Building Next-Generation Technologies for Low-Cost Gene Synthesis and High-Accuracy Genome Engineering

Abstract

The technologies that enable writing and editing of DNA form the foundation of modern molecular biology and biotechnology. However, a number of methodological barriers have limited the widespread adoption of both high-throughput de novo gene synthesis and large-scale genome alteration. Increasingly, work in the fields of synthetic biology, protein design, and gene therapy has been hindered by shortcomings in current DNA writing and editing technologies. The goal of this dissertation has been to improve both the throughput of chemical gene synthesis and the accuracy of genome editing tools.

The scale of gene synthesis is most acutely limited by the high cost of using column-synthesized oligonucleotides as the base material. It has been clear for some time that using the much cheaper microarray-synthesized DNA instead would significantly decrease costs of making long, double-stranded DNA. Unfortunately, microarrays’ high chemical complexity, high error rates, and low synthesis yield has prevented their adoption into gene synthesis workflows. We have used selective nucleotide pool amplification and enzymatic error removal to develop a synthesis pipeline that uses Agilent’s Oligonucleotide Library Synthesis microarrays to build 500-850 base pair-long double-stranded constructs. At the time of initial publication the size of the total pool (13,000 oligonucleotides encoded ~2.5 megabases of DNA) was at least one order of magnitude larger than previously reported attempts.

Following our initial progress on increasing the throughput of gene synthesis, it became apparent that using applying synthetic DNA in vivo is bottlenecked by the low efficiencies and unpredictable accuracies of existing genome engineering techniques. In an effort to build tools that can be used in a large variety of organisms we focused our efforts on engineering site-specific recombinases, many of which can function without using host-encoded proteins. Unfortunately, many groups have reported that site-specific recombinases can cause toxicity possibly due to off-target binding and recombination activities. To address this problem, we proposed that the accuracy of DNA-binding proteins can be altered through mutations in the of protein-protein interaction domains. To test this idea we obtained single-residue mutants of Cre recombinases that exhibited improved site discrimination in in vivo and in vitro recombination experiments. We have also been interested in developing rapid and reproducible assay of protein binding assays. Towards this goal, we conducted proof-of-concepts experiments that demonstrated that gel shift assays could be used to generate binding curves in a multiplexed fashion. We propose that the slope of the information content as a function of binding affinity can be used to compare binding accuracy of dissimilar proteins.