Multiple Formate Dehydrogenase Enzymes in the Facultative Methylotroph Methylobacterium extorquens AM1 Are Dispensable for Growth on Methanol

Abstract

Formate dehydrogenase has traditionally been assumed to play an essential role in energy generation during growth on C₁ compounds. However, this assumption has not yet been experimentally tested in methylotrophic bacteria. In this study, a whole-genome analysis approach was used to identify three different formate dehydrogenase systems in the facultative methylotroph Methylobacterium extorquens AM1 whose expression is affected by either molybdenum or tungsten. A complete set of single, double, and triple mutants was generated, and their phenotypes were analyzed. The growth phenotypes of the mutants suggest that any one of the three formate dehydrogenases is sufficient to sustain growth of M. extorquens AM1 on formate, while surprisingly, none is required for growth on methanol or methylamine. Nuclear magnetic resonance analysis of the fate of [¹³C]methanol revealed that while cells of wild-type M. extorquens AM1 as well as cells of all the single and the double mutants continuously produced [¹³C]bicarbonate and ¹³CO₂, cells of the triple mutant accumulated [¹³C]formate instead. Further studies of the triple mutant showed that formate was not produced quantitatively and was consumed later in growth. These results demonstrated that all three formate dehydrogenase systems must be inactivated in order to disrupt the formate-oxidizing capacity of the organism but that an alternative formate-consuming capacity exists in the triple mutant.

The classic scheme of energy metabolism during methylotrophic growth involves a formate oxidation step except in strains in which all formaldehyde is oxidized in the cyclic ribulose monophosphate cycle (1). Formate dehydrogenase (FDH) activity has been detected in most methylotrophs (3, 10, 13, 15, 22, 42), and a few FDHs have been purified and analyzed (reviewed in reference 39). In the methylotrophic yeast Candida boidinii, the FDH step was shown not to be essential for methylotrophic growth, but FDH mutants showed reduced growth on methanol (32). However, as the complete C. boidinii genome sequence is not available, the presence of other FDHs is not excluded.

Mutant-based analysis of the role of the FDH step in C₁ oxidation has not yet been attempted in methylotrophic bacteria. M. extorquens AM1 offers a convenient model to study this question. It possesses two pathways in which formaldehyde can be oxidized to formate (Fig. 1), one linked to tetrahydromethanopterin (H₄MPT) and another linked to tetrahydrofolate (H₄F) (5, 6). The enzymes involved in the two pathways have been studied in detail, and current evidence suggests that the main pathway for oxidizing formaldehyde is the H₄MPT-linked pathway (reviewed in reference 37). It has been demonstrated recently that this pathway produces formate as an intermediate, a result of a formylmethanofuran transferase/hydrolase reaction (29), and thus in this pathway one molecule of formate is formed in M. extorquens AM1 per oxidized molecule of a C₁ substrate, such as methanol or methylamine. This formate is subsequently oxidized to CO₂, presumably by FDH (Fig. 1).

FIG. 1.

C₁ metabolism of M. extorquens AM1. H₄MPT, tetrahydromethanopterin; H₄F, tetrahydrofolate; Fae, H₄MPT-dependent formaldehyde activating enzyme (39); MtdA, NADP-dependent methylene-H₄MPT dehydrogenase (8, 38); MtdB, NAD(P)-dependent methylene-H₄MPT dehydrogenase (11); Mch, methenyl-H₄MPT cyclohydrolase (30); Fhc, formyltransferase/hydrolase complex (29); Fch, methenyl-H₄F cyclohydrolase (30); FtfL, formate-H₄F ligase (23); FDH, formate dehydrogenase (20; this study).

An FDH from M. extorquens AM1 has recently been purified and characterized and shown to be a novel, tungsten-containing FDH encoded by two genes, fdh1AB (20). In this study we identified two new regions in the M. extorquens AM1 chromosome coding for two additional FDH enzymes. Using mutation analysis, we demonstrate that all three enzymes are expressed during growth on C₁ compounds but none is essential for growth on methanol, providing new insight into the energetics of C₁ metabolism in serine cycle methylotrophs.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Escherichia coli strains JM109 (Invitrogen), Top10 (Invitrogen), and S17-1 (35) were grown in Luria-Bertani (LB) medium in the presence of appropriate antibiotics as described by Sambrook et al. (33). M. extorquens AM1 was routinely grown in the minimal medium described previously (12). To test the dependence of expression of formate dehydrogenases on molybdenum and tungsten, a similar medium was used from which (NH₄)Mo₇O₂₁was omitted. To this medium, (NH₄)Mo₇O₂₁ or Na₂WO₄ or both were added at concentrations ranging from 0.3 to 3 μM, as appropriate. Succinate (20 mM) or methanol (120 mM) was used as the substrate for growth in liquid media. For growth on solid medium, succinate, methanol, methylamine (120 mM), or formate (25 mM) was added. The following antibiotic concentrations were used: kanamycin, 100 μg/ml; rifamycin, 50 μg/ml; and tetracycline, 10 mM. The following cloning vectors were used: pCR2.1 (Invitrogen, Carlsbad, Calif.) for cloning PCR fragments, pUC19 (Pharmacia) for subcloning, pCM183 (25) as a source of the loxP-flanked kanamycin resistance cassette, pAYC61 (4) and pCM184 (25) as suicide vectors, pCM130 (24) for promoter fusion construction, and pRK2013 (7) as a helper plasmid for conjugation.

DNA manipulations.

Plasmid isolation, E. coli transformation, restriction enzyme digestion, ligation, blunting ends with T4 DNA polymerase, and filling in ends with Klenow enzyme were carried out as described by Sambrook et al. (33). The chromosomal DNA for PCR amplification was isolated essentially as described by Saito and Miura (31), from 3 ml of culture.

DNA sequencing.

The 6.5X coverage genomic sequence of M. extorquens AM1 was produced by the Human Genome Center at the University of Washington and Integrated Genomics, Chicago, Ill., and is available at http://www.integratedgenomics.com/genomereleases.html#list6. Some regions involved in this study were resequenced on both strands by the Department of Biochemistry, University of Washington Sequencing Facility, on an Applied Biosystems automated sequencer.

Computer analysis.

Translation and analyses of DNA and DNA-derived polypeptide sequences were carried out with the Genetic Computer Group (Wisconsin) and ORF Finder (NCBI) programs.

Matings.

For mutant selection or plasmid transfers, biparental or triparental matings were performed overnight at 30°C on nutrient agar (Becton Dickinson & Co., Franklin Lakes, N.J.). Cells were then washed and plated onto selective plates with succinate.

Mutant generation.

The formate dehydrogenases homologous to the molybdenum-linked FDH from Ralstonia eutropha (FDH2) (9, 28) and the anaerobic FDH from Wolinella succinogenes (FDH3) (19, 21) have not yet been characterized biochemically in M. extorquens AM1. Therefore, insertion mutations were generated in all the putative genes for these two enzymes: fdh2A, fdh2B, fdh2C, fdh2D, fdh3A, fdh3B, and fdh3C. The marker exchange technique described previously (4) was used to generate insertion mutations in these genes, and the double-crossover nature of these mutations was confirmed by diagnostic PCR. fdh2A, fdh2B, fdh2C, and fdh2D mutants all had similar phenotypes and fdh3A, fdh3B, and fdh3C mutants all had similar phenotypes, suggesting that they do encode the subunits of the respective FDH enzymes (data not shown).

Deletion mutations were also generated in the two FDH enzymes. For these we used the newly described suicide vector in which the kanamycin resistance gene is flanked by loxP sites (25). A region of 2,667 bp involving fdhBA was deleted to generate a mutation in FDH2, and a region of 3,609 bp involving fdh3ABC was deleted to generate a mutation in FDH3. The resulting deletion mutants had phenotypes similar to the phenotypes of the mutants with kanamycin resistance gene insertions in separate subunits of the two enzymes.

Because FDH1 has been biochemically characterized in M. extorquens AM1 (20), only one gene was subjected to mutation. To generate this mutation, the kanamycin resistance gene flanked by loxP sites was cut out of pCM183 (25) and inserted into the HincII site in the middle of fdh1A. An unmarked variant of this mutation was created by excising the kanamycin resistance gene via specific recombination at the lox sites, as described (25). The unmarked mutant had an insertion of 150 nucleotides resulting from the recombination event. This insertion contained a number of stop codons in the fdh1A reading frame.

A double mutant lacking FDH1 and FDH2 was generated in the unmarked FDH1 background described above by introducing the marked deletion in FDH2 described above. A double mutant lacking FDH1 and FDH3 was generated by introducing the marked deletion in FDH3 described above into the unmarked FDH1 background. A double mutant in FDH2 and FDH3 was generated by first creating an unmarked version of the FDH3 deletion via specific excision (25) and then introducing the marked FDH2 deletion. The triple mutant was generated by unmarking the FDH3 deletion in the double FDH1-FDH3 mutant, followed by the introduction of the (marked) FDH2 deletion. The mutant genotypes are summarized in Table 1.

TABLE 1.

Strains of M. extorquens AM1 employed in this study

Strain	Characteristics	Reference
Wild type	Rifr derivative	25
UV26	mxaF	25
F1	fdh1A	This study
F2	Δfdh2AB::kan	This study
F3	Δfdh3ABC::kan	This study
F12	fdh1A Δfdh2AB::kan	This study
F13	fdh1A Δfdh3ABC::kan	This study
F23	Δfdh2AB::kan Δfdh3ABC	This study
F123	fdh1A Δfdh2AB::kan Δfdh3ABC	This study

Enzyme assays.

Enzyme activities were determined in crude extracts obtained by passing cells through a French pressure cell at 1.2 × 10⁸ Pa, followed by centrifugation for 10 min at approximately 15,000 × g. A standard optical assay for FDH activity was performed at 30°C by following the reduction of NAD⁺ at 340 nm (ɛ₃₄₀ = 6.2 mM⁻¹ cm⁻¹). The reaction mixture contained 50 mM tricine-KOH (pH 7.0), 30 mM sodium formate, 0.5 mM NAD⁺, and an appropriate amount of protein. As an alternative electron acceptor, benzyl viologen (2 mM; ɛ₅₇₈ = 6.25 mM⁻¹ cm⁻¹) was tested under anoxic conditions. Catechol dioxygenase was assayed as described (16). Enzyme assays were done in triplicate, and the values obtained agreed within 20%. Protein concentration was assessed spectrophotometrically (41).

¹³C NMR experiments.

Late-exponential-phase cells (OD₅₇₈ = 1.2 to 1.5) of wild-type and FDH mutant strains were centrifuged, washed, and resuspended to a final density of 15 mg (dry weight) ml⁻¹ into 6 ml of fresh medium without Mn and Fe and deprived of a carbon source. The cell suspension was transferred into an airlift nuclear magnetic resonance (NMR) tube (18). At that time, 0.6 ml of D₂O was added for the field-locking signal, and aeration at 38 ml/min was switched on. The airlift NMR tube was placed into the NMR magnet, and the initial spectrum was acquired to test for the naturally abundant signals. The labeled carbon source (99.9% [¹³C]methanol, purchased from Eurisotop, France) was then added to a final concentration of 120 mM. Spectra were accumulated in consecutive 5-min blocks of 200 scans each. All ¹³C NMR spectra were obtained at 30°C in the Fourier transform mode at 125.79 MHz on a Bruker spectrometer equipped with a dual ¹H/¹³C 10-mm probe head, with a spectral width of 15 kHz (16,000 data points) and a 90° pulse angle with an interpulse delay of 1.5 s. Proton decoupling was applied during acquisition. The free induction decays (FIDs) were exponentially multiplied (3-Hz line broadening) prior to Fourier transformation. Chemical shifts were expressed as parts per million relative to the resonance of tetramethylsilan at 0 ppm.

Formate detection in culture medium.

Strains were grown in standard medium in 2-liter Erlenmeyer flasks filled with 600 ml of medium at 180 rpm at 30°C. During growth, 1-ml samples were withdrawn, cells were pelleted, and the supernatants were used for formate detection by high-pressure liquid chromatography (HPLC) as described (27).

Estimation of methanol used to generate biomass.

A rough estimation of methanol used during formate accumulation was carried out based on biomass conversion. The 0.2 OD unit increase that occurred during formate accumulation is approximately equal to 0.05 mg (dry weight)/ml of cells (36). Assuming cells are about 50% carbon, this corresponds to about 2 mmol of cellular carbon/ml. Assuming that 50% of the methanol used is converted to biomass by Methylobacterium strains (1) and correcting for the methanol-derived CO₂ incorporation (36) and the 0.4 mM formate accumulated, these figures suggest that approximately 4 mM methanol was converted to formate during this time period.

Nucleotide sequence accession numbers.

The sequences of the three chromosomal regions encoding the three formate dehydrogenases have been deposited with GenBank under accession numbers AF489516 (FDH1), AY183757 (FDH2), and AY181035 (FDH3).

RESULTS

Three different FDH enzymes are predicted to be encoded in the genome of M. extorquens AM1.

A novel, tungsten-containing NAD-linked FDH (FDH1) has recently been purified from M. extorquens AM1, and the genes encoding the two subunits of this enzyme have been identified (fdh1BA) next to each other on the chromosome. Based on FDH activity measurements, the existence of an additional, molybdenum-bound FDH in M. extorquens AM1 was predicted (20). Whole-genome analysis by both BLAST analysis and key word searches revealed the presence of two additional gene clusters potentially encoding FDH enzymes (Fig. 2): one cluster (fdh2CBAD) potentially encoding four subunits homologous to the subunits of the well-characterized molybdenum-dependent FDH of Ralstonia eutropha (9, 28), and another (fdhABC) potentially encoding three subunits homologous to the subunits of the well-characterized anaerobic FDH of Wolinella succinogenes (19, 21).

FIG. 2.

Clustering of genes encoding the three formate dehydrogenases.

While genes encoding polypeptides closely related to the FDH1 polypeptides were only identified in the genomes of Bradyrhizobium japonicum (77% identity for the alpha subunit) and the gammaproteobacterial methanotroph Methylococcus capsulatus (63% identity for the alpha subunit; http://tigrblast.tigr.org/ufmg/), homologs of FDH2 are encoded in many known genomes, the closest again being the counterparts from M. capsulatus (72% identity for the alpha subunit), and from a range of rhizobial species (about 62% identity; http://www.ncbi.nlm.nih.gov/). The closest homologs for FDH3 are encoded in the genomes of Vibrio, Ralstonia, Shewanella, and Bordetella species (55% identity for the alpha subunit; http://www.ncbi.nlm.nih.gov/), while only about 44% identity was observed with the polypeptides composing the anaerobic FDH of W. succinogenes (21).

All three FDH enzymes are dispensable for growth on methanol, but at least one FDH is required for growth on formate.

Mutants of each of the three FDHs were constructed (F1, F2, and F3) as were double mutants (F12, F23, and F13) and a triple mutant (F123). (Details of mutant generation are given in the Materials and Methods section and in Table 1.) Surprisingly, all of the mutants grew in the presence of methanol as the sole source of carbon and energy. The growth curves obtained for the mutant strains in the presence of methanol were essentially identical to the growth curves of the wild type. All of the single and double mutants were also positive for growth on formate, and only the triple mutant was unable to grow on formate. These results confirm that formate oxidation to CO₂ is the necessary energy-generating step during growth on formate. They also provide evidence that all three FDHs are functional and suggest that they are functionally redundant under the conditions tested. However, these results also imply that the formate oxidation step may be dispensable for growth of M. extorquens AM1 on methanol.

FDH1 and FDH2 expression varies depending on the growth conditions.

Although NAD-linked FDH activity can be measured in cell extracts (20), attempts to detect the activity of FDH3 with benzyl viologen as an artificial electron acceptor were not successful under either oxic or anoxic conditions. In order to distinguish between FDH1 and FDH2, NAD-linked FDH activity was measured in the wild-type and in each of the FDH mutant strains. As shown in Table 2, activity was not detected for either of the NAD-linked enzymes in succinate-grown cultures of M. extorquens AM1, but both enzymes were expressed in the appropriate mutants grown on methanol. In the F12 mutant as well as the triple mutant, no detectable FDH was present.

TABLE 2.

NAD-linked FDH activity in wild-type M. extorquens AM1 and FDH mutants

Strain	FDH activity (mU/mg) in cells growna:
	S Mo−W−	S Mo+W+	M Mo−W−	M Mo+W−	M Mo−W+	M Mo+W+
Wild type	<1	<1	150	110	170	120
F1	<1	<1	140	100	20	100
F2	<1	<1	100	20	180	200
F3	<1	<1	90	180	190	190
F12	ND	ND	<1	<1	<1	<1
F13	ND	ND	25	25	<1	15
F23	ND	ND	65	100	150	120
F123	ND	ND	<1	<1	<1	<1

^aS, succinate; M, methanol; − and +, absence and presence, respectively, of added trace elements; ND, not determined. The concentrations of Mo and W added in this experiment were 1 μM.

In order to further assess expression, regions upstream of these gene clusters were tested for promoter activity with the promoter probe vector pCM130 and xylE as a reporter gene (24). The 511-bp region separating fdh2C from the upstream gene was found to contain high reporter activities in cells grown on the C₁ compounds methanol and methylamine or induced in the presence of formaldehyde or formate, while no activity was detected in cells grown on succinate (data not shown). Even though the highest activities of XylE were observed in the presence of methanol (up to 1 U), the natural inducer must be a substrate derived from methanol, most probably formate itself, as no increase in XylE activity was observed after exposure to methanol in a mutant (UV26) lacking methanol dehydrogenase (data not shown). This mutant was unable to oxidize methanol to formaldehyde and did not produce formate from methanol.

Even though the results showed that all three FDHs must be expressed in the appropriate mutants, no promoter activity was detected for either 200-bp or 700-bp DNA regions upstream of the fdh3ABC cluster, and only low, unregulated activity (approximately 20 mU) was detected for the 900-bp DNA region upstream of fdh1BA, while none was detected for the 200-bp fragment upstream of this cluster. It is possible that the gene clusters encoding FDH1 and FDH3 are parts of larger transcriptional units or that they are not expressed in the wild type under these growth conditions. Their expression will be assessed at the RNA level in a separate study via whole-genome expression arrays.

Since FDH1 is a W-dependent NAD-linked enzyme (20) and FDH2 has been predicted to be an Mo-dependent, NAD-linked enzyme, it was possible that W and Mo might influence the expression and/or activity of these enzymes. Therefore, FDH activity was measured in extracts from cells grown in the presence of no added Mo or W, added Mo, added W, or both. Higher activity was detected for the W-dependent FDH (FDH1) when cells were grown in the presence of added W, while the putative Mo-dependent FDH (FDH2) did not require added Mo in the growth medium for high activity except when W was added. In both cases, growth in the presence of the other trace metal resulted in low activity when the catalytic metal was not added, suggesting an inhibitory effect of the respective trace elements. FDH2 activities were low in the FDH13 double mutant under all conditions tested, but FDH1 activities were comparable in both the FDH2 and FDH23 double mutants except that in the latter the activity was higher in the presence of W.

All three FDHs are functionally expressed, as demonstrated by in vivo NMR.

One predicted phenotype of a defect in formate oxidation capacity is formate accumulation in the medium. We used in vivo ¹³C NMR spectroscopy to detect intracellular and extracellular organic compounds originating from ¹³C-labeled methanol. The analysis was performed on the wild-type M. extorquens AM1, the three double mutants, and the triple mutant. For each experiment, the initial spectrum of the cell suspensions was taken. After the addition of [¹³C]methanol, spectra were recorded in 5-min blocks to follow labeled product formation over time. In experiments involving suspensions of wild-type cells, the formation of CO₂ (125.7 ppm) and HCO₃⁻ (161.3 ppm) was observed (Fig. 3). Due to the continuous flux of air through the cell suspension, no increase in the peak size of the volatile CO₂/HCO₃⁻ could be observed. Neither formaldehyde nor formate could be detected, indicating that formaldehyde and formate oxidation steps were not limiting in the oxidation of methanol to CO₂ in the wild-type strain.

FIG. 3.

Time course showing product formation by suspensions of M. extorquens AM1 incubated with 120 mM [¹³C]methanol. Before the experiment, cell suspensions were washed and resuspended in mineral medium (12) supplemented with 1 μM each of Mo and W; 125-MHz ¹³C NMR spectra were recorded in blocks of 5 min after the addition of methanol and are shown in the region from 60 to 180 ppm. The density of the cell suspensions (6 ml) was 15 mg (dry weight)/ml, the aeration rate was 38 ml/min, and the temperature was 30°C. For each spectrum, 125 scans were collected. The chemical shift for CO₂ is at 125.7 ppm, that for HCO₃⁻ is at 161.3 ppm, and that for formate is at 172.1; the signal corresponding to methanol is at 50.1 ppm (not shown).

In contrast, with the FDH triple mutant, cells in the same experiment resulted in accumulation of formate (172.1 ppm) within 2.5 min after the addition of [¹³C]methanol (Fig. 3), and the size of the corresponding peak increased linearly until all methanol was consumed (340 min) (not shown). The signal was present in the supernatant following centrifugation of the suspensions, indicating that the labeled formate was excreted into the medium and not accumulated within the cells. No peak indicative of CO₂/HCO₃⁻ could be detected during methanol consumption, suggesting that formate was not oxidized to CO₂ in significant amounts in this mutant.

When the double FDH mutants were employed in similar experiments, no formate accumulation and wild-type production of CO₂ and HCO₃⁻ were observed for F12 and F23. However, the pattern observed for the F13 double mutant was intermediate between those of the wild-type and F123: both formate and the CO₂/HCO₃⁻ species were observed upon [¹³C]methanol addition (not shown).

These experiments show that each of the three FDHs is functionally expressed, including FDH3, whose activity was not detected in cell extracts, since any single enzyme is capable of consuming the formate produced, but the triple mutant accumulated formate. In addition, it appears that FDH1 and FDH3 alone allow formate conversion to CO₂ at a nonlimiting rate, and so no formate accumulation was observed, whereas oxidation of formate by FDH2 must have occurred at a lower rate, as formate could be detected as a labeled product in addition to CO₂/HCO₃⁻. These results are in keeping with the in vitro enzyme activity data, as FDH2 activity was low in the FDH13 mutant (Table 2).

Formate accumulation is transient during growth of the triple mutant on methanol.

A longer time course of formate excretion into the medium was measured during growth of wild-type M. extorquens AM1 and the FDH triple mutant on methanol in batch cultures. Samples were taken at different stages of growth and centrifuged to remove the cells, and the supernatant was analyzed for formate concentration by HPLC. Low formate levels (below 60 μM) were detected in samples taken during mid- to late exponential growth in wild-type M. extorquens AM1 (Fig. 4). However, in the triple FDH mutant, the concentration of formate rose significantly in early to mid-exponential phase (up to approximately 400 μM) and then decreased (Fig. 4). During the time period in which formate accumulated, the flux of methanol to formate should have generated on the order of 4 mM in the medium, based on carbon conversion values (see Materials and Methods) if all formate that was normally converted to CO₂ were excreted. These results suggest that about 10% of this formate was excreted initially but even that was consumed later in growth. The lack of detectable ¹³CO₂/H¹³CO₃ in the triple mutant during the NMR experiments described above rules out the presence of an alternative, unrecognized FDH of sufficient activity to account for the formate utilized. However, it is still possible that a more minor FDH activity is present that could not be detected by any of the methods used here.

FIG. 4.

Transient formate accumulation upon growth of M. extorquens AM1 and an FDH triple mutant in the presence of methanol. Cells (600 ml) were grown in a standard minimal medium in 2-liter Erlenmeyer flasks at 180 rpm and 30°C. Formate concentrations (indicated by bars) were determined by HPLC as described (27).

DISCUSSION

The occurrence of multiple FDHs is common in microbes. For example, three different FDHs are expressed and intricately regulated in Escherichia coli (34). In this work we identified three different FDH enzymes in the facultative methylotroph M. extorquens AM1 and investigated the role of the formate oxidation step in its C₁ metabolism. In the past, a significant role was predicted for the FDH step in energy generation in the form of NADH in serine cycle methylotrophs (1). We have shown here that all three FDHs are functional in M. extorquens AM1, and our results suggest that expression of these enzymes is regulated at least in part by the presence of Mo or W. Such a regulatory pattern may suggest that the presence of multiple FDHs provides a means for ecological fitness to organisms exposed to varied metal availability in their environments. However, in laboratory conditions, the functions of these three enzymes appear redundant, and any one of the three enzymes is sufficient both to account for growth on formate and to accommodate the flux of formate resulting from the oxidation of more reduced C₁ compounds, such as methanol. As predicted (1), formate oxidation to CO₂ was shown to be required for growth on formate.

The finding that the three FDH enzymes present in M. extorquens AM1 are not essential players in energy metabolism during growth on substrates more reduced than formate was surprising. Formate oxidation to CO₂ is predicted to produce significant amounts of reducing equivalents during methylotrophic metabolism (36), and it would be expected that if the conversion of formate to CO₂ were blocked, it would severely affect growth on methanol. However, in the triple FDH mutant, although formate did accumulate transiently it did not reach the levels predicted from biomass conversion values, and later in the growth cycle it was consumed again. Therefore, it appears that in the absence of formate conversion to CO₂ by FDH, this formate is consumed by an alternative route. The fact that the growth rate is normal under these conditions suggests that this route generates net reducing equivalents.

Since no CO₂/HCO₃⁻ was detected in the triple FDH mutant following methanol addition (Fig. 3), the presence of an unrecognized and novel FDH expressed at sufficient levels to accommodate this flux of formate did not seem plausible. The only other obvious pathway for utilizing formate is the reversible H₄F-linked C₁ transfer pathway involving formyl-H₄F ligase, methenyl-H₄F cyclohydrolase, and methylene-H₄F dehydrogenase (Fig. 1) (37). Once methylene H₄F is generated from formate, it can enter the serine cycle and produce multicarbon compounds (Fig. 1). From these intermediates, it is theoretically possible to generate net reducing equivalents by oxidation to CO₂ via derepression of the tricarboxylic acid cycle. Deregulation of the tricarboxylic acid cycle on methanol has already been observed in a mutant in which regulation of poly-β-hydroxybutyrate synthesis is affected (17). Therefore, at least one possibility exists to explain how formate might be consumed in the triple mutant. Clearly, further work will be required to determine the pathways that oxidize formate in the triple FDH mutant.