Using molecular modeling to engineer proteins with novel functions

Access full-text files




Codrea, Vlad

Journal Title

Journal ISSN

Volume Title



The RCSB Protein Data Bank currently stores over 30,000 X-ray crystal structures, information that has proven invaluable in various studies that have been undertaken to analyze these proteins. My goal is to apply molecular modeling techniques in order to add new functionality to enzymes whose 3D structures are available. The standard processes for modifying the functions of proteins have traditionally remained centered on experimental work, with a heavy reliance on directed evolution. One such approach has been used by Professor Peter Schultz from the Scripps Research Institute to introduce the unnatural amino acid 3-(2-naphthyl)-L-alanine into the amber stop codon of E. coli. This approach uses positive and negative selection to wean the cells into taking up 3-(2-naphthyl)-L-alanine. In contrast, I have used a computational method to predict mutations in aminoacyl-tRNA synthetases, the enzymes responsible for joining amino acids with their respective tRNAs, that will change the affinities of those proteins. My goal is to arrive at aminoacyl-tRNA synthetases that are capable of incorporating analogs of arginine, cysteine, phenylalanine, and tryptophan instead of the natural residues. As a follow-up, I have prepared a mutant tyrosyl-tRNA synthetase and tRNA to test how well an analog of tyrosine is incorporated into proteins. I have used an in-silico protein modeling framework developed by Professor Homme Hellinga from Duke University to tackle another outstanding problem in molecular biology: how to recreate the high affinity of the streptavidin/biotin complex. The specificity of streptavidin is changed so that it binds preferentially to biotin derivatives. Streptavidin is a tetrameric protein, similar to the avidin protein found in egg whites, but made by the Streptomyces avidinii bacteria. Streptavidin binds tightly to the vitamin D-biotin, and forms one of the strongest naturally-occurring non-covalent interactions between a protein and an organic ligand, with a dissociation constant (Kd) on the magnitude of 10-15 M. Because of this strong binding, the streptavidin/biotin combination has been used extensively in molecular and bioengineering studies that require the joining of different molecules that would not normally come together. One of the current limitations with using this combination is that it is only possible to specifically bring together two compounds (namely the compound attached to biotin and the compound attached to streptavidin) at any one step of an assay. The aim is to engineer orthologous pairs of mutant streptavidins and biotin analogs, each of which can be covalently attached to a distinct molecular payload depending on the end-user’s intended application. The members of one orthologous pair will not cross-react with the complementary member of another orthologous pair – in other words, a mutant streptavidin should have a relatively poor binding affinity to biotin. This provides an element of selectivity to parallel reactions performed in the same environment.



LCSH Subject Headings