Protein-protein interactions and assembly in the cell

In my lab, the broad goal is to understand the mechanisms and cellular control of protein-protein self-organization processes and to study general principles governing protein-protein interactions in the cell. Our approach combines theory, model building, and computer simulation, with collaboration with experimental groups.

Spatial and temporal dynamics of vesicle formation in clathrin-mediated endocytosis

Computational modeling is ideally suited to testing the mechanism of protein-coated vesicle formation as a useful complement to challenging experiments that can, at best, partially track the participating molecules. The question of clathrin coated vesicle formation is both practical and amenable to theoretical considerations, first because of the wealth of existing experimental data, and second because despite this information, the range of molecular conditions are too extensive to be tested experimentally.
The models we develop characterize the collective mechanics of the coat formation at a very high spatial (nm) and temporal (sub-us) resolution.

We aim to understand how the timing and efficiency of clathrin-coated vesicle formation (pre-scission) is initiated and controlled by distinct sets of proteins that often possess overlapping functions.  We are quantifying the role of the membrane in regulating the initiation and progression of clathrin-coat formation, and characterizing intermediates along the pathways to successful cage formation.

Membrane-mediated protein interactions

In many cellular pathways, including endocytosis, cell-division, and receptor-mediated signaling, proteins must localize to membranes as well as interact with other proteins. For proteins that can bind to one another both in solution and on the membrane, the membrane offers a reduced search space (in nearly all cases) that will thus concentrate proteins and promote more binding. We develop theory to quantify how significant a role this dimensionality reduction plays in controlling protein interactions. Quantifying this behavior is critical to understanding assembly on surfaces because 2D localization will strengthen binding reactions regardless of whether additional factors, such as curvature generation or membrane microdomains also influence binding. We are using stochastic non-spatial and reaction-diffusion simulations to characterize the dynamics of membrane-mediated assembly for proteins from diverse pathways in yeast and human cells. We are also using Molecular Dynamics simulations to study effects of 2D localization on binding kinetics.

Systems biology of protein interaction networks

Relationships between network structure and protein dynamics, function, and evolution. Protein-protein interaction networks specify which proteins bind to one another in the cell to perform their biological function. However, they do not adequately capture the competition between proteins to bind to shared partners and the cooperation between proteins to form multi-protein complexes. These details on protein binding interfaces as well as protein concentrations are critical to understand how proteins dynamically interact with one another in the cell and avoid forming spurious, nonfunctional and potentially disease causing associations. We investigate how the structure of the network and the availability of proteins contributes to their binding specificity and functional evolution.

We have recently characterized how, for a given protein interaction network, the interface interaction network has a highly specialized structure that reflects the need for proteins to maintain specificity and complementarity in both the structure and sequence of binding interface pairs. In both the clathrin-mediated endocytosis network of yeast, and the ErbB signaling network of human cells, these interface networks contain an abundance of specific network motifs that optimize chemical and structural complementarity, and a small selection of sub-optimal motifs that often indicate regulatory control of binding. The steep constraints on the structure of the interface network ultimately constrain the parent protein-protein interaction network. We found that protein networks that contain hubs provide the optimal structure for generating highly selective interface networks with a minimum number of total interfaces.

New methods for physics-based modeling of protein interactions in the cell

Many challenges exist in simulating cell-scale dimensions while retaining accuracy in describing the physics of protein association. We develop new methods to allow for high resolution spatial and temporal modeling of protein assembly through both rigorous and efficient algorithms. To study non-equilibrium processes as well as equilibrium dynamics, we use mainly reaction-diffusion techniques. Our Free-propagator Reweighting (FPR) algorithm correctly capture reaction dynamics (whether diffusion-influenced or not) in solution, on surfaces, and as recruited to surfaces. We are also developing methods for simulating membrane deformation in connection with reaction-diffusion dynamics. We compare these new methods with existing stochastic and deterministic spatial and non-spatial techniques, along with theory, for careful validation and to establish the regimes where spatial effects or diffusion are important sources for controlling reaction dynamics.