Tailoring enzyme catalysts by directed evolution

Citation

Moore, Jeffrey Charles (1996) Tailoring enzyme catalysts by directed evolution. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/szy1-1e40. https://resolver.caltech.edu/CaltechETD:etd-12222007-113626

Abstract

Directed evolution provides a nature-inspired methodology for the improvement of macromolecular properties. By imitating processes attributed to natural evolution, we have been able to improve an enzyme's activity against a desired substrate. Four rounds of random mutagenesis and screening have led to a p-nitrobenzyl (pNB) esterase with 16-fold improvement in catalytic turnover, 25-fold improvement in catalytic efficiency, and 30-fold improvement in total activity. An extensive library was screened during the fourth generation of random mutagenesis and screening which generated several clones more active than the third generation parent. These clones were recombined in a fifth generation using two techniques similar to that of recombination. A partial sampling of the resulting population contained individuals demonstrating activity two to four-fold higher than the most active variant from the fourth generation, making the total activity improvements greater than 100-fold. The mutations found to increase pNB esterase's activity in this directed evolution experiment were mapped onto a structural model. None of the effective amino acid substitutions lie in segments of the enzyme predicted to interact directly with the bound substrate. To predict in advance that these substitutions enhance pNB esterase activity would be practically impossible.

Sexual recombination provides a mechanism of information-sharing in nature, and an approximation of this method is available as DNA shuffling. The approximation was examined and compared to the natural method in terms of screening optimization. The natural method of pairwise sexual recombination appears optimal when additivity of mutational effects is assumed. In the absence of this assumption, the DNA shuffling procedure can be used to generate all the combinations of mutations present in a parent pool at the expense of requiring the screening of large numbers of clones. A compromise strategy is suggested which removes much of the additivity assumption and requires screening of substantially reduced numbers of clones.

The correlation between various enzyme properties were examined. Enzyme activities in varying concentrations of organic solvent are highly correlated, while no correlation was observed between enzyme activity and enzyme stability in oxidative environments. For the enzyme evolved, variants with increased stability were far more frequent than those with increased activity.

Thesis Files