Serratia plymuthica strain RVH1, initially isolated from an industrial food processing environment, displays potent antimicrobial activity towards a broad spectrum of Gram-positive and Gram-negative bacterial pathogens. Isolation and subsequent structure determination of bioactive molecules led to the identification of two polyamino antibiotics with the same molecular structure as zeamine and zeamine II as well as a third, closely related analogue, designated zeamine I. The gene cluster encoding the biosynthesis of the zeamine antibiotics was cloned and sequenced and shown to encode FAS, PKS as well as NRPS related enzymes in addition to putative tailoring and export enzymes. Interestingly, several genes show strong homology to the pfa cluster of genes involved in the biosynthesis of long chain polyunsaturated fatty acids in marine bacteria. We postulate that a mixed FAS/PKS and a hybrid NRPS/PKS assembly line each synthesize parts of the backbone that are linked together post-assembly in the case of zeamine and zeamine I. This interaction reflects a unique interplay between secondary lipid and secondary metabolite biosynthesis. Most likely, the zeamine antibiotics are produced as prodrugs that undergo activation in which a nonribosomal peptide sequence is cleaved off.
Citation: Masschelein J, Mattheus W, Gao L-J, Moons P, Van Houdt R, Uytterhoeven B, et al. (2013) A PKS/NRPS/FAS Hybrid Gene Cluster from Serratia plymuthica RVH1 Encoding the Biosynthesis of Three Broad Spectrum, Zeamine-Related Antibiotics. PLoS ONE 8(1): e54143. https://doi.org/10.1371/journal.pone.0054143
Editor: Arnold Driessen, University of Groningen, The Netherlands
Received: October 25, 2012; Accepted: December 6, 2012; Published: January 17, 2013
Copyright: © 2013 Masschelein et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Joleen Masschelein is a PhD fellow of the Research Foundation Flanders (www.fwo.be). This study was performed in the framework of the FWO “BaSe-ics” Research community (W0.014.12N). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Polyketides (PKs) and nonribosomal peptides (NRPs) represent two large families of structurally diverse and complex microbial metabolites that include many therapeutically valuable antibacterial drugs. Both are synthesized with similar logic by large multifunctional enzymes known as polyketide synthases (PKSs) and nonribosomal peptide synthetases (NRPSs), which link simple monomeric building blocks (acyl-CoAs and amino acids, respectively) by a cascade of condensation reactions. The enzymes are organized in a modular assembly line fashion, in which each module consists of a set of enzymatic domains, responsible for one discrete round of chain elongation and a variable set of modifications on each intermediate , . The modular architecture and the functional versatility of these enzymes, combined with post-PKS tailoring reactions, generate stereochemically complex compounds, characterized by a high level of structural variation. Depending on their biochemical properties and molecular architecture, PKSs are subdivided into two main classes. Type I PKSs consist of one or more multifunctional enzymes with active sites for each separate enzymatic reaction in the pathway. In contrast, type II synthases work iteratively, being composed of a minimal set of individual proteins that work together to create aromatic compounds .
Because of their structural and functional similarities, PKSs and NRPSs can form mixed assembly lines which give rise to hybrid polyketide-peptide natural products, as exemplified by epothilone D, rapamycin, yersiniabactin, and kalimantacin –. In addition, PKSs have also been shown to form hybrid assembly lines with the evolutionary related fatty acid synthases (FASs) with which they share striking similarities in domain organization, catalytic reactions, precursor usage and overall architectural arrangement. Recently, large-scale analysis of sequenced microbial genomes has revealed a variety of iteratively acting FAS/PKS hybrid systems closely related to the PUFA (polyunsaturated fatty acid) synthases . The pfaEABCD gene cluster is responsible for the production of long-chain polyunsaturated fatty acids such as eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in marine bacteria . The related FAS/PKS gene clusters, however, do not produce EPA or DHA but manufacture various other long chain fatty acyl products typically consisting of more than 20 carbons. Next to the type I fatty acid synthase (FASI) and the type II FAS system (FASII), whose structural organization served as the basis for the classification of PKSs, the iterative FAS/PKS mechanism represents the third bacterial fatty acid biosynthesis mechanism, referred to as “secondary lipid biosynthesis” .
Serratia plymuthica RVH1 is an opportunistic Gram-negative bacterium with strong biofilm-forming capacity, isolated from an industrial food processing environment . The strain is characterized by a LuxI/LuxR type quorum sensing (QS) system SplI/SplR that regulates the production of a number of extracellular enzymes , the switch from mixed acid fermentation to butanediol fermentation , , and the production of an antimicrobial compound  which is the research focus here. This compound is active against a broad spectrum of Gram-negative and Gram-positive bacteria and no cross-resistance to known antibiotics has been observed . Moreover, it has been shown to structurally differ from other known antimicrobials produced by Serratia strains . The production of the antibiotic has been shown to provide a selective advantage to S. plymuthica RVH1 in competition with other bacteria, e.g. in mixed biofilms .
Here, we provide evidence that the antimicrobial activity of S. plymuthica RVH1 stems from the production of the broad spectrum polyamino antibiotics identified as zeamine and zeamine II ,  (Figure 1A, 1C) and a previously uncharacterized analogue designated zeamine I (Figure 1B). Also, we report the cloning, sequencing and analysis of the gene cluster responsible for zeamine biosynthesis and propose that an unprecedented interaction between a hybrid PKS/NRPS and an iterative FAS/PKS system is responsible for the biosynthesis of the zeamine antibiotics.
Isolation and Structure Determination of the Antibiotic Compounds from S. plymuthica RVH1
During a screening of 68 Gram-negative strains, isolated from an industrial food processing line, broad spectrum antimicrobial activity was detected in one of the strongest biofilm-forming isolates designated Serratia plymuthica RVH1 . The optimal growth temperature of the strain is 30°C whereas fermentation experiments showed that antibiotic production is highest at 16°C (data not shown). Since the production of the antimicrobial compound in RVH1 is quorum sensing-regulated , fermentation cultures were supplemented with synthetic N-(3-oxo-hexanoyl)-C6-homoserine lactones which made it possible to achieve a yield of 20 mg/l of over 90% pure compounds. Using the S. aureus lawn plate assay, crude extracts were purified by silica gel chromatography leading to the isolation of the antimicrobial compounds as a yellow solid. LC/MS analysis of the purified compounds revealed two overlapping peaks (retention times: 4.05 and 4.10) with m/z of 654, and 797 and 841 respectively (Figure S1). Since the compound with m/z = 654 eluted faster than the compounds with m/z = 797 and 841, it was possible to separate the latter two compounds from the first. High-resolution mass spectrometry analysis (HRMS) (Figure S2) of these two compounds showed molecular peaks at [M+H]/e = 841.81913 and [M+H]/e = 797.79272 corresponding to molecular formulae of C49H104N6O4 and C47H100N6O3 respectively (<1 ppm error). The first molecular formula is identical to that of zeamine, isolated from Dickeya zeae DZ1 (Figure 1A) , . 1H-NMR and 13C-NMR data, combined with 1H,13C – HSQC, 1H,13C – HSQC-TOCSY, 1H,13C – HMBC and DQFCOSY analysis (Figures S3, S4, S5, S6, S7) confirmed that the structure is identical to zeamine which means that the compound with m/z = 654 must be zeamine II ([M+H]/e = 654, 6993) (Figure 1C) , . The compound with a molecular peak at [M+H]/e = 797.79272 represents an as yet uncharacterized compound which resembles zeamine and which we therefore designate as zeamine I (Figure 1B). MS-MS (Figure S8) combined with detailed NMR analysis revealed that the conserved polyamino part of zeamine I is connected via an amide bond to a 4-amino-3-hydroxy-5-methylhexanoic acid moiety instead of 6-amino-3,5-dihydroxy-7-methyloctanoic acid in the case of zeamine. This is reflected in the disappearance of a signal in the 13C-NMR spectrum at 66 ppm and one at 40 ppm. The 1H-NMR spectrum is characterized by the loss of a multiplet at 4 ppm and one at 1.6 ppm. So far, overlapping retention times have hampered the separation of zeamine from zeamine I.
Cloning, Identification and Sequencing of the Zeamine Biosynthetic Gene Cluster
To localize genes involved in zeamine biosynthesis, random Tn5 transposon mutagenesis was carried out on S. plymuthica RVH1. Tn5 insertion mutants with altered antibiotic production were identified by the soft agar halo method using S. aureus ATCC27661 as indicator strain. Of the 361 plACKO (promotorless lacZ Antimicrobial Compound Knock Out) mutants selected this way, 67 strains lost all antimicrobial activity, 127 strains showed reduced activity, and 5 strains exhibited increased antibiotic production. For nine mutants, the sites of Tn5 insertion were identified by subcloning and sequencing. In addition to genes encoding efflux systems and a putative histidine kinase, transposon insertions were found in various polyketide synthase related genes and a nonribosomal peptide synthetase gene, suggesting a hybrid PKS-NRPS assembly line (Table S1). Since PKS and NRPS genes are usually organized in gene clusters, a genomic BAC library (3840 clones, 92-fold coverage) was constructed and subsequently PCR-screened using predicted polyketide and nonribosomal peptide–related genes identified from the transposon insertion sites as template for primer design (Table S2). Based on the results, combined with restriction analysis and BAC end sequencing, the BAC clone showing the most promise of containing the full gene cluster was subcloned and sequenced by means of shotgun sequencing and primer walking.
Analysis of the Biosynthetic Gene Cluster
This combined sequencing approach resulted in a total of 170 kb of contiguous BAC-clone sequence. Several of the identified Tn5 transposon insertion sites were found to map to a 50 kb region encoding fatty acid, polyketide and nonribosomal peptide (FAS/PKS/NRPS) synthesis genes (Figure 2). With the exception of plACKO29, these transposon mutants all display a complete loss of antimicrobial activity compared to the wild type strain. The GC-content of the predicted polyketide/nonribosomal peptide operon ranges from 55% to 70% with an average of 61% which is comparable yet slightly higher than in the genus Serratia (52% to 60%) . Blast database searches revealed the presence of an almost identical, unannotated region (99% identities, 99% similarities) in Serratia plymuthica AS12 (GenBank accession number: CP002774 ) , Serratia plymuthica AS9 (CP002773)  and Serratia sp. AS13 (CP002775).
Positions of transposon insertions are marked by a dot. All but one, plACKO29, display a total loss of antibacterial activity compared to the wild type. All genes are transcribed in the same direction except for zmn5–8 and zmn22. Overlapping stop and start codons and the apparent lack of a transcriptional terminator in the short intergenic regions hints at translation coupling of these genes and suggests an operon-type organization.
Functional assignment predictions of each ORF in the FAS/PKS/NRPS region were made by comparing them with homologous sequences using BLAST (Table 1). Sequence analysis revealed 21 ORFs which include 5 PKS genes (zmn10, zmn12, zmn13, zmn14 and zmn18), 3 NRPS genes (zmn16, zmn17 and zmn19), 1 FAS/PKS gene (zmn11), 5 export related genes (zmn6, zmn7, zmn8, zmn20 and zmn21) and genes with possible post-assembly tailoring functions (Figure 2) (EMBL nucleotide sequence database accession number: HE995400). Based on these predictions, it is proposed that zmn5 and zmn22 represent both ends of the predicted cluster.
Genes Encoding FAS, PKS and NRPS
Three genes in the putative gene cluster (zmn10, zmn11 and zmn12) show homology to the pfa clusters of genes involved in the biosynthesis of PUFAs in marine bacteria such as Shewanella pneumatophori SCRC-2738 , , Moritella marina MP-1  and Photobacterium profundum SS9 . In particular, the pfaEABCD operon encodes a FAS/PKS system responsible for the production of the long chain polyunsaturated fatty acids EPA and DHA and encodes multiple conserved enzymatic domains in an operon-type organization. zmn10-12 share a common orientation, size, order, as well as domain structure (Figure 3) to pfaA, pfaC and pfaD. Zmn10 is similar to PfaA but unlike most PUFA synthases carries only one acyl carrier protein (ACP)-domain  Like PfaC, Zmn11 contains two KS domains and one FabA-like dehydratase/isomerase domain. Finally, Zmn12 contains an extra domain with putative aminotransferase activity in addition to the characteristic 2-nitropropane dioxygenase domain found in PfaD which acts as an enoyl reductase. This domain shows homology to a Class III pyridoxal phosphate (PLP)-dependent aminotransferase from Streptococcus downei F0415 (49% identities, 69% positives). No apparent pfaB ortholog is found in the vicinity of zmn10-12.
Comparison of A) part of the putative zeamine gene cluster from S. plymuthica RVH1 to the pfa clusters of genes responsible for biosynthesis of B) eicosapentaenoic acid (EPA) in S. pneumatophori SCRC-2738 and C) docosahexaenoic acid (DHA) in M. marina strain MP1(pDHA4) (C). Zeamine and pfa genes with corresponding functions and domain organization are connected by grey areas.
The involvement of a PfaA homologue (ZmsA) in zeamine biosynthesis has previously been reported by Zhou et al. . Analysis of the partial zmsA sequence data available indicates significant sequence divergence between zmn10 and zmsA. Furthermore, both homologues contain a slightly different subset of domains.
A potential associated phosphopantetheinyl transferase (PPTase) (Zmn5) is encoded in close proximity upstream of the cluster. Analysis of conserved domains points out that this Sfp-type PPTase (surfactin phosphopantetheinyl transferase) is not a PUFA-specific PPTase (PfaE-like) but rather belongs to the other group of PKS/NRPS-specific PPTases (EntD-like) , . However, all Sfp-type PPTases have been shown to enable PUFA biosynthesis . Apart from the conserved P2 (GVDIE) and P3 (KESLFKA) motifs found in both groups of PPTases, Zmn5 is characterized by the PKS/NRPS-associated 1A (KRRADY), P1a’ (DRAP) and P1b’ (GSISH) domains instead of the PUFA-related P0, P1a and P1b domains (Figure S9) , .
Immediately downstream of the pfaA, pfaC and pfaD orthologs, a putative ketoreductase and thioester reductase are encoded (Zmn13 and Zmn14 respectively). Thioester reductases are typically located at the C-terminus of PKS/NRPS assembly lines and perform reductive chain termination. Zmn14 contains the characteristic R1 domain for NAD(P)H binding as well as the five other core motifs (R2–R6) found in reductase domains (Figure S10) .
Apart from the FAS/PKS genes related to PUFA synthases, sequence analysis revealed three other multifunctional genes that encode a type I PKS module consisting of a KS, AT, KR and ACP-domain (zmn18) and two modular NRPSs composed of the characteristic condensation – adenylation – thiolation (C-A-T) domain organization in addition to a putative epimerization domain in module 3 (zmn16 and zmn17). The initiation and the first three extension modules are located on Zmn16 whereas zmn17 encodes a dimodular NRPS. Finally, zmn19 is believed to encode of a free-standing condensation domain.
All predicted KS-domains in the cluster contain the C-H-H catalytic triad  except for the second KS domain of Zmn11 which is believed to be a chain length factor . This hypothesis is supported here by the replacement of the conserved Cys residue in the active site by a Glu residue. Furthermore, the first active site His residue is replaced by Tyr (Figure S11A).
The two acyltransferase (AT) domains identified in Zmn10 and Zmn18 both contain the characteristic active site motif GXSXG (Figure S11B). Based on sequence comparison with database ATs with known substrate specificity, they are predicted to select malonyl-CoA , . This is consistent with previous feeding experiments which revealed that zeamine, except for the amino isobutyl moiety, is completely derived from acetate units .
The three putative KR-domains identified in the cluster, contain the typical Rossman fold for NADPH cofactor binding as well as the conserved SYN residues (SYK for KRZmn18) in the active site . KRZmn10 is characterized by the LDD-motif (but no apparent PxxxN motif) which correlates with B-type (“R configuration”) alcohol stereochemistry. The absence of this motif in KRZmn13 and KRZmn18 and the presence of the conserved tryptophan residue in KRZmn18 is indicative for A-type KR-domains providing a hydroxyl group of the “S” configuration (Figure S11C) , .
The first two condensation domains of Zmn16 contain a HHxxxDA and HHxxxDE core motif respectively with significant similarity to the described conserved motif HHxxxDG  present in the third C-domain of Zmn16 and the first Zmn17 C-domain. In the final Zmn17 C-domain and in Zmn19, the first histidine is replaced by a cysteine residue (CHxxxDG) and a proline residue respectively (PHxxxDG). The sequence of the epimerization domain aligns well with the sequence of the condensation domains and contains the conserved core motifs (Figure S11D).
Building block specificity of the A-domains was predicted by the support vector machine (SVM)-based NRPSpredictor toolbox and are shown in Table 2 together with the deduced specificity-conferring codes , .
All T-domains contain the highly conserved 4′-phosphopantetheinyl binding site motif LGGXS , indicating their functionality (Figure S11E). This binding motif is conserved between PCP and ACP domains which undergo the same post-translational 4′-phosphopantetheinylation of the essential serine residue. ACPZmn10 and ACPZmn18 contain a slightly different motif (LGIXS and LNGXS respectively) (Figure S11F).
Potential Tailoring, Resistance and Transport Genes
In addition to FAS, PKS and NRPS related genes, other putative gene functions possibly involved in tailoring, resistance or transport, were identified within the gene cluster.
The predicted gene product of zmn15 belongs to the carbon-nitrogen hydrolase superfamily of enzymes (EC 3.5) that catalyze condensation and hydrolysis of a wide variety of non-peptide carbon-nitrogen bonds . Members of this family include nitrilases, carbamylases, N-acyltransferases and amidases, all characterized by a α-β-β-α sandwich protein fold and classified in 13 branches according to their specific function and substrate specificity . Based on the consensus motifs flanking the conserved catalytic triad Glu-Lys-Cys, Zmn15 can be classified as a putative branch 13 nitrilase. Suffering from a lack of functional information, this enzyme class is a collection of uncharacterized carbon-nitrogen hydrolases difficult to divide into specific similarity groups.
Five putative transport genes were identified within the cluster. The gene product of zmn20 is listed as a polyamine-transporting ATPase in Serratia sp. AS12 (SerAS12_4273) whereas zmn21 encodes a putative ABC-2 type transporter. This makes them likely candidates to export zeamine and its derivatives and/or confer self-resistance to S. plymuthica RVH1.
Like Zmn20 and Zmn21, Zmn7 and Zmn8 also form a putative ABC-transporter. zmn7, which encodes a putative ABC transporter associated ATP-binding protein, shows homology to a phosphonate-transporting ATPase from Delftia sp. Cs1-4 (DelCs14_0129, 52% identities, 66% similarities) and a macrolide-specific ABC-type efflux carrier from Comamonas testosteroni S44 (CTS44_21395, 54% identities, 67% similarities) whereas Zmn8 shares similarities with ABC-transporter permeases.
The deduced product of zmn6 is a putative member of the HlyD family of secretion proteins. Together with HlyB, HlyD is responsible for the secretion of the hemolysin A toxic in uropathogenic E. coli via the type I secretion pathway .
Finally, a conserved domain search of the gene product of zmn22 revealed the presence of an acyl-aminoacyl peptidase domain, belonging to the prolyl oligopeptidase family (EC 126.96.36.199). Acylaminoacyl peptidases, or acylpeptide hydrolases are known to catalyze the cleavage of short N-acylated peptides that play a role in a variety of important biological processes .
Four genes (zmn1, 2, 4 and 9) did not possess any significant similarity to sequences in the public databases and were indicated as hypothetical proteins.
Structural analysis in the current work revealed that Serratia plymuthica strain RVH1 produces two polyamino antibiotics with the same molecular structure as zeamine and zeamine II , along with a newly discovered analogue zeamine I (Figure 1C). Zeamine I shares the conserved polyamino tail with zeamine and zeamine II but differs in the valine-derived moiety. So far, the absolute stereochemical configuration of the zeamine antibiotics remains unknown.
The combination of random transposon mutagenesis and subsequent BAC-library screening led to the identification of the zeamine biosynthesis genes. Sequence analysis revealed that the gene cluster encodes fatty acid, polyketide as well as nonribosomal peptide biosynthetic enzymes together with predicted tailoring and export proteins. Interestingly, three genes (zmn10, zmn11, zmn12) are similar to genes in the pfa gene clusters (pfaEABCD) involved in the biosynthesis of long-chain polyunsaturated fatty acids in psychrophilic bacteria prevalent in deep-sea environments , , , , (Figure 3) Recent large-scale analysis of sequenced microbial genomes has revealed a wide variety of FAS/PKS gene clusters homologous to those responsible for LCPUFA synthesis in over ten microbial phyla. These related FAS/PKS gene clusters produce various specialized long chain fatty acids consisting of more than 20 carbons in a process referred to as “secondary lipid biosynthesis” . The three zeamine genes share a common orientation, size, order, as well as domain organization with pfaA, pfaC and pfaD (Figure 3). A distinct feature of the zeamine FAS/PKS genes is the lack of a pfaB ortholog, nor is the corresponding AT-activity encoded in the pfaC ortholog (zmn11), as frequently observed in secondary lipid synthases . Also, zmn12 is the first pfaD ortholog reported to contain an aminotransferase domain. This domain belongs to the family of PLP-dependent aspartate aminotransferases (EC 188.8.131.52) and shows homology to glutamate-1-semialdehyde aminotransferases (EC 184.108.40.206). A similar PLP-dependent aminotransferase domain equally showing homology to glutamate-1-semialdehyde aminotransferases has previously been identified in MycA, a hybrid PKS/NRPS involved in the biosynthesis of the antibiotic mycosubtilin. Using glutamine as amine source, the MycA aminotransferase domain works in cis to incorporate an amine at the β-position of a growing acyl chain generating a β-aminothioester . Immediately downstream of zmn12, a predicted ketoreductase and thioester reductase are encoded which show no apparent relationship to pfa genes.
Apart from FAS/PKS genes homologous to LCPUFA biosynthesis genes, an NRPS assembly line represents a noticeable part of the putative zeamine gene cluster. The NRPS genes encode a loading module and five extension modules which are predicted to activate and incorporate 2-aminoadipic acid, phenylalanine, asparagine (2×), threonine and valine in that order. A PKS gene encoding KS, AT, ACP and KR-domains (zmn18) and a stand-alone condensation domain (zmn19) are localized immediately downstream of the NRPS genes.
Proposed Biosynthesis Model for the Zeamine Antibiotics
Bio-informatical and experimental analysis of the genes in the putative gene cluster allow us to overhaul, expand and specify the pathway for zeamine biosynthesis postulated by Wu et al.  (Figure 4). We hypothesize that a mixed FAS/PKS and a hybrid PKS/NRPS each synthesize parts of zeamine which are linked together post-assembly in the case of zeamine and zeamine I. Analogous to the PUFA synthases, the multidomain FAS/PKS-related enzymes encoded by the zeamine biosynthesis gene cluster very likely catalyse the repeated steps in the production of the polyamino chain, in which each condensation reaction is followed by a complete round of ketoreduction, dehydration/isomerization and enoyl reduction. In selected cycles however, the reductive reaction cycle is replaced with a single transamination step, possibly catalyzed by the aminotransferase domain present in Zmn12, similar to the role of the aminotransferase domain identified in MycA . In the last round of elongation, the transamination step is replaced by a ketoreduction reaction. Zmn14-catalyzed reductive chain termination would result in an aldehyde which is the substrate for a final transamination reaction. In the biosynthesis of the siderophore myxochelin B, a similar chain termination procedure is observed, in which the aldehyde aminotransferase MxcL, catalyses the exchange of the amino group . In the zeamine gene cluster, the aminotransferase domain present in Zmn12 is believed to fulfill this role.
Biosynthesis of the zeamine antibiotics can be envisaged as a two-part process. Zmn10-14 likely associate to form a multienzyme complex responsible for the coordinated formation of the polyamino tail in a process closely related to secondary lipid biosynthesis. On the other hand, the valine-derived moiety originates from the hybrid NRPS/PKS formed by Zmn16-18. In the case of zeamine and zeamine I, both parts are connected post-assembly by means of chain-terminating amide bond formation, catalyzed by Zmn19. In the course of zeamine biosynthesis, a nonribosomal pentapeptide sequence is likely cleaved off, resulting in the observed zeamine structures.
The valine-derived moiety is proposed to originate from the hybrid PKS/NRPS assembly line formed by Zmn16-18. We predict that the amino acid valine, incorporated by the C-terminal NRPS module, acts as a primer for two additional rounds of PKS chain elongation with KR-mediated reduction of the β-ketogroup, catalyzed by Zmn18. This would generate 6-amino-3,5-dihydroxy-7-methyloctanoate-S-ACP which is subsequently coupled to the polyamino chain by formation of an amide bond. A likely candidate catalyzing this linkage is the free-standing condensation domain (Zmn19) encoded directly downstream of the PKS module. Condensation domain-catalyzed amide bond formation as alternative chain termination method has been reported in the biosynthesis of the siderophore vibriobactin and has been predicted to occur in several other NRPS including those of cyclosporine, enniatin, HC-toxin, PF1022A and vibriobactin. However, while chain termination of the latter examples involves intramolecular amide bond formation and subsequent cyclization, VibH has the ability to form an amide bond between a soluble substrate acting as the acceptor nucleophile and a carrier protein-bound acyl donor , .
Taking into account the antimicrobial activity of zeamine II, the connection of the polyamino chain to the 6-amino-3,5-dihydroxy-7-methyloctanoate does not seem to be crucial for its function as antibiotic . The biosynthesis of zeamine could be envisaged as being the result of aberrant stuttering of Zmn19, a process previously observed in module 4 of 6-deoxyerythronolide B synthase and module 5 of the epothilone PKS , . This would imply that iterative processing of the intermediate is kinetically more favorable than interprotein transfer to the free-standing downstream condensation domain. As such, Zeamine I can be regarded as being the ‘intended’ product of the biosynthesis cluster.
Remarkably, the products of the first five NRPS modules are not present in the final zeamine structure even though active site residues suggest assembly line functionality. Furthermore, the reduced antimicrobial activity of transposon mutant plACKO 29 (Figure 2) indicates a likely role of Zmn16 in the biosynthesis of zeamine although possible polar effects cannot be excluded in this respect. One could speculate that the zeamine antibiotics are produced as prodrugs (Figure 1D) that undergo post-assembly activation in which five amino acids are cleaved off. A similar process has been described for the xenocoumacin antibiotics. Here, the pre-xenocoumacins are trimmed periplasmatically upon secretion at an acylated D-Asn residue by XcnG, a membrane-bound peptidase, thereby turning inactive precursor molecules into the bioactive compound. Furthermore, there are indications that this process is conserved in several other biosynthetic gene clusters . In the putative zeamine gene cluster, no XcnG homolog is found, however similar function may be exerted by Zmn22 predicted αβ hydrolase fold protein characterized by a predicted dipeptidyl aminopeptidase/acyl-aminoacyl peptidase domain. Ayg1p, a polyketide biosynthesis tailoring enzyme with a similar acyl-aminoacyl peptidase domain has been shown to hydrolytically cleave off acetoacetic acid from a heptaketide thereby giving rise to 1,3,6,8-tetrahydroxynaphthalene, the precursor of the fungal virulence factor melanin .
Alternatively, Zmn15, strategically encoded between the two biosynthesis machineries, belongs to the carbon-nitrogen hydrolase superfamily involved in amide-hydrolysis and condensation. Although some representatives of the family act on polypeptide side chains or carry out condensation reactions on the N-termini of proteins, so far no known members have been reported to cleave peptide bonds, making this enzyme an unlikely candidate for this cleavage reaction .
Apart from antibiotic properties, potent phytotoxic effects have also been attributed to the zeamine antibiotics in the phytopathogenic strain Dickeya zeae DZ1. More specifically, they have been shown to inhibit rice seed root and shoot germination . Now, with Serratia plymuthica strain RVH1 and the rapeseed plant-associated Serratia plymuthica AS12 , AS13 and AS9 , four additional phylogenetically closely related, rhizosphere associated bacterial strains have been identified that harbor the zeamine biosynthesis genes. Therefore, it is plausible that horizontal gene transfer contributed to the spread of the zeamine-producing potential among plant-associated soil bacteria.
In this study, we have identified the zeamine biosynthetic gene cluster from Serratia plymuthica RVH1, responsible for the production of three zeamine-related broad spectrum antibiotics. Analysis of the putative biosynthesis genes allowed us to propose a unique synergy between secondary lipid and secondary metabolite biosynthesis in which a mixed FAS/PKS, analogous to the PUFA synthases, and a hybrid NRPS/PKS each assemble parts of the zeamine natural products that are linked together post-assembly in the case of zeamine and zeamine I. Furthermore, indications are there that the compounds are initially synthesized as prodrugs that undergo an activation step in which a nonribosomal peptide sequence is cleaved off. The mechanism by which amino groups are incorporated into the zeamine backbone could offer prospects for the rational design of new bioactive compounds with increased functionality.
Materials and Methods
Bacterial Strains and Plasmids
S. plymuthica RVH1 served as the wild-type zeamine producer . Escherichia coli Transformax EC100 and E. coli DH5α were used for BAC-library construction and routine subcloning purposes whereas E. coli S17-1  was used for conjugative transfer of DNA to S. plymuthica RVH1. Staphylococcus aureus ATCC27661 was used in bioassays to monitor zeamine production. E. coli sdiA::Km , a kanamycin resistant strain unable to react on N-acyl-L-homoserine lactone signals, functioned as a test strain to identify random transposon mutants of S. plymuthica RVH1 with an altered production of zeamine. Bacterial strains and plasmids used during this work are listed in Table S3.
Growth and Culture Conditions
S. plymuthica RVH1 was grown at 30°C in Lysogeny Broth (LB) for liquid cultivation and on LB agar (LB broth supplemented with 1.5% w/v agar) for solid cultivation. E. coli and S. aureus strains were grown at 37°C in LB broth and on LB-agar. When appropriate, media were supplemented with antibiotics at the following concentrations: kanamycin (50 µg/ml), ampicillin (100 µg/ml), chloramphenicol (50 µg/ml).
Production and Isolation of Zeamine
S. plymuthica RVH1 was grown in liquid LB-medium, supplemented with 5 µM synthetic N-(3-oxo-hexanoyl)-C6-homoserine lactone (Sigma), at 16°C for 48 h on a rotary shaker (80 rpm). Cells were removed by centrifugation (13000 rpm, 20 min) after which the supernatant was adjusted to pH 10.0 with 1 M NaOH and extracted with n-butanol twice. After acidification to pH 3.0 with acetic acid, the extract was subjected to silica gel column chromatography (20×250 mm, MeOH/CH2Cl2 [1/8 to 1/4 to 1/2 to 1/1 to 1/0] followed by MeOH/H2O [8/1 to 4/1 to 4/1+200 mM NaCl]). Zeamine antibiotics eluted when NaCl was added to the solvent. LC/MS analysis was performed on a Finnigan Surveyor HPLC system using a C18 reverse phase column (Thermo Scientific HyPURITY 5 µm C18 30×2.1 mm) and a solvent system consisting of MeOH in H2O supplemented with 0,1% formic acid (0–2 min 20%–95%, 2–3 min 95%, 3–4 min 95%-20%, 4–8 min 20%). Peaks were monitored with a mass detector (Finnigan LCQ Advantage Max ESI/MS).
Structure Analysis of Zeamine
MS analysis was performed on a quadrupole/orthogonal acceleration time-of-flight mass spectrometer (Synapt G2 HDMS, Waters, Milford, MA). The instrument was operated at a resolution of 15000. Leucine enkephalin was used as reference for accurate mass measurements. The samples were dissolved in acetonitrile:water (1∶1) and infused at a flow rate of 5 uL/min. The capillary voltage was set to 3 kV. For NMR spectroscopy, samples were dissolved in 500 µl D2O with a final pH of 4.4. All spectra were recorded at 25°C on a Bruker Avance II 600 equipped with a 5 mm TCI HCN Z gradient cryoprobe. The Bruker Topspin 2.1 software was used to process the spectra. A 2D DQFCOSY  spectrum was measured with a spectral width of 6600 Hz in both dimensions using 64 scans and 512/2048 complex data points in t1 and t2, respectively. Heteronuclear correlation spectra were recorded using 64 scans and 1024/1024 complex data points and 200/11 ppm spectral widths in t1 and t2, respectively. The pulse sequence for the 2D gradient enhanced 1H-13C HSQC was as described by Sattler et al. . The delays Δ1 and Δ2 in the inept transfer were set to 1.67 ms (1/4JCH). The 2D HSQC–TOCSY consisted of an HSQC building block  (Δ1 = Δ2 = 1.67 ms) followed by a clean MLEV17 TOCSY transfer step  during a mixing time of 60 ms prior to detection. Decoupling during the acquisition of the HSQC and HSQC–TOCSY spectra was achieved by using the garp sequence . Sine-bell shaped gradients were applied along the z-axis during the sequences to obtain coherence selection and sensitivity enhancement. Prior to Fourier transformation, a squared sine-bell function was applied in both dimensions of 2D spectra.
Bioassay for Zeamine Production
Zeamine production was monitored by bioassay. An LB agar plate was overlaid with a lawn of stationary phase S. aureus ATCC27661 cells, diluted to OD600nm = 0,03 (5 µ/ml) and 10 µl samples of 18 h grown S. plymuthica RVH1 cultures were spotted on top. After incubation for 48 h at 16°C and 24 h at 30°C, the size of the inhibition zones was evaluated.
‘Soft Agar Halo Method
After overnight incubation of indicator strains at 30°C in LB-medium, 50 µl of the stationary cultures was added to a petridish and mixed with LB soft agar (LB broth supplemented with 0.7% w/v agar). 5 µl of an overnight stationary culture of S. plymuthica RVH1 was spotted on top. To allow the growth of strain RVH1 before that of the test strains, the plates were incubated for 2 days at 16°C before further incubation at 30°C for 24 h. This way, a clear halo surrounded by a turbid lawn was visible around the RVH1 spot in case of a sensitive test strain.
Random Transposon Mutagenesis and Identification of Genes Involved in Zeamine Biosynthesis
Random transposon mutagenesis was performed by transferring the suicide plasmid pUT mini-Tn5 lacZ1 from E. coli S17-1 λpir to an S. plymuthica RVH1 splI insertion mutant which is deficient in quorum sensing , . Using the soft agar halo method described above, mutant strains with reduced or absent growth inhibition zone were identified on LB soft agar using a kanamycin resistant test strain unable to react on signal molecules (E. coli sdiA::Km) and kanamycin as selection marker. Synthetic N-(3-oxo-hexanoyl)-C6-homoserine lactone (OHHL) signal was added to the plates to ensure maximal antibiotic production. To identify the site of transposon insertion, sonicated (Vibracell Sonics and Materials Inc.) genomic DNA of the selected mutants was ligated into SmaI digested pUC18 vectors and transformed into E. coli DH5α, followed by selection on LB-agar supplemented with ampicillin. The genes flanking the transposon insertions were identified by sequencing using the pUC24bis, pUC47 and KmKleckner primers (Table S2).
Genomic Library Construction and Screening
BAC-library construction of S. plymuthica strain RVH1 was performed as described by Osoegawa et al. . High-molecular-weight DNA was embedded in plugs, partially digested with HindIII and size selected by pulsed-field gel electrophoresis. Subsequent cloning in pIndigoBAC5 (Epicentre) vectors and transformation to E. coli Transformax™ EC100™ were performed as described by the manufacturer. This yielded a library of 3840 clones with an average insert size of 130 kb resulting in 92-fold coverage. Using the identified sequences flanking the transposon insertion sites as template for primer design, the library was screened by PCR to localize the zeamine biosynthesis genes. A HindIII restriction digest and BAC end sequencing using pINDIGO BAC-5 vector primers aided in selecting the best BAC clone for further analysis. For an overview of the oligonucleotide primers used in this study, see Table S2.
DNA Sequencing and Analysis
The selected BAC-clone was subcloned using pUC19 vectors and sequenced by a combination of primer walking and shotgun sequencing. Sequencing reactions were run using Big Dye Terminator mix (Applied Biosystems), purified and analyzed on an ABI 3130 genetic analyzer. Sequence assembly into contigs was performed using Sequencher 4.8 software (Gene Codes Corporation, USA). ORF predictions were carried out by combining comparative genomics approaches (tBLASTx) and the ORF Finder (NCBI) and Genemark.hmm algorithms .
LC-MS analysis of purified extract of S. plymuthica RVH1 containing zeamine, zeamine I and zeamine II.
HRMS analysis of zeamine and zeamine I.
1H-NMR spectrum of zeamine and zeamine I.
13C-NMR spectrum of zeamine and zeamine I.
1H,13C – HSQC spectrum of zeamine and zeamine I.
1H,13C – HSQC-TOCSY spectrum of zeamine and zeamine I overlayed with 1H,13C – HSQC.
DQFCOSY spectrum of zeamine and zeamine I.
High resolution MS-MS spectrum of zeamine I.
Amino acid sequence of zmn5 encoding a phosphopantetheinyl transferase (PPTase). Conserved motifs are underlined and indicate that Zmn5 is a PKS/NRPS-related PPTase (EntD-like) rather than PUFA-specific.
Deduced amino acid sequence of zmn14 encoding a thioester reductase domain. The R1 domain, involved in NAD(P)H binding as well as the five other core motifs (R2–R5) are underlined.
Multiple alignment and active site sequences (highlighted) of polyketide and nonribosomal peptide domains. A/Multiple sequence alignment of KS-domains. The catalytic triad is highlighted in green, conserved motifs in yellow. B/Multiple sequence alignment of AT-domains. The active site motif is highlighted in red, substrate specificity determinants in yellow. C/Multiple sequence alignment of KR-domains. The catalytic triad is marked in green, the Rossman fold involved in NADPH cofactor binding is highlighted in yellow, the residues that indicate stereospecifity in red. D/Multiple sequence alignment of C- and E-domains. The signature HHxxxDG motif is highlighted in green whereas epimerization domain-specific conserved motifs are highlighted in yellow. E/Multiple sequence alignment of PCP-domains. The conserved 4′-phosphopantetheinyl binding site motif LGGXS is highlighted in yellow. The essential serine residue in green. F/Multiple sequence alignment of ACP-domains. The conserved 4′-phosphopantetheinyl binding site motif is highlighted in yellow. The essential serine in green.
Sites of plACKO transposon insertion sites.
Oligonucleotide primers used in this study. Restriction sites are underlined.
Conceived and designed the experiments: JM WM LJG PM RVH EL JR AA CM RL. Performed the experiments: JM WM PM RVH BU CL EL JR. Analyzed the data: JM WM LJG PM RVH EL JR. Contributed reagents/materials/analysis tools: EL JR PH AA CM RL. Wrote the paper: JM.
- 1. Staunton J, Weissman KJ (2001) Polyketide biosynthesis: a millennium review. Nat Prod Rep 18: 380–416.
- 2. Marahiel MA (2009) Working outside the protein-synthesis rules: insights into nonribosomal peptide synthesis. J Pept Sci 15: 799–807.
- 3. Hertweck C, Luzhetskyy A, Rebets Y, Bechthold A (2007) Type II polyketide synthases: gaining a deeper insight into enzymatic teamwork. Nat Prod Rep 24: 162–190.
- 4. Liu F, Garneau S, Walsh CT (2004) Hybrid nonribosomal peptide-polyketide interfaces in epothilone biosynthesis. Chem Biol 11: 1533–1542.
- 5. Park SR, Yoo YJ, Ban YH, Yoon YJ (2010) Biosynthesis of rapamycin and its regulation: past achievements and recent progress. J Antibiot 63: 434–441.
- 6. Miller DA, Luo L, Hillson N, Keating TA, Walsh CT (2002) Yersiniabactin synthetase: a four-protein assembly line producing the nonribosomal peptide/polyketide hybrid siderophore of Yersinia pestis. Chem Biol 9: 333–344.
- 7. Mattheus W, Gao LJ, Herdewijn P, Landuyt B, Verhaegen J, et al. (2010) Isolation and purification of a new kalimantacin/batumin-related polyketide antibiotic and elucidation of its biosynthesis gene cluster. Chem Biol 17: 149–159.
- 8. Shulse CN, Allen EE (2011) Widespread occurrence of secondary lipid biosynthesis potential in microbial lineages. PLoS One 6: e20146.
- 9. Okuyama H, Orikasa Y, Nishida T, Morita N (2007) Bacterial genes responsible for the biosynthesis of eicosapentaenoic and docosahexaenoic acids and their heterologous expression. Appl Environ Microbiol 73: 665–670.
- 10. Van Houdt R, Aertsen A, Jansen A, Quintana AL, Michiels CW (2004) Biofilm formation and cell-to-cell signaling in Gram-negative bacteria isolated from a food processing environment. J Appl Microbiol 96: 177–184.
- 11. Van Houdt R, Moons P, Aertsen A, Jansen A, Vanoirbeek K, et al. (2007) Characterization of a luxI/luxR-type quorum sensing system and N-acyl-homoserine lactone-dependent regulation of exo-enzyme and antibacterial component production in Serratia plymuthica RVH1. Res Microbiol 158: 150–158.
- 12. Van Houdt R, Moons P, Buj MH, Michiels CW (2006) N-acyl-L-homoserine lactone quorum sensing controls butanediol fermentation in Serratia plymuthica RVH1 and Serratia marcescens MG1. J Bacteriol 188: 4570–4572.
- 13. Moons P, Van Houdt R, Vivijs B, Michiels CW, Aertsen A (2011) Integrated regulation of acetoin fermentation by quorum sensing and pH in Serratia plymuthica RVH1. Appl Environ Microbiol 77: 3422–3427.
- 14. Moons P (2008) Influence of the Serratia plymuthica RVH1 quorum sensing system on biofilm formation and interspecies interactions. PhD Thesis, Katholieke Universiteit Leuven (Belgium).
- 15. Van Houdt R, Moons P, Jansen A, Vanoirbeek K, Michiels CW (2005) Genotypic and phenotypic characterization of a biofilm-forming Serratia plymuthica isolate from a raw vegetable processing line. FEMS Microbiol Lett 246: 265–272.
- 16. Moons P, Van Houdt R, Aertsen A, Vanoirbeek K, Engelborghs Y, et al. (2006) Role of quorum sensing and antimicrobial component production by Serratia plymuthica in formation of biofilms, including mixed biofilms with Escherichia coli. Appl Environ Microbiol 72: 7294–7300.
- 17. Wu J, Zhang H-B, Xu J-L, Cox RJ, Simpson TJ, et al. (2010) 13C labeling reveals multiple amination reactions in the biosynthesis of a novel polyketide polyamine antibiotic zeamine from Dickeya zeae. Chem Comm 46: 333–335.
- 18. Zhou J, Zhang H, Wu J, Liu Q, Xi P, et al. (2011) A novel multidomain polyketide synthase is essential for zeamine production and the virulence of Dickeya zeae. Mol Plant-Microbe Interact 24: 1156–1164.
- 19. Mahlen SD (2011) Serratia infections, from military experiments to current practice. Clin Microbiol Rev 24: 755–791.
- 20. Neupane S, Finlay RD, Alström S, Goodwin L, Kyrpides NC, et al. (2012) Complete genome sequence of Serratia plymuthica strain AS12. Stand Genomic Sci 6: 165–173.
- 21. Neupane S, Högberg N, Alström S, Lucas S, Han J, et al. (2012) Complete genome sequence of the rapeseed plant-growth promoting Serratia plymuthica strain AS9. Stand Genomic Sci 6: 54–62.
- 22. Yazawa K (1996) Production of eicosapentaenoic acid from marine bacteria. Lipids S297–S300.
- 23. Hirota K, Nodasaka Y, Orikasa Y, Okuyama H, Yumoto I (2005) Shewanella pneumatophori sp. nov., an eicosapentaenoic acid-producing marine bacterium isolated from the intestines of Pacific mackerel (Pneumatophorus japonicus). Int J Syst Evol Microbiol 55: 2355–2359.
- 24. Tanaka M, Ueno A, Kawasaki K, Yumoto I, Ohgiya S, et al. (1999) Isolation of clustered genes that are notably homologous to the eicosapentaenoic acid biosynthesis gene cluster from the docosahexaenoic acid-producing bacterium Vibrio marinus strain MP-1. Biotechnol Lett 21: 939–945.
- 25. Allen EE, Bartlett DH (2002) Structure and regulation of the omega-3 polyunsaturated fatty acid synthase genes from the deep-sea bacterium Photobacterium profundum strain SS9. Microbiology 148: 1903–1913.
- 26. Orikasa Y, Nishida T, Hase A, Watanabe K, Morita N, et al. (2006) A phosphopantetheinyl transferase gene essential for biosynthesis of n-3 polyunsaturated fatty acids from Moritella marina strain MP-1. FEBS Lett 580: 4423–4429.
- 27. Sugihara S, Orikasa Y, Okuyama H (2008) An EntD-like phosphopantetheinyl transferase gene from Photobacterium profundum SS9 complements pfa genes of Moritella marina strain MP-1 involved in biosynthesis of docosahexaenoic acid. Biotechnol Lett 30: 411–414.
- 28. Sugihara S, Orikasa Y, Okuyama H (2010) The Escherichia coli highly expressed entD gene complements the pfaE deficiency in a pfa gene clone responsible for the biosynthesis of long-chain n-3 polyunsaturated fatty acids. FEMS Microbiol Lett 307: 207–211.
- 29. Gaitatzis N, Kunze B, Müller R (2001) In vitro reconstitution of the myxochelin biosynthetic machinery of Stigmatella aurantiaca Sg a15: Biochemical characterization of a reductive release mechanism from nonribosomal peptide synthetases. Proc Nat Acad Sci USA 98: 11136–11141.
- 30. Aparicio JF, Molnár I, Schwecke T, König A, Haydock SF, et al. (1996) Organization of the biosynthetic gene cluster for rapamycin in Streptomyces hygroscopicus: analysis of the enzymatic domains in the modular polyketide synthase. Gene 169: 9–16.
- 31. Haydock SF, Aparicio JF, Molnar I, Schwecke T, Khaw LE, et al. (1995) Divergent sequence motifs correlated with the substrate specificity of (methyl)malonyl-CoA:acyl carrier protein transacylase domains in modular polyketide synthases. FEBS Lett 374: 246–248.
- 32. Yadav G, Gokhale RS, Mohanty D (2003) Computational approach for prediction of domain organization and substrate specificity of modular polyketide synthases. J Mol Biol 328: 335–363.
- 33. Reid R, Piagentini M, Rodriguez E, Ashley G, Viswanathan N, et al. (2003) A model of structure and catalysis for ketoreductase domains in modular polyketide synthases. Biochemistry 42: 72–79.
- 34. Caffrey P (2003) Conserved amino acid residues correlating with ketoreductase stereospecificity in modular polyketide synthases. ChemBioChem 4: 649–662.
- 35. De Crécy-Lagard V, Marlière P, Saurin W (1995) Multienzymatic nonribosomal peptide biosynthesis: identification of the functional domains catalysing peptide elongation and epimerisation. C R Acad Sci Ser III 318: 927–936.
- 36. Rausch C, Weber T, Kohlbacher O, Wohlleben W, Huson DH (2005) Specificity prediction of adenylation domains in nonribosomal peptide synthetases (NRPS) using transductive support vector machines (TSVMs). Nucleic Acids Res 33: 5799–5808.
- 37. Röttig M, Medema MH, Blin K, Weber T, Rausch C, et al. (2011) NRPSpredictor2 - a web server for predicting NRPS adenylation domain specificity. Nucleic Acids Res 39: W362–367.
- 38. Schlumbohm W, Stein T, Ullrich C, Vater J, Krause M, et al. (1991) An active serine is involved in covalent substrate amino acid binding at each reaction center of gramicidin S synthetase. J Biol Chem 266: 23135–23141.
- 39. Pace HC, Brenner C (2001) The nitrilase superfamily: classification, structure and function. Genome Biol 2: reviews0001.1–reviews0001.9.
- 40. Brenner C (2002) Catalysis in the nitrilase superfamily. Curr Opin Struct Biol 12: 775–782.
- 41. Blight MA, Chervaux C, Holland IB (1994) Protein secretion pathway in Escherichia coli. Curr Opin Biotechnol 5: 468–474.
- 42. Barrett AJ, Rawlings ND (1992) Oligopeptidases, and the emergence of the prolyl oligopeptidase family. Biol Chem Hoppe Seyler 373: 353–360.
- 43. Metz JG, Roessler P, Facciotti D, Levering C, Dittrich F, et al. (2001) Production of polyunsaturated fatty acids by polyketide synthases in both prokaryotes and eukaryotes. Science 293: 290–293.
- 44. Duitman EH, Hamoen LW, Rembold M, Venema G, Seitz H, et al. (1999) The mycosubtilin synthetase of Bacillus subtilis ATCC6633: a multifunctional hybrid between a peptide synthetase, an amino transferase, and a fatty acid synthase. Proc Nat Acad Sci USA 96: 13294–13299.
- 45. Aron ZD, Dorrestein PC, Blackhall JR, Kelleher NL, Walsh CT (2005) Characterization of a new tailoring domain in polyketide biogenesis: the amine transferase domain of MycA in the mycosubtilin gene cluster. J Am Chem Soc 127: 14986–14987.
- 46. Silakowski B, Kunze B, Nordsiek G, Blöcker H, Höfle G, et al. (2000) The myxochelin iron transport regulon of the myxobacterium Stigmatella aurantiaca Sg a15. Eur J Biochem 267: 6476–6485.
- 47. Keating TA, Marshall CG, Walsh CT (2000) Vibriobactin biosynthesis in Vibrio cholerae: VibH is an amide synthase homologous to nonribosomal peptide synthetase condensation domains. Biochemistry 39: 15513–15521.
- 48. Keating TA, Marshall CG, Walsh CT (2000) Reconstitution and characterization of the Vibrio cholerae vibriobactin synthetase from VibB, VibE, VibF, and VibH. Biochemistry 39: 15520–15530.
- 49. Wilkinson B, Foster G, Rudd BA, Taylor NL, Blackaby AP, et al. (2000) Novel octaketide macrolides related to 6-deoxyerythronolide B provide evidence for iterative operation of the erythromycin polyketide synthase. Chem Biol 7: 111–117.
- 50. Moss SJ, Martin CJ, Wilkinson B (2004) Loss of co-linearity by modular polyketide synthases: a mechanism for the evolution of chemical diversity. Nat Prod Rep 21: 575–593.
- 51. Reimer D, Pos KM, Thines M, Grün P, Bode HB (2011) A natural prodrug activation mechanism in nonribosomal peptide synthesis. Nat Chem Biol 7: 888–890.
- 52. Fujii I, Yasuoka Y, Tsai HF, Chang YC, Kwon-Chung KJ, et al. (2004) Hydrolytic polyketide shortening by ayg1p, a novel enzyme involved in fungal melaninbiosynthesis. J Biol Chem 279: 44613–44620.
- 53. Simon R, Priefer U, Pühler A (1983) A broad host range mobilization system for in vivo genetic engineering: transposon mutagenesis in Gram-negative bacteria. Nat Biotechnol 1: 784–791.
- 54. Van Houdt R, Aertsen A, Moons P, Vanoirbeek K, Michiels C (2006) N-acyl-L homoserine lactone signal interception by Escherichia coli. FEMS Microbiol Lett 256: 83–89.
- 55. Davis AL, Laue ED, Keeler J, Moskau D, Lohman J (1991) Absortion-mode 2-dimensional NMR-spectra recorded using pulsed field gradients. J Magn Reson 94: 637–644.
- 56. Sattler M, Schwendinger MG, Schleucher J, Griesinger C (1995) Novel strategies of sensitivity enhancement in heteronuclear multi-dimensional NMR experiments employing pulsed field gradients. J Biomol NMR 6: 11–22.
- 57. Davis AL, Keeler J, Laue ED, Moskau D (1992) Experiments for recording pure –absorption heteronuclear correlation spectra using pulsed field gradients. J Magn Reson 98: 207–216.
- 58. Griesinger C, Otting G, Wüthrich K, Ernst RR (1988) Clean TOCSY for 1H spin system identification in macromolecules. J Am Chem Soc 110: 7870–7872.
- 59. Shaka AJ, Barker PB, Freeman R (1985) Computer-optimized decoupling scheme for wideband applications and low-level operation. J Magn Reson 64: 574.
- 60. De Lorenzo V, Timmis K (1994) Analysis and construction of stable phenotypes in gram-negative bacteria with Tn5- and Tn10- derived minitransposons. Methods Enzymol 235: 386–405.
- 61. Osoegawa K, Woon PY, Zhao B, Frengen E, Tateno M, et al. (1998) An improved method for construction of bacterial artificial chromosome libraries. Genomics 52: 1–8.
- 62. Borodovsky M, Mills R, Besemer J, Lomsadze A (2003) Prokaryotic gene prediction using GeneMark and GeneMark.hmm. Curr Protoc Bioinformatics Chapter 4, Unit 4.5.