1 INTRODUCTION
Directed evolution has proved to be a powerful tool to improve the activity, stability and substrate specificity of enzymes (Goldsmith et al. 2017; Romero et al. 2009; Roodveldt et al. 2005; Wong et al. 2004; Wong et al. 2007; Zeymer et al. 2018). This tool involves two crucial steps: generation of a library containing sufficient gene variants and high-throughput screening of library members with desired properties (Packer et al. 2015; Wong et al. 2006). Because of its high transformation efficiencies, rapid growth rates, well-established manipulation approaches (Crameri et al. 1996), Escherichia coliis generally employed as host organism for library creation. However, proteins are usually expressed in cytoplasm in E. coli , making the screening of library more difficult if the substrate of the protein cannot be transported into cell. Therefore, the organisms such asBacillus subtilis and Pichia pastoris that can secret proteins into medium have been developed as alternative hosts for library generation (Jiang et al. 2019; Liu et al. 2017; Reetz et al. 2007; Wang et al. 2012).
B. subtilis is an important industrial host for production of various recombinant proteins due to its GRAS (generally recognized as safe) status, excellent protein secretion ability and mature fermentation processes (Commichau et al. 2014; Marcus et al. 2004; Schumann et al. 2007; Terpe et al. 2006; Yang et al. 2011). AlthoughB. subtilis has well-developed genetic manipulation tools, its transformation frequency is still much lower than that of E. coli . Thus, the library of random mutants generated through digestion and ligation is too small to fulfill the need of direct evolution inB. subtilis . Several strategies have been adopted to address this limitation. A routine strategy is first constructing the library of variants in E. coli and then transferring the library intoB. subtilis , which is time-consuming and of low efficiency (Caspers et al. 2010; You et al. 1996; Zhao et al. 1999). Melnikov et al. (1999) attempted to clone error-prone PCR (epPCR) product via marker-replacement recombination with a structurally similar helper plasmid resident in the transformation recipient. But they found that it was difficult to recover >103transformants/μg of epPCR product. Given the phenomenon that multimeric plasmid has much higher (approximately three orders of magnitude) transformation frequency than that of monomeric plasmid in B. subtilis (Canosi et al. 1978), Shafikhani et al. (1997) generated large libraries of random mutants (3×106) in B. subtilis by PCR-based plasmid multimerization method. In this method, epPCR product was fused with linearized vector through PCR extension to generate linear plasmid multimer which could be converted to circular form through homologous recombination after entering cell; however, each cell may take several variants, decreasing screening efficiency. Increasing the number of competent cells is an effective method to create larger library. To this end, Zhang and Zhang et al. (2011) constructed a B. subtilis strain SCK6 whose competence can be induced by artificially controlling the expression of master regulator ComK. The procedure for preparing supercompetent cell of strain SCK6 was further improved by Li et al. (2017). The transformation frequencies of multimeric plasmid, monomeric plasmid and integration plasmid into strain SCK6 could reach 107, 104 and 105 transformants/μg DNA.
This study aims to construct large mutant library through inserting epPCR product into the chromosome of B. subtilis , which will solve the problems of plasmid instability and heterozygosity faced by multi-copy plasmid mediated strategies (Melnikov et al. 1999; You et al. 2012; Zhang and Zhang 2011). We accomplished the library construction within one day and the generated library had sufficient mutants (>105) for the need of direct evolution.