INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Clavulanic acid is a potent beta-lactamase inhibitor produced by the gram-positive actinomycete Streptomyces clavuligerus (27). During growth in soy fermentation medium, S. clavuligerus produces clavulanic acid and a group of structurally related but antipodal clavam metabolites, as well as conventional penicillin-type and cephamycin-type beta-lactam antibiotics. Some of the clavams possess antibacterial and antifungal activities, but they lack the beta-lactamase inhibitory activity of clavulanic acid (24). Unlike the penicillins and cephamycins, all clavam metabolites including clavulanic acid possess an oxygen-containing oxazolidine ring, and feeding studies have established that they are biosynthetically distinct from the sulfur-containing beta-lactam antibiotics (11, 15).

Clavulanic acid biosynthesis is thought to begin with the joining of arginine to a three-carbon glycolytic intermediate, likely pyruvate (25, 31, 34). The product of this condensation, N²-(2-carboxyethyl)-arginine, is then cyclized via -lactam synthetase to form the monocyclic -lactam product, deoxyguanidinoproclavaminic acid (2, 21). Deoxyguanidinoproclavaminic acid is hydroxylated by the multifunctional enzyme CAS to form guanidinoproclavaminic acid (14), which is then converted to proclavaminic acid via proclavaminate amidinohydrolase (37). This is followed by the CAS-mediated oxidative cyclization of proclavaminic acid, which forms clavaminic acid (9). The final steps of the biosynthetic pathway are poorly understood, but they most likely involve the oxidative deamination of clavaminic acid and stereochemical inversion (epimerization) at C-3 and C-5 to form the highly reactive clavulanate-9-aldehyde (22). This penultimate intermediate is then reduced by way of an NADPH-dependent dehydrogenase to clavulanic acid.

Clavaminate synthase is the best characterized of the clavulanic acid biosynthetic enzymes. It exists in S. clavuligerus as two isoenzymes, CAS1 and CAS2, that are encoded by two similar but distinct genes, cas1 and cas2, respectively (20). Sequence analysis, hybridization studies, and insertional inactivation experiments have placed the clavulanic acid biosynthetic genes in a cluster directly adjacent to the genes required for penicillin and cephamycin biosynthesis on the S. clavuligerus chromosome (1, 16, 36). In addition, these studies have shown that cas2 is located within the cluster immediately downstream of the proclavaminate amidinohydrolase-encoding gene, pah (1, 16, 37). The relative position of cas1 on the S. clavuligerus chromosome has yet to be determined, but hybridization studies have suggested that cas1 is separated from cas2 by at least 20 to 30 kb (3, 20).

The biological rationale for the possession of two CAS isozymes is unclear. One possibility is that clavulanic acid and the clavam metabolites are synthesized by separate pathways, each requiring its own CAS isozyme (20). However, a number of studies have strongly suggested that clavulanic acid and the clavams share a common pathway up to and including the clavaminic acid step (8). Another possibility is that one of the CAS-mediated biosynthetic steps is rate limiting. Duplication of an ancestral cas gene may, therefore, have been an evolutionary means of relieving the bottleneck at this point in the biosynthetic pathway (20).

Recent work has shown that the expression of cas1 is nutritionally regulated (23). When wild-type S. clavuligerus was grown in a complex soy fermentation medium, it produced both clavulanic acid and clavams, and Northern analysis revealed that transcripts of both cas1 and cas2 were present. However, in a defined starch-asparagine (SA) medium, S. clavuligerus produced clavulanic acid but failed to produce detectable levels of clavams. Northern analysis showed that only cas2 was transcribed under these conditions; no cas1-specific transcripts were detected. When a cas2 mutant of S. clavuligerus was created by targeted gene disruption, it failed to produce clavulanic acid when it was grown on SA medium. In contrast, the cas2 mutant produced both clavulanic acid and clavams when it was grown in soy medium.

These findings indicate that both cas1 and cas2 contribute to clavulanic acid biosynthesis in soy medium. However, in SA medium, in which cas1 is not expressed, cas2 becomes essential for clavulanic acid production (23). Since the cas2 mutant was capable of producing clavulanic acid and clavams in soy medium, the clavaminic acid generated through the action of CAS1 must have been freely available for the synthesis of both clavulanic acid and clavams. This strongly supports the proposal that clavaminic acid serves as a branch point precursor for both clavulanic acid and the clavams (8). The observed inability of wild-type S. clavuligerus to produce clavams when it is grown on SA medium (despite the presence of a functional cas2) suggests that additional genes specific for the biosynthesis of the clavams are under regulatory controls similar to those that govern the expression of cas1 (23).

In this paper, we report on the cloning, nucleotide sequencing, identification, and insertional inactivation of clavam-specific biosynthetic genes flanking cas1 on the S. clavuligerus chromosome. In addition, mutants disrupted either in cas1 alone or in both cas1 and cas2 were constructed by targeted gene disruption and replacement. The characteristics of these mutants with respect to clavulanic acid and clavam production are described.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. All bacterial strains and plasmids used in this study are described in Table 1. Plasmid pFDNEO-S was generously provided by R. Brzezinski (4). Cultures of Escherichia coli were maintained as described by Sambrook et al. (28). Cultures of S. clavuligerus and Streptomyces lividans were maintained as described by Paradkar and Jensen (23). For the selection and maintenance of plasmid-containing cultures, agar-based media were supplemented with ampicillin (100 µg/ml), kanamycin (100 µg/ml), thiostrepton (5 µg/ml for S. clavuligerus and 50 µg/ml for S. lividans), apramycin (25 µg/ml), or neomycin (55 µg/ml), as appropriate. For the production of clavulanic acid and/or clavam metabolites, a seed culture of S. clavuligerus was grown in Trypticase soy broth supplemented with 1% maltose on a rotary shaker (220 rpm) at 28°C for 48 h and was used to deliver a standardized inoculum to either SA medium or soy medium (23). Production cultures were grown under conditions similar to those used for seed cultures.

Chemicals. Thiostrepton was obtained from Sigma Chemical Co., St. Louis, Mo. Apramycin was generously provided by E. Seno, Eli Lilly Co., Indianapolis, Ind. Samples of clavulanic acid, clavaminic acid, and clavam-2-carboxylate for use as high-performance liquid chromatography (HPLC) standards were obtained from SmithKline Beecham, Worthing, United Kingdom.

Recombinant DNA techniques. Genomic and plasmid DNAs were extracted from Streptomyces spp. by standard techniques (12). Preparation and transformation of S. clavuligerus protoplasts were as described by Paradkar and Jensen (23). S. lividans protoplasts were prepared and transformed by standard procedures (12). For DNA sequencing, plasmid DNA was extracted from E. coli with a QIAGEN Plasmid Mini Kit (Qiagen Inc., Santa Clarita, Calif.). For the routine screening of plasmids from E. coli, standard procedures were used (28). All DNA manipulations and the transformation of competent E. coli cells were as described by Sambrook et al. (28). DNA fragments separated by agarose gel electrophoresis were eluted and purified with a QIAquick Gel Extraction Kit (Qiagen Inc.).

Isolation of cas1. A chromosomal DNA fragment carrying the cas1 open reading frame from S. clavuligerus was isolated with a 36-mer oligonucleotide, RMO1 (5'-GCGCGCAGCTCGGGGCCGTACGCGGTGCAGTCCACT). The RMO1 sequence, based on the noncoding strand of cas1 at nucleotides (nt) +9 to +44 relative to the start codon, was chosen as the probe because it possessed only 11% sequence identity with the corresponding region of the cas2 gene (20, 23).

DNA sequencing of the S. clavuligerus chromosome flanking cas1. A partial restriction map of cosmid 10D7 was generated with restriction endonucleases SacI, NcoI, and KpnI. Southern hybridization experiments between RMO1 and various digests of 10D7 DNA indicated that cas1 was located at one end of a 7-kb SacI DNA subfragment (data not shown). This fragment consisted of cas1 and approximately 6 kb of upstream DNA. The 7-kb fragment was then subcloned in the phagemid vector pBluescript II SK+ from a SacI digest of 10D7, generating the recombinant plasmid pCEC007.

View larger version (86K): [in this window] [in a new window]   FIG. 1.   Nucleotide and deduced amino acid sequences of the 5' end of cas1 and 2.6 kb of upstream DNA. The first nucleotide of each potential translational start codon is marked above, with an arrowhead indicating the direction of the open reading frame. Potential translational stop codons are indicated with an asterisk beneath the codon. The location of the BsaBI site used to create the cvm1 disruption mutation is shown above the nucleotide sequence.

View larger version (75K): [in this window] [in a new window]   FIG. 2.   Nucleotide and deduced amino acid sequences of the 3' end of cas1 and 3.6-kb of downstream DNA. The cas1-encoding portion of the presented sequence was determined for one strand only, but it agreed exactly with that previously reported by Marsh et al. (20). The sequences of both strands of the remaining 3.6 kb of DNA were determined as described in Materials and Methods. Potential translational start and stop codons are marked as indicated in the legend to Fig. 1. The locations of the two NcoI sites used to create the cvm4 and cvm5 deletion mutation are indicated above the nucleotide sequence.

Insertional inactivation of cvm1 by gene replacement. Gene disruption and replacement of cvm1 were accomplished basically as described by Paradkar and Jensen (23) with a bifunctional apramycin resistance gene (apr) cassette. The apr cassette was inserted into pCEC026 at the unique BsaBI site located within cvm1 636 bp from the translational start codon (Fig. 3). Constructs carrying the apr cassette inserted in the same orientation relative to cvm1 were isolated, ligated to the high-copy-number Streptomyces vector pIJ486 (35), and then introduced by transformation into S. lividans and from there into S. clavuligerus. The transformants were allowed to sporulate on nonselective medium and were then replica plated to identify putative mutants.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Partial restriction endonuclease map of cas1 and flanking regions from the S. clavuligerus chromosome. The thick bar at the top of the diagram represents the S. clavuligerus chromosome, and the arrows immediately below the bar represent the predicted open reading frames and the proposed direction of their transcription. The lower portion of the diagram depicts the DNA inserts from the plasmid constructs used to create the cvm1, cvm4 and cvm5, and cas1 mutants. Arrows immediately above the rectangular boxes indicate the orientation of the resistance cassette. Brackets surrounding the EcoRI site indicate that the site is vector derived and does not exist in the S. clavuligerus genome. Abbreviations: B, BsaBI; E, EcoRI; K, KpnI; N, NcoI; S, SacI.

Creation of a deletion mutation in cvm4 and cvm5 by gene replacement. A deletion mutant in cvm4 and cvm5 was created by digesting pCEC018 with NcoI to liberate a 1-kb subfragment containing most of cvm4 and a portion of cvm5 (see Fig. 3). The deleted fragment was replaced with the apr cassette oriented in the direction opposite that of cvm4 and cvm5. The plasmid was then fused with pIJ486 and was introduced by transformation into wild-type S. clavuligerus via S. lividans. The resulting transformants were allowed to sporulate under nonselective conditions and were then replica plated to identify putative mutants. Southern analysis of BstEII-digested genomic DNA from wild-type S. clavuligerus and from the mutant with pCEC029 (Table 1) as the probe showed a 4.5-kb hybridizing band in the wild type and a 2.3-kb hybridizing band in the mutants (data not shown). These results were consistent with the deletion from the S. clavuligerus chromosome of a 1-kb DNA fragment that included portions of both cvm4 and cvm5 and its replacement with the apr cassette.

Insertional inactivation of cas1 by gene replacement. A neomycin and kanamycin resistance gene (neo) cassette was subcloned on a 1-kb SacI-HindIII fragment from pFDNEO-S into pBluescript II SK+, creating pSKNeo. The neo cassette was then excised from pSKNeo as a 1-kb KpnI fragment and was inserted into pCEC007 at the KpnI site within cas1 (Fig. 3). Clones possessing recombinant plasmids carrying the neo cassette in both possible orientations within cas1 were fused to pIJ702, and these fusions were then introduced by transformation into S. lividans and from there into wild-type S. clavuligerus. Putative mutants identified after sporulation on nonselective medium were subjected to Southern analysis.

Creation of a cas1 and cas2 mutant by gene replacement. The same plasmid constructs used to generate the cas1 mutant (see above) were introduced by transformation into S. clavuligerus AP5, a mutant strain carrying the cas2 gene disrupted by insertion of apr (23). Gene disruption and replacement were confirmed by Southern analysis as described above.

HPLC. The levels of clavulanic acid and clavam metabolites were analyzed by HPLC as described by Paradkar and Jensen (23), except that the running buffer that was used consisted of 0.1 M NaH₂PO₄ plus 6% methanol (pH 3.68; adjusted with glacial acetic acid). Peaks representing clavulanic acid, clavaminic acid, and clavam-2-carboxylate in culture broths of S. clavuligerus were identified by comparison with authentic standards. 2-Hydroxymethylclavam was identified by its known chromatographic properties relative to those of clavulanic acid and clavam-2-carboxylate (3).

Estimation of beta-lactam antibiotic, clavulanic acid, and alanylclavam production. Bioassays for bioactive clavam compounds were carried out as described by Paradkar and Jensen (23), except that Klebsiella pneumoniae ATCC 29665 instead of Staphylococcus aureus N2 was used as the indicator organism in the bioassay for clavulanic acid.

Nucleotide sequence accession numbers. The nucleotide sequence of the region upstream of cas1 reported here has been deposited in the GenBank and EMBL databases under accession no. AF124928, while the region downstream of cas1 has been deposited under accession no. AF124929.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Sequence analysis of open reading frames flanking cas1. CODONPREFERENCE and FRAME analysis (data not shown) of the sequenced DNA from the region of the S. clavuligerus chromosome flanking cas1 predicted the presence of two complete open reading frames and one incomplete open reading frame upstream of cas1. All of the open reading frames were located on the opposite DNA strand relative to cas1 and were thus oppositely oriented (Fig. 3). The first open reading frame, cvm1, began at an ATG codon located 579 bp upstream of cas1 and encodes a putative 344-amino-acid polypeptide (Fig. 1). cvm1 was preceded by a potential ribosome-binding site at nt 2075 to 2079. Further examination of the DNA sequence immediately upstream of the cas1 and cvm1 open reading frames revealed a number of direct repeats, the complexity and number of which were most pronounced in the region 100 to 200 bp upstream of cas1 (data not shown).

Characterization of the cvm1 mutant. Although sequence information was obtained for part or all of the six open reading frames surrounding cas1, it was not evident from the deduced protein sequences whether any of these open reading frames might play a role in clavam biosynthesis. However, since the details of clavam biosynthesis beyond the level of clavaminic acid are essentially unknown, there was also no basis to eliminate these open reading frames as potential clavam biosynthetic genes. To gain further insights into the possible involvement of cvm1 in clavam biosynthesis, mutants were prepared by a gene replacement procedure. A cloned copy of cvm1 was disrupted by the introduction of apr oriented in the same direction as cvm1. The cvm1 disruption construct was then introduced by transformation into S. clavuligerus, where the disrupted gene exchanged with the corresponding wild-type region of the chromosome by a homologous recombination procedure. The mutants that arose from four different primary transformants (apparent frequency of mutation, 2 to 25%) were characterized.

View larger version (15K): [in this window] [in a new window]   FIG. 4.   HPLC analysis of culture filtrates from cvm1 and cvm4/cvm5 mutants of S. clavuligerus. (A) Wild-type culture filtrate. (B) cvm1 mutant culture filtrate. (C) cvm4 and cvm5 mutant culture filtrate. Peak 1, clavaminic acid; peak 2, clavam-2-carboxylic acid; peak 3, 2-hydroxymethylclavam; peak 4, clavulanic acid.

Characterization of the cvm4 and cvm5 deletion mutant. To explore the possibility that cvm4 and cvm5 were also clavam biosynthetic genes, a cvm4 and cvm5 double mutant was created. In that mutant portions of both genes were deleted and replaced with apr oriented in the direction opposite those of the cdo genes. The cvm4 and cvm5 deletion construct was introduced by transformation into S. clavuligerus, in which it exchanged with the corresponding wild-type region of the chromosome by a homologous recombination procedure. Mutants were isolated from three independent primary transformants (apparent frequency of mutation, 1 to 6%).

Characterization of the S. clavuligerus cas1 mutant. In previous studies, a cas2 disruption mutant had been shown to be conditionally defective in clavulanic acid production, with production occurring on soy medium but not on SA medium. Since this nutritional regulation was associated with cas1, a cas1 disruption mutant was prepared to examine the effect of the mutation on clavam and clavulanic acid production. The cloned cas1 gene, which was disrupted by insertion of a neo cassette in both orientations relative to the orientation of cas1, was introduced by transformation into S. clavuligerus, in which it exchanged with the resident wild-type gene at a low frequency to generate cas1 disruption mutants. Four mutants arising from two primary transformants were subjected to further analysis.

Characterization of the S. clavuligerus cas1 and cas2 mutant. To complete the analysis of the cas mutants, a cas1 and cas2 double mutant was prepared by introducing the cas1 disruption construct (described above) into the cas2 mutant created previously (23). In these double mutants, each cas gene is disrupted by the insertion of an antibiotic resistance marker: neo in cas1 and apr in cas2. The cas1 and cas2 mutants were grown in SA and soy media, and samples of culture supernatants were analyzed by HPLC at 70 and 93 h for the production of clavulanic acid, clavam-2-carboxylate, and 2-hydroxymethylclavam and were analyzed for cephamycin C and alanylclavam production by bioassay. All four of the cas1 and cas2 disruptants, the wild type, and strain AP5 produced similar amounts of cephamycin C. As previously observed (23), the wild type produced clavulanic acid in both SA and soy media but produced clavam metabolites only in soy medium. As expected, all four of the cas1 and cas2 mutants failed to produce clavulanic acid or any of the other clavam metabolites, including alanylclavam, when they were grown in either medium.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Although a basic outline for the biosynthesis of clavulanic acid exists, little is known about the production of the antipodal clavams by S. clavuligerus and Streptomyces antibioticus, and even less is understood about the clavamycins produced by Streptomyces hygroscopicus (18). Feeding studies with S. clavuligerus showed that ¹³C-labelled ornithine was incorporated with equal efficiency into both clavulanic acid and clavam-2-carboxylate. Similar results were obtained when ¹³C-labelled proclavaminate was fed to S. clavuligerus (13). Feeding studies have also shown that clavaminic acid is a precursor of clavulanic acid in S. clavuligerus (10), and when doubly labelled clavaminic acid was fed to cultures of S. antibioticus, it was incorporated at low but equal efficiencies into both valclavam and 2-(2-hydroxyethyl)clavam (8). These results strongly suggested that clavulanic acid and the antipodal clavams have a common biosynthetic pathway up to and including the clavaminic acid step.

In a previous study, we provided support for the proposal that CAS1 and CAS2 contribute to a common intracellular pool of clavaminic acid that can be shunted toward the synthesis of clavulanic acid and/or the antipodal clavams (23). When cas2 was disrupted by gene replacement, the resulting mutant still produced 2-alanylclavam and clavulanic acid when it was grown in soy medium. This suggested that the cas2 mutation was complemented by cas1 and that the clavaminic acid produced was being used as a precursor for both clavulanic acid and antipodal clavam biosynthesis. A corollary to this proposal is that a cas1 mutation should be complemented by cas2.

In the present study we tested this hypothesis by constructing mutant strains of S. clavuligerus in which cas1 had been disrupted by insertional inactivation with a neo cassette. The resulting cas1 disruptants produced lower levels of clavulanic acid, clavam-2-carboxylate, and 2-hydroxymethylclavam and normal levels of 2-alanylclavam compared to the levels produced by the wild type grown in soy medium. These results demonstrate that the defect in cas1 is at least partially complemented by cas2. To confirm that all of the clavam metabolites required CAS for their synthesis and to eliminate the possibility of a third isozyme of CAS, we created S. clavuligerus double mutants that were disrupted in both cas1 and cas2. These strains failed to produce clavulanic acid or any of the clavam metabolites, including 2-alanylclavam, when they were grown in soy medium or SA medium. These results provided conclusive evidence that CAS1 and CAS2 contribute to a common pool of clavaminic acid that serves as a precursor for the biosynthesis of all clavams produced by S. clavuligerus.

Paradkar and Jensen (23) observed that wild-type S. clavuligerus efficiently produces clavulanic acid in SA medium but is unable to produce detectable levels of 2-alanylclavam. Furthermore, when a cas2 mutant of S. clavuligerus was grown in SA medium, it produced neither clavulanic acid nor 2-alanylclavam. A Northern analysis of RNA extracted from wild-type S. clavuligerus grown in SA medium failed to detect cas1 transcripts but showed high levels of the cas2 transcript. These results showed that cas2 is essential for clavulanic acid production in SA medium, in which cas1 is not expressed. More importantly, however, a correlation was observed between cas1 expression and clavam metabolite production. This correlation might be explained if cas1 and the genes for clavam biosynthesis are coordinately downregulated when S. clavuligerus is grown in SA medium. Alternatively, if the clavaminic acid produced by way of CAS2 was somehow sequestered and could be used only to synthesize clavulanic acid, a similar effect would result.

To extend the work of Paradkar and Jensen (23), we isolated a cosmid clone that contained cas1 and surrounding regions of the S. clavuligerus chromosome. Sequence analysis of the DNA flanking cas1 revealed the presence of several open reading frames, none of which appeared to be similar to the open reading frames immediately upstream or downstream of cas2. To determine if any of the cas1-associated open reading frames were involved in the biosynthesis of clavam metabolites, we generated mutants of S. clavuligerus that were disrupted in either cvm1 or cvm4 and cvm5. An HPLC analysis of the culture supernatants showed that both mutant types failed to produce detectable levels of clavam-2-carboxylate, clavaminic acid, or 2-hydroxymethylclavam in soy medium but produced wild-type amounts of clavulanic acid. Likewise, bioassays of culture supernatants revealed that the mutants were unable to produce 2-alanylclavam but produced wild-type levels of cephamycin C. Thus, the disruption of genes both upstream and downstream of cas1 completely blocked clavam biosynthesis but had no apparent effect on clavulanic acid or cephamycin C production.

A comparison of the deduced amino acid sequences of the genes immediately upstream and downstream of cas1 with sequences found in current protein sequence databases has revealed some similarities. However, the exact role and significance of the proposed products encoded by these genes are uncertain, primarily because little is known about the biochemistry of the pathway leading from clavaminic acid to the various clavam metabolites produced by S. clavuligerus. Egan et al. (8) have proposed a tentative pathway (Fig. 5) in which clavaminic acid is successively decarboxylated and deaminated to produce a branch-point intermediate that is then converted by way of a series of oxidation, hydrolysis, and condensation reactions to clavam-2-carboxylate, 2-hydroxymethylclavam, 2-formyloxymethylclavam, and 2-alanylclavam.

View larger version (12K): [in this window] [in a new window]   FIG. 5.   Pathway for the biosynthesis of clavulanic acid and the antipodal clavams beginning at proclavaminic acid (adapted from Egan et al. [8]). 1, proclavaminic acid; 2, clavaminic acid; 3, clavulanate-9-aldehyde; 4, clavulanic acid; 5, 2-formyloxymethylclavam; 6, 2-hydroxymethylclavam; 7, clavam-2-carboxylic acid; 8, 2-alanylclavam.

In this study, preliminary evidence for a common precursor that serves as a branch-point intermediate to all of the clavam metabolites is provided by the fact that cvm4 and cvm5 disruption mutants fail to produce clavam-2-carboxylate, 2-hydroxymethylclavam, and 2-alanylclavam. This hypothesis is also supported by the lack of clavam production in cvm1 mutant cultures. In the proposed biosynthetic pathway (8), clavaminic acid is decarboxylated and deaminated by way of a pyridoxal 5'-phosphate-dependent enzyme(s). The derived amino acid sequence of cvm6 possessed significant similarity to a variety of pyridoxal 5'-phosphate-dependent transaminases, as well as to the bifunctional 2,2-dialkylglycine decarboxylase from P. cepacia (32). It may therefore be that the cvm6 product plays a part in both the decarboxylation and the deamination of clavaminic acid and commits the product to the synthesis of the clavam metabolites.