The Entner-Doudoroff Pathway Is an Essential Metabolic Route for Methylotuvimicrobium buryatense 5GB1C

The gammaproteobacterial methanotrophs possess a unique central metabolic architecture where methane and other reduced C1 carbon sources are assimilated through the ribulose monophosphate cycle. Although efforts have been made to better understand methanotrophic metabolism in these bacteria via experimental and computational approaches, many questions remain unanswered.

ABSTRACT

Methylotuvimicrobium buryatense 5GB1C, a fast-growing gammaproteobacterial methanotroph, is equipped with two glycolytic pathways, the Entner-Doudoroff (ED) pathway and the Embden-Meyerhof-Parnas (EMP) pathway. Metabolic flux analysis and ¹³C-labeling experiments have shown the EMP pathway is the principal glycolytic route in M. buryatense 5GB1C, while the ED pathway appears to be metabolically and energetically insignificant. However, it has not been possible to obtain a null mutant in the edd-eda genes encoding the two unique enzymatic reactions in the ED pathway, suggesting the ED pathway may be essential for M. buryatense 5GB1C growth. In this study, the inducible P_BAD promoter was used to manipulate gene expression of edd-eda, and in addition, the expression of these two genes was separated from that of a downstream gltA gene. The resulting strain shows arabinose-dependent growth that correlates with ED pathway activity, with normal growth achieved in the presence of ∼0.1 g/liter arabinose. Flux balance analysis shows that M. buryatense 5GB1C with a strong ED pathway has a reduced energy budget, thereby limiting the mutant growth at a high concentration of arabinose. Collectively, our study demonstrates that the ED pathway is essential for M. buryatense 5GB1C. However, no known mechanism can directly explain the essentiality of the ED pathway, and thus, it may have a yet unknown regulatory role required for sustaining a healthy and functional metabolism in this bacterium.

IMPORTANCE The gammaproteobacterial methanotrophs possess a unique central metabolic architecture where methane and other reduced C₁ carbon sources are assimilated through the ribulose monophosphate cycle. Although efforts have been made to better understand methanotrophic metabolism in these bacteria via experimental and computational approaches, many questions remain unanswered. One of these is the essentiality of the ED pathway for M. buryatense 5GB1C, as current results appear contradictory. By creating a construct with edd-eda and gltA genes controlled by P_BAD and P_J23101, respectively, we demonstrated the essentiality of the ED pathway for this obligate methanotroph. It is also demonstrated that these genetic tools are applicable to M. buryatense 5GB1C and that expression of the target genes can be tightly controlled across an extensive range. Our study adds to the expanding knowledge of methanotrophic metabolism and practical approaches to genetic manipulation for obligate methanotrophs.

INTRODUCTION

Glycolysis is a vital metabolic process of energy and building block generation for bacterial metabolism. Two widely distributed glycolytic routes among bacteria are the Embden-Meyerhof-Parnas (EMP) pathway and the Entner-Doudoroff (ED) pathway, which are partially overlapped in their lower sections. The ED pathway has two unique reactions, 6-phosphogluconate (6PG) dehydratase (Edd, encoded by edd) and 2-keto-3-deoxy-6-phosphogluconate (KDPG) aldolase (Eda, encoded by eda) (1) (Fig. 1a). It serves as the principal or even exclusive glycolytic pathway for sugar catabolism in some bacteria, for instance, Pseudomonas putida and Zymomonas mobilis (2). In other bacteria, however, the ED pathway plays an insignificant role under tested conditions (3, 4), whereas the EMP pathway functions as the predominant catabolic route. It has been noted that the ED pathway has advantages over the EMP pathway. For instance, it requires fewer protein costs than the EMP pathway to carry the same amount of flux (5).

FIG 1.

The Entner-Doudoroff pathway in M. buryatense 5GB1C. (a) Schematic representation of simplified central metabolism of M. buryatense 5GB1C in which F6P is the metabolic branch point connecting the RuMP cycle, the ED pathway, and the EMP pathway. In the ED pathway, Edd catalyzes dehydration of 6PG to form KDPG, and Eda catalyzes cleavage of KDPG to form GAP and PYR. M. buryatense 5GB1C uses pyrophosphate (PPi)-dependent F6P 1-phosphotransferase to convert F6P to FBP instead of phosphofructokinase (6). In addition, one PPi corresponds to one-third ATP (6, 37). Therefore, 3.67 ATP and 2 NADH are generated via the EMP pathway, assuming one F6P molecule is completely oxidized to two pyruvate molecules. Through the ED pathway, 2 ATP, 1 NADH, and 1 NADPH are generated per F6P molecule. (b) Genomic locus organization of edd-eda-gltA in the WT strain. 3PG, 3-phosphoglycerate; 6PG, 6-phosphogluconate; edd, gene encoding 6-phosphogluconate dehydratase; eda, gene encoding 2-keto-3-deoxy-6-phosphogluconate aldolase; F6P, fructose-6-phosphate; FBP, fructose 1,6-bisphosphate; G6P, glucose-6-phosphate; GAP, glyceraldehyde 3-phosphate; KDPG, 2-keto-3-deoxy-phosphogluconate; PPi, pyrophosphate; PYR, pyruvate.

For the obligate methanotroph Methylotuvimicrobium buryatense 5GB1C, these two glycolytic pathways are connected to the ribulose monophosphate (RuMP) cycle responsible for the assimilation of one-carbon compounds during growth on methane or methanol (6). Their starting points are both located at fructose-6-phosphate (F6P) (Fig. 1a). F6P can be converted to glucose-6-phosphate (G6P) through G6P isomerase, which may enter the ED pathway, the oxidative pentose phosphate pathway (OPPP), or the glycogen synthesis pathway. F6P can also be converted to fructose 1,6-bisphosphate (FBP) through pyrophosphate fructose 6-phosphate phosphotransferase, entering the EMP pathway (7). Since no initial sugar phosphorylation is involved in either glycolytic pathway, more net ATP can be generated in M. buryatense 5GB1C than most sugar-consuming bacteria. In addition, when one F6P molecule is converted to two pyruvate molecules through either glycolytic pathway, the EMP pathway generates 1.67 more ATP molecules than the ED pathway in theory, and these two pathways generate the same amount of reducing equivalents per F6P molecule (Fig. 1a). Therefore, the EMP pathway is more energy efficient than the ED pathway.

Our previous studies have shown that the EMP pathway is the major glycolytic pathway in M. buryatense 5GB1C. This result has been confirmed by both genome-scale metabolic model (GEM) simulation and ¹³C-labeling experimental results (7–9). GEM predicts a minimal flux through the ED pathway and a negative correlation between the specific growth rate and the absolute flux of the ED pathway in M. buryatense 5GB1C. Furthermore, our labeling experiments have revealed that, although the 3-phosphoglycerate (3PG) pool size is 20 times larger than 6PG under substrate limitation, 6PG displays a slightly slower ¹³C enrichment rate. 3PG and 6PG are key intermediates in the EMP and ED pathways, respectively (Fig. 1a), and thus, their labeling patterns can be used for qualitative comparison of fluxes through the two glycolytic pathways. If both pathways have the same amount of carbon flow, we would expect much faster ¹³C incorporation for 6PG due to its smaller metabolic pool. However, the faster ¹³C enrichment in 3PG provides strong evidence that the flux through the EMP pathway is higher than the ED pathway. This was further strengthened by subsequent flux quantification of M. buryatense 5GB1C via ¹³C isotopically nonstationary metabolic flux analysis (INST-MFA). In addition, recent studies on other gammaproteobacterial methanotrophs have implied the same metabolic dominance of the EMP pathway in, for instance, Methylotuvimicrobium alcaliphilum 20Z and Methylomonas sp. DH-1 (10–12). Correspondingly, the ED pathway seems metabolically and bioenergetically insignificant for those gammaproteobacterial methanotrophs.

To determine the essentiality of the ED pathway, a growth phenotype analysis of a knockout mutant of edd-eda would usually suffice. However, a null mutation in either edd or eda or both has never been achieved successfully in our laboratory, even though multiple methods that have been successful for this methanotroph in the past (13, 14) were utilized (Y. Fu, J. D. Groom, and M. E. Lidstrom, unpublished data), which suggests the ED pathway may be required for growth on methane. In the genome of M. buryatense 5GB1C, gltA, encoding citrate synthase, which carries out the condensation reaction of oxaloacetate (OAA) and acetyl-CoA into citrate, is located ∼240 bp downstream of edd-eda, transcribed in the same direction (Fig. 1b). As the tricarboxylic acid (TCA) cycle is essential for M. buryatense 5GB1C, citrate synthase should be essential. We therefore speculate that the gene expression of gltA might have been unexpectedly disturbed during the process of deletion of edd and/or eda, resulting in no viable mutants.

This study is focused on the essentiality of the ED pathway in M. buryatense 5GB1C. In order to decrease expression of the ED pathway without interference with gltA, we substituted the native promoter of edd-eda with the araC-P_BAD regulatory system. This construct permits tightly controlled gene expression of the ED pathway via altering the arabinose concentration in medium without altering expression of downstream gltA. Via growth tests and enzyme activity assays, we demonstrated that the ED pathway is essential for M. buryatense 5GB1C metabolism.

RESULTS

The araC-P_BAD regulatory system is functional in M. buryatense 5GB1C.

Since it was not possible to obtain null mutants in the ED pathway genes, an alternate approach involving an inducible system was used. The araC-P_BAD regulatory system has been commonly utilized as a genetic tool to control the expression of target genes (15, 16) and has been shown to function as a promoter in a different methanotroph (17). Therefore, we first tested whether this tool could regulate gene expression in M. buryatense 5GB1C. To this end, we created a strain, JG4, harboring an araC gene encoding the transcription factor that regulates the P_BAD promoter in response to arabinose, and a xylE reporter gene encoding catechol 2,3-dioxygenase (Fig. 2a) controlled by P_BAD. We observed a linear correlation between the xylE gene expression and arabinose concentration between 0 and 0.5 g/liter (Fig. 2b). Beyond that range, the gene expression plateaued at a level comparable to that of xylE under the control of the native particulate methane monooxygenase (pMMO) promoter. During growth on methane, transcripts of pmoCAB encoding the pMMO are the most abundant in M. buryatense 5GB1C (18). Hence, the araC-P_BAD regulatory system is not only functional in this methanotroph, but it also allows manipulation of gene expression across an extensive range from the background level to the pMMO gene expression level. In addition, these results confirmed that M. buryatense 5GB1C can take up sufficient extracellular arabinose to activate AraC.

FIG 2.

Expression of xylE under the control of the P_BAD promoter. (a) Map of the insertion construct used to test titration of gene expression by arabinose. araC, arabinose-responsive regulator; P_C, promoter of araC; P_BAD, promoter of araBAD operon; xylE, catechol 2,3-dioxygenase. (b) XylE enzyme activity as a function of arabinose concentration. Triangles represent data from the strain JG4 bearing P_BAD-xylE. The square represents the XylE enzyme activity under the control of the native pMMO promoter, which is used as the positive control. The diamond shows no XylE enzyme activity in the WT strain. Error bars represent standard deviations based on three biological replicates.

Attempts were made to grow M. buryatense 5GB1C on arabinose, but no growth was obtained. In addition, we also examined whether the addition of arabinose into NMS2 medium would affect the growth of wild-type M. buryatense 5GB1C, and the results indicate that an arabinose concentration between 0 and 0.5 g/liter has either no or a minor effect on bacterial growth (Fig. S1 in the supplemental material). On the other hand, our data do not rule out a possibility that a minimal amount of arabinose could be incorporated into biomass; however, it is unlikely arabinose can be oxidized to downstream building blocks to promote significant bacterial growth.

The P_BAD promoter influences the gene expression of gltA.

To control the ED pathway activity and avoid any possible disruption of downstream gltA gene expression (Fig. 1b), we replaced the native ED pathway promoter in the wild type (WT) with the araC-P_BAD regulatory system, resulting in strain LH1 (Fig. 3a and Table 1). We then tested growth of the strain LH1 in the absence and presence of arabinose. With addition of 0.1 g/liter arabinose, the specific growth rate can reach 0.23  ±  0.01 h⁻¹ (Fig. 3b), which is the same as the WT strain (13, 19). In the absence of arabinose, however, strain LH1 exhibited no significant growth after its cell density (optical density at 600 nm [OD₆₀₀]) reached 0.1 (Fig. 3b). The initial OD₆₀₀ increase was likely due to either intracellular arabinose storage in seed cultures or arabinose carryover from the inoculum.

FIG 3.

The P_BAD promoter influences the expression of gltA. (a) Genomic locus organization of edd, eda, and gltA in the LH1 strain. The araC-P_BAD was placed before edd in the LH1 strain. purE encodes N⁵-carboxyaminoimidazole ribonucleotide mutase. (b) Growth of the strain LH1 with or without 0.1 g/liter arabinose supplement in medium. (c) ED pathway activity of the WT and LH1 strains. (d) Citrate synthase enzyme activity of the WT and LH1 strains. Varied arabinose concentrations were supplied to the LH1 strain. Error bars represent standard deviations based on three biological replicates. Two-sample t tests were performed to determine significance (***, P < 0.001; **, P < 0.01; *, P < 0.05) between enzyme activities.

TABLE 1.

Strains, linear constructs, and plasmids used in this study

Strain, construct, or plasmid	Description	Source or reference no.
M. buryatense strain
5GB1C	Wild type cured of the native plasmid
JG4	5GB1C araC-PBAD-xylE	This study
FC43	5GB1C PpmoC-xylE	14
LH1	5GB1C kanr-araC-PBAD-edd	This study
LH2	5GB1C eda-T-zeor-PrpoD-gltA	This study
LH3	5GB1C eda-T-zeor-PJ23101-gltA	This study
LH4	LH1 eda-T-zeor-PrpoD-gltA	This study
LH5	LH1 eda-T-zeor-PJ23101-gltA	This study
Linear construct
cLH1	Kanr-araC-ParaC-edd	This study
cLH2	eda-T-zeor-PrpoD-gltA	This study
cLH3	eda-T-zeor-PJ23101-gltA	This study
Plasmid
pSMF7	Replicating vector harboring PJ23101-xylE	This study
pLH1	pAWP92 containing first 828 bp of gltA downstream of PrpoD; Kanr	This study
pLH2	pSMF7 containing first 828 bp of gltA downstream of PJ23101; Kanr	This study

To confirm that the gene expression of the ED pathway is tightly controlled by the P_BAD promoter, we performed enzyme activity assays on the overall ED pathway activity and the citrate synthase activity from cell extracts of M. buryatense 5GB1C harvested during the exponential growth phase. As expected, the overall ED pathway activity is correlated with arabinose concentrations (Fig. 3c). Unexpectedly, however, the same trend is observed for the citrate synthase activity (Fig. 3d), indicating that the P_BAD promoter, and possibly the native edd-eda promoter, can at least partly regulate the gene expression of gltA. Further analysis is required to investigate if edd, eda, and gltA are in the same operon. Consequently, by altering the arabinose concentrations, we manipulated two metabolic pathways simultaneously, and the growth arrest of LH1 in the absence of arabinose could be caused by perturbation of the intracellular citrate synthase activity.

Two gltA genes (MBURv2_130010 and MBURv2_130027) are present in the genome. Based on our previous RNA-seq data characterizing the transcription profiles of M. buryatense 5GB1C under different growth conditions (e.g., CH₄-limited growth conditions, O₂-limited growth conditions, slow and fast growth conditions, etc.), both gltA genes are expressed (18, 20). Generally, gltA MBURv2_130010, which is downstream of edd-eda, exhibits 2- to 3-fold higher expression than the homolog, suggesting it may be the major enzyme to carry out the citrate synthase reaction (Fig. S2). Although the measured citrate synthase activities would not distinguish between the two citrate synthase enzymes, the results unambiguously demonstrate that the overall citrate synthase activity was altered with the same trend as the ED pathway at varied arabinose concentrations.

The ED pathway is likely essential for M. buryatense 5GB1C.

To address the above issue, we replaced the intergenic spacer between eda and gltA in strain LH1 with the following construct: the Escherichia coli rrnB T1 terminator followed by the zeocin-resistant cassette and the constitutive promoter P_J23101 driving the expression of gltA. This genetic manipulation yielded strain LH5 (Fig. 4a). As a reference, we used the same design to decouple gltA and eda expression in the WT strain, which yielded strain LH3 (Fig. 4a). We have also tried using a weaker promoter, the native rpoD promoter, to control gltA; however, the resulting mutant (strains LH2 and LH4; Table 1) grew very poorly in liquid medium (data not shown). Therefore, we used only the strains LH3 and LH5 for the following analyses.

FIG 4.

The ED pathway is essential for M. buryatense 5GB1C metabolism. (a) Genomic locus organization of edd, eda, and gltA in the LH3 and LH5 strains. In the strain LH3, the intergenic spacer was substituted with T-zeo^r-P_J23101. The strain LH5 has both araC-P_BAD and T-zeo^r-P_J23101 located before edd and gltA, respectively. purE encodes N⁵-carboxyaminoimidazole ribonucleotide mutase. (b) Growth rates of the WT, LH1, LH3, and LH5 strains in response to varied arabinose conditions. The specific growth rates of the WT and LH3 strains are independent of arabinose availability. (c) ED pathway activity of WT, LH3, and LH5 strains. (d) Citrate synthase enzyme activity of WT, LH3, and LH5 strains. Varied arabinose concentrations were supplied to the LH5 strain. Error bars represent standard deviations based on three biological replicates. Two-sample t tests were performed to determine significance (***, P < 0.001; **, P < 0.01; *, P < 0.05) between enzyme activities.

We first tested the growth of WT, LH1, LH3, and LH5 in response to different arabinose concentrations (Fig. 4b). Kanamycin (50 μg/ml) was used for LH1 and zeocin (30 μg/ml) for LH3 and LH5. The growth rate of both WT and LH3, which are devoid of the araC-P_BAD regulatory system, remained the same at all tested arabinose levels (0 to 0.5 g/liter; Fig. 4b and Fig. S1). This result indicates an arabinose concentration between 0 and 0.5 g/liter neither promotes nor inhibits growth of the WT and LH3 strains. Both strains LH1 and LH5 grew optimally with a rate of 0.20 to 0.23 h⁻¹ with addition of ∼0.1 g/liter arabinose. Increasing or decreasing the optimal concentration leads to a reduction in growth rate. LH5 appears to be more sensitive to arabinose concentration, and a small variation from ∼0.1 g/liter triggers a large reduction in growth rate. Higher arabinose concentrations caused a decreased growth rate. No growth was observed for either LH1 or LH5 in the absence of arabinose (Fig. S3 and S4). The minimal requirement of arabinose supporting a significant growth of LH1 and LH5 is 0.03 g/liter and 0.07 g/liter, respectively (Fig. S3 and S4), suggesting that a gene expression threshold for the ED pathway must be met to maintain significant growth.

Enzyme activity assay of WT, LH3, and LH5 cell extracts further confirms that our genetic manipulations successfully decoupled the expression of the ED pathway and gltA and that only the ED pathway in strain LH5 was controlled by P_BAD (Fig. 4c and d). Collectively, the above-described assays provide strong evidence that the ED pathway of M. buryatense 5GB1C is essential.

Flux balance analysis reveals that high activities of the ED pathway result in reduced growth of M. buryatense 5GB1C.

To understand the metabolic flux phenotypes of M. buryatense 5GB1C carrying various flux strengths through the ED pathway, we performed flux balance analysis (FBA) simulations. Here, we introduced a flux ratio of the EMP pathway to the ED pathway as a constraint (7) so that the latter’s flux strength can be manipulated. When this ratio is sufficiently high (>10) such that the EMP pathway is the principal glycolytic pathway, the simulation predicts little variation in the overall flux phenotype and the growth rate (Fig. 5a and Table S2). In contrast, the growth rate is predicted to plummet as the flux through the ED pathway becomes strong (Fig. 5a). Analysis of the energy metabolism of M. buryatense 5GB1C reveals that the ATP budget is reduced by over 30% when the EMP/ED flux ratio drops from 100 to 0.15 (Fig. 5b), which suggests the growth defect of the strain LH5 at high expression of the ED pathway can be explained by an energy bottleneck resulting from the dominance of the ED pathway. However, when the ED pathway was “knocked out” in silico, the simulation yielded the same metabolic flux phenotype as the wild-type strain, which contradicted our experimental results. This is not surprising since the remaining EMP pathway can fulfill all the metabolic responsibilities as the ED pathway in silico without emergence of dead ends or other errors.

FIG 5.

Simulated growth rates (a) and ATP budgets (b) of M. buryatense 5GB1C at different EMP/ED flux ratios. The ratio is defined as the flux through the EMP pathway over the flux through the ED pathway. All the other constraints remain the same. The ATP budget is defined as follows: ATP budget = ATP production + 3×NADH production + 3×NADPH production − nongrowth associated ATP maintenance energy in millimole per gram per hour. The simulation yielded infeasible results when the ratio is below 0.13, and thus, those results were excluded.

DISCUSSION

In this study, we addressed a key question arising from our recent studies on metabolism of M. buryatense 5GB1C: is the ED pathway essential for growth? Substantial evidence has shown that the EMP pathway, rather than the ED pathway, plays the predominant role in M. buryatense 5GB1C, including studies of ¹³C INST-MFA, GEM simulation, and transcriptomics data (7–10, 12). In addition, intracellular fructose bisphosphate (FBP) in M. buryatense 5GB1C is abundant (8), which can activate pyruvate kinase activity and drive more carbon flow through the lower part of the EMP pathway (21, 22). Moreover, the ED pathway is bioenergetically dispensable. The ATP budget would be reduced if the ED pathway plays the major role in glycolysis since the EMP pathway is one of the key sources for ATP and NADH generation (8). Also, maintenance of high flux through the ED pathway or the OPPP would be unnecessary or might even cause redox imbalance since ample NAD(P)H can be generated through the methane oxidation and the lower EMP pathways (Fig. 1a) (8, 9). Our model simulation has confirmed a negative relationship between the ED pathway flux and the specific growth rate (7) (Fig. 5a).

Our results suggest that despite all evidence of minor flux through the ED pathway, it is essential to M. buryatense 5GB1C. Using tunable gene expression, we have shown that a functional ED pathway is required for growth at a threshold level, even under conditions in which expression of the downstream gltA is sufficient for normal growth, ruling out a possible artifact due to altered gltA expression. We also showed that higher expression of these enzymes resulted in decreased growth. Our application of araC-P_BAD in M. buryatense 5GB1C demonstrates that it can be readily utilized for tightly controlling gene expression of interest or investigating phenotypic response at varied transcription levels. Interestingly, our data also suggest that edd, eda, and gltA may be controlled by the same promoter.

Several reasons for the essentiality of the ED pathway in M. buryatense 5GB1C are possible. In the first place, the ED pathway function may be required for NADPH generation to defend against oxidative stress (23). However, although the ED pathway can contribute to NADPH production, its flux strength in M. buryatense 5GB1C is too low to support either amino acid or biomass synthesis (8). Further, the major route for reducing power generation is the substrate oxidation pathway in M. buryatense 5GB1C (8), which also invalidates the importance of the OPPP for its growth. Thus, this hypothesis unlikely holds true. Second, accumulation of intracellular KDPG has been reported to cause growth arrest in E. coli (24), and thus, a complete and functional ED pathway may be required for M. buryatense 5GB1C to avoid accumulation of KDPG at a toxic level. However, KDPG is present at low concentrations in M. buryatense 5GB1C grown on methane, often not detectable (7, 8), so this hypothesis seems unlikely.

In addition, both Edd and Eda exhibit catalytic promiscuity, and thus, they could be involved in other metabolic pathways crucial to M. buryatense 5GB1C. Some examples of such roles for the ED pathway enzymes are known. It has been shown in other bacteria that the Edd protein sequence is homologous to the native dihydroxy-acid dehydratase, an enzyme involved in the valine/isoleucine biosynthesis pathway (1, 25, 26). Eda can function as 2-keto-4-hydroxyglutarate aldolase, carrying out cleavage of 2-keto-4-hydroxyglutarate (KHG) to form pyruvate and glyoxylate or the reverse condensation reaction and can play an important role in metabolism in this way (1). This hypothesis is also unlikely since a dedicated dihydroxy-acid dehydratase (encoded by ilvD; MBURv2_60025) is present in the genome with significant constitutive gene expression (18). Another role of the KHG aldolase side reaction has been proposed in other bacteria in which this activity and several TCA cycle enzymes can form a cyclic pathway oxidizing glyoxylate into CO₂, thereby regulating the intracellular glyoxylate level (27). In M. buryatense 5GB1C, glyoxylate is generated via the partial serine cycle, and it is then converted to glycine. As the flux through the serine cycle is negligible under methane growth conditions (8, 28), the KHG aldolase activity is unlikely essential for M. buryatense 5GB1C. Another possible inhibition cascade may result from the adventitious formation of 4-erythrose phosphate (4EP) as another side reaction (29, 30). A high level of 4EP can cause inhibition of 6PG dehydrogenase and, subsequently, accumulation of 6PG, resulting in inhibition of G6P isomerase. Eventually, both the EMP pathway and the OPPP are turned off, which can be rescued by native phosphatases capable of 4EP detoxification. Under such a circumstance, the ED pathway can prevent 6PG accumulation and provide an alternative route for catabolizing sugar carbons. However, it is known that in M. buryatense 5GB1C, 6PG is of very low abundance in multiple growth conditions, on the order of magnitude of ∼10⁻⁴mmol/gDW (grams dry weight) (8). Since 6PG is not known to accumulate, it is unlikely that this route inhibits G6P isomerase or triggers a shutdown of methanotrophic metabolism.

Therefore, none of the known reasons why the ED pathway would be essential appear to apply to M. buryatense 5GB1C. It is possible that the ED pathway may have a yet unknown regulatory role sustaining a healthy and functional metabolism in this bacterium. This is consistent with our FBA simulation results. Since the FBA model does not include regulation or gene-mRNA-protein interplays, phenotypes resulting from those mechanisms are not predicted by FBA.

The FBA simulation agrees with the experimental results showing decreased growth at high ED pathway expression in that a reduced ATP budget caused by a strong ED pathway is predicted to lead to growth defects. Other possible reasons for such growth defects include reduced methane uptake rate in the mutants and overexpression burden of Edd and Eda, which competes for limited resources in cells.

In conclusion, in this study, we aimed to determine the essentiality of the ED pathway for M. buryatense 5GB1C. To this end, we constructed a mutant with edd-eda driven by the inducible P_BAD promoter and gltA driven by the constitutive P_J23101 promoter so that we can manipulate gene expression of the ED pathway without interference of citrate synthase activity. Our mutant construction was confirmed by enzyme activity assays. The mutant cannot grow without any arabinose supplement in medium, confirming that the ED pathway is essential for M. buryatense 5GB1C. In particular, a minimum gene expression of the ED pathway is required to support significant growth of M. buryatense 5GB1C. Since no known mechanism can explain the essentiality of the ED pathway, we hypothesized that the ED pathway may have a yet unknown regulatory role sustaining a healthy and functional metabolism of M. buryatense 5GB1C.

MATERIALS AND METHODS

Bacterial cultivation and growth conditions.

M. buryatense 5GB1C WT and mutants were grown in NMS2 medium (19) at 30°C. Single colonies were first inoculated into 5 ml liquid medium and grown with CH₄ as seed cultures, which were then used for subsequent tests. For strains LH1 and LH5, 0.1 g/liter arabinose was provided in the seed cultures. To minimize the influence of the arabinose carryover, seed cultures were diluted at least 100 times before being transferred to fresh media supplied with different levels of arabinose. To determine the specific growth rate, 5-ml cell cultures were grown in 25-ml glass tubes (18 mm diameter by 150 mm height) supplied with 5 ml (∼10 mM) CH₄ in the headspace. OD₆₀₀ was monitored by a spectrophotometer (Spectronic 20D+; Thermo Electron Corporation) at various time points. For enzyme activity measurements, 50-ml cell cultures were grown in 250-ml glass vials supplied with 50 ml CH₄, and they were harvested during the exponential growth phase. All cultures were shaken at 200 rpm. Either kanamycin (50 μg/liter) or zeocin (30 μg/liter) was provided for corresponding antibiotic-resistant mutants in medium. Only zeocin was provided for the strain LH5 bearing both kanamycin (Kan^r) and zeocin (Zeo^r) resistance. Two hundred grams per liter arabinose stock solution was filter sterilized before being diluted to desired concentrations.

Construction of M. buryatense 5GB1C mutants.

The iProof high-fidelity PCR kit (Bio-Rad, CA, USA) was used for PCR throughout this study. Linear construct cLH1 (Table 1) was created by overlap PCR via linking ∼800 bp of 5′ and 3′ flanking regions of the deletion target (i.e., ∼500 bp 5′ flanking region of edd [MBURv2_130008]), the kanamycin-resistant cassette, and the araC-P_BAD regulatory region (Fig. 3a). The overlap PCR cycles were described previously (31). In a similar fashion, cLH2 and cLH3 were constructed by linking ∼800 bp of the 5′ and 3′ flanking regions of the deletion target (i.e., the intergenic spacer between eda [MBURv2_130009] and gltA [MBURv2_130010]), the E. coli rrnB T1 terminator, the zeocin-resistant cassette, and one promoter, either the native rpoD promoter P_rpoD or the synthetic promoter P_J23101 (Fig. 4a). Since the promoter regions were short (≤300 bp), they were first linked with the 3′ flanking region; the 3′ flanking region was cloned downstream of P_rpoD and P_J23101 (Registry of Standard Biological Parts; http://parts.igem.org/), respectively, in plasmids pAWP92 (19) and pSMF7 via Gibson assembly, yielding plasmids pLH1 and pLH2, respectively. The plasmids were electroporated into E. coli top 10 competent cells for amplification. The PCR products of the promoter-flanking regions were fused with other fragments (i.e., the 5′ flanking region, the terminator, and the zeocin-resistant cassette) to generate cLH2 and cLH3 via overlap PCR.

The above-described linear constructs were integrated into the genome of corresponding M. buryatense 5GB1C strains through electroporation as previously described (14, 31). Mutants were screened with corresponding antibiotic sensitivity and then verified by Sanger sequencing (Genewiz, NJ, USA). For strains LH1, LH4, and LH5, 0.1 g/liter arabinose was provided in the recovery medium and subsequent antibiotic selection agar plates to ensure gene expression of the ED pathway. All strains, linear constructs, and plasmids used in this study are listed in Table 1, and all primers are listed in Table S1 in the supplemental material.

Enzyme-specific activity assay.

WT or mutant cultures were grown to an OD₆₀₀ between 0.3 and 0.5. They were then harvested by centrifugation at 3,000 × g at 4°C for 30 min. The supernatant was discarded, and cell pellets were rinsed with 1 ml cell extract buffer twice, which contained 100 mM Tris-HCl (pH 7.5), 20 mM KCl, 5 mM MnSO₄, 2 mM dithiothreitol (DTT), and 0.1 mM ethylenediaminetetraacetic acid (EDTA). The resulting cell suspensions were lysed with French Press (SLM-Aminco) at 1,000 lb/in² or 7 × 10⁶ Pa. Cell lysis was centrifuged at 10,000 × g at 4°C, and only cell extract was kept. Throughout all the following analyses, cell extract solutions were kept on ice.

Activity of the overall ED pathway was determined based on published protocols (32, 33). We added 30 μl cell extract into 300 μl buffer solution containing 200 mM Tris-HCl (pH 7.5), 5 mM 6PG, and 10 mM MgSO₄ in a 1-ml plastic cuvette. Three solutions containing the same cell extract samples were incubated at 30°C for 30, 50, and 70 min, respectively, before addition of 300 μl 0.02% (wt/vol) 2,4-dinitrophenylhydrazine solution containing 500 mM HCl. The mixtures were incubated at room temperature for another 10 min. We added 500 μl of 2 M NaOH solution to stop the reaction, and absorbance was measured spectrophotometrically at 450 nm. The standard curve was determined by using known concentrations of pyruvate solutions in similar procedures as described above. The specific activity was calculated based on the slope of measured pyruvate concentrations against time, normalized by the soluble protein added for each cell extract sample. Assays run with extracts showing no detectable activity were used as negative controls.

Measurement of the citrate synthase specific activity was modified from a published protocol (34) and carried out using microplates. The following solutions were added in sequence: 50 μl 10 mM 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) solution containing 100 mM Tris-HCl (pH 7.5) and 50 μl AcCoA (2 mM), OAA (10 mM) solution containing 100 mM Tris-HCl (pH 7.5), and, lastly, 20 μl cell extract. The kinetic absorbance was measured continuously at 412 nm at 25°C for at least 10 min by a plate reader (SpectraMax 190, Molecular Devices). The citrate synthase specific activity was determined by the increasing rate of A412 divided by the protein amount (unit, A412/min/mg soluble protein). Negative controls with no extract showed no detectable activity.

The xylE reporter gene expression assay was carried out as previously described (13, 31). Briefly, cells were grown to an OD₆₀₀ of 0.4 to 0.6, harvested at 4,000 × g for 10 min, and normalized to OD₆₀₀ of 0.5 with 50 mM Tris-HCl, pH 7.5. Ninety microliters of the cell suspensions were mixed with a final concentration of 1 mM catechol. The kinetic absorbance was measured continuously at 375 nm at 30°C for 10 min by a plate reader (SpectraMax 190, Molecular Devices).

The soluble protein concentrations were quantified using Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, USA) as instructed by the user manual.

Flux balance analysis.

The genome-scale flux balance analysis model for M. buryatense 5GB1C was constructed previously (7). In the model, three possible methane oxidation modes were included: redox arm, direct coupling, and uphill transfer modes. The formate dehydrogenase flux was constrained within 4.39 and 7.77 mmol/g/h, which was based on previous FBA simulation results (9). The methane uptake rate was 18.46 mmol/g/h, and the formate production rate was 0.123 mmol/g/h (35). The non-growth-associated ATP maintenance flux was constrained within a range from 8.39 to 21.6 mmol/g/h (9). The simulation was performed using COBRApy and GLPK solver (36).

Data availability.

All the data are presented in the main text or supplementary files.