ABSTRACT
A biosynthetic 3-hydroxy-3-methylglutaryl coenzyme A reductase (EC1.1.1.34 ), the rate-limiting enzyme of the mevalonate pathway for isopentenyl diphosphate biosynthesis, had previously been purified fromStreptomyces sp. strain CL190 and its corresponding gene (hmgr) had been cloned (S. Takahashi, T. Kuzuyama, and H. Seto, J. Bacteriol. 181:1256–1263, 1999). Sequence analysis of the flanking regions of the hmgr gene revealed five new open reading frames, orfA to -E, which showed similarity to those encoding eucaryotic and archaebacterial enzymes for the mevalonate pathway. Feeding experiments with [1-13C]acetate demonstrated that Escherichia coli JM109 harboring the hmgr gene and these open reading frames used the mevalonate pathway under induction with isopropyl β-d-thiogalactopyranoside. This transformant could grow in the presence of fosmidomycin, a potent and specific inhibitor of the nonmevalonate pathway, indicating that the mevalonate pathway, intrinsically absent in E. coli, is operating in the E. coli transformant. The hmgr gene andorfABCDE are thus unambiguously shown to be responsible for the mevalonate pathway and to form a gene cluster in the genome ofStreptomyces sp. strain CL190.
Isoprenoids, found in all organisms, play important roles such as steroid hormones in mammals, carotenoids in plants, and ubiquinone or menaquinone in bacteria (28). All these isoprenoids are synthesized by consecutive condensations of the five-carbon monomer isopentenyl diphosphate (IPP). It was generally believed that IPP is synthesized only by condensation of three molecules of acetyl coenzyme A (CoA) through the mevalonate pathway (Fig. 1A). This ubiquitous pathway consists of six enzymes: acetoacetyl-CoA synthase, 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) synthase, HMG-CoA reductase, mevalonate kinase, phosphomevalonate kinase, and pyrophosphomevalonate decarboxylase.
Mevalonate and nonmevalonate pathways for IPP biosynthesis. Fosmidomycin is a potent and specific inhibitor of DXP reductoisomerase in the nonmevalonate pathway. The reactions leading to IPP from CDP-ME2P remain unknown. MVA, mevalonate; PMVA, phosphomevalonate; DPMVA, diphosphomevalonate; TPP, thiamine diphosphate.
There is an extensive body of information concerning the mevalonate pathway in eucaryotes including rats, mice, and yeast. However, very few studies on the enzymes or genes for this pathway in an additional kingdom, Eubacteria, are available. Recently, it has turned out that the mevalonate pathway is absent in many eubacteria includingEscherichia coli and Bacillus subtilis(27). Instead of the mevalonate pathway, these eubacteria use the nonmevalonate pathway. The initial step of this newly identified pathway is the formation of 1-deoxy-d-xylulose 5-phosphate (DXP) by condensation of pyruvate and glyceraldehyde-3-phosphate catalyzed by DXP synthase (2, 14, 18, 20, 36). In the second step DXP is converted to 2-C-methyl-d-erythritol 4-phosphate (MEP) by DXP reductoisomerase as demonstrated by us previously (15, 16, 38). MEP is then cytidylylated by MEP cytidylyltransferase to give 4-(cytidine 5′-diphospho)-2-C-methyl-d-erythritol (CDP-ME) (12, 26), which is phosphorylated by CDP-ME kinase to afford 2-phospho-4-(cytidine 5′-diphospho)-2-C-methyl-d-erythritol (CDP-ME2P) (13, 21). Recently we succeeded in cloning MEP cytidylyltransferase (12) and CDP-ME kinase (13) genes from E. coli and demonstrated unequivocally that these enzymes constitute the nonmevalonate pathway (Fig. 1B). The subsequent reactions leading to IPP from CDP-ME2P, however, remain unknown.
Feeding experiments with [14C]acetyl-CoA, [14C]HMG-CoA, and [14C]mevalonate have demonstrated the involvement of the enzymes for the mevalonate pathway in some eubacteria such as Myxococcus fulvus,Lactobacillus plantarum, and Staphylococcus carnosus (7). Recent achievements in bacterial genome sequencing, however, revealed that all the eubacteria except forStaphylococcus aureus(http://www.sanger.ac.uk /Projects/S_aureus/) and the Lyme disease spirochaete Borrelia burgdorferi (5) utilized only the nonmevalonate pathway for isoprenoid biosynthesis, whereas four archaebacteria, Methananococcus jannaschii(3), Methanobacterium thermoautotrophicum(35), Pyrococcus horikoshii (10), andArchaeoglobus fulgidus (11), used only the mevalonate pathway. On the other hand, feeding experiments with13C-labeled acetate have proved that isoprenoids and hemiterpenoid metabolites such as naphterpin (31, 32), furaquinocin (6), napyradiomycin (34), and terpentecin (9) produced by the genusStreptomyces are synthesized through the mevalonate pathway. No genes encoding enzymes responsible for the mevalonate pathway, however, have been identified from these eubacteria except for the HMG-CoA reductase genes from naphterpin (37), furaquinocin (4), and terpentecin producers (4). Thus, our attention has been directed to the cloning of all other genes involved in the mevalonate pathway from naphterpin producerStreptomyces sp. strain CL190.
In this paper we describe the cloning and functional analyses of the five open reading frames (ORFs) newly found in the flanking regions of the HMG-CoA reductase gene (hmgr) fromStreptomyces sp. strain CL190. A gene cluster containing thehmgr gene and these ORFs was heterogeneously expressed inE. coli, and a labeling pattern of ubiquinone from thisE. coli transformant grown in the presence of [1-13C]acetate was analyzed with nuclear magnetic resonance (NMR). This paper is the first to describe the functional assignment of eubacterial mevalonate pathway genes expressed inE. coli.
MATERIALS AND METHODS
Materials.[1-13C]acetate (isotopic abundance, 99%) was purchased from ICON (New York, N.Y.). Fosmidomycin was a gift from Fujisawa Pharmaceutical Co., Ltd. All restriction enzymes, T4 DNA polymerase, and pUC118 were purchased from Takara Shuzo, Kyoto, Japan.
Bacterial strains and media. Streptomyces sp. strain CL190 was cultivated as described previously (33). E. coli JM109 was used as a cloning host and a host for thelacZ expression system. JM109 was grown in Luria-Bertani (LB) broth as described by Sambrook et al. (29). A minimum medium, M9 (29), which was used in the assay for fosmidomycin resistance, was always supplemented with 20 μg of thiamine/ml to meet the thiamine requirement of JM109.
Sequence analysis of the flanking regions of the hmgrgene.Total DNA from CL190, which was prepared as described previously (29), was partially digested withSau3AI, followed by size fractionation by agarose gel electrophoresis. DNA fragments larger than 20 kb were ligated to aBamHI- and phosphatase-treated pWE15 cosmid vector (Stratagene) to give a cosmid library of CL190. This cosmid library was screened by colony hybridization with a DNA fragment containing the CL190 hmgr gene (37) as a probe. Hybridization to this probe was found with 17 transformants. Seventeen cosmid clones, designated pCLC1 to pCLC17, were prepared from these positive transformants and digested with various restriction enzymes, and the positions of the hybridizing regions were defined by Southern hybridization (29) with the same probe. A series of plasmids, constructed by subcloning various hybridized DNA fragments into pUC118, were sequenced by a method described below.
Plasmid construction.A 6.7-kb fragment digested withSnaBI from cosmid pCLC13 was blunt ended with T4 DNA polymerase and then inserted into HincII- and alkaline phosphatase-treated pUC118 so that the ORFs were located in the same transcriptional direction as the lacZ promoter. E. coli JM109 was transformed with the resulting plasmid, pUMV19. This plasmid was digested with MluI, the recognition sites of which were in the targeted orfD, and then self-ligated to give deletion plasmid pUMV19ΔM (Fig.2). Thus, this deletion plasmid lacked only orfD.
Organization of ORFs in the 6.7-kbSnaBI-SnaBI fragment. The arrows indicate the extents and directions of the ORFs found in the flanking regions of thehmgr gene of Streptomyces sp. strain CL190. The arrowheads indicate the direction of the lacZ promoter in pUC118. pUMV19ΔM, derived from pUMV19, lacks only orfD.
DNA sequence analysis.DNA sequences were determined by the dideoxy chain termination method (30) with an automated sequencer (model 4000L; Li-cor) and the protocol of the supplier. A FramePlot 2.3 program (8) was used for searching ORFs. The FASTA program (19, 23) performed a homology search of protein databases. Amino acid sequences aligned by the GENETYX program (Software Development, Tokyo, Japan) were then edited visually to align consensus motifs.
Purification of ubiquinone and its NMR analysis. E. coli JM109(pUMV19) was grown at 37°C in 1 liter of LB medium containing 0.1% sodium [1-13C]acetate, 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside), and 50 μg of ampicillin/ml for 14 h. Ubiquinone (coenzyme Q8) was purified from this transformant as described previously (24) and was further purified by high-performance liquid chromatography with a PEGASIL octyldecyl silane column (4.6 by 250 mm; Senshu Scientific Co., Tokyo, Japan). Then purified ubiquinone was analyzed by13C-NMR (A-500; JEOL, Tokyo, Japan). As a control, ubiquinone purified from E. coli JM109(pUC118), which was grown under the same conditions as the transformant in the absence of [1-13C]acetate, was also analyzed.
Determination of 13C enrichment.Relative enrichments for all carbon atoms of the labeled ubiquinone were obtained by comparison of 13C integrals with those of natural-abundance standards (1.1%). The intensities of two signals at 61 ppm in 13C spectra for ubiquinone from both JM109 and JM109(pUMV19) were aligned. The signals were assigned to methoxy carbons in the ubiquinone molecule. These signals were utilized as standards for the evaluation of isotopic abundance.
Nucleotide sequence accession number.The nucleotide sequence of a 6.7-kb SnaBI-SnaBI fragment including the hmgr gene and orfABCDE has been deposited in the DDBJ, EMBL, and GenBank nucleotide sequence databases with accession no. AB037666 .
RESULTS AND DISCUSSION
DNA sequence analysis of the flanking regions of thehmgr genes.A 6.7-kb SnaBI-SnaBI DNA fragment including the flanking regions of the hmgr gene from Streptomyces sp. strain CL190 was sequenced. A FramePlot 2.3 program revealed the presence of five new ORFs,orfA to -E, in this DNA fragment. Organization of all ORFs in this 6.7-kb fragment is shown in Fig. 2. After a search of the SWISS-PROT database using a FASTA program, the products oforfA, orfB, orfC, and orfEwere deduced to be mevalonate kinase (EC 2.7.1.36 ), diphosphomevalonate decarboxylase (EC 4.1.1.33 ), mevalonate kinase (EC 2.7.1.36 ), and HMG-CoA synthase (EC 4.1.3.5 ), respectively (Table1). This search using the FASTA program, however, could not distinguish between the products of orfAand orfC; those of both ORFs showed homology with mevalonate kinase. In addition to mevalonate kinase, the mevalonate pathway requires phosphomevalonate kinase (EC 2.7.4.2 ) for phosphorylation of phosphomevalonate (Fig. 1). To reveal which of the two ORFs encodes phosphomevalonate kinase, the amino acid sequences encoded byorfA and orfC were compared with those of phosphomevalonate kinases from Saccharomyces cerevisiae(accession no. P24521 ) and Schizosaccharomyces pombe(accession no. AL109739 ). This homology search using a GENETYX program revealed that the product of orfC showed significant homology with the S. cerevisiae (26.8% identity in 153 amino acids) and S. pombe (20.6% identity in 180 amino acids) enzymes. In addition, the product of orfA showed homology (24.9% identity in 217 amino acids) with mevalonate kinase from S. pombe (accession no. Q09780 ). The products oforfC and orfA were thus deduced to be phosphomevalonate kinase and mevalonate kinase, respectively.
Deduced functions of orfA to orfE
The product of orfD showed significant homology with a hypothetical protein, ORF6, in Erwinia herbicola Eho10, which produced carotenoids (1). The associated gene,orf6, had been reported to be located in the carotenoid biosynthetic gene cluster in this organism. Since a frameshift mutation in the ORF did not affect carotenoid production (1), its biochemical function remained unclear. The function of orfDin the mevalonate pathway remains undefined in this study as well (see below).
NMR analyses of ubiquinone from an E. coli transformant harboring pUMV19.To elucidate the functions of these ORFs, the 6.7-kb SnaBI-SnaBI fragment was expressed by thelacZ system in E. coli and a feeding experiment with [1-13C]acetate was carried out on this transformant. The ability of this transformant to utilize the mevalonate pathway was proved by this feeding experiment.
The 13C-NMR spectrum of ubiquinone labeled with sodium [1-13C]acetate showed enrichments at the C-1 (isotopic abundance, 1.6%) and C-3 (isotopic abundance, 4.6%) positions of eight prenyl units in the ubiquinone molecule (Fig.3). These C-1 and C-3 carbons were derived from C-1 and C-3 of the IPP molecule, respectively (Fig. 1). On the other hand, no enrichments were found in the C-2, C-4, and C-5 positions of these prenyl units. The very large differential incorporation of the labeled acetate into mevalonate-derived ubiquinone needs to be explained, since to the best of our knowledge an unusual phenomenon such as this has never been reported. C-1 and C-3 of prenyl units of the ubiquinone side chain derive from C-1 and C-3, respectively, of acetoacetyl-CoA (Fig. 1), which unlike what occurs in the usual mevalonate pathway, is assumed to be formed by condensation of acetyl-CoA and malonyl-CoA in order to explain this unusual incorporation pattern. Since no gene of an acetoacetyl-CoA synthase (EC2.3.1.9 ) from Streptomyces sp. strain CL190 was introduced in the transformant which intrinsically lacked the mevalonate pathway, this organism should have used its own acetoacetyl-CoA synthase from fatty acid biosynthesis for the formation of acetoacetyl-CoA. Interestingly, E. coli possesses two sequences of acetoacetyl-CoA synthase (accession no. P76461 and Q46936 ) showing 57.9% identity to each other in 392 amino acids. The roles of these enzymes in acetoacetyl-CoA synthesis, however, remain undefined. The labeling pattern described above clearly suggested that [1-13C]acetate was efficiently incorporated into C-3 of mevalonate, corresponding to C-3 of IPP, via acetoacetyl-CoA. This labeled precursor was incorporated into C-5 of mevalonate, corresponding to C-1 of IPP, presumably with dilution by internal nonlabeled malonyl-CoA. An alternative explanation is to assume low activity of acetyl-CoA carboxylase to form malonyl-CoA from acetyl-CoA in E. coli. Further experiments are needed to uncover this unusual phenomenon. In any case, since two 13C atoms from two molecules of [1-13C]acetate incorporated through the mevalonate pathway label C-1 and C-3 of the IPP molecule, the labeling patterns of ubiquinone in this experiment thus proved the operation of the mevalonate pathway in E. coli JM109(pUMV19).
13C-NMR spectra of ubiquinone from JM109 and the transformant harboring pUMV19. Ubiquinone was purified from JM109 (lower), and its transformant harboring pUMV19 was grown in LB medium containing 0.1% sodium [1-13C]acetate under induction with 0.1 mM IPTG. The intensities of two signals at 61 ppm in both13C-NMR spectra were aligned. These signals were utilized as standards for the evaluation of isotopic abundance.
An E. coli transformant harboring pUMV19 was resistant to fosmidomycin.We next investigated whether E. coli JM109(pUMV19) could grow in the presence of fosmidomycin, a potent and specific inhibitor of DXP reductoisomerase in the nonmevalonate pathway (17). If this transformant could utilize the mevalonate pathway for IPP biosynthesis, it should grow in the presence of this inhibitor. E. coli JM109(pUC118) did not grow on minimum agar medium M9 containing 20 μg of fosmidomycin/ml (lower left and lower right in Fig.4). On the other hand, E. coliJM109(pUMV19) could grow on the M9 medium containing fosmidomycin when there was induction with IPTG (lower right in Fig. 4). These results unambiguously established that E. coliJM109(pUMV19) harboring orfA to -E utilized the mevalonate pathway for IPP biosynthesis. orfA to -E have thus been proved to supply IPP in E. coli. The role of orfD, however, remains to be clarified.
Phenotypes of JM109 and the transformants. All plates are shown after overnight incubation at 37°C. Minimum medium M9 used in this experiment was always supplemented with 20 μg of thiamine/ml to meet the thiamine requirement of E. coli JM109. JM109, JM109(pUMV19), and JM109(pUMV19ΔM) could not grow on the M9 plate containing 20 μg of fosmidomycin (FMM)/ml (lower left), but JM109(pUMV19) and JM109(pUMV19ΔM) could grow on the plate under induction with 0.1 mM IPTG (lower right). On the lower-right plate, the growth of JM109(pUMV19ΔM) was slower than that of JM109(pUMV19).
orfD is necessary for full operation of the mevalonate pathway.In order to gain insight into the function oforfD, pUMV19ΔM, lacking only orfD, was constructed from pUMV19. E. coli JM109 was transformed with pUMV19ΔM, and then the ability of the resulting transformant to grow in the presence of fosmidomycin was investigated. This transformant could grow on the M9 medium plate containing fosmidomycin only under induction with IPTG, but its growth seemed to be slower than that of E. coli JM109(pUMV19) (Fig. 4, lower right). The same phenomenon was also observed in a liquid M9 medium containing fosmidomycin; E. coli JM109(pUMV19) grew to an optical density at 660 nm of 2.5 (full growth), whereasE. coli JM109(pUMV19ΔM) grew to an optical density of 1.0 or less. These results indicated that orfD was not essential for operation in E. coli of the mevalonate pathway but was necessary for full operation of the pathway. A gene cluster containing orfABCE and the hmgrgene is thus shown to be essential for the operation of the mevalonate pathway in E. coli. Since wild-type E. coli has acetyl-CoA carboxylase and acetoacetyl-CoA synthase genes in the genome, it is not necessary to introduce these enzyme genes for the operation of the mevalonate pathway in E. coli.
Conclusion.In conclusion, it has been suggested thatorfA, orfB, orfC, andorfE encode mevalonate kinase, diphosphomevalonate decarboxylase, phosphomevalonate kinase, and HMG-CoA synthase, respectively, and that they form a gene cluster together with thehmgr gene in the genome of Streptomyces sp. strain CL190. We demonstrated in this study that the E. colitransformant harboring this gene cluster could utilize the mevalonate pathway for IPP biosynthesis in addition to the nonmevalonate pathway. The transformant has proved to be a useful strain for preparing mutants possessing a metabolic block(s) in the nonmevalonate pathway (12, 13), because the associated mutations could be complemented by the expression of the hmgr gene and orfABCDEcloned in this study.
ACKNOWLEDGMENTS
M. Takagi and T. Kuzuyama contributed equally to this work.
We thank Fujisawa Pharmaceutical Co., Ltd., for a sample of fosmidomycin.
This work was supported in part by a Grant-in-Aid for Scientific Research (B) from The Ministry of Education, Science, Sports and Culture, Japan (10460047), to H.S. and by a Grant-in-Aid for Encouragement of Young Scientists from The Japan Society for the Promotion of Science (11760086) to T.K.
FOOTNOTES
- Received 2 March 2000.
- Accepted 8 May 2000.
- Copyright © 2000 American Society for Microbiology
