Hydrophobic complementarity in protein-protein docking

Looking at enzymes from the inside out: the proximity of catalytic residues to the molecular centroid can be used for detection of active sites and enzyme-ligand interfaces

Analysis of the distances of the exposed residues in 175 enzymes from the centroids of the molecules indicates that catalytic residues are very often found among the 5% of residues closest to the enzyme centroid. This property of catalytic residues is implemented in a new prediction algorithm (named EnSite) for locating the active sites of enzymes and in a new scheme for re-ranking enzyme-ligand docking solutions. EnSite examines only 5% of the molecular surface (represented by surface dots) that is closest to the centroid, identifying continuous surface segments and ranking them by their area size. EnSite ranks the correct prediction 1-4 in 97% of the cases in a dataset of 65 monomeric enzymes (rank 1 for 89% of the cases) and in 86% of the cases in a dataset of 176 monomeric and multimeric enzymes from all six top-level enzyme classifications (rank 1 in 74% of the cases). Importantly, identification of buried or flat active sites is straightforward because EnSite "looks" at the molecular surface from the inside out. Detailed examination of the results indicates that the proximity of the catalytic residues to the centroid is a property of the functional unit, defined as the assembly of domains or chains that form the active site (in most cases the functional unit corresponds to a single whole polypeptide chain). Using the functional unit in the prediction further improves the results. The new property of active sites is also used for re-evaluating enzyme-inhibitor unbound docking results. Sorting the docking solutions by the distance of the interface to the centroid of the enzyme improves remarkably the ranks of nearly correct solutions compared to ranks based on geometric-electrostatic-hydrophobic complementarity scores.

Adaptation of a fast Fourier transform-based docking algorithm for protein design

Designing proteins with novel protein/protein binding properties can be achieved by combining the tools that have been developed independently for protein docking and protein design. We describe here the sequence-independent generation of protein dimer orientations by protein docking for use as scaffolds in protein sequence design algorithms. To dock monomers into sequence-independent dimer conformations, we use a reduced representation in which the side chains are approximated by spheres with atomic radii derived from known C2 symmetry-related homodimers. The interfaces of C2-related homodimers are usually more hydrophobic and protein core-like than the interfaces of heterodimers; we parameterize the radii for docking against this feature to capture and recreate the spatial characteristics of a hydrophobic interface. A fast Fourier transform-based geometric recognition algorithm is used for docking the reduced representation protein models. The resulting docking algorithm successfully predicted the wild-type homodimer orientations in 65 out of 121 dimer test cases. The success rate increases to approximately 70% for the subset of molecules with large surface area burial in the interface relative to their chain length. Forty-five of the predictions exhibited less than 1 A C(alpha) RMSD compared to the native X-ray structures. The reduced protein representation therefore appears to be a reasonable approximation and can be used to position protein backbones in plausible orientations for homodimer design.

Modeling oligomers with Cn or Dn symmetry: Application to CAPRI target 10

The abundance of oligomeric proteins makes them a frequent target for structure prediction. However, homologous proteins sometimes adopt different oligomerization states, rendering the prediction of structures of whole oligomers beyond the scope of comparative modeling. This obstacle can be overcome by combining comparative modeling of the single subunit of an oligomer with docking techniques, designed for predicting subunit-subunit interfaces. We present here algorithms for predicting the structures of homo-oligomers with C(n) or D(n) (n > 2) symmetry. The prediction procedure includes a symmetry-restricted docking step followed by a C(n) or D(n) oligomer-forming step, in which the dimers from the docking step are assembled to oligomers. The procedure is applied to each of the crystallographically independent subunits in 8 C(n) and 3 D(n) oligomers, producing very accurate predictions. It is further applied to a single monomer of the tick-borne encephalitis virus coat protein E (Target 10 of the CAPRI experiment). The predicted trimer ranked 30, obtained via rigid-body geometric-hydrophobic docking followed by C(n) oligomer formation, is very similar to the experimentally observed trimer formed by domain II of this protein. Furthermore, the predicted trimer formed from the separated domain I is also close to the experimental structure.

Docking to single-domain and multiple-domain proteins: Old and new challenges

The diverse selection of targets in the CAPRI experiments provides grounds for determining the limits of our rigid-body docking program MolFit, and for extending it. We find that the sensitivity of MolFit is high, enabling it to produce reasonably accurate docking solutions when the structures undergo moderate local conformation changes upon complex formation or when the docked molecules are modeled. Yet the ranks of these solutions are sometimes too low to meet the requirements of CAPRI assessment. This indicates that the selectivity of MolFit, which was optimized for docking of unbound X-ray structures, and which relies on the availability of external data from biochemical and bioinformatic sources, needs readjustment in order to meet the challenges presented by NMR or modeled structures. A different challenge is presented by large global conformation changes such as movements of domains. We show that such changes can be accommodated within the rigid-body approximation by employing rigid multibody multistage docking procedures. We also address the difficulty of ranking results from 2-body and multibody docking scans in cases in which there are no external data favoring one option over the other.

The effect of resolution-dependent global shape modifications on rigid-body protein-protein docking

Docking unbound molecules presents a challenge in the case where no prior biological or bioinformatic knowledge exists. This is mainly due to differences between the structures of the molecules when in a complex and in the free state. Presumably, these differences interfere with the ability of protein-protein docking algorithms, which rely on a dominant shape descriptor, to identify the correct solution and rank it higher than false solutions. In this study we verify the notion that small discords in the molecular fit can be eliminated by using appropriately designed low-resolution shape descriptors, thereby improving the docking results. We exploit the inherent gradual resolution dependency of Fourier transforms and formulate a resolution-dependent shape descriptor by truncating selected Fourier transform terms. Thus, different levels of shape modification are attained, affecting the degree of detail in the depiction of the molecular surface. We applied the modified descriptor to a selection of 23 protein-protein systems, using the unbound structures where possible. The docking results obtained with the new geometric descriptor were considerably superior to former results, improving the ranks of nearly correct solutions for 17 systems. Unification of the results of scans in which different resolutions were employed further improved the ranks of nearly correct solutions to less than 100 for 12 of the 23 systems and less than 300 for 20 systems. The new geometric descriptor can be combined with other descriptors, which typify the electrostatic or hydrophobic character of the molecular surface, and with external experimental or bioinformatic data.

Analysis of ordered and disordered protein complexes reveals structural features discriminating between stable and unstable monomers

Most proteins exist in the cell as multi-component assemblies. However, which proteins need to be present simultaneously in order to perform a given function is frequently unknown. The first step toward this goal would be to predict proteins that can function only when in a complexed form. Here, we propose a scheme to distinguish whether the protein components are ordered (stable) or disordered when separated from their complexed partners. We analyze structural characteristics of several types of complexes, such as natively unstructured proteins, ribosomal proteins, two-state and three-state complexes, and crystal-packing dimers. Our analysis makes use of the fact that natively unstructured proteins, which undergo a disorder-to-order transition upon binding their partner, and stable monomeric proteins, which exist as dimers only in their crystal form, provide examples of two vastly different scenarios. We find that ordered monomers can be distinguished from disordered monomers on the basis of the per-residue surface and interface areas, which are significantly smaller for ordered proteins. With this scale, two-state dimers (where the monomers unfold upon dimer separation) and ribosomal proteins are shown to resemble disordered proteins. On the other hand, crystal-packing dimers, whose monomers are stable in solution, fall into the ordered protein category. While there should be a continuum in the distributions, nevertheless, the per-residue scale measures the confidence in the determination of whether a protein can exist as a stable monomer. Further analysis, focusing on the chemical and contact preferences at the interface, interior and exposed surface areas, reveals that disordered proteins lack a strong hydrophobic core and are composed of highly polar surface area. We discuss the implication of our results for de novo design of stable monomeric proteins and peptides.

On proteins, grids, correlations, and docking

The activity of a living cell can be portrayed as a network of interactions involving proteins and nucleic acids that transfer biological information. Intervention in cellular processes requires thorough understanding of the interactions between the molecules, which can be provided by docking techniques. Docking methods attempt to predict the structures of complexes given the structures of the component molecules. We focus hereby on protein-protein docking procedures that employ grid representations of the molecules, and use correlation for searching the solution space and evaluating putative complexes. Geometric surface complementarity is the dominant descriptor in docking. Inclusion of electrostatics often improves the results of geometric docking for soluble proteins, whereas hydrophobic complementarity is more important in construction of oligomers. Using binding-site information in the scan or as a filter helps to identify and up-rank nearly correct solutions.