DNA binding by the ETS protein TEL (ETV6) is regulated by autoinhibition and self-association

Abstract

The ETS protein TEL, a transcriptional repressor, contains a PNT domain that, as an isolated fragment in vitro, self-associates to form a head-to-tail polymer. How such polymerization might affect the DNA-binding properties of full-length TEL is unclear. Here we report that monomeric TEL binds to a consensus ETS site with unusually low affinity (K(d) = 2.8 x 10(-8) M). A deletion analysis demonstrated that the low affinity was caused by a C-terminal inhibitory domain (CID) that attenuates DNA binding by approximately 10-fold. An NMR spectroscopically derived structure of a TEL fragment, deposited in the Protein Data Bank, revealed that the CID consists of two alpha-helices, one of which appears to block the DNA binding surface of the TEL ETS domain. Based on this structure, we substituted two conserved glutamic acids (Glu-431 and Glu-434) with alanines and found that this activated DNA binding and enhanced trypsin sensitivity in the CID. We propose that TEL displays a conformational equilibrium between inhibited and activated states and that electrostatic interactions involving these negatively charged residues play a role in stabilizing the inhibited conformation. Using a TEL dimer as a model polymer, we show that self-association facilitates cooperative binding to DNA. Cooperativity was observed on DNA duplexes containing tandem consensus ETS sites at variable spacing and orientations, suggesting flexibility in the region of TEL linking its self-associating PNT domain and DNA-binding ETS domain. We speculate that TEL compensates for the low affinity, which is caused by autoinhibition, by binding to DNA as a cooperative polymer.

FIGURE 1.

Engineering of TEL monomeric and dimeric configurations. A, proposed model of wild-type TEL forming open-ended polymers via head-to-tail PNT domain interactions. The ETS domain, which mediates DNA binding, is also shown attached by a proposed flexible linker (this report). B, two point mutations (A94D, oval with horizontal lines, or V113E, oval with vertical lines) in the PNT domain block polymerization by disrupting one of the two surfaces necessary for subunit interaction (21). TEL species bearing both point mutations are monomeric, denoted here TEL^M. C, a combination of the two mutant proteins each bearing a single point mutation yields a stable dimer, denoted here TEL^D.

FIGURE 2.

TEL DNA binding is autoinhibited by sequences C-terminal to the ETS domain. A, schematic representation of TEL deletion mutants. PNT domain with polymer-blocking point mutations A94D and V113E (*), linker inhibitory damper (LID), ETS domain, and C-terminal inhibitory domain (CID) indicated. B, equilibrium DNA binding studies of TEL^M (upper) and ΔN331:ΔC426 (lower). Proteins were titrated over a concentration range of 10⁻¹³ to 10⁻⁷ m with each lane representing a 2.5-fold increase. The leftmost lane is a DNA-only control. C, DNA-binding isotherms were generated, as described under “Experimental Procedures,” for TEL^M (open circles) and ΔN331:ΔC426 (filled circles). Data points with error bars and K_d values represent the mean ± S.D. from four independent experiments. D, schematic of TEL autoinhibition by the CID with dampening by the LID.

FIGURE 3.

Inhibitory helix H5 blocks the TEL DNA binding interface. A, unpublished NMR-derived tertiary structure of ΔN334:ΔC436 in Protein Data Base (2DAO.pdb). The three α-helices (H1–H3) and four β-strands (S1–S4) that comprise the ETS domain (red), as well as two additional C-terminal helices (H4 and H5) (purple) of the CID are indicated. Conserved acidic residues Glu-431 and Glu-434 are shown as sticks (green). Arg-426 and Arg-429, which are sites of trypsin cleavage, are highlighted (orange). B, the DNA binding surface of an ETS domain is illustrated by a crystal structure of an Ets-1-DNA complex (selected residues 331–437 from 1MDM.pbd (42)): helix H3 lying in the major groove and the N terminus of helix H1 contacting the phosphodiester backbone, ETS domain (red); helix H4 and H5 of Ets-1 inhibitory module (purple), additional inhibitory helices N-terminal to the ETS domain are not illustrated for simplicity. C, schematic representation of ΔN334:ΔC436, with helices (cylinders) and β-strands (arrows) identified. Potential trypsin cleavage sites at all arginine and lysine residues are depicted as vertical lines, and Arg-426 and Arg-429 are indicated (*). A sequence alignment of C-terminal TEL residues (426–436) shows the position of helix H5 and the negatively charged residues Glu-431 and Glu-434. D, partial trypsin proteolysis of ΔN334:ΔC436 (upper) and ΔN334:ΔC436(E431A/E434A) (lower) with increasing amounts of trypsin. The positions of 10- and 15-kDa molecular mass markers are shown. Fragments with cleavage at Arg-426 and Arg-429 were identified via ESI/MS.

FIGURE 4.

TEL dimers bind cooperatively on DNA duplexes with a double ETS site. A, the ETS binding sites (bold) within the 57-bp DNA duplex used for EMSAs. Arrows indicate the orientation of 5′-GGA-3′ core sequences in DNA. B, DNA binding patterns of full-length, autoinhibited TEL^M (left) and TEL^D (right) as described (Fig. 2), except here each lane represents a reaction with a 10-fold higher protein concentration. Labels, (1x)TEL^M, (2x)TEL^M, and (1x)TEL^D, refer to the proposed identity (M, monomer or D, dimer) and number (1x or 2x) of independent binding species. The darker appearance of the right panel represents a longer exposure of the radioactive gel and is inconsequential as all assays had equivalent amounts of DNA. C, analysis of ΔC426 species, as in B. This fragment lacks autoinhibition and, thus, exhibits higher affinity DNA binding.

FIGURE 5.

TEL dimers form more stable complexes than TEL monomers on tandem ETS binding sites. Dissociation of protein-DNA complexes were measured by EMSAs. Labels, (1x)ΔC426^M, (2x)ΔC426^M, and (1x)ΔC426^D, defined as in Fig. 4. Free DNA is only shown (A, upper). A, stability of ΔC426^M (upper) and ΔC426^D (lower) bound to a single ETS consensus site (shown). B, stability of ΔC426^M (upper) and ΔC426^D (lower) bound to a direct repeat of an ETS consensus site (shown). C, exponential decay plots of data from A and B. The upper x axis is adjusted by 15 s to account for the time required for the mixing of the sample and the sample to enter the gel. ΔC426^D bound to a direct repeat (filled circles) and ΔC426^D bound to a single ETS binding site (open circle) are shown. Similar to ΔC426^D on a single site, dissociation plots for ΔC426^M indicate a half-life of <10 s on either a direct repeat or a single site. These data are not shown for clarity of the graph. See Table 2 for t½ values of ΔC426^D on a single ETS site or a variety of double ETS site duplexes.

FIGURE 6.

Model of dimeric TEL binding to DNA with a double ETS site. Determinants of cooperative dimeric binding suggest flexibility of TEL^D between the self-associating PNT domain and the DNA-binding ETS domain. Horizontal arrows indicate binding site positions and orientation of the core 5′-GGA-3′ recognition sequence. Domains and point mutations are illustrated as in Fig. 1.