INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Coumermycin A₁ is produced by Streptomyces rishiriensis DSM 40489 (13). Together with novobiocin and clorobiocin (Fig. 1), it belongs to the coumarin antibiotics. Novobiocin (Albamycin; Pharmacia & Upjohn) is licensed in the United States as an antibiotic for the treatment of infections with multiresistant gram-positive bacteria, such as Staphylococcus epidermidis and Staphylococcus aureus (27, 34, 47).

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Structures of the coumarin antibiotics coumermycin A1, novobiocin, and clorobiocin.

Bacterial DNA gyrase is the target of the aminocoumarin antibiotics (25). X-ray crystallographic examinations (8, 20, 45, 48) have demonstrated that the aminocoumarin moiety and the substituted deoxysugar moiety of these compounds are essential for their binding to the gyrase B subunit of bacterial gyrase. The coumermycin A₁ molecule (Fig. 1) contains two of these active aminocoumarin-deoxysugar moieties and has been shown to stabilize a dimer form of the 43-kDa fragment of GyrB (1, 10). Therefore, coumermycin A₁ is likely to cross-link the two gyrase B subunits of the intact gyrase heterotetramer, which consists of two gyrase A and two gyrase B subunits. Consequently, the affinity of this antibiotic for intact gyrase is extremely high: 50% inhibition of gyrase is achieved by coumermycin A₁ at a concentration of only 0.004 µM, compared to 0.1 µM for novobiocin, 1.8 µM for norfloxacin, and 110 µM for nalidixic acid (32). Likewise, coumermycin A₁ has been found to exhibit much higher antibacterial activity than novobiocin (36).

In coumermycin A₁, a methylpyrrole-2-carboxylic acid unit is attached to the 3-OH of each deoxysugar moiety (Fig. 1). The same pyrrole unit is also contained in clorobiocin (Fig. 1) and has been shown to result in a higher affinity for gyrase than has the carbamoyl group found in the corresponding position of novobiocin (45). In contrast, the prenylated 4-hydroxybenzoate group of novobiocin and clorobiocin, which is absent in coumermycin A₁, appears not to be essential for biological activity (2, 11, 18). These features make coumermycin A₁ a most interesting starting compound for the development of new aminocoumarin antibiotics, which may serve as anti-infective agents against multiresistant gram-positive bacteria. Recently, the chemical synthesis of a series of new aminocoumarin antibiotics has been published (9, 18, 19, 31, 33).

Combinatorial biosynthesis with the biosynthetic gene clusters for the aminocoumarin antibiotics could provide additional possibilities for the discovery of novel anti-infective agents. The genetic manipulation of the biosynthesis of polyketide and peptide antibiotics has already succeeded in the production of new and even clinically useful "hybrid" antibiotics (4, 14, 16, 26, 28, 39, 42).

Knowledge of the biosynthetic genes for aminocoumarin biosynthesis is a prerequisite for the production of new aminocoumarin antibiotics by combinatorial biosynthesis. Our group has recently reported the identification of the novobiocin biosynthetic gene cluster (41). Molecular biological studies of coumermycin A₁ biosynthesis are yet unpublished. We present now the cloning and sequencing of the coumermycin A₁ biosynthetic gene cluster from S. rishiriensis DSM 40489 and its functional identification by insertional gene inactivation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and culture conditions. The bacterial strains and plasmids used in this study are listed in Table 1. S. rishiriensis DSM 40489 was cultivated at 28°C and 175 rpm for 2 to 4 days in baffled flasks. For isolation of chromosomal DNA, the organism was grown in liquid medium containing 1.0% malt extract, 0.4% yeast extract, 0.4% glucose, and 1.0 mM CaCl₂ (pH 7.3). For preparation of protoplasts, S. rishiriensis was grown for 44 to 48 h in CRM medium, containing 10.3% sucrose, 2.0% tryptic soy broth, 1.0% MgCl₂ · 6H₂O, 1.0% yeast extract, and 0.4% glycine (pH 7.0). Protoplasts were prepared as described by Steffensky et al. (41) and regenerated on R2YE medium (12).

DNA isolation and manipulation. Standard methods for DNA isolation and manipulation were performed as described by Hopwood et al. (12) and Sambrook et al. (37). DNA fragments were isolated from agarose gels using a QIAEX II gel extraction kit (Qiagen, Hilden, Germany). Isolation of cosmids and plasmids was carried out with ion-exchange columns (Nucleobond AX kits; Macherey-Nagel, Düren, Germany) according to the manufacturer's protocol.

Construction and screening of the cosmid library. Chromosomal DNA of S. rishiriensis DSM 40489 was partially digested with Sau3AI, dephosphorylated, ligated into cosmid vector pOJ446 (which had been digested with HpaI), dephosphorylated, and restricted with BamHI. The ligation products were packaged with Gigapack III XL (Stratagene, Heidelberg, Germany) and transduced into E. coli XL1 Blue MRF'.

DNA sequencing and computer-assisted sequence analysis. Restriction fragments of approximately 300 to 3,000 bp from cosmids 4-2H and 4-7D were subcloned into pBluescript SK(). Sequencing was performed by the dideoxynucleotide chain termination method on a LI-COR automatic sequencer (MWG-Biotech AG, Ebersberg, Germany).

Construction of the vector for insertional gene inactivation. Vector pZK4, for proB disruption, was constructed by insertion of neomycin resistance gene aphII into the sequence of proB as follows. The aphII gene was obtained as a 0.99-kb EcoRI-HindIII fragment from plasmid pNeo4 (Table 1) and ligated into the same sites of pZW331 (Table 1), which contained the 5' region of proB, to give plasmid pZW2. The 3' region of proB was obtained as a 0.96-kb HindIII-XhoI fragment from pZW32 (Table 1) and ligated into the same sites of pZW2, resulting in plasmid pZW3. A 2.65-kb PstI fragment of pZW3, containing the disrupted proB gene, was cloned into the same sites of pBluescript SK() to give the inactivation vector pZK4. In pZK4, the aphII gene fragment had the same orientation as the proB gene and the orientation opposite that of the bla resistance gene of the vector.

Transformation of S. rishiriensis DSM 40489. Transformation of S. rishiriensis with pZK4 was carried out by polyethylene glycol-mediated protoplast transformation. Two grams of S. rishiriensis mycelia was incubated in 7 ml of P buffer (12) containing 1 mg of lysozyme per ml for 20 to 40 min at 30°C. For transformation, pZK4 was propagated in E. coli ET12567 (23), and the resulting double-stranded plasmid DNA was denaturated by alkaline treatment (30). The denaturated DNA (10 to 20 µg) was mixed with 200 µl of P buffer containing 10⁹ S. rishiriensis protoplasts (200 µl) and with 500 µl of T buffer (12) containing 25% (wt/vol) polyethylene glycol 1000 (Roth, Karlsruhe, Germany). The resulting suspension was plated on R2YE plates. After 16 to 20 h at 25°C, the plates were overlaid with 3 ml of soft nutrient agar (12) containing neomycin (33.3 µg/ml) for selection of integration mutants.

Determination of the production of coumermycin A₁ and other secondary metabolites in S. rishiriensis. Metabolites of S. rishiriensis were analyzed by modifications of the methods of Kawaguchi et al. (13) and Berger and Batcho (5).

Nucleotide sequence accession number. The nucleotide sequence reported in this study is available in the GenBank database under accession no. AF235050.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of the coumermycin A₁ biosynthetic gene cluster. We have previously cloned the novobiocin biosynthetic gene cluster from S. spheroides NCIB 11891 by screening of a cosmid library of the producer with a probe for a dTDP-glucose 4,6-dehydratase gene, which is involved in the biosynthesis of the deoxysugar moiety of the antibiotic (41). Since coumermycin A₁ contains the same deoxysugar moiety, we subjected genomic DNA of the coumermycin producer, S. rishiriensis DSM 40489, to Southern hybridization with the same probe. A single hybridizing band was detected. Likewise, hybridization with a probe from the novobiocin resistance gene gyrB^r, encoding a non-novobiocin-sensitive gyrase B subunit (44), resulted in a single hybridizing band. Therefore, a cosmid library of the coumermycin producer was established in vector pOJ446 and screened with both probes.

Sequencing of the cosmids and identification of ORFs. From a core region of 30.8 kb, both strands were sequenced. This sequencing revealed the presence of 28 complete open reading frames (ORFs) upstream of the aminocoumarin resistance gene gyrB^r (Fig. 2). The cluster showed striking similarity to the novobiocin biosynthetic gene cluster: 15 of the identified ORFs were found to have, on average, 84% identity to corresponding ORFs of the novobiocin cluster at the amino acid level (Table 2), and all of these ORFs were arranged in both clusters in identical order (Fig. 2). Table 2 shows the homologies found between the genes of the coumermycin A₁ and novobiocin clusters, as well as homologies to other GenBank entries.

View larger version (14K): [in this window] [in a new window]   FIG. 2.   Comparison of the biosynthetic gene clusters for coumermycin A1 (A) and novobiocin (B). The genes which are homologous in both clusters are shown as black arrows.

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Hypothetical biosynthetic pathway of coumermycin A1. CoA, coenzyme A.

Insertional inactivation of the proB gene in S. rishiriensis. The striking similarity between the gene cluster identified in this study and the previously identified gene cluster for novobiocin biosynthesis made it very likely that we had indeed cloned the coumermycin biosynthetic gene cluster. Functional proof for this hypothesis was provided by an insertional gene inactivation experiment. The gene proB, located in the central region of the cluster and possibly involved in pyrrole biosynthesis (see above), was chosen for this experiment.

View larger version (29K): [in this window] [in a new window]   FIG. 4.   Insertional gene inactivation in the coumermycin A1 biosynthetic gene cluster. (A) Schematic presentation of the gene replacement. The 0.87-kb DNA fragment used as a probe is indicated as a thin black bar. Relevant restriction sites: P, PstI; B, BamHI; X, XhoI; E, EcoRI. aphII, neomycin resistance gene. (B) Southern blot analysis of the proB mutants. Lanes 1, PstI; lanes 2, BamHI; lanes 3, XhoI. Expected bands: wild type (WT), 1.66 kb (PstI), 2.45 kb (BamHI), and 2.66 kb (XhoI); mutants resulting from gene replacement; 2.65 kb (PstI), 3.44 kb (BamHI), and 3.65 kb (XhoI).

View larger version (108K): [in this window] [in a new window]   FIG. 5.   TLC analysis of secondary metabolites in S. rishiriensis strains. Lane 1, coumermycin A1 standard; lane 2, extract of DSM 40489 (wild type); lanes 3 and 4, extracts of proB mutants ZW20 and ZW21, respectively. The plate (Silica Gel 60 F254) was developed with dichloromethane-methanol-formic acid (45:2:1). Blue spots on a yellow background were observed in daylight after spraying with 10% fresh ferric chloride-potassium ferricyanide (5 g each in 100 ml of water).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The coumarin antibiotics (Fig. 1) are closely related to each other in their chemical structure. The biosynthetic gene cluster of novobiocin, the best-known member of this group, has recently been identified (41). Coumermycin A₁ is of special pharmaceutical interest due to its pronounced antibacterial activity and its extremely high affinity for bacterial gyrase (32, 36).

In the present study, we have cloned and sequenced the biosynthetic gene cluster of coumermycin A₁ from S. rishiriensis DSM 40489. The cluster showed striking similarity to the novobiocin biosynthetic gene cluster (Fig. 2). In both clusters, nearly all of the ORFs are oriented in the same direction. At the 3' end of each cluster, a gene encoding a coumarin-resistant gyrase B subunit is located. This gene has been described previously as the principal novobiocin resistance gene in the novobiocin producer S. spheroides (44), and expression of the gyrB^r resistance gene from the coumermycin cluster in Streptomyces lividans showed that it also conferred resistance to both novobiocin and coumermycin A₁ (E. Schmutz, unpublished data).

Immediately upstream of the resistance gene, both clusters contained five highly homologous genes in identical order (cumUVWXY and novSTUVW). Based upon their homology to known genes of deoxysugar biosynthesis, we previously assigned these genes to the first five steps required for the biosynthesis of the deoxysugar moiety of novobiocin (Fig. 1 and 3), and functional proof for this hypothesis was provided by an inactivation experiment with novT (41). Coumermycin A₁ contains the same deoxysugar moiety as novobiocin, and in the coumermycin A₁ biosynthetic gene cluster, exactly the same deoxysugar biosynthetic genes were found. This finding provides additional support to our functional assignment of these genes (Fig. 3).

One of the last steps of novobiocin biosynthesis is the O methylation of the 4-OH of the deoxysugar moiety, and we have putatively assigned the gene novP to this reaction. The same O-methylation reaction is required in coumermycin A₁ biosynthesis; indeed, a gene homologous to novP, i.e., cumN, was found at the corresponding position of the coumermycin cluster. The gene novO, immediately upstream of novP, had been assigned to the C-methylation reaction at position 8 of the coumarin ring of novobiocin, and this assignment is now supported by the presence of a highly homologous gene, cumM, likely to carry out the same reaction in coumermycin biosynthesis.

Attachment of the deoxysugar to the 7-OH of the aminocoumarin ring requires very similar glycosyltransferases in novobiocin and coumermycin biosyntheses; indeed, two very similar glycosyltransferase genes, novM and cumH, are found at the same relative positions of both clusters.

In novobiocin (Fig. 1), the aminocoumarin ring and the substituted benzoate ring are linked by an amide bond. Recently, the gene novL has been functionally identified, by overexpression and purification, as encoding the amide synthetase responsible for both adenylation of the substituted benzoyl moiety and its transfer to the amino group (40). In the structure of coumermycin, two corresponding amide bonds are present, linking the two aminocoumarin rings to a central 3-methylpyrrole-2,4-dicarboxylic acid moiety. The gene cumG shows high homology to novL and is located at the same relative position of the gene cluster. It is most probably involved in the formation of these amide bonds. It cannot be decided at present, however, whether CumG catalyzes the adenylation and transferase reactions for the formation of both amide bonds in the coumermycin molecule or whether additional gene products are involved in these reactions.

The gene cumC displays distinct homology to peptide synthetase genes, and its deduced amino acid sequence shows the presence of the typical conserved motifs of peptide synthetases described by Marahiel et al. (24), including the 4-phosphopantetheinyl attachment site required for covalent binding of the acyl substrate in the form of a thioester (data not shown). cumC is very similar to novH, which is located at the same relative position of the novobiocin cluster. We have recently proven that novH is not involved in the formation of the amide bond between the aminocoumarin ring and the substituted benzoate ring of novobiocin, and we have speculated that NovH may catalyze the activation of tyrosine, or a derivative thereof, during the biosynthesis of one of the two aromatic rings of novobiocin (40). The presence of a very similar gene in the cluster for coumermycin, which contains the aminocoumarin gene but not the substituted benzoate ring, suggests that novH and the corresponding cumC may be involved in the biosynthesis of the aminocoumarin ring found in both antibiotics.

Immediately downstream of cumC is found the gene cumD, which shows homology to genes for cytochrome P-450 enzymes: the very recently described NikQ catalyzes the -hydroxylation of histidine during nikkomycin biosynthesis (17), and the product of ORF20 of the chloroeremomycin biosynthetic gene cluster (46) may be involved in the -hydroxylation of tyrosine (17). The biosynthesis of the aminocoumarin moieties of novobiocin and coumermycin, which are derived from tyrosine (15, 21), requires the introduction of an oxygen at the -position of tyrosine. The cytochrome P-450 enzyme CumD and the corresponding enzyme NovI of the novobiocin cluster are likely candidates for the catalysis of this reaction.

The ring oxygen of the aminocoumarin moiety of novobiocin has been shown to be derived from the carboxy group of tyrosine, rather than from molecular oxygen (7), suggesting the formation of the aminocoumarin ring by a unique oxidative cyclization mechanism rather than by ortho-hydroxylation of tyrosine followed by simple lactonization. The formation of the aminocoumarin ring may therefore require the oxidation of a (hypothetical) -hydroxytyrosine derivative to a -ketotyrosine derivative and subsequent oxidative cyclization. CumE (and the corresponding NovJ) show homology to 3-ketoacyl-(ACP) reductase and may catalyze the first of the two oxidation steps. The adjacent CumF (and the corresponding NovK) share homology to redox enzymes and may catalyze the oxidative cyclization step.

The genes for the biosynthesis of the characteristic aminocoumarin ring of the aminocoumarin antibiotics must be present in both the novobiocin and the coumermycin clusters, and a comparison of both clusters is therefore an obvious method for identifying possible candidate genes for the biosynthesis of this moiety. This comparison leads to the suggestion that the gene products of cumCDEF corresponding to those of novHIJK may catalyze the formation of the aminocoumarin ring in a reaction sequence such as that shown in Fig. 3, i.e., activation of tyrosine and covalent binding in the form of a thioester, -hydroxylation, oxidation to a -ketoacyl intermediate, and oxidative cyclization. Further experiments are now in progress to provide functional proof for this hypothesis and to identify whether the C-methylation reaction catalyzed presumably by CumM (Fig. 3) occurs before or after coumarin ring formation. It is noteworthy that in the chloroeremomycin cluster, a gene with homology to a peptide synthetase gene, ORF19, is situated immediately upstream of the P-450 gene ORF20, suggesting that -hydroxylation may require prior activation of tyrosine in the biosynthesis of chloroeremomycin as well. Likewise, -hydroxylation of free histidine could not be demonstrated after heterologous expression of the P-450 enzyme NikQ mentioned above, and the involvement of an additional enzyme was postulated (17).

NovN has been suggested to catalyze the transfer of the carbamoyl group to the deoxysugar moiety of novobiocin (Fig. 1). Coumermycin does not contain this carbamoyl group, and no gene with similarity to novN is found in the coumermycin cluster. However, at the same relative position is found the gene cumJ, which shows homology to acyltransferase genes and which may catalyze the transfer of the pyrrole-2-carboxylic acid moieties to the deoxysugar moieties of coumermycin A₁. Immediately downstream of cumJ are found the genes proA, proB, and proC, which share homology with genes supposedly involved in the conversion of proline to pyrrole-2-carboxylic acid in the biosynthesis of pyoluteorin in P. fluorescens Pf-5 (29). These genes may be tentatively assigned to the biosynthesis of the pyrrole-2-carboxylic acid moieties of coumermycin A₁ (Fig. 3). The incorporation of proline into coumermycin A₁ has been demonstrated previously (38). Further experiments are now in progress to elucidate the exact mechanism of the biosynthesis of these pyrrole rings.

For several other genes in the coumermycin cluster, so far no function can be suggested, e.g., for cumOPQRT. Some of these genes may be involved in the formation of the central 3-methylpyrrole-2,4-dicarboxylic acid unit of coumermycin.