c-Myb-binding sites mediate G(1)/S-associated repression of the plasma membrane Ca(2+)-ATPase-1 promoter.

We demonstrate that two Myb-binding sites of the mouse plasma membrane Ca(2+)-ATPase-1 (PMCA1) promoter are required for G(1)/S cell cycle stage-associated repression of PMCA1 promoter activity. Nuclear run-on experiments revealed G(1)/S-associated repression of PMCA1 transcription. Ribonuclease protection assays revealed two transcription initiation sites between two Myb-binding sites in the PMCA1 promoter. Gel shift assays showed that c-Myb can bind to wild-type but not point mutated Myb binding sequences of the PMCA1 promoter. Transient transfection assays using cell cycle-synchronized vascular smooth muscle cells (VSMC) and PMCA1 promoter-luciferase constructs showed a 2-fold decrease in reporter activity at G(1)/S as compared with G(0). Overexpression of wild-type c-Myb severely repressed PMCA1 promoter activity at both G(0) and G(1)/S while co-transfection of a dominant negative c-Myb, or a construct encoding an anti-c-Myb neutralizing antibody, completely abolished the repression seen at G(1)/S. Single nucleotide substitutions in the first, second, or both Myb-binding sites alleviated the G(1)/S-associated repression of PMCA1 promoter activity in transformed rat VSMC and primary mouse VSMC cultures. We conclude that c-Myb mediates G(1)/S-associated transcriptional repression of the PMCA1 Ca(2+) pump in rodent VSMC by direct binding to the PMCA1 promoter.

We demonstrate that two Myb-binding sites of the mouse plasma membrane Ca 2؉ -ATPase-1 (PMCA1) promoter are required for G 1 /S cell cycle stage-associated repression of PMCA1 promoter activity. Nuclear run-on experiments revealed G 1 /S-associated repression of PMCA1 transcription. Ribonuclease protection assays revealed two transcription initiation sites between two Myb-binding sites in the PMCA1 promoter. Gel shift assays showed that c-Myb can bind to wild-type but not point mutated Myb binding sequences of the PMCA1 promoter. Transient transfection assays using cell cyclesynchronized vascular smooth muscle cells (VSMC) and PMCA1 promoter-luciferase constructs showed a 2-fold decrease in reporter activity at G 1 /S as compared with G 0 . Overexpression of wild-type c-Myb severely repressed PMCA1 promoter activity at both G 0 and G 1 /S while co-transfection of a dominant negative c-Myb, or a construct encoding an anti-c-Myb neutralizing antibody, completely abolished the repression seen at G 1 /S. Single nucleotide substitutions in the first, second, or both Myb-binding sites alleviated the G 1 /S-associated repression of PMCA1 promoter activity in transformed rat VSMC and primary mouse VSMC cultures. We conclude that c-Myb mediates G 1 /S-associated transcriptional repression of the PMCA1 Ca 2؉ pump in rodent VSMC by direct binding to the PMCA1 promoter.
The myb family of transcription factors is a polyphyletic group whose members possess a conserved DNA-binding domain with a helix-turn-helix like motif (recently termed the "Myb box" (1)). Myb box proteins carry out a variety of functions including positive and negative transcriptional regulation, modulation of mRNA stability (2), and the regulation of telomere length (3). They have diversified enormously in plants where they represent the largest known regulatory gene family (4). In vertebrates, only three family members (A-, B-, and c-myb) are known, of which c-myb and its encoded gene product c-Myb have been well characterized. The c-Myb DNA-binding domain recognizes the consensus hexanucleotide sequence (C/ T)AAC(G/T)G. c-Myb also contains transactivation and C-terminal negative regulatory domains. c-Myb expression is vitally important for the control of cell proliferation in a variety of cell types, and c-Myb is also involved in the regulation of apoptosis (reviewed in Ref. 5). Vascular smooth muscle cells (VSMC) 1 express c-Myb at the late G 1 phase of the cell cycle (6,7).
VSMC are dynamic structural and functional components of blood vessel walls. As contractile cells they are capable of effecting vascular tone. However, as proliferative cells they participate in vaso-occlusive processes such as atherosclerosis and post-angioplasty restenosis (8). In studying the mechanisms regulating VSMC proliferation, we have focused on the role played by c-Myb in the regulation of intracellular Ca 2ϩ concentrations ([Ca 2ϩ ] i ) at the G 1 /S cell cycle interface. We have shown that c-Myb activity regulates [Ca 2ϩ ] i in both VSMC and fibroblasts (9 -11). Experiments employing either wild-type or dominant negative forms of c-Myb, in which cell cycle-associated Ca 2ϩ homeostasis was monitored, implicated the plasma membrane Ca 2ϩ -ATPase (PMCA) family of Ca 2ϩ efflux pumps as critical mediators of c-Myb-dependent [Ca 2ϩ ] i (9,10).
The PMCAs are high affinity, low capacity, Ca 2ϩ pumps that extrude cytosolic Ca 2ϩ to the extracellular space (12)(13)(14). Each of the 4 known PMCA genes (PMCA1-4) gives rise to multiple isoforms through alternative splicing (14,15). These genes differ in their 5Ј-untranslated regions (14,16) which may contain regulatory sequences that direct tissue-specific expression. PMCA1 and PMCA4 are the ubiquitously expressed members of this 4-gene family (17,18) and Western blot analysis has shown PMCA1 expression to exceed that of PMCA4 (19).
It is known that Ca 2ϩ transporting membrane proteins play a major role in modulating Ca 2ϩ signaling (20). Indeed, the resting [Ca 2ϩ ] i in many cell types is critically regulated by the level of PMCA activity (9,21,22). However, despite the obvious importance of PMCAs to cellular Ca 2ϩ handling, little is known of the mechanisms regulating their expression. A study on rat endothelial cells suggested that PMCA1 is regulated at the transcriptional level via a protein kinase C-dependent pathway (23). Our studies on VSMC showed that expression levels of PMCA1 are inversely correlated with levels of c-Myb activity, such that suppression of c-Myb activity led to 2-4-fold increases in the level of PMCA1 mRNA and protein (10). These manipulations resulted in a 30% reduction in mean resting [Ca 2ϩ ] i , a 42% decrease in mean S phase entry, and a 36% decrease in the mean cell proliferation rate of synchronized rat VSMC populations (10). We also demonstrated that a direct 2-fold overexpression of PMCA1, independent of any manipulation of c-Myb activity, resulted in similarly significant reduc-tions in [Ca 2ϩ ] i , G 1 to S transitions, and rate of cell proliferation (9). Having demonstrated the physiological importance of PMCA1 regulation in VSMC, we now examine more closely the molecular interaction between c-Myb and the PMCA1 gene.
Du et al. (24) have previously cloned and sequenced 1010 bp of the mouse PMCA1 promoter. In their series of primer-extension and promoter-reporter assays carried out in mouse neural cells, they found the PMCA1 promoter to lack a TATA box, possess numerous SP1 sites, one long GC repeat within a CpG island, and an untranslated first exon that was at least 19 kilobases away from the translational start within exon 2. We have re-cloned and sequenced a 975-bp fragment of the above murine PMCA1 promoter and have defined two VSMC-specific transcription initiation sites located between a pair of Mybbinding sites: Myb-binding site-1 and site-2 at positions ϩ440 and ϩ528, respectively (numbering is relative to the neural cell-specific transcriptional start site). We used ribonuclease protection assays, gel shift, and luciferase reporter assays to delineate the function of these two Myb-binding sites in the mouse PMCA1 promoter. Our experiments involved restriction digested deletants, single substitution point mutants in one or both Myb-binding sites, and effector constructs expressing either c-Myb, a dominant negative form of c-Myb, or an anti-Myb neutralizing antibody. Our studies show that the repression of the murine PMCA1 gene at the G 1 /S interface of rodent VSMC requires both Myb-binding sites flanking the two transcriptional initiation sites.

MATERIALS AND METHODS
DNA Constructs-The reporter plasmid pPM1-luc and its mutants are described below. p3090, the full-length murine c-Myb expression construct, was a kind gift of Dr. Michael Kuehl, Bethesda, MD (25). The ⌬5-myb, a dominant negative c-Myb mutant lacking amino acid residues 109 to 185 which constitute the major portion of the c-Myb DNAbinding domain, has been described elsewhere (10). psFV23, a mammalian expression plasmid encoding a single chain neutralizing antibody raised against the transactivation domain of c-Myb, was a generous gift of Dr. D. T. Curiel and Dr. K. Kasono, Birmingham, AL (26).
Cell Culture and Cell Cycle Synchronization-An optimal number (differing according to the size of the culture vessel) of SVE cells (ATCC number CRL-2018) were seeded and allowed to attach overnight in Dulbecco's modified Eagle's medium, 10% fetal bovine serum, and 200 g/ml G418 at 37°C and 5% CO 2 . Cells were washed twice with phosphate-buffered saline and incubated in Dulbecco's modified Eagle's medium containing 0.25% fetal bovine serum for 48 h at which point the cells were either harvested (G 0 stage cells), or the medium was replaced with Dulbecco's modified Eagle's medium and 10% fetal bovine serum for 8 (G 1 stage cells), 16 (G 1 /S stage cells), or 24 h (G 2 /M stage cells) prior to harvest. The degree and extent of cell cycle synchronization, as assessed by DNA quantitative flow-cytometry, was identical to previously published reports (data not shown) (6,9,10). Primary VSMC were isolated by the method of Cornwell and Lincoln (27) using 10 -12 aortas from C57Bl6 mice (Charles River Laboratories, Inc., Wilmington, MA). They were then cultured and synchronized for cell cycle stages as described above.
Nuclear Run-on-Run-on assays were carried out according to Nevins (28). Nuclei were isolated from cell cycle synchronized cultures at the G 0 and G 1 /S stages by employing 0.3% Nonidet P-40 for cell lysis. Nascent RNA was labeled for 10 min at 37°C, extracted with the RNeasy kit (Qiagen), DNased, and re-extracted with RNeasy before being hybridized to 5 g of immobilized PMCA1 cDNA probe (either single or double stranded) at 65°C for 2-3 days, washed and autoradiographed. Band intensities were quantitated via NIH Image software. Band intensities were corrected for filter background and normalized to intensities of Actin bands.
Mouse PMCA1 Promoter-Reporter Constructs-PCR primers were designed using the 1010-bp mouse PMCA1 promoter sequence (Gen-Bank accession number U16707) and the public domain software Primer 3.0. 2 The forward PCR primer sequence corresponded to numbers 29 -50 and the reverse PCR primer to nucleotide numbers 964 -985 in GenBank accession number U16707. Both primers had a T m of 75°C. The mouse PMCA1 promoter was amplified using these primers and 200 ng of genomic DNA from C57Bl6 mice. The PCR reaction used 10 pmol of each primer, 200 M of each dNTP, and 0.5 units of Taq DNA Polymerase (Qiagen) in a 25-l volume. Hot start PCR was employed as follows: 95°C for 5 min (Pre-PCR); 35 cycles of 95°C for 1 min, 65°C for 1 min, and 72°C for 1 min; and a final extension of 72°C for 10 min.
The 975-bp PCR product was gel eluted and blunt ligated into SmaIcut pGL3-Basic (Promega) to generate the plasmid pPM1-luc. The pPM1-luc insert was sequenced on an automated sequencer (ABI Prism Model 377) using vector-specific primers Glprimer2 and Rvprimer3 (Promega) as well as an insert based primer (corresponding to position 389 -410 in GenBank accession number U16707) and contig alignment was done using Sequencher 4.0 software (Gene Codes Corp.). This sequence has been deposited in the GenBank data base under accession number AF162783. 3Ј-and/or 5Ј-ends of the pPM1-luc insert were deleted via restriction endonuclease digestions, blunted, and self-ligated to generate various deletion mutants. p⌬-5ЈElement lacks the AatII-NheI fragment spanning Ϫ418 to Ϫ68, p⌬-myb2 lacks the Myb site-2 containing BstEII-BglII fragment spanning ϩ488 to ϩ557, and p⌬-myb1ϩ2 lacks the PmlI-BglII fragment spanning ϩ301 to ϩ557 which contains both Myb sites-1 and -2. All deletions were confirmed by multiple restriction digestions and sizing of digested products on agarose gels.
Site-directed Mutagenesis-Site-directed mutagenesis of one or both Myb-binding sites in the mouse PMCA1 promoter was carried out as described (29) with minor modifications (30). Briefly, uridylated single stranded pPM1-luc DNA (noncoding strand) was annealed slowly over a period of 5 h in a thermal cycler with either oligo ⌬M1-AS (introduces an A to G substitution in Myb-binding site-1 at ϩ440) or oligo ⌬M2-AS (introduces an A to G substitution in Myb-binding site-2 at ϩ528) or with both primers simultaneously; see Table I for primer sequences. Primer extension of the annealed templates in the presence of T4 DNA polymerase and T4 DNA ligase yielded uridylated-nonuridylated relaxed, circular hybrids which were transformed into Escherichia coli DH5␣MCR competent cells and four transformants from each mutagenesis reaction were screened via sequencing. A small restriction fragment (PmlI-BglII fragment spanning ϩ301 to ϩ557) bearing the mutated Myb-binding site(s) was cut out of the screened mutants and was used to replace the identical restriction fragment in the wild-type pPM1-luc construct to minimize chance mutations at sites other than the Myb-binding site. This generated three point mutants, viz. pPt-myb1 (Myb site-1 mutated), pPt-myb2 (Myb site-2 mutated), and pPt-myb1ϩ2 (both Myb sites mutated). The region spanning ϩ301 to ϩ557 from all three point mutants was sequenced in an automated sequencer (ABI Prism 377) and the mutant sequences were aligned with the wild-type sequence (GenBank accession number AF162783) using Sequencher 4.0 software (Gene Codes Corp.) to confirm point mutations.
Ribonuclease Protection Assays-A stably transformed mouse VSMC line was constructed by isolating mouse aortic smooth muscle cells from 10 C57Bl mice as described (27). Isolated cells were stained with smooth muscle ␣-actin antibody (Sigma; data not shown) to confirm that only smooth muscle cells were present. These primary cultures were immortalized with a retrovirus carrying the SV40 large T antigen and the G418 resistance marker as described (31) and selected for 14 days with 400 g/ml G418. Total RNA was isolated from T-75 flasks containing cell cycle synchronized G 1 stage mouse VSMC (5 h postserum stimulation) by using the RNeasy kit (Qiagen) and DNase treated. Three different length probes were generated (see Fig. 1A). The reverse PCR primer used to PCR clone the mouse PMCA1 promoter fragment (positions 964 to 985 in GenBank accession number U16707) was annealed to AatII-digested pPM1-luc DNA and extended with Klenow DNA polymerase in the presence of [␣-32 P]dCTP to generate a prematurely terminated, uniformly labeled single stranded DNA probe: Probe 1 spanned ϩ437 to ϩ557. Probes 2 and 3 were generated by using a recombinant version of Klenow enzyme lacking both 3Ј to 5Ј and 5Ј to 3Ј exonuclease activities (MBI Fermentas): Probe 2 arose by primer extension of the reverse PCR primer from BamHI-digested pPM1-luc (ϩ182 to ϩ557) and Probe 3 by extension of the antisense primer EMSA M1 (see Table I) after annealing to AatII-digested pPM1-luc (Ϫ63 to ϩ460). All three probes were resolved on 5% denaturing polyacrylamide gels along with end-labeled GeneRuler 1-kilobase ladder (MBI Fermentas), gel purified, and quantitated by measuring Cerenkov counts. 10 5 cpm of labeled probe was annealed with 1.25, 2.5, 5.0, and 10 g of total RNA overnight at 42°C and digested with a 200-fold dilution of the nuclease mixture according to the manufacturer's instructions (Ambion). Protected RNA products were resolved on a 5% denaturing polyacrylamide gel with dideoxy sequencing reactions carried out with 2 Copyright S. Rozen and H. J. Skaletsky. either the reverse PCR primer and pPM1-luc or antisense EMSA M1 and pPM1-luc to help size the RNA products.
Transient Transfection Promoter-Luciferase Assay-Approximately 0.5 ϫ 10 6 rat SVE cells or mouse primary VSMC were seeded per well in 6-well dishes and allowed to attach overnight. Cells were washed twice with phosphate-buffered saline and transfected with a standard combination of plasmids (unless otherwise stated) which consisted of 1 g of reporter plasmid (pPM1-luc or one of its mutants), 50 ng of reference reporter plasmid (pRL-TK, carrying the renilla luciferase gene under the thymidine kinase promoter; Promega), and 1 g of effector plasmid (where applicable) with the help of 3 l of Lipo-fectAMINE (Life Technologies, Inc.) in a 1-ml overlay of Opti-MEM (Life Technologies, Inc.) serum-free medium according to the manufacturer's instructions (5 h incubation at 37°C). The overlay was removed, the cells washed twice with phosphate-buffered saline and then allowed to recover overnight in complete medium. Cells were serum starved as described above and different transfected cell populations were stimulated with serum for 0 (G 0 stage) or 16 h (G 1 /S stage) and harvested by scraping in 250 l of Passive Lysis Buffer (Promega). Cell lysates were subjected to 2 freeze thaws and 20 l (10 l in case of primary cultures) was used to measure firefly and Renilla luciferase activity in a luminometer (Lumat LB 9501; EG&G Berthold) as per a commercially available kit protocol (Dual Luciferase Assay System; Promega).
Raw relative luminescence units were corrected for auto-luminescence of the firefly and Renilla luciferase substrates (detected in mock transfected cells which were transfected with pGem7 plasmid) and normalized for transfection efficiency by dividing by the corrected Renilla relative luminescence unit values for each sample. With both rat SVE cells as well as mouse primary VSMC, mean normalized (f/r) relative luminescence unit values for the wild-type promoter-reporter construct (pPM1-luc) at the G 0 stage (mean of at least two experiments) was set as a reference value. This reference value was then used to derive ratios for individual G 0 and later stage samples (sample f/r value divided by reference value). Ratios (relative normalized relative luminescence units) derived in this way from two or more experiments were used to compute mean Ϯ S.E. In a few experiments, where a parallel comparison with the wild-type pPM1-luc reporters activity was not performed, the G 0 values of the tested constructs were used as the reference value.
Gel Mobility Shift Assays-G 1 /S stage SVE cells, as well as asynchronous cultures of a human leukemia cell line known to overexpress c-Myb (K562: ATCC number CCL-243), were used to prepare nuclear extracts as described elsewhere (32). Slight modifications to the protocol included lysis of cells with 0.2-0.5% Nonidet P-40 after swelling in Buffer A (see Dignam et al. (32)) and 10 -20 passages through a 21gauge needle as well as Dounce homogenization.
Oligos (EMSA-M1, EMSA-M2, EMSA Pt-M1, EMSA Pt-M2; see Table I for details) corresponding to wild-type and single point mutant Myb-binding site sequences of mouse PMCA1 promoter (GenBank accession number AF162783) were synthesized. 8 pmol each of sense and antisense EMSA-M1 oligo and sense and antisense EMSA-M2 oligo were end-labeled with [␥-32 P]ATP and polynucleotide kinase, annealed to form double stranded oligos EMSA-M1 and EMSA-M2, resolved on a 15% native polyacrylamide gel, excised out of the gel, and purified on Sep-Pak C 18 columns (Waters) as described (30).
Nuclear extracts were mixed with binding buffer and poly(dI-dC) on ice prior to addition of 100-fold molar excess of unlabeled double stranded oligo corresponding to either wild-type or point mutant Mybbinding sites from the murine PMCA1 promoter. After incubating this mixture for 15 min at 25°C, 0.5 ng of purified end-labeled double stranded EMSA-M1 or EMSA-M2 oligo was added and incubation continued at 25°C for an additional 15 min. 2 l of 10 ϫ gel shift loading dye was added and the DNA-protein complexes were resolved on a 5% native Tris glycine gel run at 30 mA for 1 h at 4°C (33). The gels were dried and autoradiographed.
Statistical Analysis-Luciferase assay data are shown as mean Ϯ S.E. and represent results from at least two separate experiments. Student's t test was used to make pairwise comparisons between results. Statistical significance was defined as p Յ 0.05.

RESULTS
PMCA1 Transcription-We previously showed in immortalized rat VSMC that steady state mRNA and protein levels of PMCA1 were decreased by approximately 50% at the G 1 /S cell cycle stage as compared with at G 0 (9). In the present study we performed numerous nuclear run-on assays in the same cell type to determine whether this 2-fold regulation was occurring at the transcriptional level. These attempts showed either no change in the rate of new PMCA1 message transcription from G 0 to G 1 /S or a slight transcriptional repression of 30% in some experiments (mean G 0 to G 1 /S ratio of 1 (Ϯ0.07): 0.75 (Ϯ0.05)). Why some experiments showed no reduction in PMCA1 transcription may be explained by the poor sensitivity of nuclear run-on assays for detecting small (i.e. 2-fold) changes (34,35). While the overall observed reduction in the rate of PMCA1 transcription at G 1 /S as compared with G 0 was small, it remains possible that this decrease is sufficient to account for the 2-fold lowering of steady state mRNA levels that is known to occur. Accordingly, we proceeded to examine the structure and function of the PMCA1 gene promoter in the context of cell cycle progression in rodent VSMC.
Wild-type Mouse PMCA1 Promoter Sequence-A 975-bp region of the mouse PMCA1 promoter, spanning position Ϫ418 to ϩ557 (numbering based on the most 5Ј transcriptional start site mapped in mouse neural cells; see Fig. 1A) was amplified from C57Bl6 mouse genomic DNA and sequenced. Alignment of opposing strand sequences generated via vector-specific primers (see "Materials and Methods") showed only 5 nucleotide differences. These were clustered in the region from ϩ78 to ϩ138 (nucleotides 496 to 556 in the 975-bp insert) and likely arose from gel resolution ambiguities in the longest chain terminated products derived from vector-specific primers. Indeed, 4 of these differences were resolved by sequence generated from insert-specific primers (see "Materials and Methods") and the last was resolved by comparison with the sequence reported by Du et al. (24) (GenBank accession number U16707).
By contrast, when the mouse PMCA1 promoter sequence deduced in our lab (GenBank accession number AF162783) was aligned with the sequence reported earlier (GenBank accession number U16707), there were 54 positions in which the two sequences differed and most of these were clustered in the middle of the insert. In our opinion, these could have resulted from a combination of causes, viz. differences in the mouse strains used to clone each sequence and gel resolution ambiguities in the longest chain terminated products characterized by Du et al. (24).
Mapping the Site of Transcription Initiation-Ribonuclease protection assays were used to definitively map the transcripts arising from the mouse PMCA1 gene in a stably transformed mouse VSMC line. As shown in Fig. 1B, when a 121-nucleotide probe covering positions ϩ437 to ϩ557 is used, two main transcripts are protected which map to ϩ475 and ϩ485, just downstream of Myb site-1 (ϩ440). Similarly, when another probe spanning ϩ182 to ϩ557 is used, the same two transcripts (initiated at ϩ475 and ϩ485) are protected (data not shown). No transcripts are protected when a probe spanning Ϫ63 to ϩ460 is used.
In mouse neuroblastoma cells (24), four PMCA1 transcription start sites occur (ϩ1, ϩ27, ϩ47, and ϩ64) and these appear to be specific to neural tissue. Our ribonuclease protection data show those neural-specific transcription initiation sites to be absent in PMCA1 transcripts of VSMC; instead, two smooth muscle-specific transcription initiation sites were found 475 and 485 bases downstream of the most 5Ј neural-specific start site. The authenticity of these two transcription initiation sites is further bolstered by additional evidence presented below: deletion of the region between the two Myb sites (as in the deletant (p⌬-myb1ϩ2)) leads to a 60% drop in transcription levels from the PMCA1 promoter.
Transient Transfection Luciferase Assays-We next employed transient transfection of promoter-luciferase expression constructs to evaluate the importance of the two Myb-binding sites and the core promoter element in the functioning of the c-Myb Represses PMCA1 Gene Activity at G 1 /S mouse PMCA1 promoter during cell cycle progression in rodent VSMC. Analysis of our mouse PMCA1 promoter sequence (GenBank accession number AF162783) with public domain software (MatInspector V2.2 (36)) revealed two putative Mybbinding sites in opposite orientations and only 82 bp apart (Fig.  1). Myb site-1 is located at ϩ440 to ϩ445 and Myb site-2 at ϩ528 to ϩ533. Based on this analysis, three promoter deletion mutants were generated: p⌬myb2 with a deletion spanning ϩ488 to ϩ557 lacks Myb binding site-2; p⌬myb1ϩ2 with a deletion from ϩ301 to ϩ557 is missing both Myb-binding sites; and p⌬5Јelement which, having lost Ϫ418 to Ϫ68, lacks a core PMCA1 promoter previously defined in neural cells (see below) but retains 68 bp of upstream sequence, all four neural-specific transcriptional start sites, and all downstream sequences in-cluding both Myb-binding sites (see Figs. 1 and 2). With regards to the latter construct (p⌬5Јelement), Du et al. (24) had defined the region Ϫ256 to Ϫ117 (nucleotide positions as in GenBank accession number U16707) as a core promoter element responsible for ϳ66% of PMCA1 transcription in murine neuroblastoma cells.
As a single A to G substitution in the consensus Myb-binding site has been shown in gel-shift assays to prevent the binding of purified c-Myb (37, 38), we created point mutants of the mouse PMCA1 promoter in which one or both Myb-binding sites had undergone single A to G substitutions: pPt-myb1 carries a single point mutation in Myb binding site-1 (ϩ440); pPt-myb2 carries a single point mutation in Myb binding site-2 (ϩ528); and pPt-myb1ϩ2 carries the same single substitutions in both Myb-binding sites (see Fig. 4 and Table I).
As detailed under "Materials and Methods," the mean G 0 luciferase activity of the wild-type promoter-luciferase construct was used to normalize values of all other variants of this promoter at both G 0 and G 1 /S cell cycle stages. Of note, the G 0 value itself for the wild-type promoter was always within 10% of the mean value (Figs. 2-4). The wild-type PMCA1 promoterreporter construct, pPM1-luc, showed a 50% reduction in luciferase activity at G 1 /S as compared with G 0 (G 0 versus G 1 /S ϭ 0.99 Ϯ 0.09 versus 0.46 Ϯ 0.01; p ϭ 0.03; Fig. 2). Deletion of a core promoter element defined in neural cells (p⌬-5Јelement) did not change the repression seen at G 1 /S (p⌬5ЈElement G 1 / S ϭ 0.41 Ϯ 0.07; p ϭ 0.04; Fig. 2). Furthermore, persistent luciferase activity in the absence of the core promoter element defined in neural cells argues for the presence of additional promoter elements active in VSMC (p⌬5ЈElement G 0 versus pPM1-luc G 0 ϭ 0.87 Ϯ 0.17 versus 0.99 Ϯ 0.09; p ϭ NS; Fig. 2). Importantly, deletion of the restriction fragment bearing Myb site-2 (p⌬-myb2) completely abrogated the G 1 /S stage repression of PMCA1 promoter activity (p⌬-myb2 G 0 versus G 1 /S ϭ 1.40 Ϯ 0.32 versus 1.09 Ϯ 0.19; p ϭ NS; Fig. 2). This result supported the importance of Myb binding site-2 in bringing about G 1 /S-associated repression of PMCA1. However, deletion of the restriction fragment bearing both Myb site-1 and Myb site-2 (p⌬-myb1ϩ2; lacking the region from ϩ301 to ϩ557) produced a 2-fold down-regulation of PMCA1 promoter-dependent luciferase activity at the G 0 stage which persisted even at G 1 /S (pPM1-luc G 0 versus p⌬-myb1ϩ2 G 0 ϭ 0.99 Ϯ 0.09 versus 0.39 Ϯ 0.03; p ϭ 0.02; Fig. 2). This finding supported the presence of a VSMC-specific promoter element active in VSMC in the region spanning ϩ301 to ϩ557. The two PMCA1 transcripts mapped from mouse VSMC are also located within this deleted region (at ϩ475 and ϩ485). The alternative possibility that Myb site-1 is involved in promoting basal levels of PMCA1 expression was refuted by using point mutants as described below.
Co-transfections with either wild-type or dominant negative c-Myb expression constructs were next used to increase or decrease functional c-Myb levels in VSMC transfected with the wild-type PMCA1 promoter-reporter construct (Fig. 3). c-Myb over-expression reduced G 0 stage PMCA1 promoter activity by 60% and G 1 /S stage PMCA1 transcription by 75% as compared with the G 0 stage cells transfected with pPM1-luc alone (pPM1luc G 0 versus pPM1-luc with Myb overexpression G 1 /S ϭ 0.99 Ϯ 0.09 versus 0.19 Ϯ 0.18; p ϭ 0.05; Fig. 3). Expression of the dominant negative c-Myb mutant completely relieved the G 1 /S stage repression of luciferase activity in a dose dependent manner, with higher doses of the mutant producing greater derepression. A mammalian expression construct encoding an anti-c-Myb single chain antibody has recently been shown to inhibit the transactivation activity of c-Myb (26). When this anti-c-Myb antibody construct was co-transfected into SVE FIG. 1. A, structure of the mouse PMCA1 promoter. The 975-bp insert (GenBank accession number AF162783) is shown. The positions of the mouse neural tissue-specific transcription start site (ϩ1) and two vascular smooth muscle-specific transcription start sites (ϩ475 and ϩ485) are shown. Also shown are key restriction endonuclease sites, two Myb-binding sites, SP1 and CAAT box sites, as well as the positions and lengths of single stranded DNA probes employed in ribonuclease protection assays to map transcription initiation sites. B, mapping transcription initiation sites with ribonuclease protection assays. Total RNA from the G 1 stage of a stably transformed mouse VSMC line was hybridized with a uniformly labeled, single stranded DNA probe (probe 1 shown in Fig. 1A), digested with a mixture of nucleases and resolved on a 5% denaturing polyacrylamide gel with a sequencing ladder generated by using the same primer which was used to generate the probe (see "Materials and Methods" for details). Lane 1, undigested, intact probe; lanes 2-5, 1.25, 2.5, 5.0, and 10 g of total RNA hybridized to 10 5 cpm of probe, respectively. The coding strand sequence corresponding to the two transcription start sites is shown with boxes around the first nucleotide of each transcript. c-Myb Represses PMCA1 Gene Activity at G 1 /S cells with the wild-type PMCA1 promoter, it caused a marked de-repression of PMCA1 promoter-dependent luciferase activity at the G 1 /S stage. Taken together, these results strongly support a role for c-Myb as a functionally important repressor of PMCA1 gene activity.
We then employed single nucleotide substitutions in either one or both Myb-binding sites of the PMCA1 promoter to dissect out the contribution of these sites to net PMCA1 promoter activity in the SVE cell line (Fig. 4). Point mutation of single nucleotides in Myb site-1, Myb site-2, or both Myb sites-1 and -2, alleviated the G 1 /S stage repression of luciferase activity seen in the wild-type promoter. While the luciferase activities of these point mutants at G 0 appear slightly higher than the G 0 activity of the wild-type promoter, these differences were not statistically significant. Furthermore, the fact that pPt-myb1 did not reduce the basal level activity of the PMCA1 promoter as did p⌬-myb1ϩ2, suggests that Myb binding site-1 has no role in promoting basal level PMCA1 expression. These data further suggest the presence of a VSMC-specific promoter ele-ment in the deleted region of p⌬-myb1ϩ2.
To test the functional specificity of our point mutants, we co-transfected pPt-myb1ϩ2 with the anti-Myb encoding construct. Reduction in functional c-Myb activity had no effect on reporter levels which confirmed that the double point mutant had lost c-Myb responsiveness (data not shown). Finally, the alleviation of G 1 /S-associated PMCA1 repression was also observed when the point mutants were tested in primary cultures of mouse VSMC (data not shown).
Gel Shift Studies-In order to determine whether putative Myb-binding sites-1 and -2 of the murine PMCA1 promoter can actually bind c-Myb protein, we next employed the gel shift assay (Fig. 5). Nuclear extracts from a human leukemic cell line (K562 cells) known to express high levels of c-Myb (33) were used for gel shift assays with end-labeled Myb site-1 or Myb site-2 double stranded oligonucleotides. Increasing amounts of the nuclear extract led to the formation of higher amounts of c-Myb⅐DNA complexes. Excess unlabeled Myb site-1 oligo successfully competed with labeled Myb site-1 DNA for binding to c-Myb while excess unlabeled mutant Myb site-1 oligo bearing a single point mutation did not (Fig. 5A). Myb site-2 double stranded oligo showed similar results. An excess of unlabeled Myb site-2 oligo could out-compete labeled Myb site-2 oligo in binding to c-Myb while an excess unlabeled mutant Myb site-2 oligo could not (Fig. 5B). We also carried out gel shift assays using nuclear extracts made from rat SVE cells at the G 1 /S stage. Although these experiments always yielded results identical to the ones obtained with extracts from the human cell line, the intensities of the shifted bands were always weaker (data not shown). This finding is consistent with the markedly lower amounts of c-Myb in rat VSMC as compared with K562 cells, and argues against nonspecific binding of non-Myb proteins to the target oligos.

DISCUSSION
Results of numerous studies have revealed that the c-Myb transcription factor regulates [Ca 2ϩ ] i during the G 0 to S phase cell cycle progression of VSMC and fibroblasts (6, 7, 9 -11, 39). Overexpression of wild-type c-Myb has been shown to increase [Ca 2ϩ ] i at both the G 0 and G 1 /S cell cycle stages as compared with control transfected cells (10), while the use of dominant negative c-Myb mutants demonstrated that reductions in c-Myb activity abolished the normal rise in both resting and stored [Ca 2ϩ ] i as cells moved to the G 1 /S transition (10). This latter effect was due to a marked increase in the Na e ϩ -independent Ca 2ϩ efflux rate, and was associated with important reductions in the rate of S-phase entry and cell proliferation

5Ј-CCCGGAGCTGGCCGcTGCCCCTTGAGCT-3Ј
3Ј-GGGCCTCGACCGGCgACGGGGAACTCGA-5Ј FIG. 2. Transient transfection luciferase assays with wild-type and deleted PMCA1 promoters. The 975-bp mouse PMCA1 promoter was ligated upstream of the firefly luciferase cDNA to make the construct labeled pPM1-luc. The luciferase vector without any upstream promoter was termed "Promoterless." ⌬-myb2 has been deleted of a restriction fragment spanning ϩ488 to ϩ557 and lacks Myb site-2. ⌬-myb1ϩ2 lacks a fragment spanning ϩ301 to ϩ557 which contained both Myb binding sites-1 and -2. The ⌬-5Јelement lacks a promoter region known to be active in murine neural cells (24). The constructs were used to transiently transfect rat VSMC with luciferase activities being measured at G 0 and G 1 /S stages. Values were corrected for background luminescence as well as transfection efficiency and normalized to the mean luciferase activity of the wild-type promoter at the G 0 stage (see "Materials and Methods"). Data shown are the mean Ϯ S.E. of at least two experiments c-Myb Represses PMCA1 Gene Activity at G 1 /S (9). It was subsequently shown that c-Myb activity levels are inversely related to mRNA and protein levels of PMCA1 (9), which normally show a ϳ50% reduction at G 1 /S as compared with G 0 . Overexpression of PMCA1, independent of any manipulations in c-Myb activity, also resulted in increased rates of Ca 2ϩ efflux and significant reductions in [Ca 2ϩ ] i , G 1 /S progression and proliferation (9). Thus, abolition of the normal repression of endogenous PMCA1 expression at G 1 /S, by either dominant negative c-Myb constructs or overexpression of a transfected PMCA1, prevented the normal fall in the Ca 2ϩ efflux rate and disabled the accumulation of intracellular Ca 2ϩ (9,10). Together, the above data led to our definition of the PMCA1 Ca 2ϩ efflux pump as an end-effector of the c-Myb-dependent elevation in [Ca 2ϩ ] i critically required for G 1 /S transitions in VSMC.
Previous investigations into PMCA1 gene regulation have suggested the existence of agonist-and tissue-specific signaling pathways (23,24,40). Phorbol ester or angiotensin II treatments of non-synchronized rat aortic endothelial cell cultures were shown to produce protein kinase C-dependent increases in PMCA1 mRNA (8 -20-fold) and protein (3-4-fold) within 4 -6 h of agonist treatment (23). A protein kinase A-dependent pathway was implicated in both cAMP-and thapsigargin-induced increases in PMCA1 expression in non-synchronized endothelial cells derived from resistance vessels of the rat brain (40). Of interest, rat aortic endothelial cells did not show increases in PMCA1 levels in response to these latter agents (40). Non-synchronized mouse neuroblastoma cells also responded to a 4-h treatment with phorbol ester by exhibiting a 5-fold increase in steady state PMCA1 mRNA (24). Importantly, the regulation of this response appeared to depend on transcriptional activation of the PMCA1 gene via a core promoter segment (Ϫ442 to ϩ169) of the murine PMCA1 gene (24). When cell cycle-synchronized rat VSMC are studied, the early stages of G 1 progression (i.e. 0.5, 1, 2, and 4 h post-serum stimulation) also show significantly increased levels of PMCA1 mRNA as compared with either G 0 or G 1 /S levels (9). Of interest, these elevations in PMCA1 expression are coincident with the expression of early-response genes and with known mitogenmediated elevations in [Ca 2ϩ ] i (41). The above studies suggest that a variety of signal transduction pathways, early response genes, and/or transient increases in [Ca 2ϩ ] i during the early part of G 0 to G 1 progression may act to increase PMCA1 expression. However, the subsequent down-regulation of PMCA1 Single nucleotide substitutions were made in either the first, second, or both Myb-binding sites of the mouse PMCA1 promoter ligated upstream of the firefly luciferase cDNA to produce three point mutants, viz. Pt-myb1, Pt-myb2, and Pt-myb1ϩ2, respectively. These point mutants were used to transfect rat VSMC and luciferase activity was measured at the G 0 and G 1 /S stages. Data shown have been corrected for background luminescence and transfection efficiency and normalized to the mean luciferase activity of the wild-type promoter at the G 0 stage. Data shown are the mean Ϯ S.E. of at least two experiments.
c-Myb Represses PMCA1 Gene Activity at G 1 /S levels and PMCA1-mediated Ca 2ϩ efflux rates at the G 1 /S interface (9) requires an opposing effect on PMCA1 expression at this point in the cell cycle. The present study has elucidated a molecular mechanism through which c-Myb mediates this opposing effect.
While it has recently been shown that c-Myb activity can decrease the half-life of thrombospondin 2 mRNA in mouse fibroblasts, the mechanism through which c-Myb mediated this effect was not elucidated (2). Bein et al. (2) have speculated that c-Myb may transactivate the expression of a labile ribonuclease which in turn could account for the enhanced degradation of thrombospondin 2 mRNA. Although a c-Myb-dependent reduction in the stability of PMCA1 mRNA could conceivably contribute to the down-regulation of steady state PMCA1 mRNA levels at the G 1 /S interface, the findings of the present study suggest a transcriptional mechanism of PMCA1 regulation. Our nuclear run-on data show at least a 25% decrease in the rate of PMCA1 transcription at G 1 /S as compared with G 0 (data not shown). As it is difficult to document 2-fold differences in transcription rates with nuclear run-on assays, the above result is likely an underestimate of the reduction in PMCA1 transcription.
Using MatInspector V2.2 (36), our analysis of the human PMCA1 gene sequence reveals two putative c-Myb-binding sites at position ϩ988 and ϩ1182. Indeed, a comparison of the human and murine PMCA1 promoter regions has revealed several other similarities as well (16). Both genes share features such as a large first intron (20 -30 kilobases), a translational start in exon 2, and now two putative Myb-binding sites downstream of transcriptional start. Of interest, comparisons between pig and human PMCA1 cDNA sequences also show a high degree of sequence homology (85-90%) (16). While an apparently significant mismatch at the exon 1-exon 2 interface is due to an exon in pig PMCA1 cDNA being absent from the human sequence, the missing human exon (termed exon 1*) is actually located in intron 1 of the human gene (16). Hilfiker et al. (16) have speculated that some vertebrate tissues may express PMCA1 with exon 1* as the first exon instead of exon 1 by virtue of an additional downstream promoter. Indeed, multiple SP1-binding sites and a consensus CCAAT box are present upstream of exon 1* in intron 1 of the human PMCA1 gene (16).
Although the 5Ј regulatory region of the murine PMCA1 gene does have certain characteristics typical of housekeeping genes, such as the absence of a TATA box, multiple SP1binding sites, and a high frequency of CpG dinucleotides (42), our data point to further complexity within this region. An earlier study (24) of the PMCA1 promoter in mouse neural cells showed the presence of a neural-specific core promoter element (inactive in our VSMC line) while our data point to an additional promoter active in VSMC further downstream. More importantly, when the region spanning ϩ301 to ϩ557 was deleted (deletant p⌬myb1ϩ2), we found a 60% reduction in luciferase activity at both G 0 and G 1 /S when compared with the wild-type promoter which implies the loss of a major promoter element. This deletant lacks both Myb-binding sites and should not have been down-regulated at the G 1 /S stage via Mybmediated repression. Also, this deletant's down-regulation at G 0, when c-Myb levels are low, further suggests loss of a promoter element. Indeed, MatInspector analysis confirms that the region deleted from p⌬myb1ϩ2 contains putative binding sites for several transcription factors including SP1, AP1, AP4, MyoD, MZF1, CREB, and a GC box element. Hence the lowering of luciferase activities at G 0 and G 1 /S in this deletant is most likely due to the loss of a major cis-acting regulatory module promoting transcription initiation at ϩ475 and ϩ485. Deletion of the smooth muscle-specific promoter from p⌬myb1ϩ2 does not lead to a complete loss of transcriptional activity in this mutant and we speculate that this may be due to the presence of a cryptic promoter element in the remaining 5Ј-flanking region within this mutant construct. There are many reports in the literature documenting the activation of cryptic promoters either upstream (43)(44)(45)(46)(47) or downstream (48 -52) of the major native promoter once this major promoter has been deleted or inactivated. Thus our data, together with those reported earlier (24), suggest the presence of two promoters in the mouse PMCA1 gene: one element (spanning Ϫ256 to Ϫ118) active in neural cells and the other (between ϩ301 to ϩ485) active in VSMC.
The preceding discussion has shown that the presence of multiple promoter elements with distinct transcriptional start sites is possible in the human PMCA1 gene and highly probable in the murine gene. This is analogous to the recent realization from transgenic mouse studies that multiple, independent, cisregulatory modules are required to direct the developmental expression of muscle-specific genes. More specifically, a gene may have one promoter for skeletal muscle expression, another for cardiac expression, and one or more for expression in smooth muscle cells of various types (53). The tissue-specific activation of each promoter element would then be accomplished by unique sets of transcription factors. Indeed, this type of combinatorial regulation of gene expression appears to be a general strategy for tissue-specific expression in other genes as well (54,55).
Several genes involved in cell proliferation are known to be transcriptionally activated by c-Myb. These include c-myc (56,57), cdc2 (58), topoisomerase II␣ (59), DNA polymerase ␣ (60), c-kit (61), and the gene for proliferating cell nuclear antigen (PCNA) (62). In addition, c-Myb is also known to repress genes such as c-erbB-2 (63), the mouse N-ras promoter as tested in avian fibroblasts (64), the monocytic gene MRP14 in HL-60 cells (65), the 5-lipoxygenase gene during HL-60 differentiation (66), the c-kit gene in non c-Kit-expressing cells (67), and the promoter function of both human and mouse c-fms genes (68). Of the above six genes known to be negatively regulated by c-Myb, four are cytoplasmic proteins with roles in cell-cell c-Myb Represses PMCA1 Gene Activity at G 1 /S interactions or cell differentiation, while two are proto-oncogenes encoding transmembrane proteins involved in signal transduction (c-erbB2 and N-ras).
The present study highlights similar consequences of c-Myb activity in rodent VSMC, i.e. control of a transmembrane protein involved in cell proliferation via transcriptional repression. Our studies conclude that the PMCA1 pump is a physiologically important c-Myb-responsive cell proliferation gene over which c-Myb mediates cell cycle-specific negative regulation through specific c-Myb binding sites in the PMCA1 promoter. The c-Myb-mediated down-regulation of PMCA1 levels allows sustained elevations in resting and releasable [Ca 2ϩ ] i at the G 1 /S interface (9), which in turn facilitate G 1 to S cell cycle transitions through a variety of putative Ca 2ϩ -responsive effectors (9,41).