Our laboratory investigates physical properties of biological macromolecules and seeks to understand how they define biological activity within the cell. An essential component of our work is the development of algorithms and computational software for extracting accurate biological insights from three-dimensional structural information. These downloadable programs and web-based tools are based on conceptual insights developed in our lab into the functional nature of protein structure, and have proved to be highly useful in facilitating novel insights into a variety of observed biological phenomena. Our lab’s work focuses on the following general themes:

Electrostatic interactions

Classical electrostatics, as reflected in the Poisson-Boltzmann (PB) equation, were used to study macromolecular electrostatics for many years. Our laboratory recognized that an accurate description of electrostatic effects requires that molecules be described in atomic detail because their shape, as well as their charge distribution, determines their electrostatic properties. This insight led to the development of DelPhi, which paved the way for our current understanding of the electrostatic properties of biological macromolecules.

During the development of DelPhi, we demonstrated, though numerous applications to a variety of biological phenomena, that electrostatic interactions play a central role in the structure and function of biological macromolecules, a fact that had not previously been fully appreciated. In a series of papers, we established the importance of electrostatics in enhanced diffusion, enzyme mechanisms, protein stability, protein-protein interactions, nucleic acid function, and the binding of proteins to charged systems such as nucleic acids and membrane surfaces. 

Within the context of electrostatics, we also developed a number of numerical tools to introduce a new approach, embodied in the GRASP program, to graphically describe the surface of macromolecules and to map various properties on that surface. GRASP is now widely used in the visual representation of molecules in structural biology. Publications of new protein structures typically contain GRASP surfaces depicting electrostatic potentials, and patches of enhanced positive or negative potential are often reliable predictors of functionally important regions.    

Protein-DNA recognition

Work in our lab has revealed a novel mode of sequence-specific protein-DNA recognition that is mediated by electrostatic interactions. In a collaborative series of papers with Richard Mann, we showed, through computational analysis of structural databases and electrostatic calculations, that many transcription factors achieve their DNA recognition specificity via the binding of positively charged amino acids to narrow regions of the minor groove in a sequence-dependent fashion. The underlying mechanism is based on the effect of molecular shape on electrostatic potentials, a phenomenon we had previously discovered.

The algorithmic and conceptual advances contained in these papers have enabled a new generation of structure-based tools for identifying transcription factor binding sites.

Cell-cell adhesion

In a long-term collaboration with crystallographer Lawrence Shapiro, we have revealed that evolution has conferred adhesion receptors such as cadherins precisely defined homophilic and heterophilic differential binding affinities that determine cell sorting behavior. This discovery was made utilizing a combination of biophysical studies, statistical mechanical theory, and multi-scale simulation.

We determined, for example, that adherens junctions, cell-cell junctions whose formation is mediated by the dimerization of cadherins located on apposed cells, are formed via a two-dimensional phase transition, involving trans (apposed cell) and cis (same cell) interactions.

To establish the underlying mechanism, we solved a long-standing problem of general importance by developing a new theory and simulation methodology that transforms binding affinities measured in 3D solution to 2D affinities that are relevant to the constrained environment of a membrane surface. The simulations revealed how adhesion receptors form ordered two-dimensional supramolecular assemblies following cell-cell contact and provide a new conceptual framework for studying the organization of membrane-mediated macromolecular signaling complexes.

We believe that these findings hold great potential for understanding cell-cell adhesion and could enable myriad biomedical applications.

Structural bioinformatics

After writing a series of papers on the physical forces that drive protein folding and association, we undertook a theoretical program aimed at combining biophysics and bioinformatics in the prediction of protein structure and function. We developed new homology methods in protein structure prediction including structure-based sequence alignment, side chain prediction, loop prediction and model generation that all introduced algorithmic advances that have since been widely adopted.

Most recently, we have proposed that protein structure space is not adequately utilized in the detection of functional and evolutionary relationships between proteins. We have been developing methods that use local geometric similarity to search broadly for novel functional relationships throughout protein-structure space using what we call the “Structural Blast” concept to identify small structural fragments of proteins that reveal functional information.

The enormous potential of this research direction is evident in a 2012 paper describing a new method, called PrePPI, for predicting protein-protein interactions for entire genomes. PrePPI uses structural information on an unprecedented scale, paving the way for the integration of structural biology and systems biology.

The PrePPI database, accessible from the Honig lab web site, contains over 300,000 predicted high confidence protein-protein interaction for the human genome and will soon contain separate databases for fly, mouse, maize and many other genomes. Its accuracy has been shown to be comparable to that obtained from high throughput experimental methods but the information is purely in silico and is publically available.