Nikolay V. Dokholyan

he physical nature of interactions between subatomic particles, atoms, molecules, cells, and organisms has always intrigued me. The underlying question throughout my research is how these interactions shape the complex organization, behavior, and evolution of biomolecules and organisms. To approach this question I have been studying the following topics, among others: (i) statistical properties of genetic material of various species --- DNA molecules, (ii) physical properties of the workers of the cell --- proteins, (iii) physics of cell motility. Such a broad approach is necessary to tie together the diverse pieces of knowledge of molecular properties and evolution that is available to us. My present principal effort is directed towards understanding the nature of physical interactions between amino acids in proteins and the impact of these interactions on the chemical and biological properties of proteins and, at a higher level, cells and organisms. Next I describe selected parts of my past and present research and tentative future proposals.

I. Present and Past Research

The physics of protein folding: As a physicist, my first questions relate to the physical properties of proteins (Refs. [17, 20, P7] in my Curriculum Vitae). Specifically, what mechanisms are responsible for the rapid (1ms - 1s) folding transition of proteins to their unique native (ground) state? The fact that proteins find their native state much faster than would be expected in a random search was first pointed out by Levinthal in 1968. Among several mechanisms proposed to explain protein folding, I tested the nucleation scenario --- the formation of a specific set of contacts, the nucleus, that is necessary and sufficient for the rapid descent of the protein to its native state --- in small protein models using molecular dynamics simulations [18]. The methods I developed with coworkers allowed us to study nucleation processes in real proteins, such as Src SH3 [P10] and c-Crk SH3 [P9] domain proteins. The surprising outcome of our studies of SH3 domain proteins was that we predicted that (i) a specific contact at the folding transition between amino acids L24 and G54 in Src SH3 domain, and (ii) the formation of a hydrogen bond network between E16, M48, and L18 in c-Crk SH3 domain are crucial for the formation of the native state. I plan to test this prediction experimentally in collaboration with Dr. David Baker (University of Washington).

I am also searching for alternatives to computationally heavy molecular dynamics simulations in studies of the protein folding kinetics. Specifically, I have used experimentally determined structural and physical protein properties to understand protein folding kinetics. By mapping protein 3D structures to graphs, we demonstrated that protein graphs constitute a special class, called ``small-world networks''. Using graph theory, we identified key kinetically important residues for a set of real proteins [P8]. In a separate study, using graph theory, we dissected the transitional state ensembles of two proteins Src SH3 domain and Chymotrypsin Inhibitor 2 and identified the only topological features that distinguish the pre- and post transitional states of these proteins [P6].

Protein evolution: One can understand how evolution controls biological organism diversity on the macroscopic scale by controlling protein evolution on the much smaller scale of biological molecules. The complexity of this effect is phenomenal since its span scales from Angstroms (length scale of an amino acid) to meters (the length scale of an organism), which is a difference of the magnitude of 10¹⁰. An understanding of protein evolution will shed light on the macroscopic evolution of organisms and will impact such disciplines as sociology and behavioral science. In my first attempt [P3,21] to uncover the physical mechanisms underlying evolution, I proposed a model that provides an explanation for the hierarchical organization of proteins into families of structurally similar proteins --- fold families. The model, which is based on the evolutionary selection of proteins by their native state stability, reproduces patterns of amino acids conserved across protein families.

There are proteins that share similar 3D structures but have few identical amino acids in common in the corresponding positions of their aligned sequences. Some pairs of proteins sharing the same fold have sequence similarity as low as expected for random sequences 8 -- 9%. I estimated the number of protein sequences that correspond to a given 3D structure [P5]. I predicted that this number depends strongly on the thermodynamic stability of a given structure --- the more stable the structure, the less sequences there are that correspond to that structure. I tested the predictions against data available in the FSSP database on the families of structurally similar proteins and found that the actual data is in agreement with my predictions.

Protein engineering: The most difficult and challenging tests of our understanding of proteins are the abilities (i) to predict a protein's 3D structure from its sequence, (ii) to fold ab initio a given protein in molecular dynamics simulations, and (iii) to engineer proteins with desired physical, chemical and biological properties. The principal task is to identify a set of amino acid interaction parameters that would satisfy all three tests (i) -- (iii).

I have derived amino acid interaction parameters for a single protein, Src SH3 domain. I tested the derived potential using discrete molecular dynamics simulations by refolding SH3 from high temperatures to lower ones. In most simulations, it reaches its native state domain. The overall final conformations have the same topology as the native state of a protein and the best root-mean-square displacement of the aligned structures of Src SH3 domain obtained from the computer simulations and its native crystal structure is around 3.6 A. Further side-chain packing simulations at the all-atom level will be necessary for the refinement of the ``designed'' potential parameters. This work is an ongoing project.

Properties of DNA sequences: Another way to understand proteins is to study the genetic material of organisms --- DNA molecules. The question is what is specific to protein coding DNA sequences? I studied repetitive patterns of coding and noncoding regions of DNA. I found that n-tuple repeats have different properties in coding versus noncoding DNA. I developed two models based on biochemical mechanisms inside the living cell, to which can be attributed the difference in properties of the n-tuple repeats in coding and noncoding DNA [2-7]. The difference in the the distribution of the n-tuple repeats in coding and non-coding DNA may aid gene recognition.

II. Future Directions

I plan to continue and broaden my study of physical interactions and their effect on the organization of biomolecules and organisms and their evolution.

The physics of ligand binding: Understanding the interaction of proteins with other molecules in cells is cardinal to our perception of the life of organisms. This understanding is thus crucial for developing effective pharmaceuticals. I plan to study the phyics of ligand binding in molecular dynamics simulations and by means of bioinformatics for homology modeling of the binding sites. I aim to uncover the physical and chemical principles underlying interactions of proteins with ligands and to predict the binding properties of a given system of proteins and ligands. To achieve this aim I plan to develop fast and reliable ab initio folding methods relying on more detailed models of protein energetics (e.g. distant-dependent potentials, explicit side-chains). Such models can be created by the joint efforts of computational biologists, crystallographers, NMR scientists and microscopists. Simulations can be of great importance from both fundamental and therapeutic points of views.

Protein engineering: I will direct my effort toward engineering the interaction potential for several proteins in order to determine general and more refined interaction parameters that can be applied in the structure prediction of an arbitrary protein. I would like to build a research group that will help me to attack this problem en masse for a large number of proteins computationally. In addition, I will establish collaborations with experimental groups to test computationally derived predictions by means of site-directed mutagenesis.

I plan to apply and further refine the methodology of protein property manipulation to other proteins of higher complexity. Ultimately, I will pursue my hopes for understanding the complex web of interactions of a protein with other molecules in the cell and I will pursue the challenge of altering the biological properties of cells. Such an achievement may lead to a deeper understanding of organism physiology and aid in the discovery of more effective approaches to medicine.

Protein evolution: I am intrigued by the large amount of structural and sequence data available to scientists. I would like to pursue the problem of understanding and predicting protein function by the comparative analysis of orthologous and paralogous proteins from fully sequenced genomes. I would also like to unveil the basic principles of protein function evolution. Specifically, I would like to study the co-evolution of interacting networks of proteins that is crucial for signal transduction processes. The goal is to identify the specificity-determining residues in pairs of interacting proteins, and to predict interaction partners in the genome. My ultimate goal is to understand how the evolution of biomolecules shapes the evolution of organisms and societies.

DNA Promoter recognition: I have developed a passion during my graduate studies for searching for regularities in noisy systems using the knowledge of the physical properties of these systems. One idea I wish to pursue is the identification of promoters --- genetic regulatory system --- crucial for the identification of protein coding regions in DNA molecules. I believe this can be done using the joint knowledge of the physical properties of DNA sequences and the field of bioinformatics.

Download Research Statement: [Adobe PDF] [PostScript]