Present addresses: Biochemistry Department, Stanford University, CA, USA. ‡Institut für Mikrobiologie, TU München, Weihenstephaner Berg 3, 85354 Freising, Germany.
Steroids, such as cholesterol, are synthesized in almost all eukaryotic cells, which use these triterpenoid lipids to control the fluidity and flexibility of their cell membranes. Bacteria rarely synthesize such tetracyclic compounds but frequently replace them with a different class of triterpenoids, the pentacyclic hopanoids. The intriguing mechanisms involved in triterpene biosynthesis have attracted much attention, resulting in extensive studies of squalene-hopene cyclase in bacteria and (S)-2,3-oxidosqualene cyclases in eukarya. Nevertheless, almost nothing is known about steroid biosynthesis in bacteria. Only three steroid-synthesizing bacterial species have been identified before this study. Here, we report on a variety of sterol-producing myxobacteria. Stigmatella aurantiaca is shown to produce cycloartenol, the well-known first cyclization product of steroid biosynthesis in plants and algae. Additionally, we describe the cloning of the first bacterial steroid biosynthesis gene, cas, encoding the cycloartenol synthase (Cas) of S. aurantiaca. Mutants of cas generated via site-directed mutagenesis do not produce the compound. They show neither growth retardation in comparison with wild type nor any increase in ethanol sensitivity. The protein encoded by cas is most similar to the Cas proteins from several plant species, indicating a close evolutionary relationship between myxobacterial and eukaryotic steroid biosynthesis.
Myxobacteria are Gram-negative eubacteria that are micropredators or saprophytes in the soil environment (Reichenbach, 1993). They show complex social interactions and morphological differentiation resulting in the production of fruiting bodies. Myxobacteria are able to move by gliding, produce swarm colonies (Reichenbach and Dworkin, 1992; Dworkin, 1996) and have been described as a rich source of secondary metabolites with various biological activities (for reviews, see Reichenbach and Höfle, 1993; 1999). Additionally, Nannocystis excedens and Polyangium sp. have been shown to produce steroids (Kohl et al., 1983), tetracyclic triterpenes that have rarely been found in prokaryotes. In fact, their occurrence has only been reported in one further bacterial species, Methylococcus capsulatus (Bird et al., 1971). In contrast, almost all eukaryotes use steroids in order to control the fluidity and flexibility of their cell membranes (Ourisson and Nakatani, 1994). As an alternative, bacteria often use pentacyclic terpenes of the hopanoid type as building blocks in their membranes (Ourisson et al., 1987; Kannenberg and Poralla, 1999). The biosynthesis of hopanoids and steroids starts from squalene, a linear precursor that is formed by condensation of six isopentenyl units. The biosynthetic steps giving rise to the cyclic products belong to the most complicated reactions in nature catalysed by one single protein, squalene-hopene cyclase (Shc), and different forms of (S)-2,3-oxidosqualene cyclases (Osc) respectively. The latter type of enzyme is responsible for the formation of the protosteryl cation, the first cyclization product of (S)-2,3-oxidosqualene (see Fig. 1; Michal, 1999). A variety of further reactions leads to different cyclic triterpenes, which are the precursors of membrane sterols, such as the zoosterine cholesterol, the mycosterine ergosterol (both derive from lanosterol) and the phytosterine sitosterol (derives from cycloartenol). Triterpenoid cyclases for the biosynthesis of tetracyclic steroid skeletons share a high degree of similarity. In fact, the amino acid residues that promote the formation of cycloartenol, lanosterol and parkeol have been identified (Meyer et al., 2000), and directed evolution generated cycloartenol synthase mutants producing lanosterol (Meyer et al., 2002).
Co-existence of both Shc and Osc activity has been demonstrated in M. capsulatus (Rohmer et al., 1980), and the shc gene has been cloned and characterized from this organism (Tippelt et al., 1998). The authors could not detect an osc gene using hybridization experiments, although M. capsulatus produces both steroids and hopanoids (Bird et al., 1971). Thus, cloning and characterization of a bacterial gene involved in steroid biosynthesis is still an unaccomplished task.
Because two of the three bacterial steroid-producing species described are myxobacteria, we set out to analyse whether the ability to form steroids is widespread among this group of microorganisms. As demonstrated in this report, this is in fact the case. In addition, we aimed at the identification of the key enzyme in triterpene biosynthesis in the model species S. aurantiaca in order to provide data revealing the evolutionary relationship of bacterial and eukaryotic (5)2,3-oxidosqualene cyclases. S. aurantiaca strain Sg a15 was shown to produce cycloartenol, a steroid only reported from plants and algae to date. The gene encoding the cycloartenol cyclase was cloned and characterized, revealing a close evolutionary relationship between bacterial and plant 2,3-oxidosqualene cyclases.
Several species of myxobacteria produce steroids
A total of 88 strains from 11 different genera of the order Myxococcales were tested for the production of steroids, squalene and its dehydroderivatives (Table 1; Zeggel, 1993). All compounds were identified by comparative thin-layer chromatography (TLC) and gas chromatography (GC) and by their mass spectrometry (MS) data. Additionally, nuclear magnetic resonance (NMR) spectra of pure compounds isolated from large-scale cultivations were recorded. Besides the already known compounds, squalene and 8(9)-cholesten-3β-ol, nine additional steroids were identified in Nannocystis sp., most of them in strain Na e485. Stigmatella sp. produce squalene and cycloartenol, and all Cystobacter sp. were found to contain squalene and its dehydroderivatives as well as 8(9),24-lanostadien-3β-ol and cycloartenol. Squalene but no steroids could be detected in Corallococcus sp.. Additionally, di- and tetrahydrosqualene could be found in Angiococcus sp., Melitangium sp. and Archangium sp. respectively. No steroids and only trace amounts of squalene were found in Myxococcus sp., and neither steroids nor squalene were produced by Sorangium sp., P. vitellinum sp. and Chondromyces sp.
Table 1. . Strains used and compounds identified in this study.
. Not detected in Na e624.
. Not detected in Na e646.
. Not detected in Na e645.
Na e1, Na e158, Na e485, Na e584, Na e620, Na e624, Na e629, Na e633, Na e637, Na e641, Na e645, Na e646, Na e651, Na e652, Na e654, Na e656, Na e662, Na e665, Na e666, Na e667, Na e672, Na e674, Na e674, Na e688, Na e689
Cc c133, Cc c222, Cc c385, Cc c666, Cc c667, Cc c668, Cc c679, Cc c680, Cc c681, Cc c686, Cc c697, Cc c699, Cc c742, Cc c743, Cc c744, Cc c745, Cc c746
An d1, An d4
Ar 2280, Ar 4908
Mx f1, Mx f2, Mx f174, Mx f311
Mx s1, Mx s2
Mx v4, Mx v80
Mx x8, Mx x61, DK1622
So ce26, So ce90, So ce377, So ce427, So ce516, So ce635
Pl vt1, Pl vt4, Pl vt5
Cm a2, Cm a5, Cm a14
The highest amounts of steroids and squalene were found in Nannocystis sp. and Me. lichenicola Me l26 with 1.5–2.2% and 0.7% of total dry cell weight respectively. Whereas the total amount of the summarized steroids in the investigated Nannocystis strains did not show high variations, the percentage of the individual steroids was different from strain to strain (Fig. 2). In most cases, 7-cholesten-3β-ol followed by 8(9)-cholesten-3β-ol were the main steroids in Nannocystis sp. with the corresponding cholestadienols as minor components (see Fig. 2, strains Na e1, Na e666, Na e667 and Na e684). However, this pattern changed in strain Na e654, which produces barely any 8(9)-unsaturated steroids, and strain Na e646, which does not produce reduced side-chain steroids at all.
Biosynthesis and inhibition studies
To prove that the identified terpenoids are biosynthesized de novo, feeding experiments with 14C-labelled acetate or mevalonate were carried out, and terpenoids were analysed for the incorporation of radioactivity. At least one strain was investigated from every genus producing squalene, steroids or both, and all showed incorporation of the precursors indicating de novo biosynthesis.
Furthermore, detailed feeding experiments with various myxobacterial compounds have revealed leucine to be a good precursor of isoprenoids in myxobacteria (H. B. Bode and R. Müller, unpublished results), which is presumably because of the degradation of leucine to 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA; Anke and Dieckmann, 1971; Michal, 1999; Ginger et al., 2001). Accordingly, 14C-labelled leucine was incorporated efficiently into cycloartenol in S. aurantiaca Sg a15.
To gain further insight into myxobacterial steroid biosynthesis, inhibition studies with well-known inhibitors of isoprenoid and steroid biosynthesis were conducted with S. aurantiaca Sg a15 and Nannocystis excedens Na e485. At concentrations of 2 µM, fluvastatin, a clinically used inhibitor of the eukaryotic HMG-CoA reductase (Manzoni and Rollini, 2002), affected neither growth nor production of cycloartenol and squalene in S. aurantiaca Sg a15, whereas complete growth inhibition was observed at concentrations of 20 µM, probably as a result of unspecific effects of the inhibitor. Treatment with terbinafin (up to 50 mg l−1), a clinically used inhibitor of fungal squalene epoxidase (Balfour and Faulds, 1992), did not result in the expected accumulation of squalene. Instead, an accumulation of 8(9)-unsaturated sterols and an overall increase in the steroid content of N. exedens Na e485 was observed (Fig. 3). After treatment with miconazole, an inhibitor of the P450-dependent C14-demethylation in fungi (Hitchcock, 1991), the expected accumulation of lanosterols was observed (at 0.8 mg l−1 miconazole: Fig. 4). The previously described (Hitchcock, 1991) accumulation of 4-methyl and 4,4-dimethyl sterols at 0.6 mg l−1 miconazole, which were normally only detected in traces, was also observed. No change in the steroid pattern or growth effects were observed with AMO 1618, an inhibitor of eukaryotic oxidosqualene cyclase (Douglas and Paleg, 1978), at concentrations up to 10 µM.
Cloning and mutagenesis of the cycloartenol synthase (cas) of S. aurantiaca Sg a15
A DNA fragment encoding part of a putative open reading frame (ORF) with homology to oxidosqualene cyclases was detected on plasmid pCBS85 during our screening for secondary metabolism genes in S. aurantiaca Sg a15 (cf. Beyer et al., 1999; Silakowski et al., 2000a; 2001). Owing to the fact that S. aurantiaca was not known to make steroids, we analysed the strain for its biosynthetic capacity and detected cycloartenol as a triterpenoid produced (see above). In order to prove that the cas gene found is involved in cycloartenol biosynthesis, gene inactivation was performed as described previously (Beyer et al., 1999). A 439 bp fragment of the gene was amplified and cloned into a vector containing a kanamycin resistance gene (pCBS95). This knock-out plasmid was electroporated into S. aurantiaca Sg a15, kanamycin-resistant colonies were selected and analysed for correctness of the integration site after homologous recombination by Southern blot experiments (data not shown). The integration leads to a gene duplication of cas, with the resulting two copies of the gene carrying either a 3′ or a 5′ deletion. Thus, both copies were assumed to be non-functional. This assumption was tested by analysing cycloartenol production in mutants CBS95-1 and CBS95-2. They were indeed shown to be cycloartenol negative by TLC analysis of cell extracts after large-scale cultivation. Next, growth of the mutants was compared with wild type, which revealed that loss of steroid biosynthesis does not lead to slower growth rates in the medium tested. Wild-type and mutant cells were analysed using transmission electron microscopy, which provided no indication of a difference in membrane structure.
Development of S. aurantiaca Sg a15 and its descendants
The S. aurantiaca strain used in this study (Sg a15) only forms aggregates but no fruiting bodies. This is presumably because of the loss of developmental factors during many passages in liquid culture. Nevertheless, aggregation and swarming can be monitored in a fruiting body assay as has been described for S. aurantiaca DW4/3-1 (Plaga et al., 1998). Using this assay, no difference between wild type and mutants became obvious.
Influence of temperature changes, ethanol concentration and light on the sterol pattern
Nannocystis excedens Na e485 was cultivated at 30°C and 37°C for 2 and 3 days, but no significant change could be observed in the overall steroid content or in the steroid composition. The influence of the steroid content on ethanol tolerance was investigated with both cycloartenol non-producing mutants of S. aurantiaca Sg a15 (mutant strains CBS95-1 and CBS95-2) and a kanamycin-resistant S. aurantiaca Sg a15 reference strain (NGS 12; Gaitatzis et al., 2002 ) in order to minimize antibiotic-dependent growth differences. The strains were cultivated with concentrations of 0%, 0.1%, 0.25%, 0.5%, 1.0% and 2.0% ethanol, and no difference in growth could be detected between sterol-producing and non-producing strains. Up to 0.5% ethanol, growth was not influenced at all, whereas at 1%, a significant decrease was observed in the cell number. Almost no growth could be seen after the addition of 2% ethanol. Also, fatty acid analyses of cells from these cultures according to previously published procedures ( Mahmud et al., 2002 ) showed no differences in the fatty acid pattern after growth in different ethanol concentrations (data not shown). Furthermore, there was no difference in the steroid content for strain Sg a15 when grown in light compared with growth in darkness (data not shown), as described for carotenoid production in some myxobacteria ( Reichenbach, 1993 ).
Analysis of the DNA region isolated, and phylogeny of triterpene cyclases
The complete sequence of the cas gene from S. aurantiaca Sg a15 was determined after a vector recovery strategy from chromosomal DNA of mutant CBS95. The method described in Experimental procedures allows the rapid cloning of the genomic sequence of cas, together with adjacent nucleotide stretches. The analysis of the determined sequence reveals that the cas gene is preceded by an unfinished ORF with significant similarity to farnesyl diphosphate farnesyl transferases (squalene synthases). The ORF (of which a 300-amino-acid sequence can be predicted) is 31% identical and 43% similar at the amino acid level to the squalene synthase from M. capsulatus (Tippelt et al., 1998) and 26% identical and 39% similar to the FdtD protein of Halobacterium sp. NRC-1 (AE005014). The cas gene and this ORF seem to be translationally coupled, because the stop codon of the ORF is directly preceded by a start codon in the reading frame of the cas gene (ATGA; stop codon of the ORF underlined). The codon bias in both genes (ORF 68/49/77%; and cas 70%/51%/77% G+C in the first/second/third position of each codon) and the overall G+C content of the DNA region sequenced (66%) are in accordance with other genes from myxobacteria (Shimkets, 1993). The molecular weight of Cas is predicted to be 72.8 kDa, and the protein contains the non-tandem QW motif that can be found in all squalene and oxidosqualene cyclases (Fig. 5; Poralla et al., 1994). The protein is similar to different forms of eukaryotic oxidosqualene cyclases and, to a significantly lower extent, also similar to bacterial squalene hopene cyclases (see Table 2, Fig. 5). The bacterial Cas protein reported here is more closely related to the functionally equivalent plant cycloartenol synthases than to bacterial Shc proteins.
Table 2. . Comparison of the Cas protein from S. aurantiaca Sg a15 with squalene and oxidosqualene cyclases from other sources.
In mammals, the final product of steroid biosynthesis is cholesterol, with lanosterol as the first oxidosqualene cyclization product, whereas the first steroid in plants is cycloartenol (Michal, 1999). The steroid pattern in myxobacteria is species and strain specific. In S. aurantiaca Sg a15, only cycloartenol and no additional steroids could be detected. Cystobacter sp. are able to produce lanosterol and cycloartenol. It will be interesting to see whether both products result from oxidosqualene cyclization catalysed by the same enzyme, as has been demonstrated to be possible by Meyer et al. (2002) for eukaryotic enzymes. The most evolved biosynthetic pathway among the myxobacteria studied in this work can be found in N. exedens strains, which are also the best producers of steroids. Almost all known intermediates and side-products of the route to cholesterol in eukaryotes can be found under different cultivation conditions (Figs 1 and 2), which underlines the biochemical similarity of myxobacterial pathways to those of more developed organisms. However, some significant differences can be found; eukaryotes usually use one major steroid (e.g. mammals use up to 99% cholesterol), with other steroids being only minor components. In contrast, Nannocystis sp. produce at least two or three major products (Fig. 2).
Studies with different types of steroid biosynthesis inhibitors underline the differences in the biochemical pathways; fluvastatin, a HMG-CoA reductase inhibitor, does not affect cycloartenol biosynthesis in S. aurantiaca, and no inhibition of squalene epoxidase by terbinafin or AMO 1618 was detected in Nannocystis sp.. This might result from the presence of enzymes differing in structure from the well-characterized eukaryotic ones. It will be interesting to see why the myxobacterial enzymes are not affected by the inhibitors, which might be of consequence for the characterization of resistance mechanisms in other organisms, especially fungi. The missing sensitivity to fluvastatin might also be a consequence of the presence of the alternative methylerythritol pathway to isopentenylpyrophosphate in myxobacteria. This pathway has been described in a variety of bacteria and other organisms (Rohdich et al., 2001), but no evidence for its presence or absence is available in myxobacteria. However, the incorporation of leucine, which can serve as an alternative precursor for HMG-CoA (Anke and Dieckmann, 1971; Michal, 1999; Ginger et al., 2001), into cycloartenol in the presence of fluvastatin strongly supports the fact that there are structural differences between eukaryotic and myxobacterial HMG-CoA reductases. Although it has been shown that the mevalonate pathway is active in Myxococcus xanthus (Horbach et al., 1993), it cannot be excluded that myxobacteria possess both biosynthetic routes, as has been described, e.g. for some streptomycetes. The growth inhibition observed with some inhibitors at high concentrations seems to be a result of non-specific effects. However, terbinafin clearly inhibits the 8(9)- to 7-double bond isomerization in Nannocystis sp., a previously unknown mode of action. Only miconazole showed the mode of action expected from studies with different fungi, as judged by the enrichment of biosynthetic intermediates after treatment with the compound.
The data presented in this paper provide evidence of the sterol biosynthetic pathway in myxobacteria (Fig. 1). 2,3-Oxidosqualene is cyclized to either cycloartenol and/or lanosterol, which can be demethylated at C14 to give 4,4-dimethylcholestadien-3β-ol. Stepwise C4 demethylation results in the formation of 7- and 8(9)-cholesten-3β-ol as the final products. The reduction of the 24,25-double bond seems to be possible for every intermediate, because all corresponding reduced compounds have been found during the course of this study. At least the reduction of 7,24- and 8(9),24-cholestadien-3β-ols to the corresponding cholestenols is assumed to be performed by an unspecific enzyme, which could be lacking in N. excedens strain Na e646 that accumulates the former compounds (see Fig. 2).
The pattern of steroids isolated after miconazole treatment shows that the C14-demethylation is necessary for the C4-demethylation and Δ8→Δ7 isomerization, because no corresponding lanosterol derivative was detected at higher miconazole concentrations (Fig. 4). Furthermore, miconazole seems to inhibit the demethylation of 4-methylcholestadien-3β-ol to zymosterol at lower concentrations (0.6 mg l−1) than required for the inhibition of the C14-demethylation from lanosterol to 4,4-dimethylcholestadien-3β-ol (≥0.8 mg l−1), resulting in the observed accumulation of 4-methyl and 4,4-dimethyl cholesterols and lanosterol at the lower concentration.
Although our data clearly show that myxobacteria can produce as much steroids as eukaryotic organisms, the function of those compounds remains a mystery. Under vegetative growth conditions, blocked production does not cause any negative effect on the reproduction time, as shown by comparable growth of cycloartenol knock-outs and wild-type strains of S. aurantiaca. Furthermore, no change in membrane lipid composition (as judged by the fatty acid pattern) or ethanol tolerance could be observed in these strains. No change in the amount or pattern of the steroids could be detected after cultivation of N. excedens Na e485 under different temperatures. In contrast, it has been demonstrated that bacterial growth is reduced and that ethanol sensitivity is increased in hopanoid-producing microorganisms, if Shc is inhibited (Kannenberg and Poralla, 1999). Additionally, it has been observed that hopanoid content increases with rising growth temperatures and acidic conditions (Poralla et al., 1984; Poralla and Kannenberg, 1987). Although there is some speculation that hopanoids are sterol surrogates in bacteria, their physiological function is poorly understood (Kannenberg and Poralla, 1999). Both hopanoids and steroids are most probably located in the membranes on account of their highly hydrophobic properties. Similar to eukaryotes, they might have a function in controlling membrane fluidity (Ourisson and Nakatani, 1994). Interestingly, myxobacteria already have branched chain, saturated and unsaturated fatty acids as membrane constituents (Mahmud et al., 2002) that would allow them to increase their membrane fluidity over a broad range of temperatures. As the gliding motility of myxobacteria involves contact between cells and their substrate, membrane fluidity might have influenced their gliding pattern. However, no changes in swarm pattern or aggregation were observed in the mutant. Perhaps the recent observation that polar slime extrusion and polar pili retraction drive cell movement (Wolgemuth et al., 2002) would imply that the membrane does not play a major role in gliding. An additional function of the myxobacterial steroids might be as signalling molecules similar to steroidal hormones in man or to cholesterol modification of the hedgehog family of secreted signalling proteins (Porter et al., 1996), and we are currently testing this hypothesis in selected myxobacterial steroid producers.
In addition to the mysterious function of these compounds in bacteria, this study raises the intriguing question of the evolutionary origin of steroid biosynthetic pathways in eukaryotes and bacteria. Owing to the fact that the cas gene described here is the first prokaryotic oxidosqualene cyclase, only a very limited phylogenetic analysis is possible. Nevertheless, the data presented in Table 2 and the comparison of S. aurantiaca Cas with representatives of functionally related proteins (Fig. 5) show that the bacterial Cas protein is more closely related to eukaryotic oxidosqualene cyclases than to prokaryotic triterpene synthases of the Shc type. At this time, the question of whether bacteria or eukarya invented steroid biosynthesis first cannot be answered. We are in the process of cloning further steroid biosynthetic genes from myxobacteria for a more detailed phylogenetic analysis. In addition, future studies will focus on the function of these compounds in prokaryotes.
Myxobacterial strains described in this study came from the strain collection at the GBF. All chemicals and most steroid standards were purchased from either Merck or Sigma. 8(9)-Cholesten-3β-ol, 8(14)-cholesten-3β-ol and 8(9)-lanosten-3β-ol were prepared from 5,8(9)-cholestadien-3β-ol, 5,7-cholestadien-3β-ol and 8(9),24-lanostadien-3β-ol by standard hydrations, and their purity was determined by GC/MS analysis. Cycloartenol, which was originally isolated in the laboratory of Professor Dr Rohmer (Strasbourg, France), was kindly supplied by Professor Dr K. Poralla and Dr E. Kannenberg (Tübingen, Germany). Terbinafin was obtained from Dr N. S. Snyder (Sandoz, Vienna, Austria), AMO 1618 from Serva and fluvastatin (Cranoc 20; AstraZeneca) from a local pharmacist.
All strains except for Archangium sp. and S. aurantiaca were kept on VY/2 agar (5 g l−1 fresh bakers’ yeast, 1 g l−1 CaCl2 × 2H2O, 0.005 g l−1 cyanobobalamin, 15 g l−1 agar, adjusted to pH 7.2). Archangium sp. were kept on YH medium (5 g l−1 yeast extract, 1 g l−1 MgSO4 × 7H2O, 1 g l−1 CaCl2−2H2O, 10 g l−1 Hepes, 15 g l−1 agar); S. aurantiaca strains were kept on tryptone starch medium (Mahmud et al., 2002) with 60 µg ml−1 kanamycin when appropriate. For the isolation of steroids and lipids, all strains were cultivated in liquid cultures, which were inoculated with a 2–20% inoculum from a liquid preculture and incubated for 3–5 days at 30°C in a gyratory shaker at 160 r.p.m. S. aurantiaca strains were inoculated with 1 × 107 cells ml−1, and EtOH was added as indicated. Most strains were cultivated in MD1 liquid medium [3 g l−1 casitone (Difco), 2 g l−1 MgSO4 × 7H2O, 0.5 g l−1 CaCl2 × 2H2O, pH 7.2]. Archangium sp. were cultivated in CY medium [3 g l−1 casitone (Difco), 1 g l−1 yeast extract, 1 g l−1 CaCl2 × 2H2O, pH 7.2], Corallococcus sp. in MD1 liquid medium with 0.1% soluble starch, and Polyangium sp. in Pol medium [4 g l−1 Probion (experimental single cell protein, Hoechst), 3 g l−1 soluble starch, 2 g l−1 MgSO4 × 7H2O, 0.5 g l−1 CaCl2 × 2H2O, 0.005 g l−1 cyanocobalamin, 11.9 g l−1 Hepes, 1 ml l−1 trace elements solution (0.1 g l−1 ZnSO4 × 7H2O, 0.03 g l−1 MnCl2 × 4H2O, 0.3 g l−1 H3BO3, 0.01 g l−1 CuCl2 × 2H2O, 0.02 g l−1 NiCl2 × 6H2O, 0.03 g l−1 Na2MoO4 × 2H2O), pH 7.0]. Large-scale cultivations of N. excedens were performed in Probion medium (10 g l−1 Probion, 1 g l−1 MgSO4 × 7H2O, 1 g l−1 CaCl2 × 2H2O).
Isolation of squalene and steroids
Small- and large-scale extraction of squalene and steroids was performed as described previously (Bligh and Dyer, 1959; Radin, 1981) from wet or freeze-dried cell pellets with either MeOH–CHCl3 (1:1) or toluene–hexane (1:1). All extracts were concentrated 20-fold before analysis. Squalene, di- and tetrahydrosqualene and sterols were identified by comparative TLC against standards (Radin, 1981) using either rhodamine 6G or ZAK reagent for detection (Avigan et al., 1963; Claude, 1966), by their retention time in GC (Hewlett-Packard HP 5890 A series II) and GC-MS analysis (GC: Carlos Erba; MS: Kratos MS 50 or Finnigan MAT GCQ) of the extracts after silylation with MSTFA (Donike, 1969) or acetylation with acetic acid anhydride or trifluoroacetic acid anhydride against various standards as described previously (Ogunkoya, 1981). Preparative isolation of sterols from N. excedens Na e485 was performed as described previously (Sperry, 1963; Vroman and Cohan, 1967), and the structures of the main compounds, 8(9)-cholesten-3β-ol, 7-cholesten-3β-ol, 8(9)-cholestadien-3β-ol, 7,24-cholestadien-3β-ol, were determined by EI- and HREI-MS (Finnigan MAT 95) and NMR spectroscopy (Bruker AM-300, WM-400, AM-600). Cycloartenol was enriched by preparative TLC before silylation and GC-MS analysis. Quantitative analysis of squalene and steroids was performed after addition of the internal standards squalene, 5-cholesten-3β-ol and 8(9)-lanosten-3β-ol for the determination of squalene, C27–C29 sterols and C30 sterols respectively. The dry weight of the cell pellet was determined after heating to 105°C for 24 h. The relative dry weight is a measure of the growth inhibition caused by the addition of each inhibitor. It gives the ratio of the dry weight of any sample to the dry weight of the control sample generated without the addition of inhibitor.
All media were analysed for their squalene and steroid content, and only Probion-containing media were shown to contain trace amounts of squalene. This were taken into consideration for the determination of squalene.
Feeding and inhibition experiments
The de novo triterpenoid biosynthesis in myxobacteria was investigated by feeding 10 µCi of RS-[2-14C]-mevalonate DBED salt, sodium [2-14C]-acetate or [U-14C]-leucine to 50 ml cultures of the triterpene-producing strains. The incorporation was determined after lipid extraction and TLC separation by scintillation analysis using a TLC scanner Tracemaster 20 (Berthold Technologies) and a PhosphoImager (Fujifilm BAS-2500) with the basreader 3.14 (Fuji) analysis software. In order to inhibit triterpene biosynthesis in N. excedens Na e485, terbinafin and miconazole were added to liquid cultures of the strain as ethanolic solutions, and the lipids were extracted and analysed as described. Inhibition of HMG-CoA reductase in S. aurantiaca Sg a15 by fluvastatin was performed as follows. The strain was grown in 50 ml of medium for 3 days before the addition of 10 µCi of [U-14C]-leucine and/or 0, 2 or 20 µM of fluvastatin respectively. After incubation for three more days, the cells were harvested and analysed as described above.
DNA manipulations, analysis, sequencing and PCR
Chromosomal DNA of S. aurantiaca strains was prepared as described previously (Neumann et al., 1992). Polymerase chain reaction (PCR) was carried out using HotStarTaq Polymerase (Qiagen) according to the manufacturer's protocol. DMSO was added to a final concentration of 5%. The conditions using the Eppendorf Gradient Mastercycler were as follows: 15 min at 95°C for activation of the polymerase, denaturation for 30 s at 95°C, annealing for 30 s at 57–60°C and extension for 45 s at 72°C; 30 cycles and a final extension for 10 min at 72°C. PCR products were purified with the High Pure PCR product purification kit (Roche Molecular Biochemicals). Sequencing of plasmids was performed using primer walking.
Southern analysis of DNA was performed using the standard protocol for homologous probes of the DIG High Prime DNA labelling and detection starter kit II (Roche Molecular Biochemicals). All other DNA manipulations were performed according to standard protocols (Sambrook et al., 1989). Amino acid and DNA alignments were done using the program clustal W (Thompson et al., 1994) and the lasergene software package (DNASTAR).
Cloning of the cycloartenol synthase gene (cas) and construction of the S. aurantiaca Sg a15 cas mutant
Using the primers MXCHS3 (5′-CATGCTCAGTGCCTCAG GCC-3′) and MXCHS4 (5′-GTCACCATCGCAGCAACACC-3′) and chromosomal DNA from S. aurantiaca Sg a15 as the template, a 1.1 kb fragment harbouring part of cas was amplified and cloned into pCR2.1-TOPO using the TOPO TA cloning kit (Invitrogen), resulting in plasmid pCBS85. Primers STEPCR1 (5′-GACGCGGTGCAGTTCATCCTCTC-3′) and STEPCR2 (5′-CAGCTTTCGACGACCTCACTCCAG-3′) amplify a 439 bp fragment of cas in plasmid pCBS85, which was cloned into pCR2.1TOPO, resulting in pCBS95, which was electroporated after denaturation of the DNA into S. aurantiaca Sg a15 (Silakowski et al., 2000b), resulting in mutant CBS95. The correctness of the integration of the plasmid was verified using Southern hybridization (data not shown).
The insert of pCBS95 was labelled and used for genomic Southern hybridizations with DNA from S. aurantiaca DW4/3-1, S. aurantiaca Sg a15 and M. xanthus DK1622 according to the DIG users’ guide (Roche Molecular Biochemicals) under the following conditions: the hybridization was performed at 38°C in standard buffer, and washing was done at 58°C. In contrast to M. xanthus DK1622, both S. aurantiaca strains gave clean signals with the probe (data not shown).
Genomic DNA from mutant strain CBS95 was prepared, and 1 µg of DNA was digested with ClaI, religated and transformed into Escherichia coli. This procedure allowed the isolation of plasmids harbouring pCBS95 and the adjacent genomic nucleotide sequences up to the next ClaI restriction site, including the complete cas gene from S. aurantiaca. The inserts of these plasmids were subcloned using EcoRI and ClaI into pSK(–), resulting in plasmids pSW16 (0.6 kb insert) and pSW95 (15 kb insert). Both plasmids were sequenced by primer walking.
The assay for development of S. aurantiaca Sg a15 and its descendants was performed as described previously (Plaga et al., 1998). Between 107 and 108 cells were spotted and analysed using a stereoscope.
Nucleotide accession number
The nucleotide sequence of the cas gene reported here has been submitted to the EMBL database under accession number AJ494839.
This work was supported by grants Bo 1834/1-1 and Mu 1254/3 from the Deutsche Forschungsgemeinschaft to H.B.B. and R.M. respectively. The authors would like to thank H. Lünsdorf for the comparison of wild-type and mutant cells by electron microscopy, N. Nimtz for help with the mass spectroscopic analyses of triterpenoids, and E. Nudleman for critical reading of the manuscript. We gratefully acknowledge valuable comments from D. Kaiser, K. Poralla and two unknown reviewers regarding this publication.