|
Structural and Chemical Biology @ Vanderbilt University |
back to research projects
Computational Design of Novel Protein Therapeutics
Members: Andrew Morin, Nicole Shen
link to poster pdf
Vancomycin is a small-molecule beta-lactam glycopeptide antibiotic which binds and sequesters the free -D-Alanine-D-Alanine C-terminus of a key gram-positive bacterial cell wall component, thereby inhibiting proper cell wall biosynthesis and consequently rendering the bacteria vulnerable to osmotic lysis. Although vancomycin is often considered an “antibiotic of last resort”, whose use is highly controlled, bacterial resistance to vancomycin and other beta-lactam antibiotics has already become widespread. While a small number of next-generation antibiotics capable of treating these resistant strains are either currently available or in the development pipeline, the pace of new therapeutic development over the last several decades, and into the foreseeable future, cannot not keeping pace with the rate of emergence of resistant pathogens.
The most common mechanism of acquired resistance observed in pathogenic bacterial strains is through the substitution of a -D-Lactate in place of the -D-Ala at the free C-terminus of the bacterial peptide. This single replacement of the C-terminal amino linkage by an ester linkage of the lactate, destroying a single hydrogen bond between vancomycin and the peptide, is enough to destroy binding of vancomycin to its bacterial peptide ligand and render the bacteria resistant.
The objective of my dissertation research is to develop and validate computational methods for designing novel protein therapeutics. Because the molecular and structural bases for both vancomycin binding and resistance have previously been well classified, the re-design of a protein binding pocket to bind the resistant D-Ala-D-Lac peptide motif seems an ideal proof-of-concept experiment.
The experimental process will be divided into three distinct but overlapping phases.
Phase one is the in-silico re-design of the ligand binding pocket on a panel of four selected, naturally occurring protein scaffolds to bind the resistant D-Ala-D-Lac bacterial peptide. These four protein scaffolds, each of which has a high resolution structure available, were chosen based on properties related to stability, size and conformation of the binding pocket, anticipated immunogenicity and ease of genetic manipulation. These four scaffold proteins include thermophilic TIM-barrel, beta-propeller and twisted beta-sheet folds. Atomic level models of each of the scaffold proteins then underwent successive rounds of in-silico ligand-docking and side-chain design using the Rosetta program. Rosetta uses a combination of monte-carlo sampling and database scoring methods to direct design of output structures. Each round of computational dock/design sampled several hundred million structures that were then filtered using calculated ligand binding energies. Output structures were further filtered using custom developed computer programs on both quantitative and qualitative measures of binding. Approximately 10,000 structures from each round were then carried over into the next iterative round of dock/design, and the process repeated. In each of five iterative rounds, calculated ligand binding energies improved along with the design of the peptide binding site.
Phase two is the experimental validation of the in-silico designs. Approximately one dozen of the best protein designs from the computational rounds, with representatives from each protein scaffold, are currently being finalized for protein expression. Expression vectors will be created from amino acid sequences output by Rosetta and optimized for cloning into vectors using both multiple mutation reaction and gene assembly methods. Current computational results indicate that as many as 14 mutations form the wild-type scaffold protein may be necessary for some designs. Histidine tag sequences are to be added to each protein to facilitate purification by Ni+ affinity chromatography from an E.coli expression cell line. Assays for binding affinities of the designed proteins for both the wild-type and resistant bacterial peptides will be evaluated. Binding assay techniques under consideration include isothermal titration calorimetry (ITC), surface plasmon resonance spectroscopy (SPS) and polarization fluorescence anisotropy.
Phase three is an anticipated refinement and re-evaluation round to address any flaws or deficiencies revealed by the experimental phase, possibly followed by further experimental validation and development.
Obtaining experimental evidence of high affinity binding of the vancomycin resistant bacterial peptide by one or more of our re-designed proteins would potentially elevate our designs to drug candidate status and be a significant advance in the field of computational protein design.
While the development of a replacement for vancomycin to address the growing problem of bacterial resistance is an important and worthy goal in its own right, the underlying objective of my work is to develop a broadly applicable and repeatable computational method for design and re-design of protein binding sights as related to therapeutic action. Designing a replacement for vancomycin is a good first step in achieving that aim. However, we anticipate that the techniques and methods developed in the pursuit of a vancomycin replacement will greatly enhance our ability to apply this process to other problems in protein therapeutic design.
Computational Design of Novel Protein Therapeutics
Members: Andrew Morin, Nicole Shen
link to poster pdf
Vancomycin is a small-molecule beta-lactam glycopeptide antibiotic which binds and sequesters the free -D-Alanine-D-Alanine C-terminus of a key gram-positive bacterial cell wall component, thereby inhibiting proper cell wall biosynthesis and consequently rendering the bacteria vulnerable to osmotic lysis. Although vancomycin is often considered an “antibiotic of last resort”, whose use is highly controlled, bacterial resistance to vancomycin and other beta-lactam antibiotics has already become widespread. While a small number of next-generation antibiotics capable of treating these resistant strains are either currently available or in the development pipeline, the pace of new therapeutic development over the last several decades, and into the foreseeable future, cannot not keeping pace with the rate of emergence of resistant pathogens.
The most common mechanism of acquired resistance observed in pathogenic bacterial strains is through the substitution of a -D-Lactate in place of the -D-Ala at the free C-terminus of the bacterial peptide. This single replacement of the C-terminal amino linkage by an ester linkage of the lactate, destroying a single hydrogen bond between vancomycin and the peptide, is enough to destroy binding of vancomycin to its bacterial peptide ligand and render the bacteria resistant.
The objective of my dissertation research is to develop and validate computational methods for designing novel protein therapeutics. Because the molecular and structural bases for both vancomycin binding and resistance have previously been well classified, the re-design of a protein binding pocket to bind the resistant D-Ala-D-Lac peptide motif seems an ideal proof-of-concept experiment.
The experimental process will be divided into three distinct but overlapping phases.
Phase one is the in-silico re-design of the ligand binding pocket on a panel of four selected, naturally occurring protein scaffolds to bind the resistant D-Ala-D-Lac bacterial peptide. These four protein scaffolds, each of which has a high resolution structure available, were chosen based on properties related to stability, size and conformation of the binding pocket, anticipated immunogenicity and ease of genetic manipulation. These four scaffold proteins include thermophilic TIM-barrel, beta-propeller and twisted beta-sheet folds. Atomic level models of each of the scaffold proteins then underwent successive rounds of in-silico ligand-docking and side-chain design using the Rosetta program. Rosetta uses a combination of monte-carlo sampling and database scoring methods to direct design of output structures. Each round of computational dock/design sampled several hundred million structures that were then filtered using calculated ligand binding energies. Output structures were further filtered using custom developed computer programs on both quantitative and qualitative measures of binding. Approximately 10,000 structures from each round were then carried over into the next iterative round of dock/design, and the process repeated. In each of five iterative rounds, calculated ligand binding energies improved along with the design of the peptide binding site.
Phase two is the experimental validation of the in-silico designs. Approximately one dozen of the best protein designs from the computational rounds, with representatives from each protein scaffold, are currently being finalized for protein expression. Expression vectors will be created from amino acid sequences output by Rosetta and optimized for cloning into vectors using both multiple mutation reaction and gene assembly methods. Current computational results indicate that as many as 14 mutations form the wild-type scaffold protein may be necessary for some designs. Histidine tag sequences are to be added to each protein to facilitate purification by Ni+ affinity chromatography from an E.coli expression cell line. Assays for binding affinities of the designed proteins for both the wild-type and resistant bacterial peptides will be evaluated. Binding assay techniques under consideration include isothermal titration calorimetry (ITC), surface plasmon resonance spectroscopy (SPS) and polarization fluorescence anisotropy.
Phase three is an anticipated refinement and re-evaluation round to address any flaws or deficiencies revealed by the experimental phase, possibly followed by further experimental validation and development.
Obtaining experimental evidence of high affinity binding of the vancomycin resistant bacterial peptide by one or more of our re-designed proteins would potentially elevate our designs to drug candidate status and be a significant advance in the field of computational protein design.
While the development of a replacement for vancomycin to address the growing problem of bacterial resistance is an important and worthy goal in its own right, the underlying objective of my work is to develop a broadly applicable and repeatable computational method for design and re-design of protein binding sights as related to therapeutic action. Designing a replacement for vancomycin is a good first step in achieving that aim. However, we anticipate that the techniques and methods developed in the pursuit of a vancomycin replacement will greatly enhance our ability to apply this process to other problems in protein therapeutic design.