One possible method of genetic transfer between archaea and eubacteria is the assimilation of archaeal extracellular DNA by a eubacterium in an effort to survive. This would allow a eubacterium to adapt to new metabolites or to utilize different cofactors for normal or new metabolic processes. The genetic location of the acquired DNA would then be of great use to explain the original event. If the DNA was integrated into the genome there could have been a genetic recombination or an insertional event. If the DNA is located extrachromosomally, then the eubacterium possibly stabilized the extrachromosomal element for a phenotype essential for survival.

One eubacterial phenotype that requires an archaeal cofactor is the ability to grow on propylene and other short-chain aliphatic alkenes as a source of carbon and energy (1). Xanthobacter strain Py2, a gram-negative facultative methylotroph, and Rhodococcus rhodochrous (Rhodococcus corallinus; Nocardia corallina) B276, a gram-positive actinomycete, initiate propylene metabolism by a monooxygenase reaction that inserts an oxygen atom into the olefin bond forming epoxypropane (22, 27). Epoxypropane is then further metabolized by using CoM as a metabolic carrier molecule (1). An enzyme designated epoxyalkane:CoM transferase conjugates CoM to R- and S-enantiomers of epoxypropane to form the R- and S-enantiomers of 2-hydroxypropyl CoM, which are then dehydrogenated to form 2-ketopropyl CoM (1, 3, 4). 2-Ketopropyl CoM is subsequently carboxylated by a novel NADPH:disulfide oxidoreductase/carboxylase to generate acetoacetate and release CoM (8).

The presence of CoM in eubacteria is a fairly recent discovery, and accordingly the genetic location of the genes of CoM biosynthesis has not yet been determined. In both Xanthobacter strain Py2 and R. rhodochrous, CoM biosynthesis is coordinately regulated with the expression of the enzymes of alkene and epoxide metabolism, suggesting that the genes for CoM biosynthesis, alkene oxidation, and epoxide metabolism may be clustered and may be under the control of a common regulatory element (19). Importantly, Saeki et al. recently showed that the genes encoding the alkene monooxygenase of R. rhodochrous were located extrachromosomally on a linear megaplasmid 185 kb in length (25). Repeated subculturing of R. rhodochrous under nonselective conditions, i.e., on rich medium in the absence of propylene, resulted in the loss of the linear megaplasmid and a propylene-negative phenotype. Based on the large size of the linear megaplasmid and the coregulation alluded to above, it is plausible that the additional genes of alkene metabolism and CoM biosynthesis are located on this megaplasmid as well.

The discovery of a linear megaplasmid involved in alkene oxidation in R. rhodochrous warrants an investigation of whether a similar situation exists in Xanthobacter strain Py2, an organism phylogenetically distinct from R. rhodochrous. In this paper we demonstrate that the genes for alkene and epoxide metabolism are indeed on a linear megaplasmid in Xanthobacter strain Py2, demonstrating a conserved strategy for the maintenance of this eubacterial phenotype. Based on multiple sequence alignments, a gene on the sequenced portion of the linear megaplasmid is shown to have high identity with a methanogenic gene of unknown function but which is adjacent to a gene recently shown to be involved in CoM biosynthesis in Methanococcus jannaschii (16). Thus, eubacterial aliphatic alkene oxidation is a phenotype requiring selective pressure, maintenance of an extrachromosomal element, and the biosynthesis of a specialized cofactor that was, until only recently, thought to be restricted to the methanogenic archaea.

Materials. Oligonucleotides were obtained from Operon Technologies, Inc., Alameda, Calif. Ready-to-go PCR beads were obtained from Amersham Pharmacia, Piscataway, N.J.

Growth conditions and isolation of a propylene-negative mutant. Xanthobacter strain Py2 was grown in shake flasks using the medium and conditions described previously (2, 19). For isolation of a propylene-negative strain, cells were subcultured successively in mineral salts medium containing glucose as the sole source of carbon and energy. Cells were grown to stationary phase and then subcultured into fresh medium at a dilution of 1:25. After ten transfers, cells were unable to grow using propylene as the source of carbon. A pure isolate was obtained by plating the glucose-grown cells on rich medium and selecting a single colony. This propylene-negative isolate was designated strain Py2.101.

Induction of alkene monooxygenase and epoxide carboxylase activities. Wild-type and propylene-negative Xanthobacter strains were subcultured successively with acetate as the sole carbon source. The third subcultures were grown to early log phase (A₆₀₀ = ~2.0), at which time the cells were harvested by centrifugation and resuspended to an A₆₀₀ of 3.0 in fresh medium containing acetate. Samples (1 ml) of the cell suspensions were assayed for the ability to degrade propylene and epoxypropane as described previously (15). CoM concentrations in cell suspensions and spent media were determined as described previously (19). When present, the protein synthesis inhibitors chloramphenicol and rifampin were added to 0.2 mg · ml⁻¹ and 0.4 mg · ml⁻¹, respectively.

Preparation of high-molecular-weight DNA. Frozen cell paste (wild-type and propylene-negative) was thawed in 10 volumes of minimal medium containing 1% (wt/vol) each of glycine and glucose followed by incubation with shaking for 3 h at 30°C. The cells were then harvested by centrifugation and resuspended in 1 volume each of EET (0.1 M EDTA, 10 mM EGTA, 10 mM Tris [pH 8.0]) and phosphate-buffered saline–agarose (8.0 g of NaCl, 0.2 g of KCl, 1.44 g of NaHPO₄ · 7H₂O, and 0.24 g of KH₂PO₄ per liter, with 1.6% [wt/vol] agarose, pH 7.2) at 60°C. The cell suspensions were embedded into agarose plug molds and lysed as described by McClelland et al. (20).

CHEFE. Contour-clamped homogeneous-field electrophoresis (CHEFE) was performed in a Bio-Rad CHEF-DR II system using the conditions described by Saeki et al. (25). DNA bands were stained with ethidium bromide and visualized using a UV transilluminator. The linearity and size of pEK1 were determined by altering the pulse times and monitoring the migrational rates according to concatemers of lambda phage high-molecular-weight DNA markers using protocols described by Ravel et al. (24).

Isolation of DNA from CHEFE gels. Desired DNA bands were excised from CHEFE gels and centrifuged in plasmid prep microspin cups (Stragene, San Diego, Calif.) in a microcentrifuge at 14,000 rpm for 10 min. After centrifugation, water (100 μl) was added to the cup, and the centrifugation was repeated. The DNA in the filtrate was precipitated with ammonium acetate (26), washed with 70% ethanol, and resuspended in 32 μl of H₂O. The DNA was further purified by precipitation with 8 μl of 4 M NaCl and 40 μl of 13% (wt/vol) polyethylene glycol 8000. After a 20-min incubation on ice, the DNA was pelleted by centrifugation at 4°C in a microcentrifuge (14,000 rpm for 15 min), washed with 70% ethanol, and resuspended in 20 μl of water. DNA concentrations were determined by A₂₆₀ (26).

Oligonucleotide synthesis. Oligonucleotides for xecC, the NADPH:2-ketopropyl CoM oxidoreductase/carboxylase gene, and xamoA, the alkene monooxygenase alpha subunit gene, were designed by complementing the published DNA sequences (28, 32). Oligonucleotides for PCR amplification of the ribulose 1,5-bisphosphate carboxylase/oxygenase (rubisco) gene were designed by complementing the DNA sequence for the cfxL gene, which encodes the rubisco large subunit in Xanthobacter flavus strain H4-14 (21).

Probing by PCR amplification. The DNAs used for PCR amplification were total high-molecular-weight DNA, CHEFE-resolved genomic DNA, and CHEFE-resolved linear megaplasmid pEK1. PCR mixtures contained 100 ng of DNA from each source, 1 μM each primer, 2.0% (vol/vol) glycerol, and a PCR ready-to-go bead (Amersham Pharmacia) in a total volume of 25 μl. Cycling parameters for amplification of xamoA, 1.6 kbp, and xecC, 1.6 kbp, were the same, with 1 cycle at 95°C for 5 min followed by touchdown PCR (13) of 20 cycles, decreasing the annealing temperature by 1 degree each cycle at 95°C for 30 s, 70°C for 30 s, and 72°C for 30 s. After touchdown PCR, the reaction mixture was subjected to 30 cycles of normal PCR with 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s followed by a final elongation at 72°C for 7 min and then a final hold at 4°C. Cycling parameters for amplification of rubisco, 1.5 kbp, were quite similar, except reverse touchdown PCR was used instead of touchdown PCR: 1 cycle at 95°C for 5 min followed by reverse touchdown PCR of 20 cycles, increasing the annealing temperature by 1 degree each cycle at 95°C for 30 s, 50°C for 30 s, and 72°C for 30 s (13). After reverse touchdown PCR the reaction mixture was subjected to 30 cycles of normal PCR with 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s followed by a final elongation at 72°C for 7 min and then a final hold at 4°C. Amplification was examined by agarose gel electrophoresis using established protocols (26).

Sequencing of the PCR probe products. The PCR product from each reaction was sequenced to ensure proper identification of the amplified probe. The DNA was isolated from the agarose gel as described for the CHEFE gel DNA isolation and was subjected to Big-Dye sequencing analysis at the Utah State University Biotechnology Center.

Completed sequence of xecG. Total DNA was used to make a cosmid library with the SuperCos 1 from Stratagene and was amplified per protocol. Individual clones were picked and resuspended in 100 μl of sterile water, and 10 μl was used in the PCR probing reaction. Positive clones were identified by PCR amplification of xecA and xecC using the protocol for amplification of xecC as described above. Five positive clones were identified, and all were sequenced beyond xecG in a double-stranded nonambiguous manner.

Multiple sequence alignments. Sequence similarities between xecG and the genes that code other proteins in the database were identified using the Basic Local Alignment Search Tool (BLAST). Multiple sequence alignment was performed using NPS@:Multalin version 5.3.2 (9, 10).

Nucleotide sequence accession number. The sequence of xecG has been deposited with the GenBank data bank under accession no. AY024334.

Isolation and characterization of propylene-negative mutants of Xanthobacter strain Py2. Repeated subculturing of Xanthobacter strain Py2 with glucose as the sole carbon source resulted in the loss of the ability to grow subsequently by using propylene or epoxypropane as growth substrates. To determine whether the propylene-negative phenotype resulted from a loss of alkene monooxygenase and/or epoxide carboxylase enzyme activities, the wild-type and spontaneous propylene-negative strains were examined for expression of alkene- and epoxide-degrading activities when exposed to propylene and epoxypropane. In agreement with previous studies (15), wild-type Xanthobacter strain Py2 that had been grown for several generations with an alternative carbon source, in this case acetate, did not have detectable levels of propylene and epoxypropane degradation activities at the initiation of the experiment (Fig. 1). After a 2- to 3-h lag period, both alkene- and epoxide-degrading activities were expressed in the wild-type strain, a result that was prevented by the addition of the RNA and protein synthesis inhibitors rifampin and chloramphenicol (Fig. 1). In contrast to this result, the spontaneous propylene-negative strain did not induce either alkene- or epoxide-degrading activities, even after long exposure times (Fig. 1). Thus, the propylene-negative phenotype correlates with the loss of both key activities of alkene metabolism. The key to maintaining the propylene-positive phenotype was found to lie in the number of nonselective subcultures the bacteria were subjected to prior to transferring to propylene-containing medium. In our hands, propylene-dependent growth could be routinely restored in cultures transferred 2 to 4 times in nonselective medium but not in cultures transferred 6 to 10 times. In addition, propylene-positive cultures could not be recovered from agar plates or slants containing rich medium after extended storage periods (6 months or longer) unless the plates or slants were incubated under an atmosphere of propylene. Clearly, selective pressure is necessary to maintain the propylene-positive phenotype of Xanthobacter strain Py2.

FIG. 1 Induction of alkene monooxygenase and epoxide carboxylation activities in acetate-grown Xanthobacter strain Py2 by propylene or epoxypropane. (A) Propylene remaining; (B) epoxypropane remaining. Symbols: □, wild-type strain Py2; ■ wild-type (more ...)

Quantitation of CoM in wild-type and propylene-negative Xanthobacter strain Py2. Recently, the expression of the alkene and epoxide metabolizing enzymes of Xanthobacter strain Py2 and R. rhodochrous B276 was shown to be coordinately regulated with the biosynthesis of CoM, the methanogenic archaeal cofactor that serves as the carrier of activated carbon units involved in epoxide-to-β-keto acid conversion (19). The only eubacterial function that has been ascribed to date to CoM is as the carrier molecule in epoxide carboxylation, and CoM does not accumulate to detectable levels in cultures of Xanthobacter strain Py2 or R. rhodochrous grown with other carbon sources (19). To extend these observations, the concentrations of CoM in cell suspensions and spent culture media of the wild-type and propylene-negative strains were measured. No CoM could be detected (detection limit of ~0.1 μM) in either cell suspensions or spent media of the propylene-negative strain either before or after prolonged (greater than 3 days) exposure to the inducer, propylene. In agreement with our previous results (19), wild-type cells, which had no detectable CoM prior to induction, accumulated significant levels of CoM (>10 μM) over the same time course.

Identification of a linear megaplasmid in Xanthobacter strain Py2. As mentioned in the Introduction, Saeki et al. have observed a similar loss of the propylene-positive phenotype of R. rhodochrous B276 when subcultured on rich medium (25). They further demonstrated that this phenotype correlated with the loss of a 185-kb linear megaplasmid containing the alkene monooxygenase genes (25). In order to determine whether a similar situation is responsible for the propylene-negative phenotype of Xanthobacter strain Py2, total high-molecular-weight DNA was fractionated by CHEFE under the appropriate conditions for identification and resolution of linear megaplasmids (24, 25). As shown in Fig. 2, a single 320-kb linear megaplasmid, designated pEK1, was resolved by this fractionation for wild-type Xanthobacter strain Py2 (Fig. 2, lane 2). In contrast, the 320-kb megaplasmid was not found in the propylene-negative mutant strain (Fig. 2, lane 1). Whereas R. rhodochrous B276 was shown to have four resolvable linear megaplasmids, one of which correlated with propylene-dependent growth, only the single 320-kb linear megaplasmid was detected in Xanthobacter strain Py2.

FIG. 2 Identification of a linear DNA molecule in Xanthobacter strain Py2 by CHEFE. Lane 1, fractionated high-molecular-weight DNA in propylene-negative strain Py2.101; lane 2, fractionated high-molecular-weight DNA in wild-type strain Py2; lane 3, molecular (more ...)

The alkene monooxygenase and epoxide carboxylation enzymes of Xanthobacter strain Py2 are located on the linear megaplasmid pEK1. The alkene monooxygenase of Xanthobacter strain Py2 is a four-component enzyme system encoded by six clustered genes arranged as shown in Fig. 3A (33). The genes encoding the four key enzymes of epoxide metabolism, i.e., epoxyalkane:CoM transferase, R- and S-hydroxypropyl-CoM dehydrogenases, and NADPH:2-ketopropyl-CoM oxidoreductase/carboxylase, are likewise clustered in an operon in Xanthobacter strain Py2, with the genes designated xecA, xecD, xecE, and xecC encoding the respective four enzymes (Fig. 3B) (28). The spatial relationship between the alkene monooxygenase genes and epoxide carboxylation enzyme genes has not been reported for Xanthobacter strain Py2. While R. rhodochrous B276 has an epoxide carboxylation system consisting of proteins biochemically very similar to those of Xanthobacter strain Py2, the genes encoding these proteins have not been cloned.

FIG. 3 Genetic map of alkene monooxygenase (xamo) and epoxide carboxylase (xec) genes of Xanthobacter strain Py2. Representative genes are drawn to scale. xamoA, xamoB, xamoE, genes encoding the alpha, gamma, and beta subunits, respectively, of the alkene monooxygenase (more ...)

PCR amplification of xamoA and xecC, which encode the alpha subunits of the alkene monooxygenase and NADPH:2-ketopropyl-CoM oxidoreductase/carboxylase, respectively, was used to probe the location of the genes and to determine whether they were present in the propylene-negative strain (Table 1). As shown in Fig. 4, PCR amplification of total high-molecular-weight DNA from wild-type Xanthobacter strain Py2 with primers designed for xamoA and xecC resulted in the amplification of the respective genes, as verified by sequence analysis of the PCR product (Fig. 4A and B, lanes 1). In contrast, no amplification of xamoA or xecC was observed when total DNA from the propylene-negative strain was used (Fig. 4A and B, lanes 1). When CHEFE-resolved DNA fractions from wild-type Xanthobacter strain Py2 were probed, xamoA and xecC were amplified only from the DNA corresponding to the linear megaplasmid (Fig. 4A and B, lanes 2 and 3), demonstrating that both genes are on the megaplasmid. As a control, the genetic location of the large subunit of rubisco, a gene expected to be located on a chromosome, was probed using primers designed to the sequence of the X. flavus gene cfxL (Table 1). As expected, cfxL was amplified from the genomic DNA fractions from both the wild-type and mutant strains, while no amplification was seen from the linear megaplasmid.

TABLE 1 Oligonucleotides used in this work

FIG. 4 PCR amplification of xamoA (A), xecC (B), and cfxL (C). Lanes 1, total high-molecular-weight DNA from wild-type strain Py2; lanes 2, CHEFE-purified genomic DNA from wild-type strain Py2; lanes 3, CHEFE-purified linear megaplasmid pEK1; lanes 4, total (more ...)

XecG is a homolog of proteins present in methanogens and Bacillus subtilis. The DNA fragment used by Swaving et al. (28) to sequence the genes involved in epoxide carboxylation contains a truncated gene that encodes the first 150 amino acids of a hypothetical protein, herein referred to as XecG, that is downstream of the assigned epoxide carboxylation genes but which is of unknown function (Fig. 3). An initial examination of this truncated protein by BLAST search revealed amino acid sequence similarity to three hypothetical proteins present in the genomes of three prokaryotes: MTH1674 from Methanobacterium thermoautotrophicum (strain Delta H), MJ0255 from M. jannaschii, and YitD from B. subtilis. In order to strengthen the evidence that XecG is a homolog of these three proteins, the entire gene was cloned and sequenced. Multiple sequence alignments of the four hypothetical proteins are presented in Fig. 5, and the sequence identities among the proteins are presented in Table 2. The high degree of sequence identity between the four proteins and the lack of homologs to these proteins in other prokaryotes suggest that the proteins have similar, highly specialized functions. Graupner and coworkers have described a plausible biosynthetic pathway for CoM that begins with phosphoenolpyruvate and proceeds through l-sulfolactate phosphate, l-sulfolactate, sulfopyruvate, sulfoacetaldehyde, and sulfoethylcysteine as intermediates (16). In B. subtilis, l-sulfolactate has been shown to be a major constituent (up to and greater than 5% dry weight) of sporulating cells and mature spores (6). It is not known why B. subtilis spores accumulate this novel metabolite, but it is noteworthy that it appears to be absent in vegetative cells and thus appears to be involved in the sporulation process (6). Xanthobacter strain Py2, M. jannaschii, M. thermoautotrophicum, and B. subtilis thus share a common requirement for the production of the novel metabolite l-sulfolactate. It is tempting to speculate that xecG, yitD, and the methanogen homologs encode proteins that play a role in the synthesis of this metabolite.

FIG. 5 Multiple sequence alignment with Xanthobacter strain Py2 (Xanthobacter) hypothetical protein XecG, M. thermoautotrophicum (strain Delta H) (M. thermoauto) hypothetical protein MTH1674, M. jannaschii hypothetical protein MJ0255, and B. subtilis hypothetical (more ...)

TABLE 2 Sequence identities among XecG homologs

Graupner and coworkers have recently shown that two proteins encoded by the MJ0256 open reading frame in M. jannaschii have sulfopyruvate decarboxylase activity, an activity required for CoM biosynthesis via the proposed pathway, when cloned and expressed in Escherichia coli (16). Interestingly, MJ0256 is the gene immediately adjacent to MJ0255, the XecG homolog described above. A further intriguing observation is that YitC, the B. subtilis hypothetical protein encoded by the gene adjacent to yitD, has 20% identity with MJ1140, another methanogen protein of unknown function. In order to determine whether homologs of yitC or MJ0256 are located near xecG, the DNA immediately adjacent to xecG was sequenced. An open reading frame immediately adjacent to xecG was identified and found to encode an apparent argininosuccinate lyase (data not shown). Thus, if homologs of these other proteins are present on the linear megaplasmid of Xanthobacter strain Py2, then they are separated by at least one gene that has no apparent connection to sulfolactate or CoM metabolism.

Implications of these studies. The bacterial metabolism of short-chain aliphatic alkenes such as propylene involves a complex sequence of reactions resulting in oxygenation and carboxylation of the alkene to produce the corresponding β-keto acid, which then enters central metabolism. One of the intriguing questions of this system is why CoM was chosen as the nucleophile for epoxide ring opening and as the C₃ carrier in the pathway rather than a more conventional cofactor. The presence of linear megaplasmids in two phylogenetically distinct bacteria that use a conserved strategy for alkene metabolism suggests that the capability to synthesize CoM may have resulted from the acquisition of the requisite genes from methanogenic archaea. Linear megaplasmids have in recent years been demonstrated to play an important role in the acquisition of diverse specialized catabolic activities by other strains of Rhodococcus, Xanthobacter, and other eubacteria (5, 11, 17, 18, 24, 29). For Xanthobacter strain Py2, the available evidence suggests that both the CoM biosynthetic genes and alkene and epoxide metabolic genes are located on the single 320-kb linear megaplasmid. This raises questions about the origin of the metabolic enzymes of alkene and epoxide metabolism and their relation to CoM and methanogenesis, since alkenes and epoxides are not known to be metabolized by methanogens or other archaea. It will be interesting to determine what other genes reside on the linear megaplasmid of Xanthobacter strain Py2 and how these genes relate to other catabolic, regulatory, and biosynthetic activities of the methanogenic archaea and other organisms.

This work was supported by National Institutes of Health grant GM51805.

We thank Robert H. White of Virginia Polytechnic University for sharing information on the CoM biosynthetic enzymes of M. jannaschii prior to publication. We also thank Dennis Welker for technical assistance with CHEFE.