Nucleoprotein Interactions Governing Cell Type-dependent Repression of the Mouse Smooth Muscle α-Actin Promoter by Single-stranded DNA-binding Proteins Purα and Purβ*

Purα and Purβ are structurally related single-stranded DNA/RNA-binding proteins implicated in the control of cell growth and differentiation. The goal of this study was to determine whether Purα and Purβ function in a redundant, distinct, or collaborative manner to suppress smooth muscleα-actin gene expression in cell types relevant to wound repair and vascular remodeling. RNA interference-mediated loss-of-function analyses revealed that, although Purβ was the dominant repressor, the combined action of endogenous Purα and Purβ was necessary to fully repress the full-length smooth muscle α-actin promoter in cultured fibroblasts but to a lesser extent in vascular smooth muscle cells. The activity of a minimal core enhancer containing a truncated 5′ Pur repressor binding site was unaffected by knockdown of Purα and/or Purβ in fibroblasts. Conversely, gain-of-function studies indicated that Purα or Purβ could each independently repress core smooth muscle α-actin enhancer activity albeit in a cell type-dependent fashion. Biochemical analyses indicated that purified recombinant Purα and Purβ were essentially identical in terms of their binding affinity and specificity for GGN repeat-containing strands of several cis-elements comprising the core enhancer. However, Purα and Purβ exhibited more distinctive protein interaction profiles when evaluated for binding to enhancer-associated transcription factors in extracts from fibroblasts and vascular smooth muscle cells. These findings support the hypothesis that Purα and Purβ repress smooth muscle α-actin gene transcription by means of DNA strand-selective cis-element binding and cell type-dependent protein-protein interactions.

Phenotypic reprogramming of adventitial fibroblasts and medial vascular smooth muscle cells (VSMCs) 4 plays a critical role in arterial remodeling associated with de novo atherosclerosis, restenosis after per-cutaneous coronary intervention, transplant arteriosclerosis, and veingraft disease (1)(2)(3)(4)(5). Although the notion of phenotypic modulation of vascular cell types has existed for several decades, the molecular mechanisms leading to pathologic changes in cell migration, proliferation, or apoptosis are largely unknown. Among the well characterized markers of VSMC differentiation status, smooth muscle ␣-actin (SM␣A) is unique in that its physiological role is not entirely restricted to regulation of vascular contractility (6). Indeed, experimental evidence indicates that reduced levels of SM␣A may alter the tissue-invasive and motile properties of certain fibroblastic cell types in vitro (7,8) and may be a contributing factor in elevating the risk of plaque rupture in vulnerable fibroatheroma in vivo (9,10). On the other hand, enhanced SM␣A expression is crucial to myofibroblast differentiation and physiological wound repair (11,12) and, together with other contractile proteins such as smooth muscle myosin, may limit neointimal hyperplasia during vascular remodeling induced by mechanical injury or proatherogenic stimuli (13)(14)(15). Thus, identification of factors that modulate SM␣A expression in myofibroblasts and VSMCs could provide useful targets for therapies designed to either stabilize, or perhaps favorably alter, the phenotypic composition of vascular lesions.
Among the multitude of transcription factors implicated in SM␣A gene regulation (16,17), a group of functionally novel single-stranded DNA (ssDNA)-binding proteins known as Pur␣, Pur␤, and MSY1 have been reported to restrain muscle-CAT (MCAT) enhancer-dependent SM␣A gene transcription in both fibroblasts and VSMCs (18). These repressor proteins were originally identified based on their ability to interact with opposing strands of a highly conserved, asymmetric polypurine-polypyrimidine (Pur/Pyr) tract located in the 5Ј region of the mouse SM␣A gene (19). In its double-stranded configuration, the Pur/ Pyr sequence contains a consensus MCAT enhancer motif, which serves as a recognition site for transcription enhancer factor 1 (TEF-1) (18). Importantly, computer modeling of the extended Pur/Pyr tract coupled with ssDNA-selective chemical modification of genomic DNA in cultured fibroblasts indicated that this region has the propensity to assume a partially unpaired configuration in response to transforming growth factor ␤1 (TGF␤1) signaling (20). These findings coupled with results of in vitro DNA binding and cell-based SM␣A promoter mutagenesis studies, led to the development of two alternative models for cryptic MCAT enhancer regulation. The salient feature of both models involves ssDNA-specific interactions by endogenous Pur␣, Pur␤, and MSY1, which either mask TEF-1 from other components of the transcription machinery or alter cis-element structure in such a way as to preclude TEF-1 binding to double-stranded DNA (dsDNA) (18). Although speculative, the latter scenario is plausible because several independent studies have demonstrated that Pur and Y-box proteins are individually capable of promoting DNA strand separation in vitro (21)(22)(23)(24). However, based on more recent work, it is now clear that Pur␣ and Pur␤ have additional targets and mechanistic roles to play in suppressing smooth and cardiac muscle-specific gene transcription depending upon the cell type studied, Pur expression levels, and/or TGF␤1 signaling (25)(26)(27).
There are four known Pur family members in mice and men, Pur␣, Pur␤, and two isoforms of Pur␥ (28 -30). Although a comprehensive analysis of the tissue expression of Pur paralogs at the protein level has not yet been reported, evidence from RNA transcript analysis suggests that Pur␣ and Pur␤ are ubiquitously expressed while Pur␥ exhibits a more restricted distribution (30). All Pur family members exhibit a substantial degree of homology particularly in domains that confer nucleic acid binding capacity (25,31,32). By virtue of their ability to interact with high affinity and specificity with single-stranded nucleic acid sequences rich in purine nucleotides of the form (GGN) n , Pur proteins appear to have evolved to regulate cellular processes essential to genetic metabolism, including DNA replication and transcription, RNA transport and localization, and mRNA translation (33). The founding member of this family, Pur␣, has also been implicated in regulation of cell cycling and proliferation in vitro (34 -38) and in brain and myeloid development in vivo (39). Despite their structural relatedness, it is unknown at present whether Pur␣ and Pur␤ are functionally redundant or distinct. Hence, the objective of this study was to perform a rigorous comparative evaluation of mouse Pur␣ and Pur␤ with the use of gainof-function and loss-of-function strategies in SM␣A-expressing cell types. In combination with results of biochemical assays conducted with purified recombinant proteins, our findings provide insight into the physical and functional interactions that account for the ability of Pur␣ and Pur␤ to repress SM␣A gene expression in a cell-type-specific manner.

EXPERIMENTAL PROCEDURES
Cell Culture, Transient Transfection, and Reporter Gene Assay-Aortic segments from C57BL/6J mice were obtained following protocols approved by the University of Vermont Institutional Animal Care and Use Committee. VSMCs were isolated by cell outgrowth from aortic tissue explants and characterized as previously described (40,41). Primary VSMCs were cultivated in a 90% air/10% CO 2 incubator at 37°C in growth medium consisting of Dulbecco's modified Eagle's medium, 1ϫ insulin-transferrin-selenium supplement (Invitrogen), and 20% v/v heat-inactivated fetal bovine serum (FBS, HyClone). For transient transfection studies, primary VSMCs were seeded in six-well plates at a density of 10,000 cells/well and transfected with the use of jetPEI TM reagent at a ratio of 2 l/g of plasmid DNA as directed by the manufacturer (Qbiogene). AKR-2B mouse embryonic fibroblasts (MEFs) or rat A7r5 VSMCs were cultured and transiently transfected as previously described (42). Briefly, subconfluent AKR-2B or A7r5 cells seeded in 60-mm dishes were transfected with the use of GenePORTER TM reagent (Gene Therapy Systems) at a ratio of 3 l/g of plasmid DNA. After 48-h incubation in growth medium, cells were washed with phosphate-buffered saline and extracted using 1ϫ reporter lysis buffer (Roche Applied Science) supplemented with protease inhibitors. In some cases, transfected cells were growth-arrested by incubation in serum-free medium for 48 h then re-stimulated with either 20% FBS or 5 ng/ml TGF␤1 (R&D Systems) for 6 h prior to harvesting cell lysates. Total protein in transfected cell lysates was determined by bicinchoninic acid assay (Sigma) using bovine serum albumin (BSA) as a protein standard. Commercial immunoassays were used to measure chloram-phenicol acetyltransferase (CAT) or ␤-galactosidase reporter proteins as directed by the manufacturer (Roche Applied Science). Reporter values were corrected for total protein content. Transfections were typically performed in triplicate and repeated two to three times to ensure reproducibility. Data sets were subjected to one-way analysis of variance to identify differences among group means at the p Ͻ 0.05 significance level. Post-hoc comparisons were performed using the Bonferroni adjustment for multiple comparisons.
Construction of Promoter-Reporter Plasmids and Expression Vectors-Mouse SM␣A promoter-reporters and mammalian expression plasmids encoding His-epitope-tagged versions of mouse Pur␣ and Pur␤ were described previously (18). A methodological description of the design and construction of plasmid-and lentiviral-based short hairpin RNA (shRNA) expression vectors targeting mouse Pur␣ or Pur␤ can be found in the supplemental material. All plasmids used for transfection were purified from Escherichia coli cultures by double cesium chloride gradient centrifugation.
Western Blotting of Transgene-expressed and Endogenous Proteins-Ectopic expression of His-tagged Pur␣ or Pur␤ was monitored by Western blotting of lysed cell protein with an RGS(H) 4 monoclonal antibody (Qiagen) as previously described (25). Expression of endogenous Pur proteins was similarly assessed with a rabbit polyclonal antibody that recognizes a conserved sequence present in both mouse Pur␣ and Pur␤ (19). Commercial monoclonal antibodies were used for detection of SM␣A (clone 1A4, Sigma) and glyceraldehyde-3-phosphate dehydrogenase (clone 6C5, Research Diagnostics Inc.).
Purification of His-Pur␣ and His-Pur␤ Expressed in E. coli-Bacterial expression constructs encoding mouse Pur␣ and Pur␤ with N-terminal His epitope tag were described previously (19). Detailed methods used for the expression, purification, and quality control of His-tagged Pur␣ and Pur␤ are presented in the supplement materials. His-tagged MSY1 expressed in E. coli was purified and quantified as previously described (19).
Biochemical Assays for DNA Binding and Protein-Protein Interaction-A discontinuous solid-phase DNA binding assay was performed as previously described with some minor modifications (43). Recombinant His-Pur␣ or His-Pur␤ was diluted to a final concentration of 50 nM in coating buffer consisting of 20 mM HEPES, pH 7.5, 150 mM NaCl, 5 g/ml BSA. Proteins were passively adsorbed to microtiter wells (Costar EIA/RIA plate, 96-well Easywash TM high binding) by applying 100 l of coating solutions and incubating overnight at 4°C for 16 -24 h. Coating solution was removed by aspiration, and wells were rinsed twice with wash buffer consisting of 20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.05% v/v Tween 20. Blocking buffer containing 2% w/v BSA (Sigma, EIA grade) dissolved in HEPES buffer was then applied to each well. After 1-h incubation at room temperature, blocking buffer was removed and 100 l of fluid phase competitor (Pur␣ or Pur␤ or unlabeled oligonucleotide) diluted in binding buffer (0.2% w/v BSA dissolved in wash buffer) was applied in triplicate. After 10-min incubation at 37°C, 100 l of 2 nM 3Ј-biotinylated ssDNA probe diluted in binding buffer was added. Plates were incubated for 1 h at 37°C. Binding solutions were removed by aspiration, and wells were washed three times. This was followed by addition of 100 l of ExtraAvidin-HRP (Sigma) diluted 1:2000 in wash buffer to each well and incubation for 30 min at 37°C. Wells were washed three times, and 100 l of ABTS (2,2Ј-azinobis[3-ethylbenzothiazoline-6-sulfonic acid]) substrate solution (Chemicon) was added. After 15-min incubation at room temperature, absorbance readings at 405 nm were obtained with a microplate reader. In another set of experiments, the DNA binding assay was performed the same way as described above except that the fluid phase contained dif-fering amounts of 3Ј-biotinylated single-or double-stranded probes corresponding to selected sequences found in the mouse SM␣A 5Ј-flanking region. Biotinylated or unlabeled oligonucleotides were made by chemical synthesis and purified by reverse phase cartridge (Sigma-Genosys). dsDNA probes were generated by annealing complementary oligonucleotides in which only one of the strands was biotinlabeled. Annealed dsDNA probes were purified by electrophoresis on 4% agarose gels.
An enzyme-linked immunosorbent assay (ELISA) for detection protein-protein interaction was performed as previously described (42). In these experiments, microtiter wells were coated with selected concentrations of His-Pur␣ or His-Pur␤. Nuclear or cytosolic extracts derived from either AKR-2B or A7r5 cells served as a source of Pur-interacting factors. Primary and secondary antibodies used for the detection of protein-protein complexes were obtained commercially (Santa Cruz Biotechnology). Rabbit polyclonal MSY1 and TEF-1 antibodies were produced by this laboratory (18,19).

Functional Consequences of Pur␣ or Pur␤ Knockdown in Fibroblasts
and VSMCs-To investigate the functional properties of endogenous Pur␣ and Pur␤ in cultured fibroblasts and VSMCs, a loss-of-function approach utilizing RNA interference was undertaken. The effectiveness and specificity of shRNA-mediated knockdown of both exogenous and endogenous Pur␣ and Pur␤ in fibroblasts or VSMCs was confirmed by transfection and immunoblot analyses (see supplemental data, Fig. S1). Given previous studies indicating that these proteins function as corepressors of the SM␣A promoter (18), we hypothesized that knockdown of Pur␣ and/or Pur␤ in non-differentiated fibroblasts would result in promoter activation. A series of co-transfection experiments was conducted in AKR-2B MEFs with the use of a full-length SM␣A promoter-CAT reporter known as VSMP8 (Fig. 1A) and selected Pur␣or Pur␤-specific shRNA expression plasmids. The VSMP8 promoter construct was chosen, because it contains mouse SM␣A sequence (Ϫ1074 through the first intron) required for smooth muscle-specific  (18,26). VSMP4 lacks the ϳ2.5-kbp intron 1 (triangle), 5Ј-flanking region from Ϫ1074 to Ϫ192, and is transcriptionally activated due to exposure of a cryptic MCAT enhancer (⌬K d ) (18). B, specific and dose-dependent activation of VSMP8 by Pur␤ shRNA-I. AKR-2B cells were co-transfected with 2.8 g of VSMP8, 0.2 g of pCMV␤, selected amounts of plasmid encoding Pur␤ shRNA-I or scrambled control sequence, and filler DNA to 5.0 g/dish. Whole cell extracts were prepared 48 h after transfection and assayed for total protein and CAT reporter. C, extracts of AKR-2B MEFs transduced with lentiviral vectors encoding Pur␤ shRNA-I (lane 1) or scrambled sequence (lane 2) were analyzed by Western blotting for detection of Pur␣ and Pur␤ (10 g of protein/lane) or SM␣A (1 g of protein/lane). Note that the faster migrating Pur␤ band is specifically reduced while SM␣A is increased in shRNA-expressing cells. The SM␣A blot was reprobed for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a loading control. Numbers on the left denote the size of prestained protein markers in kilodaltons. D, de-repression of the SM␣A promoter by Pur␤ shRNA-I is promoter context-dependent. AKR-2B cells were co-transfected with VSMP8 or VSMP4 Ϯ 1 g of the indicated shRNA plasmids then assayed as described above. B and D, promoter activity is expressed as CAT reporter divided by total cell protein (mean Ϯ S.D.). *, p Ͻ 0.025; **, p Ͻ 0.01 compared with reporter only control.

FIGURE 2. De-repression of the SM␣A promoter in MEFs and VSMCs in response to knockdown of Pur␣ and/or Pur␤ and the requirement for Pur/Pyr element integrity.
A and B, AKR-2B MEFs or primary C57BL/6J VSMCs were transiently transfected with 2.8 g of VSMP8 or VSMP4 reporters, 0.2 g of pCMV␤, and 1.0 g of expression plasmid encoding the indicated shRNA or scrambled control sequence. pBLCAT3 was used as filler to equalize the amount of DNA transfected at 5.0 g/dish. Whole cell extracts were prepared 48 h after transfection and assayed for total protein and CAT reporter. To compare the relative effect of shRNAs or scrambled controls on each promoter, corrected CAT values were normalized to values obtained in transfectants in which VSMP8 or VSMP4 were co-transfected with pBLCAT3 filler DNA only (control activity defined as 1 for each reporter). Results are expressed as -fold de-repression (mean Ϯ S.E.). **, p Ͻ 0.01; ***, p Ͻ 0.001 compared with reporter only control. Inset, AKR-2B MEFs were co-transfected with the indicated SM␣A promoterreporter constructs and a combination of Pur␣ plus Pur␤ shRNA-I plasmids or scrambled control plasmids then assayed as described above. To assess the relative level of shRNAmediated induction of each reporter, CAT values obtained in Pur␣ plus Pur␤ shRNA co-transfectants were divided by CAT values measured in scrambled control co-transfectants. Results are expressed as relative promoter induction (mean Ϯ S.E.). **, p Ͻ 0.01; ***, p Ͻ 0.001 compared with VSMP4. C and D, deletion mutants of the SM␣A Pur/Pyr element were tested for their ability to compete for binding of full-length biotinylated ssDNA probe (ssPE32-bF, Ϫ195 to Ϫ164 5Ј-GGGAGCAGAACAGAGGAATGCAGTGGAAGAGA-3Ј-biotin, 1 nM) to purified recombinant Pur␣ or Pur␤ in a solid-phase colorimetric ssDNA binding assay. Truncating competitor oligonucleotide to Ϫ191 reduced competition efficiency by ϳ10-fold.
transgene expression in vivo (45) and, as such, exhibits very weak transcriptional activity in transfected AKR-2B cells due, in part, to negative regulation by endogenous Pur repressor proteins. As shown in Fig. 1B, co-transfection of Pur␤ shRNA-I resulted in a dose-dependent increase in VSMP8 promoter activity, whereas a scrambled control construct (Pur␤ Scm) had no effect on this reporter. Corroborative results were obtained in lentiviral-transduced AKR-2B MEFs where stable shRNAmediated knockdown of Pur␤ augmented SM␣A protein expression (Figs. 1C and S1, see supplemental data).
ToevaluatewhethertheeffectofPur␤knockdownwaspromotercontextdependent, a truncated SM␣A reporter known as VSMP4 was also tested for responsiveness to Pur␤ deficiency. As previously documented (18), this mutant reporter exhibits unrestricted MCAT enhancer activity due to the absence of 5Ј-nucleotides required for strand-specific Pur/Pyr element recognition by endogenous Pur repressors (see Fig. 1A for schematic). Hence it was not surprising to find that expression of Pur␤ shRNA-I had no discernable effect on the transcriptional activity of VSMP4 in transfected MEFs (Fig. 1D). To assess whether knockdown of Pur␣ would yield analogous results, shRNA constructs demonstrating the maximum efficiency of Pur␣ or Pur␤ knockdown (Fig. S1, Pur␣ shRNA-I and Pur␤ shRNA-I) were transfected either individually or in combination into AKR-2B MEFs together with VSMP8 or VSMP4 reporters ( Fig. 2A). Knockdown of Pur␣ alone resulted in a modest ϳ2-fold enhancement in VSMP8 activity, whereas knockdown of Pur␤ alone induced VSMP8 by ϳ4-fold. Importantly, a synergistic response (ϳ12-fold activation over reporter only control) was observed in AKR-2B MEFs expressing both Pur␣ and Pur␤ shR-NAs implying that endogenous Pur repressors likely function in a collaborative manner to regulate the transcriptional activity of the fulllength SM␣A promoter in this cell type. Transient transfection of the same Pur␣ or Pur␤ shRNA constructs in primary mouse VSMCs yielded similar results, although the extent of de-repression was less pronounced than seen in AKR-2B MEFs (Fig. 2B). Relative to VSMP8, VSMP4 exhibited little or no responsiveness to combined Pur␣ and Pur␤ knockdown in both cell types. Evaluation of other SM␣A reporter constructs containing differing lengths of 5Ј-flanking region indicated that significant promoter induction in response to co-knockdown of Pur␣ and Pur␤ in AKR-2B cells minimally required 5Ј sequence extending to Ϫ195 ( Fig. 2A, inset, compare ⌬195 to VSMP4). This result coincides with biochemical studies indicating that high affinity binding by purified Pur␣ or Pur␤ to the purine-rich strand of a Pur/Pyr element is impaired by deleting the GGGA motif from Ϫ195 to Ϫ192 (Fig. 2, C and D).
Functional Consequences of Pur␣ or Pur␤ Overexpression in Fibroblasts and VSMCs-In an earlier study, we reported that Pur␣ and Pur␤ demonstrate vastly different repressor activity toward the full-length VSMP8 reporter when ectopically expressed in rat A7r5 VSMCs (25). To ensure that this finding was not entirely cell line-dependent, analogous transfection studies were conducted in exponentially growing primary mouse VSMCs. As shown in Fig. 3, whereas Pur␤ showed potent repressor activity toward VSMP8, Pur␣ failed to significantly inhibit the promoter. This result was not due to an overt difference in transgenemediated expression of Pur␣ as Western blotting of lysed transfectants revealed comparable expression of His-tagged Pur␣ and Pur␤ (inset of Fig. 3). Curiously, the VSMP4 reporter, which contains elements comprising the minimal core SM␣A enhancer, was also repressed by forced Pur␤ expression, but not by Pur␣, in primary mouse VSMCs.
To ascertain whether repression of VSMP4 by exogenous Pur␤ was unique to primary cells and to test if Pur␣ would repress VSMP4 when similarly overexpressed in a less differentiated cell type, titration studies were performed in AKR-2B MEFs and A7r5 VSMCs. Consistent with results obtained in primary VSMCs, Pur␤ showed more potent VSMP4 inhibitory activity than Pur␣ when overexpressed in A7r5 cells (Fig. 4A). However, exogenous Pur␣ and Pur␤ exhibited equally strong repressor activity toward VSMP4 in AKR-2B cells (Fig. 4B). Comparable dose-dependent expression of His-Pur␣ or His-Pur␤ in each cell type was confirmed by Western blotting (insets of Fig. 4, A and B). Adjusting cell culture conditions did not appear to influence the cell type-dependent activity of exogenous Pur␣ or Pur␤, because similar results were obtained in synchronized cells that were either serum-starved or stimulated with serum or TGF␤1 after growth arrest (Fig. 4, C and D). Relative to A7r5 cells, which exhibit a stable adult VSMC phenotype (46), the VSMP4 reporter predictably showed a higher degree of serum and TGF␤1 inducibility in synchronized AKR-2B cells consistent with transition to a myofibroblast-like phenotype (20). In growth factor-stimulated AKR-2B MEFs, exogenous Pur␣ and Pur␤ were equally effective in attenuating VSMP4 induction (Fig. 4D). In contrast, only Pur␤ inhibited VSMP4 activity in synchronized A7r5 VSMCs, whereas Pur␣ had a modest potentiating effect on VSMP4 in serum-free and TGF␤1treated cells (Fig. 4C).
Binding Affinity and Specificity of Recombinant Pur␣ and Pur␤ for SM␣A-derived Cis-elements-In an attempt to uncover a physical basis for the differences in transcriptional activity of Pur␣ and Pur␤ observed in gain-of-function studies, quantitative nucleoprotein interaction assays were conducted in which biotinylated DNA probes in either single-or double-stranded configurations were assessed for binding to His-Pur␣ or His-Pur␤ immobilized on microtiter wells. Care was taken to ensure that purified preparations of recombinant Pur␣ and Pur␤ were devoid of contaminating nucleic acid and functional in terms of DNA binding in solution (supplemental Figs. S2 and S3). As shown in Fig. 5 (A and B), His-Pur␣ and His-Pur␤ exhibited identical DNA-binding properties when probed with either the purine-rich strand (ssPE30-bF) or complementary pyrimidinerich strand (ssPE30-bR) of a minimal 30-nucleotide SM␣A Pur/Pyr element. High affinity binding (apparent K d ϳ 1.7 nM) was observed to the purine-rich strand, whereas weaker and non-saturable binding characterized the interaction of His-Pur␣ and His-Pur␤ with the pyrimidine-rich strand. The A 405 nm readings generated with the double-stranded form of the Pyr/Pur element containing the canonical MCAT motif was near or below background (BSA-only-coated wells) consistent with the notion that Pur proteins preferentially bind to single-stranded nucleic acid sequences rich in purine residues. As a control, purified His-tagged MSY1 was also tested against the same probes. In agreement with past findings (18,19), recombinant MSY1 demonstrated specific and high affinity interaction with the pyrimidine-rich strand only (Fig. 5C).
Results of previous SM␣A promoter-reporter transfection studies coupled with in vitro DNA-binding assays using cell extract-derived proteins suggested that Pur proteins can physically and functionally interact with other cis-elements in the core enhancer region of the SM␣A gene. These sites include a TGF␤1-hypersensitive region (THR, Ϫ176 to Ϫ145) (47) and a Sp1/3 and Pur␣/␤-interacting element (SPUR, Ϫ59 to Ϫ28) (26). To quantitatively characterize how Pur proteins associate with each of sites relative to the MCAT-containing Pur/ Pyr element (Ϫ195 to Ϫ164), biotinylated DNA probes of equivalent 32 nucleotide length in either single-or double-stranded configurations were evaluated for binding to recombinant mouse Pur␣ or Pur␤ by colorimetric assay (Fig. 6). Results indicated that Pur proteins preferentially interact with the purine-rich strand of each of these elements albeit with distinctly different affinities (ssSPUR32-bF Х ssPE32-bF Ͼ ssTHR32-bR from highest to lowest affinity, apparent K d ϳ1 nM to ϳ20 nM). Pur␣ or Pur␤ binding to double-stranded PE32, SPUR32, and THR32 probes in which the high affinity strand was annealed to its lower affinity biotin-labeled complement was either not detectable or very weak under the assay conditions employed. Although Pur␣ and Pur␤ exhibited comparable affinities for each purine-rich ssDNA probe tested, a significant difference was noted in terms of apparent binding capacity (A 405 reading at saturation) for PE and SPUR versus the THR element. Although disparities in surface orientation of the biotin moiety in the solid-phase ssDNA-protein complex could account for such a result, an equally plausible interpretation is that the SPUR and PE interaction stoichiometry may differ from that of the THR.
Interaction Profile of Recombinant Pur␣ and Pur␤ with SM␣A Promoter-associated trans-Acting Factors-Results of DNA binding studies utilizing purified recombinant proteins suggested that Pur␣ and Pur␤ are indistinguishable in terms of affinity and specificity for relevant SM␣A ciselements in vitro. To ask whether the same is true for known MCAT, CArG, THR, and SPUR/TCE element-binding trans-acting factors, ELISAbased profiling of Pur␣ and Pur␤ interaction partners was conducted using protein extracts from proliferating AKR-2B and A7r5 cells. In this assay system, equimolar amounts of recombinant His-Pur␣ or His-Pur␤ were immobilized on microtiter wells. A control ELISA performed with the use of a pan Pur polyclonal antibody verified that the coating efficiencies of His-Pur␣ and His-Pur␤ were virtually identical over a broad range of concentrations (supplemental Fig. S4). A surface-saturating coating concentration of 100 nM was selected for all subsequent protein interaction studies in which a fixed amount of nuclear or cytosolic protein extract from either fibroblasts or VSMCs was incubated with wells pre-coated with His-Pur␣ or His-Pur␤. BSA-only-coated wells served as a control for binding speci- ficity. After washing to remove unbound proteins, solid-phase transcription factor-Pur complexes were detected by ELISA with the use of primary antibodies specific for MSY1, TEF-1, SRF, Sp1, Sp3, pRB, or Smad2/3 followed by the appropriate HRP-coupled secondary antibody for colorimetric signal generation. As a further control for the fidelity of this detection system, each antibody was also independently screened for cross-reactivity with immobilized Pur␣ or Pur␤ in the absence of added cell extract. This was done so that any background signal due to nonspecific primary or secondary antibody binding could be subtracted from absorbance readings obtained in Pur␣-or Pur␤-coated wells incubated with cell extract. To account for the variability inherent in cell extract preparation, binding profiles were determined using at least two different preparations from each cell type.
As shown in Fig. 7, protein interaction profiles for Pur␣ or Pur␤ were dependent upon the cell type and subcellular compartment from which factors were derived. Striking differences between Pur␣ and Pur␤ in Numbers in parentheses designate the calculated concentration of DNA probe where half-maximal binding was achieved. Sense (bF) and antisense (bR) strands were initially tested to identify the highest and lowest affinity strands comprising each element. To guard against any possible contribution from contaminating high affinity ssDNA, dsDNA probes were created by annealing complementary strands in which only the low affinity strand was biotin-labeled and then purified by gel electrophoresis.  (19). Insets, representative profiles generated with the use of cytosolic extracts from the same cell type are shown for comparison (mean Ϯ S.E.).
terms of binding capacity for certain factors were evident both within and between cell types. For example, MSY1 demonstrated ϳ2-fold greater preference for Pur␤ than for Pur␣ in nuclear extracts from both AKR-2B MEFs and A7r5 VSMCs (Fig. 7, compare A and B). Relative to Pur␣, Pur␤ also demonstrated increased binding activity toward nuclear Sp3 in both cell types and nuclear Sp1 particularly in A7r5 cells. Curiously, these differences were less pronounced or entirely absent when cytosolic proteins were used as a source of Pur interacting factors (compare insets in Fig. 7, A and B). In fact, Pur␣ exhibited a trend toward preferential binding to cytosol-derived TEF-1, Sp3, and Smad2/3, whereas Pur␤ preferred to associate with the nuclear forms of these proteins in AKR-2B cells (Fig. 7A). In contrast, Pur␣ and Pur␤ were comparable in terms of their ability to bind TEF-1, SRF, and Smad2/3 present in the nucleus of A7r5 cells. The MSY1, TEF-1, and Sp3 interaction profiles of Pur␣ and Pur␤ were virtually identical using cytosolderived proteins from A7r5 cells (Fig. 7B). Consistent with the rapid growth status of cells at the time of extract preparation, interaction of pRB with Pur␣ or Pur␤ was not detectable in this assay system.

DISCUSSION
Murine Pur␣ and Pur␤ are structurally related proteins exhibiting ϳ70% sequence identity at the amino acid level (28). The highest level of similarity exists within the central modular region of each molecule where the minimal nucleic acid-binding domain resides (25,29,48). Interestingly, sequence elements within the more divergent amino and carboxyl termini have been implicated in providing stability and/or specificity of Pur␣ or Pur␤ interaction with certain protein and/or nucleic acid-binding partners (25,29). This has led us to speculate that certain nucleoprotein interactions modulated by Pur paralog-specific sequence elements may account for the distinct transcriptional properties ascribed to Pur␣ and/or Pur␤ in different cell types and promoter contexts (49 -60). However, because most cell-based studies to date have focused on a single Pur paralog, we decided to conduct an in-depth comparative analysis of the physical and functional properties of Pur␣ and Pur␤ in the context of SM␣A gene regulation in fibroblasts and VSMCs.
Experimental evidence from multiple laboratories indicates that Pur␣ and Pur␤ can associate with purine-rich ssDNA in a sequencespecific fashion as a monomer, dimer, or possibly higher order oligomers depending upon the sequence context of the target element and in vitro binding conditions (19,24,61). Self-association of Pur␣ (62) and interaction between Pur␣ and Pur␤ in the absence of ssDNA has also been reported (19,43), although the biophysical parameters of Pur oligomerization and relevance to nucleic acid binding remains to be determined. Nonetheless, based on these known physical properties, negative regulation of the MCAT enhancer in the SM␣A promoter could conceivably occur by either the independent or combined action of Pur␣ and Pur␤. To resolve this issue, we turned to RNA interference as a means of testing whether only one or both of the endogenous Pur proteins was required to maintain the full-length promoter in a repressed state in fibroblasts as suggested by previous cis-element mutation studies (18). Results indicated that, whereas Pur␤ appeared to play the dominant role in repression, Pur␣ also contributed to a significant extent under conditions in which Pur␤ expression was knocked down simultaneously and an intact Pur/Pyr element between Ϫ195 and Ϫ165 was present in the promoter (Figs. 1 and 2). These observations are consistent with earlier in vivo footprinting analyses that showed the genomic SM␣A region spanning Ϫ210 to Ϫ150 to be a hotspot for modification by chemical agents that selectively target unpaired bases (20). In short, new findings based on RNA interference-mediated loss-of-function offer proof of principle that Pur␣ and Pur␤ do indeed function as authentic repressors of the SM␣A promoter despite their preferential affinity and specificity for ssDNA sequences. Independent support for the notion that Pur proteins have the intrinsic ability to recognize and/or induce formation of non-B-DNA structures in the regulatory regions of certain genes is also provided by in vitro footprinting studies of functional Pur recognition elements located in the c-myc P1 and PDGF-A gene promoters (24,59).
Although functional cooperation between endogenous Pur␣ and Pur␤ is consistent with the previously proposed model of cryptic MCAT enhancer regulation (18), synergistic activation of the fulllength SM␣A promoter in response to combined knockdown of both factors in AKR-2B cells implied the possible unmasking of additional cis-elements and associated trans-activators under negative Pur control in fibroblasts (26,47). Such a scenario may explain the lesser response observed in primary VSMCs where knockdown of Pur␣ and/or Pur␤ resulted in only modest induction of the full-length SM␣A promoter.
An alternate possibility is that the steady-state levels of endogenous Pur␣ and Pur␤ were suboptimal for robust gene repression in differentiated VSMCs, a setting in which the SM␣A promoter would be expected to be in a constitutively activated state. To test this idea, we performed gain-of-function experiments and found that forced expression of Pur␤, but not Pur␣, repressed the full-length promoter and the minimal core SM␣A enhancer in primary mouse VSMCs and clonal rat A7r5 cells (Figs. 3 and 4). In the case of Pur␣ overexpression, definitive repressor activity was restricted to the core enhancer in AKR-2B MEFs in a manner consistent with gain-of-function analysis of Pur␣ in other fibroblast cell lines (26). These results suggested that exogenous Pur␣ and Pur␤ regulate transcription by cell type-and promoter contextspecific mechanism(s). In support of this contention, forced expression of Pur␣ actually increased core enhancer activity in growth-arrested or TGF␤1-treated A7r5 VSMCs but not in asynchronous or serum-treated cultures of the same cell type (Fig. 4). Moreover, in keeping with the neutral action of exogenous Pur␣ in asynchronous VSMCs, co-expression of Pur␣ had no substantive effect on Pur␤-mediated repression of the full-length SM␣A promoter in A7r5 cells (supplemental Fig. S5). This contrasted with the co-repressor activity demonstrated by MSY1 when co-expressed with Pur␤ in the same cell type (25). Taken together, findings based on gain-of-function and loss-of-function assays underscore the importance of taking into account cell type, culture conditions, promoter context, and cofactor requirements when attributing transcriptional regulatory activity to endogenous or exogenous Pur proteins.
To more rigorously define the molecular targets through which Pur proteins repress SM␣A promoter activity, we endeavored to characterize the interaction properties of purified recombinant Pur␣ and Pur␤ with the use of cell-free DNA and protein binding assays in vitro. In the case of DNA binding, experiments were conducted with biotinylated probes corresponding to specific cis-regulatory sequences present in SM␣A 5Ј-flanking region previously shown to serve as binding sites for cell extract-derived Pur␣/␤ by qualitative band shift or pull-down assays (18,26,47). Pilot studies indicated that reliable measurement of binding affinity of Pur␣ and Pur␤ for ssDNA in vitro requires that care be taken to prevent fortuitous nucleic acid contamination in recombinant protein preparations (supplemental Figs. S2 and S3). Although seemingly a mere technical issue, the presence of extraneous nucleic acid may explain some discrepancies in the literature regarding the reported binding affinity, specificity, and/or stoichiometry of Pur family members for certain single-stranded nucleic acid sequences (24, 25, 63, 64). Our quantitative analyses revealed that recombinant E. coli-derived Pur␣ and Pur␤ were essentially identical in terms of their strand preference and binding affinity toward each SM␣A promoter-derived element tested (Figs. 5 and 6). Consistent with our previous findings (25), and those of others (24), high affinity binding to ssDNA (apparent K d ϳ 1-2 nM) was dictated by the presence of nucleotide repeats of the general form (GGN) n , where N is not a G. At least two repeats were minimally required for ssDNA binding. Annealed dsDNA probes were poor interaction partners under the assay conditions utilized. Thus, the preponderance of evidence does not support a mechanism whereby disparities in cis-element recognition could account for functional differences observed between Pur␣ and Pur␤ in overexpression studies. However, these results do not exclude the possibility that one or more paralogspecific post-translational modifications such as phosphorylation (65) might significantly alter Pur DNA-binding affinity and specificity in a cellular milieu. In this regard, our report, implicating the lysine acetyltransferases cAMP-response element-binding protein (CREB)-binding protein (CBP) and p300 as co-activators of the core SM␣A enhancer in fibroblasts and VSMCs (42), points to site-specific acetylation as another modification of potential relevance to modulation of Pur␣ and/or Pur␤ structure-function.
Although resolution of the physiological significance of post-translational modification to the molecular biology of Pur proteins will require further investigation, the importance of protein-protein interaction as a significant contributor to the cellular effects of Pur␣ and/or Pur␤ is becoming more apparent. Reported Pur interaction partners fall into several functional classes and include gene regulatory factors such as human JC polyomavirus large T-antigen (66), E2F1 (67), YB-1/MSY1 (19,50), Sp1 (26,51), Sp3 (18), SRF (18), TEF-1 (18), Sox10 (68), heterogeneous nuclear ribonucleoprotein K (68), and Smad2/3 (26), cell cycle regulators, including certain cyclin/cdk complexes (38,69) and pRB (31), and proteins involved in RNA transport (70 -72). Some Pur binding partners such as YB-1/MSY1 and heterogeneous nuclear ribonucleoprotein K are known to play important roles at multiple levels of RNA metabolism, including transcription, transport, and translation. Thus, it very likely that the gene-regulatory properties of Pur proteins are not restricted to control of transcription initiation but may extend to other levels of information metabolism particularly mRNA translation (27,43,(73)(74)(75). In view of the expanding repertoire of regulatory factors that apparently interact with Pur␣ and/or Pur␤ and the functional distinctiveness exhibited by these proteins when overexpressed in different SM␣A-expressing cell types, an effort to compare the binding capacity of Pur␣ and Pur␤ for relevant fibroblast and VSMC-derived transcription factors seemed warranted.
An ELISA approach to assess protein-protein interaction was selected over more traditional co-immunoprecipitation assays, because a known amount of Pur protein could be used as the solid-phase probe. Practical and theoretical limitations of this approach include the capacity to detect low affinity interactions, reliance on immobilized rather than soluble Pur proteins as probes, and use of an antibody-based detection system that requires epitope recognition in solid-phase complexes. Because immobilization did not appear to alter their ability of Pur␣ or Pur␤ to interact specifically and with high affinity to known ssDNA target sequences, we felt justified in using an ELISA to assess proteinprotein interaction. However, we restricted our analysis to proteins implicated in SM␣A gene regulation and previously identified as Purinteracting factors by co-immunoprecipitation or pull-down assay (18,26). Unlike the results of comparative ssDNA binding, differences in the protein binding profile of purified recombinant Pur␣ and Pur␤ were obvious both within and between cell types (Fig. 7). Moreover, binding profiles changed depending on the subcellular source (cytosol versus nucleus) of extracted factors suggesting that functionally relevant Pur interactions, which affect promoter activity, may not be restricted to the nuclear compartment of a cell. Although these findings may simply be a reflection of the steady-state levels and subcellular distribution of specific regulatory proteins in AKR-2B MEFs versus A7r5 VSMCs, the existence of multiple transcription factor interaction partners reinforces an emergent concept that Pur repressor activity toward the SM␣A promoter is probably not determined by ssDNA binding alone (18,25,26). Accordingly, an alternative or supplemental mode of repressor action of Pur␣ and Pur␤ may entail protein-protein interaction-mediated sequestration of certain transcriptional activators in a way that obstructs their recognition of canonical DNA motifs and/or interaction with other co-activators (26,42).
Although mechanistically plausible, the concept that Pur repressors can impair activator function by virtue of direct physical interaction is predicated on the assumption that Pur proteins are present at intracellular concentrations sufficient to drive complex formation with transcription factors such as Sp1, Sp3, SRF, TEF-1, and/or Smad2/3. Experimentally, such conditions likely existed in cells transfected to overexpress Pur␣ or Pur␤ as evidenced by the fact that the core SM␣A enhancer was repressed (Figs. 3 and 4) despite the absence of 5Ј-nucleotides required for high affinity binding by Pur␣ and Pur␤ to the purinerich strand of the MCAT-containing Pur/Pyr element. Hence, at high local concentrations, such as those achieved in gain-of-function studies, Pur␣ or Pur␤ may disrupt cooperative interactions among TEF-1, SRF, and Sp1/3 thereby impairing efficient assembly of a composite enhanceosome complex (42). Importantly, support for the proposition that steady-state concentrations of endogenous Pur␣ and Pur␤ are adequate to allow inhibitory physical interaction with specific activators comes from a study of SM␣A enhancer induction during TGF␤1-induced myofibroblast differentiation (26). In this more physiologically germane model, Pur repressors appear to play a key role in restricting enhancer activity by interacting with Sp1/3 and Smad co-activators in a manner affecting protein occupancy at the THR and SPUR elements.
In conclusion, we propose that Pur␣ and Pur␤ act to limit SM␣A gene transcription in fibroblasts and VSMCs by a combination of protein-DNA and protein-protein interactions typified by 1) high affinity strand-specific binding to cis-regulatory elements exhibiting distinct purine/pyrimidine asymmetry and 2) cell type-and compartment-dependent binding to certain SM␣A-associated transcriptional regulatory proteins. A corollary of this two-pronged model is that the relative intracellular concentrations and localization of Pur␣ and Pur␤ will necessarily affect the stability and viability of inhibitory interactions with DNA and protein targets, and perhaps, the degree of SM␣A gene repression. Pur␣ and Pur␤ may thus serve as molecular rheostats allowing fibroblasts and VSMCs to meet the physiological demand for phenotypic plasticity during wound repair and arterial remodeling (76). Ex vivo analyses of mouse and human tissue specimens bolster this argument by providing circumstantial evidence that in vivo levels of Pur␣ and Pur␤ change in a manner consistent with progressive repression of SM␣A and ␣-myosin heavy chain gene expression in vascular and cardiac cell types during developmental or pathological remodeling of the heart (26,27,77). Identification of environmental cues that alter Pur␣ and/or Pur␤ expression, subcellular localization, and/or structure-function will be crucial to validating this conceptual model of SM␣A gene regulation.