BIOGRAPHY

Konstantin Zeldovich graduated from Moscow State University in Russia in 1998 with a M.S. in Physics. In 2001, he received his PhD in physics and mathematics also from Moscow State University. His thesis research, supervised by Prof. Alexei Khokhlov, was focused on the theoretical study of gels and adsorbed layers of charged polymers, or polyelectrolytes. Between 2002 and 2004, Dr. Zeldovich was a postdoctoral fellow at the Institut Curie in Paris, France, where he worked in the area of statistical physics of molecular motors, such as kinesin and myosin in the laboratory of Profs. J.-F. Joanny and J.Prost. Later, from 2004 till August 2008, Dr. Zeldovich was a postdoctoral fellow at the laboratory of Prof. Eugene Shakhnovich at the Department of Chemistry and Chemical Biology, Harvard University. His research there focused on developing novel, physics-inspired models of molecular evolution, as well as protein thermostability and their sequence-structure relationships.

On September 1, 2008, Dr. Zeldovich was appointed tenure-track assistant professor at the Program in Bioinformatics and Integrative Biology at the University of Massachusetts Medical School. His research deals with molecular evolution, and protein folding, stability, and interactions.

Physics-based models of molecular evolution

Earlier, we have proposed a new model of molecular evolution, which considers evolution as a diffusion process in the space of stabilities of the organism’s proteins [Zeldovich, Chen, Shakhnovich, PNAS 2007]. Indeed, when a protein undergoes a mutation, its thermodynamic stability (folding free energy dG) changes by a small amount. In present-day proteins, these changes are normally detrimental, decreasing the stability. However, a small fraction of mutations makes proteins more stable. Looking at each protein, this process can be thought of as a biased diffusion along the coordinate of dG. Now, consider the organism as a whole, containing N proteins in the genome. Exceptional cases aside, all of the essential proteins encoded by the organism’s genome must be stable (dG<0) in order to work properly. Otherwise, the organism may not be viable. A mathematical treatment of this problem leads to a diffusion equation with an absorbing boundary, which allows an exact solution. In agreement with bioinformatics data, the model predicts that organisms with higher mutation rates have shorter genomes (e.g. RNA-based vs DNA-based viruses), organisms living at elevated temperatures (thermophiles) have shorter genomes, and the probability distribution of stability of evolved proteins has a certain shape.

Future research along these lines includes explicit consideration of the genetic code, intrinsically unstructured proteins, and incorporation of epistasis, where effects of mutation in one gene can be significantly altered by a mutation in another gene.

Protein Thermostability

What can one say about a bacterium just by looking at its genome? It turns out that the temperature of the natural environment of the bacterium can be inferred just by looking at the amino acid composition of the bacterium’s proteins [Zeldovich, Berezovsky, Shakhnovich, PLoS Comp Biol 2007]. The sum of the fractions of amino acids IVYWREL in the genome can be used to predict the environmental temperature up to about 10 degree C. A simple model of lattice proteins [Berezovsky, Zeldovich, Shakhnovich, PLoS Comp Biol 2007] suggested that amino acids responsible for high thermostability must come from the opposite ends of the hydrophobicity spectrum, i.e. the magic combination should contain both very hydrophobic and very hydrophilic amino acids. Unfortunately, we are still lacking the microscopic understanding as to why exactly these amino acids are so tightly linked to thermostability.

One of the research projects of the lab involves extensive Monte-Carlo simulations of proteins under different temperatures, aimed at the precise, quantitative understanding of the mechanisms of protein thermostability evolved in the thermophiles’ inhospitable world.

Protein Folding in Crowded Environments

Most of the conventional models of protein folding deal with a single polypeptide chain, sometimes with an explicit consideration of the surrounding water molecules. However, as it is now understood, these conditions are seldom realized in vivo, as the volume fraction of macromolecules inside a living cell can reach 0.2. Thus, collisions and interactions between proteins (all kinds – folded, misfolded, and nascent chains) are the norm, rather than exception. Previous research, both experimental and computation, has shown that putting mechanical constraints on a folding protein may significantly alter its folding kinetics and maybe even the stability of the native state. Also, avoidance of promiscuous interactions in a dense protein environment might have imposed specific evolutionary constraints on protein sequences.

One of the research projects in the lab includes analytical calculations and computer modeling of protein folding in crowded environments and looking for the possible “crowding-mitigating” signatures in real proteins.