A new hydrogen-bonding potential for the design of protein-RNA interactions predicts specific contacts and discriminates decoys

Abstract

RNA-binding proteins play many essential roles in the regulation of gene expression in the cell. Despite the significant increase in the number of structures for RNA-protein complexes in the last few years, the molecular basis of specificity remains unclear even for the best-studied protein families. We have developed a distance and orientation-dependent hydrogen-bonding potential based on the statistical analysis of hydrogen-bonding geometries that are observed in high-resolution crystal structures of protein-DNA and protein-RNA complexes. We observe very strong geometrical preferences that reflect significant energetic constraints on the relative placement of hydrogen-bonding atom pairs at protein-nucleic acid interfaces. A scoring function based on the hydrogen-bonding potential discriminates native protein-RNA structures from incorrectly docked decoys with remarkable predictive power. By incorporating the new hydrogen-bonding potential into a physical model of protein-RNA interfaces with full atom representation, we were able to recover native amino acids at protein-RNA interfaces.

Figure 1

Schematic representation of the geometric parameters used to describe hydrogen-bond geometry. δ_HA represents the distance between the hydrogen and acceptor atoms; Θ, the angle at the hydrogen atom describes the linearity of hydrogen bond; Ψ, the angle at the acceptor atom; (X represents the dihedral angle given by rotation around the acceptor–acceptor base bond; for sp² hybridized acceptors, it is a measure of the planarity of the hydrogen bond. A, acceptor; D, donor; H, hydrogen; AB, acceptor base; and R₁, R₂, reference atoms bound to the acceptor base.

Figure 2

Distance (δ_HA), linear angle (Θ) and angular (Ψ) distributions for selected hydrogen bonds at protein–RNA/DNA interfaces: (a) phosphate oxygens to protein side-chain NH/NH₂; (b) base N to protein side-chain NH/NH₂; and (c) base O to protein side-chain NH/NH₂.

Figure 3

Dihedral angular distributions (X) for hydrogen bonds at the interface of proteins and RNA/DNA: (a) base N to protein side-chain NH/NH₂ donors; (b) base NH/NH₂ to protein side-chain sp² hybridized acceptors; (c) phosphate O to protein backbone NH; and (d) phosphate O to protein side-chain NH/NH₂.

Figure 4

Hydrogen bonding-potential of mean force for interactions between base N and protein side-chain NH/NH₂ donors. (a) Distance δ_HA; (b) angle Θ; (c) angle Ψ; and (d) angle X. The knowledge-based potentials were calculated from the negative logarithm of the observed frequency distributions (see Methods).

Figure 5

Native protein sequence recovery at protein–RNA interface derived from a test set of 17 protein–RNA complexes. Different energy functions are used to test the substitution profile: red bars, complete energy function, as described in the text; light blue bars, energy function with the angular terms of hydrogen-bonding potential turned off; yellow bars, the hydrogen-bonding potential was substituted with a purely Coulombic interaction model. The bars show how often the native amino acids are calculated to be energetically most favorable at each interfacial position probed. (a) Charged amino acids; (b) polar amino acids; and (c) polar aromatic amino acids.

Figure 5

Native protein sequence recovery at protein–RNA interface derived from a test set of 17 protein–RNA complexes. Different energy functions are used to test the substitution profile: red bars, complete energy function, as described in the text; light blue bars, energy function with the angular terms of hydrogen-bonding potential turned off; yellow bars, the hydrogen-bonding potential was substituted with a purely Coulombic interaction model. The bars show how often the native amino acids are calculated to be energetically most favorable at each interfacial position probed. (a) Charged amino acids; (b) polar amino acids; and (c) polar aromatic amino acids.

Figure 6

Recovery of hydrophobic amino acids at the interface of protein–RNA complexes using the complete energy function including the hydrogen-bonding potential.

Figure 7

Scatter plots obtained by scoring five sets of protein–RNA decoys using either hydrogen-bonding potential (a, c, e, g and i) or a Coulombic potential with distance-dependent dielectric constant (b, d, f, h and j). A total of 2000 decoys are created for each of the five test structures using the small perturbation method. The scores of the native structures are highlighted using red circles: (a and b) 1CVJ; (c and d) 1EC6; (e and f) 1FXL; (g and h) 1JID; and (i and j) 1URN.

Figure 7

Scatter plots obtained by scoring five sets of protein–RNA decoys using either hydrogen-bonding potential (a, c, e, g and i) or a Coulombic potential with distance-dependent dielectric constant (b, d, f, h and j). A total of 2000 decoys are created for each of the five test structures using the small perturbation method. The scores of the native structures are highlighted using red circles: (a and b) 1CVJ; (c and d) 1EC6; (e and f) 1FXL; (g and h) 1JID; and (i and j) 1URN.