Sequence of cDNAs Encoding Components of Vascular Actin Single-stranded DNA-binding Factor 2 Establish Identity to Purα and Purβ*

Transcriptional repression of the mouse vascular smooth muscle α-actin gene in fibroblasts and myoblasts is mediated, in part, by the interaction of two single-stranded DNA binding activities with opposite strands of an essential transcription enhancer factor-1 recognition element (Sun, S., Stoflet, E. S., Cogan, J. G., Strauch, A. R., and Getz, M. J. (1995) Mol. Cell. Biol. 15, 2429–2436). One of these activities, previously designated vascular actin single-stranded DNA-binding factor 2 includes two distinct polypeptides (p44 and p46) which specifically interact with the purine-rich strand of both the enhancer and a related element in a protein coding exon of the gene (Kelm, R. J., Jr., Sun, S., Strauch, A. R., and Getz, M. J. (1996) J. Biol. Chem. 271, 24278–24285). Expression screening of a mouse lung cDNA library with a vascular actin single-stranded DNA-binding factor 2 recognition element has now resulted in the isolation of two distinct cDNA clones that encode p46 and p44. One of these proteins is identical to Purα, a retinoblastoma-binding protein previously implicated in both transcriptional activation and DNA replication. The other is a related family member, presumably Purβ. Comparative band shift and Southwestern blot analyses conducted with cellular p46, p44, and cloned Pur proteins synthesized in vitro and in vivo, establish identity of p46 with Purα and p44 with Purβ. This study implicates Purα and/or Purβ in the control of vascular smooth muscle α-actin gene transcription.

Transcriptional activation or repression of RNA polymerase II-dependent genes is often mediated by sequence-specific DNA-binding proteins (i.e. transcription factors) that respond to extracellular signals by interacting with critical cis-acting regulatory elements. Activator or repressor proteins generally bind to double-stranded DNA recognition sites distal to the TATA box and influence promoter activity by interacting, either directly or through an adaptor protein, with factor(s) com-prising the RNA polymerase II basal transcription complex (1,2). Although most transcription factors demonstrate preferential binding to double-stranded DNA, a number of recent studies have associated sequence-specific, single-stranded DNA (ssDNA) 1 -binding factors with both transcriptional activating (3)(4)(5)(6) and repressing elements (7)(8)(9)(10)(11)(12). In particular, some ssDNA-binding proteins, such as the far upstream elementbinding protein and heterogeneous nuclear ribonucleoprotein K, have been shown to utilize cis elements, prone to forming non-B-DNA structures, as docking sites to effect gene expression (13)(14)(15).
In support of a role for ssDNA-binding proteins as transcription factors, our study of the vascular smooth muscle (VSM) ␣-actin gene has revealed the integral involvement of several ssDNA-binding proteins in the control of promoter activity (16 -18). Activation and repression of the mouse VSM ␣-actin promoter in both fibroblasts and undifferentiated myoblasts is mediated, in part, by positive and negative trans-acting factors that bind to a 30-base pair polypurine-polypyrimidine tract containing an inverted, muscle-specific M-CAT consensus motif (AGGAATG) (16,17). While activation likely requires the M-CAT-binding protein, transcription enhancer factor-1 (TEF-1) (17), repression appears to result from the interaction of two, tissue-restricted, ssDNA binding activities which putatively function by stabilizing local alterations in DNA secondary structure that preclude TEF-1 binding to double-stranded DNA (16). We have recently reported that one of these ssDNA binding activities, designated vascular actin single-stranded DNAbinding factor 2 (VACssBF2), also interacts with a conserved 30-base coding sequence element (CE) which borders the 3Ј end of intron 2 in the mouse VSM ␣-actin gene (18). Interestingly, this CE sequence was found to repress promoter activity when positioned 5Ј and adjacent to either a TEF-1 or activator protein-1 enhancer element (18), further confirming the ability of VACssBF2 to function as a transcriptional repressor.
Our initial efforts to biochemically characterize VACssBF2 prompted us to search the literature for known ssDNA-binding proteins that exhibited a similar preference for purine-rich sequences and possessed a molecular weight comparable to the p46 and p44 components of VACssBF2. Our search revealed a protein known as Pur␣ (19). This ssDNA-binding protein was originally identified as a HeLa cell nuclear protein that interacted with a region of stably-bent DNA in the 5Ј-region of the human c-myc gene, designated the PUR (for purine-rich) element (20). This sequence element is present in several eukary-* This work was supported in part by a grant from the National Institutes of Health (R01 HL54281) and the Mayo Foundation. 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.
The nucleotide sequences reported in this paper have been submitted to GenBank™ otic zones of initiation of DNA replication as well as in the flanking regions of a number of different genes (20). Although the exact role of Pur␣ in mammalian DNA replication is still unknown, recent studies have implicated Pur␣ in the transcriptional activation of several different genes (5,(21)(22)(23). Moreover, the identification and partial cloning of a cDNA encoding the C-terminal region of a homologous protein, termed Pur␤ (19), suggested that a family of Pur proteins may exist in higher eukaryotes. In this study, we have utilized both biochemical and molecular cloning techniques to reveal the identity of the p46 and p44 components of VACssBF2 as mouse Pur␣ and Pur␤, respectively.

EXPERIMENTAL PROCEDURES
Oligonucleotide Probes-Oligonucleotide CE-PrM 4 , a multimer of the VACssBF2 exon-binding site (18) (GGGAGTAATGGTTGGAATGGGC-CAAAAAGA) 4 was synthesized on a Applied Biosystems model 394 DNA/RNA synthesizer and gel filtered over a NAP-25 column (Pharmacia) in distilled water. The full-length 120-base oligonucleotide was purified from smaller synthesis products by polyacrylamide gel electrophoresis. CE-PrM 4 , end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase was used as the primary probe in expression library screening and Southwestern blotting. A 30-base mutant form of the VACssBF2 exon-binding site, designated CE-PrMmu2 (18) or CE-Fmu2, was used as a nonspecific competitor. Oligonucleotides corresponding to the forward and reverse strands of the human c-myc PUR element (20), PUR-F, GGAGGTGGTGGAGGGAGAGAAAAG, and PUR-R, CTTT-TCTCTCCCTCCACCACCTCC, were synthesized as described above. Biotinylated oligonucleotides corresponding to the PUR element and the VSM ␣-actin promoter element (16,18) were also prepared by chemical synthesis using a biotin phosphoramidite containing a mixed polarity triethylene glycol spacer (BioTEG, Glen Research, Sterling, VA).
Preparation AKR-2B Fibroblast Nuclear Extract-Nuclear extract was prepared according to the method of Sealey and Chalkley (24) with some minor modifications. AKR-2B fibroblasts were harvested from monolayer cultures by scraping in the presence of ice-cold phosphatebuffered saline and then collected by centrifugation. Following two additional washes with phosphate-buffered saline, the cell pellet was suspended in 10 volumes of 10 mM HEPES pH 8, 50 mM NaCl, 0.5 M sucrose, 1.0 mM EDTA, 0.5 mM spermidine, 0.15 mM spermine, 0.2% (v/v) Triton X-100, 7 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, 0.5 g/ml pepstatin, 0.5 g/ml aprotinin. After a 20-min incubation on ice with frequent (every 5 min) vortexing, nuclei were collected by centrifugation. The nuclear pellet was suspended by vigorous vortexing in 6 volumes of extraction buffer (10 mM HEPES pH 8, 350 mM NaCl, 0.1 mM EDTA, 0.5 mM spermidine, 0.15 mM spermine, 25% (v/v) glycerol, and protease inhibitors). After centrifugation at 12,000 ϫ g for 10 min, the supernatant was saved and the pellet was extracted a second time with 6 volumes of extraction buffer containing 530 mM NaCl. Following centrifugation, the 350 and 530 mM NaCl supernatants were pooled, assayed for total protein, then aliquoted, and stored at Ϫ80°C.
Fractionation of AKR-2B Nuclear Proteins-AKR-2B nuclear protein (ϳ10 mg) was diluted to 0.1 M NaCl and applied to 3 ml of heparinagarose resin (Sigma) equilibrated in 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, 0.1 M NaCl. Fractions were collected and monitored for protein by absorbance measurement at 280 nm. After washing to a baseline A 280 reading with equilibration buffer, heparin-binding proteins were eluted by increasing the concentration of NaCl to 0.6 and 1.0 M in a stepwise fashion. Selected fractions were assayed for VACssBF2 DNA binding activity by electrophoretic mobility shift assay and Southwestern blotting as described previously (16,18). All Southwestern blots reported in this study were probed with 10 6 cpm/ml 32 P-CE-PrM 4 supplemented with 10 g/ml poly(dI-dC) and a 50 -100-fold molar excess of nonspecific competitor oligonucleotide, CE-PrMmu2.
Screening of a Mouse Lung Expression Library for cDNAs Encoding VACssBF2 p46 and p44 and Sequencing of Binding-site Clones-A mouse lung cDNA expression library (Stratagene) was plated in Escherichia coli host strain, XL1-Blue, on 10 150-mm plates at a density of 25,000 plaque forming units per plate. After a 4-h incubation at 42°C, the plates were overlaid with nitrocellulose filters saturated with 10 mM isopropyl-␤-D-thiogalactopyranoside. Following an additional 16-h incubation at 37°C, filters were lifted from the plates, placed in a blocking buffer of 25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA (TNE) containing 5% Carnation nonfat dry milk, and gently agitated for 1 h at room temperature. After a 15-min equilibration in binding buffer (1% nonfat dry milk-TNE), filters were then incubated for 3 h with 1-2 ϫ 10 6 cpm/ml 32 P-CE-PrM 4 in the presence of a 160-fold molar excess of CE-PrMmu2 diluted in binding buffer containing 10 g/ml poly(dI-dC). Filters were washed three times with TNE containing 0.05% Tween 20 and one time with TNE alone. Filters were then air dried and autoradiographed for 1-2 h at Ϫ80°C. Potential positive plaques were selected, titered, replated on 100-mm dishes, and rescreened as above. Two clones, designated 9-5 and 10-1, were positive in the secondary screen. After a final tertiary screen to ensure clonal homogeneity, single plaques were selected and cDNA inserts were excised as pBluescript phagemids by co-infection with helper phage. The upper and lower strands of clones 9-5 and 10-1 were sequenced by semiautomated dideoxy termination using an Applied Biosystems Model 377A DNA sequencer and the appropriate T3, T7, and internal primers.
In Vitro Synthesis of Mouse Pur␣ and Pur␤-Proteins encoded by clones 9-5 (Pur␣) and 10-1 (Pur␤) were synthesized in vitro using a T3-dependent, transcription/translation-coupled wheat germ extract system (TnT, Promega). Cesium chloride-purified pBluescript plasmid (empty vector or vector containing cDNA insert), 1 g, was combined with 25 l of wheat germ extract, 20 M amino acid mixture, 40 units of ribonuclease inhibitor, and 10 units of T3 RNA polymerase in a final volume of 50 l. After a 1-h incubation at 30°C, an 8-l aliquot from each TnT reaction was removed, combined with an equal volume of 2 ϫ SDS sample preparation buffer with 5% (v/v) ␤-mercaptoethanol and heated at 90°C for 5 min. TnT products were analyzed for VACssBF2 DNA binding activity by Southwestern blotting.
Construction and Transfection of Mouse Pur␣  and Pur␤ (10-1) Expression Vectors-The cDNAs encoding mouse Pur␣ (clone 9-5) and mouse Pur␤ (clone 10-1) were cut from pBluescript by EcoRI and KpnI. Agarose gel purified cDNAs were then subcloned into a mammalian expression vector, pCI (Promega). Ligated plasmids were transformed into E. coli HB101 cells using standard CaCl 2 techniques. All expression vectors used in transfection experiments were purified by double cesium chloride gradient centrifugation and sequenced by semi-automated dideoxy DNA sequencing to confirm the orientation and fidelity of the cDNA insert. Transient transfections of mouse embryo-derived AKR-2B fibroblasts were performed using 10 g of total plasmid DNA per 10 6 cells as described previously (18). After a 16 -18-h recovery period, the transfected cells were rendered quiescent by incubating 48 h in serum-free MCDB402 medium (JRH Biosciences, Lenexa, KS). Cells were restimulated for 6 h with fresh MCDB402 medium supplemented with 20% fetal bovine serum (Hyclone, Logan, UT). Cells were then washed three times with phosphate-buffered saline and extracted by adding 1 ml of hypotonic lysis buffer (10 mM MOPS, pH 6.5, 10 mM NaCl, 1.5 mM MgCl 2 , 1% (v/v) Triton X-100) supplemented with 0.5 mM phenylmethylsulfonyl fluoride and 0.5 g/ml each leupeptin, pepstatin, and aprotinin. Following centrifugation at 14,000 ϫ g, cell extracts were assayed for total protein by BCA dye-binding assay (Pierce) using bovine serum albumin as a standard. Equivalent amounts of cellular protein were evaluated for p46 (Pur␣) and p44 (Pur␤) ssDNA binding activity by Southwestern blotting and band shift assay.

RESULTS
VACssBF2 Is Composed of Three, Electrophoretically Distinct, ssDNA-binding Proteins-To facilitate the molecular characterization of VACssBF2, we performed experiments aimed at defining the polypeptide composition of VACssBF2. Nuclear proteins from AKR-2B fibroblasts were applied to heparin-agarose resin, eluted stepwise with NaCl, and the fractions monitored for VACssBF2 by band shift assay. As shown in Fig. 1 (top panel), VACssBF2-ssDNA binding activity was released from heparin-agarose by 0.6 M NaCl and resolved as three distinct bands. Southwestern blotting of pooled, VACssBF2-enriched fractions (0.6 M NaCl eluate) revealed three, electrophoretically distinct, ssDNA-binding proteins with relative apparent molecular weights of ϳ115,000, 46,000, and 44,000 (Fig. 1, lower panel). These data, coupled with previous UV cross-linking experiments (18), suggested that the M r ϳ 115,000 protein is the major component of the more slowly migrating band shift complex, while the M r ϳ 46,000 and 44,000 proteins comprise the more rapidly migrating bandshift doublet (Fig. 1, top panel). It should be noted that these molecular weights are slightly different from those previously reported (121,000, 51,000, and 48,000) (18). This was due to a change in the equation used to fit the relative electrophoretic mobilities of the marker proteins (linear versus nonlinear re-gression, r ϭ 0.95 versus r ϭ 0.99). Multiple Southwestern analyses conducted with variously sized marker proteins yielded consistent molecular weight estimates for the components of VACssBF2 (115,00 Ϯ 5,000, 46,400 Ϯ 1,000, 44,100 Ϯ 1,000, n ϭ 3). Based on these data, we have designated the individual VACssBF2 components as p115, p46, and p44.
VACssBF2 p46 and p44 Interact with the Human c-myc PUR Element-To determine whether VACssBF2 p46 and p44 are related to the Pur family in terms of ssDNA-binding specificity, we tested the ability of the human c-myc PUR element sense strand (PUR-F) to function as a competitor in band shift assays using an AKR-2B fibroblast nuclear extract and 32 P-probe corresponding to the VACssBF2 coding element recognition site (CE-PrMss or CE-F). As shown in Fig. 2, PUR-F (lanes 5-8) was as effective, if not better, at inhibiting the formation of the faster migrating p46/p44 ssDNA-binding complexes (Fig. 2, double arrows) than the VSM ␣-actin coding element and promoter element sense strand oligonucleotides, CE-F (lanes 1-4) and PE-F (lanes 13-16). The absence of competition by the reverse strand of the PUR element (PUR-R, lanes 9 -12) and by the CE mutant (CE-Fmu2, lanes [17][18][19][20] indicate that the competitive properties of PUR-F were specific. Consistent with previous results, the ssDNA binding activity of the slower migrating, p115 complex (Fig. 2, single arrow) showed no apparent sequence specificity. When used as a band shift probe, 32 P-PUR-F exhibited a band shift pattern that was virtually identical to that displayed by 32 P-CE-F and 32 P-PE-F (data not shown). Moreover, streptavidin-paramagnetic particles coupled with either biotin-PUR-F or biotin-PE-F were equally efficient at "capturing" p46 and p44 from AKR-2B nuclear extracts (Fig. 3, lanes 1 and 2). Importantly, biotin-PE-R bound particles failed to capture p46 and p44 (Fig. 3, lane 3). Since equivalent amounts of fibroblast protein and oligonucleotide- coupled particles were used in each instance, these results were not attributable to a loading artifact but represent actual differences in ssDNA-binding affinity. The absence of p115 in this experiment is indicative of its weak ssDNA-binding affinity and specificity (16,18). The stringency of the binding and washing conditions and/or competition by p46/p44 probably contributed to the failure of p115 to bind the DNA-affinity particles. Collectively, these results indicated that p46 and p44 bind to purine-rich ssDNA with substantially greater affinity and specificity than p115. Furthermore, the data suggested that VACssBF2 p44 and p46 might be related to the Pur family of ssDNA-binding proteins.
Isolation and Sequence Analysis of VACssBF2 cDNA Clones-Owing to the enrichment of VACssBF2 in the lung (17), a mouse lung cDNA expression library was selected as a source for isolating cDNA clones encoding p44 and p46 using a ssDNA-binding site screening strategy. Initially, 250,000 plaques were screened with a 32 P-end labeled oligonucleotide multimer of the VACssBF2 exon recognition element (CE-PrM 4 ) in the presence of excess, unlabeled mutant oligonucleotide (CE-Fmu2) deficient in p46/p44 binding ( Fig. 2 and Ref. 18). Two clones, designated 9-5 and 10-1, which were positive in the second round of screening, were each replated at 100 plaque forming units per plate and lifted onto isopropyl-␤-Dthiogalactopyranoside-saturated filters. Each filter was cut in half and each half was screened with 32 P-CE-PrM 4 in the presence of a 160-fold molar excess of either unlabeled wildtype (CE-F) or unlabeled mutant (CE-Fmu2) oligonucleotide. For both clones, CE-F effectively competed for probe binding while CE-Fmu2 did not (data not shown). These results confirmed that each clone encoded a protein whose ssDNA binding properties were consistent with that of VACssBF2 p44 or p46.
The amino acid sequence deduced from the corresponding nucleotide sequence of each clone is represented in Fig. 4. The open reading frame of clone 9-5 encodes a protein of 321 amino acids (M r ϭ 34,884), while clone 10-1 encodes a protein of 324 amino acids (M r ϭ 33,901). In both clones, the AUG codon encoding the putative start methionine is in a favorable context for translational initiation based upon Kozak consensus criteria (25). Alignment of the deduced amino acid sequences suggested that clones 9-5 and 10-1 encode structurally related proteins (71% identity) (Fig. 5). A search of the GenBank data base revealed that the nucleotide sequence encoding the open reading frame of clone 9-5 is identical to the cDNA sequence previously reported for mouse Pur␣ (26). Moreover, the COOHterminal sequence (amino acids 213-324) encoded by 10-1 is virtually identical to the amino acid sequence encoded by a partial cDNA clone of human Pur␤ (19) (Fig. 5). Based upon these results, it seems likely that clone 10-1 encodes full-length mouse Pur␤. Comparison of the two cloned mouse Pur proteins reveal some remarkable similarities as well as some equally striking differences (Fig. 5). Mouse Pur␤ exhibits the same modular structure as originally described for the human Pur␣ protein (19). The basic, 23-amino acid class I repeats (Fig. 5, single line) and acidic, 26-amino acid class II repeats (Fig. 5, double lines) are highly conserved between Pur␣ and Pur␤. Because these Expression screening of a mouse lung cDNA library using a tetramer of the VACssBF2 exon recognition element as a probe resulted in the isolation of two clones, designated 9-5 and 10-1. The upper and lower stands of each clone were sequenced independently. The amino acid sequences deduced from the corresponding upper strand nucleotide sequence are represented using single letter code. Clone 9-5 encodes a 321-amino acid protein that is identical to mouse Pur␣ (26). Clone 10-1 encodes a related 324-amino acid protein, presumably Pur␤. Stop codons are indicated by a "*" symbol.
repeat modules have been implicated in the ssDNA binding of human Pur␣ (27), it is noteworthy that the second class II repeat is interrupted by a glycine-rich stretch in Pur␤. The "psycho" motif ( Fig. 5, dotted line), which was identified on the basis of homology to the retinoblastoma protein-binding region of SV-40 large T-antigen, other viral transforming proteins, and certain cellular proteins involved in replication and transcription (26), is also conserved in mouse Pur␤, albeit with some differences. This region has been shown to be important in modulating protein-protein interaction between retinoblastoma protein and human Pur␣ (27). Interestingly, Pur␤ lacks a potential casein kinase II phosphorylation site present in the C-terminal region of the psycho motif of Pur␣ as well as a potential N-linked glycosylation site between the first class I and first class II repeats (amino acid 95 in Pur␣, amino acid 81 in Pur␤). However, the most striking difference between Pur␣ and Pur␤ lies in the NH 2 -and COOH-terminal ends of the proteins. Although the overall glycine-rich composition of their NH 2 -terminal regions are relatively similar, the glycine-rich stretch of Pur␤ (19 of 26 residues) begins at amino acid 11 and is interrupted by a 6-amino acid sequence ( 23 FQPAPR 28 ) while Pur␣ exhibits an almost uninterrupted stretch of glycine residues (17 of 18 residues) starting at amino acid 31. Moreover, the COOH-terminal glutamine-rich stretch of Pur␣ ( 296 QQQQQQQ 302 ) is absent in Pur␤.
Identification of p46 as Pur␣ and p44 as Pur␤-Several previous studies have noted that Pur␣ exhibits an electrophoretic mobility in SDS-polyacrylamide gels consistent with a protein of 42 (28) or 45-47 kDa (27). In at least one study, the anomalous electrophoretic mobility of Pur␣ was attributed to post-translational processing (27) although another report ruled out N-glycosylation and phosphorylation as contributing to its unusual migration in SDS gels (28). To confirm the identity of cellular p46 and p44 as Pur␣ and Pur␤, respectively, the electrophoretic mobility of both in vitro and in vivo translation products of clones 9-5/Pur␣ and 10-1/Pur␤ were compared with cellular p46 and p44. As shown in Fig. 6C, the in vitro transcription/translation products of clones 9-5 and 10-1 migrated similarly to cellular p46 and p44, respectively, when analyzed by Southwestern blotting. Moreover, transient transfection of cytomegalovirus enhancer driven-expression vectors harboring clones 9-5 (Pur␣) and 10-1 (Pur␤) into AKR-2B fibroblasts resulted in the enhanced expression of p46 in cells transfected with pCI 9-5␣ (Fig. 6A, lanes 2-4), while p44 was elevated in fibroblasts transfected with pCI 10-1␤ (Fig. 6A,  lanes 6 -8). Cotransfection of both pCI 9-5␣ and pCI 10-1␤ resulted in the enhanced expression of the p46/p44 bands relative to endogenous p46 and p44 (Fig. 6B, compare lanes 2 and  5). Corroborative results were obtained when selected transfectants were analyzed by band shift assay. As shown in Fig. 7  (lanes 5-8), cotransfection of pCI 9-5␣ and pCI 10-1␤ resulted in the enhanced expression of the p46/p44 gel shift doublet. Importantly, only the faster migrating band of the doublet was augmented in cells transfected with pCI 10-1␤ alone (Fig. 7,  lanes 1-4). These data establish the identity of slower migrating band as the Pur␣/p46-ssDNA complex and the faster migrating species as the Pur␤/p44-ssDNA complex. Interestingly, the overall expression of transfected Pur␤/p44 was consistently greater than that of Pur␣/p46 (Fig. 6, A and B, and Fig. 7). This was probably due to some inherent difference in mRNA or protein stability, since both expression constructs utilized the same parent vector. In any case, these transfection results support the conclusion that p46 is Pur␣ and p44 is Pur␤.  10-1). A, amino acid sequences were aligned using the Gap algorithm (Genetics Computer Group). The 71% overall sequence identity between mouse Pur␣ and mouse Pur␤ is due to conservation of several sequence modules previously identified within the context of human Pur␣ (19). These include three basic class I repeats (single underline), two acidic class II repeats (double underline), and a predicted ␣-helical region later designated the "psycho motif" (dotted line) (26). B, the COOH-terminal region of mouse Pur␤ is virtually identical to the amino acid sequence encoded by a partial clone of human Pur␤ previously reported (19).

DISCUSSION
The varied expression of VSM ␣-actin in phenotypically altered smooth muscle cells and stromal fibroblasts associated with such pathological processes as arteriosclerosis (29 -31), vascular and cutaneous wounding (32,33), and tumor cell invasion (34, 35) suggests that these cell types have evolved transcriptional mechanisms to both activate and repress VSM ␣-actin gene expression in response to external stimuli. Indeed, a number of studies conducted with chicken, mouse, and human VSM ␣-actin promoters have indicated that cell context is critically important in determining the operative regulatory elements and DNA-binding proteins that mediate transcriptional activation and repression (36 -39). Our previous analyses of the mouse VSM ␣-actin promoter pointed to a novel molecular model in which sequence-specific VACssBFs repress promoter activity in cells of both myogenic and fibroblastic lineage by interacting with opposing strands of an essential TEF-1 enhancer element (16 -18). One of these factors, VACssBF1, has been cloned and is identical to the mouse Y-box protein MSY1 (40). Interestingly, the human homolog, YB-1, has also been reported to function as a transcriptional repressor within the context of the major histocompatibility class II DRA (41) and grp78 (42) promoters. In the mouse VSM ␣-actin promoter, MSY1 interacts with the pyrimidine-rich, antisense strand of the TEF-1 enhancer. The other factor, VACssBF2, binds to the purine-rich sense strand of the TEF-1 enhancer and a related element in a protein-coding exon of the gene (18). While initial data suggested that VACssBFs function in a cooperative manner to disrupt TEF-1 enhancer function (16), recent experiments conducted with chimeric promoters have indicated that VACssBF2 can repress enhancer activity in the absence of detectable VACssBF1/MSY1-binding to the opposing strand (18).
Owing to the relative paucity of reports documenting a role for ssDNA-binding proteins in transcriptional repression, we sought to characterize VACssBF2 using biochemical and molecular cloning techniques. Southwestern blotting of heparinagarose fractions enriched in VACssBF2 revealed three prominent bands of M r ϳ115,000, 46,000, and 44,000 (Fig. 1).
However, only the p46 and p44 species demonstrated high affinity, purine-rich ssDNA-binding specificity (Figs. 2 and 3). Expression screening of a mouse lung cDNA library with a multimer of VACssBF2 exon recognition element yielded two cDNA clones (9-5 and 10-1) that encode related ssDNA-binding proteins (Figs. 4 and 5). Clone 9-5 encodes Pur␣, a purine-rich ssDNA-binding protein previously shown to interact with (GGN) n repeats in sequence elements functionally linked to both DNA replication and transcription (5,19,20,23,26,28,43). Based upon homology to a partial cDNA clone of human Pur␤ previously reported (19), clone 10-1 encodes mouse Pur␤. These results were not surprising given the prior finding that cellular p46 and p44 could also bind to an authentic Pur␣ ssDNA-recognition element in vitro (Figs. 2 and 3). However, clones 9-5 and 10-1 encode full-length proteins with predicted molecular masses of 34.9 and 33.9 kDa, respectively. While these values differ substantially from the apparent molecular weights of p46 and p44, both in vitro and in vivo synthesized Pur␣ and Pur␤ co-migrate with cellular p46 and p44 in SDSpolyacrylamide gels (Fig. 6). Thus, the apparent discrepancies in molecular weight are a result of anomalous electrophoretic migration.
Because the cloning and detection strategies employed in these studies were based on affinity for a VACssBF2 ssDNA recognition element, it might be argued that cellular p46 and p44 are not identical to the Pur proteins but simply exhibit similar ssDNA-binding properties. We believe this is highly unlikely for the following reasons. First, with the exception of p115, no other bands are detectable in band shift, Southwestern blot, or biotinylated ssDNA capture experiments using fibroblast nuclear extracts and either VACssBF2 or human c-myc Pur␣ ssDNA oligonucleotide probes (Figs. 1-3, and data not shown). Second, and most importantly, Pur␣ and Pur␤ exactly co-migrate with cellular p46 and p44 when overexpressed in fibroblasts and subjected to SDS-polyacrylamide gel electrophoresis (Fig. 6). Thus, any non-Pur proteins which exhibit Pur-like ssDNA binding activity would have to exactly co-migrate with authentic Pur␣ and Pur␤. Based on these considerations, we conclude that the p46 and p44 components of the VACssBF2 complex correspond to Pur␣ and Pur␤, respectively.
With the molecular cloning of the p46 and p44 components of VACssBF2, we have now identified 4 of the 5 proteins previously shown to interact with single-or double-stranded forms of an essential VSM ␣-actin TEF-1 enhancer element (16,17). These include a TEF-1 related protein (17), MSY1 (or VACssBF1), and now Pur␣ and Pur␤ (or VACssBF2 p46 and p44). Only the relatively weak and nonspecific ssDNA-binding p115 component of VACssBF2 remains to be cloned and identified. Curiously, both cellular Pur␣ and YB-1 (the human homolog of MSY-1) have also been implicated in the activation of early (Pur␣) and late (YB-1) gene transcription of human JC polyomavirus in glial cells of the central nervous system (22). In this context, Pur␣ and YB-1 were found to function cooperatively in that each protein was capable of modulating the binding of the other to its respective viral promoter ssDNAbinding site (23). Since certain astroglial cells are known to also express VSM ␣-actin (44,45), the functional association of Y-box and Pur proteins may be common to VSM ␣-actin expressing cell types.
It is interesting that human Pur␣ and YB-1 function as activators of JC polyomavirus gene transcription in glial cells while mouse Pur␣ and/or Pur␤, and MSY1 apparently function as repressors of VSM ␣-actin gene transcription in myoblasts and fibroblasts. While it is not uncommon for a transcription factor, such as Pur␣ or YB-1, to act as either an activator or as FIG. 7. The slower and faster migrating bands of the VACssBF2 band shift doublet correspond to Pur␣ and Pur␤, respectively. 32 P-Oligonucleotide corresponding to the sense strand of the VSM ␣-actin promoter element (PE-PrMss, ϳ20 fmol, 30,000 cpm) was incubated with 4 g of lysed cell protein from AKR-2B fibroblasts transfected with varying amounts of pCI 10-1␤ alone (lanes 1-4) or pCI 10-1␤ plus pCI 9-5␣ (lanes 5-8) as described in the legend to Fig. 6. Protein-ssDNA complexes were resolved by band shift assay. Arrows (from top to bottom) show Pur␣ and Pur␤ protein-ssDNA complexes. FP, free probe. a repressor depending upon the influence of other cell and/or promoter-specific gene regulatory proteins, the structural differences between mouse Pur␣ and Pur␤ are nevertheless provocative (Fig. 5). Because glutamine-rich sequences have been associated with transcriptional activation domains (1), the absence of a Q 7 sequence and other differences in the COOHterminal region of Pur␤ may be of functional significance in terms of repression. Perhaps Pur␤ antagonizes activation by Pur␣ via competitive ssDNA binding. Elucidation of the individual functional roles of MSY1, Pur␣, and Pur␤, in repression of the VSM ␣-actin gene promoter in myoblasts and fibroblasts will obviously require further investigation. Several questions to be addressed include 1) do these proteins possess the ability to modulate TEF-1 enhancer topology, 2) do they interact with the enhancer in a mutually-dependent or independent manner, 3) what are the consequences of such interactions in terms of enhancer function, 4) are these proteins differentially expressed or post-translationally modified in a cell-or tissuespecific manner, and 5) does Pur␤, like Pur␣ (27), interact with the retinoblastoma protein? Answers to such questions should lead to an improved understanding of the mechanisms that regulate VSM ␣-actin gene expression during smooth muscle myogenesis, cutaneous wound healing, vascular injury, fibrocontractive disease, and stromal responses to neoplasia.