• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of jbacterPermissionsJournals.ASM.orgJournalJB ArticleJournal InfoAuthorsReviewers
J Bacteriol. Jun 2006; 188(11): 4024–4036.
PMCID: PMC1482889

Structural and Functional Characterization of Three Polyketide Synthase Gene Clusters in Bacillus amyloliquefaciens FZB 42


Although bacterial polyketides are of considerable biomedical interest, the molecular biology of polyketide biosynthesis in Bacillus spp., one of the richest bacterial sources of bioactive natural products, remains largely unexplored. Here we assign for the first time complete polyketide synthase (PKS) gene clusters to Bacillus antibiotics. Three giant modular PKS systems of the trans-acyltransferase type were identified in Bacillus amyloliquefaciens FZB 42. One of them, pks1, is an ortholog of the pksX operon with a previously unknown function in the sequenced model strain Bacillus subtilis 168, while the pks2 and pks3 clusters are novel gene clusters. Cassette mutagenesis combined with advanced mass spectrometric techniques such as matrix-assisted laser desorption ionization-time of flight mass spectrometry and liquid chromatography-electrospray ionization mass spectrometry revealed that the pks1 (bae) and pks3 (dif) gene clusters encode the biosynthesis of the polyene antibiotics bacillaene and difficidin or oxydifficidin, respectively. In addition, B. subtilis OKB105 (pheA sfp0), a transformant of the B. subtilis 168 derivative JH642, was shown to produce bacillaene, demonstrating that the pksX gene cluster directs the synthesis of that polyketide.

Environmental Bacillus amyloliquefaciens strain FZB 42 is distinguished from the domesticated model organism Bacillus subtilis 168 (23) by several features important for rhizosphere competence particularly by its abilities to suppress competitive organisms present in the plant rhizosphere (17, 21) and to promote plant growth (16). In a previous contribution (20), we have reported that B. amyloliquefaciens FZB 42 is a producer of three families of lipopeptides, surfactins, bacillomycins D, and fengycins, which are well-known secondary metabolites with mainly antifungal activity. They are also produced by numerous B. subtilis strains (48). Furthermore, three giant gene clusters containing genes with homology to polyketide synthase (PKS) genes of modular organization were identified but not assigned functional roles. Mutants of FZB 42 deficient in the synthesis of cyclic lipopeptides were unable to suppress phytopathogenic fungi but still retained their antibacterial potency.

Polyketides belong to a large family of secondary metabolites that include many bioactive compounds with antibacterial, immunosuppressive, antitumor, or other physiologically relevant bioactivities. Their biosynthesis is accomplished by stepwise decarboxylative Claisen condensations between the extender unit and the growing polyketide chain, generating enzyme-bound β-ketoacyl intermediates. Before a subsequent round of chain extension, a variable set of modifying enzymes can locally introduce structural variety. Similar to the nonribosomal synthesis of peptides, the PKS multienzyme system uses acyl carrier proteins (ACPs) that are posttranslationally modified with the 4′-phosphopantetheine prosthetic group to channel the growing polyketide intermediate during elongation processes (3). Type I PKSs are modularly organized giant synthases, each module of which usually contains a β-ketoacyl synthase (KS), an acyltransferase (AT), and the ACP as essential and basic domains that may be complemented by a variable set of additional domains. The order of the modules dictates the sequence of biosynthetic events, and the generally observed colinearity between PKS structure and biosynthetic steps was shown to permit combinatorial manipulation of type I PKSs in order to generate novel compounds (7).

Although the model strain B. subtilis 168 has been shown to contain a large PKS cluster designated the pksX system (1), it is unable to synthesize polyketides because of a mutation in the sfp gene (27-29) encoding 4′-phosphopantetheine transferase. Sfp not only phosphopantetheinylates the peptidyl carrier proteins of surfactin and other lipopeptides but displays broad substrate specificity for other carrier proteins as well, including ACPs of PKSs (27). Production of polyketide-like compounds with antibacterial activity by wild-type isolates of B. subtilis has been described previously (15). The polyene antibiotics difficidin and oxydifficidin are highly unsaturated 22-member macrolides with a rare phosphate group (51). Another antibiotic, bacillaene, was demonstrated to be a conjugated hexaene with a linear structure (33), but its chemical structure is still unknown. Since the discovery of the pksX system in the genome of B. subtilis 168, its function has been subject to ample speculation, with difficidin commonly being discussed as its metabolic product (42). So far, no genetic structure could be functionally assigned to the biosynthesis of this or any other polyketide in bacilli, except for the zwittermicin system, in which a part of the genes contributing to its biosynthesis have recently been uncovered (12).

Here we characterize the giant modular PKS gene clusters pks1 and pks3, which are responsible for the biosynthesis of the antibiotics bacillaene and difficidin or oxydifficidin in B. amyloliquefaciens. Another gene cluster, pks2, is involved in the synthesis of an additional but not yet characterized polyketide that displayed weak antibacterial activity in our assays. Notably, all three PKS systems exhibit the unusual trans-AT architecture described recently; i.e., all PKS modules lack an AT domain and are complemented by ATs encoded on isolated genes (34). The clusters display a high degree of homology, suggesting that they might have evolved from a single ancestral operon.


Strains, growth conditions, and DNA transformation.

B. amyloliquefaciens strain FZB 42 was described previously (17). The indicator strains used for bioassays were Bacillus megaterium (laboratory stock) and Erwinia carotovora (B. Krebs, FZB Biotech, Berlin, Germany). Difficidin or oxydifficidin producer strain B. subtilis ATCC 39320 (51) was used for comparison of matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectra. The strains used in this study are summarized in Table Table1.1. Bacteria were cultivated routinely on Luria broth (LB) medium solidified with 1.5% agar. For polyketide production and MS analysis, the bacteria were grown in Landy medium (24). The media and buffer used for DNA transformation of Bacillus cells were prepared according to the method of Kunst and Rapoport (22). Competent cells were prepared as previously described (20).

Strains and plasmids used in this study

Mass spectrometric (MS) analysis. (i) MALDI-TOF MS.

Polyketides of B. amyloliquefaciens FZB 42 were identified in extracts of lyophilized culture filtrates and XAD7 extracts by MALDI-TOF MS. Mass spectra were recorded with a Bruker Daltonik Reflex MALDI-TOF instrument containing a 337-nm nitrogen laser for desorption and ionization. One hundred to 200 single scans were accumulated for every spectrum. 2,5-Dihydroxbenzoic acid was used as the matrix. For MS analysis, 1- to 2-μl portions of extracts were mixed with an equal volume of matrix solution, spotted onto the target, and air dried.

Positive-ion detection and reflector mode were used. The acceleration and reflector voltages were 20 and 23.4 kV in pulsed ion extraction mode. A molecular mass gate of 300 Da improved the measurements by filtering out most matrix ions. Monoisotopic mass numbers were obtained.

(ii) High-performance liquid chromatography-electrospray ionization (HPLC-ESI) MS.

HPLC-ESI MS was performed on a QTRAP 2000 system (Applied Biosystems, Darmstadt, Germany) coupled with an Agilent 1100 HPLC system (Agilent, Waldbronn, Germany). Aliquots of acetonitrile (ACN)-water extracts of the culture filtrates of wild-type and mutant strains were fractionated by reversed-phase HPLC on a Merck LiChroCART C18 5-μm column (125 by 4 mm) at a flow rate of 1.5 ml/min with a gradient of 0% ACN plus 0.1% formic acid to 100% ACN plus 0.1% formic acid in 10 min. The eluent was split to a flow rate of 300 μl/min and fed online into the spray chamber. Every sample was measured in the negative and positive modes, and mass spectra were acquired in an m/z range of 300 to 800 at a scan rate of 1,000 atomic mass units/s.

Construction of mutants defective in polyketide synthesis.

A summary of all of the mutant strains and plasmids used in this study is given in Table Table1.1. To generate sfp mutant CH3 (Δsfp::ermAM), a 2.1-kb PCR fragment was amplified with primers sfp-1 (5′-TCAACGTGTCCAACGTCAAG) and sfp2-2 (5′-AGTGATTAAGGATTTGGCGAAC) and cloned into pGEM-T. Plasmid pCH3 was obtained by insertion of ermAM isolated from plasmid pMX39, a derivative of pDB101 (5), into the central sfp gene region as previously described (20). Mutant CH3 was obtained after transformation of FZB 42 by the linearized plasmid.

Mutant CH6 (Δpks1KS1::cat) was obtained as follows. With primers pks1Up (5′-AAAGGAGCGGAGTGCAACATC) and pks1Dw (5′-TGAGATGATGCCGTCCTCTTC), a 3.3-kb fragment containing the first KS domain of the pks1 gene cluster was amplified by PCR and cloned into vector pGEM-T. A central 1.4-kb fragment of the insert was removed by digestion with EcoRI and HpaI and replaced with a chloramphenicol resistance cassette, yielding pCH6. The ScaI-linearized plasmid was recombined via double homologous recombination into competent B. amyloliquefaciens FZB 42 cells, yielding mutant CH6.

Mutant CH7 (Δpks2KS1::cat) was obtained as follows. A 3.0-kb PCR fragment containing the first KS domain of pks2 was amplified with primers pks2Up (5′-AGCTCATTGACAGCGATGCTGC) and pks2Dw (5′-ATGATCGCCGCTTCCTCATCAG) and cloned into vector pGEMT. An internal 1.3-kb fragment was removed by EcoRI-KpnI digestion and replaced with the 1.3-kb cat cassette. This plasmid was linearized with ScaI and transformed into FZB 42 to gain pks2 knockout mutant CH7.

Mutant CH8 (Δpks3KS1::ermAM) was obtained as follows. The pks3 gene cluster was disrupted by insertion of an Emr gene cassette. A 2.2-kb fragment containing the first KS domain of the biosynthesis gene cluster was amplified with primers pks3Up (5′-ACATTGACCGCATCCTCAATCTG) and pks3Dw (5′-TCAGCTGCTGTTCGGAATGTG), cloned into vector pGEM-T, and then digested with EcoRI and Eco72I, which removed a central 376-bp fragment. An Emr gene cassette (1.9 kb) derived from pMX39 was cloned into the deleted construct, yielding pCH8. The pks3 mutant CH8 was obtained after transformation of competent FZB 42 cells with ApaI-linearized pCH8.

Strain CH11 (Δpks1KS1::cat Δpks3KS1::ermAM), impaired in the synthesis of bacillaene and difficidin, was generated by transformation of strain CH8 with chromosomal DNA isolated from strain CH6.

Strain CH12 (Δpks2KS1::cat Δpks3KS1::ermAM), impaired in the synthesis of difficidin and with a knockout mutation of the pks2 gene cluster, was generated by transformation of strain CH7 with chromosomal DNA isolated from strain CH8.

Strain CH14 (Δpks1KS1::cat Δpks2KS1::neo), impaired in the synthesis of bacillaene and with a knockout mutation of the pks2 gene cluster, was generated by transformation of strain CH6 with chromosomal DNA isolated from strain CH13. In order to avoid identical antibiotic resistances in the double-mutant strain, the antibiotic resistance of strain CH8 was previously changed to neomycin resistance by means of marker exchange plasmid pECE73 (44), yielding strain CH13 (Table (Table11).

All mutants were selected from media containing the appropriate antibiotics at the concentrations recommended by Cutting and van der Horn (8) and confirmed by Southern hybridization and colony PCR with appropriate primers.


Bioautographs were prepared as previously described (15). In short, supernatants from bacteria grown in Landy medium were loaded onto an XAD7 (Sigma) resin column, washed, and eluted with methanol. After drying, the samples obtained were redissolved in methanol, spotted onto silica gel 60 F254 thin-layer chromatography (TLC) aluminum sheets (20 by 20 cm; Merck, Darmstadt, Germany), and separated by TLC chromatography (chloroform-methanol-water at 65:25:4, vol/vol). Afterwards, strips of the TLC plate were placed for 2 h at room temperature on the surface of a top agar layer containing the indicator strain. The agar plates were then incubated overnight at 37°C. Zones of inhibition documented the positions of antibiotics separated by TLC. The experiments were repeated at least three times.

Analysis of domain structure.

The program package SEARCHPKS of the modular polyketide synthase database available at http://linux1.nii.res.in/~pksdb/DBASE/page.html and BLAST comparison with annotated domains of similar PKSs were used to detect conserved active-site motifs.

Construction of phylogenetic trees.

The amino acid sequences of discrete Bacillus ATs and of the KS domains extracted from modular Bacillus PKSs were used for BLASTP comparison in order to detect their closest orthologs. Sequences were aligned by the ClustalW program (47) accessible at http://www.ebi.ac.uk/clustalw/. A distance matrix was calculated from this alignment by protein distance matrix calculation (PROTDIST program), and the matrix was then transformed into a tree by the NEIGHBOR program. In order to verify the accuracy of the tree, multiple data sets were generated with the SEQBOOT program using 1,000 bootstrap replicates. A tree was built from each replicate with the PROTDIST program, and then bootstrap values were computed with the CONSENSE program. The phylogenetic tree was visualized with the TreeView32 program (http://taxonomy.zoology.gla.ac.uk/rod/treeview.html). The programs used to construct the phylogenetic tree were obtained from the PHYLIP package, v.3.65 (13), which is accessible at http://evolution.gs.washington.edu/phylip.html.

Evaluation of oligonucleotide usage (OU) pattern.

Computational analysis to determine local OU variance (OUV) was performed as described by Reva and Tummler (39).


Discovery of three large trans-AT PKS gene clusters in B. amyloliquefaciens.

Three PKS gene clusters together covering 196,340 bp (Fig. (Fig.1)1) were found at sites around 1.4 Mbp (pks2), 1.7 Mbp (pks1), and 2.3 Mbp (pks3) distant clockwise from the replication origin of the B. amyloliquefaciens genome (20), which is 3,916 kb in size (X.-H. Chen and R. Borriss, unpublished data). All three gene clusters contain the giant open reading frames typical of type I PKS systems. In addition, the pks1 gene cluster contains two hybrid PKS-nonribosomal peptide synthetase (NRPS) genes. The organization of gene cluster pks1 is very similar to that of the pksX gene cluster of B. subtilis 168 (23), but a pksA-like sequence corresponding to a putative transcriptional regulator is missing within the pks1 gene cluster. However, we detected a homolog of pksA adjacent to a β-ketoacyl-ACP synthase (KAS) in a region 0.97 Mbp distant from the FZB 42 replication origin (Table (Table2).2). The question of whether this probable regulator, a member of the TetR family, does regulate in trans expression of the pks1 gene cluster remains open to investigation. Gene clusters pks2 and pks3 contain upstream regulatory elements resembling transcriptional regulators. Notably, the pks3 gene cluster is preceded by two genes encoding a putative LysR-like transcriptional regulator and an antiterminator protein (Table (Table2).2). None of the three clusters encodes PKS modules harboring AT domains that are a feature of most other known modular PKSs. Instead, discrete AT genes were detected upstream of the megasynthase open reading frames (Table (Table2).2). This architecture, which has been recently described as trans-AT PKS (34), is also known from a smaller number of other systems, such as the pederin and onnamide PKSs from bacterial endosymbionts of beetles (34) and sponges (36), respectively; the leinamycin (lnm) PKS from Streptomyces atroolivaceus (7); the mupirocin PKS from Pseudomonas fluorescens (11); the myxovirescin (TA) PKS from Myxococcus xanthus (30); and others (4, 19, 40). Experimental evidence presented for the LnmG AT suggested that the discrete ATs act iteratively by loading malonyl coenzyme A (CoA) onto all PKS modules during polyketide synthesis (7). The unexpected presence of three members of this small PKS family in the B. amyloliquefaciens genome indicates that trans-AT PKSs are much more common than previously anticipated.

FIG. 1.
Organization of the gene clusters involved in polyketide synthesis in B. amyloliquefaciens FZB 42 (pks1, pks2, and pks3) and B. subtilis 168 (pksX). Discrete ATs are indicated by filled arrows, and modular PKS are indicated by bold arrows. The NRPS portions ...
Organization of the three polyketide biosynthesis gene clusters

Characterization of the polyketides produced by B. amyloliquefaciens strain FZB 42 by MALDI-TOF MS and HPLC-ESI MS.

The DNA sequence data imply that strain FZB 42 has the biosynthetic machinery for the production of at least three different kinds of polyketides. Polyketide formation by FZB 42 was investigated by MALDI-TOF MS and HPLC-ESI MS, which yielded complementary results. MALDI-TOF MS measurements were performed with culture filtrates without further purification. Polyketides of B. amyloliquefaciens FZB 42 were investigated at three growth times. Cells were harvested at 12, 24, and 48 h. By means of MALDI-TOF MS, the highest polyketide yield was detected after cultivation for 12 h. Figure Figure2A2A shows MALDI-TOF mass spectra of wild-type strain FZB 42. Polyketide mass peaks were found in an m/z range of 450 to 650. These signals disappeared in culture filtrates of mutant CH3 (Δsfp::ermAM), which is a derivative of FZB 42 with a disrupted sfp gene (Fig. (Fig.2B).2B). We demonstrated that this mutant is deficient in polyketide production because in vivo Sfp protein functions as a 4′-phosphopantetheinyl transferase in the conversion of the apo into the holo form of both NRPSs and PKSs, as already verified in vitro (27).

FIG. 2.
MALDI-TOF mass spectra of polyketide compounds produced by FZB 42 and mutant CH3 after 12 h of cultivation in Landy medium (24; see Materials and Methods). For MS analysis, 1- to 2 μl portions of lyophilized culture filtrates were used. Experimental ...

In the MALDI-TOF mass spectra, several groups of related mass signals were observed that represent protonated, as well as mono- and dialkali, ion adducts (data not shown). The corresponding compounds were identified by comparison with MS data available for the polyketides of B. subtilis ATCC 39320 and B. subtilis 1A/3 (15, 51) and attributed to the polyene antibiotics bacillaene and difficidin (33, 51). Two biosynthetic variants of bacillaene were observed with molecular masses differing by 2 mass units ([M + H]+ = 581.3 and 583.3). Apparently, in the latter compound one double bond of the polyene system of bacillaene is saturated. A characteristic feature in the MALDI-TOF mass spectra of this polyketide is the appearance of dehydrated species with [M + H − 18]+ = 563.3 and 565.3 originating from the parent ions by the loss of H2O.

Difficidin and its oxidized form oxydifficidin appeared mainly as their alkali ion adducts, whereas the intensity of the mass peaks of their protonated forms at m/z 545.3 and 561.3 are very low. Oxydifficidin bears a hydroxyl group at position 5 of the difficidin ring system (51) showing mass signals 16 mass units higher than those of the main metabolite. In particular, peaks were observed at m/z 583.3, 605.3, 621.3, and 637.3 which were attributed to mono- and dialkali ion adducts of these compounds (data not shown). Wilson et al. (51) showed that difficidin is phosphorylated. In agreement, alkali adducts of the dephosphorylated species were detected at m/z 487.4 and 503.3 (difficidin), as well as at m/z 503.4 and 519.4 (oxydifficidin). Presumably, the phosphate group is responsible for attachment of multiple alkali ions. This pattern of polyketide signals was the same as that found for strain ATCC 39320, which has been used by Wilson et al. (51) for isolation and structure analysis of the polyene antibiotic difficidin or oxydifficidin.

For HPLC-ESI MS analysis, lyophilization products of culture filtrates of wild-type B. amyloliquefaciens FZB 42 and the mutants derived therefrom were extracted with acetonitrile-0.1% formic acid. Extracts were submitted to analytical HPLC coupled with an ESI-QTrap mass spectrometer. The data obtained from these HPLC-ESI MS runs are summarized in Table Table3.3. In positive mode, bacillaene variants with molecular masses of [M + H]+ = 581.5 and 583.5, as well as their dehydrated species ([M + H − 18]+ = 563.5 and 565.5), were found at retention times of 7.61 and 7.71 min, respectively. In addition to MS characterization, UV-visible light spectra were obtained by diode array detection, which clearly confirmed the identities of bacillaenes compared to UV-visible light spectra published previously (33). Difficidin and oxydifficidin could only be detected by ESI MS in their deprotonated forms ([M − H] = 543.4 and 559.3) in the negative mode at retention times of 10.91 and 8.54 min, respectively. The MS data obtained by MALDI-TOF MS (Fig. (Fig.2)2) and HPLC-ESI MS (Table (Table3)3) and the bioautographs performed with enriched culture filtrates of wild-type FZB 42 (Fig. (Fig.3)3) indicate that difficidin or oxydifficidin represents the main antibacterial compound of this strain.

FIG. 3.
Survey of antibacterially acting metabolites through bioautography. Supernatants prepared from strain FZB 42 and mutant strains (A, C) and B. subtilis JH642 (sfp0) and OKB105 (sfp+) (B, D) grown for 12 h in Landy medium (24) were separated by ...
Molecular masses of polyketide products of B. amyloliquefaciens FZB 42 and mutant strains detected by LC-ESI mass spectrometrya

Assignment of biological functions to polyketide gene clusters on the basis of gene disruption.

Disruptions of all three pks target genes were carried out by insertion of an antibiotic resistance cassette via homologous recombination (see Materials and Methods). Successful inactivation of gene clusters was proven by MS analysis and bioautography of mutants bearing gene replacement mutations in the respective first KS domains of pks1 (CH6 [Δpks1KS1::cat]), pks2 (CH7 [Δpks2KS1::cat]), and pks3 (CH8 [Δpks3KS1::ermAM]). Extracted ion chromatograms of HPLC-ESI MS runs (Table (Table33 and data not shown) and MALDI-TOF mass spectra (data not shown) from B. amyloliquefaciens FZB 42 and the respective mutant strains, including double-knockout mutants CH11 (Δpks1KS1::cat Δpks3KS1::ermAM), CH12 (Δpks2KS1::cat Δpks3KS1::ermAM), and CH14 (Δpks1KS1::cat Δpks2KS1::neo), yielded complementary results. Mutant CH6 (Δpks1KS1::cat) harbored a mutation of the pks1 system, and as a consequence the bacillaene-specific molecular mass peaks ([M + H]+ = 581.3 and 583.3 and [M + H − 18]+ = 563.3 and 565.3) of the dehydrated species are absent in the MALDI-TOF mass spectrum compared to the wild-type mass spectra. In contrast, difficidin or oxydifficidin biosynthesis remains unaffected, which is shown by signals in HPLC-ESI MS in negative mode ([M − H] = 543.4 and 559.3). MALDI-TOF mass spectra yield m/z = 605.3 and 621.3, apparently indicating adducts with two alkali ions. Mutant CH7 (Δpks2KS1::cat), with a gene disruption in the pks2 cluster, showed the same pattern of bacillaene- and difficidin- or oxydifficidin-related molecular masses as found previously for the wild-type strain. However, we did not identify any molecular mass that could be related to a putative pks2 product; for mutant CH8 (Δpks3KS1::ermAM), only bacillaene-specific molecular masses were detected. CH11 (Δpks1KS1::cat Δpks3KS1::ermAM) is a double mutant with gene disruptions in both the pks1 and pks3 clusters. For this strain, no molecular masses were identified that would indicate the presence of polyketides. Both the bacillaene- and difficidin- or oxydifficidin-related peaks were absent. In CH12 (Δpks2KS1::cat Δpks3KS1::ermAM), pks2 and pks3 were inactivated and only the molecular masses characteristic for bacillaene were observed. Finally, CH14 (Δpks1KS1::cat Δpks2KS1::neo), bearing mutations in the pks1 and pks2 gene clusters exclusively, showed the difficidin- or oxydifficidin-specific molecular mass peaks (data not shown).

On the basis of these results, also confirmed by HPLC-ESI MS (Table (Table3),3), the pks regions in the genome of B. amyloliquefaciens FZB 42 were assigned as follows. pks3 codes for the production of difficidin or oxydifficidin and is designated the dif gene cluster, while pks1 is responsible for the biosynthesis of bacillaene and is designated the bae gene cluster. Since pks1 is an ortholog of the pksX system in B. subtilis 168 (more details are given below), it was challenging to confirm unambiguously that bacillaene is the natural product of this silent gene cluster. To this end, B. subtilis OKB105 (pheA sfp+), known as a potent producer of the lipopeptide surfactin (27), was tested for the ability to produce bacillaene. Bioautographs performed with B. subtilis JH642 (pheA trpC2 sfp0) and its derivative OKB105 demonstrated that the presence of a functional sfp gene enabled B. subtilis to produce that polyketide (Fig. (Fig.3).3). The function of pks2 in B. amyloliquefaciens is less clear insofar as no pks2-specific compound was detected by HPLC-ESI MS or MALDI-TOF MS. However, in bioautographs (Fig. (Fig.3)3) we observed the disappearance in the pks2 mutant CH7 and CH12 supernatants of a weak antibiotic activity, suggesting that this gene cluster might be responsible for the synthesis of an unknown polyketide (tentatively named PK2) that is weakly expressed and/or has low bioactivity.

Remarkably, the growth of phytopathogenic bacteria such as E. carotovora was strongly suppressed by difficidins and bacillaene as the most active agents to suppress bacterial growth. As mentioned above, sfp mutation blocked nearly the entire antibacterial activity. However, bioautographs demonstrated that an antibacterial compound with an unknown structure, putatively produced by conventional ribosomal synthesis, remained present in lipopeptide- and polyketide-deficient mutants (Fig. (Fig.33).

Analysis of PKS architecture. (i) Modular organization of Bacillus PKSs is of unusual architecture.

In most of the PKS systems studied so far, the modular architecture strictly reflects the order and structure of building blocks present in the polyketide chain. This feature has been described as the colinearity rule and is the underlying paradigm that should allow the rational redesign of PKS architectures in order to generate novel polyketides (18, 37). In recent years, however, a growing number of PKSs have been discovered that do not follow the classic biosynthetic rules (50). Domain analysis of the B. amyloliquefaciens PKSs reveals a situation similar to those systems described previously. In none of the three systems, for example, do the deduced PKS proteins feature the textbook architecture consisting of an N-terminal KS domain and a C-terminal ACP domain. Instead, modules with incomplete domain sets occupy the protein termini, suggesting that single modules are encoded on two distinct proteins. An example is the C-terminal KS residing on BaeL that presumably interacts with the remaining part of the module at the N terminus of BaeM in trans, yielding a complete PKS module (Table (Table2).2). A similar domain organization has recently been reported for the lnm biosynthetic gene cluster (45). A second notable feature is that several proteins exhibit unusual domain arrangements, such as ketoreductase (KR) or enoylreductase domains located behind ACP domains (DifJ and -L and Pks2F, -G, and -H) or tandem dehydratase (DH) domains preceding a KS domain (DifK).

The pks1 (bae) gene cluster encodes a hybrid NRPS-PKS system.

Only one of the B. amyloliquefaciens PKSs resembles systems identified earlier. The bae gene cluster exhibits striking architectural and sequence similarity to the pksX gene cluster of B. subtilis 168, which clearly indicates that it is its ortholog (Fig. (Fig.1).1). The core region of both gene clusters differs from that of gene clusters pks2 and dif by the occurrence of two hybrid NRPS-PKS orthologs (pksJ/baeJ and pksN/baeN, Table Table2).2). The N-terminal part of PksJ and its ortholog BaeJ consists of a loading module for an unknown starter unit, while the subsequent module displays features typical of NRPS with consecutive condensation, amino acyl adenylation, and peptidyl carrier protein domains. Downstream of the NRPS module are two trans-AT PKS modules. A second hybrid NRPS-PKS protein, PksN (BaeN), contains in its N-terminal part a complete NRPS-type module, followed by a total of three further PKS modules in which the last one is incomplete, consisting only of the KS and KR domains.

Although with bacillaene for the first time a secondary metabolite can be attributed to this gene cluster, the sparse available chemical data do not allow clear assignment of gene functions to specific synthesis steps of a molecule with an as-yet-unknown structure. It has been reported that bacillaene is a conjugated hexaene with a molecular formula of C35H48O7, as deduced from MS data (33). However, the presence of two NRPS modules in the bae cluster is not in accordance with this formula, since the incorporation of amino acids would require the presence of at least two nitrogen atoms. In two independent studies, Stachelhaus, Challis, and coworkers (6, 43) have identified amino acid residues within the adenylation domain of NRPSs that allow deduction of the substrate specificity of NRPS modules. Therefore, in order to predict the amino acids incorporated into the bacillaene structure, the nonribosomal code was extracted from the two bae adenylation domains (Table (Table4).4). For the first NRPS module of BaeJ, the analysis gave 100% identity with the glycine-specific code (31, 46) while no similarity to known codes was found for the other BaeN module. These data give rise to the assumption that bacillaene should contain at least one glycine residue.

Specificity-conferring code of A domains residing in hybrid NPRS/KS of the pks1 (bae) and pksX biosynthesis gene clusters

Model for difficidin or oxydifficidin synthesis.

Similar to the bae genes, the dif system exhibits an unusual architecture that sets it apart from most other PKSs. Nevertheless, a plausible biosynthetic pathway can be proposed for difficidin which accounts for the presence of most of the features of the dif system (Fig. (Fig.4).4). The starting point would be a C3 precursor, such as pyruvate, which is converted to an ACP-bound acrylyl moiety by reduction (perhaps catalyzed by DifE), dehydration, and transfer by the acyl-CoA ligase DifD to the N-terminal loading ACP of DifF. The incorporation of pyruvate and other α-oxygenated three-carbon starter units into polyketides is not unprecedented (14, 26); however, to our knowledge an acrylate starter moiety has not yet been reported. The subsequent biosynthetic steps that extend the starter unit can be largely rationalized by the colinearity rule, with several notable deviations. A number of modules lack domains that would be predicted from the polyketide structure. Those are a KR domain in module 3, two DH domains in modules 4 and 9, and two enoylreductase domains in modules 2 and 8. It is possible that these activities are provided by other proteins acting in trans. Such a complementation has also been suggested for other polyketides, such as chalcomycin (49) and epothilone (46). Thus, in order to explain the presence of a double bond in epothilone, Tang and coworkers proposed a mechanism consisting of module skipping, use of a DH domain of a downstream module, and subsequent iterative use of the same module for the next round of elongation (46). However, in difficidin biosynthesis the architecture of most downstream modules would not allow such a scenario. Further studies are necessary to determine whether elements within the dif system are responsible for complementation or whether the activities are provided from somewhere else in the genome, for example, the bae or pks2 system. Another deviation from the colinearity rule is the presence of two superfluous modules, module 5, jointly encoded by difG and difH, and module 11, a product of difJ and difK, which are obviously skipped during biosynthesis. Closer inspection of the KS domains, which are an essential part of functional PKS modules, revealed that most of them contained the functional residues characterized as necessary for their catalytic activity, i.e., C163, D306, E309, H298, K328, and H333 (numbering according to E. coli type I β-ketoacyl-ACP synthase) (25, 52). However, the KSs of modules 5 and 11 both feature an H298 → N substitution and should therefore be inactive, since they would be unable to catalyze the decarboxylation of malonyl units. Similar mutations have been previously reported for the pederin, onnamide, and mupirocin systems and are also present in the bae and pksX clusters. In addition, we were unable to identify an ACP located inside module 11 of the dif PKS, indicating that this module might be a partially decayed evolutionary remnant.

FIG. 4.
Model for difficidin biosynthesis. Domains predicted to be inactive are crossed out. Some activities are predicted to be provided in trans by unknown enzymes, which are shown above some of the modules. Domain abbreviations not defined in the text: ER, ...

An interesting feature of the difficidin structure is the occurrence of four double bonds with a Z configuration, a geometry that is very rare among polyketides (2, 32, 41, 49). It has been proposed that an aspartate residue in the KR domain plays a crucial role in defining the stereochemistry of the ketoreduction step, which in turn would influence the double-bond configuration after dehydration by the downstream DH domain (38). An alignment of dif KR domains revealed an agreement with this model for the synthetic assembly of most of the double bonds except for the KRs of DifJ. These KRs contain sequence features that would lead to a prediction of the respective opposite configurations. Similar deviations from the Asp rule have been observed previously for the phoslactomycins, which are products of a cis-AT PKS (32), suggesting the need for further refinement of the model.

The likely final steps in difficidin biosynthesis are the conversion of a keto function into an exomethylene group, subsequent cyclization by the thioesterase domain, and attachment of the rare phosphate moiety to the hydroxyl group at C-15. The formation of exomethylene or other carbon-containing functionalities from keto groups has been described for a limited number of other polyketides, such as pederin-type compounds (34, 36), myxovirescin (31), and the jamaicamides (10). It has been proposed that these groups are generated by joint action of a 3-hydroxy-3-methylglutaryl-CoA synthase (HMGS) and an enoyl-CoA hydratase (EH) homolog, as well as free-standing KS and ACP units, which are invariably found encoded within the corresponding gene clusters (10, 36). We identified proteins with similarity to HMGSs, EHs, and ACPs encoded by difN, difO, and difC, respectively, but no individual KS gene. A candidate for the likely final step in difficidin biosynthesis, phosphorylation, is DifB, which resembles (30% identity and 50% similarity) a hypothetical bacterial protein kinase (accession no. ZP_00829240) and contains typical kinase-like sequence profiles (smart00220, COG0661).

Phylogenetic analysis of pks gene clusters.

While the bae gene cluster harbors genes that are clearly orthologs of the B. subtilis 168 pksX operon (Fig. (Fig.1),1), favoring the idea that this cluster is part of the B. subtilis or B. amyloliquefaciens core region, the origin of the additional pks2 and dif gene clusters is less obvious. One possibility is that pks2 and bae might have been acquired by horizontal gene transfer from other soil bacteria. Alternatively, pks2 and dif might have evolved from an ancestral pks operon by several gene duplication events, or—more likely—by both events. Recently, it has been shown that the operon encoding pederin biosynthesis in the bacterial symbiont of Paederus fuscipes beetles is part of a genomic island characterized by several decayed insertion sequence elements (35). However, no genetic features typical of horizontally acquired DNA were identified in the pks regions of B. amyloliquefaciens. The G+C contents of bae (49.47%), pks2 (43.39%), and dif (49.48%) were not clearly distinct from the average G+C content of the whole genome of 46.2% (Chen and Borriss, unpublished). Moreover, computational analysis of the local OUV (39) of the pks gene clusters did not reveal an island-specific OU architecture. Local deviations in the tetranucleotide pattern from the signature of the whole FZB 42 genome were only 2.47 for bae and 1.91 for dif. A slightly higher value was determined for pks2 (3.18). Moreover, features indicating events of horizontal gene transfer such as remnants of phage-, insertion sequence element-, or transposase-like sequences were not detected, except for a phage-like sequence residing at the flanking region of the pks2 gene cluster (Table (Table22).

Sequence alignments of the discrete ATs of all of the pks gene clusters revealed similarities to the probably functionally related PksC, PksD, and PksE ATs of B. subtilis 168 (23); the PedC and PedD ATs involved in pederin biosynthesis (34); the MmpIII AT1/AT2 of P. fluorescens involved in mupirocin biosynthesis (11); and the LnmG AT of S. atroolivaceus, an iteratively acting AT in leinamycin biosynthesis (7). A phylogenetic tree constructed for this group of discrete malonyl-CoA-specific ATs revealed that the hybrid AT-oxidoreductase (AT-OR) proteins present in all hitherto known Bacillus PKS gene clusters form, together with the mono domain ATs PksC (pksX) and BaeC, a distinct clade not including other members of the family of discrete ATs (Fig. (Fig.5A).5A). Therefore, we hypothesize that the discrete AT-ORs have evolved from a PksE-like ancestor protein residing in ancient bacilli. Gene duplication events might have generated the paralogous mono domain ATs PksC and BaeC. In contrast, the third pair of discrete ATs, PksD and BaeD, might have evolved independently.

Phylogenetic tree analyses of discrete ATs and integrated KS domains. Bootstrap values (n = 1,000) for branches are given. The scale bar represents the number of substitutions. (A) Phylogenetic tree analysis of discrete ATs occurring in bacterial ...

In order to analyze PKS evolution in more detail, we used a similar phylogenetic approach using all of the KS domains extracted from the Bacillus PKSs and hybrid NRPS-PKSs and their closest relatives identified by BLASTP comparison. In total, 100 KS sequences extracted from trans-AT PKS operons were included in this phylogenetic analysis. At least five clusters consisting of paralogous Bacillus KS domains were identified. Notably, a distinct clade (group 2) is formed by a group of KS domains located in hybrid Bacillus NRPS-PKS proteins adjacent to the NRPS modules. As expected, the shortest evolutionary distances were detected between the corresponding KS domains derived from bae and pksX, but several KSs derived from the dif operon were also involved in clades formed by related members of the Bacillus pks gene clusters (Fig. (Fig.5B).5B). Especially striking was the close relatedness between domain KS-6 residing at DifH and the corresponding KS domains residing at PksL (pksX) and BaeL (76% identity, 87% similarity), for example, suggesting that at least parts of the bacillaene and difficidin biosynthesis gene clusters have the same origin. It is tempting to speculate that processes of evolution of novel pks operons were completed by several events of homologous recombination leading to gene duplications. Because of the presence of sequences with repeated homology located at distinct sites of the modularly organized PKSs, such events are considered to be very likely. Homologies of 67% identity and 81% similarity detected between domains KS-2 and KS-3 residing at DifF with their counterparts KS-7 and KS-9 residing at DifI support such an assumption. Gene duplication events are particularly visible in gene cluster pks2, where many of the KS domains appeared as a separate branch of the phylogenetic tree and a high degree of similarity between KS domains residing at distinct proteins was detected. Similarity between the Bacillus pks gene clusters is also apparent in tailoring enzymes, e.g., between the adjacent HMGS- and EH-encoding genes present in the pksX (pksGHI), bae (baeGHI), and dif (difNO) gene clusters (Table (Table2).2). While these genes are arranged colinearly within the pksX and bae gene cluster, they have been reshuffled to a position downstream from the PKS-encoding genes within the dif gene cluster (Fig. (Fig.11).

FIG. 5

Taken together, these results support the idea that the bae and dif gene clusters might have the same origin but also several gene duplication events might have contributed to the evolution of the recent biosynthesis gene clusters. Lower similarity exists between pks2 and the other Bacillus gene clusters. Despite its different G+C content, its different OUV, and the presence of an adjacent phage-like sequence, we cannot exclude the possibility that major parts of the pks2 operon have been acquired by horizontal gene transfer. The greater evolutionary distances between the pks2 ATs and KSs within the phylogenetic trees constructed from type I trans-AT PKS systems support this notion.


The antibacterial polyketides bacillaene and difficidin or oxydifficidin were identified in the culture filtrate of B. amyloliquefaciens FZB 42 by MALDI-TOF MS and HPLC-ESI MS. It has been previously suggested that biosynthesis of those polyketides should be accomplished by giant operons similar to that of the inactive pksX operon in B. subtilis 168. Here we present clear genetic evidence that the pks1 and pks3 gene clusters encoding modular type I PKSs of the trans-AT architectural group are responsible for the biosyntheses of bacillaene and difficidin, respectively, while the type I PKS of the pks2 gene cluster seems to be involved in the synthesis of an as-yet-unknown polyketide with weak antibacterial activity. Bacillaene and difficidin are the main antibacterial secondary metabolites of environmental B. amyloliquefaciens FZB 42. We have defined for the first time the complete gene clusters involved in their synthesis and have presented a model for the synthesis of difficidin based on the structure of the pks3 (dif) gene cluster. We suggest that the impressive biosynthetic capacity of strain FZB 42 to produce antagonistically acting polyketides and lipopeptides enables FZB 42 to cope successfully with competing organisms, e.g., phytopathogenic bacteria and fungi, within its natural environment. Structural similarity detected between gene clusters pks1(bae) and pks3 (dif), which are involved in the synthesis of two different polyketides, should inspire future experiments in which swapping of domains occurring in the distinct but related gene clusters of B. amyloliquefaciens might allow study of their unusual enzymologies and generation of a variety of novel bioactive polyketides.


This work was supported by funds from the competence network Genome Research on Bacteria (GenoMik), financed by the German Ministry for Education and Research (R.B. and G.G.), and by grants from the Lower Saxony Ministry of Science and Culture (G.G.). The work of K.S. and R.D.S. was supported by an Emmy-Noether fellowship from the DFG (SU 239/2).

An anonymous referee is thanked for the suggestion to investigate the bacillaene production of OKB105. We thank Mohammed Marahiel for supplying strain OKB105. Oleg Reva is thanked for introducing us to OUV analysis of genomic regions.


1. Albertini, A. M., T. Caramori, F. Scoffone, C. Scotti, and A. T. Galizzi. 1995. Sequence around the 159° region of the Bacillus subtilis genome: the pksX locus spans 33.6 kb. Microbiology 141:299-309. [PubMed]
2. August, P. R., L. Tang, Y. J. Yoon, S. Ning, R. Muller, T. W. Yu, M. Taylor, D. Hoffmann, C. G. Kim, X. Zhang, C. R. Hutchinson, and H. G. Floss. 1998. Biosynthesis of the ansamycin antibiotic rifamycin: deductions from the molecular analysis of the rif biosynthetic gene cluster of Amycolatopsis mediterranei S699. Chem. Biol. 5:69-79. [PubMed]
3. Cane, D. E. 1997. A special thematic issue on polyketide and nonribosomal polypeptide biosynthesis. Chem. Rev. 97:2463-2706. [PubMed]
4. Carvalho, R., R. Reid, N. Viswanathan, H. Gramajo, and B. Julien. 2005. The biosynthetic genes for disorazoles, potent cytotoxic compounds that disrupt microtubule formation. Gene 359:91-97. [PubMed]
5. Ceglowski, P., and J. C. Alonso. 1994. Gene organization of the Streptococcus pyogenes plasmid pDB101: sequence analysis of the ORF eta-copS region. Gene 145:33-39. [PubMed]
6. Challis, G. L., J. Ravel, and C. A. Townsend. 2000. Predictive, structure-based model of amino acid recognition by nonribosomal peptide synthetase adenylation domains. Chem. Biol. 7:211-224. [PubMed]
7. Cheng, Y.-Q., G.-L. Tang, and B. Shen. 2003. Type I polyketide synthase requiring a discrete acyltransferase for polyketide biosynthesis. Proc. Natl. Acad. Sci. USA 100:3149-3154. [PMC free article] [PubMed]
8. Cutting, S. M., and P. B. van der Horn. 1990. Genetic analysis, p. 27-74. In C. R. Harwood and S. M. Cutting (ed.), Molecular biological methods for Bacillus. Wiley-Interscience, Chichester, United Kingdom.
9. Du, L., C. Sanchez, M. Chen, D. J. Edwards, and B. Shen. 2000. The biosynthetic gene cluster for the antitumor drug bleomycin from Streptomyces verticillus ATCC 15003 supporting functional interactions between nonribosomal peptide synthetases and a polyketide synthase. Chem. Biol. 7:623-642. [PubMed]
10. Edwards, D. J., B. L. Marquez, L. M. Nogle, K. McPhail, D. E. Goeger, M.A.Roberts, and W. H. Gerwick. 2004. Structure and biosynthesis of the jamaicamides, new mixed polyketide-peptide neurotoxins from the marine cyanobacterium Lyngbya majuscula. Chem. Biol. 11:817-833. [PubMed]
11. El-Sayed, A. K., J. Hothersall, S. M. Cooper, E. Stephens, T. J. Simpson, and C. M. Thomas. 2003. Characterization of the mupirocin biosynthesis gene cluster from Pseudomonas fluorescens NCIMB 10586. Chem. Biol. 10:419-430. [PubMed]
12. Emmert, E. A. B., A. K. Klimowicz, M. G. Thomas, and J. Handelsman. 2004. Genetics of zwittermicin A production by Bacillus cereus. Appl. Environ. Microbiol. 70:104-113. [PMC free article] [PubMed]
13. Felsenstein, J. 1989. PHYLIP: phylogeny inference package. Cladistics 5:164-166.
14. Hildebrand, M., L. E. Waggoner, H. Liu, S. Sudek, S. Allen, C. Anderson, D. H. Sherman, and M. Haygood. 2004. bryA: an unsual polyketide synthase gene from the uncultivated bacterial symbiont of the bryozoan Bugula neritina. Chem. Biol. 11:1543-1552. [PubMed]
15. Hofemeister, J., B. Conrad, B. Adler, B. Hofemeister, J. Feesche, N. Kucheryava, G. Steinborn, P. Franke, N. Grammel, A. Zwintscher, F. Leenders, G. Hitzeroth, and J. Vater. 2004. Genetic analysis of the biosynthesis of nonribosomal peptide- and polyketide-like antibiotics, iron uptake and biofilm formation by Bacillus subtilis A1/3. Mol. Genet. Genomics 272:363-378. [PubMed]
16. Idriss, E. E. S., H. Bochow, H. Ross, and R. Borriss. 2004. Use of Bacillus subtilis as biocontrol agent. 6. Phytohormone like action of culture filtrates prepared from plant growth promoting Bacillus amyloliquefaciens FZB24, FZB 42, FZB45 and Bacillus subtilis FZB37. J. Plant Dis. Prot. 111:583-597.
17. Idriss, E. E. S., O. Makarewicz, A. Farouk, K. Rosner, R. Greiner, H. Bochow, T. Richter, and R. Borriss. 2002. Extracellular phytase activity of Bacillus amyloliquefaciens FZB 45 contributes to its plant-growth-promoting effect. Microbiology 148:2097-2109. [PubMed]
18. Katz, L., and S. Donadio. 1993. Polyketide synthesis: prospects for hybrid antibiotics. Annu. Rev. Microbiol. 47:875-912. [PubMed]
19. Kopp, M., H. Irschik, S. Pradella, and R. Muller. 2005. Production of the tubulin destabilizer disorazol in Sorangium cellulosum: biosynthetic machinery and regulatory genes. ChemBioChem 6:1277-1286. [PubMed]
20. Koumoutsi, A., X.-H. Chen, A. Henne, H. Liesegang, G. Hitzeroth, P. Franke, J. Vater, and R. Borriss. 2004. Structural and functional characterization of gene clusters directing nonribosomal synthesis of bioactive cyclic lipopeptides in Bacillus amyloliquefaciens strain FZB 42. J. Bacteriol. 186:1084-1096. [PMC free article] [PubMed]
21. Krebs, B., B. Höding, S. M. Kübart, A. Workie, H. Junge, G. Schmiedeknecht, P. Grosch, H. Bochow, and M. Heves. 1998. Use of Bacillus subtilis as biocontrol agent. 1. Activities and characterization of Bacillus subtilis strains. J. Plant Dis. Prot. 105:181-197.
22. Kunst, F., and G. Rapoport. 1995. Salt stress is an environmental signal affecting degradative enzyme synthesis in Bacillus subtilis. J. Bacteriol. 177:2403-2407. [PMC free article] [PubMed]
23. Kunst, F., N. Ogasawara, I. Moszer, et al. 1997. The complete genome sequence of the gram-positive bacterium Bacillus subtilis. Nature 390:249-256. [PubMed]
24. Landy, M., G. H. Warren, S. B. Roseman, and L. G. Colio. 1948. Bacillomycin, an antibiotic from Bacillus subtilis active against pathogenic fungi. Proc. Soc. Exp. Biol. Med. 67:539-541. [PubMed]
25. McGuire, K. A., M. Siggaard-Andersen, M. G. Bangera, J. G. Olsen, and P. von Wettstein-Knowles. 2001. β-Ketoacyl-acyl carrier protein synthase I of Escherichia coli: aspects of the condensation mechanism revealed by analysis of mutations in the active site pocket. Biochemistry 40:9836-9845. [PubMed]
26. Mochizuki, S., K. Hiratsu, M. Suwa, T. Ishii, F. Sugino, K. Yamada, and H. Kinashi. 2003. The large linear plasmid pSLA2-L of Streptomyces rochei has an unusually condensed gene organization for secondary metabolism. Mol. Microbiol. 48:1501-1510. [PubMed]
27. Mootz, H. D., R. Finking, and M. A. Marahiel. 2001. 4′-Phosphopantetheine transfer in primary and secondary metabolism of Bacillus subtilis. J. Biol. Chem. 276:37289-37298. [PubMed]
28. Nakano, M. M., M. M. Marahiel, and P. Zuber. 1988. Identification of a genetic locus required for biosynthesis of the lipopeptide antibiotic surfactin in Bacillus subtilis. J. Bacteriol. 170:5662-5668. [PMC free article] [PubMed]
29. Nakano, M. M., N. Corbell, J. Besson, and P. Zuber. 1992. Isolation and characterization of sfp: a gene that functions in the production of the lipopeptide biosurfactant, surfactin, in Bacillus subtilis. Mol. Gen. Genet. 232:313-321. [PubMed]
30. Paitan, Y., G. Alon, E. Orr, E. Z. Ron, and E. Rosenberg. 1999. The first gene in the biosynthesis of the polyketide antibiotic TA of Myxococcus xanthus codes for a unique PKS module coupled to a peptide synthetase. J. Mol. Biol. 286:465-474. [PubMed]
31. Paitan, Y., E. Orr, E. Z. Ron, and E. Rosenberg. 1999. Genetic and functional analysis of genes required for the post-modification of the polyketide antibiotic TA of Myxococcus xanthus. Microbiology 145:3059-3067. [PubMed]
32. Palaniappan, N., B. S. Kim, Y. Sekiyama, H. Osada, and K. A. Reynolds. 2003. Enhancement and selective production of phoslactomycin B, a protein phosphatase IIa inhibitor, through identification and engineering of the corresponding biosynthetic gene cluster. J. Biol. Chem. 278:35552-35557. [PubMed]
33. Patel, P. S., S. Huang, S. Fisher, D. Pirnik, C. Aklonis, L. Dean, E. Meyers, P. Fernandes, and F. Mayerl. 1995. Bacillaene, a novel inhibitor of prokaryotic protein synthesis produced by Bacillus subtilis: production, taxonomy, isolation, physico-chemical characterization and biological activity. J. Antibiot. 48:997-1003. [PubMed]
34. Piel, J. 2002. A polyketide synthase-peptide synthetase gene cluster from an uncultured bacterial symbiont of Paederus beetles. Proc. Natl. Acad. Sci. USA 99:14002-14007. [PMC free article] [PubMed]
35. Piel, J., I. Höfer, and D. Hui. 2004. Evidence for a symbiosis island involved in horizontal acquisition of pederin biosynthetic capabilities by the bacterial symbiont of Paederus fuscipes beetles. J. Bacteriol. 186:1280-1286. [PMC free article] [PubMed]
36. Piel, J., D. Hui, G. Wen, D. Butzke, M. Platzer, N. Fusetani, and S. Matsunaga. 2004. Antitumor polyketide biosynthesis by an uncultivated bacterial symbiont of the marine sponge Theonella swinhoei. Proc. Natl. Acad. Sci. USA 101:16222-16227. [PMC free article] [PubMed]
37. Rawlings, B. J. 2001. Type I polyketide biosynthesis in bacteria (part B). Nat. Prod. Rep. 18:231-281. [PubMed]
38. Reid, R., M. Piagentini, E. Rodriguez, G. Ashley, N. Viswanathan, J. Carney, D. V. Santi, C. R. Hutchinson, and R. McDaniel. 2003. A model of structure and catalysis for ketoreductase domains in modular polyketide synthases. Biochemistry 42:72-79. [PubMed]
39. Reva, O. N., and B. Tummler. 2004. Global features of sequences of bacterial chromosomes, plasmids and phages revealed by analysis of oligonucleotide usage patterns. BMC Bioinformatics 5:90. [Online.] doi:.10.1186/1471-2105-5-90 [PMC free article] [PubMed] [Cross Ref]
40. Royer, M., L. Costet, E. Vivien, M. Bes, A. Cousin, A. Damais, I. Pieretti, A. Savin, S. Megessier, M. Viard, R. Frutos, D. W. Gabriel, and P. C. Rott. 2004. Albicidin pathotoxin produced by Xanthomonas albilineans is encoded by three large PKS and NRPS genes present in a gene cluster also containing several putative modifying, regulatory, and resistance genes. Mol. Plant-Microbe Interact. 17:414-427. [PubMed]
41. Schupp, T., C. Toupet, N. Engel, and S. Goff. 1998. Cloning and sequence analysis of the putative rifamycin polyketide synthase gene cluster from Amycolatopsis mediterranei. FEMS Microbiol. Lett. 159:201-207. [PubMed]
42. Scotti, C., M. Piatti, A. Cuzzoni, P. Perani, A. Tognoni, G. Grandi, A. Galizzi, and A. M. Albertini. 1993. A Bacillus subtilis large ORF coding for a polypeptide highly similar to polyketide synthases. Gene 130:65-71. [PubMed]
43. Stachelhaus, T., H. D. Mootz, and M. A. Marahiel. 1999. The specificity-conferring code of adenylation domains in nonribosomal peptide synthetases. Chem. Biol. 6:493-505. [PubMed]
44. Steinmetz, M., and R. Richter. 1994. Plasmids designed to alter the antibiotic resistance expressed by insertion mutations in Bacillus subtilis, through in vivo recombination. Gene 142:79-83. [PubMed]
45. Tang, G.-L., Y.-Q. Cheng, and B. Shen. 2004. Leinamycin biosynthesis revealing unprecedented architectural complexity for a hybrid polyketide synthase and nonribosomal peptide synthetase. Chem. Biol. 11:33-45. [PubMed]
46. Tang, L., S. L. Ward, L. Chung, J. R. Carney, Y. Li, R. Reid, and L. Katz. 2004. Elucidating the mechanism of cis double bond formation in epothilone biosynthesis. J. Am. Chem. Soc. 126:46-47. [PubMed]
47. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680. [PMC free article] [PubMed]
48. Vater, J., X. Gao, G. Hitzeroth, C. Wilde, and P. Franke. 2003. “Whole cell”—matrix-assisted laser desorption ionization-time of flight mass spectrometry, an emerging technique for efficient screening of biocombinatorial libraries of natural compounds—present state of research. Comb. Chem. High Throughput Screen. 6:557-567. [PubMed]
49. Ward, S. L., Z. H. Hu, A. Schirmer, R. Reid, W. P. Revill, C. D. Reeves, O. V. Petrakovsky, S. D. Dong, and L. Katz. 2004. Chalcomycin biosynthesis gene cluster from Streptomyces bikiniensis: novel features of an unusual ketolide produced through expression of the chm polyketide synthase in Streptomyces fradiae. Antimicrob. Agents Chemother. 48:4703-4712. [PMC free article] [PubMed]
50. Wenzel, S. C., and R. Muller. 2005. Formation of novel secondary metabolites by bacterial multimodular assembly lines: deviations from textbook biosynthetic logic. Curr. Opin. Chem. Biol. 9:447-458. [PubMed]
51. Wilson, K. E., J. E. Flor, R. E. Schwartz, H. Joshua, J. L. Smith, B. A. Pelak, J. M. Liesch, and D. Hensens. 1987. Difficidin and oxydifficidin: novel broad spectrum antibacterial antibiotics produced by Bacillus subtilis. 2. Isolation and physicochemical characterization. J. Antibiot. 40:1682-1691. [PubMed]
52. Witkowski, A., A. K. Joshi, and S. Smith. 2002. Mechanism of the beta-ketoacyl synthase reaction catalyzed by the animal fatty acid synthase. Biochemistry 41:10877-10887. [PubMed]
53. Zweerink, M. M., and A. Edison. 1987. Difficidin and oxydifficidin: novel broad spectrum antibacterial antibiotics produced by Bacillus subtilis. III. Mode of action of difficidin. J. Antibiot. 40:1692-1697. [PubMed]

Articles from Journal of Bacteriology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...