Deletion of COM donor and acceptor domains and the interaction between modules in bacillomycin D produced by Bacillus amyloliquefaciens.

Bacillomycin D is a cyclic lipopeptide produced by Bacillus amyloliquefaciens fmbJ. At present, no relevant report has described the combinatorial biosynthesis of bacillomycin D. Due to the strong biosynthetic potential of the communication-mediating (COM) domains, its crosstalk between NRPS subunits has been studied to some extent, but the interaction of COM domain between modules is rarely reported. Therefore, in this study, we conducted the combinatorial biosynthesis of bacillomycin D through the deletion of the COM donor and acceptor domains between the modules and elucidated the interaction between the NRPS modules. The results showed that the deletion of the donor domain between modules 2 and 3 did not affect catalysis by upstream modules, but prevented downstream modules from catalysing the extension of the lipopeptide product, ultimately resulting in mutant complexes that could form linear dipeptides with the sequence β-NH2FA-Asn-Tyr. However, the engineered hybrid bacillomycin D NRPSs lacking the donor domains between modules 3 and 4 and modules 6 and 7 could form multiple assembly lines that produced bacillomycin D and its analogs (linear tripeptides, cyclic hexapeptides and linear hexapeptides). In addition, all the acceptor domain deletion strains failed to produce bacillomycin D, only truncated peptides produced by module interruption (except for the acceptor domain deletion strains between modules 3 and 4, which also produced cyclic hexapeptides). In conclusion, deletion of the inter-module donor domains led to a more flexible hybrid biosynthetic system for the production of diverse peptide products; compared with the inter-subunit donor domain deletion strains that could only produce truncated peptides, the former had a greater biosynthetic capacity. Meanwhile, the acceptor domains between modules were an important part of module-module interactions and efficient communication within bacillomycin D synthetase.

Introduction

Bacillus amyloliquefaciens fmbJ belongs to the genus Bacillus and is a biocontrol microorganism with strong antagonistic effects on pathogenic fungi such as Aspergillus flavus, Aspergillus ochraceus, Fusarium graminearum and Rhizopus stolonifer [[1], [2], [3], [4]]. Amphiphilic antimicrobial lipopeptides with β-hydroxy or β-amino fatty acid chains synthesized by nonribosomal pathways are widely present in Bacillus [[5], [6], [7]]. Due to their special amphiphilic structure, antimicrobial lipopeptides possess both antibacterial activity and biosurfactant properties [8], mainly consisting of three major families of surfactin, fengycin and iturin [9,10]. Bacillomycin D, belonging to the iturin family, is an antifungal compound with a cyclic lipopeptide structure composed of a β-amino aliphatic chain linked to a heptapeptide [[11], [12], [13]]. As shown in Fig. 1a, the synthase operon of bacillomycin D consists of four open reading frames, bamD, bamA, bamB and bamC, and bacillomycin D synthase is also a PKS/NRPS hybrid enzyme [14].

Fig. 1

Overview of bacillomycin D biosynthesis. (a) The bacillomycin D synthase operon and its chemical structure, in which the three pairs of black symbols represent the three pairs of COM domains in bacillomycin D, (b) complete amino acid sequence of the COM domain in bacillomycin D, (c) COM domain between TycA and TycB subunits in tyrocidine synthase, and (d) COM domain between PpsA and PpsB subunits in plipastatin synthase.

According to the molecular logic employed by NRPS, the biosynthesis of specific products relies on the selectivity of individual modules and their coordinated interactions [15,16]. Namely, in multienzyme complexes, the immobilized assembly line is required for interactions between adjacent modules while preventing interactions between nonadjacent modules. In general, the short terminal structure provides a pathway for the synthesis of intermediates along the correct assembly line, similar to the docking domain in PKS synthase, known as the COM domain in NRPS [17]. As a linker, the COM domain is divided into a donor and an acceptor. The pairing of donor and acceptor domains promotes the correct positioning of the NRPS subunit in the multienzyme complex. The donor-acceptor structure of the COM domain usually consists of 15–30 amino acid residues [18], and sequence analysis predicts that both the donor and acceptor parts of the COM domain have an α-helical conformation [19,20], which is considered an important structure for the mutual selection of donors and acceptors. Furthermore, selective communication is presumed to be dominated by polar and electrostatic interactions [17]. Hahn [17,19] and colleagues mapped the position of the COM domain, revealing the highly conserved sequence TPSD within the E domain that is used as a fusion site between the E domain and the C-terminal donor domain (the exact fusion site is in italics). The conserved sequence L(T/S) P(M/L) QEG also exists at the N-terminus of the acceptor module and was identified as the fusion site between the C domain and the N-terminal COM acceptor domain. Therefore, three pairs of COM domains are present in bacillomycin D and located between module 2 and module 3 (COMDBamB1-COMABamB2), module 3 and module 4 (COMDBamB2-COMABamB3), and module 6 and module 7 (COMDBamC1-COMABamC2) (see Fig. 1a and b). This differs from the positions of previously studied COM domains between subunits, such as the COM domains between TycA-TycB subunits and PpsA-PpsB subunits (see Fig. 1c and d). Therefore, throughout this paper, the comparison and discussion are carried out through the two expressions of inter-module COM domains and inter-subunit COM domains (the COM domains involved in this study are all located at the interface of the E-C domain, and the inter-modules here are located within subunits).

At present, the combinatorial biosynthesis of synthases such as surfactin [21], fengycin [22] and tyrocidine [23] has been widely reported, but the combinatorial biosynthesis strategy for bacillomycin D, a hybrid enzyme system, has not been studied. Previous researchers have successfully generated new analogs through the mutation, deletion and replacement of COM domains between NRPS subunits and suggested that the COM domain plays a decisive role in the interaction between NRPS subunits [18,22,24]. This makes it possible to uses COM domains to mediate module interactions and promote diverse polypeptide production. In this study, we analyzed the interaction between modules by deleting three pairs of COM domain donors and acceptors in bacillomycin D produced by B. amyloliquefaciens fmbJ. In addition, we determined whether the different positions of the donor and acceptor exerted different effects on the synthesis of bacillomycin D, whether the deletion of the COM domain affected the interaction between the multiple NRPS modules, and whether the compatibility of the COM domains between modules was consistent with the COM domains between NRPS subunits. Thus, our study complements the combinatorial biosynthesis mediated by the COM domain, and is expected to provide an additional reference for the versatility of the COM domain.

Materials and methods

Strains and culture conditions

B. amyloliquefaciens fmbJ, the cloned host strain Escherichia coli JM 109 and integration plasmid pKS2 (eliminated the BamHI site) were all preserved by the Enzyme Engineering Laboratory of Nanjing Agricultural University. The demethylation host Escherichia coli JM110 was purchased from TransGen Biotech. Bacterial strains were routinely grown on Luria–Bertani (LB) agar plates or in LB broth at 37 °C. The seed medium contained 5.0 g L−1 beef extract, 5.0 g L−1 yeast extract, 10.0 g L−1 peptone and 5.0 g L−1 NaCl. The fermentation medium used for lipopeptide production was described previously [2,25].

Construction of deletion plasmids

The plasmids with donor and receptor domain deletions in bacillomycin D synthase are all derivatives of pKS2. The corresponding donor and acceptor domains were deleted using plasmids pKS2ΔCOMDBamB1, pKS2ΔCOMDBamB2, pKS2ΔCOMDBamC1, pKS2ΔCOMABamB2, pKS2ΔCOMABamB3 and pKS2ΔCOMABamC2. Primers were designed according to the genome sequence of B. amyloliquefaciens fmbJ [25], and the primer sequences are shown in Table 1. For pKS2ΔCOMDBamB1, the upstream (548 bp) and downstream (527 bp) DNA fragments of the donor COMDBamB1 were amplified by PCR with the primer pairs COMDBamB1-UF/R and COMDBamB1-DF/R, respectively. These two fragments were used as templates for splice overlap extension polymerase chain reaction (SOE-PCR) with primers COMDBamB1-UF and COMDBamB1-DR. The upstream and downstream fusion fragments of COMDBamB1 were obtained and cloned into the SalI-KpnI sites of pKS2 using a one-step recombinase to obtain pKS2ΔCOMDBamB1. The obtained plasmid was sequenced and verified by GenScript Bio. The other deletion plasmids, pKS2ΔCOMDBamB2, pKS2ΔCOMDBamC1, pKS2ΔCOMABamB2, pKS2ΔCOMABamB3 and pKS2ΔCOMABamC2, were constructed in the same manner as pKS2ΔCOMDBamB1.

Table 1

Primer sequences used to construct donor and acceptor deletion strains.

Name	Sequence (5'∼3′)	Restriction site
pKS2-F	GgtcgacGGTATCGATAAGCTT	SalI
pKS2-R	ggtaccCAATTCGCCCTATAGTG	KpnI
COMDBamB1-UF	TATAGGGCGAATTGggtaccCGGCCGAGCCATTCAATATATTG	KpnI
COMDBamB1-UR	AGCAGACTTCTCTCACACCAGAAAATATTTATCCATTAACGCCGATGC
COMDBamB1-DF	TGGTGTGAGAGAAGTCTGCTTCCGGCTTACGCAATGTG
COMDBamB1-DR	GCTTATCGATACCgtcgacCTCGTGAATTTGGAAGGCCAC	SalI
COMABamB2-UF	TATAGGGCGAATTGggtaccCAATAAGCCGTGGCCTCTTTC	KpnI
COMABamB2-UR	CGCGTGAATTAGGCGAAGCGCCGATGCAGAAGGGCATG
COMABamB2-DF	CGCTTCGCCTAATTCACGCGTCTGCTCCAAAAGCTGTTCC
COMABamB2-DR	GCTTATCGATACCgtcgacCCACAAGCCAGTACCCTGTC	SalI
COMDBamB2-UF	TATAGGGCGAATTGggtaccGCTGCCGATGACGGTTCTG	KpnI
COMDBamB2-UR	AACAATCGGAACTGACGCCAGTTGAAAACGTGTATCCGCTCAC
COMDBamB2-DF	TGGCGTCAGTTCCGATTGTTCTTTATGAACGCAATGCGTAATGATGG
COMDBamB2-DR	GCTTATCGATACCgtcgacCCGCATTGGTCAGGACTGGAAC	SalI
COMABamB3-UF	TATAGGGCGAATTGggtaccGGCGGTCAAGCCACTCGATATAG	KpnI
COMABamB3-UR	AACTGCCGCATGCAGGTGAGCCGATGCAGAAAGGAATGCTC
COMABamB3-DF	CTCACCTGCATGCGGCAGTTGAATGAGGAGCTGATCCAATTCGTC
COMABamB3-DR	GCTTATCGATACCgtcgacCGTATCCGGTTGTGCTGCACG	SalI
COMDBamC1-UF	TAGGGCGAATTGggtaccGGGGCTGTATGGAGTGACGGC	KpnI
COMDBamC1-UR	ATCAAATTCACCTGACGCCAATTGAAAATATATACCCATTAACCCCGATGC
COMDBamC1-DF	TGGCGTCAGGTGAATTTGATCTTGAGCATCACAATGTGCG
COMDBamC1-DR	TATCGATACCgtcgacCTCGGCTATTCATCGTTGGACCGG	SalI
COMABamC2-UF	TAGGGCGAATTGgttaccGGCCAATAGGCTGACGCCTG	KpnI
COMABamC2-UR	AGACGAGTCATCTCGGTGACCCGATGCAGAAAGGAATGTTGTTC
COMABamC2-DF	GTCACCGAGATGACTCGTCTGCTGCACAAATTGATCCAATTCAC
COMABamC2-DR	TATCGATACCgtcgacCTCGGTGGGATGGTTCACAAGCCT	SalI

Lowercase bases represent the restriction site, and underlined bases represent the homologous sequence.

Construction of Bacillus amyloliquefaciens deletion mutants

Each deletion plasmid was electroporated into B.amyloliquefaciens fmbJ [26]. Using the kanamycin and erythromycin markers present in the pKS2 plasmid, the positive transformants were screened with 20 μg mL−1 kanamycin and 5 μg mL−1 erythromycin at 30 °C. The deletion mutants were selected using a two-step replacement recombination procedure described previously [5,27]. The strain was grown at 37 °C (a temperature not allowed for plasmid replication) in the presence of kanamycin and erythromycin to select clones in which the plasmid was integrated into the chromosome between the target gene and a homologous sequence on the plasmid through a single crossover [5]. Afterward, a separate clone of the integrant was cultured in LB medium at 30 °C for 5 generations (12 h per generation) to induce a second crossover event and excise the plasmid. The obtained kanamycin- and erythromycin-sensitive strains with original or deleted sequences were verified by PCR and sequencing (see Table 2 for the primers used in this experiment). The mutant strains were named fmbJD1, fmbJD2, fmbJD3, fmbJA1, fmbJA2 and fmbJA3 (see Table 3).

Table 2

Primer sequences used to verify donor and acceptor deletion strains.

Name	Sequence(5'∼3′)
ΔCOMDBamB1-F	CGCCTAATTCACGCGTCTG
ΔCOMABamB2-F	GGCGTTAATGGATAAATATTTTCCGCTTC
ΔCOMDBamB2-F	CCAATTCGTCAATGGACATGCCT
ΔCOMABamB3-F	AGTGAGCGGATACACGTTTTCAACC
ΔCOMDBamC1-F	TGATCCAATTCACCGATCGTTATACC
ΔCOMABamC2-F	CGGGGTTAATGGGTATATATTTTCAATGTC

The upstream primer sequences all contain the corresponding donor and acceptor sequences. If no bands appear after gel electrophoresis, the donor or acceptor domains have been deleted; the downstream primer sequences are the corresponding DR primers listed in Table 1 above.

Table 3

The strains used in this study and their lipopeptide products.

Deletion strategies		Strains	Lipopeptide products	Molecular formula	Mass(m/z)	Peptide sequence	Yield(%)
Wild type		fmbJ	bacillomycin D	C48H74N19O15	1031.5413	C14-17β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser-Thr	100
				C49H76N19O15	1045.5570
				C50H78N19O15	1059.5726
				C51H80N19O15	1073.5883
Donor deletion	the donor between modules 2 and 3 (COMDBamB1)	fmbJD1	linear dipeptide	C27H44N4O6	521.3327	C14-16β-NH2FA-Asn-Tyr	100
				C28H46N4O6	535.3483
				C29H48N4O6	549.3625
	the donor between modules 3 and 4 (COMDBamB2)	fmbJD2	bacillomycin D	C48H74N19O15	1031.5404	C14-17β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser-Thr	89.05
				C49H76N19O15	1045.5569
				C50H78N19O15	1059.5721
				C51H80N19O15	1073.5881
			linear tripeptide	C31H50N6O8	635.3770	C14-16β-NH2FA-Asn-Tyr-Asn	9.75
				C32H52N6O8	649.3930
				C33H54N6O8	663.4085
			cyclic hexapeptide	C43H67N9O14	934.4902	C14-18β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr	1.20
				C44H69N9O14	948.5037
				C45H71N9O14	962.5222
				C45H73N9O14	976.5385
				C46H75N9O14	990.5536
	the donor between modules 6 and 7 (COMDBamC1)	fmbJD3	bacillomycin D	C48H74N19O15	1031.5404	C14-17β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser-Thr	91.88
				C49H76N19O15	1045.5571
				C50H78N19O15	1059.5723
				C51H80N19O15	1073.5874
			linear hexapeptide	C44H69N9O14	948.5020	C14-16β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser	8.12
				C45H71N9O14	962.5234
				C46H73N9O14	976.5341
Acceptor deletion	the acceptor between modules 2 and 3(COMABamB2)	fmbJA1	linear dipeptide	C27H44N4O6	521.3337	C14-16β-NH2FA-Asn-Tyr	100
				C28H46N4O6	535.3492
				C29H48N4O6	549.3650
	the acceptor between modules 3 and 4(COMABamB3)	fmbJA2	linear tripeptide	C31H50N6O8	635.3776	C14-16β-NH2FA-Asn-Tyr-Asn	57.85
				C32H52N6O8	649.3937
				C33H54N6O8	663.4080
			cyclic hexapeptide	C43H67N9O14	934.4950	C14-16β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr	42.15
				C44H69N9O14	948.5126
				C45H71N9O14	962.5281
	the acceptor between modules 6 and 7(COMABamC2)	fmbJA3	linear hexapeptide	C44H69N9O14	948.5079	C14-15β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser	100
				C45H71N9O14	962.5276

The yield here is only the relative proportion of different lipopeptides produced by the same strain, which is obtained by integrating the peak area, not the real yield of lipopeptides.

Extraction and identification of lipopeptide antibacterial substances

Single colonies of the wild-type fmbJ and mutant strains obtained after activation were inoculated into the seed medium, cultured at 37 °C and 180 rpm to the logarithmic growth phase, inoculated into the fermentation medium at an inoculum concentration of 5% and cultured at 30 °C for 120 h with shaking at 180 r·min−1. After centrifugation, the fermentation broth supernatant was collected, and its pH was adjusted to 2.0 with 6 M HCl. Then, the solution was stored at 4 °C until further treatment. Subsequently, the precipitate was collected after centrifugation, and the supernatant was discarded. Methanol was added to dissolve the pellet, and then the pH was adjusted to 7.0 with NaOH and the sample was centrifuged at 10,000 g for 10 min to obtain the crude lipopeptide extract [1,25]. After the crude extract was filtered through a 0.22 μm membrane, a Waters Xevo G2-XS Q-TOF mass spectrometer was used for high-resolution liquid chromatography-electrospray ionization-mass spectrometry (LC-ESI-MS). HPLC and mass spectrometry conditions were described in the studies by Gong and Qian et al. [1,2].

Results

Donor and acceptor knockout strain acquisition and HPLC detection

As shown in Table 3, we deleted three pairs of COM domain donors and acceptors in bacillomycin D to study the interaction between modules. First, the constructed deletion plasmids were electroporated into the target strain B. amyloliquefaciens fmbJ, and then the mutant strains fmbJD1, fmbJD2, fmbJD3, fmbJA1, fmbJA2 and fmbJA3 were obtained through markerless deletion using temperature-induced homologous recombination double crossovers. The kanamycin- and erythromycin-sensitive strains were verified by PCR, and the strains that verified correct were sent to GenScript Bio Company for sequencing. According to the sequencing results, the donor and acceptor domain deletion strains have been obtained (supplementary materials: Fig. S1).

The crude antibacterial lipopeptide extracts obtained by the fermentation of the 6 mutant strains described above were analyzed using HPLC. It was found that the presence of bacillomycin D was only detected in the fermentation crude extracts of the donor deletion strains fmbJD2 and fmbJD3. The bacillomycin D production level of fmbJD2 was comparable to that of wild-type fmbJ, while the bacillomycin D production of fmbJD3 was significantly decreased (see Fig. 2a). The expression level of TE domain in fmbJD3 was significantly decreased at all times during the fermentation process, especially at 36 h of fermentation; while the donor deletion strain fmbJD2 did not show a decrease in the expression of the TE domain throughout the fermentation process (see Fig. 2b). This was due to the fact that the donor COMDBamC1 was located between modules 6 and 7, and the TE domain that mediated product release was located in module 7. And the deletion of the donor COMDBamC1 affected the interaction with the downstream module (module 7) to a certain extent, resulting in a significant decrease in the production of bacillomycin D.

Fig. 2

Deletion of the donor domain COMDBamB2 does not affect the production of bacillomycin D. (a) HPLC analysis of fermentation crude extracts. The production of bacillomycin D (four homologs of C14-17) in the fermentation crude extracts of strains fmbJ, fmbJD2 and fmbJD3 were 331.51 ± 5.94, 335.69 ± 10.31 and 49.73 ± 11.93 mg L−1, respectively. (Standard curve: y = 7.6396x-2.3576; x, content of bacillomycin D, mg·L−1; y, peak area, mAU·h), and (b) determination of relative expression of TE domain.

Identification of lipopeptide analogs in fermentation products of donor deletion strains

Pure HPLC analysis is unable to determine whether the donor deletion strains fmbJD1, fmbJD2 and fmbJD3 synthesize bacillomycin D and its analogs; therefore, we further detected and analyzed the crude lipopeptide extracts using high-resolution ESI-LC-MS. Both fmbJD2 and fmbJD3 synthesized bacillomycin D, but fmbJD1 did not synthesize bacillomycin D. In addition, the fermentation product of the mutant strain fmbJD1 also exhibited a signal with an ionic molecular weight of m/z 521.3327, as shown in Fig. 3a. This signal was identical to the preliminarily predicted molecular weight of a novel linear dipeptide (C14β-NH2FA-Asn-Tyr) whose predicted molecular formula was C27H44N4O6. Signals were also observed at m/z 535.3483 and m/z 549.3625, which were the molecular weights of C15 and C16, respectively. The precursor ion at m/z 521.3327 was selected for MS/MS analysis and the dipeptide sequence was determined to be C14β-NH2FA-Asn-Tyr. As shown in Fig. 3c, the C-terminal y-ion fragments m/z 296.1230 and m/z 182.0805 in the MS/MS spectrum were consistent with the predicted fragmentation values of linear dipeptide molecules, and fragment ion peaks at m/z 521.3318 and m/z 115.0501 were detected; thus, we determined that the assembly line of the lipopeptide analog was C14β-NH2FA-Asn-Tyr. These results indicated that the deletion of the donor COMDBamB1 did not affect catalysis by upstream modules, but prevented the downstream modules from catalysing the extension of the lipopeptide product, ultimately resulting in mutant complexes that could form linear dipeptides with the sequence β-NH2FA-Asn-Tyr.

Fig. 3

LC-MS/MS analysis of the crude extract from the mutant strain fmbJD1. (a) Chromatogram corresponding to m/z 521.3327, (b) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJD1, and (c) ESI-MS/MS fragment ion analysis at m/z 521.3327.

The fermentation product of the donor deletion strain fmbJD2 exhibited a signal with an ionic molecular weight of m/z 934.4902 in addition to a signal with an ionic molecular weight of m/z 635.3770, as shown in Fig. 4a and b. The aforementioned signal was the same as the preliminarily predicted molecular weight of a novel linear tripeptide (C14β-NH2FA-Asn-Tyr-Asn) and a novel cyclic hexapeptide (C14β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr). The molecular formulas were C31H50N6O8 and C43H67N9O14, respectively. In addition, signals at m/z 649.3930 and m/z 663.4085 were detected as linear tripeptides with molecular weights of C15 and C16 and at m/z 948.5037, m/z 962.5222, m/z 976.5385 and m/z 990.5536 as cyclic hexapeptides with molecular weights of C15, C16, C17 and C18. We speculated that the deletion of COMDBamB2 may cause the assembly line of bacillomycin D to undergo module skipping, namely, to skip module 4, to generate a new cyclic hexapeptide assembly line.

Fig. 4

LC-MS/MS analysis of the crude extract from the mutant strain fmbJD2. (a) Chromatogram corresponding to m/z 635.3770, (b) chromatogram corresponding to m/z 934.4902, (c) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJD2, (d) ESI-MS/MS fragment ion analysis at m/z 635.3770, and (e) ESI-MS/MS fragment ion analysis at m/z 934.4902.

The precursor ions m/z 635.3770 and m/z 934.4902 were selected for MS/MS analysis. As shown in Fig. 4d, the N-terminal b-ion fragments m/z 503.3228 and m/z 340.2594 and the C-terminal y-ion fragments m/z 410.1662, m/z 296.1245 and m/z 133.0611 were consistent with the predicted fragment values of linear tripeptide molecules. The assembly line of the lipopeptide analogs was determined to be C14β-NH2FA-Asn-Tyr-Asn. The MS/MS analysis of the ion molecular weight of m/z 934.4902 is shown in Fig. 4e. Fragment ion peaks m/z 617.3665, m/z 503.3227 and m/z 340.2594 were N-terminal b-ion fragments, in addition to the fragment ion peaks m/z 243.0758, m/z 216.0663 and m/z 188.0710, which were consistent with the predicted fragment values of cyclic hexalipopeptide molecules. From this information, we deduced that the compound structure of m/z 934.4902 was β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr. After the deletion of the donor domain COMDBamB2, new assembly lines, namely, the linear tripeptide assembly line and cyclic hexapeptide assembly line, were generated due to module interruption and module skipping in the presence of bacillomycin D. This suggested that in this complex biosynthetic system, a series of modules upstream of module 4 retained the ability to assemble the precursor tripeptide chain. However, due to the different strengths of the interactions between the modules, competition for connection resulted in different assembly lines for the production of different lipopeptides. Among them, the yield of bacillomycin D could account for 89.05%, and the yield of linear tripeptides and cyclic hexapeptides accounted for 9.75% and 1.2% (see Table 3); that is, the interaction between modules 3 and 4 was stronger than the hydrolysis of the TE domain and the interaction between modules 3 and 5.

Similarly, the fermentation product of the mutant strain fmbJD3 also had the predicted molecular weight of m/z 948.5020 due to module interruption, as shown in Fig. 5a. We preliminarily speculated that the assembly line was disconnected from module 6 and module 7, resulting in the assembly line of the linear hexapeptide β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser, in addition to the existence of linear hexapeptide homologs at C15 and C16. The MS/MS analysis of the ion molecular weight of m/z 948.5020 is shown in Fig. 5c. A series of ion fragment peaks (609.2510 → 446.1871 → 332.1452 and 617.3663 → 503.3229 → 340.2590) was consistent with the fragment values for the novel linear hexapeptide C14β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser. Thus, after the deletion of the donor COMDBamC1 in bacillomycin D, a series of modules located upstream of module 7 retained the ability to assemble the precursor hexapeptide chain. However, since the interaction between modules 6 and 7 was stronger than the hydrolysis of the TE domain, two assembly lines were generated to direct the production of bacillomycin D (91.88%) and linear hexapeptide (8.12%) (see Table 3).

Fig. 5

LC-MS/MS analysis of the crude extract from the mutant strain fmbJD3. (a) Chromatogram corresponding to m/z 948.5020, (b) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJD3, and (c) ESI-MS/MS fragment ion analysis at m/z 948.5020.

Identification of lipopeptide analogs in fermentation products of acceptor deletion strains

No bacillomycin D signal was detected in the fermentation product of the acceptor deletion strain fmbJA1, but it also showed the ionic molecular weight of the linear dipeptide m/z 521.3337, as shown in Fig. 6a, and the predicted molecular formula was C27H44N4O6. Signals were also detected at m/z 535.3492 and m/z 549.3650, which were the molecular weights of C15 and C16, respectively. The precursor ion at m/z 521.3337 was selected for MS/MS analysis, and the results are shown in Fig. 6c. The C-terminal y-ion fragments m/z 296.1232 and m/z 182.0805 present in the MS/MS spectrum were consistent with the predicted fragment values of linear dipeptide molecules. In addition to the fragment ion peaks m/z 521.3335 and m/z 115.0501, we determined that the assembly line of the lipopeptide analog is C14β-NH2FA-Asn-Tyr. The aforementioned results showed that the deletion of the acceptor COMABamB2 also prevented the downstream modules from catalysing the extension of the lipopeptide product, which ultimately guided the synthesis of linear dipeptides by the assembly line β-NH2FA-Asn-Tyr.

Fig. 6

LC-MS/MS analysis of the crude extract from the mutant strain fmbJA1. (a) Chromatogram corresponding to m/z 521.3337, (b) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJA1, and (c) ESI-MS/MS fragment ion analysis at m/z 521.3337.

The fermentation product of the acceptor deletion strain fmbJA2 was further detected and analyzed using high-resolution ESI-LC-MS, and bacillomycin D was also absent. However, the product exhibited the same ion molecular weight signals of m/z 635.3776 and m/z 934.4950 as the donor deletion strain fmbJD2 (see Fig. 7a and b). We speculated that the deletion of the acceptor COMABamB3 not only caused module interruption but also skipped module 4 to generate the assembly line of the cyclic hexapeptide. The predicted molecular formulas were C31H50N6O8 and C43H67N9O14. In addition, signals were detected at m/z 649.3937 and m/z 663.4080 as the molecular weights of the linear tripeptides at C15 and C16, respectively, and signals at m/z 948.5126 and m/z 962.5281 as the molecular weights of the cyclic hexapeptides at C15 and C16, respectively.

Fig. 7

LC-MS/MS analysis of the crude extract from the mutant strain fmbJA2. (a) Chromatogram corresponding to m/z 635.3776, (b) chromatogram corresponding to m/z 934.4950, (c) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJA2, (d) ESI-MS/MS fragment ion analysis at m/z 635.3776, and (e) ESI-MS/MS fragment ion analysis at m/z 934.4950.

The precursor ions m/z 635.3776 and m/z 934.4950 were selected for MS/MS analysis. As shown in Fig. 7d, the predicted linear tripeptide b- and y-fragment ion peaks at m/z 503.3226, m/z 340.2572 and m/z 410.1655, m/z 296.1237 were present in the MS/MS spectrum. Thus, the assembly line of the lipopeptide analog was determined to be C14β-NH2FA-Asn-Tyr-Asn. The MS/MS analysis of the ion with molecular weight m/z 934.4950 is shown in Fig. 7e. Fragment ion peaks at m/z 746.4059, m/z 619.3723, m/z 503.3210 and m/z 340.2600 were N-terminal b-ion fragments, and the fragment ion peak at m/z 656.3706 was consistent with the predicted fragment value for the cyclic hexalipeptide molecule. Based on this information, the structure of the compound at m/z 934.4950 was determined to be C14β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr. The acceptor domain COMABamB3 deletion strain generated a hybrid enzyme complex that produced two types of bacillomycin D analogs. Among them, the hydrolysis of the TE domain was slightly stronger than the interaction of modules 3 and 5, resulting in a slightly higher yield of linear tripeptides (57.85%) than that of cyclic hexapeptides (42.15%) (see Table 3).

Similarly, no bacillomycin D signal was detected in the fermentation product of the acceptor deletion strain fmbJA3, and only the predicted linear hexapeptide ion with a molecular weight m/z 948.5079 produced by the interruption of the module was present, as shown in Fig. 8a. A signal was detected at m/z 962.5276 for the linear hexapeptide homolog of C15. The ion with a molecular weight of m/z 948.5079 was selected as the precursor ion for MS/MS analysis, as shown in Fig. 8c. A series of ion fragment peaks (843.4588 → 714.4067 → 617.3638 → 503.3219 → 340.2557 and 609.2522 → 446.1873) were consistent with the fragment values for the novel linear hexapeptide C14β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser. These results showed that after the deletion of the acceptor COMABamC2 in bacillomycin D, the assembly line was unable to pass the substrate from module 6 to module 7, and could only produce linear hexapeptides (C14-15β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser).

Fig. 8

LC-MS/MS analysis of the crude extract from the mutant strain fmbJA3. (a) Chromatogram corresponding to m/z 948.5079, (b) ESI-LC-MS total chromatogram of the crude extract from the mutant strain fmbJA3, and (c) ESI-MS/MS fragment ion analysis at m/z 948.5079.

Discussion

With the advent of synthetic biology, knowledge of the modularity of the NRPS system, and a better understanding of the structure of the assembly line of NRPSs [28], combinatorial biosynthesis has emerged as a powerful strategy for the reprogramming of NRPSs [29,30]. In the past few decades, various reprogramming strategies have been developed for NRPSs. The first was gatekeeper domain exchange (deletion), which mainly focused on the replacement of the gatekeeper domain (A-PCP or A domain) [5,31,32]. The purpose was to load different monomers into the assembly line to obtain new analogs. The second strategy was extended to different domains, such as deletion and advancement of the TE domain in the module [33,34]. The third involved the exchange (deletion) of single or multiple complete modules within the assembly line [5,35]. Fourth, some researchers recently proposed the concept of XUs [36], and by exchanging XU units, they were able to create novel peptides at high yields. However, one limitation of this approach is the amino acid specificity of the downstream C domain. Therefore, authors identified a new fusion point within the C domain to overcome this problem, proposing another concept, XUCs [37]. This strategy was considered to have the potential to generate peptide libraries. While all of the aforementioned methods might lead to the synthesis of the desired NRP product, a significant reduction in product yield is often observed. Meanwhile, the current knowledge of the NRPS systems was mostly based on domain, dual domain and single NRPS module information [38,39]. However, little was known about the interactions between modules in multimodule NRPSs [40].

Bioinformatics analyses unveiled the prevalence of both inter- and intra-subunit COM interfaces throughout biosynthetic assembly lines, including two types of PCP domain-bridging interfaces not previously characterized [41]. Fage and co-workers validated the importance of the PCP-COMD domain in inter-subunit communication, demonstrating the versatility of COM regions in promoting functional domain-domain interactions beyond those of the canonical E and C domain pair [41]. Through the exchange of E-C COM domains between modules 4 and 5 and modules 5 and 6, Kaniusaite and co-workers demonstrated that the E-C COM domain between modules was similar to the E-C COM domain between subunits in that a successful domain exchange required compatible linkers to connect the upstream and downstream modules of interest [42]. In this study, we first located the specific position and sequence of the COM domain in bacillomycin D through the conserved sequences of the donor and acceptor domains at the E-C domain interface determined by Hahn [17,19] et al. (see Fig. 1). The position of the COM domain in bacillomycin D was located between modules, similar to the E-C COM domains between modules 4 and 5 and between modules 5 and 6 in teicoplanin non-ribosomal peptide synthetase [42], all of which belong to the intra-subunit COM interfaces reported by Fage [41]. To further clarify whether the compatibility of COM domain donors and acceptors between modules was consistent with the compatibility between subunits. And whether the different positions of the donor and acceptor in the assembly line exerted different effects on the synthesis of bacillomycin D. We explored these by deleting the donor and acceptor domains of three pairs of COM domains at different positions in bacillomycin D.

Regarding the deletion of the donor domain, deletion of COMDBamB1 (in module 2), COMDBamB2 (in module 3), and COMDBamC1 (in module 6) all resulted in module disruption, resulting in truncated peptides, namely, linear dipeptides (C14-16β-NH2FA-Asn-Tyr), linear tripeptides (C14-16β-NH2FA-Asn-Tyr-Asn) and linear hexapeptides (C14-16β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser) (the synthetic mechanisms are shown in Fig. 9a, b and 9d). These results were consistent with previous reports that the selectivity of the ppsC acceptor was blocked and led to a novel assembly line for the synthesis of hexalipopeptides (ppsA/ppsB/ppsC) when the ppsC donor was deleted [22]. However, in our study, we found that in addition to COMDBamB1-deleted strains, COMDBamB2- and COMDBamC1-deleted strains had the original bacillomycin D assembly line. In particular, the production of bacillomycin D was not significantly reduced after the deletion of COMDBamB2. Therefore, the position of the donor domain exerted different effects on the entire assembly line, and the COM domain at the front end played a more important role in the synthesis of bacillomycin D. Similar results have also been reported in previous studies of the deletion of the T domain at various positions in plipastatin. Deletion of the T domain of module 6 completely inactivated the plipastatin complex, whereas deletion of the T domain of module 7 resulted in the presence of a linear hexapeptide [5]. Thus, the T domain of module 6 played an important role in the overall structural conformation of the plipastatin NRPS complex. Based on these results, the deletion of the donor domain between modules or between NRPS subunits altered the interaction of upstream and downstream modules to varying degrees. However, the donor domains between NRPS subunits appear to play a more important role in the overall structural conformation of the NRPS complex than the donor domains between modules [22]. Interestingly, we also discovered a new assembly line for the synthesis of cyclic hexapeptides (C14-18β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr) in the fermentation product of the COMDBamB2-deleted strain (see Fig. 9c). This product has never been obtained from previous donor and acceptor deletions between NRPS subunits. The existence of a more flexible hybrid biosynthetic system suggests that the inter-module donor domain deletion strains have greater biosynthetic capacity than the inter-subunit donor domain deletion strains.

Fig. 9

Generation mechanism of four novel lipopeptides. (a) Formation of linear dipeptide due to premature hydrolysis of the thioesterase (TE) domain, (b)formation of linear tripeptide due to premature hydrolysis of the TE domain, (c) formation of cyclic hexapeptide by complete module 4 skipping, and (d) formation of linear hexapeptide due to premature hydrolysis of the TE domain.

In contrast, deletion of the acceptor domains between modules showed lower compatibility. After deleting COMABamB2 (in module 3), COMABamB3 (in module 4) and COMABamC2 (in module 7), we found that all strains were incapable of producing bacillomycin D and all strains generated truncated peptides due to module interruption: linear dipeptides (C14-16β-NH2FA-Asn-Tyr), linear tripeptides (C14-16β-NH2FA-Asn-Tyr-Asn) and linear hexapeptides (C14-15β-NH2FA-Asn-Tyr-Asn-Pro-Glu-Ser). These results were similar to those obtained through donor domain deletions. Thus, the deletion of acceptors did not affect the catalysis of the upstream module, but prevented the downstream module from catalysing the elongation of the lipopeptide product, ultimately leading to assembly line disruption. By comparing donor and acceptor substitutions, we found that donor substitution and donor deletion only generated truncated peptides. However, acceptor replacement only resulted in a reduction in yield and did not affect the original assembly line [18,22]. Therefore, we postulated that the acceptor domain between modules was more important for the overall structural conformation of the NRPS complex enzyme than the acceptor domain between NRPSs, indicating that the acceptor domain was an important part of module-to-module interactions and efficient communication within bacillomycin D synthetase.

In addition, we also found the C14-16β-NH2FA-Asn-Tyr-Asn-Glu-Ser-Thr assembly line for the synthesis of cyclic hexapeptide in the COMABamB3-deleted strain. This result indicated that the deletion of either the donor domain or the acceptor domain in modules 3 and 4 would lead to the skipping of module 4, resulting in the connection of module 3 and module 5. This suggested a strong interaction between module 3 and module 5 in bacillomycin D. The previous correct connection was attributed to the correct pairing of the COM domain donor and acceptor. When either the donor or acceptor was absent, the interaction between module 3 and module 4 was affected, resulting in this unusual assembly line (cyclic hexapeptides, Fig. 9c). Unusual biosynthetic pathways have also been shown in studies of teicoplanin synthase reprogramming, resulting in the presence of pentapeptides (M5-6a + M3+M4+M3) due to the strong interaction mediated by the M3/M4 interface [42].

Conclusion

In this study, the COM domains between the bacillomycin D synthetase modules were reported for the first time, and the interaction between NRPS modules was elucidated by the deletion of the donor and acceptor domains. In conclusion, deletion of the intermodular COM donor and acceptor domains might be used to reprogram bacillomycin D synthase in B. amyloliquefaciens, resulting in novel lipopeptides, namely, linear dipeptides, linear tripeptides, linear hexapeptides, and cyclic hexapeptides. Among them, cyclic hexapeptides were generated by module skipping. The presence of multiple assembly lines suggested that the combinatorial biosynthetic potential of the COM domains was greater than previously reported. In particular, the donor domain between modules did not affect the original bacillomycin D yield at all, which had never been reported before. However, the intermodule acceptor domain was an important part of module-module interactions and efficient communication in bacillomycin D synthetase. This study further complemented the current lack of studies on the interaction between NRPS modules and provided a theoretical basis for the generation of novel lipopeptides mediated by the intermodule COM domains.

CRediT authorship contribution statement

Ziyan Lv: Conceptualization, Methodology, Formal analysis, Investigation, Writing – original draft. Wenjie Ma: Methodology. Ping Zhang: Formal analysis. Zhaoxin Lu: Resources, Supervision. Libang Zhou: Validation. Fanqiang Meng: Validation. Zuwei Wang: Investigation. Xiaomei Bie: Conceptualization, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no conflict of interest.