Exploring Protein Fold Space

The model of protein folding proposed by Ptitsyn and colleagues involves the accretion of secondary structures around a nucleus. As developed by Efimov, this model also provides a useful way to view the relationships among structures. Although somewhat eclipsed by later databases based on the pairwise comparison of structures, Efimov’s approach provides a guide for the more automatic comparison of proteins based on an encoding of their topology as a string. Being restricted to layers of secondary structures based on beta sheets, this too has limitations which are partly overcome by moving to a more generalised secondary structure lattice that can encompass both open and closed (barrel) sheets as well as helical packing of the type encoded by Murzin and Finkelstein on small polyhedra. Regular (crystalline) lattices, such as close-packed hexagonals, were found to be too limited so pseudo-latticses were investigated including those found in quasicrystals and the Bernal tetrahedron-based lattice that he used to represent liquid water. The Bernal lattice was considered best and used to generate model protein structures. These were much more numerous than those seen in Nature, posing the open question of why this might be.

Shared unfolding pathways of unrelated immunoglobulin-like β-sandwich proteins

The Dynameomics project contains native state and unfolding simulations of 807 protein domains, where each domain is representative of a different metafold; these metafolds encompass ~97% of protein fold space. There is a long-standing question in structural biology as to whether proteins in the same fold family share the same folding/unfolding characteristics. Using molecular dynamics simulations from the Dynameomics project, we conducted a detailed study of protein unfolding/folding pathways for 5 protein domains from the immunoglobulin (Ig)-like β-sandwich metafold (the highest ranked metafold in our database). The domains have sequence similarities ranging from 4 to 15% and are all from different SCOP superfamilies, yet they share the same overall Ig-like topology. Despite having very different amino acid sequences, the dominant unfolding pathway is very similar for the 5 proteins, and the secondary structures that are peripheral to the aligned, shared core domain add variability to the unfolding pathway. Aligned residues in the core domain display consensus structure in the transition state primarily through conservation of hydrophobic positions. Commonalities in the obligate folding nucleus indicate that insights into the major events in the folding/unfolding of other domains from this metafold may be obtainable from unfolding simulations of a few representative proteins.

A Large-Scale Structural Classification of Antimicrobial Peptides

Antimicrobial peptides (AMPs) are potent drug candidates against microbial organisms such as bacteria, fungi, parasites, and viruses. AMPs have abundant sequences and structures, two fundamental resources for bioinformatics researches, but analyses on how they associate with each other are either nonexistent or limited to partial classification and data. We thus present A Database of Anti-Microbial peptides (ADAM), which contains 7,007 unique sequences and 759 structures, to systematically establish comprehensive associations between AMP sequences and structures through structural folds and to provide an easy access to view their relationships. 30 distinct AMP structural fold clusters with more than one structure are detected and about a thousand AMPs are associated with at least one structural fold cluster. According to ADAM, AMP structural folds are limited—AMPs only cover about 3% of the overall protein fold space.

Dynameomics: protein dynamics and unfolding across fold space

All currently known structures of proteins together define ‘protein fold space’. To increase the general understanding of protein dynamics and protein folding, we selected a set of 807 proteins and protein domains that represent 95% of the currently known autonomous folded domains present in globular proteins. Native state and unfolding simulations of these representatives are now complete and accessible via a novel database containing over 11 000 simulations. Because protein folding is a microscopically reversible process, these simulations effectively sample protein folding across all of protein fold space. Here, we give an overview of how the representative proteins were selected and how the simulations were performed and validated. We then provide examples of different types of analyses that can be performed across our large set of simulations, made possible by the database approach. We further show how the unfolding simulations can be used to compare unfolding of structural elements in isolation and in different structural contexts, using as an example a short, triple stranded β-sheet that forms the WW domain and is present in several larger unrelated proteins.