Non-ribosomal peptide synthetase (NRPS) is a microbial systemic enzyme that produces secondary metabolites such as lipopeptide antibiotics (Shokrollahi et al., 2021). In general, NRPS enzymes consist of multiple modules, with each module capable of accommodating several domains, including the condensation domain, adenylation (A) domain, thiolation (T) domain, epimerization (E) domain, and thioester (TE) domain (Marahiel, 2016; Izoré et al., 2021). In certain literature, the T domain is alternatively referred to as the peptidyl carrier protein (PCP) domain (Rüschenbaum et al., 2022). Additionally, it has been demonstrated that the C domains contain two catalytic tunnels that connect the donor-PCP and acceptor-PCP domain-binding sites to the active site. These tunnels serve as the pathway for the donor and acceptor substrates to access the active site (Samel et al., 2007).

The structure of a three-module NRPS contains an initiation module (core A-T domain), an elongation module (core C-A-T domain) that can be followed by an E domain, and a termination module (core C-A-T-TE domain) (Calcott and Ackerley, 2015; Izoré et al., 2021). The bio-synthesis process of peptide moiety can be divided by these domains into several steps, beginning with post-translational modification of each T domain by attachment of a 4′-phosphopantetheine (4′-pp) cofactor. The 4′-pp group serves as a flexible arm that facilitates the coordination of substrate movement between catalytic sites (Kittilä et al., 2016; Zhang et al., 2022). Within each module, the A domain plays a crucial role in recognizing a specific monomer and activating it as an aminoacyl adenylate. Subsequently, it attaches the activated monomer to the 4′-pp group located on the adjacent T domain (Schwarzer et al., 2002; Zhu et al., 2019). In a sequential fashion, beginning with the first module, each C domain facilitates the formation of peptide bonds by linking the donor substrate, which is attached to the upstream T domain, with the acceptor substrate, which is attached to the downstream T domain (Wheadon and Townsend, 2021). Before the condensation reaction takes place, it is possible that each T domain may require interaction with additional tailoring domains, such as the E domain. The E domain plays a role in modifying the substrate carried by each module, including the ability to catalyze racemization (Kim et al., 2022). Once the peptide chain reaches the termination module, the product is released by a thioesterase (TE) domain through hydrolysis or intramolecular cyclization. Following this, the synthesis cycle can be repeated to generate multiple copies of the same peptide (Marahiel, 2016; Hühner et al., 2019).

Fusaricidin is an important antibiotic family produced by Paenibacillus polymyxa with great potential use in medical and agricultural applications (Kajimura and Kaneda, 1996; Lee et al., 2012; Haron et al., 2019; Jeong et al., 2019). So far, 14 fusaricidin members have been reported, consisting of an invariable 15-guanidino-3-hydroxypentadecanoic acid and a variable cyclic hexapeptide (Deng et al., 2011; Vater et al., 2015; Reimann et al., 2017). The variable cyclic hexadepsipeptide moiety of fusaricidin usually contains six amino acid residues: L-Thr¹, X², X³, D-allo-Thr⁴, X⁵, and D-Ala⁶ (Han et al., 2012). Among the family, fusaricidin LI-F07a has better antimicrobial activity than other fusaricidin analogs. Based on the current knowledge, L-Thr¹, D-Val², L-Phe³, D-allo-Thr⁴, D-Asn⁵, and D-Ala⁶ constitute the cyclic hexadepsipeptide moiety within LI-F07a (Reimann et al., 2017).

The discovery of penicillin in 1928 started the golden age of natural product antibiotic discovery that peaked in the mid-1950s (Hutchings et al., 2019). Since then, a gradual decline in antibiotic discovery and development and the evolution of antibiotic resistance in many human pathogens has led to the current antimicrobial resistance crisis (Larsson and Flach, 2022). We believe that the future of antibiotic findings looks bright as new technologies such as genome mining and editing are deployed to discover new antibiotics with diverse bioactivities (Albarano et al., 2020).

In our previous studies, we reported the strain WLY78 produces fusaricidin, and the fermented strain could be used for bio-controlling the Fusarium head blight in wheat (Li and Chen, 2019; Li et al., 2021). While acknowledging the effectiveness of this biocontrol strain, we are not content with maintaining the present status. Our primary motivation is to proactively prevent the potential emergence of microbial resistance issues, which serves as a driving force for our continuous advancements. Our objective is to optimize the NRPS enzymes involved in the synthesis of existing antibiotics strategically. This optimization aims to enhance the effectiveness of antibiotics. In our current study, we have created a new fusaricidin antibiotic by modifying the FusA. Briefly, we analyzed the function of FusA and confirmed it could biosynthesizes fusaricidin LI-F07a. Then, each module within FusA was reorganized by gene modification. Moreover, we found that one of the engineered strains produced a novel fusaricidin derivative. The structure of this novel compound was identified as [ΔAla⁶] fusaricidin LI-F07a. Meaningfully, this new compound exhibits stronger activity than fusaricidin LI-F07a in inhibiting the pathogen fungi F. asiaticum. Finally, this novel substance was successfully applied to prevent plants from being infected by pathogens: Fusarium oxysporum and Botrytis cinerea.

The Paenibacillus species is a powerful bio-fungicide and has many potential applications. So far, 14 fusaricidin analog members have been reported to be produced by P. polymyxa (Kuroda et al., 2001; Choi et al., 2008). In the present study, by synthetase engineering, we created a biosynthetic pathway for the production of a novel fusaricidin derivative, which could serve as an antibiotic replacement solution for preventing pathogen fungi infection in agriculture.

In the past, to elucidate the structure of a compound produced by a microorganism required fermentation, purification, and then identification. Now, the genome mining method has rapidly paved the way for predicting secondary metabolites within microorganisms. Similarly, we used the genome mining strategy to predict the genome and found the FusA coding gene, suggesting that FusA producing fusaricidin LI-F07a is a great possibility (Figure 1A). The A domain within NRPS is reported to be essential in recognizing the substrate that consists of lipopeptide in a 1:1 manner (Marahiel et al., 1997; Stachelhaus et al., 1999). According to the principle above, a lot of known A domains were aligned with our A domains within FusA, and then, we acquired the substrate of our A domains, which matches the structure of peptide moiety within fusaricidin LI-F07a (Figure 1B). Furthermore, we also noticed the “ten code residues” located at positions 235, 236, 239, 278, 299, 301, 322, 330, 331, and 517 within the A domain. It has been previously reported that these residues serve as the binding site for the A domain when interacting with the substrate in the three-dimensional structure (Conti et al., 1997; Stachelhaus et al., 1999). Since the “ten code residues” are extracted from the A domains, the alignment of either the “A domain” or the “ten code residues” supports the same conclusion (Figure 1C).

Mass spectrometry fragment ion analysis is a reliable detection method and particularly powerful for the structural characterization of analogous compounds (Kurusu et al., 1987; Han et al., 2012). This method has successfully characterized more than 10 fusaricidin members (Vater et al., 2015). With a similar method, we demonstrated that our strain could produce fusaricidin LI-F07a by LC–MS–MS fragment ion analysis (Figure 2). All of the fusaricidin members are lipid-modified non-ribosomal cyclic hexadepsipeptides containing four D-amino acids and two L-amino acids. This particular ω-functionalized lipid side chain is of key importance for the antibiotic activity of the fusaricidins and their selective inhibition of fungal cells due to the interaction with phospholipid cell membranes (Bionda et al., 2012). However, these fusaricidin members with different activity share the same lipid side chain except the cyclic hexadepsipeptide (Kurusu et al., 1987). How much the cyclic hexadepsipeptide of fusaricidin contributes to its antifungal activity still needs to be explored. Therefore, we intend to modify the cyclic hexadepsipeptide of fusaricidin LI-F07a to see whether its antifungal activity is affected.

In past decades, a rational design of novel peptide antibiotics method has been developed by modifying one domain and domain fusion within NRPS in Bacillus sp. and Paenibacillus sp. (Stachelhaus et al., 1995; Mootz et al., 2000; Yakimov et al., 2000; Han et al., 2012). However, only modifying one domain may cause a decrease in substrate-recognition ability due to the incompatibility of other domains in the same module. This is because, in addition to the A domain in NRPS, other domains such as C, T, and E also exhibit specific substrate recognition. If we could knock out an entire module from the NRPS without causing frameshift mutations, some instances might result in the production of a peptide that is missing a corresponding amino acid.

Therefore, we did not replace any A domains but directly deleted each of modules that consist of the A domain and its neighboring domains (Figure 3A). Excluding the pleiotropic effect of a mutation in prokaryotes is generally considered easier compared to eukaryotic cells. In prokaryotes, such as our strain, the genome is typically continuous without introns. Introns are non-coding regions in the genome of eukaryotes that can contribute to the pleiotropic effects observed during gene knockout by introducing complexity in gene expression regulation and splicing. However, in prokaryotes, the absence of introns means that gene knockout does not involve the splicing or regulatory complexities associated with introns (Setubal et al., 2018). Moreover, frameshift mutations are one of the main factors that can contribute to the pleiotropic effects of gene knockout. When designing primers for homologous recombination double exchange, we have meticulously considered and designed the primers to effectively prevent any subsequent frameshift mutations resulting from the knockout. Even if the subsequent primary amino acid sequence changes due to a frameshift mutation, a frameshift protein may still retain its structure and functionality. In fact, deleterious frameshift mutations have been proposed to be potential drivers for molecular evolution. Based on this shiftability, many genes and certain genomes are naturally optimized for frameshift tolerance (Wang et al., 2022). Importantly, we ensured the absence of frameshift mutations by conducting thorough sequencing analysis. We successfully acquired a strain with engineered ΔM6 that showed a more enhanced antifungal activity than the wild-type strain with FusA (Figure 3B). However, a former study had created a mutant with inactivated 4′-phosphopantetheinyl transferase, which could disrupt the production of NRPS/NPKS-dependent metabolites such as fusaricidins and still kept its antagonistic activity against Fusarium culmorum. They emphasized that the biofilm matrix formation may be of major importance in its antagonism (Timmusk et al., 2019). There could potentially exist an antagonistic relationship between fusaricidins and exopolysaccharides production. It has been demonstrated that fusaricidin inhibits Bacillus subtilis by specifically targeting the mechanism of exopolysaccharide production (Yu et al., 2012). Additionally, we observed a similar phenomenon where the capacity for biofilm synthesis was enhanced in ΔM2, ΔM5, and ΔM6 mutants. This enhancement may be associated with increased polysaccharides production. However, except for ΔM6, none of these strains produced antibacterial effects in our study. Therefore, we suppose that the enhanced effect of this engineered strain should be attributed to the new compound [ΔAla⁶] fusaricidin LI-F07a (Figure 4C). Indeed, both in vitro and in vivo, [ΔAla⁶] fusaricidin LI-F07a did show stronger ability against F. oxysporum and B. cinerea than fusaricidin LI-F07a (Figure 5).

Until the completion of our study, it remained inconclusive whether the fungicidal activity of our mutants was affected by small peptides or the failure of peptide chains to cyclize due to the early peptide chain breakage caused by knocking out. Usually, the production of fusaricidin LI-F07a ranges from 12 mg/L to 76 mg/L (Raza et al., 2010; Han et al., 2012). In our wild-type strain, the maximum yield of fusaricidin LI-F07a is ~60 mg/L. In the engineered strain M6, the production of [ΔAla⁶] fusaricidin LI-F07a is ~55 mg/L. Based on the comparison of the yield data before and after the genetic modification, it can be observed that the synthesis efficiency of the derivative is reduced by approximately 10 percent. The engineered ΔM6 shows promising potential as a sustainable choice that can meet diverse industrial and agricultural demands. [ΔAla⁶] fusaricidin LI-F07a can be applied effectively to prevent plants from being infected by Fusarium species pathogens.

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding authors.

YL performed the research and drafted the manuscript. SC made crucial revisions of the manuscript. All authors contributed to the article and approved the submitted version.

The funders of this work are Chengdu NewSun Crop Science Company Limited (No. SWHCRR001) and the National Key Research and Development Program of China Award (No. 2017YFD0200807).

YL was employed by Chengdu NewSun Crop Science Co. Ltd.

The authors declare that this study received funding from Chengdu NewSun Crop Science Co. Ltd. This funder was engaged in the purifications, the greenhouse trails and the writing of this article.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.