INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Streptomyces clavuligerus produces an array of -lactam compounds including cephamycin C, clavulanic acid, and at least four other clavam metabolites. Clavulanic acid has considerable chemotherapeutic and economic value because of its -lactamase-inhibitory activity. In contrast, the other clavam metabolites are ineffective as -lactamase inhibitors. These clavam metabolites have the same nuclear structure, a fused bicyclic -lactam-oxazolidine ring system, as does clavulanic acid. However, the stereochemistry of the ring system is 5R in the clavams rather than 5S as in clavulanic acid, and the clavam metabolites also have different side chain substituents at C-2 (Fig. 1).

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Structures of the clavam metabolites produced by Streptomyces clavuligerus. (A) Clavulanic acid; (B) other clavam metabolites.

The biosynthetic pathway leading to clavulanic acid has been partially elucidated (Fig. 2), and evidence has been obtained to indicate that both clavulanic acid and the antipodal clavam metabolites share a common pathway, at least to the level of proclavaminate or clavaminate (5, 9). The structural similarity between clavulanic acid and the other clavam metabolites, together with the evidence for a shared pathway, has led to the assumption that these compounds comprise a family of biosynthetically related metabolites.

View larger version (22K): [in this window] [in a new window]   FIG. 2.   Biosynthetic pathway leading to clavulanic acid and the other clavams. CEAS, N-carboxyethylarginine synthase; BLS, -lactam synthetase; PAH, proclavaminate amidinohydrolase; CAS, clavaminate synthase; CAD, clavaldehyde dehydrogenase.

The genes encoding enzymes involved in clavulanic acid biosynthesis are known to be clustered in S. clavuligerus and located adjacent to the genes involved in cephamycin C biosynthesis (1, 7, 23). In early studies, we localized the gene that encodes proclavaminate amidinohydrolase (pah) and showed that clavulanic acid production on starch asparagine (SA) medium was lost when pah was disrupted (1). Involvement in clavulanic acid biosynthesis was further substantiated when the gene encoding one of the isozymes of clavaminate synthase (cas2 [11]) was found to be located immediately downstream of pah. In subsequent studies, additional genes involved in clavulanic acid biosynthesis have been located through sequence analysis of DNA on either side of pah and cas2. The gene orf2 (S. E. Jensen, K. A. Aidoo, and A. S. Paradkar, U. S. patent application 08/134,018) has very recently been shown to encode a carboxyethylarginine synthase (10) responsible for condensing 3-phosphoglyceraldehyde with arginine as the first dedicated step in the clavulanic acid biosynthetic pathway. The gene bls (7) encodes a -lactam synthetase which forms the monocyclic -lactam ring at an early stage in the pathway (2, 12), while the gene cad (18) encodes clavaldehyde dehydrogenase, which oxidizes clavaldehyde to clavulanic acid as the final step in the pathway (14). Late pathway steps involved in the inversion of stereochemistry of the bicyclic ring nucleus and modification of the C-2 side chain have yet to be elucidated, and candidate genes have not been identified. Similarly, the biosynthetic enzymes and corresponding genes responsible for the formation of the clavam metabolites are unknown.

An intriguing aspect of this biosynthetic gene cluster is the observation that cas2 is only one of a pair of paralogous biosynthetic genes that encode very similar enzymes (11). The location of cas1, the second of this pair of genes, is unknown, but it is more than 25 kb away from cas2. The rationale for having two separate genes encoding isozymes with very similar substrate specificities and kinetic properties is unclear. However, the two genes are known to be regulated differently, since cas1 is expressed in a soy-based (Soy) medium but not in SA medium, whereas cas2 is expressed in both media (17). Hypotheses put forward to explain the existence of two genes have included the suggestion that two genes may be required to ensure sufficient supplies of this potentially rate-limiting enzyme, since clavaminate synthase is responsible for three separate steps in the pathway. Alternatively, it has been suggested that one form of Cas may be associated with the biosynthesis of clavulanic acid while the second form is involved in the biosynthesis of other clavam metabolites. In this regard, Mosher et al. (13) have recently shown that the DNA sequence flanking cas1 contains a number of open reading frames (ORFs) and that disruption of these genes interferes specifically with the production of the clavam metabolites, while clavulanic acid production is unaffected.

In the present study we show that, like cas, several more of the genes involved in early steps of clavulanic acid biosynthesis have paralogues located elsewhere in the genome and that these paralogues are associated with the biosynthesis of the clavam metabolites rather than with that of clavulanic acid.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. S. clavuligerus NRRL 3585 was obtained from the Northern Regional Research Laboratories, Peoria, Ill., and maintained either on MYM agar (21) or on ISP medium no. 3 (Difco, Detroit, Mich.). Escherichia coli MV1193 (25), used as the host strain for all of the cloning and subcloning experiments, was grown in Luria broth (LB [19]) or on LB agar plates containing ampicillin (50 µg/ml) or tetracycline (10 µg/ml). The indicator organism E. coli ESS was obtained from A. L. Demain, Massachusetts Institute of Technology, Cambridge), Staphylococcus aureus N2 was obtained from the Department of Biological Sciences, University of Alberta, and Bacillus sp. strain ATCC 27860 was obtained from the American Type Culture Collection, Manassas, Va.

DNA isolation, manipulation, and Southern analyses. Plasmid and genomic DNA preparations from Streptomyces spp. were isolated as described by Hopwood et al. (8). Protoplasts of S. clavuligerus were prepared, transformed, and regenerated as described previously (17). Plasmid DNA isolation from E. coli cultures, restriction enzyme digestion and analysis, and ligation and transformation of E. coli were all performed using standard techniques (19).

DNA sequencing and analysis. Sequence information was obtained for approximately 15 kb of S. clavuligerus DNA extending downstream from pcbC and ending at a BglII site. Ordered sets of deletions were generated (19) using fragments of the DNA insert from the cosmid clone K6L2 (1) and subcloned into the E. coli plasmids pUC118 and pUC119. The deletion-generated fragments were sequenced in both orientations by the dideoxynucleotide chain termination method of Sanger et al. (20), using Sequenase (version 2.0) DNA polymerase and commercially available universal primers. Areas of compression in the sequence band pattern were relieved by carrying out reactions using 7-deaza-dGTP in place of dGTP. The nucleotide sequence data were analyzed for the presence of restriction sites, ORFs, and codon usage by the PC-gene programme (Intelligenetics Corp.). Similarity searches were accomplished using the online BLAST program at the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/BLAST/) and the online FASTA program available through European Bioinformatics Institute (http://www2.ebi.ac.uk/fasta3).

Insertional inactivation of ORFs in the 15 kb of DNA sequence. In order to determine the roles of the various ORFs contained within the 15 kb of DNA, mutants were constructed by a gene replacement procedure based on that described by Paradkar and Jensen (17).

HPLC analysis of clavulanic acid and other clavam metabolites. High-pressure liquid chromatography (HPLC) analyses were performed as described previously (13). In brief, samples of culture supernatants were derivatized with an imidazole reagent that reacts specifically with -lactam-containing compounds. Derivatized samples were analyzed by HPLC under conditions which allow detection of clavulanic acid and the other clavam metabolites; they were compared to authentic standards of clavulanic acid, clavaminic acid, and clavaminic acid-2-carboxylate (all generously provided by SmithKline Beecham Pharmaceuticals). Although no standard was available for 2-hydroxymethylclavam, the relevant peak could also be identified based on its chromatographic properties relative to the known standards. Alanylclavam was not eluted under these chromatographic conditions and could be detected only by bioassay.

Bioassays of -lactam antibiotics and of alanylclavam. Bioassays were conducted by the agar plate disk diffusion method with appropriate indicator organisms. Total antibiotic was bioassayed using E. coli ESS as the indicator organism, clavulanic acid was bioassayed using S. aureus N2 as the indicator organism, and alanylclavam was bioassayed using Bacillus sp. strain ATCC 27860 as the indicator organism (17).

Nucleotide sequence accession number. The sequence data obtained for approximately 15 kb of S. clavuligerus DNA have been deposited in GenBank under accession no. AF205427.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Nucleotide sequence of the clavulanic acid gene cluster. The location of the clavulanic acid biosynthetic genes adjacent to the cephamycin gene cluster in S. clavuligerus was first reported by Ward and Hodgson (23). In parallel to these studies, we mapped pah to a region about 5.7 kb downstream of pcbC, the gene encoding isopenicillin N synthase (1). In view of the tight linkage of pah to the cephamycin cluster, we undertook a sequence analysis of the DNA flanking pah to look for additional genes encoding clavulanic acid biosynthetic enzymes. Ordered sets of deletion subclones were generated which extended from the 3' end of pcbC downstream for a distance of about 15 kb, ending at a BglII site. The deletion-generated subclones were sequenced on both strands, and additional sequence information was obtained to cross all subclone junctions and ensure that no small fragments were inadvertently missed. This sequence information was originally documented as part of a patent application (Jensen et al., U.S. patent application 08/134,018) and has now been deposited in GenBank under accession no. AF205427.

View larger version (6K): [in this window] [in a new window]   FIG. 3.   Arrangement of ORFs in the clavulanic acid region of the supercluster.

Mutation of ORFs involved in clavulanic acid biosynthesis. The involvement of each of the ORFs contained in the 15-kb region in clavulanic acid production was investigated by preparing gene disruption mutants and examining the effect on production. In each case, a cloned copy of the gene of interest was disrupted by insertion of an antibiotic resistance marker. The disrupted gene was then crossed into the chromosome by homologous recombination replacing the wild-type copy of the gene (17). Mutants were isolated based on their antibiotic resistance phenotype and then confirmed by Southern analysis. A diagrammatic representation of the gene disruption strategy is shown in Fig. 4A, using orf2 as an example. Southern analysis of the resulting mutants is shown in Fig. 4B. Genomic DNA isolated both from the wild type and from orf2 mutants was digested with EcoRI/NruI, separated by agarose gel electrophoresis, blotted onto nylon membranes, and then probed with a radiolabeled orf2 DNA fragment. Whereas wild-type DNA showed hybridization to a 2.0-kb EcoRI/NruI fragment, the orf2 mutant showed hybridization to a 3.5-kb fragment. The increased size of the hybridizing fragment corresponded to the 1.5-kb size of the apr cassette used to disrupt the gene. When the same blot was hybridized with a probe derived from the apr cassette, only the 3.5-kb band in the orf2 mutants was labeled. The orf2 mutant is designated orf2::apr to indicate that apr was used to create the disruption. In other cases, a thiostrepton resistance gene (tsr) or a neomycin resistance gene (neo) was used for the disruption, and those mutants are designated accordingly. In this way, orf2::apr, bls::apr, pah::tsr, cas2::apr, orf6::apr, orf7::neo, claR::apr, cad::apr, and orf10::apr mutants were created as described in Materials and Methods and confirmed by Southern analysis.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   Disruption of the orf2 gene. (A) The stippled arrow represents orf2, the solid arrow represents the disrupting apramycin resistance cassette, and the open boxes represent DNA flanking orf2. The solid lines represent chromosomal or plasmid DNA. (B) Southern analysis of genomic DNA from S. clavuligerus orf2::apr mutants M1, M2, and M3 and from the wild-type strain using an orf2-specific probe and an apr-specific probe.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   HPLC analysis of culture filtrates from wild-type and bls::apr strains of S. clavuligerus grown on SA medium. Samples of culture filtrate were analyzed by HPLC either after derivatization with imidazole reagent or without derivatization. The data presented are difference chromatograms in which underivatized traces have been subtracted from the derivatized traces. (A) bls::apr mutant; (B) wild type. CA, clavulanic acid; 2-HMC, 2-hydroxymethyl clavam; C-2-C, clavam 2-carboxylate; CM, clavaminic acid.

View larger version (20K): [in this window] [in a new window]   FIG. 6.   HPLC analysis of culture filtrates from wild-type and bls::apr strains of S. clavuligerus grown on Soy medium. Samples of culture filtrate (diluted fivefold with water in the case of the wild-type culture, undiluted in the case of the bls::apr mutant) were analyzed by HPLC either after derivatization with imidazole reagent or without derivatization. The data presented are difference chromatograms in which underivatized traces have been subtracted from the derivatized traces. (A) bls::apr mutant; (B) wild type. For abbreviations, see the legend to Fig. 5.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A 15-kb stretch of chromosomal DNA located downstream from pcbC in the cephamycin gene cluster of S. clavuligerus was subjected to sequence analysis. Ten ORFs were identified, one associated with self-resistance to penicillins and nine associated with clavulanic acid biosynthesis. orf2, bls, pah, cas2, cad, and claR are structural or regulatory genes already known to be involved in clavulanic acid biosynthesis. orf6 and part of orf7 have been sequenced previously, but it is unclear how they are involved in clavulanic acid biosynthesis (7). Finally, orf10, which resembles a cytochrome P450, is a new ORF that has not been described previously.

Evidence for the importance of this region in clavulanic acid production was obtained by creating mutants with disruptions of each of the ORFs and analyzing their phenotypes. Individual disruption of each of the ORFs resulted in a blockage of clavulanic acid production when the mutants were grown on SA medium, indicating that all were involved in the production of clavulanic acid. However, when the same mutants were grown on Soy medium, clavulanic acid production was restored at least partially for mutants defective in orf2, bls, pah, cas2, and orf6.

This pattern of failure to produce on SA medium, but ability to produce on Soy medium, is the same phenotype that was seen previously for cas2 mutants, and it suggests that, as for cas2, paralogues exist for each of these genes. These paralogues are regulated differently from the genes in the clavulanic acid cluster, such that they are not expressed in SA medium but are expressed in Soy medium (17). The existence of a paralogue for cas2 is well established, since the cas1 gene has been cloned and sequenced, and Cas1 has been purified and characterized. A compelling case can also be made for the existence of paralogues for orf2, bls, and pah, because these genes are known to encode essential enzymes in the clavulanic acid biosynthetic pathway. The fact that orf2, bls, and pah disruption mutants, in which these genes have been destroyed by insertion of antibiotic resistance markers, can still form clavulanic acid and other clavams implies that they must have alternative genes that produce equivalent enzymes. The only other simple explanation that comes to mind is that the complex Soy medium already contains intermediates of the clavulanic acid pathway, thereby obviating the need for all of these biosynthetic enzymes. However, the complex chemical structures of these intermediates (Fig. 3) make such an explanation extremely unlikely. The enzymatic steps catalyzed by Orf2, Bls, and Pah are also sufficiently unusual that it is very unlikely that they could be carried out by other enzymes of broad specificity. The case for the existence of a paralogue of orf6 is less compelling because the role of this gene in the biosynthesis of clavulanic acid and the clavams is unknown. However, the striking similarity in phenotype of all five of these mutant classes (blocked on SA medium but producing on Soy medium) argues for a similar mechanism at work in all cases. Of the group of mutants showing evidence for the presence of paralogues, orf2 is the most ambiguous because production of clavulanic acid and clavams was very poor, even on Soy medium. However, since the orf2 gene product is required for clavulanic acid production, the production even of small amounts of clavulanic acid and clavams in the knockout mutant is possible only if a paralogue exists.

The locations of the paralogues relative to the genes in the cephamycin-clavulanic acid supercluster is unknown, just as the location of cas1 relative to cas2 is unknown. Inspection of the DNA sequence in the region flanking cas1 shows no evidence of candidate paralogues, but only a limited amount of sequence information is available at present (13). However, the same studies have shown that the genes surrounding cas1 are involved in clavam biosynthesis, since disruption of the genes flanking cas1 specifically affects clavam production.

While paralogues apparently exist for orf2, bls, pah, cas2, and orf6, individual disruption of orf7, claR, cad, and orf10 results in mutants in which the defect in clavulanic acid production is not growth medium dependent. These mutants are defective in clavulanic acid production on both SA and Soy media. In a recent study, claR was shown to regulate the transcription of orf7, cad, and orf10, as well as regulating its own transcription (16). In contrast, orf2, bls, pah, cas2, and orf6 were transcribed independently from claR. Since mutants defective in claR were unable to produce clavulanic acid but did accumulate clavaminic acid, this suggests that orf2, bls, pah, cas2, and orf6 are involved in early steps of the clavulanic acid biosynthetic pathway, before the level of clavaminic acid. Certainly, this is known to be the case for orf2, bls, pah, and cas2, but it can only be inferred for orf6, since its role in clavulanic acid biosynthesis is not known. All of the early steps in clavulanic acid biosynthesis leading to clavaminic acid are now understood, and yet the role of orf6 remains unclear. The similarity of the Orf6 protein to acetyltransferase type enzymes suggests intriguingly that an as-yet-unrecognized acetylation step may be involved in the early stages of the pathway.

Based on these results, it appears that the genes encoding early enzymes of the biosynthetic pathway, the part of the pathway that is common to both clavulanic acid and the clavam metabolites, all have paralogues. In contrast, the genes encoding biosynthetic enzymes specific for the steps that convert clavaminic acid into clavulanic acid are unique. The simplest explanation for these data would be that S. clavuligerus contains two biosynthetic gene clusters, one for the production of clavulanic acid and a second for the production of the other clavam metabolites. Inasmuch as these two groups of structurally related compounds share a common biosynthetic pathway to the level of clavaminic acid, mutations in the early steps of one pathway can be compensated by the corresponding paralogue from the other pathway. Mutations in late steps of the pathway specifically knock out production of the metabolite involved, while having little effect on production of the product(s) of the other pathway.