Protein-Protein Interactions: What Are the Preferred Ways in Which Proteins Interact?

Ma B, Elkayam T, Wolfson H, and Nussinov R. Protein-protein interactions: structurally conserved residues distinguish between binding sites and exposed protein surfaces. Proc Natl Acad Sci U S A 100: 5772–7, 2003.

hat are the preferred ways in which proteins interact? The understanding and the ability to predict the preferred mode of protein associations can provide tremendous insight into molecular and cellular biology. Protein structures can associate in different ways. Mapping protein interactions on a genomic scale is essential for effective drug design. Protein-protein interactions are crucial to the understanding of practically all in vivo functions—cellular regulation, biosynthetic and degradation pathways, signal transduction, initiation of DNA replication, transcription and translation, multimolecular associations, packaging, the immune response, and oligomer formation. However, despite the broad recognition of the importance of deciphering the complex nature of protein interactions, they are still not entirely understood.

Assisted by our novel, amino acid sequence, order-independent multiple structural comparison algorithms (Shatsky M et al. Proteins 56: 143–56, 2004), we have been able to derive a structural dataset of protein-protein interfaces (Keskin O et al. Protein Science 13: 1043–55, 2004) to superimpose the multiple members of the interface clusters and obtain conserved residues (Ma B et al. Proc Natl Acad Sci U S A 100: 5772–7, 2003). We have observed that conserved residues correlate with the residue hot spots identified by alanine scanning mutagenesis (DeLano WL. Curr Opin Struct Biol 12: 14–20, 2002). Hot spots are residues that, when mutated to alanine, lead to an increase in the absolute binding energy by more than 2 kcal/mol (Bogan AA and Thorn KS. J Mol Biol 280: 1–9, 1998). We have further observed that the hot spots occur predominantly at the interfaces of macromolecular complexes, distinguishing protein binding sites from the remainder of the surface. Consequently, hot spots can be used to define binding epitopes—areas on proteins to which ligands bind. In particular, conservation of tryptophan on the protein surface indicates that it is highly likely that there is a binding site at this location. To a lesser extent, conservation of phenylalanine and methionine also implies a binding site. Figure 1 illustrates a protein-protein interface, a complex, and a hot spot residue.

Figure 1. The left-hand side illustrates an interface between two protein molecules. Only the backbone chains are shown. One protein chain is in red, the other in yellow. The residue hot spots (red and yellow balls, in the corresponding chains) are conserved in homologous interfaces. The homologous interfaces are shown in different colors (light blue, dark blue, green, and violet) superimposed on the red and yellow chains. Top, right-hand side: The entire protein-protein complex, in a space-filled representation. The backbones are highlighted as ribbons. Bottom, right-hand side: A residue hot spot (red ball) and the surrounding residues in the protein-protein interface.

We have investigated the organization of the hot spots and the structurally conserved residues (Keskin O et al. J Mol Biol 345: 1281–94, 2005). Interestingly, the hot spots are clustered within locally, highly packed regions, explaining the conservation of these residues and their large stabilizing contributions. Within the dense clusters, they form a network of interactions, and consequently, their contributions to the stability of the complex are cooperative. However, the contributions of independent clusters are additive. This suggests that the binding free energy is not a simple summation of the single hot spot residue contributions and highlights the similarity between binding and folding where conserved residues occur in folding nuclei.

Furthermore, this hot spot organization reconciles the apparent conflicting observations: On the one hand, electrostatic interactions are well known to be extremely important in protein associations (Sheinerman FB et al. Curr Opin Struct Biol 10: 153–9, 2000), yet on the other hand, we observed charge-charge conserved residue couples to be underrepresented (Halperin I et al. Structure [Camb] 12: 1027–38, 2004). The densely packed regions effectively screen the electrostatic interactions between the conserved charged hot spot residues. Thus, packing—a major player in folding—has a crucial role in binding. Conserved polar residues at the binding interfaces confer rigidity to minimize the entropic cost of binding, whereas the surrounding residues form a flexible cushion. Hence, the picture that emerges is that protein-protein associations are optimized locally, with the clustered, networked, highly packed structurally conserved residues contributing dominantly and cooperatively to the stability of the complex. When addressing the crucial question of “what are the preferred ways in which proteins interact,” these findings point toward a critical involvement of hot regions in protein-protein interactions.

Comparisons of different crystal packing interfaces or of mutant variants in the crystallographic database illustrate that similar protein structures can associate in different ways. However, the clustered protein-protein interface dataset further illustrates that proteins that have different functions, different sequences, and different global structures can associate in similar ways. It is intriguing that proteins, regardless of their family origins and functions, have preferred organizations. Binding and folding are thus similar processes, refined by evolution for function.

Preferred organization is a key in chemistry and in protein science, whether in amyloid microfilaments or in globular protein-protein associations. Evolution re-utilizes preferred, favorable patterns and modulates these toward different functions. This highlights the potential use of the protein-protein interface dataset toward binding site prediction. Structural comparisons of the interfaces against protein structures in the structural database are likely to identify currently unknown sites and identify (or validate) proposed functions, pathways, and cellular networks. Combined with other existing cellular-pathway and protein-interaction databases, such approaches will integrate biology at different levels. Above all, we envision that observations such as those described here and those attained from future work along such integrative lines will provide insight into the answer to one of the most profound of questions: Can we predict the ways proteins will likely interact?

Buyong Ma, PhD
Ozlem Keskin, PhD
Ruth Nussinov, PhD
Nanobiology Program
NCI-Frederick, Bldg. 469/Rm. 149
Tel: 301-846-5579
Fax: 301-846-5598
ruthn@ncifcrf.gov

back to top