Accurate peptide-protein docking tools

A main focus of our group is the development of protocols for the accurate modeling of peptide-mediated interactions. In terms of application, these bring in reach the detailed study of interactions that were previously out of reach. In terms of basic research, these reveal basic strategies used by Nature to allow refined regulation.

Rosetta FlexPepDock was developed to address the challenge of modeling the conformational flexibility upon binding encountered in peptide-mediated interactions.

Rosetta FlexPepDock was one of the first dedicated peptide-protein docking protocols that explicitly model the flexibility of a peptide during binding. It has allowed the structural characterization of many peptide-mediated interactions, and raised the interest in peptide-protein interactions in the CAPRI docking community and beyond.

Rosetta FlexPepDock has motivated the development of a variety of additional peptide- docking protocols (summarized in a Book “Modeling Peptide-Protein Interactionsedited by us).

Blind global ab initio peptide docking

PIPER-FlexPepDock is the first ab initio peptide docking protocol to allow high-resolution modeling of peptide-protein complexes when only the sequence of the peptide and the receptor structure are known.

The protocol follows the original Rosetta ab initio concept of the use of fragment libraries to accelerate search, based on the observation that bound peptide conformations can frequently be found among sequence - and (predicted) secondary structure-similar fragments from solved monomer structures. This allows to decouple peptide conformational search from docking:

  1. Determine a set of fragments to represent the conformational ensemble of a peptide (Rosetta Fragment Picker)
  2. Rigid-body dock each fragment to the receptor (PIPER FFT-based exhaustive docking)
  3. Refine top-ranking models to model induced fit (using Rosetta FlexPepDock)

The results of a representative benchmark indicate substantial improvements over existing protocols.

References:

[Alam et al.], [Proter et al.], [Raveh et al.] and [Kozakov et al.].

Structure-based identification of peptide substrates

Structure-based identification of substrates can complementsequence-motif-based identification of readers, writers and erasers of post-translational modifications.

Rosetta FlexPepBind uses FlexPepDock to model different peptide substrates into the receptor binding pocket and to evaluate their ability to bind in a (catalysis-competent) low-energy conformation.

The approach is calibrated for a specific system based on a template substrate-receptor structure and a small set of characterized substrates/non-substrates.

Using this protocol, we have been able to successfully identify, and experimentally validate, many new substrates for a range of post-translational modifications such as prenylation, deacetylation, proteolytic cleavage and many more. These substrates provide valuable information about new connections in the network of protein interactions, and cross-regulation between different post-translational modifications.

Hot linear segments at the protein interface

In addition to classical peptide-protein interactions, our work has pinpointed the many additional interactions between structured domains that are dominated by one dominant linear segment - a peptide.

Rosetta PeptiDerive detects such “hot segments” that can serve as starting point for the design of peptides and peptidomimetics for specific manipulation of an interaction.

Peptide Design using FlexPepDock

Rosetta FlexPepDock is easily extended to design. Extension of an inhibitory peptide into nearby binding pockets is a promising strategy to increase binding affinity and specificity of inhibitory peptides.

In a first step we identify potential peptide binding pockets using our PeptiMap protocol. PeptiMap is based on the observation that solvents often map to binding sites on a protein. Using efficient solvent mapping, FTmap was developed by our collaborators (Dima Kozakov and Sandor Vajda) to identify computationally ligand binding site. PeptiMap is an adapted version that is taylored specifically to peptide ligands.

We then apply FlexPepDesign, an extension of Rosetta FlexPepDock that includes a design step during the docking process to, design a peptide extension into that pocket. As an example case, we have characterized the hotspots of an inhibitory peptide of a tick protease, based on a solved structure. Starting from this structure, we have applied this strategy to increase the binding affinity by two orders of magnitude, to sub-nanomolar affinity, this an ongoing Collaboration with the Mares Lab, Prague University. Results from this work will be published in the near future.

References:

[Lavi et al.], [Bohnuud et al.], [Hanova et al.]

Integration strategies of multiple peptide-protein interactions in cellular regulation

To appreciate the regulatory role and complexity of peptide-mediated interactions, it is crucial to consider the broader context: peptide- protein interactions do not occur alone, but involve often many peptide-domain connections between two protein partners (and more). We have chosen the WW domain as a prototype for a peptide-binding domain that often occurs in tandem in a protein to study the different integration strategies of information from multiple peptide binding domains.

Specifically, we focus on two key regulatory proteins, the tumor suppressor WWOX that competes with YAP for target binding, to determine cell fate. Both WWOX and YAP contain two tandem WW domains, but the interplay between these domains differs significantly. Interestingly, in WWOX the second domain, WW2, does NOT bind to any peptide motif, and its contribution is therefore unclear. Using a range of different experimental approaches, including NMR, CD, ITC, and more, we have investigated how WW2 affects protein stability and peptide binding. Our results reveal a complex stabilization mechanism whereby both WW2 and the peptide stabilize the WW1 domain using overlapping and distinct regions (under preparation for publication).

References:

[Dobson et al.], [Abu-Remaileh et al.]