Molecular Interactions between Single-stranded DNA-binding Proteins Associated with an Essential MCAT Element in the Mouse Smooth Muscle α-Actin Promoter*

Transcriptional activity of the mouse vascular smooth muscle α-actin gene in fibroblasts is regulated, in part, by a 30-base pair asymmetric polypurine-polypyrimidine tract containing an essential MCAT enhancer motif. The double-stranded form of this sequence serves as a binding site for a transcription enhancer factor 1-related protein while the separated single strands interact with two distinct DNA binding activities termed VACssBF1 and 2 (Cogan, J. G., Sun, S., Stoflet, E. S., Schmidt, L. J., Getz, M. J., and Strauch, A. R. (1995) J. Biol. Chem. 270, 11310–11321; Sun, S., Stoflet, E. S., Cogan, J. G., Strauch, A. R., and Getz, M. J. (1995) Mol. Cell. Biol. 15, 2429–2936). VACssBF2 has been recently cloned and shown to consist of two closely related proteins, Purα and Purβ (Kelm, R. J., Elder, P. K., Strauch, A. R., and Getz, M. J. (1997)J. Biol. Chem. 272, 26727–26733). In this study, we demonstrate that Purα and Purβ interact with each other via highly specific protein-protein interactions and bind to the purine-rich strand of the MCAT enhancer in the form of both homo- and heteromeric complexes. Moreover, both Pur proteins interact with MSY1, a VACssBF1-like protein cloned by virtue of its affinity for the pyrimidine-rich strand of the enhancer. Interactions between Purα, Purβ, and MSY1 do not require the participation of DNA. Combinatorial interactions between these three single-stranded DNA-binding proteins may be important in regulating activity of the smooth muscle α-actin MCAT enhancer in fibroblasts.

Eukaryotic gene transcription requires the coordinated assembly of upstream cis-element binding proteins, intermediary cofactors, and components of the basal transcription machinery into a multicomponent complex competent to initiate transcription. During this process, sequence-specific DNA-binding transcriptional activators and/or repressors play a pivotal role in modulating the cell-type specific expression of genes. While most such proteins bind to double-stranded DNA target sequences, a small but intriguing subclass has been identified that show enhanced affinity and specificity for either the sense or antisense strands of certain cis-regulatory elements required for promoter-specific activation (1)(2)(3)(4) or repression (5)(6)(7)(8)(9). We have recently cloned and identified two single-stranded DNA (ssDNA) 1 -binding proteins, Pur␣ and Pur␤, that interact with the purine-rich strand of an essential transcription control sequence upstream of the mouse vascular smooth muscle (VSM) ␣-actin gene promoter (10).
The involvement of ssDNA-binding proteins in VSM ␣-actin gene transcription was discovered as a consequence of promoter mapping studies that led to the identification of a conserved 5Ј-flanking sequence required for both activation and repression of promoter activity in fibroblasts and undifferentiated myoblasts (11,12). This proximal promoter element (PE) sequence (Ϫ195 to Ϫ165) exhibited polypurine-polypyrimidine asymmetry, an inverted muscle-specific MCAT (AGGAATG) enhancer element, and bound at least three distinct DNA binding activities in a sequence and strand-specific manner. The two ssDNA binding activities, formerly designated vascular actin single-strand binding factors, VACssBF 1 and 2, appeared to play a role in repression (11) while a transcription enhancer factor 1-related protein was implicated in activation (12,13). Although the mechanism of repression remains to be formally established, a hypothetical model involving VACssBFmediated disruption of MCAT element base pairing and competition for transcription enhancer factor 1 binding was proposed (11). Interestingly, an additional binding site for VACssBF2 was later identified on the purine-rich coding strand of a GGAATG-containing sequence element located in a downstream VSM ␣-actin exon (14). This coding element sequence functioned as a VACssBF2-dependent repressing element when positioned 5Ј and adjacent to a transcription enhancer factor 1-or activator protein 1-dependent enhancer element in chimeric promoter constructs (14). Because the noncoding strand of the coding element sequence lacked detectable VACssBF1 binding affinity (14), these data suggested that VACssBF2 binding was necessary and sufficient for repression. Screening of a mouse lung cDNA expression library with the exonic VACssBF2-binding site ultimately resulted in the isolation of two clones encoding the purine-rich ssDNA-binding proteins, Pur␣ and Pur␤ (10). Biochemical analyses of the cloned mouse Pur proteins expressed in fibroblasts confirmed that Pur␣ and Pur␤ corresponded to the p46 and p44 components of VACssBF2 that bind to the purine-rich strand of the PE and presumably down-regulate VSM ␣-actin gene expression (10).
In the present study, we used a similar binding site screen to confirm the identity between VACssBF1 and the mouse Y-box * This work was supported by National Institutes of Health Grant R01 HL54281 (to M. J. G.), Minnesota Affiliate of the American Heart Association Grant MN-97-F-20 (to R. J. K.), 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  protein, MSY1. By utilizing recombinant proteins and isoformspecific immune reagents, we demonstrate highly specific protein-protein interactions between Pur␣, Pur␤, and MSY1 which seem likely to have functional significance. These include the binding of Pur proteins to ssDNA as both homo-and heterodimeric complexes and the formation of heterotrimeric complexes between all three proteins in the absence of DNA. These data suggest that protein-protein interactions between Pur␣, Pur␤, and MSY1 play an important role in regulating transcriptional activity of the VSM ␣-actin gene in fibroblasts.

EXPERIMENTAL PROCEDURES
Cloning of a cDNA Encoding a PE-MCAT Strand Binding Protein, MSY1-A mouse lung cDNA expression library (Stratagene) was screened for cDNA-encoded proteins that interact with the pyrimidinerich strand of the VSM ␣-actin MCAT enhancer element using a binding site cloning methodology described previously (10). Eight independent clones were isolated from 250,000 plaques initially screened. DNA sequencing by semi-automated dideoxy termination indicated that all eight clones encoded the mouse Y-box protein, MSY1 (15).
Construction of Bacterial Expression Vectors and Purification of 6xHis-tagged Pur␣, Pur␤, and MSY1-The cDNAs encoding the open reading frame minus the start methionine of mouse Pur␣, Pur␤ (10), and MSY1 (clone 7-1, this study) were amplified by polymerase chain reaction using primers which generated 5Ј BamHI and 3Ј KpnI cloning sites. The polymerase chain reaction products were gel purified, cut with restriction enzymes, and subcloned into pQE-30 (Qiagen) to generate fusion constructs encoding a N-terminal 6xHis tag. The resultant plasmids were transformed into Escherichia coli strain JM109 and the orientation and fidelity of the polymerase chain reaction-amplified cDNA inserts were determined by DNA sequencing. For protein preparation, 1 liter of terrific broth containing 100 g/ml ampicillin was inoculated (1:50) with an overnight bacterial culture and incubated for 5-7 h at 37°C. Recombinant protein synthesis was induced by the addition of isopropyl-␤-D-thiogalactopyranoside to 2.0 mM and an additional 4-h growth period. E. coli were collected by centrifugation at 5000 ϫ g for 10 min and resuspended in 14 ml of cold 50 mM sodium phosphate, pH 8.0, 300 mM NaCl (lysis buffer) supplemented with 0.5 mM phenylmethylsulfonyl fluoride, and 0.5 g/ml each pepstatin, leupeptin, and aprotinin. Cells were lysed by sequential 30-min incubations on ice with lysozyme added to 1.0 mg/ml followed by Triton X-100 added to 0.13% (v/v). Lysates were centrifuged at 100,000 ϫ g for 30 min. Supernatants were collected, combined with 4 ml of packed Ni-NTA agarose resin (Qiagen) equilibrated in lysis buffer, and mixed overnight at 4°C. The resin was washed sequentially at room temperature with 50 mM sodium phosphate, pH 8.0, buffer containing 0.3, 1.0, and 2.0 M NaCl. The resin was then packed into a column and washed with 50 mM sodium phosphate, 2.0 M NaCl, pH 8.0, until the A 280 nm of the flow-through was Յ0.02. His-tagged protein was eluted with a step gradient of 50 mM phosphate, 2.0 M NaCl containing 20, 40, 80, and 500 mM imidazole. Fractions were analyzed by SDS-polyacrylamide gel electrophoresis and Coomassie Blue R-250 staining. Pur protein or MSY1-enriched fractions were pooled, dialyzed versus 50 mM sodium phosphate, 1.0 M NaCl, pH 8.0, and chromatographed a second time on Ni-NTA agarose to ensure optimal purity. His-tagged protein eluates were dialyzed against 25 mM HEPES, 0.5 M NaCl, pH 8, aliquoted, and stored at Ϫ20°C. Recombinant proteins were also purified as above but in buffers supplemented with 8 M urea. Purification under such conditions enhanced the yield of His-tagged protein. No differences were observed in the in vitro protein-binding properties between His-tagged proteins purified under nondenaturing or denaturing conditions after dilution and/or dialysis into aqueous buffer. Recombinant protein concentration was estimated by optical density measurement using molar extinction coefficients and molecular weights of 18,610 and 35,000 for Pur␣, 18,610 and 34,000 for Pur␤, and 26,170 and 36,000 for MSY1. In some cases, proteins were quantified by dye-binding assay (Bio-Rad) using bovine serum albumin as a standard.
Preparation of Peptide-specific Polyclonal Antibodies against Mouse Pur␣, Pur␤, and MSY1-Peptides corresponding to amino acids 42-69, 210 -229 and 302-324 of mouse Pur␤ (10), 149 -175 and 291-313 of mouse Pur␣ (10), and 85-110, 139 -165, 242-267, and 276 -302 of MSY1 (this study) were synthesized using modified Merrifield solidphase chemistry and purified by reverse-phase high performance liquid chromatography by the Mayo Protein Core Facility. The composition of each peptide was confirmed by amino acid analysis. Each peptide was synthesized with a cysteine residue at either the N or C terminus to facilitate coupling to maleimide-activated KLH (Pierce) and iodoacetylagarose (Sulfolink, Pierce). KLH-coupled peptides were used as immunogens and rabbit polyclonal antisera production was carried out by a commercial vendor (Cocalico) using a 60-day standard protocol. Polyclonal rabbit IgGs were affinity purified using peptide-coupled agarose columns. Briefly, whole IgG was enriched from rabbit antisera by 40% ammonium sulfate precipitation. Following centrifugation for 10 min at 5000 ϫ g, the IgG-rich pellet was reprecipitated, dissolved in phosphate-buffered saline, and dialyzed. Rabbit IgG was then applied to the appropriate 2-ml peptide-agarose (0.5-1.0 mg of peptide/ml) column equilibrated with phosphate-buffered saline. The flow-through fraction was collected and reapplied. After washing the column with phosphatebuffered saline, peptide-bound IgG was eluted with 0.1 M glycine, pH 2.5, and immediately neutralized with 1 M Tris, pH 9.5. Affinity purified IgG was precipitated by the addition of solid ammonium sulfate to 75% saturation. The pellet was collected by centrifugation, dissolved in 50% (v/v) glycerol/phosphate-buffered saline, and stored at Ϫ20°C. Rabbit IgG from pooled preimmune serum was purified on Protein A/G-agarose (Calbiochem). IgG concentration was estimated by optical density measurement based upon a molar extinction coefficient and molecular weight of 210,000 and 150,000, respectively.
Screening of Peptide Affinity-purified Antibodies by Western (Immuno) blotting-Cellular Pur␣, Pur␤, or MSY1 were enriched from 1 mg of AKR-2B fibroblast nuclear protein by selective capture on biotinylated-ssDNA (PE-F for Pur proteins and PE-R for MSY1) coupled streptavidin-paramagnetic particles as described previously (10). DNA-bound proteins were eluted with 1% SDS, resolved on a 10% (29:1) polyacrylamide mini-curtain gel, and electrotransferred to a polyvinylidene difluoride membrane (Immobilon-P) for 90 min (300 mA) in 25 mM Tris, 192 mM glycine, 20% (v/v) methanol at 4°C. After blocking overnight in 25 mM Tris, 150 mM NaCl, pH 7.5 (TBS), with 5% (w/v) Carnation nonfat dry milk at 4°C, the membrane was fitted into a multiscreen apparatus (Mini-PROTEAN II, Bio-Rad) and selected channels were loaded with 0.6 ml of rabbit anti-Pur or MSY1 peptide antibody diluted to 2.0, 0.5, and 0.1 g/ml in 2% nonfat dry milk/TBS. Following a 1-h incubation at ambient temperature with gentle mixing, the antibody solutions were aspirated, and each channel was washed once with TBS, containing 0.05% Tween 20 (TBST). The initial wash solution was aspirated, the apparatus disassembled, and the entire blot washed three more times (5 min/25-ml wash). Goat anti-rabbit IgG-HRP (Santa Cruz) diluted 1:2000 in TBST was then applied for 1 h. The blot was washed four times (30 min total) and chemiluminescence reagent (ECL, Amersham) was applied for 1 min. Immune complexes were visualized on x-ray film (XAR-5, Kodak) following a 5-10-s exposure.
Electrophoretic Mobility Shift Assay for Protein-DNA Binding-Band shift assays were performed as described previously (11,14). For antibody supershift experiments, rabbit IgGs (0.25-1.0 g) were preincubated for 20 min with AKR-2B nuclear protein (10) in binding buffer containing poly(dI-dC) (11). A 32 P-ssDNA probe corresponding to the purine-rich coding strand of the PE was then added (ϳ1 nM final) and mixtures were incubated for an additional 20 min prior to electrophoresis on a 6% nondenaturing polyacrylamide gel.
ELISA for Protein-Protein Interaction-His-tagged Pur␤, MSY1, or dihydrofolate reductase (DHFR)-coated microtiter wells (50 nM application as described above) were incubated with varying amounts of AKR-2B nuclear protein (10) (100 l/well) diluted in binding buffer (HBST with 0.1% bovine serum albumin) for 16 -18 h at 4°C. Wells were aspirated and washed 3 times with HBST and 100 l of anti-Pur or anti-MSY1 peptide IgG diluted to 1.0 g/ml in binding buffer was applied for 1 h at room temperature. Primary antibody solution was aspirated and wells were washed three times with HBST. Goat antirabbit IgG-HRP (Santa Cruz) diluted 1:2000 in HBST was then applied for 1 h. Secondary antibody solution was aspirated and wells were washed four times with HBST. Immune complexes were detected using 100 l of ABTS chromogenic substrate. Absorbance readings at 405 nm were determined after 45 min.
Immunoprecipitation Assay for Protein-Protein Interaction-Mouse AKR-2B fibroblasts were transiently transfected with mouse Pur␣ and Pur␤ expression vectors as described previously (10). All subsequent steps including cell synchronization, serum-stimulation, harvest, extraction, and protein assay have been detailed previously (10,14). Whole cell protein extract (100 g) from transfected cells or nuclear extract from nontransfected rapidly growing AKR-2B fibroblasts (10) was combined with 2.5 g of selected rabbit anti-Pur or MSY1 peptide IgGs in a final volume of 250 l. After a 1-h incubation at room temperature, ϳ10 7 sheep anti-rabbit IgG-coupled magnetic dynabeads (Dynal) were added and the mixtures incubated for an additional 90 min. In some experiments, goat anti-rabbit IgG-biotin (Santa Cruz) coupled to streptavidin-coated paramagnetic beads (Promega) were used. The beads were then captured with a magnet and washed three times with HBS. Rabbit IgG-bound protein was specifically eluted by adding a vast excess of free peptide (20 l at 50 M) and incubating for 30 min at room temperature. Eluates were supplemented with Laemmli SDS sample preparation buffer and 5% (v/v) ␤-mercaptoethanol, and subjected to electrophoresis on a 10% (29:1) polyacrylamide mini-gel. Immunoprecipitates were evaluated for the presence Pur proteins and MSY1 via immunoblotting as described above.

RESULTS
Development and Characterization of Isoform-specific Immune Reagents to Pur␣ and Pur␤-A repertoire of immune reagents based upon cDNA-deduced amino acid sequences of Pur␣ and Pur␤ were produced to assist in defining determinants of protein-DNA and putative protein-protein interactions. Synthetic peptides corresponding to both conserved and unique sequences within the Pur proteins were used as immunogens in rabbits (Fig. 1). IgGs were enriched from rabbit antisera and then subjected to affinity purification using peptide-coupled agarose columns. The resultant affinity purified IgGs were tested for reactivity using both recombinant (Histagged) and cellular Pur␣ and Pur␤ (Figs. 2 and 3). Assessment of antibody binding to immobilized recombinant Pur␣ and Pur␤ by ELISA (Fig. 2) indicated that several antibodies possessed remarkable specificity for either Pur␣ (anti-A291-313) or Pur␤ (anti-B210 -229 and anti-B302-324) while another antibody directed against a conserved region (anti-B42-69) cross-reacted with Pur␣ and Pur␤ (Fig. 2). The specificity of these antibodies was also evaluated by Western blotting of cellular Pur proteins enriched from an AKR-2B fibroblast nuclear extract by selective capture on paramagnetic particles coupled with the purine-rich strand of the VSM ␣-actin MCAT element, PE-PrMss, or PE-F (10). Consistent with ELISA data, anti-B42-69 demonstrated similar reactivity toward fibroblast-derived Pur␣ (p46, M r ϳ 46,000) and Pur␤ (p44, M r ϳ 44,000) which migrate as a closely spaced doublet (Fig. 3, lanes 4 -6). Anti-A291-313 recognized only the slower migrating Pur␣ (p46) isoform (Fig. 3, lanes 16 -18) while both anti-B210 -229 and anti-B302-324 preferentially detected the faster migrating Pur␤ (p44) isoform (Fig. 3, lanes 10 -12 and 13-15). As expected, preimmune rabbit IgG failed to detect the Pur proteins in both screening assays.
Effect of Anti-Pur Antibodies on Protein-DNA Complex Formation-Band shift assays were conducted in the presence of anti-Pur peptide antibodies to confirm the identity of Pur protein-ssDNA complexes previously suggested by overexpression studies (10). Initial experiments utilized nuclear protein from AKR-2B fibroblasts as a source of cellular Pur␣ and Pur␤ and Purified 6xHis-tagged mouse Pur␣ or Pur␤ at 50 nM was applied to polystyrene microtiter wells (100 l/well) and incubated 16 h at 4°C. Following washing and blocking steps, rabbit anti-mouse Pur peptide IgG (1.0 -0.016 g/ml, 100 l/well) was applied to the Pur protein-coated wells. After a 2-h incubation at room temperature, solid-phase immune complexes were detected by ELISA using a goat anti-rabbit IgG-HRP conjugate as the secondary antibody. a 32 P-oligonucleotide probe corresponding to the purine-rich coding strand of the PE (PE-PrMss or PE-F) (11). As shown in earlier studies, Pur␣ and Pur␤ -ssDNA complexes migrate as a closely spaced doublet (Fig. 4A, lane 2). A slower migrating complex (NS) that is also detected is composed of an unrelated, nonspecific DNA-binding protein (11,14). As illustrated in Fig.  4A, three out of the four anti-Pur antibodies tested (lanes 4 -6) were found to selectively supershift the two major Pur protein-ssDNA complexes into two slower migrating complexes, designated SS1 and SS2. These supershifted complexes were not formed when preimmune IgG or anti-B42-69 were included in the reaction mixtures (lanes 2 and 3). The inability of anti-B42-69 to supershift suggests that this antibody is unable to bind Pur proteins in their native (i.e. nondenatured) state since immobilization on polystyrene (Fig. 2) or denaturation by SDS (Fig. 3) did not interfere with epitope recognition.
Closer inspection of the band shift patterns obtained using isoform-specific antibodies provided unexpected results. While anti-A291-313 appeared to only partially supershift the major Pur␣-ssDNA complex (Fig. 4A, lane 6, arrow), a minor and slower migrating complex (denoted by a *) was completely supershifted by this antibody. Moreover, anti-B210 -229 and anti-B302-324 clearly supershifted the most rapidly migrating Pur␤ complex and, surprisingly, the major (middle) Pur␣-containing complex as well (Fig. 4A, lanes 4 and 5). These data suggested that the major (middle) Pur␣ complex is heterogeneous and likely contains Pur␤ while the minor, slowest migrating complex is composed exclusively of Pur␣. Owing to the low abundance of this minor Pur␣-containing complex, we also performed antibody supershift analyses using extracts from AKR-2B fibroblasts transfected with mouse Pur␣ and/or Pur␤ expression vectors (10). As shown in Fig. 4B, overexpression of Pur␣ enhanced formation of the minor (slowest migrating) Pur␣-ssDNA complex (lane 2, *) which was supershifted by anti-A291 (lane 5) but not by anti-B302 (lane 8). Overexpression of Pur␤ enhanced formation of the major (most rapidly migrating) Pur␤-ssDNA complex (Fig. 4B, lane 3) which was supershifted by anti-B302 (lane 9) but not by anti-A291 (lane 6). Co-expression of both proteins enhanced formation of the two major (middle and most rapidly migrating) Pur protein complexes but not the minor (slowest migrating) Pur␣ complex (Fig. 4B, lane 4). Both major complexes were efficiently super-shifted by anti-B302 (compare lanes 4 and 10) while only the middle Pur␣/␤ complex was effected by anti-A291 (compare lanes 4 and 7). Importantly, formation of all three Pur protein-ssDNA complexes was abolished by low concentrations of sodium deoxycholate, a mild ionic detergent known to disrupt protein-protein interactions (16) (Fig. 4C, lanes 2-5). In contrast, the nonspecific ssDNA-binding complex (NS) was largely unaffected by deoxycholate while the Pur protein-ssDNA complexes were unaffected by Triton X-100 (lanes 6 -9). It is important to note that no faster migrating ssDNA complex was detected in samples treated with deoxycholate implying that monomeric binding of the Pur proteins to ssDNA does not occur under these conditions. Together, these data provide strong evidence that mouse Pur␣ and Pur␤ bind to the purine-rich strand of the MCAT enhancer in the form of homo-and heterodimers. No evidence was obtained for the existence of monomeric ssDNA-binding complexes. These data are consistent with a recent finding that human Pur␣ binds to a recognition element in the myelin basic protein gene promoter as a homodimer (17) and suggest that dimerization between Pur proteins may be a necessary prerequisite for functional activity.
Pur␣ and Pur␤ Associate in the Absence of ssDNA-To determine whether Pur protein dimerization requires coincident interaction with a ssDNA recognition element, we evaluated the ability of the Pur proteins to associate in the absence of ssDNA. Initially, we performed immunoprecipitation experi- DNA-bound proteins were eluted, resolved on a 10% SDS-polyacrylamide mini-curtain gel, and electrotransferred to a polyvinylidene difluoride membrane. Pur␣ (p46 band) and Pur␤ (p44 band) were detected by immunoblotting using a multiscreen apparatus (Bio-Rad) in which selected channels were loaded with anti-Pur peptide antibody diluted to 2.0, 0.5, and 0.1 g/ml.

FIG. 4. Anti-Pur peptide antibodies selectively supershift specific Pur protein-ssDNA complexes (A and B) and deoxycholate blocks Pur protein-ssDNA complex formation (C).
A, B, and C, band shift assays were performed using the purine-rich forward (coding) strand of the PE (PE-F, ϳ1 nM) as the 32 P-ssDNA probe (11). A, the indicated rabbit IgGs (0.5 g) were preincubated for 20 min with AKR-2B nuclear protein (3 g). Probe was then added and mixtures were incubated for an additional 20 min prior to electrophoresis. B, the indicated rabbit IgGs (0.5 g) were preincubated for 20 min with lysed whole cell protein (2 g) obtained from AKR-2B fibroblasts transiently transfected with pCI-Pur␣ (10 g), pCI-Pur␤ (10 g), or both (5 g each) expression vectors (10). Reaction mixtures were supplemented with probe and incubated as above. Lane 1 contains lysed cell protein from empty vector (pCI) transfected cells and thus represents ssDNA-binding of endogenous Pur proteins. C, AKR-2B nuclear protein (2 g) was preincubated for 20 min with varying concentrations of sodium deoxycholate (DOC) or Triton X-100 (TX-100) prior to the addition of the ssDNA probe. Arrows and labels designate Pur␣, Pur␤, and antibody supershifted (SS1 and SS2) protein-ssDNA complexes. The "*" highlights a minor Pur␣-containing complex (A) which is enhanced upon forced expression of Pur␣ (B). ments using whole cell extracts of AKR-2B fibroblasts transiently co-transfected with Pur␣ and Pur␤ expression vectors (10). Isoform-specific anti-Pur IgGs were used to specifically capture Pur proteins from the cell extract while anti-B42-69, which cross-reacts with both Pur␣ and Pur␤, was employed to assess the composition of immunoprecipitates by Western blotting. As shown in the top panel of Fig. 5, both anti-B210 -229 (lane 3) and anti-B302-324 (lane 4) were found to co-immunoprecipitate Pur␣ and Pur␤. Anti-A291-313 was also able to co-immunoprecipitate Pur␣ and Pur␤ (Fig. 5, lane 5) albeit with reduced efficiency relative to the anti-Pur␤ antibodies (lanes 3 and 4). As expected, preimmune rabbit IgG failed to immunoprecipitate the Pur proteins (Fig. 5, lane 2). Owing to the isoform specificity of the precipitating antibodies, these results indicate that Pur␣ and Pur␤ can associate via proteinprotein interaction.
As a more quantitative test for Pur protein interaction, we examined the ability of recombinant Pur␤ to selectively bind cellular Pur␣ in a discontinuous, solid-phase binding assay. Polystyrene microtiter wells coated with His-tagged Pur␤ or His-tagged DHFR were incubated with varying amounts of nuclear protein extracted from rapidly growing AKR-2B fibroblasts. After removal of unbound nuclear protein, solid-phase Pur␣-Pur␤ complexes were detected by ELISA using a Pur␣specific antibody, anti-A291-313. As shown in the bottom panel of Fig. 5, the colorimetric signal obtained from Pur␤-coated wells was dose-dependent and saturable. The specificity of this interaction is demonstrated by the total absence of color generated by DHFR-coated control wells. These data reinforce the conclusion drawn from the immunoprecipitation and band shift experiments and confirm that Pur␣ and Pur␤ can form a specific and stable protein-protein complex. Moreover, these data demonstrate that the Pur-protein complex formation does not require the presence of exogenous ssDNA.
Identification of MSY1 as a PE-MCAT Strand-binding Protein-Screening of a mouse lung cDNA expression library with a 32 P-end-labeled tetramer of the pyrimidine-rich strand of the PE yielded eight independent PE-MCATss-binding clones. Each clone was tested for its ssDNA binding specificity using wild-type and mutant oligonucleotide competitors in a tertiary filter binding assay. All eight clones produced identical results. Fig. 6 shows the data for one of the clones, 7-1, where excess wild-type oligonucleotide (PE-MCATss) completely inhibited the binding of the 32 P-tetramer in comparison to a mutant oligonucleotide (PE-MCATmu2) lacking VACssBF1 binding affinity (11). DNA sequencing revealed that each clone contained overlapping nucleotide sequences that were virtually identical to the cDNA sequence encoding MSY1 previously reported by Tafuri and co-workers (15). The full-length cDNA sequence of clone 7-1 and the published MSY1 cDNA sequence differ by only a single nucleotide within the open reading frame. Alignment of the deduced amino acid sequences illustrates that clone 7-1 encodes a glycine residue rather than an alanine residue at codon 29 (Fig. 7). The reason for this discrepancy is probably due to a polymorphism although the glycine codon is conserved in the rat and human Y-box homologues, EFI A (18), dpbB (19), and YB-1 (20) (Fig. 7).
Synthetic peptides corresponding to several sequences of predicted antigenicity were used as immunogens to derive MSY1-specific antibodies. Affinity purified rabbit IgGs were tested for specificity as described for the panel of anti-Pur antibodies (data not shown), and two, anti-MSY242-267 and anti-MSY276 -302, were selected for use in further experiments. Importantly, neither antibody exhibited detectable cross-reactivity with either of the Pur proteins. These antibodies were used to test whether cellular, as opposed to recombinant, MSY1 would bind to the pyrimidine-rich strand of the MCAT enhancer. As shown in Fig. 8, an MSY1 immunoreactive Transfectants were rendered quiescent and serum-stimulated for 6 h. Cell extracts were prepared and 100 g of lysed cell protein was supplemented with 2.5 g of the indicated rabbit IgG and incubated for 1 h. Following an additional incubation with sheep anti-rabbit-coupled magnetic beads, immune complexes were captured on a magnet, washed, and rabbit IgG bound protein eluted with free peptide. Eluates were analyzed by immunoblotting using anti-B42-69 to detect Pur␣ and Pur␤. Lane 1 shows Pur protein immunoreactivity observed in 2 g of whole cell extract prior to immunoprecipitation. Bottom, His-tagged Pur␤ and DHFR-coated microtiter wells (50 nM application) were incubated with varying amounts of AKR-2B nuclear protein for 16 h at 4°C. Wells were aspirated, washed 3 times, and solid-phase Pur␣-Pur␤ complexes were detected by ELISA using anti-PurA 291-313 as the primary antibody. Absorbance readings at each point were corrected by subtracting a background A 405 nm reading generated with His-tagged protein-coated wells and binding buffer. Screening of a mouse lung cDNA expression library (250,000 plaques) with a 32 P-end labeled tetramer of the pyrimidine-rich strand of the PE yielded eight independent PE-MCATss-binding clones. Each clone was tested for its ssDNA-binding specificity in a filter binding assay. All eight clones produced identical results. The data for one of the clones, 7-1, is shown. Excess (150-fold molar) wild-type oligonucleotide (right side) completely inhibited the binding of the 32 P-tetramer while a mutant oligonucleotide deficient in VACssBF1 binding did not (left side). species was captured from a crude AKR-2B fibroblast nuclear extract by paramagnetic particles coupled with the pyrimidinerich, reverse strand of the enhancer (PE-R, lane 3) but not by particles coupled with the opposing, purine-rich forward strand (PE-F, lane 2). In contrast, the purine-rich forward strand, but not the pyrimidine-rich reverse strand, effectively captured Pur␣ and Pur␤ (compare lanes 5 and 6). These data validate the expression cloning results and provide strong evidence that the ssDNA-binding complex previously termed VACssBF1 (11,12), is identical, or closely related to the mouse Y-box protein, MSY1. The anomalous electrophoretic mobility of MSY1 (M r ϳ55,000) is consistent with previous findings for other Ybox proteins (21).
Interaction of Pur␣ and Pur␤ with MSY1-The identification of MSY1 as a pyrimidine-rich strand, VSM ␣-actin MCAT enhancer-binding protein was particularly intriguing given a previous report implicating the human homolog, YB-1, as a transient Pur␣-binding protein in the context of a different promoter element (22). As an independent evaluation of the potential for protein-protein interaction between mouse Pur␣ and/or Pur␤ and MSY1, quantitative binding studies were conducted with purified, recombinant proteins. The binding of cellular Pur␣ and Pur␤ to His-tagged MSY1 passively immobilized on polystyrene microtiter wells was evaluated by ELISA. Fluid-phase AKR-2B nuclear protein was incubated with both MSY1 and DHFR-coated wells. After removal of unbound nuclear protein, solid-phase Pur␣-MSY1 and Pur␤-MSY1 complexes were detected using antibodies that specifically recognize the C-terminal region of either Pur␣ or Pur␤. While virtually no signal was obtained from wells coated with DHFR, the colorimetric signal generated with MSY1-coated wells was dose-dependent and saturable (Fig. 9A). Similar results were obtained when the assay was performed using Histagged Pur␣ or Pur␤ as the solid-phase ligands and an MSY1specific antibody to detect complex formation (Fig. 9B). Although MSY1 binding was not completely saturable in this orientation (owing to decreased accessibility of binding sites, lower affinity of the detecting antibody, and/or limiting fluidphase MSY1), the immobilized Pur proteins were nonetheless indistinguishable in terms of their ability to specifically partner with cellular MSY1. These data demonstrate that both Pur␣ and Pur␤ can form a stable, heterotrimeric complex with MSY1 in the absence of ssDNA.
To test whether or not such a heterotrimeric complex could be detected in a nuclear extract without disturbing the equilibrium by exposure to immobilized recombinant ligand, immunoprecipitation experiments were performed with anti-Pur and anti-MSY1 antibodies and a nuclear extract from nontransfected AKR-2B fibroblasts. The Pur protein and MSY1 composition of each immunoprecipitate was analyzed by Western blotting using anti-PurB42-69 to detect both Pur isoforms, followed by anti-MSY276 -302 to detect MSY1. Consistent with the results obtained using extracts from transfected fibroblasts (Fig. 5), Pur␣ and Pur␤ were co-immunoprecipitated by the Pur␤-specific antibodies, anti-B210 -229 and anti-B302-324 (Fig. 10, lanes 1 and 2). The Pur␣-specific antibody, anti-A291-313, preferentially captured Pur␣ and a marginal amount of Pur␤ (lane 3). Importantly, an anti-MSY1 immunoprecipitate (M242-267) included Pur␣ and a small but detectable amount of Pur␤ (lanes 6). Subsequent immunoblotting for MSY1 revealed the presence of an additional M r ϳ 55,000 band (MSY1) in the anti-MSY242-267 and anti-B302-324 immunoprecipitates (lanes 7 and 8) but not the anti-B210 -229 and anti-A291-313 immunoprecipitates (data not shown). The M r ϳ 60,000 band present in all lanes, except the no IgG control lane (Fig.  10, lane 4), is a nonspecific band (NS) likely corresponding to the heavy chain of rabbit IgG. It is noteworthy that these heterodimeric and heterotrimeric complexes composed of Pur␣-Pur␤ and Pur␣-Pur␤-MSY1 could be detected without manipulating the cellular levels of Pur proteins and/or MSY1 via the use of expression vectors.  (19), and human YB-1 (20) are shown. Clone 7-1 encodes a glycine residue rather than an alanine residue at codon 29. This glycine residue is conserved in the other Y-box homologues. The highly conserved "cold shock" or DNAbinding domain is underlined.

DISCUSSION
Because Pur␣ and/or Pur␤ appear to function as repressors of VSM ␣-actin MCAT enhancer activity (11,14) and because Pur␤ has not previously been characterized, we felt it important to study potential molecular interactions between these two ssDNA-binding proteins. Recombinant mouse Pur proteins and rabbit polyclonal antibodies directed against specific domains both common and unique to Pur␣ and Pur␤ were used in these studies (Figs. 1-3). Surprisingly, we found that several antibodies specific for the Pur␤ isoform supershifted protein-ssDNA complexes composed of both Pur␣ and Pur␤ implying that the Pur proteins can transiently associate via proteinprotein interaction (Fig. 4). This conclusion was validated by both immunoprecipitation and ELISA-based, protein-protein binding experiments which indicated that a specific and stable Pur␣-Pur␤ complex can form in the absence of a ssDNA recognition element (Fig. 5). This interaction likely underlies the previously unrecognized ability of Pur␣ and Pur␤ to form het-erodimeric ssDNA-binding complexes and is likely of consequence given the diverse functional roles attributed to Pur␣. For example, Pur␣ has been previously implicated in DNA replication of several viral genomes (23)(24)(25) and transcriptional activation of both viral (22,26) and mammalian promoters (27)(28)(29)(30)(31)(32). In contrast, our results suggest that mouse Pur␣ and/or Pur␤ are able to repress the activity of a nearby enhancer in the context of both natural and chimeric promoters (11,14). While it is not uncommon for a transcription factor to function as either an activator or repressor depending on promoter context, it is equally likely that the properties of a homodimeric Pur␣ ssDNA-binding complex differ substantially from a homodimeric Pur␤ complex, or from a Pur␣/Pur␤ heterodimer. This could easily explain why the VSM ␣-actin MCAT enhancer is negatively regulated despite the presence of a transcriptional activator like Pur␣. This possibility is lent additional credence by the structural differences between Pur␣ and Pur␤, most notably the absence of a C-terminal polyglutamine sequence, a potential transactivation domain, in Pur␤ ( Fig. 1).
In an earlier study, a binding site screen of a human astroglioma cell cDNA expression library was used to tentatively identify a member of the Y-box family of nucleic acid-binding proteins as the pyrimidine-rich ssDNA binding activity previously termed VACssBF1 (33). In the present study, we confirmed this result using a cDNA library from a mouse tissue enriched in smooth muscle. This screen yielded eight independent clones, all encoding the mouse Y-box protein MSY1. Moreover, MSY1 was selectively captured from fibroblast nuclear extracts by the pyrimidine-rich strand of the MCAT enhancer coupled to paramagnetic beads (Fig. 8). Together, these studies provide convincing evidence that MSY1 does indeed represent the ssDNA binding activity which interacts with the strand of the MCAT enhancer opposing the Pur protein recognition site. Thus, it may be highly significant that MSY1 specifically interacts with both Pur␣ and Pur␤ in vitro (Fig. 9), and indeed, can be co-immunoprecipitated from fibroblast nuclear extracts in the form of an MSY1-Pur protein complex (Fig. 10).
While the functional significance of these interactions to transcriptional regulation of the VSM ␣-actin gene remains to be established, it is noteworthy that the human Y-box protein homologue, YB-1, has been shown to similarly interact with Pur␣ to reciprocally modulate each others binding to the JC polyomavirus lytic control element (22). Similar cooperative interactions seem likely to occur within the context of the FIG. 9. Detection of cellular Pur␣ and Pur␤ binding to recombinant MSY1 by ELISA. A, His-tagged MSY1 and DHFR-coated microtiter wells (50 nM application) were incubated with varying amounts of AKR-2B nuclear protein for 16 h at 4°C. Wells were aspirated, washed 3 times, and solid-phase Pur␣-MSY1 or Pur␤-MSY1 complexes were detected by ELISA using anti-PurA 291-313 or anti-PurB 302-324, respectively, as the primary antibodies. B, His-tagged Pur␣, Pur␤, or DHFR-coated microtiter wells (50 nM application) were incubated with varying amounts of AKR-2B nuclear protein for 16 h at 4°C. Wells were aspirated, washed 3 times, and solid-phase MSY1-Pur␣ or MSY1-Pur␤ complexes were detected by ELISA using anti-MSY1 242-267 as the primary antibody. Absorbance readings at each point were corrected by subtracting a background A 405 nm reading generated with His-tagged protein-coated wells and binding buffer. FIG. 10. Immunoprecipitation of a Pur␣-Pur␤-MSY1 complex from an AKR-2B fibroblast nuclear extract. Nuclear extract from rapidly growing AKR-2B fibroblasts (136 g of protein) was combined with selected rabbit anti-Pur or MSY1 peptide IgGs (2.5 g) and incubated for 1 h. Following an additional incubation with goat anti-rabbit IgG-biotin coupled streptavidin-paramagnetic particles, immune complexes were captured with a magnet, washed, and rabbit IgG bound protein eluted with free peptide. Eluates were assayed by immunoblotting with anti-B42-69 to detect Pur␣ and Pur␤ followed by anti-MSY 276 -302 to detect MSY1.
MCAT enhancer element. Because Pur␣ and/or Pur␤ appear capable of repressing MCAT enhancer activity independently of MSY1 (14), protein-protein interactions with MSY1 might serve to antagonize this effect by virtue of sequestering one or both of the Pur proteins into an inactive complex. Alternatively, MSY1 may potentiate the effect(s) of the Pur proteins owing to the fact that vertebrate Y-box proteins have been functionally implicated in both transcriptional activation (18, 34 -36) and repression (37,38). Other possibilities can easily be envisioned but can only be resolved through experimentation.
While potential combinatorial interactions between Pur␣, Pur␤, MSY1, and their respective ssDNA recognition motifs are numerous, they are not the only interactions which may be important to the regulation of MCAT enhancer activity. In particular, the binding of Pur␣ to a ssDNA recognition element has also been shown to be modulated by specific protein-protein interaction with the retinoblastoma tumor suppressor protein, Rb (39). Whether Rb similarly interacts with Pur␤ is not known. However, a Pur␣ sequence implicated in Rb binding, termed the "psycho" motif (39,40), is largely conserved in Pur␤, albeit with modification (10). We are currently exploring the potential involvement of Rb in regulating VSM ␣-actin gene transcription. Irrespective of the outcome of these experiments, it seems quite clear that complex combinations of proteinprotein and protein-ssDNA interactions are likely to be important to the ability of Pur␣, Pur␤, and MSY1 to modulate the activity of the VSM ␣-actin MCAT enhancer element. Delineation of the effects of such interactions on both the topology and activity of this element is an important priority.