Mechanism of strand-specific smooth muscle alpha-actin enhancer interaction by purine-rich element binding protein B (Purbeta)

Abstract

Expression of the smooth muscle alpha-actin gene in growth-activated vascular smooth muscle cells and stromal fibroblasts is negatively regulated by members of the Pur family of single-stranded DNA/RNA-binding proteins. In particular, Puralpha and Purbeta are postulated to repress transcription by forming helix-destabilizing complexes with the sense strand of an asymmetric polypurine-polypyrimidine tract containing a canonical MCAT enhancer motif in the 5' region of the gene. Herein, we establish the mechanism of Purbeta binding to the purine-rich strand of the enhancer using quantitative methods and purified components. Initial evaluation of DNA-binding specificity and equilibrium stoichiometry via colorimetric-, autoradiographic-, and fluorescence-based assays suggested that Purbeta interacts with two distinct G/A-rich sites within the nominal single-stranded enhancer element to form a high-affinity 2:1 protein:DNA complex. Statistical mechanical analyses of band shift titrations of the nominal element in conjunction with DNase I footprint titrations of the extended smooth muscle alpha-actin 5'-flanking region demonstrated that assembly of the nucleoprotein complex likely occurs in a sequential, cooperative, and monomer-dependent fashion. Resolution of the microscopic energetics of the system indicated that monomer association with two nonidentical sites flanking the core MCAT motif accounts for the majority of the intrinsic binding affinity of Purbeta with intersite cooperativity contributing an approximately 12-fold increase to the stability of the nucleoprotein complex. These findings offer new insights into the mechanism, energetics, and sequence determinants of Purbeta repressor binding to a biologically relevant, contractile phenotype-regulating cis-element while also revealing the thermodynamic confines of putative Purbeta-mediated effects on DNA structure.

FIGURE 1

Schematic of the polypurine-polypyrimidine tract containing a cryptic MCAT enhancer element in the 5′ flanking region of the mouse SMαA gene. Prior evidence for ssDNA-dependent structural rearrangements () in conjunction with ssBP-dependent repression of a core MCAT enhancer (italicized letters) () makes this region a focus of analysis in this study. The purine-rich probe, PE32-F, contains the minimum sequence supporting high affinity binding by Purβ and conferring MCAT enhancer repression. Nucleotides identified with a “•” were previously implicated in the ssDNA-binding specificity of Purβ extracted from cultured mammalian cell nuclei. Underlined sequences are homologous to a proposed minimal Purα recognition element (GGGAGAG) (). Numbers denote nucleotide positions relative to the transcriptional start site.

FIGURE 2

Analysis of the nucleotide determinants of PE32-F binding by N-HisPurβ. Results of fluid-phase competitor titrations in an ELISA-based ssDNA-binding assay are shown for a series of oligonucleotides harboring heptadeoxythymidylate (T7) mutations of putative purine-rich binding motifs in PE32-F. (A) Competition isotherms were fit to equation 1 to resolve IC₅₀ values for each competitor. Sequences of oligonucleotides are presented in Table 1. Points represent individual measurements made in a representative assay. (B) IC₅₀ values plotted as best fit value ± 67% confidence interval. The competitor marked with an asterisk indicates very low affinity as evidenced by an irresolvable IC₅₀ value.

FIGURE 3

Evaluation of the assembly and stoichiometry of N-HisPurβ:PE32-F complexes. (A) At least three electrophoretic species are detectable in band shift titrations of N-HisPurβ with 2.0 nM PE32-F* indicative of sequential formation of nucleoprotein complexes with differing putative stoichiometries (1:1, 2:1, and 3:1). Concentrations of N-HisPurβ in lanes 2 through 19 increase 1.5-fold from 0.41 nM (lane 2) to 400 nM (lane 19). (B) Limited serial dilution of a mixture of N-HisPurβ and PE32-F* (5.0 nM and 1.0 nM, respectively) was performed and samples were subjected to EMSA. The intensity of the Free Probe band was quantified by densitometry and standardized to known quantities of PE32-F* (lanes 1 and 20) to determine the concentration of free DNA, [D_free], in each lane. The concentration of nucleoprotein complex [P_nD] with stoichiometry n was determined from the equation [P_nD] = [D_total] – [D_free]. The concentration of free protein was estimated using the equation [P_free] = [P_total] – n[P_nD], in which n is an assumed integer value of 1, 2, or 3. (C) Isotherms of ln[P_nD/D_free] versus ln[P_free] with assumed integer values of n were plotted. Each point represents the mean of duplicate experiments. Dashed lines represent the least-squares regression fits of each dataset to equation S11. Numbers in parentheses reflect the returned regression fit value of n ± s.d. (D) Fluorescence anisotropy analysis of the binding of N-HisPurβ to 50 nM PE32-F-3FLC. The dashed line represents a non-linear least squares fit of the data to equation S6. Fixing K_r at near-zero values (infinite affinity, solid line) verified the equivalency transition at an R_P/D value of 2:1. Symbols show titrations using two different preparations of N-HisPurβ.

FIGURE 4

Titration analysis of N-HisPurβ binding to PE32-F by quantitative EMSA. (A) N-HisPurβ was diluted over a concentration range of 10⁻¹¹ to 10⁻⁸ M and equilibrated with 25 pM PE32-F* prior to subjecting reaction mixtures to EMSA. (B) Densitometric analysis of each lane of the gel shown in (A) verifies the presence of three pixel intensity peaks indicative of separate electrophoretic species. Note that the putative 1:1 complex does not accumulate significantly compared to the free probe or the 2:1 complex, suggestive of a sequential and cooperative binding mechanism. (C) Individual band intensities are plotted as a function of [N-HisPurβ]_free(●, Free Probe; ▼, 1:1 complex; ■, 2:1 complex). Each point represents the mean ± s.d. of quadruplicate experiments. Lines are global nonlinear least squares fits of individual species data points to equations S7–S9. (D) Band intensity data from (A) presented as Fraction Bound, Ȳ ([D_total] – [D_free]) / [D_total]) versus [N-HisPurβ]_free were fit to equation 4. The Hill coefficient, α_H, was held constant at values of 1.5 (dashed line) and 1.0 (dotted line) to reflect the dependency of this variable on goodness of fit.

FIGURE 5

Titration analysis of N-HisPurβ binding to SMP382-F by quantitative DNase I footprinting. Representative footprint titration analysis of N-HisPurβ binding to SMP382-F* shows two regions of protection adjacent to the core MCAT motif and within the nominal PE32-F sequence (marked as 3′ site and 5′ site). The protected sites within the PE32-F sequence are separated by a band with protein-independent pixel intensity, when normalized to pixel intensity of the control region (–208 to –201) suggesting this intervening region is not protected by N-HisPurβ. Other sites of protection are evident within or near the THR, TCE, and CArG boxes 1 and 2 as well as a previously uncharacterized upstream region (−218 to −210), and a region adjacent to the TATA box. A positional map of the above noted cis-elements is shown on the right next to the lanes containing dideoxy NTP sequencing reactions (G, A, T, and C).

FIGURE 6

Delineation of a cooperative binding model describing the interaction of N-HisPurβ with the MCAT-containing enhancer based on DNase I footprint data. Individual 3′ and 5′ site data points implying differential N-HisPurβ affinity were systematically and globally fit to equations (Table S3) describing a two-site model for either non-identical, independent binding sites (A) or non-identical, interacting binding sites (B). Blue symbols represent N-HisPurβ binding to the 3′ site. Red symbols represent binding to the 5′ site. Each point represents the mean ± s.d. of five independent experiments and the lines are best fit isotherms. Residual analysis and fit statistics support a cooperative binding model involving two non-identical binding sites.

FIGURE 7

Monte Carlo error simulations to assess model confidence. Reiterative error simulations (1000) were performed on individual site isotherms shown in Figure 6 to yield error-incorporated isotherms that were then globally fit to various two site models. Box-and-whisker plots representing the distributions of fitting statistics (∑residuals² ) for each model are shown when error is introduced at the level observed in experiments described herein (±13%, left panel). Reducing error to ±5% leads to higher model confidence as indicated by resolution of box-and-whisker plots (right panel). Boxes represent 25th-75th percentiles, whiskers represent 10th-90th percentiles. Median is marked by a line across the box, and mean is denoted as (+). Coop, cooperatively-interacting binding sites; Indy, independent non-interacting binding sites.