Methylamine Utilization via the N-Methylglutamate Pathway in Methylobacterium extorquens PA1 Involves a Novel Flow of Carbon through C₁ Assimilation and Dissimilation Pathways

Abstract

Methylotrophs grow on reduced single-carbon compounds like methylamine as the sole source of carbon and energy. In Methylobacterium extorquens AM1, the best-studied aerobic methylotroph, a periplasmic methylamine dehydrogenase that catalyzes the primary oxidation of methylamine to formaldehyde has been examined in great detail. However, recent metagenomic data from natural ecosystems are revealing the abundance and importance of lesser-known routes, such as the N-methylglutamate pathway, for methylamine oxidation. In this study, we used M. extorquens PA1, a strain that is closely related to M. extorquens AM1 but is lacking methylamine dehydrogenase, to dissect the genetics and physiology of the ecologically relevant N-methylglutamate pathway for methylamine oxidation. Phenotypic analyses of mutants with null mutations in genes encoding enzymes of the N-methylglutamate pathway suggested that γ-glutamylmethylamide synthetase is essential for growth on methylamine as a carbon source but not as a nitrogen source. Furthermore, analysis of M. extorquens PA1 mutants with defects in methylotrophy-specific dissimilatory and assimilatory modules suggested that methylamine use via the N-methylglutamate pathway requires the tetrahydromethanopterin (H₄MPT)-dependent formaldehyde oxidation pathway but not a complete tetrahydrofolate (H₄F)-dependent formate assimilation pathway. Additionally, we present genetic evidence that formaldehyde-activating enzyme (FAE) homologs might be involved in methylotrophy. Null mutants of FAE and homologs revealed that FAE and FAE2 influence the growth rate and FAE3 influences the yield during the growth of M. extorquens PA1 on methylamine.

INTRODUCTION

Methylated amines are a large group of biologically, industrially, and environmentally relevant compounds. The simplest methylated amine, (mono-)methyl amine (MA), is produced by anaerobic bacteria that decarboxylate nitrogenous organic matter in the gastrointestinal tract or animal waste and decompose methylated osmolytes, such as glycine betaine in the marine environment (1, 2). In addition, a growing fraction of MA is being released during the degradation of nitrogen-containing herbicides and pesticides (3), biomass combustion (1), and industrial processes, such as fish processing. MA also plays an important role in the global carbon and nitrogen budget (2, 4). However, MA does not accumulate in aerobic environments (4) because it gets rapidly consumed by methylotrophs that oxidize reduced single-carbon (C₁) substrates to produce biomass and CO₂ (5) and serves as a nitrogen source for several classes of aerobic microorganisms (6, 7).

There are three well-characterized enzymes/pathways used by methylotrophs to oxidize MA. In many Gram-negative methylotrophs, such as Methylobacterium extorquens AM1, MA is oxidized by a periplasmic methylamine dehydrogenase (MaDH) (8, 9) to formaldehyde. Methylamine oxidase, a Cu-containing enzyme present in Gram-positive methylotrophs (10) and methylotrophic yeast strains (11), oxidizes MA to formaldehyde (12) in a single step as well. In contrast, the N-methylglutamate (NMG) pathway, originally described in Pseudomonas spp. (now Aminobacter spp. [13]), is a complex, multistep pathway in which the methyl group from MA is transferred to glutamate to form two novel amino acid derivatives, NMG and γ-glutamylmethylamide (GMA), which are then oxidized to release formaldehyde (14). The NMG pathway involves at least three unique enzymes: N-methylglutamate synthase (NMGS), N-methylglutamate dehydrogenase (NMGDH), and γ-glutamylmethylamide synthetase (GMAS) (3, 15–17). While the enzymes involved in the NMG pathway were purified and characterized over 3 decades ago (18–23), the genes were discovered very recently (3, 15, 17). Genes encoding the enzymes of the NMG pathway are often clustered in an ∼10-kb genomic region (Fig. 1). A sarcosine oxidase-like cluster of four genes encodes NMGDH (mgdABCD), a glutamate synthase-like cluster of three genes encodes NMGS (mgsABC), and a glutamine synthetase-like gene (gmaS) encodes GMAS. Since the discovery of the NMG pathway gene cluster, sequence-based analysis has revealed that the NMG pathway is present in a large fraction of methylotrophs (even those containing MaDH) as well as nonmethylotrophs, where it serves to extract the nitrogen from MA (6).

FIG 1.

(A) Arrangement of genes encoding the N-methylglutamate pathway on the chromosome of nonmethylotrophic proteobacteria (like Agrobacterium tumefaciens) and methylotrophic proteobacteria (like Hyphomicrobium sp. strain MC1 and Methylobacterium extorquens PA1). Blue, genes encoding the four subunits of NMGDH (mgdABCD); red, genes encoding the three subunits of NMGS (mgsABC); orange, a single gene encoding GMAS (gmaS). TR, transcriptional regulator; purU, gene encoding formyl-tetrahydrofolate hydrolase; folD, gene encoding bifunctional methylene tetrahydrofolate dehydrogenase and methenyl tetrahydrofolate cyclohydrolase; TCA, tricarboxylic acid.

Recent genetic analyses of the NMG pathway have produced somewhat contradictory results in terms of its topology. A study using Methyloversatilis universalis FAM5 (a facultative methylotroph of the betaproteobacteria) (15) showed that the enzyme γ-glutamylmethylamide synthetase was not essential during growth on low concentrations of MA. This observation supported a notion that γ-glutamylmethylamide synthetase is a bidirectional enzyme that is essential only at high concentrations of MA because it acts as a gatekeeper and keeps the flux of methylamine under control (15). However, in Methylocella silvestris BL2 (a facultative methanotroph of the alphaproteobacteria) (3) and, very recently, Methylobacterium extorquens DM4 (17), γ-glutamylmethylamide synthetase was shown to be essential for growth on MA. These studies supported a topology for the NMG pathway in which γ-glutamylmethylamide synthetase plays an essential role, directly or indirectly, in the formation of N-methylglutamate (24).

Additionally, uncertainty surrounding the end product of methylamine oxidation via the NMG pathway has occluded our understanding of how the NMG pathway feeds into the C₁ assimilatory and dissimilatory modules. For a long time, it was assumed that methylamine oxidation mediated by the NMG pathway led to the production of formaldehyde (20). Then, as observed during methanol growth in M. extorquens species (25), the tetrahydromethanopterin (H₄MPT)-dependent formaldehyde oxidation pathway (26, 27) would serve as the primary dissimilatory module. Formate, thus produced, would be incorporated into biomass through the tetrahydrofolate (H₄F)-dependent C₁ assimilation pathway (28–31) and the serine cycle (32, 33) (see Fig. S1 in the supplemental material). Of late, genes encoding a dissimilatory version of the H₄F-dependent C₁ transfer pathway (34) (characterized by the presence of folD and purU instead of mtdA, fch, and ftfL) have been found in the vicinity of the NMG pathway gene cluster (15) (Fig. 1). Additionally, N-methylglutamate dehydrogenase (mgdC) contains an H₄F binding domain and produces methylene-H₄F (CH₂=H₄F) in the presence of H₄F in vitro (15). In light of these recent findings, CH₂=H₄F, instead of formaldehyde, has been proposed to be the end product of the NMG pathway. Like in the case of chloromethane metabolism by M. extorquens CM4 (34, 35), CH₂=H₄F as the end product of primary oxidation would seem to circumvent the need for an H₄MPT-dependent formaldehyde oxidation pathway. A portion of the CH₂=H₄F would get assimilated via the serine cycle, while the rest would be oxidized by the dissimilatory branch of the H₄F-dependent C₁ transfer pathway (see Fig. S2 in the supplemental material). However, with burgeoning sequence data, we now know that methylotrophs that encode the NMG pathway often do not contain either folD or purU (data not shown). How do these methylotrophs oxidize CH₂=H₄F to formate, then? Or does MA oxidation via the NMG pathway lead to the production of some free formaldehyde, instead?

In order to better understand how the NMG pathway operates in vivo, we genetically dissected its physiology in M. extorquens PA1 (36). Although PA1 utilizes methanol in the same way as the better-characterized AM1 strain (25), here we find that it lacks methylamine dehydrogenase and possesses only the NMG pathway for MA use. By creating mutant strains with lesions in one or more genes of the NMG pathway and assaying the growth patterns of each of these strains on MA, we uncovered that the NMG pathway has a linear topology in PA1 akin to that very recently suggested for DM4 (17) and M. silvestris BL2 (3). Methylamine growth of null mutants with mutations in genes encoding enzymes of modules involved in C₁ transfer pathways revealed that the H₄MPT-dependent formaldehyde oxidation pathway was required for C₁ dissimilation, whereas an intact H₄F-mediated C₁ transfer pathway was not required for either formate assimilation or dissimilation of CH₂=H₄F. Further, we unveiled that FAE homologs, often found in the genomes of methylotrophs (FAE2, FAE3) (37, 38), play a functional role during methylamine growth using the NMG pathway. The presence of either FAE or FAE2 significantly increased the growth rate but not the yield on methylamine, whereas the presence of FAE3 increased the yield on methylamine.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

Strains relevant to this study included the following: the Δcel strain of the pink-pigmented wild-type (WT) stock of M. extorquens AM1 (CM2720) and the Δcel strain of the pink-pigmented wild-type stock of M. extorquens PA1 (CM2730), as described elsewhere (39). Unless specified, all growth conditions utilized a modified version of Hypho minimal medium, consisting of 100 ml phosphate salts solution (25.3 g of KH₂PO₄ plus 22.5 g Na₂HPO₄ in 1 liter deionized water), 100 ml sulfate salts solution [5 g of (NH₄)₂SO₄ and 2 g of MgSO₄·7 H₂O in 1 liter deionized water], 799 ml of deionized water, and 1 ml of trace metal solution (40). A nitrogen-free version of the modified Hypho minimal medium used the same recipe described above but with a nitrogen-free sulfate salts solution (7 g of MgSO₄·7 H₂O in 1 liter deionized water). All components were autoclaved separately before mixing under sterile conditions. Filter-sterilized carbon sources were added just prior to inoculation in liquid minimal medium with (final concentrations) 3.5 mM sodium succinate, 15 mM methylamine hydrochloride, 15 mM methanol, and 3.5 mM sodium succinate and with 7.66 mM methylamine as a nitrogen source.

Growth rate measurements.

Apart from growth on methylamine as a carbon source, all other growth measurements were conducted using an automated growth platform (41). All growth regimes consisted of three cycles consisting of inoculation, acclimation, and growth measurement. All strains were stored in vials at −80°C in 10% dimethyl sulfoxide; growth was initiated by transferring 10 μl freezer stock into 10 ml of Hypho medium with 3.5 mM succinate in 50-ml flasks (covered with a 50-ml plastic beaker) in an incubator shaking at 225 rpm, and the strains were maintained at 30°C. Upon reaching stationary phase (∼2 days), the cultures were transferred 1:64 into fresh medium (in 48-well microtiter plates [CoStar 3548]) with the carbon source to be tested to a final volume of 640 μl, allowed to reach saturation in this acclimation phase, and again diluted 1:64 into fresh medium for the measured (experimental) growth. The 48-well microtiter plates were incubated in a room maintained at 30°C and 80% humidity in an incubation tower (Liconic USA LTX44 with custom-fabricated cassettes) shaking at 650 rpm (39). The increase in the optical density at 600 nm (OD₆₀₀) for strains grown in 48-well microtiter plates was measured using an automated, robotic culturing and monitoring system (41). A series of robotic instruments (including a shovel, a transfer station, and a twister arm), all controlled by the Clarity manager, were used to move the 48-well plates from the incubation tower to a Perkin-Elmer Victor2 plate reader for OD₆₀₀ measurements. The dynamics and specific growth rates of the cultures were calculated from the log-linear growth phase using open-source, custom-designed growth analysis software called CurveFitter (available at http://www.evolvedmicrobe.com/CurveFitter/).

Growth on methylamine as a carbon source was below the detection threshold of the automated growth platform described above. Growth from freezer stocks at −80°C was initiated by transferring 10 μl freezer stock into 10 ml of Hypho medium with 3.5 mM succinate in 50-ml flasks (covered with a 50-ml plastic beaker) in an incubator shaking at 225 rpm, and the culture was maintained at 30°C. Upon reaching stationary phase (∼2 days), cultures were transferred 1:16 into 9.4 ml fresh medium with 15 mM methylamine hydrochloride, allowed to reach saturation in this acclimation phase (∼3 days), and again diluted 1:16 into 9.4 ml fresh medium with 15 mM methylamine for the measured (experimental) growth (∼3 days). A 50-μl aliquot of three replicate cultures for each strain was sampled every 8 to 10 h during the growth phase. The OD₆₀₀ of the culture was measured using a spectrophotometer (Bio-Rad). The dynamics and specific growth rates of the cultures were calculated from the log-linear growth phase. The yield was measured as the maximum OD₆₀₀ during the growth phase. The growth rate and yield reported for each strain and condition are the means calculated from three biological replicates, unless otherwise noted.

Methanol and methylamine shock experiments.

Ten microliters of the freezer stock of strains CM3803 (CM2730 ΔmptG) and CM2730 were grown in 48-well microtiter plates with 3.5 mM succinate as the carbon source. Upon reaching stationary phase (∼2 days), CM3803 and CM2730 were transferred 1:64 into 630 μl of fresh medium with 3.5 mM succinate in two 48-well microtiter plates. After 12 h of incubation, a pulse of methanol resulting in final concentrations of (i) 1 mM methanol, (ii) 5 mM methanol, (iii) 10 mM methanol, and (iv) 50 mM methanol was added to triplicate cultures of CM3803 and CM2730 in one 48-well plate. Similarly, after 12 h of incubation, a pulse of methylamine resulting in final concentrations of (i) 1 mM methylamine, (ii) 5 mM methylamine, (iii) 10 mM methylamine, and (iv) 50 mM methylamine was added to triplicate cultures of CM3803 and CM2730 in the other 48-well plate. Triplicate control wells of CM2730 and CM3803 (to which no methanol and methylamine were added) were maintained in each 48-well plate as well. The increase in the OD₆₀₀ was measured as described earlier.

Generation of mutant strains.

M. extorquens PA1 deletion mutants lacking mgsABC, fae2, fae3, gmaS, and mgdABCD were generated on the genetic background of CM2730 using the allelic exchange vector pCM433 (44). The double-deletion mutants lacking fae and fae2 or fae3 were generated on the genetic background of CM3753 (CM2730 Δfae) using the allelic exchange vector pCM433 (44). The double-deletion mutants lacking fae2 and fae3 were generated on the genetic background of CM3757 (CM2730 Δfae2) using the allelic exchange vector pCM433 (44). The double-deletion mutants lacking mgsABC and gmaS were generated on the genetic background of CM3733 (CM2730 ΔmgsABC) using the allelic exchange vector pCM433 (44).The triple-deletion mutants lacking fae, fae2, and fae3 were generated on the genetic background of CM4208 (CM2730 Δfae Δfae3) using the allelic exchange vector pCM433 (44). A region upstream and downstream of each of these genes or operons of ∼0.5 kb was amplified using PCR. The forward primer for the upstream flank was designed to have a 30-bp-long sequence at the 5′ end homologous to the sequence upstream of the NotI cut site in pCM433. The reverse primer for the upstream flank was designed to have a 30-bp sequence at the 5′ end homologous to the first 30 bp of the downstream flank. The reverse primer for the downstream flank was designed to have a 30-bp-long sequence at the 5′ end homologous to the sequence downstream of the NotI cut site in pCM433. The PCR products representing the upstream and downstream flanks were simultaneously ligated into the pCM433 vector cut with NotI using a three-part Gibson assembly protocol described elsewhere (45). Cloning of the upstream and downstream flanks for fae2, fae3, mgsABC, gmaS, and mgdABCD in pCM433 resulted in pDN51, pDN52, pDN53, pDN54, and pDN55, respectively. Mutant strains of M. extorquens PA1 were made by introducing the appropriate donor constructs through conjugation by a triparental mating between NEB 10β competent cells (New England BioLabs, Ipswich, MA) containing the donor construct, an Escherichia coli strain containing the conjugative plasmid pRK2073, and PA1, as described elsewhere (44). All mutant strains were confirmed by a diagnostic PCR analysis and validated by Sanger sequencing of the mutant locus. All strains and plasmids used and generated for this study are listed in Table 1.

TABLE 1.

M. extorquens strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
Strains
CM2730	M. extorquens PA1 Δcel	39
CM3733	CM2730 ΔmgsABC	This study
CM3753	CM2730 Δfae	25
CM3757	CM2730 Δfae2	This study
CM3761	CM2730 Δfae3	This study
CM3765	CM2730 ΔgmaS	This study
CM3769	CM2730 ΔmgdABCD	This study
CM3773	CM2730 ΔftfL	25
CM3799	CM2730 ΔglyA	25
CM3803	CM2730 ΔmptG	25
CM3835	CM2730 ΔmgsABC ΔgmaS	This study
CM4208	CM2730 Δfae Δfae3	This study
CM4230	CM2730 Δfae Δfae2	This study
CM4232	CM2730 Δfae2 Δfae3	This study
CM4244	CM2730 Δfae Δfae2 Δfae3	This study
Plasmids
pCM433	Allelic exchange vector (Tetr Sucs)	44
pDN50	pCM433 with Δfae upstream and downstream flanks	25
pDN51	pCM433 with Δfae2 upstream and downstream flanks	This study
pDN52	pCM433 with Δfae3 upstream and downstream flanks	This study
pDN53	pCM433 with ΔmgsABC upstream and downstream flanks	This study
pDN54	pCM433 with ΔgmaS upstream and downstream flanks	This study
pDN55	pCM433 with ΔmgdABCD upstream and downstream flanks	This study
pRK2073	Helper plasmid (IncP tra)	53

RESULTS

γ-Glutamylmethylamide synthetase is conditionally essential during methylamine growth in M. extorquens PA1.

Our strain of PA1 lacking the cel locus (encoding cellulose biosynthesis), to prevent clumping and permit accurate measurement of growth characteristics, is referred to here as the WT (39) and could grow on 15 mM MA with an average growth rate of 0.042 h⁻¹ and an average final OD₆₀₀ of 0.712 (three biological replicates were used throughout, unless otherwise noted). A 10-kb genomic region encoding enzymes of the NMG pathway was recently characterized in DM4 (17). In PA1, a completely syntenic cluster of nine genes (of which eight had >95% amino acid identity and one had ∼93% amino acid identity to the corresponding genes in DM4) encoding the enzymes of the NMG pathway was observed as well (Fig. 1). To determine if the NMG pathway is active and involved in methylamine oxidation, we made and tested the growth of null mutants lacking individual enzymes of the NMG pathway on MA. No measurable change in the OD₆₀₀ was observed for the ΔmgdABCD mutant, the ΔmgsABC mutant, or the ΔgmaS mutant on MA as the sole carbon source. Contrary to the findings of studies in other methylotrophs (15), we did not detect growth for the ΔgmaS mutant on any concentration of MA ranging from 1 to 25 mM after 7 days or after a 14-day incubation on 15 mM MA. The essential role of all three enzymes (especially GMAS) for methylamine oxidation supported a linear topology for the NMG pathway in PA1.

In order to even more sensitively test whether MA could be converted by strains with defects in different enzymes of the NMG pathway, we took advantage of the fact that MA can also be used as a nitrogen source, as ammonia (NH₄⁺) ions are a by-product of MA oxidation (Fig. 2B). When MA is used only as a nitrogen source, the stoichiometric ratio of biomass (46) and carbon yield of the serine cycle (43) indicates that cells need to take in 15- to 20-fold more carbon than nitrogen. Therefore, growth on succinate with MA as a nitrogen source is 5-fold faster than that on MA as a carbon source but is still contingent on a functional NMG pathway. We leveraged these fast growth conditions to detect subtle phenotypes for NMG pathway-null mutants and further verify the linear topology of the NMG pathway. While the ΔmgsABC mutant and the ΔmgsABC ΔgmaS double mutant could not use MA as a nitrogen source (Fig. 2A), the ΔgmaS mutant grew poorly in succinate with MA as the nitrogen source, lacking the typical log-linear growth characteristics (Fig. 2A). The marginal ability of the ΔgmaS mutant to use MA as a nitrogen source indicated that the conversion of MA to NMG (with the concomitant release of NH₄⁺) catalyzed by NMGS is physiologically relevant only when GMAS is absent. Additionally, the inability of the ΔmgsABC mutant to use MA as a nitrogen source corroborated the lack of a GMA-dissimilating enzyme in PA1. Altogether, these data suggest an essential role for GMAS during growth on MA as a carbon source but not during growth on MA as a nitrogen source in PA1, as indicated in Fig. 2B.

FIG 2.

(A) OD₆₀₀-versus-time plots for three replicate cultures of (i) the Δcel wild-type strain of M. extorquens PA1, (ii) the ΔmgsABC mutant, (iii) the ΔgmaS mutant, and (iv) the ΔmgsABC ΔgmaS double mutant in nitrogen-free medium with 3.5 mM succinate and 7.66 mM methylamine. (B) The topology of the N-methylglutamate pathway in M. extorquens PA1 is linear; NMGS is capable of synthesizing NMG from GMA as well as methylamine, albeit only in the ΔgmaS mutant.

The H₄MPT-dependent formaldehyde oxidation pathway is required during growth on MA with the NMG pathway.

It was recently proposed that, instead of formaldehyde, the oxidized C₁ compound emerging from the NMG pathway might actually be CH₂=H₄F (15). If so, this end product should obviate the H₄MPT-mediated formaldehyde oxidation pathway during growth on MA. We tested the hypothesis that methylamine growth mediated by the NMG pathway avoids formaldehyde production by examining a mutant lacking mptG (encoding β-ribofuranosylaminobenzene 5′-phosphate synthase, the first enzyme of the H₄MPT biosynthesis pathway [42]). Previous work with AM1 has shown that methanol oxidation leads to formaldehyde buildup and rapid cell death in the ΔmptG mutant and that MA has the same effect, albeit to a lesser extent (26). We measured the capacity of the ΔmptG mutant of PA1 to withstand conditions that might lead to the production of variable amounts of formaldehyde at different rates. First, we tested the capacity of the ΔmptG mutant to tolerate different doses of methanol shock and MA shock during growth on a multicarbon substrate (succinate). As had been seen with AM1, even a pulse of 1 mM methanol completely inhibited the growth of the ΔmptG mutant of PA1 on succinate (Fig. 3A). In contrast, a pulse of 50 mM MA did not impede the growth of the ΔmptG mutant in succinate (Fig. 3B). Since the resilience of the ΔmptG mutant to MA shock might also be an artifact of tight regulation of the NMG pathway (17), we then tested whether this mutant could use MA as the sole nitrogen source. The WT, the ΔmptG mutant, and the ΔmgsABC mutant were grown to saturation on succinate with NH₄⁺ as the nitrogen source and were then diluted 64-fold in medium with succinate as the carbon source and MA as the sole nitrogen source. After ∼16 h, the ΔmgsABC mutant could no longer grow, whereas the WT and the ΔmptG mutant continued to grow (Fig. 3C). These growth data suggested that once the internal nitrogen reservoir ran out, the ΔmptG mutant could still use MA as a nitrogen source; growth was moderately compromised (the growth rate was 23% lower on succinate with MA as the nitrogen source than with NH₄⁺ as the nitrogen source; P < 0.001 for the difference in the mean growth rate of three biological replicates compared using the Student t test). The growth defect incurred by using MA versus NH₄⁺ as a nitrogen source was more severe for the ΔmptG mutant than for the WT, where the growth rate on succinate with MA as the nitrogen source was merely 11% lower than that on succinate with NH₄⁺ as the nitrogen source (P = 0.001) (see Table S1 in the supplemental material).

FIG 3.

(A) OD₆₀₀-versus-time plots for three replicate cultures of the Δcel wild-type strain of PA1 and the ΔmptG mutant in medium with 3.5 mM succinate. A pulse of 1 mM methanol was added to the growth medium 12 h after incubation (as indicated by the arrow). (B) OD₆₀₀-versus-time plots for three replicate cultures of the Δcel wild-type strain of PA1 and the ΔmptG mutant in medium with 3.5 mM succinate. A pulse of 50 mM methylamine was added to the growth medium 12 h after incubation (as indicated by the arrow). (C) Three replicate cultures of the Δcel wild-type strain of PA1, the ΔmptG mutant, and the ΔmgsABC mutant, grown to saturation in medium with 3.5 mM succinate, were transferred (1/64-fold dilution) to nitrogen-free medium with 3.5 mM succinate and 50 mM methylamine (as indicated by the arrow). The OD₆₀₀-versus-time plots follow the growth of each of these three genotypes in nitrogen-free medium with 3.5 mM succinate and 50 mM methylamine.

Next, we examined the ability of the ΔmptG mutant strain to grow on MA as the sole carbon source. No observable growth or significant change in the OD₆₀₀ was detected for the ΔmptG mutant strain on MA as a carbon source (Fig. 4). In order to quantify the extent of H₄MPT-dependent formaldehyde oxidation required during growth on MA, we used a strain lacking fae (encoding a formaldehyde-activating enzyme that catalyzes the condensation of free formaldehyde with H₄MPT [47]). A marginal rate of CH₂=H₄MPT production can be sustained by the spontaneous condensation of formaldehyde and H₄MPT in a Δfae mutant (47). Therefore, the fae deletion only partially abolishes the H₄MPT-dependent formaldehyde oxidation pathway and leads to an intermediate mutant phenotype in terms of methanol sensitivity (26). Unlike the ΔmptG strain, the strain with the fae deletion showed WT-like growth on MA (Fig. 4). During growth on MA, the absolute requirement for the H₄MPT-mediated formaldehyde oxidation pathway indicates that at least some of the carbon from MA is ending up as formaldehyde, but the low rate of formaldehyde oxidation needed can be sustained in the absence of FAE.

FIG 4.

Ratios of growth rates of mutants with lesions in various methylotrophy-specific modules to the growth rate of the Δcel wild-type strain PA1 on 15 mM methylamine. The Δfae and ΔmptG mutants represent strains with partial and complete lesions in the H₄MPT-dependent pathway for formaldehyde oxidation (or C₁ dissimilation), respectively. The ΔftfL mutant represents a strain with a lesion in the H₄F-dependent pathway for formate assimilation, and the ΔglyA mutant represents a strain with a lesion in the serine cycle for C₁ assimilation. Error bars represent the 95% confidence intervals of the mean ratio for triplicate cultures for each condition.

An intact H₄F pathway is not required for assimilation or dissimilation during methylamine growth of M. extorquens PA1.

The growth of M. extorquens on any methylotrophic compound yet tested requires an intact H₄F pathway for either assimilation (i.e., growth on methanol) or dissimilation (i.e., growth on chloromethane by CM4). Given that at least some of the carbon from MA appears to require the H₄MPT pathway for dissimilation, we tested whether a complete H₄F-dependent C₁ transfer pathway was also required for growth on MA (Fig. 4). A strain lacking ftfL (encoding formyl-tetrahydrofolate ligase, which catalyzes the reduction of formate and free H₄F to formyl-H₄F in an ATP-dependent manner [29]) cannot grow on C₁ compounds like methanol and formate (25) because of an early block in assimilation (29). We recently reported that a ΔftfL mutant of PA1 has the same phenotype as the ΔftfL mutant of AM1 (25). Contrary to the findings for other C₁ compounds, the ΔftfL mutant could grow on MA at a rate indistinguishable from that of the WT (P = 0.8887). In contrast, a mutant lacking glyA (encoding serine hydroxymethyltransferase [48]) was incapable of growth on MA, confirming that C₁ units are still assimilated via the serine cycle. These data, especially the nonessentiality of ftfL, indicate that use of the NMG pathway for growth on MA involves the flow of carbon through the C₁-specific metabolic pathways such that the H₄MPT-dependent pathway handles oxidation and CH₂=H₄F enters directly into the serine cycle. This observation is consistent with either formaldehyde or CH₂=H₄F (or both) being a branch point of metabolism during growth on MA, as depicted in Fig. 5C and D, in contrast to formate, which is the branch point of metabolism during growth on methanol (30, 31), as depicted in Fig. 5A.

FIG 5.

(A) Schematic representation of methanol dehydrogenase (MDH)- and MaDH-mediated methylamine metabolism in M. extorquens. Primary oxidation generates free formaldehyde, which is oxidized to formate by the H₄MPT-dependent C₁ dissimilation pathway. Formate is reduced by the H₄F-dependent C₁ assimilation pathway and assimilated using the serine cycle. The reaction highlighted in red, mediated by FtfL (formyl-tetrahydrofolate ligase), is unique to this topology. (B) Schematic representation of chloromethane metabolism in M. extorquens CM4. Primary oxidation produces H₄F derivatives, using an H₄F-dependent C₁ dissimilation pathway, and completely bypasses formaldehyde production. Methylene tetrahydrofolate (CH₂=H₄F) is assimilated using the serine cycle. The reactions highlighted in orange, which are catalyzed by enzymes of the H₄F-dependent C₁ dissimilatory pathway (MetF, FolD, and PurU), are unique to this topology. (C) Schematic representation of the metabolic network involved in methylamine growth if the end product of the N-methylglutamate pathway is formaldehyde. The condensation of formaldehyde and H₄F to form CH₂=H₄F, highlighted in blue, is unique to this topology. (D) Schematic representation of the metabolic network involved in methylamine growth if the end product of the N-methylglutamate pathway is CH₂=H₄F. CH₂=H₄F enters the H₄MPT-mediated C₁ assimilation pathway either by dissociation to formaldehyde (I) or by a direct cofactor switch (II). These two alternate reactions, indicated by green arrows, are unique to this topology.

FAE homologs play a functional role during growth on MA in PA1.

Do any unique gene products play a role in rerouting carbon through the C₁-specific metabolic pathways when the NMG pathway operates? Many methylotrophs, including PA1, have two distant FAE homologs, fae2 (Mext_3143) and fae3 (Mext_1450), in the genome, yet their role in methylotrophy, if any, remains unknown (37, 49). We hypothesized that these gene products may play a role during growth on MA, perhaps catalyzing some of the novel reactions in Fig. 5C and D. To test our hypothesis, we deleted fae, fae2, and fae3 individually and in all possible combinations (Fig. 6A and B) and tested the growth of each of these mutants on 15 mM MA. Neither fae2 nor fae3 was essential for growth on MA since all the single-, double-, and triple-knockout mutants could grow on MA. The growth rates of strains with either fae or fae2 were mostly indistinguishable from the growth rate of PA1 or significantly higher in the case of the Δfae2 mutant (12%; P = 0.0108). However, mutant strains lacking both fae and fae2 showed a significant growth defect on MA (Fig. 6A). In addition, all mutant strains lacking fae3 showed a significant yield (maximum OD₆₀₀) defect relative to the yield of the WT on MA (Fig. 6B).

FIG 6.

(A) Red, blue, and green arrows represent the deletion of fae, fae2, and fae3, respectively, in the genomic background from which the arrow originates. The numbers and signs on the arrows represent the percent increase/decrease in growth rate as a result of the gene deletion. For percentage values shown in bold and parentheses, the P value is <0.05 (with the difference in the growth rates of three replicate cultures of the two strains being compared using the Student t test). For percentage values shown in gray, the difference in growth rate between the two strains is not significant (P > 0.05). (B) Red, blue, and green arrows represent the deletion of fae, fae2, and fae3, respectively, in the genomic background from which the arrow originates. The numbers and signs on the arrows represent the percent increase/decrease in yield as a result of the gene deletion. For percentage values shown in bold and parentheses, the P value is <0.05 (with the difference in the maximum OD₆₀₀ of three replicate cultures of the two strains being compared using the Student t test). For percentage values shown in gray, the difference in yield between the two strains is not significant (P > 0.05).

To determine whether the growth characteristics of the fae-, fae2-, and fae3-null mutants described in Fig. 6A and B are limited to growth on MA via the NMG pathway, we tested the growth of each of these mutants on another C₁ compound (methanol), a multi-C compound (succinate), and succinate with MA as the nitrogen source (see Fig. S3 in the supplemental material). As expected, strains with a fae deletion did not grow on methanol. Deletion of either fae2 or fae3 led to a 5 to 10% growth rate defect on methanol (with no significant change in yield), but the Δfae2 Δfae3 mutant recovered to be indistinguishable in terms of growth rate (P = 0.644) and yield (P = 0.6889) from the WT (see Fig. S3 in the supplemental material). All mutants grew significantly slower than the WT on succinate, whereas only the double- and triple-knockout mutants grew significantly slower than the WT on succinate while using MA as a nitrogen source (see Fig. S3 in the supplemental material). No significant change in yield (relative to that of the WT) was observed for any mutant during growth on succinate (with either NH₄⁺ or MA as the nitrogen source). FAE and FAE homologs influence a combination of growth and yield on many different compounds, but the phenotypic effect of each mutant varies on the basis of the growth conditions tested.

DISCUSSION

In this study, we dissected the genetics of the NMG pathway mediating slow growth on MA in PA1. All three enzymes of the NMG pathway—NMGS, NMGDH, and GMAS—were essential for using MA as a carbon source, but only NMGDH and NMGS were absolutely essential for using MA as a nitrogen source. These results corroborate the findings of a recent study of the NMG pathway in DM4 (17). In DM4, though, it was suggested that a distant gmaS homolog (found in all sequenced M. extorquens species) partially compensates for a gmaS deletion and enables the ΔgmaS mutant to use MA as a nitrogen source, albeit at much lower rates than the WT. We observed that the slow, steadily decelerating growth of the ΔgmaS mutant (Fig. 2A) in succinate with MA as the nitrogen source was completely abolished in a ΔgmaS ΔmgsABC mutant. Our data support the notion that NMGS has a physiologically relevant, albeit low, activity for the conversion of MA to NMG that manifests only in the absence of GMAS. In the presence of GMAS, instead of MA, GMA might serve as the true substrate for NMGS. The precision of our automated growth measurement platform (41), along with a screen that coupled fast growth to a reduced demand upon the NMG pathway (using MA as a nitrogen source in the presence of an alternate carbon source), revealed that the NMG pathway has a linear topology in PA1 akin to that proposed for M. silvestris BL2 (3) and, very recently, DM4 (17). This genetic evidence for a linear topology for the NMG pathway (Fig. 2B) might also explain (i) the extremely low specific activity observed for the conversion of MA and glutamate to NMG by NMGS in vitro (24) and (ii) the accumulation of GMA in Δmgs mutants (15).

While it has been demonstrated in other systems (15) that the NMG pathway produces CH₂=H₄F, neither FolD nor PurU, enzymes known to work in the dissimilatory direction of the H₄F-dependent C₁ transfer pathway to generate formate, are encoded by the PA1 genome. We also observed that the H₄MPT-dependent C₁ dissimilatory module was essential but that deletion of a complete H₄F-dependent C₁ assimilation pathway had no effect upon growth on MA (Fig. 4). This result led us reevaluate if CH₂=H₄F really was the end product of the NMG pathway in PA1. The extreme formaldehyde sensitivity of a mutant incapable of synthesizing H₄MPT (the ΔmptG mutant), a cofactor involved in formaldehyde oxidation (26, 42), was leveraged to identify whether formaldehyde is generated and oxidized by the NMG pathway. The ΔmptG mutant could not grow on MA as a carbon source but could use MA as a nitrogen source (Fig. 3C and 4A). Additionally, a Δfae strain with a partial lesion in the H₄MPT-mediated formaldehyde oxidation pathway could still grow on MA. These data suggest that at least some formaldehyde is generated during growth on MA, but unlike growth on methanol (where the Δfae mutant cannot grow), it is likely that either (i) less than 100% of MA is oxidized to formaldehyde or (ii) the rate of formaldehyde production is low enough to be sustained by a spontaneous reaction with H₄MPT and H₄F. Additionally, the rate of formaldehyde production is likely to be even lower when MA is used as a nitrogen source, perhaps low enough that the flux can be handled by nonspecific aldehyde dehydrogenases. However, given the reactive nature of formaldehyde, it is almost impossible to resolve whether formaldehyde is generated as the direct end product (Fig. 5C) or by the dissociation of CH₂=H₄F (Fig. 5D). Alternately, we cannot rule out a third scenario where CH₂=H₄F is the end product of the NMG pathway and a methylene transfer occurs to generate CH₂=H₄MPT. In such a scenario, mptG but not fae would be essential for growth on MA, which is consistent with our data as well.

Typically, on the basis of whether formaldehyde or CH₂=H₄F is produced as the end product of primary oxidation, one of two distinct topologies is used for methylotrophic metabolism in M. extorquens species (Fig. 5A and B). The key difference between the two topologies is whether C₁ flux is channeled through an assimilatory or dissimilatory branch of the H₄F-dependent C₁ transfer pathway. In this study, we have uncovered a novel topology of C₁ use in PA1 that completely avoids the H₄F-dependent C₁ pathway for net dissimilation or assimilation (Fig. 5C and D).

Finally, this paper is the first to report functional roles for FAE homologs (FAE2 and FAE3) that are often found in the genomes of methylotrophs. FAE2 and FAE appear to play somewhat redundant roles during growth on MA, positively reinforcing growth rates but not yield (Fig. 6A). Three of the four key residues that make and stabilize contact with the methyl groups of H₄MPT in FAE (50) are conserved or similar in FAE2 (see Fig. S4 in the supplemental material). Therefore, FAE2 could be either an enzyme like FAE but with different kinetic parameters or perhaps a regulator that changes the expression of the pathway, sensing H₄MPT, CH₂=H₄MPT, or both. FAE3 appears to play a role distinct from that of FAE and FAE2 by impacting yield only during growth on MA (Fig. 6B) and thus likely has a different biochemical function. Specific residues that allow FAE to distinguish H₄MPT and H₄F (50) are not conserved in FAE3. Therefore, FAE3 is unlikely to be involved in the H₄MPT-dependent formaldehyde oxidation pathway and may work as either an enzyme or a regulator of the H₄F pathway. Further investigation will be required to uncover the precise physiological roles and underlying biochemistry of FAE2 and FAE3.

The extremely slow growth on MA (doubling time, ∼1 day) when MA oxidation is mediated by the NMG pathway had occluded genetic and physiological analysis, until recently. Of late, though, the NMG pathway is going through a phase of rediscovery, in light of its importance in several environmental regimes (7, 51, 52). While various studies with different model systems (15–17) have dissected the genomics, genetics, and physiology of the NMG pathway in isolation, we extended this genetic approach to understand how NMG-dependent MA metabolism links to central C₁-specific metabolic pathways in PA1. Unlike all previously documented instances of methylotrophy, we found that cells simultaneously require the H₄MPT-dependent formaldehyde oxidation pathway and the serine cycle but do not need to move C₁ units through an intact H₄F-dependent C₁ transfer pathway for either assimilation or dissimilation. Our work highlights that a single set of metabolic modules can be used in distinct configurations to route C₁ flux in different ways, depending upon the growth substrate in question.