Table 5 Summary enzyme engineering strategies discussed
Enzyme Engineering Strategy | ||||
|---|---|---|---|---|
Rational Design | Computational Design | Semi-Rational Design | Random Mutagenesis | |
Strengths | • Precisely targets protein residues/regions that are predicted to be most impactful • Small libraries need only simple, low-effort screening methods | • Can rapidly screen mutants in silico prior to any laborious experimentation • Can potentially create de novo function with limited previous data | • Precisely targets protein residues/regions that are predicted to be most impactful • Tests varied chemistry at positions • Screening can be done without ultra-high-throughput methods | • Can probe entire sequence space, finding rare, surprising mutations • Requires limited structural and evolutionary information • Efficiently navigates good vs. bad mutation combinations (epistatic effects) |
Limitations | • Requires knowledge of enzyme structure and/or homologous enzymes and mutations • Can only probe limited sequence space: no surprising mutations | • Requires experimental validation: predictions may be inaccurate • Requires reliable models • Screening large, combinatorial mutation spaces may be resource-intensive | • Requires knowledge of enzyme structure and/or homologous enzymes and mutations • Limited to mutation at the positions selected • Requires a high-throughput screen | • Requires a high-throughput screen to find improved variants • Mechanistic insight into mutations may not be readily apparent |
Contexts | • Can complement other strategies, especially for first-generation engineering of newly discovered enzyme scaffolds • Many mutatons are currently known that improve key functions (e.g., Tm) | • Can complement other strategies, e.g., using experimental datasets to train better models • Models and methods can be generalized to many different scaffolds | • Can complement other strategies, e.g., by targeting hotspots found through random mutagenesis • Screens exist for its successful execution | • Can complement other strategies, e.g, by gathering large-scale sequence-function data for AI/ML models • Screens exist for its successful execution |
