Characterization of the Complete Zwittermicin A Biosynthesis Gene Cluster from Bacillus cereus

Bacillus cereus UW85 produces the linear aminopolyol antibiotic zwittermicin A (ZmA). This antibiotic has diverse biological activities, such as suppression of disease in plants caused by protists, inhibition of fungal and bacterial growth, and amplification of the insecticidal activity of the toxin protein from Bacillus thuringiensis. ZmA has an unusual chemical structure that includes a d amino acid and ethanolamine and glycolyl moieties, as well as having an unusual terminal amide that is generated from the modification of the nonproteinogenic amino acid β-ureidoalanine. The diverse biological activities and unusual structure of ZmA have stimulated our efforts to understand how this antibiotic is biosynthesized. Here, we present the identification of the complete ZmA biosynthesis gene cluster from B. cereus UW85. A nearly identical gene cluster is identified on a plasmid from B. cereus AH1134, and we show that this strain is also capable of producing ZmA. Bioinformatics and biochemical analyses of the ZmA biosynthesis enzymes strongly suggest that ZmA is initially biosynthesized as part of a larger metabolite that is processed twice, resulting in the formation of ZmA and two additional metabolites. Additionally, we propose that the biosynthesis gene cluster for the production of the amino sugar kanosamine is contained within the ZmA biosynthesis gene cluster in B. cereus UW85.Bacillus cereus strain UW85 was isolated for its ability to suppress disease in alfalfa caused by the plant pathogen Phytophthora medicaginis (17). This antiprotist activity was subsequently found to be associated with the filtrate of fully sporulated B. cereus UW85 (37). Analysis of this filtrate identified two antiprotist antibiotics, zwittermicin A (ZmA) and kanosamine (28, 37). Of the two antibiotics, ZmA has shown the more interesting biological activities, having not only antiprotist activity, but also antibiotic activity against gram-positive and gram-negative bacteria, as well as fungi (32, 38). ZmA was also found to potentiate the activity of the toxin protein of Bacillus thuringiensis against insects (3).A preliminary chemical structure of ZmA was determined by the Handelsman and Clardy groups (18). More recently, Rogers and colleagues performed a series of elegant structural studies that compared ZmA produced from B. cereus with synthetic ZmA derivatives that had varied stereocenters (32, 33). From this work, the chemical structure of ZmA with the appropriate stereocenters has been determined (Fig. ​(Fig.1).1). The antibiotic has a number of unusual structural components. First, ZmA is one of only a few linear aminopolyol natural products to be identified. Second, the core of ZmA is formed from ethanolamine and glycolyl moieties that are rarely seen in natural products. Third, the N terminus of ZmA is formed from d-serine (d-Ser), not l-Ser, as initially expected. This suggests that the amino acid either is incorporated as the d isomer or is incorporated as the l isomer and is then isomerized at some point during its biosynthesis. Finally, ZmA is the only natural product that we are aware of that contains an unusual 2-aminosuccinamide moiety. This moiety is likely to come from the amino acid β-ureidoalanine (β-Uda) that has had its carboxylic acid replaced by a terminal amide.Open in a separate windowFIG. 1.Chemical structure of ZmA. Numbers have been added to identify the sites of hydroxyl groups as discussed in the text.We have been investigating how B. cereus UW85 assembles this antibiotic to gain insights into how production of ZmA can be improved and how the unusual structural components of ZmA are formed. We previously proposed that the biosynthesis of ZmA involves the condensation of the amino acids l-Ser and l-2,3-diaminopropionate (l-Dap), along with the carboxylic acid precursors malonyl-coenzyme A (CoA), (2S)-aminomalonyl-acyl carrier protein (ACP), and (2R)-hydroxymalonyl-ACP (4, 12). The proposal for l-Ser incorporation was made prior to the full elucidation of the stereochemistry of ZmA. The condensation of amino acids and carboxylic acids suggests that ZmA is assembled via a nonribosomal peptide synthetase (NRPS) and polyketide synthase (PKS) megasynthase. These megasynthases are modular enzymes with a set of catalytic domains, or modules, for each precursor incorporated into the natural product (reviewed in references 13 and 40). Support for the involvement of this type of enzymology in ZmA biosynthesis comes from a combination of genetic and biochemical studies. Transposon mutagenesis of B. cereus UW85 identified insertions in genes coding for NRPS and PKS enzymology that abolished ZmA production (12). Sequencing of a 19-kb fragment of the ZmA biosynthesis gene cluster identified genes coding for NRPS and PKS enzymology (12). Furthermore, other groups investigating ZmA biosynthesis in B. thuringiensis strains have identified genes coding for NRPS modules that are essential for ZmA biosynthesis in these strains (35, 49, 50). Finally, we used biochemistry and mass spectrometry to establish the existence of two ACP-linked PKS extender units, (2S)-aminomalonyl-ACP and (2R)-hydroxymalonyl-ACP (4). All of these data support the hypothesis that the backbone of ZmA is assembled by an NRPS/PKS megasynthase.In addition to the mixed amino acid and carboxylic acid backbone, ZmA also contains a terminal amide (Fig. ​(Fig.1).1). How these amide groups are formed was investigated by Müller and colleagues and Silakowski and colleagues as they deciphered how myxothiazole is biosynthesized (29, 36). Briefly, the NRPS/PKS megasynthase that assembles the backbone of myxothiazole forms a product that is 1 amino acid longer than myxothiazole. This results in a biosynthetic intermediate that contains a glycyl residue at the C terminus of myxothiazole, while the intermediate remains thioesterified to the peptidyl carrier protein (PCP) domain of the terminal NRPS module. The α-carbon of the glycine is hydroxylated by a flavin-dependent monooxygenase, a modification that results in an unstable intermediate that spontaneously releases the myxothiazole backbone, with the nitrogen of the terminal amide coming from the glycine. The terminal PCP domain contains the glyoxyl group left after C-N bond cleavage, and this product is released from the PCP domain by the neighboring thioesterase (Te) domain. Based on this precedent, the terminal amide of ZmA may be produced by a similar mechanism.Here, we present the identification of the complete ZmA biosynthesis gene cluster from B. cereus UW85. The biosynthesis gene cluster was identified by locating the previously reported biosynthesis genes and by mapping the locations of transposon insertions that abolished the ability of B. cereus UW85 to produce ZmA. As expected, the gene cluster codes for NRPS and PKS enzymology that is likely to be involved in ZmA assembly from its amino acid and carboxylic acid precursors. Surprisingly, we fiound that ZmA not only is likely to be processed at its C terminus to generate the terminal amide by a mechanism similar to that seen in myxothiazole biosynthesis, but it appears to also be processed at its N terminus. These two processing events potentially lead to the biosynthesis of two additional metabolites besides ZmA. Furthermore, the kanosamine biosynthesis gene cluster appears to be fully contained within the ZmA biosynthesis gene cluster. A mechanism for ZmA production is presented, along with proposals for how three additional metabolites are produced by the enzymes encoded by this unusual gene cluster.