With the increasingly serious drug resistance of Acinetobacter baumannii, there is an increasingly urgent need for new antibacterial drugs. Phage lysin PlyAB1 has a bactericidal effect on drug‐resistant A. baumannii, which has the potential to replace antibiotics to fight infection caused by A. baumannii. However, its application is limited by its thermal stability and lytic activity. To solve these problems, molecular dynamics (MD) simulations combined with Hotspot wizard 3.0 were used to identify key residue sites affecting thermal stability, and evolutionary analysis combined with multiple sequence alignment was used to identify key residue sites affecting lytic activity. Four single‐point variants with significantly increased thermal stability and four single‐point variants with significantly lytic activity were obtained, respectively. Furthermore, by superimposing mutations, we obtained three double‐point variants, G100Q/K69R, G100R/K69R, and G100K/K69R, with significantly improved thermal stability and improved lytic activity. At 45°C, the lytic activity and half‐life of the optimal variant G100Q/K69R were 1.51‐ and 24‐fold higher than those of the wild PlyAB1, respectively. These results deepen our understanding of the structure and function of phage lysin and contribute to the application of phage lysin in antibiotic substitution.

With the continuous emergence and spread of drugresistant and multi-drug-resistant Staphylococcus aureus, traditional antibiotic treatment has gradually lost its effect. There is an urgent need to develop and study new and effective bio-green inhibitors to control S. aureus. In this study, the S. aureus phage DZ25 was isolated from milk and the lysin LysDZ25 with excellent tolerance to serum and NaCl solution was identified. Subsequently, to improve the lytic activity and thermal stability of LysDZ25, RoseTTAFold was used to construct threedimensional (3D) structures, molecular dynamics (MD) simulation was used for conformational acquisition, and the MDL strategy previously developed in our lab was used to rationally design variants. After two rounds of rational design, the optimal variant with improved thermal stability, S333V/N245R/D299L, was obtained, and its half-life time was 4.0-fold that of wild-type LysDZ25. At 37, 40, 45, and 50 °C, the lytic activity of the optimal triple-point variant S333V/N245R/ D299L was increased by 17.3-, 26.7-, 20.2-, and 50.1-fold compared with that of the wild-type LysDZ25, respectively. Finally, cell count was used to evaluate the lytic activity, and the results showed that the optimal variant S333V/N245R/D299L could drop about 3.5 log 10 values compared with the control and about 2.6 log 10 values compared with the wild-type LysDZ25.