Manipulation of conformational change in proteins by single-residue perturbations

Abstract

Using the perturbation-response scanning (PRS) technique, we study a set of 25 proteins that display a variety of conformational motions upon ligand binding (e.g., shear, hinge, allosteric). In most cases, PRS determines single residues that may be manipulated to achieve the resulting conformational change. PRS reveals that for some proteins, binding-induced conformational change may be achieved through the perturbation of residues scattered throughout the protein, whereas in others, perturbation of specific residues confined to a highly specific region is necessary. Overlaps between the experimental and PRS-calculated atomic displacement vectors are usually more descriptive of the conformational change than those obtained from a modal analysis of elastic network models. Furthermore, the largest overlaps obtained by the latter approach do not always appear in the most collective modes; there are cases where more than one mode yields comparable overlap sizes. We show that success of the modal analysis depends on an absence of redundant paths in the protein. PRS thus demonstrates that several relevant modes can be induced simultaneously by perturbing a single select residue on the protein. We also illustrate the biological relevance of applying PRS to the GroEL, adenylate kinase, myosin, and kinesin structures in detail by showing that the residues whose perturbation leads to precise conformational changes usually correspond to those experimentally determined to be functionally important.

Figure 1

C_α displacement results for (A) ADK, (B) AR, (C) UCE, and (D) FD. Gray line represents the experimental C_α displacements between open and closed structures, black curve is the prediction of PRS, and dashed curve is obtained from the best normal-mode vector in ANM. Area under each curve is normalized to unity. Respective correlations between the experimental conformational change and theoretical binding-induced fluctuation profiles of PRS and ANM are 0.96 and 0.91 for ADK, 0.87 and 0.65 for AR, 0.95 and 0.88 for UCE, and 0.91 and 0.52 for FD. Insets show ribbon diagrams of superimposed proteins, with the open form in black and the closed form in gray.

Figure 2

(A) Average redundancy index versus variance plot for the proteins listed in Table 1, colored according to the mode that best represents the conformational change between the apo and holo forms. Experimental binding-induced conformational change in proteins with low redundancy index (and low variance) are well represented by one of the four slowest modes. (B and C) Probability distribution of redundancy index (B) and average path length versus average redundancy/residue (C) for the residues of ADK and OTF.

Figure 3

(A and B) Ribbon diagrams of GroEL colored according to overlaps between the theoretical and experimental conformational changes of the T→R transition (A) and R″→T transition (B). (C and D) Ribbon diagrams of ADK for the apo form (C) and the holo form (D) crystallized with inhibitor AP₅A. Each residue in the ribbon diagrams is colored according the overlap between the experimental binding induced conformational change and the profile of the theoretical response upon perturbing that residue, from darkest (highest overlap) to lightest (lowest overlap). The perturbed residues with the highest overlaps are shown as balls. Also displayed are normalized H⁻¹ matrices for open (E) and closed states (F). Normalization is carried out so that the displacement in each direction of every residue is equal. Color scale is from dark to light gray in print version.