Conformational variants of duplex DNA correlated with cytosine-rich chromosomal fragile sites

Abstract

We found that several major chromosomal fragile sites in human lymphomas, including the bcl-2 major breakpoint region, bcl-1 major translocation cluster, and c-Myc exon 1-intron 1 boundary, contain distinctive sequences of consecutive cytosines exhibiting a high degree of reactivity with the structure-specific chemical probe bisulfite. To assess the inherent structural variability of duplex DNA in these regions and to determine the range of structures reactive to bisulfite, we have performed bisulfite probing on genomic DNA in vitro and in situ; on duplex DNA in supercoiled and linearized plasmids; and on oligonucleotide DNA/DNA and DNA/2'-O-methyl RNA duplexes. Bisulfite is significantly more reactive at the frayed ends of DNA duplexes, which is expected given that bisulfite is an established probe of single-stranded DNA. We observed that bisulfite also distinguishes between more subtle sequence/structural differences in duplex DNA. Supercoiled plasmids are more reactive than linear DNA; and sequences containing consecutive cytosines, namely GGGCCC, are more reactive than those with alternating guanine and cytosine, namely GCGCGC. Circular dichroism and x-ray crystallography show that the GGGCCC sequence forms an intermediate B/A structure. Molecular dynamics simulations also predict an intermediate B/A structure for this sequence, and probe calculations suggest greater bisulfite accessibility of cytosine bases in the intermediate B/A structure over canonical B- or A-form DNA. Electrostatic calculations reveal that consecutive cytosine bases create electropositive patches in the major groove, predicting enhanced localization of the bisulfite anion at homo-C tracts over alternating G/C sequences. These characteristics of homo-C tracts in duplex DNA may be associated with DNA-protein interactions in vivo that predispose certain genomic regions to chromosomal fragility.

FIGURE 1.

Correlation of bcl-2 MBR and bcl-1 MTC bisulfite reactivity with sequence features. A, bisulfite reactivity of the bcl-2 MBR has been published and is plotted in the top panel (13). The x axis denotes every base position along the 528-bp fragment (which includes A and T in addition to C and G). C→T and G→A conversion percentages are plotted, calculated as the number of molecules converted at that particular position, and divided by the total number of molecules sequenced for that strand. Bases in the 2nd panel are aligned with those in the top panel, and plot the C or G “string” length, or the length of consecutive Cs or Gs within which a particular C or G position is located. For instance, ACGGGGCCGGT would have values of 0, 1, 4, 4, 4, 4, 2, 2, 2, 2, 0 for each base position, from left to right. The bottom two graphs show the C, G, C+G, A, T, and A+T densities calculated using a moving window of 21 bp. Although there is some correlation between C and C+G density and bisulfite reactivity, the fine peak structure of reactivity matches best with the string length. B, bisulfite reactivity at the bcl-1 MTC. Reactivity and sequence features for the bcl-1 MTC are plotted similar to A, but using the data from supplemental Fig. S1 for the three bisulfite plots.

FIGURE 2.

Bisulfite reactivity on double-stranded DNA containing GGGCCC, GGCGCC, and GCGCGC sequences. Bisulfite reactivities for each cytosine on each duplex DNA substrate are shown in (A-C). Boxed in dark gray are the sequences of interest, and each black circle denotes a bisulfite-catalyzed C to U conversion detected after cloning and sequencing. 30 nt of the long strand sequence have been truncated for space.

FIGURE 3.

Bisulfite reactivity on dsDNA containing GGCC, GCGC, and CGCG sequences. Bisulfite reactivities for each cytosine on each duplex DNA substrate are shown in A-C. Boxed in dark gray are the sequences of interest, and each black circle denotes a bisulfite-catalyzed C to U conversion detected after cloning and sequencing. 30 nt of the long strand sequence have been truncated for space.

FIGURE 4.

Circular dichroism of nucleic acids containing GGGCCC, GCGCGO, and GGCGCC sequences. Circular dichroism spectra of B-form DNA (catGCGCGCatg), A-form RNA (cauGCGCGCaug), and two B/A-form intermediate sequences (catGGCGCGatg and catGGGCCCatg) in 100 mm Na⁺ are shown.

FIGURE 5.

Bisulfite is minimally reactive on DNA-2′-O-methyl RNA hybrid duplexes. Bisulfite reactivities for each cytosine on each hybrid duplex DNA-2′-O-methyl RNA substrate are shown. Notations are as in Fig. 2.

FIGURE 6.

Double-stranded DNA conformations of d(GGGCCC)₂. Structures of d(GGGCCC)₂ (G = red, C = green) in canonical B-DNA (A); an averaged structure from molecular dynamics simulation (see below) (B); x-ray structure of a B/A-intermediate structure (PDB code 1DC0) (C); and canonical A-DNA (D). The structures are viewed with strand 1 running from top to bottom in a 5′ to 3′ direction and the major groove on the left of each helix. The simulated structure is an average of 2500 structures collected between 4.5 and 5 ns of the simulation and has the following helical parameters: shift = 0.0 Å, slide = -0.8 Å, rise = 3.4 Å, tilt = 0.0°, roll = 4.3°, and twist = 32.6°. Comparison with the data in Table 2 shows that this structure has parameters that are similar to those averaged over 10 ns. All structures are shown as 12-bp duplexes (for the simulated structure, the terminal base pairs were deleted). The central region is numbered with reference to the six cytosine bases, as d(G₆G₅G₄C₁C₂C₃)·(G₃G₂G₁C₄C₅C₆).

FIGURE 7.

Calculated localization of bisulfite ions on double-stranded DNA containing GGGCCC (A) and GCGCGC (B) sequences. Red ellipse in A indicates the electropositive patch that is observed along the major groove edge of the three consecutive cytosine bases.