C1 Transfer Enzymes and Coenzymes Linking Methylotrophic Bacteria and Methanogenic Archaea

See allHide authors and affiliations

Science  03 Jul 1998:
Vol. 281, Issue 5373, pp. 99-102
DOI: 10.1126/science.281.5373.99


Methanogenic and sulfate-reducing Archaea are considered to have an energy metabolism involving C1 transfer coenzymes and enzymes unique for this group of strictly anaerobic microorganisms. An aerobic methylotrophic bacterium, Methylobacterium extorquens AM1, was found to contain a cluster of genes that are predicted to encode some of these enzymes and was shown to contain two of the enzyme activities and one of the methanogenic coenzymes. Insertion mutants were all unable to grow on C1 compounds, suggesting that the archaeal enzymes function in aerobic C1metabolism. Thus, methylotrophy and methanogenesis involve common genes that cross the bacterial/archaeal boundaries.

Methylobacteriumstrains are obligately aerobic bacteria capable of growing on one-carbon (C1) compounds (methylotrophs) (1). They contain a well-known pathway for interconverting C1compounds that involves a folate coenzyme, tetrahydrofolate (H4F), and nicotinamide adenine dinucleotide phosphate [NAD(P)] (Fig. 1A) (1–3). It has been postulated that in Methylobacterium strains this pathway operates in the oxidative direction to convert formaldehyde to CO2, generating reduced pyridine nucleotides and serving as a major energy-generating pathway during growth on C1compounds (2, 3). The strictly anaerobic Archaea that produce methane (methanogens) or that grow on acetate by sulfate reduction (Archaeoglobus) carry out analogous reactions that use a different coenzyme, tetrahydromethanopterin (H4MPT), and a furan-type cofactor, methanofuran (MFR), and involve either H2 or the electron acceptor/donor F420 rather than NAD(P) (Fig. 1B). In the methanogenic Archaea, these reactions are part of the pathway that reduces CO2 to methane, the central pathway for energy metabolism in methanogens (4). In the sulfate-reducing archaeon Archaeoglobus, the H4MPT- and MFR-linked reactions shown in Fig. 1B operate in the oxidative direction, as part of the energy metabolism for utilizing acetate (5). Thus, the H4MPT/MFR-linked enzymes have until now only been found in methanogenic and sulfate-reducing Archaea, whereas the H4F-linked enzymes have been found in Bacteria, Eukarya, and some Archaea. This separation has suggested that the two sets of pathways are evolutionarily distinct, with the H4MPT/MFR-linked pathways arising after the separation of the methanogenic and sulfatereducing Archaea from other evolutionary branches.

Figure 1

(A) Tetrahydrofolate (H4F)-mediated C1 transfer pathway found in Bacteria, some Archaea, and Eukary-otes and (B) tetrahydromethanopterin/ methanofuran (H4MPT/ MFR)–mediated C1transfer pathway found in methanogenic and sulfate-reducing Archaea. MH4F DH, meth-ylene H4F dehydrogenase; MH4MPT DH, methylene H4MPT dehydrogenase; MH4F CH, methenyl H4F cyclohydrolase; MH4MPT CH, methenyl H4MPT cyclohydrolase; FH4FS, formyl H4F synthase; FMFR:H4MPT FT, formyl MFR:H4MPT formyltransferase; FDH, formate dehydrogenase; FMFR DH, formyl MFR dehydrogenase. The two alternative pathways are proposed to operate inM. extorquens AM1, except that MH4MPT DH is NAD(P)-linked. The known genes of M. extorquens AM1 for both pathways are shown in italics.

One of the aerobic methylotrophs that has been well studied is the α-proteobacterium Methylobacterium extorquens AM1. This bacterium contains nine clusters of genes involved in growth on C1 compounds (6). The largest of these is ∼30 kb and contains the gene for NADP-dependent methylene H4F dehydrogenase (mtdA) as well as a number of other methylotrophy genes (6). We recently cloned and sequenced a 14-kb region adjacent to one end of this 30-kb gene cluster (7) and found that it contains 13 new open reading frames (Fig. 2). The translated products of 12 of these are similar to those of genes in methanogens (8) or in Archaeoglobus fulgidus (9), or both, but not to any other entries in the databases, whereas the amino acid sequence translated from one (orfX) is similar to the sequence of M. extorquens AM1 MtdA but not to any archaeal polypeptide entries (percent identical amino acids is shown in Table 1). The products of six of these methanogen-like genes (orfY,orf4, orf7, orf9,orf17, and orf18) are similar to unidentified gene products in archaeal genome sequences. The other six are similar to the translated products for known genes of the H4MPT/MFR pathway in methanogens and A. fulgidus, and to another archaeal protein (Fig. 1B and Table 1): FfsA is similar to formyl MFR:H4MPT formyltransferase; OrfZ to methenyl H4MPT cyclohydrolase; Orf1, Orf2, and Orf3 to three of the six subunits for formyl MFR dehydrogenase, including the catalytic (B) subunit (10); and Orf5 to ribosomal protein S6 modification protein (a methylase) inMethanococcus jannaschii (8). Alignments of the putative methenyl H4MPT cyclohydrolase and formyl MFR:H4MPT formyltransferase sequences with known sequences in the database reveal strong conservation of groups of amino acid residues between all of the sequences (Fig. 3) (11).

Figure 2

(Bottom) Physical map of the DNA fragment described in this work. Shaded boxes represent putative genes whose translated products are similar to archaeal polypeptides (26 to 44% identical amino acids). Striped box represents a gene similar to another M. extorquens AM1 gene,mtdA (32% identical amino acids). Arrows show direction of transcription. (Top) DNA fragments cloned for complementation studies.

Figure 3

Alignments of amino acid sequences for (A) methenyl H4MPT cyclohydrolase and (B) formyl MFR:H4MPT formyltransferase from methanogenic (Methanosarcina barkeri, M. thermoautotrophicum, M. jannaschii,Methanopyrus kandleri) and sulfate-reducing (A. fulgidus) Archaea with the sequences translated from M. extorquens AM1 orfZ and ffsA, respectively.

Table 1

Characteristics of insertion mutants described in this study.

View this table:

Insertion mutations have been generated in the chromosomal copies of all newly identified genes by means of a kanamycin-resistant cassette and allelic exchange with the suicide vector pAYC61 (12) (Table 1). Mutants with insertions inorfY, orf5, andorf18 were all double-crossover (exchange-type) insertions. These all grew normally on C1compounds and were not studied further. Mutants with insertions in the remaining 10 genes were all unable to grow on C1 compounds but grew normally on succinate, suggesting that these genes were required for C1 metabolism. Mutants with insertions in three unidentified genes (orfX,orf4, and orf7) were the result of double-crossover recombination events and had a straightforward genotype. However, double-crossover mutants were not obtained for five of the genes that show similarity to known genes (ffsA, orfZ,orf1, orf2, andorf3) as well as for two unidentified genes (orf9 and orf17), even though 200 or more independent kanamycin-resistant colonies were screened for loss of the vector in each instance. Instead, all insertions in these genes were the result of single-crossover recombination events that produced both a mutated and a normal copy of the targeted genes, separated by vector sequences. We have previously observed a similar phenomenon for mtdA mutants, in which single-crossover insertion mutants exhibited a C1-negative phenotype (2). In that instance, the single-crossover event apparently separated the normal gene copy from its native promoter, which resulted in reduced expression of the gene product in the mutant. This low-level constitutive activity was evidently sufficient for growth on succinate but not for growth on C1 compounds (2). For representatives of the mutants described here, hybridization experiments and polymerase chain reaction (PCR) analysis involving specific primers (13) suggested that the vector separated the normal gene from upstream chromosomal sequences. Therefore, the C1-negative phenotype of these mutants may reflect the need for greater expression of these genes during growth on C1 compounds than on multicarbon compounds.

The inability to obtain null mutants in these genes might be due to polarity effects of the insertions on nearby essential genes. To test this possibility, we generated subclones containing specific genes for complementation studies, for each of the mutants that showed a C1-negative phenotype (Fig. 2). In each instance these subclones complemented the appropriate mutant. These results demonstrate that the C1-negative phenotypes were due to alterations in the mutated genes, and not due to polarity effects of the insertions on nearby genes.

Because in methanogens, the methenyl H4MPT cyclohydrolase and the formyl MFR:H4MPT formyltransferase are encoded by single genes and corresponding enzyme activities can be detected in Escherichia coli containing the methanogen genes (14), assays were carried out for these enzymes in both M. extorquens AM1 and in E. colicontaining the M. extorquens AM1 genes (Table 2). The formyltransferase was assayed with MFR and H4MPT isolated from the methanogen Methanobacterium thermoautotrophicum and the cyclohydrolase with methenyl H4MPT isolated from the same methanogen (15). Activities similar to those found in methanogens (16) were present in M. extorquens AM1 grown on methanol, and lower activities were found in succinate-grown cells. Methylobacterium extorquens AM1 cells containing pALS8, a plasmid containing both ffsA(predicted to encode the formyltransferase) and orfZ(predicted to encode the cyclohydrolase), showed increased levels of the two enzyme activities. Escherichia coli containing theorfZ fragment (Fig. 2) transcribed by thelac promoter showed significant activities of the cyclohydrolase and no detectable formyltransferase. Escherichia coli containing the ffsA fragment (Fig. 2), also transcribed by the lac promoter, demonstrated formyltransferase activity and no cyclohydrolase activity. These results show that genes encoding these enzymes were present on the expected DNA fragments. These results also strongly suggest that the products of ffsA and orfZ in M. extorquens AM1 must be carrying out reactions with H4MPT and MFR or with unknown cofactors that are readily interchangeable with H4MPT and MFR. It seems likely that FfsA and OrfZ function in an oxidative pathway similar to the H4F-linked pathway for formaldehyde oxidation (Fig. 1).

Table 2

Activities of tetrahydromethanopterin-linked enzymes in M. extorquens AM1, the double-crossover mutant ORFX, andE. coli transformed with orfZ orffsA. Activities are shown in units per minute per milligram of protein and were determined with H4MPT and MFR isolated from M. thermoautotrophicum as coenzymes. The activity of methylene H4F dehydrogenase is given as a control.

View this table:

To substantiate these results we purified methenyl H4MPT cyclohydrolase from M. extorquens AM1 and examined whether the methylotroph contains H4MPT. Cell extracts of M. extorquens AM1 were found to catalyze the hydrolysis of both methenyl H4MPT (3 U/mg) and methenyl H4F (0.03 U/mg). The two activities were separated by chromatography on DEAE Sephacel and Q-Sepharose yielding a more than 100-fold purified methenyl H4MPT cyclohydrolase preparation that did not catalyze the hydrolysis of methenyl H4F. SDS–polyacrylamide gel electrophoresis revealed the presence of a 33-kD polypeptide. By determination of its mass and NH2-terminal amino acid sequence, the polypeptide was identified as the orfZ gene product (17).

The cell extracts were also found to contain the coenzyme H4MPT. Because of higher stability of the methenyl form (18), H4MPT was first converted to methenyl H4MPT (19), which was then purified by reversed-phase high-performance liquid chromatography (5). The compound obtained showed an ultraviolet (UV)/visible spectrum identical to that of methenyl H4MPT from M. thermoautotrophicum and significantly different from that of methenyl H4F. Matrix-assisted laser desorption/ionization–time-of-flight mass spectrometry (MALDI/TOFMS) revealed that the compound had a mass of 577 daltons, identical to that of methenyl dephospho H4MPT, a form of H4MPT also present in some methanogens (20) and exhibiting the same UV/visible spectrum as H4MPT (21). Using an extinction coefficient ɛ335 = 21.6 mM−1 cm−1 for methenyl H4MPT (18), we estimated the amount of dephospho H4MPT in M. extorquens AM1 to be 1.6 nmol per milligram of protein, equal to an intracellular concentration of 0.2 mM. This concentration is lower than in most methanogens (22) but is the same order of magnitude as in Archaeoglobus species (23).

The formyltransferase (FfsA) and the cyclohydrolase (OrfZ) are not the only two enzymes present in M. extorquens AM1 that use H4MPT as coenzyme. Cell extracts of the methylotroph catalyze the dehydrogenation of methylene H4MPT with NADP (7.3 U/mg) and NAD (0.6 U/mg) as electron acceptors. Two enzymes were separated by chromatography on Phenyl Sepharose and Q-Sepharose (13). The NADP-dependent enzyme did not show activity with NAD, but exhibited activity with both H4MPT and H4F. However, the H4MPT-dependent specific activity of the NADP-specific enzyme was 10 times greater than the activity with H4F. The other enzyme was specific for H4MPT and showed highest activity with NAD. However, it also showed some activity with NADP. This NAD-specific enzyme appears to be encoded byorfX, one of the genes for which double crossover C1-negative insertion mutants were obtained (Table 1) and the translated product of which is similar to MtdA. The NAD-dependent enzyme activity was below detection level in theorfX insertion mutant, and it was increased inM. extorquens AM1 carrying the corresponding gene in pALS8 (Table 2).

The results presented here indicate that the coenzyme H4MPT, in a dephospho form, and H4MPT-dependent enzymes previously thought to be unique to anaerobic methanogenic and sulfate-reducing Archaea are present and functionally required in the aerobic proteobacterium M. extorquens AM1. Our preliminary results show that these activities are also present in other methylotrophs (13). The methanogenic and sulfate-reducing Archaea and methylotrophic bacteria have in common an energy metabolism based on the interconversion of C1compounds. It may therefore not be unexpected that for the same catabolic interconversion they also use closely related coenzymes and enzymes, even with the large evolutionary distance between Archaea and Proteobacteria. The genes encoding these enzymes have either been conserved, because these organisms evolved from a common ancestor, or they have been transferred horizontally between more recent ancestors. The latter possibility is especially intriguing in light of the recent hypothesis that eukaryotes may have arisen as a result of a symbiotic association between a methanogenic archaeon and a proteobacterium (24). It is possible that the presence of these archaeal-like genes in Proteobacteria reflects processes that were involved in such an event. Therefore, the detailed functional analysis of all genes involved in C1 metabolism in M. extorquens AM1 and other methylotrophic bacteria is of significant interest to understanding the evolutionary history of C1oxidation and reduction pathways in all organisms.

  • * To whom correspondence should be addressed. E-mail: lidstrom{at}


View Abstract

Stay Connected to Science

Navigate This Article