Asymmetric dimerization in a transcription factor superfamily is promoted by allosteric interactions with DNA

Abstract

Transcription factors, such as nuclear receptors achieve precise transcriptional regulation by means of a tight and reciprocal communication with DNA, where cooperativity gained by receptor dimerization is added to binding site sequence specificity to expand the range of DNA target gene sequences. To unravel the evolutionary steps in the emergence of DNA selection by steroid receptors (SRs) from monomeric to dimeric palindromic binding sites, we carried out crystallographic, biophysical and phylogenetic studies, focusing on the estrogen-related receptors (ERRs, NR3B) that represent closest relatives of SRs. Our results, showing the structure of the ERR DNA-binding domain bound to a palindromic response element (RE), unveil the molecular mechanisms of ERR dimerization which are imprinted in the protein itself with DNA acting as an allosteric driver by allowing the formation of a novel extended asymmetric dimerization region (KR-box). Phylogenetic analyses suggest that this dimerization asymmetry is an ancestral feature necessary for establishing a strong overall dimerization interface, which was progressively modified in other SRs in the course of evolution.

Graphical Abstract

Figure 1.

(A) Domain organization of ERRα. (B) ERRα DBD protein sequence and schematic secondary structure representation. (C) Sequence of the 26 bp DNA fragment, embERRE/IR3, used for crystallization where the two binding sites for DBD1 and DBD1 are outlined with blue and orange boxes, respectively. (D) Cartoon representation of the homodimer of ERRα DBD bound to the embERRE/IR3 response element, with DBD1 in light blue and DBD2 in orange color. The KR-box indicated in (B) is highlighted on the structure, in lemon green for DBD1 and teal for DBD2.

Figure 2.

The asymmetry of the dimerization interface involves a conserved KR-box motif. (A) Enlarged view of the residues belonging to two D-boxes (DBD1 and DBD2) that form the canonical dimerization interface. (B) Cartoon representation of the KR-box region (lemon green for DBD1 and teal for DBD2) for the subunits superimposed, with DBD1 in light blue and DBD2 in orange color. (C) Multiple sequence alignment of the human ERR DBD sequences with those of the other steroid NRs (NR3 subfamily (AncSR1, ER, AncSR2, AR, GR, MR, PR) and NRs of the subgroups NR1 (TR, RAR, PPAR, ROR, VDR), NR2 (RXR and HNF4), NR4 (NGF-IB, SF1), showing KR box conservation. The sequence numbering corresponds to the human ERRα sequence and the colors relate to the percentage of sequence identity. The top panel represents the secondary structure elements of ERRα. The KR box region of the ERRs is highlighted in green. (D) Stick representation of the interactions between the KR box residues Arg128 and Lys129 of DBD1 with the DNA phosphate backbone of G17 of the complementary strand (chain B). (E, F) Enlarged view of the dimerization interface interactions between DBD1 and DBD2 looking at the region of the KR-box of (E) DBD2 and (F) DBD1. Residues are shown with a stick representation, with carbon atom in light blue for DBD1 and orange for DBD2, red for oxygen, blue for nitrogen and yellow for sulfur atom.

Figure 3.

Major groove interactions in DBD1 and DBD2. (A, B) The residues Glu97, Lys100, Lys104 and Arg105 in H1 interact with the major groove half-site via base-specific interactions, in a similar manner for DBD1 (A) and DBD2 (B), but the interaction network is much stronger in DBD1 than in DBD2, as indicated by the structural water molecules seen in the electron density of DBD1, but not of DBD2. In addition, the side chains of residues Lys100 and Lys104 are only well defined in DBD1, but not in DBD2 (see Supplementary Figure S3A and B). (C) Enlarged view of the superimposition of one subunit of ERRα DBD (DBD1 in light blue) with the corresponding ERα DBD in the region encompassing H1 and H2. The core of ERRα DBD is stabilized by a pocket of hydrophobic residues. In ERα, Phe136, Leu140 and Thr106 are replaced by Ile, Tyr, and Ser, respectively, weakening the contribution of hydrophobic contacts. The Phe136 residue in ERRα, replaced by Leu in ERα, further strengthens the core interactions through an intricate network of π-π stacking interactions.

Figure 4.

TCA and TAA 5′ extensions are key in the conformation of the A-box. (A) In DBD1, the conformation of the T-box (residues 145–154) adjacent to the A-box (residues 155–161) is well defined and helps the A-box loop to penetrate into the minor groove of the 5′ extension. (B) Interaction pattern of the CTE residues of DBD1 with the TCA 5′extension. (C) Interaction pattern of the CTE residues of DBD2 with the TAA 5′extension. Only a few H-bond interactions are observed, demonstrating the weak interaction at TAA 5′extension. (D) EMSA analyses show the significant role of 5′- extension TNA nucleotides. EMSA gel for different concentration of MgCl₂ 25, 50 100, 250 μM (gels I to IV); lane 1: marker, lane 2; free DNA control; lane 3 TCA ERRE; lane 4: TTA ERRE; lane 5: TGA ERRE; lane 6: TAA ERRE and lane 7: ARE (Androgen receptor RE) DNA as a control DNA sequence (see Materials and Methods for sequences).

Figure 5.

The KR-box motif is an ancestral feature crucial for the asymmetric organization of ERRα DBD on DNA compared to ERα DBD and AncSR1 DBD. (A, B) Superimposition of ERRα DBD and ERα DBD on DNA (2 orientations, with a 90° rotation between them). (C, D) Superimposition of ERRα DBD and AncSR1 DBD on DNA, with (C) a global view and (D) an enlarged view in the KR-box region of DBD2. ERRα DBD1 is colored in light blue, ERRα DBD2 in orange, ERα DBDs in green, and AncSR1 in magenta. The views of the superimposition along the DNA axis (in B and in D, for ERα and AncSR1, respectively) emphasize the shift of ERRα DBD2 positioning on DNA.

Figure 6.

(A–D) DNA geometrical parameters of the embERRE/IR3 DNA bound by ERRα, showing the (A) minor groove width, (B) major groove width, (C) roll, and (D) twist parameters. (E–H) DNA geometrical parameters of the ERE-IR3 bound by ERα DBD homodimer (PDB code 1HCQ, (38)), showing the (E) minor groove width, (F) major groove width, (G) roll and (H) twist parameters.

Figure 7.

A representation of the ERRα DBD/ERRE complex, summarizing our main structural observations. The dimerization interface of ERRα DBD homodimer is made by interactions between the D box of each subunit. Furthermore, it is reinforced by the presence of an extended interaction area arising from the KR-box region of DBD2 that unfolds as a loop and interacts with the dimerization region of DBD1, as indicated by yellow arrows. The dimerization interface is asymmetric with different KR-box conformations for the two subunits (highlighted by the lemon green and teal colours) and it is required for proper stabilization of the DBD homodimer on DNA (as suggested by the red arrows), suggesting the importance of the DNA, which acts as allosteric driver in accommodating the receptor in a fully functional state suitable for gene expression regulation. The subunit bound to the extended half-site, DBD1, is shown in blue colour and DBD2 is shown in orange colour.