Wilms tumor protein recognizes 5-carboxylcytosine within a specific DNA sequence

Abstract

In mammalian DNA, cytosine occurs in several chemical forms, including unmodified cytosine (C), 5-methylcytosine (5 mC), 5-hydroxymethylcytosine (5 hmC), 5-formylcytosine (5 fC), and 5-carboxylcytosine (5 caC). 5 mC is a major epigenetic signal that acts to regulate gene expression. 5 hmC, 5 fC, and 5 caC are oxidized derivatives that might also act as distinct epigenetic signals. We investigated the response of the zinc finger DNA-binding domains of transcription factors early growth response protein 1 (Egr1) and Wilms tumor protein 1 (WT1) to different forms of modified cytosine within their recognition sequence, 5'-GCG(T/G)GGGCG-3'. Both displayed high affinity for the sequence when C or 5 mC was present and much reduced affinity when 5 hmC or 5 fC was present, indicating that they differentiate primarily oxidized C from unoxidized C, rather than methylated C from unmethylated C. 5 caC affected the two proteins differently, abolishing binding by Egr1 but not by WT1. We ascribe this difference to electrostatic interactions in the binding sites. In Egr1, a negatively charged glutamate conflicts with the negatively charged carboxylate of 5 caC, whereas the corresponding glutamine of WT1 interacts with this group favorably. Our analyses shows that zinc finger proteins (and their splice variants) can respond in modulated ways to alternative modifications within their binding sequence.

Keywords: 5-carboxylcytosine; DNA modification; epigenetics.

Figure 1.

Egr1/Zif268 and WT1 bind methylated and 5mC oxidized DNA. Binding affinities of Egr1/Zif268 and WT1 with oligos containing varied forms of cytosine, as measured by fluorescence polarization assays. (A,B) Oligos fully methylated at both sites (M/M), unmethylated at both sites (C/C), or methylated in only the bottom strand at both sites (C/M). (C,D) Oligos modified in only the top strand at both sites with 5mC (M), C, 5hmC, 5fC, or 5caC. 5mC was present in the bottom strand at both sites in all cases. (E,F) Oligos modified in only the top strand at the 3′ site. All other sites contained 5mC (M). (G,H) Comparison of Egr1 and WT1 with oligos containing 5caC in the top strand at both sites (5caCx2) or only the 3′ site (5caCx1) using the data from C–F. (I) The “QQ” variant (Q369 and E427Q) of WT1 displays enhanced binding with 5caCx2 oligos.

Figure 2.

Egr1 binds methylated and unmodified DNA. (A) Schematic representation of the ZnF1–3 DNA-binding domain of Egr1/Zif268. The sequence and the secondary structure are shown. Arrows represent β strands, lines represent loops, and ribbons represent α helices. Two cysteine and two histidine residues (C2H2) in each finger are responsible for Zn²⁺ ligand binding (top connecting lines). Amino acids at positions −1, −4, and −7 (highlighted) relative to the first histidine interact specifically with the DNA bases shown below. The sequence of the oligos used for this study is shown with the top strand (magenta) oriented left to right from 3′ to 5′ with a 5′ overhanging adenine. The complementary strand (black) has a 5′ overhanging thymine. (B) Arg351 at position −7 of ZnF1 forms a methyl–Arg–Gua triad with the top strand of XpG (X = 5mC, 5hmC, or 5fC). The 2Fo − Fc electron density, contoured at 1σ above the mean, is shown in gray. (C) E354 at position −4 of ZnF1 is in van der Waals contact with the methyl group of 5mC (red arrow; distance shown in angstroms). The simulated annealing omit electron densities (meshed lines), contoured at 10σ and 4σ above the mean, respectively, for omitting the methyl group of 5mC and the side chain of E354 are shown. (D,E) The side chain of E354 becomes disordered with 5hmC (D) or 5fC (E). (E) An intrabase H-bond is present between the formyl oxygen of 5fC and the N4 group. Simulated annealing omit electron densities (meshed lines), contoured at 10σ and 4σ above the mean, respectively, for omitting the hydroxyl group of 5hmC (D) (or the carbonyl oxygen atom of 5fC [E)] and the side chain of E354 are shown. (F,G) Conformation of E354 indicates ∼90° side chain rotations with 5hmC (green) and 5fC (cyan) compared with 5mC (magenta). (H,I) E410 at position −4 of ZnF3 adopts two conformations with unmodified cytosine, engaging in van der Waals and weak C-H…O type H-bond interactions with the ring carbon atoms.

Figure 3.

WT1 binds 5caC and 5mC. (A) Schematic of ZnF2–ZnF4 DNA-binding domain WT1 (−KTS isoform), depicted as in Figure 2. (B) The side chain of Q369 at position −4 of ZnF2 adopts different conformations with (from top to bottom) 5mC, 5hmC, 5fC, and 5caC. The 2Fo − Fc electron density, contoured at 1σ above the mean, is shown in gray. (C) Q369 is in van der Waals contact with the methyl group of 5mC (red arrow; distance in angstroms). The simulated annealing omit electron densities (meshed lines), contoured at 10σ and 4σ above the mean, respectively, for omitting the methyl group of 5mC and the side chain of Q369 are shown. (D) Q369 forms an H-bond (dotted line) with the N4 atom of 5hmC and is in van der Waals contact (red arrow) with the CH₂ group. (E) Superimposition of C and D showing side chain conformation of Q369 with 5mC (magenta) and 5hmC (green). The two conformations are related by rotations of χ1 = 120°, χ2 = 90°, and χ3 = 90°. (F) Q369 interacts with 5hmC via the side chain carbonyl oxygen and with the 5′ Gua via the amide group. (G) R372 interaction with Gua7 in the presence of 5mC (magenta) and 5hmC (green). (H) Q369 interaction with 5fC. An intrabase H-bond is present in 5fC as in Egr1 (Fig. 2E). (I) Superimposition of C and H showing side chain conformation of Q369 with 5mC (magenta) and 5fC (cyan). (J) The three phosphate groups immediately surrounding 5fC are mobile. (K) One carboxylate oxygen of 5caC forms an H-bond with the side chain amide group of Q369. The other forms an intrabase H-bond with the N4 group. (L) Water-mediated interactions surrounding 5caC and Q369. The simulated annealing omit electron densities (meshed lines), contoured at 10σ and 4σ above the mean, respectively, for omitting the carboxylate group of 5caC and the side chain of Q369 are shown. (M) Superimposition of C and K showing Q369 with 5mC (magenta) and 5caC (gray). The two side chain conformations are related by a 70° rotation of the χ3 torsion angle. (N) Electrostatic sandwich between the negatively charged carboxylate group of 5caC and the positively charged guanidino groups R372 and R366.

Figure 4.

The WT1 +KTS isoform binds most strongly to 5caC DNA. (A) Human WT1 contains a C-terminal ZnF DNA-binding domain comprising four fingers in tandem. For the study described here, we used a fragment of WT1 containing ZnF2, ZnF3, and ZnF4 without KTS (the −KTS isoform) and with KTS (the +KTS isoform). (B) Comparison of the ±KTS isoforms on oligos containing unmodified C or 5mC (fully or hemimethylated). (C) The two KTS isoforms have a similar, relatively low affinity for 5caC-containing DNA but substantially different affinities for 5mC-containing DNA. (D,E) The +KTS isoform binds most strongly to 5caC-containing DNA. Affinity is uniformly low in 300 mM NaCl (D) but considerably higher in 200 mM NaCl (E).

Figure 5.

A WT1 mutant variant with high preference for 5mC. (A) The “PP” variant (Q369P and E427P) of WT1 prefers 5mC to all other forms by a factor of 40—140. (B) Structural comparison of WT1 wild type and the Q369P mutant. (C) The methyl group of 5mC forms a van der Waals contact with the proline at −4 position. The simulated annealing omit electron density (blue lines), contoured at 4σ above the mean, for omitting the methyl group of 5mC are shown.