Machine learning-guided evolution of pyrrolysyl-tRNA synthetase for improved incorporation efficiency of diverse noncanonical amino acids
Qunfeng Zhang, Ling Jiang, Yadan Niu, Yujie Li, Wanyi Chen, Jingxi Cheng, Haote Ding, Binbin Chen, Ke Liu, Jiawen Cao, Junli Wang, Shanli Ye, Lirong Yang, Jianping Wu, Gang Xu, Jianping Lin, Haoran Yu
Abstract
The pyrrolysyl-tRNA synthetase (PylRS) is widely used to incorporate noncanonical amino acids (ncAAs) into proteins. However, the yields of most ncAA-containing protein remain low due to the limited activity of PylRS variants. Here, we apply machine learning to engineer the tRNA-binding domain of PylRS. The FFT-PLSR model is first applied to explore pairwise combinations of 12 single mutations, generating a variant Com1-IFRS with an 11-fold increase in stop codon suppression (SCS) efficiency. Deep learning models ESM-1v, Mutcompute, and ProRefiner are then used to identify additional mutation sites. Applying FFT-PLSR on these sites yields a variant Com2-IFRS showing a 30.8-fold increase in SCS efficiency, and up to 7.8-fold improvement in the catalytic efficiency (kcat/KmtRNA). Transplanting these mutations into 7 PylRS-derived synthetases significantly improves the yields of proteins containing 6 types of ncAAs. This paper presents improved PylRS variants and a machine learning framework for optimizing the enzyme activity. The low activity of PylRS hinders the efficient production of ncAA containing proteins. Here, authors apply machine learning to engineer the tRNA-binding domain of PylRS, obtaining variants with improved stop codon suppression efficiency and achieving higher yields of proteins containing various ncAAs.