As alternatives to chemical pesticides, biological agents for controlling

Transcription

1 Nonribosomal Peptide Synthase Gene Clusters for Lipopeptide Biosynthesis in Bacillus subtilis 916 and Their Phenotypic Functions Chuping Luo, a,b Xuehui Liu, c Huafei Zhou, a,b Xiaoyu Wang, a Zhiyi Chen a,b College of Plant Protection, Nanjing Agriculture University, Nanjing, China a ; Institute of Plant Protection, Jiangsu Academy of Agricultural Sciences, Nanjing, China b ; Institute of Biophysics, Chinese Academy of Sciences, Beijing, China c Bacillus cyclic lipopeptides (LPs) have been well studied for their phytopathogen-antagonistic activities. Recently, research has shown that these LPs also contribute to the phenotypic features of Bacillus strains, such as hemolytic activity, swarming motility, biofilm formation, and colony morphology. Bacillus subtilis 916 not only coproduces the three families of well-known LPs, i.e., surfactins, bacillomycin Ls (iturin family), and fengycins, but also produces a new family of LP called locillomycins. The genome of B. subtilis 916 contains four nonribosomal peptide synthase (NRPS) gene clusters, srf, bmy, fen, and loc, which are responsible for the biosynthesis of surfactins, bacillomycin Ls, fengycins, and locillomycins, respectively. By studying B. subtilis 916 mutants lacking production of one, two, or three LPs, we attempted to unveil the connections between LPs and phenotypic features. We demonstrated that bacillomycin Ls and fengycins contribute mainly to antifungal activity. Although surfactins have weak antifungal activity in vitro, the strain mutated in srfaa had significantly decreased antifungal activity. This may be due to the impaired productions of fengycins and bacillomycin Ls. We also found that the disruption of any LP gene cluster other than fen resulted in a change in colony morphology. While surfactins and bacillomycin Ls play very important roles in hemolytic activity, swarming motility, and biofilm formation, the fengycins and locillomycins had little influence on these phenotypic features. In conclusion, B. subtilis 916 coproduces four families of LPs which contribute to the phenotypic features of B. subtilis 916 in an intricate way. As alternatives to chemical pesticides, biological agents for controlling crop diseases have drawn more and more attention from microbiologists and plant pathologists. Due to the production of abundant secondary metabolites with a broad spectrum of antimicrobial activities, many Bacillus strains are widely used in the biocontrol of crop diseases (1 5). One extensively used and intensively studied microorganism, Bacillus subtilis, has the potential to produce antimicrobial compounds. Among these compounds, cyclic lipopeptides (LPs) generated by nonribosomal peptide synthases (NRPSs) have the well-recognized potential to be used in biotechnology and biopharmaceutical applications because of their antimicrobial and surfactant properties (6, 7). Three families of Bacillus cyclic LPs surfactins, iturins, and fengycins are well described with regard to most of the mechanisms that account for the biocontrol effect of different Bacillus strains to date (5, 8, 9). Surfactins, iturins, and fengycins are synthesized by large multienzyme complexes and show striking product microheterogeneity due to variations in both length and branching of fatty acid chains and in amino acids of peptide sequences (6, 7). All the biochemical and genetic results showed that these three families of LPs are synthesized by modular organization of NRPSs and obey the linear rule. It is not difficult to make novel LP analogues with certain peptide substitutions by manipulation of existing LP synthetases, e.g., via subunit exchange, module exchange, and combinatorial biosynthesis (10, 11). The prospect of creating numerous bioactive LPs by genetic engineering has stimulated the search for new lipopeptide synthetases from Bacillus strains. With the development of genome mining strategies, a large number of new natural products with good antimicrobial activities were identified from traditional biocontrol strains (12, 13). Therefore, the opportunities to find novel LPs and synthetases will be largely enhanced with the number of Bacillus strain genomes sequenced recently. B. subtilis 916 was isolated from paddy soil in Jurong County, Jiangsu Province, China, in The genome sequence of B. subtilis 916 was recently analyzed, and four NRPS gene clusters were identified (14). In addition to the three conventional gene clusters, srf, bmy, and fen, B. subtilis 916 also contains a new gene cluster, loc, which is responsible for the biosynthesis of a new family of LP called locillomycin. In comparison to other Bacillus strains, it is unusual for B. subtilis 916 to have the potential to coproduce four families of LPs, since the other Bacillus strains with good control efficiency usually coproduce 2 or 3 families of LPs (6, 7, 15). Recent research showed that the LPs were also involved in multicellular behaviors in terms of swarming motility, biofilm formation, and colony morphology (3, 5, 6, 16 19). B. subtilis strains can use swimming- and sliding-type mechanisms to swarm on surfaces. Swarming motility enables cells to spread on plant surfaces and is an important component of the initial development of microbial biofilms (17, 20). Swarming motility and biofilm forma- Received 6 September 2014 Accepted 22 October 2014 Accepted manuscript posted online 31 October 2014 Editor: R. M. Kelly Address correspondence to Chuping Luo, luochuping@163.com, or Zhiyi Chen, chzy @163.com. Citation Luo C, Liu X, Zhou H, Wang X, Chen Z Nonribosomal peptide synthase gene clusters for lipopeptide biosynthesis in Bacillus subtilis 916 and their phenotypic functions. Appl Environ Microbiol 81: doi: /aem Supplemental material for this article may be found at /AEM Copyright 2015, American Society for Microbiology. All Rights Reserved. doi: /aem aem.asm.org Applied and Environmental Microbiology January 2015 Volume 81 Number 1

2 LP Biosynthesis and Phenotypic Features in B. subtilis TABLE 1 B. subtilis strains and plasmids used Strain or Source or plasmid Description a reference B. subtilis strains 916 Producer of surfactin, fengycin, bacillomycin L, and locillomycin CGMCC no BSBM bmyd::nm r ; BGG105 Laboratory stock BSFM fena::cm r This study BSLM locd::em r This study BSSM srfa::spec r This study BSBFM bmyd::nm r fena::cm r This study BSBLM bmyd::nm r locd::em r This study BSFLM fena::cm r locd::em r This study BSBFLM bmyd::nm r fena::cm r locd::em r This study Plasmids puc19 Cloning vector; Ap r Laboratory stock pbest501 pgem4 carrying the PrepU promoter BGSC and neo gene from pub110, Nm r pbac105 puc19 carrying the PrepU promoter Laboratory stock and the 750-bp and 800-bp fragments from the bmyd operon psg1164 Integrated vector; Amp r Cm r BGSC psgfen psg1164 carrying a 729-bp fragment This study of fena pmutin4 Integrated vector; Amp r Em r BGSC pmutinloc pmutin4 carrying a 812-bp This study fragment of locd pdg1728 Integrated vector; Amp r SpeC r BGSC pucsc puc19 carrying the spectinomycin This study cassette from pdg1728 pucscsrf pucsc carrying a 908-bp fragment of srfaa This study a Ap r, ampicillin resistance; Em r, erythromycin resistance; Cm r, chloramphenicol resistance; Spec r, spectinomycin resistance. tion are major adaptive colonization strategies of Bacillus strains in new environments (17, 21). While many research results suggested that surfactins played a major role in influencing the multicellular behaviors of B. subtilis (3, 5, 22), results concerning the functions of the other LPs remain inconclusive. Here, we report the identification, molecular characterization, and functional domain analysis of the four NRPS gene clusters in the genome of B. subtilis 916. We also report the construction of a series of single, double, and triple mutants which are disrupted in different NRPS gene clusters. We demonstrated that the four NRPS gene clusters are responsible for the productions of the four families of LPs surfactins, bacillomycin Ls, fengycins, and locillomycins. Function investigation of the four families of LPs showed that they are involved in the phenotypic features of B. subtilis 916 in an intricate way. MATERIALS AND METHODS Bacterial strains, plasmids, and media. The strains and plasmids used in this study are listed in Table 1. Luria-Bertani (LB) broth medium (10 g/liter tryptone, 5 g/liter yeast extract, 5 g/liter NaCl) was used for cultivation of Escherichia coli and B. subtilis. When necessary, antibiotics were added at the following concentrations: ampicillin, 100 g/ml; spectinomycin, 100 g/ml; erythromycin, 1 g/ml; neomycin, 20 g/ml; chloramphenicol, 5 g/ml. MSgg medium was used for biofilm formation and colony morphology and contained 100 mm morpholinepropanesulfonic acid (MOPS) (ph 7), 50 g/ml tryptophan, 50 g/ml phenylalanine, 2 mm MgCl 2, 700 M CaCl 2,50 M FeCl 3,50 M MnCl 2,2 M thiamine, 1 M ZnCl 2. The fungal strains were cultured at 28 C on potato dextrose agar (PDA) medium containing (per liter) 200 g of potato infusion, 20 g of glucose, and 20 g of agar (ph 7.0). Genome sequence, DNA analysis, and domain structure. The draft genome of B. subtilis 916 was sequenced with Illumina/Solexa HiSeq 2000 as reported previously (14). On the basis of the draft genome, the complete genome of B. subtilis 916 was sequenced using the Pacbio sequencing platform. The genome bioinformatics analysis was performed by Shanghai Hanyubiotech company. DNA analyses and translation were conducted with the Vector NTI 10 and DANMAN software packages. BLAST with annotated domains of similar nonribosomal peptide synthetases (NRPSs) was used to detect conserved active-site motifs. The amino acid sequences of discrete Bacillus adenylation domains, thiolation or peptidyl carrier domains, and condensation domains were extracted from modular Bacillus NPRSs and were used for BLASTP comparison in order to detect their closest orthologs. Construction of mutants deficient in four families of LPs synthesis. B. subtilis 916 mutants were generated according to a modified protocol originally developed for B. subtilis 168 (23). The bmyd gene was disrupted by insertion of a neomycin cassette via double-crossover homologous recombination, as reported previously (24), and the mutant BSBM was selected. Disruption of the fena gene was achieved by insertion of a chloramphenicol cassette. A PCR product of 729 bp, obtained with the primers FenAF (5=-TTTCTCGAGGTCTTGATGGTGCAGTCAGA-3=) and FenAR (5=-TTTGAATTCCTGGACCTGTTTGTCTTTGT-3=) (XhoI and EcoRI restriction sites are underlined), was cloned into psg1164, resulting in psgfen. psgfen was transformed into B. subtilis 916, and the mutant BSFM was selected. The locd gene was disrupted by insertion mutagenesis with an erythromycin cassette derived from pmutin4. A PCR product of 812 bp, obtained with the primers LocDF (5=-TTTAAGCTTTCAGGTAC CAACGATGAACA-3=) and LocDR (5=-TTTGGATCCTTGTCCATTAC AGCTACGGT-3=) (HindIII and BamHI restriction sites are underlined), was cloned into pmutin4, resulting in pmutinloc. pmutinloc was transformed into B. subtilis 916, and the mutant BSLM was selected. The srfaa gene was disrupted by insertion mutagenesis with a spectinomycin cassette derived from pdg1728 and puc19. A PCR product of 1,182 bp containing the spectinomycin resistance gene and promoter was obtained with the primers SpecF (5=-TTTGGATCCCTGCAGCCCTGGC GAATG-3=) and SpecR (5=-TTTGAATTCAGATCCCCCTATGCAAGG- 3=) (BamHI and EcoRI restriction sites are underlined) from pdg1728. After digestion with BamHI and EcoRI, the spectinomycin cassette was cloned into puc19, resulting in pucsc. A 908-bp PCR product from the srfaa gene region was amplified with the primers SrfA-AF (5=-TTTAAG CTTACACAGATATCAGGCAAGC-3=) and SrfA-AR (5=-TTTGGATCC GTCCCATCGTTCCTTCACA-3=) (HindIII and BamHI restriction sites are underlined) and inserted into pucsc, yielding the vector pucscsrf. pucscsrf was transformed into B. subtilis 916, and the mutant BSSM was selected. To obtain double and triple mutants, the vectors constructed above were transformed into competent B. subtilis 916 cells one by one, and the mutants BSBFM ( bmyd::nm r fena::cm r ), BSBLM ( bmyd:: Nm r loca::em r ), BSFLM ( fena::cm r loca::em r ), and BSBFLM ( bmyd::nm r fena::cm r loca::em r ) were selected. All the single, double, and triple disruptions of genes described above were demonstrated in the resistant colonies by PCR with appropriate primers and by Southern hybridization (data not shown). Isolation and purification of LPs and HPLC-MS analysis. Surfactins, bacillomycin Ls, and locillomycins were isolated by adding concentrated hydrochloric acid to the culture broth of BSFM after the biomass was removed by centrifugation. Precipitates were formed at ph 2.0 which could be collected, dried, and extracted with methanol (MeOH). The solvents were removed under reduced pressure and white solids were collected after the solvents were removed. The white solids were dissolved in methanol followed by charcoal treatment and passed through a m-pore-size filter. The MeOH extractions were added to a 3-fold vol- January 2015 Volume 81 Number 1 Applied and Environmental Microbiology aem.asm.org 423

3 Luo et al. ume of H 2 O and titrated to ph 7.0 with 5 M NaOH. The extracted impurities were added to an Agilent amino solid-phase extraction column and washed with 50% (vol/vol) MeOH-H 2 O, 100% MeOH, 1% (vol/vol) formic acid-meoh, and 2% (vol/vol) formic acid-meoh step by step. The 1% (vol/vol) formic acid-meoh elutions were concentrated by nitrogen drying to partial purification of surfactin and bacillomycin. The 2% (vol/ vol) formic acid-meoh elutions were concentrated by nitrogen drying to partial purification of locillomycin. All the partially purified samples were then loaded on an Agilent C 18 solid-phase extraction column separately and washed with 40%/50% (vol/vol) MeOH-H 2 O for locillomycin, with 50%/70% (vol/vol) MeOH-H 2 O for bacillomycin L, and with 75%/90% (vol/vol) MeOH-H 2 O for surfactin. The 50%, 70%, and 90% (vol/vol) MeOH-H 2 O elution products were concentrated by nitrogen drying to purify locillomycins, bacillomycin Ls, and surfactins, respectively. Fengycin purification was the same as surfactin purification except that it started with BSSM culture broth. The extraction and purification products were analyzed by high-performance liquid chromatography (HPLC) with an acetonitrile-water-trifluoroacetic acid solvent system (40:60:0.5 [vol/vol/ vol] for bacillomycin Ls and locillomycins, 50:50:0.5 [vol/vol/vol for fengycins, and 80:20:0.5 [vol/vol/vol] for surfactins) with a 0.5-ml/min flow rate. Except for the locillomycins, which were monitored at 230 nm, the LPs (i.e., the surfactins, fengycins, and bacillomycins) were monitored at 210 nm. All the samples were further analyzed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MS) with an Agilent 1100 series HPLC-MS/MS system. When necessary, tandem MS spectrometry was also performed, in which a precursor ion was mass selected and the parent ion was divided into the daughter ions to give structurally significant product ions using an Agilent 1100 series LC-MS/MS system (Agilent Technologies). Assay of the antifungal activities against Fusarium oxysporum. The activities of the wild-type and mutant strains against spore germination of F. oxysporum were assessed as follows. Spores of F. oxysporum were diluted to 10 6 ml 1 and inoculated by using 500 l to flood LB medium plates. The excess liquid was removed, and the plates were allowed to dry under a laminar flow hood for 30 min. Portions (50 l) of culture strains were deposited in 7-mm-diameter wells created in the solidified medium using sterile glass tubes. The plates were inoculated at 28 C, and inhibition zones were measured after 1 to 3 days. To further determine the antifungal activity of the four purified LPs, five doses of each family of LPs were added separately to three aliquots each containing 25 ml potato dextrose agar at 45 C, mixed rapidly, and poured into three separate plates. After the agar had cooled, mycelial plugs were added to the plates in equal amounts. Samples consisting of buffer only served as controls, and all the plates were inoculated at 28 C. When the mycelial colony of the control had grown to almost fill the plate, the area of the colony was measured, and the inhibition of fungal growth on the other plates was determined by calculating the percent reduction of the area of the mycelial colony. Evaluation of hemolytic activities. To evaluate the hemolytic activities of wild-type B. subtilis 916 and the mutants, strains were inoculated onto the blood agar plates with 5% defibrinated sheep blood by streaking. Hemolytic activities were visualized by development of a clear halo around the growth of the strains after incubation at 37 C for 1 to 4 days. In all cases, three replicate plates were used for each strain, and the experiment was repeated once. Evaluation of colony architecture, swarming motility, and biofilm formation. For colony architecture, 1 l of starting culture was spotted onto the surface of an MSgg agar plate containing 20 g/ml Congo red and 10 g/ml Coomassie brilliant blue and incubated at 30 C for 48 h, as previously described (25). To evaluate swarming motility, 1 l of an overnight culture was used to inoculate the center of LB plates containing 20 g/ml Congo red and 10 g/ml Coomassie brilliant blue solidified with 0.7% agar. The plates were incubated at 37 C and were evaluated for colony spread over time. For pellicle (floating biofilm) formation analysis, each strain was grown in 4 ml of LB at 37 C overnight, and 100 l of this starting culture was used to inoculate 4 ml of MSgg containing 20 g/ml TABLE 2 Calculated mass values of M, M H, and M Na ions corresponding to identified isoforms of surfactins, bacillomycin Ls, locillomycins, and fengycins in culture extracts from B. subtilis 916 Mass value a Lipopeptide M M H M Na Surfactin A (C 13 ) 1, ,008.6* 1,030.6 Surfactin B (C 14 ) 1, ,022.7* 1,044.7* Surfactin C (C 15 ) 1, ,036.7* 1,058.7* Bacillomycin LA (C 14 ) 1, ,021.5* 1,043.5* Bacillomycin LB (C 15 ) 1, ,035.5* 1,057.5* Bacillomycin LC (C 16 ) 1, ,049.5* 1,071.5 Locillomycin A (C 13 ) 1, ,146.6* 1,168.6 Locillomycin B (C 14 ) 1, ,160.6* 1,182.6* Locillomycin C (C 15 ) 1, ,174.6* 1,196.6 Fengycin A (C15/Ala-6) 1, ,449.8* 1,471.9 Fengycin B (C 16 /Ala-6) 1, ,463.9* 1,485.9 Fengycin C (C 17 /Ala-6) 1, ,477.9* 1,499.9 Fengycin D (C 16 /Val-6) 1, ,491.9* 1,513.9 Fengycin E (C 17 /Val-6) 1, ,505.9* 1,527.9 a The data were compiled from whole cells grown on LB agar. Peaks presented in Figure S1 in the supplemental material are indicated with asterisks. Congo red and 10 g/ml Coomassie brilliant blue in 12-well microtiter plates, which was then incubated without agitation at 37 C for 24 h. Nucleotide sequence accession numbers. The draft genome sequence and complete genome sequence of B. subtilis 916 and the new gene cluster for biosynthesis of locillomycins described here have been submitted to GenBank under accession numbers AFSU , CP009611, and KF866134, respectively. RESULTS Identification and quantification analysis of four families of LPs produced by B. subtilis 916. The LPs of B. subtilis 916 were investigated by HPLC-MS. Four groups of mass peaks were detected (see Fig. S1 in the supplemental material), and their mass numbers are summarized in Table 2. Families 1, 2, and 4 of the LPs were identified as surfactins, bacillomycin Ls, and fengycins (Fig. 1) by comparing their mass data with those previously obtained by MS analysis of the LPs of numerous Bacillus strains. The novel family 3 of the LPs were identified as locillomycins by evaluation of the fragment spectra obtained from electrospray ionization (ESI)- MS/MS and nuclear magetic resonance (NMR) spectra (C. Luo, Z. Chen, J. Y. Guo, X. Liu, X. Wang, Y. Liu, and Y. Liu, U.S. patent application 14/190,817; C. Luo and Z. Chen, unpublished data). B. subtilis 916 produces C 13 to C 15 surfactins, C 14 to C 16 bacillomycin Ls, and C 15 to C 17 fengycins. This pattern of LPs corresponds to the metabolite spectra found for most surfactin-, bacillomycin-, and fengycin-producing Bacillus strains. Unexpectedly, B. subtilis 916 also produces a novel family of LPs called locillomycins, which are unique nonapeptides with fatty acid side chains of 13 to 15 carbon atoms (Fig. 1). Each purified LP was further analyzed by HPLC-MS, and different peaks representing different isomers of LPs were obtained (see Fig. S5 to S8 in the supplemental material). The different peaks produced from each preparation were used to quantify the four families of LPs produced by B. subtilis 916 in the agitated Erlenmeyer flasks for 3 days. The production of surfactins was detected at the first8hofgrowth and revealed the early synthesis 424 aem.asm.org Applied and Environmental Microbiology January 2015 Volume 81 Number 1

4 LP Biosynthesis and Phenotypic Features in B. subtilis FIG 1 Structures of four families of LPs surfactins, bacillomycin Ls, locillomycins, and fengycins. Amino acid residues are in the three-letter code. The L or D configuration is indicated with a subscript before the amino acid residue designation. The fatty acid moieties are shaded, and the number of carbon atoms is given. The surfactin family has a polar molecule structure, which is characterized by a core cyclic peptide with 7 amino acid residues and an exocyclic acyl group consisting of a -hydroxyl fatty acid moiety with 13 to 15 carbon atoms. The bacillomycin L family is characterized by a chemical structure comprising of a core cyclic peptide with 7 amino acid residues and an exocyclic acyl group consisting of a -amino fatty acid moiety with 14 to 17 carbon atoms. The novel locillomycin family is characterized by a chemical structure comprising a core cyclic peptide with 9 amino acid residues and a -hydroxyl fatty acid moiety with 13 to 15 carbon atoms. The fengycin family is characterized by a chemical structure comprising a core cyclic peptide with 10 amino acid residues and an exocyclic acyl group consisting of either a saturated or unsaturated -hydroxyl fatty acid moiety with 16 to 17 carbon atoms. of surfactins by B. subtilis 916. In contrast to surfactins, other three LPs were not observed during the first 12 h of growth, which was expected since the biosynthesis of these LPs is known to occur only after the exponential growth phase. The highest rates of production of surfactins and bacillomycin Ls were 22.8 mg/liter and 19.7 mg/liter, respectively, at 36 h of growth. However, the highest levels of production of fengycins and locillomycins were only 4.8 mg/liter and 3.6 mg/liter, respectively (Table 3). Organization of four nonribosomal peptide synthetase (NRPS) gene clusters in the B. subtilis 916 genome. The B. subtilis 916 genome contains four nonribosomal gene clusters, srf, fen, bmy, and loc, which are responsible for the synthesis of three wellknown LPs (surfactin, fengycin, and bacillomycin L) and the new locillomycin family, respectively (see Fig. S2 in the supplemental material). While the gene clusters srf, fen, and bmy in B. subtilis 916 locus are closely related to the corresponding gene clusters srf, fen, and bam in Bacillus amyloliquefaciens FZB42, respectively, the loc gene cluster appears to be a single-site insertion relative to B. amyloliquefaciens FZB42. Like the fen and bmy gene clusters, srf and loc are also close to each other on the chromosome of B. subtilis 916. In addition, the locus of the bmy gene cluster in B. subtilis 916 is the same as that of the iturin A gene cluster in B. subtilis RB14 (26). The fen locus in B. subtilis 916 is related to pps in B. subtilis 168, fen in B. subtilis F29-3, and myc in B. subtilis ATCC 6633 (15). The fact that the gene clusters fen, pps, and myc occupy essentially the same locus on the genomic DNA suggests that different NRPS gene clusters could be integrated into these loci as an insertion or as a replacement of existing NRPS gene clusters. Interestingly, although the genes of the loc cluster have no homologs in the B. amyloliquefaciens FZB42 genome, the nucleotide sequences outside loc are approximately 98% identical and contain two highly conserved genes, glms and ybbr. Near the insertion site of loc in B. subtilis 168 there is a skf gene cluster which is responsible for the biosynthesis of sporulation killing factor, and B. subtilis ATCC 6633 contains a rhi gene cluster which is responsible for the biosynthesis of an antifungal phosphonate oligopeptide of rhizocticin (27). The fact that the loc, skf, and rhi gene clusters occupy the same locus on the genomic DNA may be due to the highly conserved genes glms and ybbr, which can be easily swapped among different Bacillus strains through homologous recombination. In general, the locillomycin, sporulation killing factor, and rhizocticin in B. subtilis 916, B. subtilis 168, and B. subtilis ATCC 6633 have been found in identical loci, and it is further suggested that either the NRPS or polyketide synthase (PKS) gene cluster is also interchangeable among different B. subtilis strains. Schematic representation and functional domain analysis of the four nonribosomal peptide synthetases (NRPSs). Genetic and biochemical analyses have revealed that the arrangement of the modules of most LP synthetases is linear with amino acid sequences of LPs (6, 7, 28). As expected, the modular organization of NRPSs involved in biosynthesis of surfactins, bacillomycin Ls, and fengycins in B. subtilis 916 is similar to that of their counterparts in other Bacillus strains, and all the NRPSs obey the linear rule (Fig. 2; also, see Fig. S11 and Table S1 in the supplemental material). That is to say that the order and specificity of the modules within proteins encoded by srf, bmy, and fen in B. subtilis 916 determine the amino acid sequence of the surfactins, bacillomycin Ls, and fengycins, respectively (see Fig. S3 and Table S1 in the supplemental material). In particular, the bmy gene cluster responsible for bacillomycin L biosynthesis in B. subtilis 916 is very similar to the bam gene cluster responsible for bacillomycin D biosynthesis in B. amyloliquefaciens FZB42. Although the amino acid sequences encoded by bmy share high similarity ( 97%) with their counterparts encoded by bam, a low-similarity ( 60%) region in bmy and bam was found. Further detailed bioinformation analysis showed that this region encoded adenylation domains which were responsible for activation of Ser-4 for bacillomycin L TABLE 3 Production of four families of LPs surfactins, bacillomycin Ls, locillomycins, and fengycins by B. subtilis 916 grown in LB medium a LP production (mg liter 1 ) LPs 8 h 12 h 24 h 36 h 72 h 96 h Surfactins 5.4 (0.24) 12.9 (2.1) 19.1 (2.5) 22.8 (3.0) 21.2 (2.9) 22.6 (3.1) Bacillomycin Ls (2.4) 19.7 (2.6) 16.3 (1.8) 15.5 (1.6) Locillomycins (0.3) 3.6 (0.5) 3.4 (0.5) 3.2 (0.4) Fengycins (0.3) 4.8 (0.6) 4.2 (0.5) 4.4 (0.7) a The values are means from three experiments, and the values in parentheses are standard deviations. January 2015 Volume 81 Number 1 Applied and Environmental Microbiology aem.asm.org 425

5 Luo et al. and Pro-4 for bacillomycin D (see Fig. S4 in the supplemental material). To our surprise, the biosynthesis of locillomycins by the loc gene cluster deviates from the linear mechanism and exhibits a nonlinear assembly (29, 30). The loc gene cluster encodes 4 proteins (LocD, LocA, LocB, and LocC) which constructed a hexamodular NRPS, but they biosynthesize cyclic nonapeptides, the locillomycins. Based on further analysis of the organization of the function modules in loc and the structure of locillomycins, we propose that the biosynthesis of locillomycins exhibits a rare nonlinear mechanism, with the middle three domains in LocB being used iteratively and a KAS domain FIG 3 HPLC spectrograms of four families of LPs produced by B. subtilis 916 and its mutants. (a) HPLC spectrograms of fengycins. BSFM, which is disrupted in fena, was deficient in production of fengycins, and BSSM, which is disrupted in srfaa, had a significant decrease in the production of fengycins. (b) HPLC spectrograms for bacillomycin Ls. BSBM, which is disrupted in bmyd, and BSSM, which is disrupted in srfaa, were deficient in production of bacillomycin Ls. (c) HPLC spectrograms for surfactins. BSSM is deficient in production of surfactins. (d) HPLC spectrograms for locillomycins. BSLM, which is disrupted in locd, was deficient in production of locillomycins. 426 aem.asm.org Applied and Environmental Microbiology January 2015 Volume 81 Number 1 FIG 2 Gene clusters of surfactin, locillomycin, bacillomycin L, and fengycin. The schematic representation of the entire gene cluster for srf, bmy, loc, and fen comprises the ORFs and the domains corresponding to NRPSs and amino acids incorporated by the different modules, which encode catalytic machineries responsible for the biosynthesis of surfactins, bacillomycin Ls, locillomycins, and fengycins. The order and specificity of the modules within NRPSs encoded by srf, bmy, and fen determine the amino acid sequence of the surfactins, bacillomycin Ls, and fengycins, respectively. However, the hexamodular NRPSs encoded by loc carry out biosynthesis of locillomycin nonapeptides.

6 LP Biosynthesis and Phenotypic Features in B. subtilis in LocD being skipped (this observation will be detailed elsewhere). Consecutive disruption of bmy, fen, srf, and loc gene clusters yields series of LP-deficient phenotypes. To confirm that the srf, bmy, fen, and loc gene clusters are responsible for surfactin, bacillomycin L, fengycin, and locillomycin biosynthesis, we disrupted srfaa, bamd, fena, and locd in the B. subtilis 916 genome with homologous recombination. Double and triple mutant strains were also obtained. Analysis of the mutant strains by HPLC-MS confirmed that BSBM ( bac::nm r ), BSFM ( fena::cm r ), and BSLM ( loca::em r ) were deficient in bacillomycin L, fengycin, and locillomycin production, respectively (Fig. 3). Unexpectedly, BSSM ( srfa::spec r ) was deficient in producing not only surfactins but also bacillomycin Ls, and significantly decreased production of fengycins was also observed (Fig. 3; also, see Fig. S9 to S12 and Table S2 in the supplemental material). In contrast to BSSM, while the mutant BSBM lacked bacillomycin L biosynthesis, it produced 43.4% more surfactins and 25.6% more fengycins than wild-type B. subtilis 916 (see Table S2 and Fig. S10 and S12 in the supplemental material). Like BSBM, while BSFM lacked fengycin biosynthesis, it produced 18.5% more surfactins than the wild type (see Table S2 and Fig. S12 in the supplemental material). As expected, the double and triple mutant strains failed to produce two and three LPs, respectively. The production of the four LPs by the double and triple mutant strains was also assessed by HPLC- MS. The results showed that while the mutant strains were deficient in producing some of the LPs, the production of the others could be enhanced significantly (see Fig. S9 to S12 and Table S2 in the supplemental material). Evaluation of antifungal activities and hemolytic activities of wild-type and mutant strains. In this study, the antifungal activities of mutant strains against spore germination and mycelium growth of F. oxysporum were also investigated (Fig. 4; also, see Fig. S13 in the supplemental material). Compared to the wild type, BSBM and BSFM had significantly decreased antifungal activity. In contrast to the single mutants BSBM and BSFM, the double mutant BSBFM had further-decreased antifungal activity. As expected, the single mutant BSSM was unable to inhibit the growth F. oxysporum in a manner similar to that of BSBFM. Compared to the wild type, BSLM had no significant change in its antifungal activity. For the triple mutant BSBFLM, which retained the ability to produce surfactins, the ability to suppress the growth of F. oxysporum was completely abolished. The toxicities of the four LPs for F. oxysporum were also investigated in in vitro assays. F. oxysporum showed high sensitivity to the bacillomycin Ls and fengycins, and the 50% inhibitory concentrations (IC 50 s) for hyphal growth were below 2 g/ml. F. oxysporum showed moderate sensitivity to the locillomycins, and the IC 50 for hyphal growth was 18.8 g/ml. Further, F. oxysporum showed the lowest sensitivity to surfactins, and the IC 50 for hyphal growth was over 50.0 g/ml. The results above strongly suggest that the bacillomycin Ls and fengycins contribute mainly to the antifungal activity of B. subtilis 916 against F. oxysporum. The hemolytic activities of B. subtilis 916 and the mutants were also compared (Fig. 5). While the wild type, BSFM, and BSLM were able to induce a clear halo surrounding the streak-inoculated region on blood agar plates at 24 h of growth, BSBM was not able to formed a clear halo until 48 h of growth. In particular, BSSM was unable to induce a hemolytic halo around the inoculated region before 36 h of growth and formed only a vague halo at 72 h of FIG 4 Antifungal activities of B. subtilis 916 and its mutants against F. oxysporum. (a) A volume of 50 l culture broth of B. subtilis 916 and its mutant strains was dropped into agar plates which contain spores of F. oxysporum. The plates were incubated for 1 to 3 days at 28 C. Two inhibition zones were formed. The inner inhibition zones are marked by red circles, and the outer inhibition zones are marked by green circles. WT, wild-type B. subtilis 916; BSBM, single mutant disrupted in bmyd; BSFM, single mutant disrupted in fena; BSLM, single mutant disrupted in locd; BSSM, single mutant disrupted in srfaa; BSBFM, double mutant disrupted in both bmyd and fena; BSFLM, double mutant disrupted both in fena and locd; BSBLM, double mutant disrupted in both bmyd and locd; BSBFLM, triple mutant disrupted in bmyd, fena, and locd. (b) Measurement the inhibition zones of B. subtilis 916 and its mutant strains against F. oxysporum. Whereas bacillomycin L contributed mainly to the inner inhibition zones, the fengycins contributed conclusively to the outer inhibition zones. Data are average diameters standard deviations for three replicates in two independent experiments. Means in the same column with different letters are significantly different (P 0.05) according to Duncan s multiple range tests. growth. The double and triple mutant strains have less hemolytic activity than their parent strains. The hemolytic activities of the four purified families of LPs were also investigated. As expect, the surfactins and bacillomycin Ls showed strong hemolytic activities, but the fengycins and locillomycins showed weak hemolytic activities. Thus, we draw the conclusion that the bacillomycin Ls and surfactins play much more important roles in the hemolytic activities of B. subtilis 916 than the locillomycins and fengycins do. Evaluation of swarming motility, biofilm formation, and colony morphology. B. subtilis 916 formed colonies with dense wrinkles and compact structure on MSgg plates, showed good swarming motility on semisolid LB plates, and formed a thick and wrinkled floating biofilm in MSgg broth (Fig. 6, 7, and 8). The January 2015 Volume 81 Number 1 Applied and Environmental Microbiology aem.asm.org 427

7 Luo et al. mutant BSSM, like other strains deficient in production of surfactins, formed flat colonies, had significantly less swarming motility, and formed only a very thin and fragile floating biofilm (3, 5). Interestingly, the mutant BSBM also showed changes in colony morphology and decreased floating biofilm formation, but its swarming motility was unexpectedly significantly enhanced. The mutant BSLM showed no difference in swarming motility and biofilm formation, and the colony morphology of BSLM was significantly different from that of the wild type, similar to that of FIG 6 Comparison of colony morphology of B. subtilis 916 and its mutant strains on LB plates with Congo red and Coomassie brilliant blue dyes. The wild-type (WT) strain B. subtilis 916 and single mutant BSBFM disrupted in fena formed the highly structured colonies. The single mutants BSBM, BSLM, and BSSM disrupted in bmyd, locd, and srfaa, respectively, formed less structured and flatter colonies than WT. Furthermore, the double mutants BSBFM, BSFLM, and BSBLM and triple mutant BSBFLM formed flatter colonies than their parent strains. Unlike the disruption of fena (encoding fengycins), the disruptions of srfaa, bmyd, and locd all had an important influence on the colony morphology of B. subtilis 916 and its mutant strains. 428 aem.asm.org BSBM (Fig. 6 to 8). However, the mutant BSFM showed no distinctive change in any of the three phenotypical features. Restoration of the multicellular behavior of mutant strains by adding exogenous LPs was also investigated (Fig. 6 to 8). The addition of exogenous LPs failed to restore colony morphology in any of the mutant strains. However, the swarming motility of FIG 7 Floating-biofilm formation of the B. subtilis 916 and its mutant strains with or without adding exogenous surfactins and bacillomycin Ls in MSgg broth with Congo red and Coomassie brilliant blue dyes. WT B. subtilis 916 formed a thick and wrinkled floating biofilm. Like the WT, the mutants BSFM (disrupted in fena) and BSLM (disrupted in locd) also formed thick and wrinkled floating biofilms. However, biofilm formation was weakened significantly in mutant BSBM (disrupted in bmyd), which formed a thin and fragile floating biofilm. Furthermore, BSSM (disrupted in srfaa) formed a thinner and more fragile floating biofilm than BSBM. The floating biofilm was restored significantly in BSBM with either bacillomycin Ls (BSBM BL) or surfactins (BSBM SRF) at 20 g/ml. However, the floating biofilm was restored in the mutant BSSM only faintly with bacillomycin Ls (BSSM BL) and surfactins (BSSM SRF). Applied and Environmental Microbiology January 2015 Volume 81 Number 1 FIG 5 Hemolytic activities of wild-type B. subtilis 916 and its mutant strains. B. subtilis 916 and its mutant strains were inoculated on blood agar plates by streaking. Hemolytic activities were visualized by development of a clear halo around the growth of the strains after incubation at 37 C for 1 to 3 days. WT, wild-type B. subtilis 916, which coproduces four LPs; BSBM, single mutant deficient in production of bacillomycin Ls; BSFM, single mutant deficient in productions of fengycins; BSLM, single mutant deficient in the production of locillomycins; BSSM, single mutant deficient in production of both surfactins and bacillomycin Ls and decreased in production of fengycins; BSBFM, double mutant deficient in production of both bacillomycin Ls and fengycins; BSFLM, double mutant deficient in production of both fengycins and locillomycins; BSBLM, double mutant deficient in production of both bacillomycin Ls and locillomycins; BSBFLM, triple mutant deficient in production of bacillomycin Ls, fengycins, and locillomycins. Surfactins and bacillomycin Ls produced by B. subtilis 916 and its mutant strains contributed mainly to their hemolytic activities.

8 LP Biosynthesis and Phenotypic Features in B. subtilis FIG 8 Swarming motility of the B. subtilis 916 and its mutants with or without the addition of exogenous surfactins and bacillomycin Ls, as assessed in 0.7% agar LB plates with Congo red. The WT strain B. subtilis 916 showed good swarming motility on semisolid LB plates. Swarming motility of BSFM and BSLM, which are disrupted in fena and locd, respectively, showed no difference from that of the WT. The swarming motility of BSSM disrupted in srfaa decreased significantly compared to that of the WT. The swarming motility of BSBM increased significantly compared to that of the WT. The swarming motility of BSBM was further enhanced by adding exogenous bacillomycin Ls at 20 g/ml (BSBM BL). As expected, swarming motility was restored in BSSM upon addition of exogenous surfactins at 20 g/ml (BSSM SRF). To our surprise, swarming motility of BSBM decreased significantly upon addition of exogenous surfactins at 20 g/ml (BSBM SRF). Similarly, swarming motility of BSSM further decreased upon addition of exogenous bacillomycin Ls (BSSM BL). BSSM and BSBM was restored or even enhanced upon addition of exogenous surfactins and bacillomycin Ls, respectively. The swarming motility of the double mutant BSBSM was also restored significantly by the addition of both surfactins and bacillomycin Ls. To our surprise, unlike other mutant strains disrupted in srfaa reported previously, which were able to restore biofilm formation, the mutant BSSM was unable to restore biofilm formation when surfactins and bacillomycin Ls were added. Interestingly, BSBM was able to restore the formation of floating biofilm upon the addition of either surfactins or bacillomycin Ls. DISCUSSION The endospore-forming rhizobacterium B. subtilis and closely related species are well-known candidates for developing efficient biopesticide products mainly due to their production of two dozen antibiotics with an amazing variety of structures, enabling them to cope with competing organisms within food crop tissues (4, 6, 7, 31). Usually, B. subtilis has an average of 4 to 5% of its genome devoted to antibiotic synthesis (6). Recently, the complete genome of B. subtilis 916 was sequenced (14). The most distinctive feature of this genome is that a considerable part ( 10%, 400 kb, organized in 9 operons) is devoted to the biosynthesis of polyketide and peptide antibiotics. Five gene clusters encoding PKSs and four gene clusters involved in NRPS biosynthesis of four families of LPs have been identified. In contrast to other biocontrol B. subtilis and B. amyloliquefaciens strains, whose genomes contain only 2 or 3 NRPS gene clusters, the genome of B. subtilis 916 contains four family NRPS gene clusters. In addition, B. subtilis 916 is naturally competent for uptake of naked DNA and homologous recombination and ideally suitable for genetic manipulation. Thus, B. subtilis 916 provides an excellent model for research on the biosynthesis and functions of LPs. The disruption of bmy, fen, and loc genes blocked production of the corresponding LPs, suggesting that these gene clusters are responsible for their biosynthesis. To our surprise, unlike the other B. subtilis strains disrupted in srfaa, which had only the production of surfactins blocked, a B. subtilis 916 mutant with a disruption of srfaa had the production of bacillomycin Ls as well as surfactins blocked and had impaired production of fengycins (3, 5, 9). In contrast to the disruption of srf, the disruption of bmy enhanced the production of the other LPs significantly. Further investigations are necessary to clarify the entangled biosynthesis of bacillomycin Ls and fengycins. The double and triple mutant strains further confirmed that the four NRPS gene clusters are responsible for the four families of LPs in B. subtilis 916 and that they interact with and affect one another. Intriguingly, in contrast to the other Bacillus LPs reported previously, locillomycins showed a novel molecular architecture (6, 7). While the length of the fatty acid chains of the locillomycins is similar to that of other Bacillus LPs, the amino acid moiety of locillomycins is different from moieties of the iturins, surfactins, and fengycins. Moreover, unlike other Bacillus LP biosynthesis, biosynthesis of locillomycins used a nonlinear pathway, where a nonapeptide is assembled by hexamodular NRPSs. With the possibility of making novel LPs by rationally manipulated NRPSs, the unusual structure and unique biosynthesis pathway of locillomycins could inspire novel means to construct hybrid NRPSs, which could be used to synthesize new LP derivatives (10, 32, 33). Our results showed that the bacillomycin Ls, together with fengycins, mainly contribute to the antifungal activity of B. subtilis 916. In addition, bacillomycin Ls and surfactins obviously contributed to the hemolytic activity of this strain. Though the surfactins and locillomycins showed limited antifungal activities, they showed strong antibacterial and antiviral activities (data not shown). Particular emphasis should be placed on the observation that locillomycins and fengycins exhibited low hemolytic activities and strong antimicrobial activities, which could give rise to potential therapeutic applications. The antifungal and hemolytic activities of double and triple mutant strains were further decreased compared to those of their parent strains. In summary, the four LP families contributed individually, differently, and synergistically to the antimicrobial and hemolytic activity of B. subtilis 916. Recently, LPs were proven to function as signal molecules in the multicellular behaviors of B. subtilis in terms of swarming motility, biofilm formation, and colony morphology (9, 16, 19, 34). Surfactins contributed to the swarming motility and biofilm formation by acting as wetting agents for reducing the surface tension and acting as pheromones for quorum sensing (3, 5, 9, 20). In contrast to the data on surfactins, the reports on bacillomycins which facilitated swarming motility and were involved in biofilm formation are fewer and inconclusive (5, 7, 18). In this study, we demonstrated unambiguously that both surfactins and bacillomycin Ls are involved in swarming motility and biofilm formation of B. subtilis 916. It is interesting that while attempts to restore biofilm formation of srfaa mutant by adding both surfactins and bacillomycin Ls were unsuccessful, the biofilm formation of bacd mutant was restored significantly upon the addition of either surfactins or bacillomycin Ls. On the basis of these results, we presume that surfactins and bacillomycin Ls trigger mature January 2015 Volume 81 Number 1 Applied and Environmental Microbiology aem.asm.org 429

9 Luo et al. biofilm formation involved in different quorum-sensing pathways. While the locillomycins have no significant influence on the swarming motility and biofilm formation of B. subtilis 916, the colony morphology of the loc mutant, like that of srf and bmy mutants, changed significantly compared to that of wild-type B. subtilis 916. The results strongly suggest that locillomycins, like surfactins and bacillomycin Ls, take part in the multicellular behaviors of B. subtilis 916. However, it seems that the fengycins contribute only to the antifungal activity and are not involved in other phenotypical features of B. subtilis 916. In view of the different contributions of surfactins, bacillomycin Ls, and locillomycins to multicellular behaviors of B. subtilis 916, it will be interesting to further decipher the detailed mechanisms of different LPs with regard to how they trigger active kinase expression and regulate the swarming motility, biofilm formation, and colony morphology of B. subtilis 916 (21, 22, 34 39). In conclusion, we identified and characterized four NRPS gene clusters responsible for the biosynthesis of four families of LPs surfactins, bacillomycin Ls, fengycins, and locillomycin in B. subtilis 916. Functions of these LPs with regard to the phenotypic features were evaluated. Deficiencies in producing one or more of these four families of LPs led to different results, some of which are straightforward while others are elusive. It will be interesting to know how the genes and chemicals involved in these phenotypic features are affected by the four families of LPs. An investigation at the genomics level on this question will be our next challenge. ACKNOWLEDGMENTS We thank Jun Yao Guo for the linguistic revision and critical review of the manuscript. This work was supported by the National High-tech R&D Program of China (2011AA10A201), National Natural Science Foundation of China (grant ), and the Science Foundation of the Jiangsu Academy of Agricultural Sciences [grant CX(12)5001]. REFERENCES 1. Asaka O, Shoda M Biocontrol of Rhizoctonia solani damping-off of tomato with Bacillus subtilis RB14. Appl Environ Microbiol 62: Xu Z, Zhang R, Wang D, Qiu M, Feng H, Zhang N, Shen Q Enhanced control of cucumber wilt disease by Bacillus amyloliquefaciens SQR9 by altering the regulation of its DegU phosphorylation. Appl Environ Microbiol 80: Bais HP, Fall R, Vivanco JM Biocontrol of Bacillus subtilis against infection of Arabidopsis roots by Pseudomonas syringae is facilitated by biofilm formation and surfactin production. Plant Physiol 134: Ahimou F, Jacques P, Deleu M Surfactin and iturin A effects on Bacillus subtilis surface hydrophobicity. Enzyme Microb Technol 27: Zeriouh H, de Vicente A, Pérez-García A, Romero D Surfactin triggers biofilm formation of Bacillus subtilis in melon phylloplane and contributes to the biocontrol activity. Environ Microbiol 16: Stein T Bacillus subtilis antibiotics: structures, syntheses and specific functions. Mol Microbiol 56: Ongena M, Jacques P Bacillus lipopeptides: versatile weapons for plant disease biocontrol. Trends Microbiol 16: / /j.tim Romero D, de Vicente A, Rakotoaly RH, Dufour SE, Veening J-W, Arrebola E, Cazorla FM, Kuipers OP, Paquot M, Pérez-García A The iturin and fengycin families of lipopeptides are key factors in antagonism of Bacillus subtilis toward Podosphaera fusca. Mol Plant Microbe Interact 20: Ongena M, Jourdan E, Adam A, Paquot M, Brans A, Joris B, Arpigny J-L, Thonart P Surfactin and fengycin lipopeptides of Bacillus subtilis as elicitors of induced systemic resistance in plants. Environ Microbiol 9: Sugimoto Y, Ding L, Ishida K, Hertweck C Rational design of modular polyketide synthases: morphing the aureothin pathway into a luteoreticulin assembly line. Angew Chem Int Ed Engl 53: Mootz HD, Schwarzer D, Marahiel MA Construction of hybrid peptide synthetases by module and domain fusions. Proc Natl Acad Sci USA97: Challis GL Mining microbial genomes for new natural products and biosynthetic pathways. Microbiology 154: Velásquez JE, van der Donk WA Genome mining for ribosomally synthesized natural products. Curr Opin Chem Biol. 15: Wang X, Luo C, Chen Z Genome sequence of the plant growthpromoting rhizobacterium Bacillus sp. strain 916. J Bacteriol 194: Duitman EH, Hamoen LW, Rembold M, Venema G, Seitz H, Saenger W, Bernhard F, Reinhardt R, Schmidt M, Ullrich C, Stein T, Leenders F, Vater J The mycosubtilin synthetase of Bacillus subtilis ATCC 6633: a multifunctional hybrid between a peptide synthetase, an amino transferase, and a fatty acid synthase. Proc Natl Acad Sci U S A 96: Shank EA, Kolter R Extracellular signaling and multicellularity in Bacillus subtilis. Curr Opin Microbiol 14: Houry A, Briandet R, Aymerich S, Gohar M Involvement of motility and flagella in Bacillus cereus biofilm formation. Microbiology 156: Xu Z, Shao J, Li B, Yan X, Shen Q, Zhang R Contribution of bacillomycin D in Bacillus amyloliquefaciens SQR9 to antifungal activity and biofilm formation. Appl Environ Microbiol 79: López D, Vlamakis H, Losick R, Kolter R Cannibalism enhances biofilm development in Bacillus subtilis. Mol Microbiol 74: http: //dx.doi.org/ /j x. 20. Kinsinger RF, Shirk MC, Fall R Rapid surface motility in Bacillus subtilis is dependent on extracellular surfactin and potassium ion. J Bacteriol. 185: Vlamakis H, Chai Y, Beauregard P, Losick R, Kolter R Sticking together: building a biofilm the Bacillus subtilis way. Nat Rev Microbiol 11: López D, Fischbach Ma, Chu F, Losick R, Kolter R Structurally diverse natural products that cause potassium leakage trigger multicellularity in Bacillus subtilis. Proc Natl Acad Sci U S A 106: Kunst F, Rapoport G Salt stress is an environmental signal affecting degradative enzyme synthesis in Bacillus subtilis. J Bacteriol 177: Luo C, Chen Z, Liu Y, Zhang J, Liu Y, Wang X, Nie Y Construction and function analysis of Bac operon mutants of bio-control strain Bacillus subtilis Bs-916. Wei Sheng Wu Xue Bao 49: Romero D, Aguilar C, Losick R, Kolter R Amyloid fibers provide structural integrity to Bacillus subtilis biofilms. Proc Natl Acad Sci U S A 107: Tsuge K, Akiyama T, Shoda M Cloning, sequencing, and characterization of the iturin A operon. J Bacteriol 183: Borisova Sa, Circello BT, Zhang JK, van der Donk WA, Metcalf WW Biosynthesis of rhizocticins, antifungal phosphonate oligopeptides produced by Bacillus subtilis ATCC6633. Chem Biol 17: Kessler N, Schuhmann H, Morneweg S, Linne U, Marahiel Ma The linear pentadecapeptide gramicidin is assembled by four multimodular nonribosomal peptide synthetases that comprise 16 modules with 56 catalytic domains. J Biol Chem 279: /jbc.m Moss SJ, Martin CJ, Wilkinson B Loss of co-linearity by modular polyketide synthases: a mechanism for the evolution of chemical diversity. Nat Prod Rep 21: Sieber SA, Marahiel MA Molecular mechanisms underlying non- 430 aem.asm.org Applied and Environmental Microbiology January 2015 Volume 81 Number 1