DNA sequence-dependent deformability deduced from protein-DNA crystal complexes

Abstract

The deformability of double helical DNA is critical for its packaging in the cell, recognition by other molecules, and transient opening during biochemically important processes. Here, a complete set of sequence-dependent empirical energy functions suitable for describing such behavior is extracted from the fluctuations and correlations of structural parameters in DNA-protein crystal complexes. These elastic functions provide useful stereochemical measures of the local base step movements operative in sequence-specific recognition and protein-induced deformations. In particular, the pyrimidine-purine dimers stand out as the most variable steps in the DNA-protein complexes, apparently acting as flexible "hinges" fitting the duplex to the protein surface. In addition to the angular parameters widely used to describe DNA deformations (i.e., the bend and twist angles), the translational parameters describing the displacements of base pairs along and across the helical axis are analyzed. The observed correlations of base pair bending and shearing motions are important for nonplanar folding of DNA in nucleosomes and other nucleoprotein complexes. The knowledge-based energies also offer realistic three-dimensional models for the study of long DNA polymers at the global level, incorporating structural features beyond the scope of conventional elastic rod treatments and adding a new dimension to literal analyses of genomic sequences.

Figure 1

Scatter plots in the Twist–Roll plane of base step parameters in protein-bound and B–DNA crystal complexes. Dots correspond to the “unperturbed” P⋅DNA sample used in averages (Table 1) and derived deformabilities (Table 2) and circles to states of extreme bending, twisting, and stretching (included in B– and P′⋅DNA samples). Rectangles enclose points lying within three rms (3Δθ) deviations of 〈Twist〉 and 〈Roll〉. Ellipses are projections of the six-dimensional equi-potential surfaces on the Twist–Roll plane obtained from the 2 × 2 Twist–Roll covariance matrix (see text); these contours correspond to energies of 4.5 k_BT (“3Δθ ellipses”). Histograms on the edges of the scatter plots are scaled with respect to a value of unity for the most populated angular ranges (422–474 occurrences in P⋅DNA and 162–168 for B–DNA).

Figure 2

Scatter plots in the Twist–Roll plane of observed parameters and derived energy contours of pyrimidine–purine (YR), purine–purine (RR), and purine–pyrimidine (RY) dimer steps. See legend to Fig. 1. Note the gradually decreasing areas of the 3Δθ ellipses from left to right. Corresponding plots for B–DNA are found at the following URL: http://rutchem.rutgers.edu/∼olson/pdna.html.

Figure 3

Sequence-dependent motions along the longest principal axes of P⋅DNA dimer steps: pyrimidine–purine (YR), purine–purine (RR), and purine–pyrimidine (RY) steps. Nonequilibrium forms, corresponding to the parametric changes in Table 2, are superimposed on the average dimer structures (thickened bonds). Perturbed conformations correspond to states deformed along the longest principal axes of the derived energy functions, with increments equal to changes of 〈λ₁²〉^−1/2 (Table 2) and energies ranging from 0 to 12.5 k_BT for ±5〈λ₁²〉^−1/2 deviations. Opposing directions of fluctuations are distinguished by color-coded (Y, light blue; R, red) and gray images. Motions illustrated with respect to a reference frame in the 5′ base pair (i.e., the M in MpN dimers). Views from the leading strand. Base pairs represented as ideal Watson–Crick pairs with C1′ atoms of rest structures noted by circles. Note the decreasing range of motions from left to right.

Figure 4

Scatter plots in the Slide–Roll plane of observed parameters and derived energy contours of individual pyrimidine–purine (YR) dimer steps. Note the negative coupling of CA and TA parameters. See legend to Fig. 1.