The highly conserved beta-hairpin of the paired DNA-binding domain is required for assembly of Pax-Ets ternary complexes

Abstract

Pax family transcription factors bind DNA through the paired domain. This domain, which is comprised of two helix-turn-helix motifs and a beta-hairpin structure, is a target of mutations in congenital disorders of mice and humans. Previously, we showed that Pax-5 (B-cell-specific activator protein) recruits proteins of the Ets proto-oncogene family to bind a composite DNA site that is essential for efficient transcription of the early-B-cell-specific mb-1 promoter. Here, evidence is provided for specific interactions between Ets-1 and the amino-terminal subdomains of Pax proteins. By tethering deletion fragments of Pax-5 to a heterologous DNA-binding domain, we show that 73 amino acids (amino acids 12 to 84) of its amino-terminal subdomain can recruit the ETS domain of Ets-1 to bind the composite site. Furthermore, an amino acid (Gln22) within the highly conserved beta-hairpin motif of Pax-5 is essential for efficient recruitment of Ets-1. The ability to recruit Ets proteins to bind DNA is a shared property of Pax proteins, as demonstrated by cooperative DNA binding of Ets-1 with sequences derived from the paired domains of Pax-2 and Pax-3. The strict conservation of sequences required for recruitment of Ets proteins suggests that Pax-Ets interactions are important for regulating transcription in diverse tissues during cellular differentiation.

FIG. 1

The Pax-5 paired domain (residues 1 to 149) and the ETS domain of Ets-1 (residues 333 to 440) exhibit cooperative binding to the wild type mb-1 promoter. (A) Double-stranded oligonucleotide probe sequences used in this study. Only the sense sequence is shown, numbered relative to +1 of the wild-type mb-1 promoter (45). Nucleotides contacted by Pax-5 and those contacted by Ets-1, as estimated by methylation interference and footprinting studies (20, 24a), are highlighted at the top. Boxed lowercase letters represent mutations. (B) Relative DNA binding by Pax-5 and ETS DNA-binding domains. DNA probes and inclusion of proteins are indicated at the top. wt, wild type; TC, ternary complexes; F, free probe.

FIG. 2

A model of the Pax-5 and Ets-1 DNA-binding domains bound to a portion of the mb-1 promoter sequence. The murine Ets-1 ETS domain and flanking carboxy-terminal α-helix (red) and amino-terminal subdomain of Drosophila Paired (blue) are shown docked on idealized B-DNA (yellow). The guanosine common to the Pax-5 and Ets-1 binding sites in mb-1 is indicated as Gua. Positions of the aspartic acid (Asp) following the recognition helix in Ets-1 and the glutamine (Gln) in the β-hairpin of Pax-5 are indicated. These residues are identified as playing a key role in ternary complex formation. The amino termini of the two protein fragments are identified by N.

FIG. 3

DNA binding by truncated Pax-5 polypeptides. (A) Schematic representation of truncated polypeptides used for panels B and C. The secondary structure of Pax-5 was predicted from the crystal structure of Drosophila Paired (53). Arrows are regions of β-sheet, boxes are α-helices, and τ1 and τ2 are turns. The β-hairpin is formed by β1-τ1-β2, and the type II turn (β-turn) is τ2. Relative to the diagram, Pax-5(1–149) includes six additional carboxy-terminal amino acids. (B) Control EMSA showing binding of truncated Pax-5 polypeptides to the Sγ2a probe. The lysate in lane pET was prepared from E. coli transformed with empty pET-11a vector. The gel was exposed to X-ray film for 15 h. F, free probe. (C) EMSA showing binding of truncated Pax-5 polypeptides to the mb-1 probe with or without added Ets-1 ETS domain. The concentration of Pax-5(1–149) used in lanes 2 and 3 is approximately one-fourth that of other Pax-5 polypeptides used in this experiment. Binding of the Ets-1 ETS domain by itself was detected due to the extended period of autoradiography (3 days) necessary for detection of weak ternary complex formation. Lane pET is as in panel B. TC, ternary complexes; F, free probe.

FIG. 4

DNA binding and recruitment of the Ets-1 ETS domain by LEF-1–Pax-5 amino-terminal subdomain hybrid polypeptides. (A) DNA probes used in these assays. Pax-5 and Ets-1 binding sites in the wild-type (wt) probe are highlighted as in Fig. 1A. LEF-1 recognition sequences are circled. LPE (0) was the sequence predicted by molecular modeling to be optimal for LEF-1–Pax-5 binding. (B) EMSA of LEF-1–Pax-5(12–84) binding to the five LPE probes. TC, ternary complexes; F, free probe. (C) Control experiments. EMSA of LEF-1–HMG (residues 296 to 381 of murine LEF-1), LEF-1–linker [the HMG domain of LEF-1 and the (Gly₃Ser/Thr)₄ linker], or LEF-1–Pax-5(12–84) binding to the wild-type mb-1 or LPE (−1) probe. Ternary complexes are detected only with Pax-5, Ets-1, and the wild-type mb-1 probe. F, free probe. (D) Ternary complex assembly with the hybrid protein requires the Ets-1 binding site. EMSA was performed with the wild-type or mut 1 (Fig. 1A) version of the LPE (−1) probe together with the LEF-1–Pax-5(12–84) protein and Ets-1 ETS domain. TC, ternary complexes; F, free probe. The gels in panels B to D were exposed to X-ray film for 12 to 15 hours.

FIG. 5

Gln22 of Pax-5 is required for efficient assembly of ternary complexes with Ets-1. EMSA was performed with wild-type (wt) or mutated Pax-5(1–149) polypeptide and either the CD19 or mb-1 DNA probe, as indicated. Lysates in lanes pET were prepared from E. coli transformed with empty pET-11a vector. TC, ternary complexes; F, free probe.

FIG. 6

The glutamine required for efficient recruitment of Ets proteins (and four flanking amino acids) has been perfectly conserved throughout the evolution of Pax proteins. Amino acid sequences comprising the amino-terminal subdomains and linker regions of Pax proteins were aligned for comparison as shown. Secondary structure of Drosophila Paired is shown with arrows indicating β-strands, boxes indicating α-helical regions, and τ1 and τ2 indicating turns identified in the crystal structure of Paired (53). The β-hairpin is formed by strands β1 and β2 and the type I β-turn τ1, while τ2 is a type II β-turn. Shaded vertical bar indicates the completely conserved region of the β-hairpin, including the glutamine residue analyzed in this report. Sequences were derived from Homo sapiens Pax-1 (EMBL/GenBank accession no. P15863), Pax-3 (P23760), Pax-5 (M96944), Pax-6 (M77844), Pax-7 (Z35141), Pax-8 (L19606), and Pax-9 (S36115); Mus musculus Pax-2 (280984) and Pax-4 (P32115); Branchiostoma floridae (amphioxus) AmphiPax-6 (AJ223440); Halocynthia roretzi (ascidian) Pax-37 (D84254) and HRPax-258 (AB006675); Lineus sanguineus (ribbonworm) Ls-Pax-6 (X95594); Acropora millepora (coral) Pax-C (AF053459); Chrysaora quinquecirrha (sea nettle) Pax-A1 (U96195) and Pax-B (U96197); Drosophila melanogaster Paired (P06601), Gooseberry proximal (Gsb-p; P09083), Gooseberry distal (Gsb-d; P09082), Sparkling (AF010256), Eyeless (X79492), Pox-meso (P23757), and Pox-neuro (P23758); and C. elegans Pax homologs C04G2.7 (Z70718) and F27E5.2 (Z48582) and Pax-6 homolog vab-3 (U31537). The dot in the Pax-4 sequence represents a gap in the alignment.

FIG. 7

Recruitment of the Ets-1 ETS domain by Pax-2. (A) The paired domain of Pax-2 (residues 1 to 148) binds the mb-1 probe and recruits Ets-1. EMSA was performed with the mb-1 probe. Binding assays were performed with rabbit reticulocyte lysates programmed with synthetic RNA encoding the Pax-2 paired or Ets-1 ETS domain or with a translation without added RNA (No RNA). Both, Pax-2 and Ets-1 were both added after their separate translation; TC, ternary complexes; F, free probe.

FIG. 8

DNA binding and recruitment of Ets-1 by LEF-1–Pax-3(31–107) fusion protein. EMSA was performed with the LPE (−1) probe. Faster-migrating bands are likely due to limited proteolysis during preparation of proteins from E. coli. TC, ternary complexes; F, free probe; wt, wild type.