the APEG-1 transcription start site. Two 3.3-kb mouse APEG-1 promoter constructs, p(Ϫ3336/ϩ76) and p(Ϫ3336/ϩ76)Rev, were made as described under "Experimental Procedures." The promoter activity of the two constructs in RASMC was only 20% of the activity of the 2.7-kb APEG-1 promoter construct p(Ϫ2663/ϩ76).

FIG. 7.
Deletion and mutation analysis of mouse APEG-1 promoter. Reporter plasmids were constructed from the APEG-1 promoter as described under "Experimental Procedures." The four 5Ј deletion constructs p(Ϫ1073/ϩ76), p(Ϫ479/ϩ76), p(Ϫ355/ϩ76), and p(Ϫ122/ ϩ76) were made from p(Ϫ2663/ϩ76). They show that most of the APEG-1 promoter activity is contained within p(Ϫ122/ϩ76). The two 3Ј deletion constructs p(Ϫ479/ϩ38) and p(Ϫ122/ϩ38) were made from p(Ϫ479/ϩ76) and p(Ϫ122/ϩ76). They show minimal promoter activity. The p(Ϫ479/ϩ76)Emut and p(Ϫ122/ϩ76)Emut constructs contain a 2-bp mutation that changes the E box motif in exon 1 from CAGCTG to CAGCAC. The diagram on the left shows the relative lengths of the constructs and the positions of the CArG boxes (white boxes) and the E box (black ovals). The E box mutation is indicated by the hatched ovals. Transfection experiments were repeated at least three times for each construct, and promoter activity is expressed as a percentage of p(Ϫ2663/ϩ76) activity.
promoter activity but not sufficient to confer SMC specificity. A similar scenario is found in CArG box-mediated SMC-specific gene expression (13,(31)(32)(33). By electrophoretic mobility shift assay, the CArG box is bound by serum response factor, whose expression is not cell type restricted. Thus, it appears that additional SMC-specific accessory factors may be required to interact with either the APEG-1 E box-binding protein or serum response factor to direct SMC specific transcription.
Although the APEG-1 E box and the MCK E box share the same MyoD-type E box (49), they formed distinct DNA-protein complexes when incubated with nuclear extracts of RASMC or C2C12 myotubes (Fig. 8, B and C). For example, incubation of RASMC nuclear extracts resulted in a more prominent DNAprotein complex with the APEG-1 E box than the MCK E box. Furthermore, an E2A-containing complex was present when C2C12 nuclear extracts were incubated with the MCK but not APEG-1 E box. These observations indicate that additional E box-flanking sequences are required to determine the binding preferences of different nuclear proteins in the same cell type, which may contribute to cell type-specific gene expression.

The APEG-1 E Box Motif and Its Flanking Sequences Confer
VSMC-specific Enhancer Activity to a Heterologous SV40 Promoter-We cloned (in both orientations) a 74-bp exon 1 sequence (ϩ2 to ϩ76) containing the APEG-1 E box upstream of the SV40 promoter in pGL3-Promoter. The resulting plasmids pGL3-E1Ebox and pGL3-E1Ebox.Rev were then transfected into several cell types. The SV40 enhancer in pGL3-Control (positive control) increased SV40 promoter activity in all cell types tested (Fig. 9). Only in RASMC, in contrast, did the APEG-1 E box and its flanking sequence (in either orientation) increase SV40 promoter activity markedly (Fig. 9). These findings indicate that the E box motif and its flanking sequence are both necessary and sufficient for a VSMC-specific enhancer.
To our knowledge, this work on the APEG-1 promoter is the first demonstration that an E box plays an important role in VSMC-specific gene transcription. Future studies to identify the trans-acting factors responsible for E box-dependent APEG-1 promoter activity and down-regulation of promoter activity by VSMC de-differentiation will contribute to our understanding of the molecular mechanisms regulating VSMCspecific gene expression and phenotypic modulation.
FIG. 8. Electrophoretic mobility shift assays. A, two 18-bp oligonucleotide probes derived from the mouse APEG-1 exon 1 sequence with an E box (E) or mutated E box (Emut) motif were end-labeled with [␥-32 P]ATP and incubated with or without RASMC (SM) nuclear extract (N.E.). Arrow indicates the mobility-shifted probe and its binding nuclear protein complex. Three competitors (E, Emut, and Oct-1) were added to the binding reaction at a 100-fold molar excess to demonstrate E box-binding specificity. B, monoclonal anti-E2A protein antibody (␣-E2A) was added to the binding reaction but was not able to change the mobility of the E box-binding complex in RASMC (SM) nuclear extract (N.E.). As a positive control, a mouse MCK enhancer E box (MCK E box) was used as probe to incubate with differentiated C2C12 myotube nuclear extract (C2) in the presence or absence of ␣-E2A. The MCK E box probe was bound weakly by RASMC nuclear extract. C, nuclear extracts (N.E.) from RAMSC (SM), C2C12 myotubes (C2), and NIH-3T3 cells (3T3) were incubated with the APEG-1 exon 1 E box to demonstrate that the E box-binding protein is present in all three cell types.
FIG. 9. Activation of transcription on a heterologous SV40 promoter by the APEG-1 E box motif and its flanking sequences. A 74-bp DNA fragment containing mouse APEG-1 E1 sequence from bp ϩ4 to ϩ76 was cloned into the SmaI site of pGL3-Promoter in the correct (pGL3-E1Ebox) and the reverse (pGL3-E1Ebox.Rev) orientations. The two constructs, along with pGL3-Promoter and pGL3-Control, were each transfected into the indicated cell types and their reporter luciferase activities were analyzed. The diagram on the left shows the constructs with the SV40 promoter (P) and an SV40 enhancer region (En). Arrowhead indicates orientation of the E box-containing mouse APEG-1 E1 sequence. Normalized luciferase activity is presented as a percentage of pGL3-Control activity.