Protein structure and design by genetic experiments

Abstract

Gene sequences refined by natural selection contain evidence of the interactions essential for function. By modeling natural sequences, we can now solve protein and RNA 3D folds and design new molecules and systems. Now we are also beginning to create genetic data in the lab faster than nature does, and we can explore genetic sequences and phenotypes nature hasn’t. How can laboratory selection be leveraged to discover underlying biology and design new genes? Here we show that the 3D folds of proteins and RNAs can be solved ab initio using deep mutation scans. We also design small, targeted mutation scans to solve protein structures efficiently. Second, we solve the structure of human vitamin K epoxide reductase with natural sequences and use mutation scans to dissect the mechanisms of drug-interfering mutants. Lastly, we determine mechanisms of thioesterase specificity and design enhanced enzymes for production of fuels and plastics using laboratory recombination. As DNA technology continues to accelerate, we propose such methods will in turn accelerate basic biology and biological design.