How do soluble proteins bind to biological membranes?

Proteins are Nature's building blocks but not only, they are dedicated and skilled workers, each fulfilling its own specific role. They are fairly large molecules, containing thousands of atoms. When one or several key proteins are not able to fulfill their function in the human body, this might potentially result in diseases. Although proteins are large for being molecules, they are rather small in size and are difficult to observe directly. Hence it is often handy to build models, using computers, and simulate their behavior. The models are built from physics and chemistry principles, and the simulations have proven to be useful to learn more about how, for example, proteins can bind to cell membranes or if and how drugs can bind to proteins and affect their behavior. In this project we focused on a particular type of proteins, which perform their function at the surface of the cells and have the ability to stick to the membrane of the cells for just the right amount of time to achieve their tasks there, and leave again when the job is done. We aimed at finding out which are the forces that bring these proteins to the membrane, and how the duration of that stay is regulated by the same forces. We made a better tool for simulations of proteins binding to membranes, an atomistic force field for molecular dynamics simulations which we have improved and tested so that it is reliable for interactions between choline chemical groups and aromatic amino acids. We also did a similar work on a less detailed force field (coarse grained). Using the improved atomistic force field we could get a detailed picture of the energetics of insertion of aromatic amino acids at the interface region of choline-containing lipid bilayers. In addition to that, we developed a mathematical model to find what differentiates proteins that bind to cell membranes from those that don?t. Applying our model to a large dataset of proteins, we could show which structure or shape, and which amino acids need to be present on the surface of a protein for it to bind to a membrane. The results of that project can be used to design proteins so that they have the ability to bind membranes, or to design small molecules that can inhibit protein-membrane binding.

The project looked into the modeling of cation-pi interactions in the context of peripheral protein-membrane binding. The force field parameters developed in this project are included with the official CHARMM distribution, and in the coarse grain MARTINI force field. They will allow members of our community to more reliably model aromatics-choline interactions. This is relevant not only in the context of protein-membrane interactions, but also for ligand design and drug discovery, for example. The understanding of the role of aromatic amino acids in peripheral membrane binding brought by this project is crucial in our understanding of peripheral protein-membrane interactions. It shows that aromatics are not simply hydrophobic anchors and will influence how we predict peripheral membrane binding sites. Through their work on this project, the youngest member of the staff have developed new skills and been exposed to international collaborations.

The study of protein-lipid interactions is of fundamental importance but data on protein-membrane interfaces is still scarce compared to what we know about protein-protein interactions. Our knowledge of the structural fingerprint of protein-membrane interfaces relies on an old model that needs to be updated. The text-book description of a prototypical interfacial binding site consists of clusters of basic amino acids and a few hydrophobic amino acids. Relying on observations we and others made on a few membrane-binding proteins1-6 we believe that a distinction should be made between aliphatic amino acids and the three aromatic amino acids, and particularly between these three. They have different physicochemical properties, anchor at different depths, and could play an underestimated role in lipid specificity. Furthermore our work indicates that the role of positively charged amino acids is possibly too simplistic in the current model. Encouraged by unpublished preliminary results, and unlike what has been attempted so far, we intend to study a large protein dataset to evaluate to which extent our findings can be generalized. We will use a multidisciplinary approach combining bioinformatics, molecular simulations and biophysics. In particular we propose to improve the current model by (1) mapping the interactions between aromatic amino acids and membrane phospholipids and (2) quantifying the spectra of nonspecific electrostatics interactions among membrane-binding proteins. We will for these purposes also contribute methodological developments relevant to biomolecular modeling and theoretical chemistry. Numerous human diseases are attributed to alterations of membrane-protein interactions. In the era of lipidomics and with the current efforts aiming at drawing protein-lipid interactome maps7, our project is timely and relevant for molecular life science at large. It will have impact on the annotation of lipid-binding domains.