The Ethylmalonyl-CoA Pathway Is Used in Place of the Glyoxylate Cycle by Methylobacterium extorquens AM1 during Growth on Acetate*

Background: The ethylmalonyl-CoA pathway represents an alternative to the glyoxylate cycle. Results: The ethylmalonyl-CoA pathway operates during acetate growth of Methylobacterium extorquens and represents one of three entry points into central metabolism. Conclusion: Isocitrate lyase-positive and -negative bacteria differ substantially in overall metabolic flux distribution. Significance: Tight coordination must exist for operation of the citric acid cycle in conjunction with the ethylmalonyl-CoA pathway. Acetyl-CoA assimilation was extensively studied in organisms harboring the glyoxylate cycle. In this study, we analyzed the metabolism of the facultative methylotroph Methylobacterium extorquens AM1, which lacks isocitrate lyase, the key enzyme in the glyoxylate cycle, during growth on acetate. MS/MS-based proteomic analysis revealed that the protein repertoire of M. extorquens AM1 grown on acetate is similar to that of cells grown on methanol and includes enzymes of the ethylmalonyl-CoA (EMC) pathway that were recently shown to operate during growth on methanol. Dynamic 13C labeling experiments indicate the presence of distinct entry points for acetate: the EMC pathway and the TCA cycle. 13C steady-state metabolic flux analysis showed that oxidation of acetyl-CoA occurs predominantly via the TCA cycle and that assimilation occurs via the EMC pathway. Furthermore, acetyl-CoA condenses with the EMC pathway product glyoxylate, resulting in malate formation. The latter, also formed by the TCA cycle, is converted to phosphoglycerate by a reaction sequence that is reversed with respect to the serine cycle. Thus, the results obtained in this study reveal the utilization of common pathways during the growth of M. extorquens AM1 on C1 and C2 compounds, but with a major redirection of flux within the central metabolism. Furthermore, our results indicate that the metabolic flux distribution is highly complex in this model methylotroph during growth on acetate and is fundamentally different from organisms using the glyoxylate cycle.

Acetyl-CoA assimilation was extensively studied in organisms harboring the glyoxylate cycle. In this study, we analyzed the metabolism of the facultative methylotroph Methylobacterium extorquens AM1, which lacks isocitrate lyase, the key enzyme in the glyoxylate cycle, during growth on acetate. MS/MS-based proteomic analysis revealed that the protein repertoire of M. extorquens AM1 grown on acetate is similar to that of cells grown on methanol and includes enzymes of the ethylmalonyl-CoA (EMC) pathway that were recently shown to operate during growth on methanol. Dynamic 13 C labeling experiments indicate the presence of distinct entry points for acetate: the EMC pathway and the TCA cycle. 13 C steady-state metabolic flux analysis showed that oxidation of acetyl-CoA occurs predominantly via the TCA cycle and that assimilation occurs via the EMC pathway. Furthermore, acetyl-CoA condenses with the EMC pathway product glyoxylate, resulting in malate formation. The latter, also formed by the TCA cycle, is converted to phosphoglycerate by a reaction sequence that is reversed with respect to the serine cycle. Thus, the results obtained in this study reveal the utilization of common pathways during the growth of M. extorquens AM1 on C1 and C2 compounds, but with a major redirection of flux within the central metabolism. Furthermore, our results indicate that the metabolic flux distribution is highly complex in this model methylotroph during growth on acetate and is fundamentally different from organisms using the glyoxylate cycle.
The ability to oxidize two-carbon compounds (C2) 2 to carbon dioxide for energy generation and assimilation of these compounds into cellular material are physiological features shared by various microorganisms. The question of how carbon assimilation is accomplished is not only important with respect to growth on C2 substrates such as acetate, but also for numerous substrates that share a common initial conversion to acetyl-CoA as the entry point into central carbon metabolism. These include abundant substrates such as fatty acids, alcohols, esters, terpenes, waxes, alkenes, and polyhydroxyalkanoates.
Oxidation of acetyl-CoA under aerobic conditions generally occurs via the TCA cycle and does not allow carbon assimilation because of decarboxylation reactions. To circumvent the steps that interfere with the net synthesis of C4 compounds from acetyl-CoA, many organisms use the glyoxylate cycle that was identified more than 50 years ago and that operates in conjunction with the TCA cycle (1,2). Instead of being decarboxylated, the TCA cycle intermediate isocitrate is cleaved by isocitrate lyase to succinate and glyoxylate. Malate synthase, the second enzyme of the glyoxylate cycle, condenses glyoxylate and another molecule of acetyl-CoA to malate, a C4 compound.
Isocitrate lyase-negative organisms that grow on acetate but lack an operating glyoxylate cycle, including Methylobacterium extorquens (3), Rhodopseudomonas (4), Streptomycetes (5), and Paracoccus (6), must use another strategy for anaplerosis of C2 compounds. This alternative pathway has long remained a mystery (7); using biochemical studies together with mutant analysis, Erb, Alber and co-workers (8 -12) discovered the EMC pathway in Rhodobacter sphaeroides. This pathway generates one glyoxylate and one succinyl-CoA molecule from two acetyl-CoA and two carbon dioxide molecules. Condensation of glyoxylate with acetyl-CoA produces malate. Recently, the operation of the EMC pathway during methylotrophic growth and its connection to the serine cycle in the Alphaproteobacterium M. extorquens AM1 was demonstrated by dynamic and steady-state 13 C labeling experiments (13). Under methylotrophic conditions, the EMC pathway is required to generate glyoxylate from acetyl-CoA, the carbon fixation product of the serine cycle, to restore the acceptor molecule glyoxylate. Furthermore, it was shown that the TCA cycle is incomplete during growth on methanol because no flux through 2-oxoglutarate dehydrogenase was observed (14).
The metabolic network topology of the central carbon metabolism involved in substrate oxidation, and production of precursor metabolites during growth on acetate is well understood in isocitrate-positive organisms such as Escherichia coli (15)(16)(17)(18)(19) and Corynebacterium glutamicum (20,21) but has never been studied in isocitrate lyase-negative organisms. This study is the first to provide a detailed examination of the central metabolism of an isocitrate lyase-negative organism during growth on acetate. M. extorquens AM1 is a well studied and biotechnologically relevant model organism with respect to its methylotrophic lifestyle (22). Here, we address three different aspects of C2 metabolism in M. extorquens. First, we addressed the question of how redox equivalents in the facultative methylotroph are generated during heterotrophic growth on acetate. The second goal was to identify the pathway(s) involved in acetyl-CoA assimilation and, in particular, to understand the function of the EMC pathway and its potential connection to the TCA cycle. The third aim of this study was to understand the role of the serine cycle during growth on acetate because a number of intermediates, such as serine, glycine, phosphoglycerate, and phosphoenolpyruvate (PEP), are important branching points for biomass production. In addition to methylene (m)-tetrahydrofolate (H 4 F), these precursor metabolites must be synthesized from acetate. To investigate the operation of the metabolic network, with a focus on the role and function of the TCA cycle, the EMC pathway, and the serine cycle during growth of M. extorquens AM1 on acetate, we performed comparative proteome analysis (acetate versus methanol) and dynamic and steady-state 13 C labeling experiments.

EXPERIMENTAL PROCEDURES
Chemicals-[ 13 C]Sodium acetate (99%) was purchased from Cambridge Isotope Laboratories, and D 2 O (99.8% and 99.97%) was purchased from Eurisotop. All other chemicals were purchased from Sigma.
Medium Composition and Cultivation Conditions-M. extorquens AM1 was grown on minimal medium (23) supplemented with 5 mM sodium acetate. This substrate concentration was chosen because growth was inhibited by higher acetate concentrations. A preparatory culture was grown on minimal medium plates (1.5% agar); batch cultures were inoculated with cells washed from plates with minimal medium. All of the cultures were grown in a 500-ml bioreactor (Infors-HT) with a working volume of 400 ml at 28°C, with an aeration rate of 0.2 liters/min and stirring at 1000 rpm. The pH was kept constant at 7.0 by the addition of acetic acid (0.5 M), which allowed for substrate feeding and maintenance of the substrate concentration. The cells were harvested by centrifugation at 6000 ϫ g at room temperature and frozen in liquid nitrogen until analysis.
Quantification of Acetate-Quantification of acetate was performed with a Waters Alliance 2690 HPLC and a UV-VIS detector (DAD) at 210 nm. Cell cultures were harvested with a 2-ml plastic syringe and filtered through an RC syringe filter (0.2 m); tartaric acid was added to the cell-free supernatant as an internal standard. Organic acids were separated using a Phenomenex Rezex ROA organic acid Hϩ (8%, 300 ϫ 7.8 mm) column with a flow rate of 0.5 ml/min and detected at a wavelength of 210 nm. An isocratic method was employed as described by the supplier, using 2.5 mM H 2 SO 4 as the solvent. Acetate was quantified with a 0.3 mM standard deviation based on technical replicates.
Biomass Composition-The analysis of biomass composition was conducted from three biological replicates as described previously (14). Biomass yield (g of [C]/cell dry weight (CDW)/g of [C] per substrate) was calculated from the biomass production rate and the substrate uptake rate.
Proteome Analysis-Cells were resuspended in deionized water supplemented with a protease inhibitor mixture (Complete; Roche Applied Science) and passed through a small French press cell three times, followed by centrifugation (5 min, 8000 ϫ g) to remove cell debris. Proteins were separated by one-dimensional SDS/PAGE (Criterion Tris-HCl gel, 10.5-14%, 13.3 ϫ 8.7 cm; Bio-Rad) and analyzed after tryptic digestion (trypsin; Promega) by reversed phase high performance liquid chromatography coupled to high accuracy mass spectrometers as described previously (24). MS/MS spectra were searched against a data base using Mascot (Matrix Science) and X!tandem. A data base containing all annotated proteins in the M. extorquens AM1 genome was downloaded from the Genoscope website (25). Using a decoy data base, the false positive rate was found to be less than 0.01%. To determine significant changes in protein levels during growth on acetate compared with methanol, spectral counts of each protein were normalized to the sum of all spectra detected for each sample. Average values for each substrate were obtained from three biological replicates and were used to calculate fold changes of normalized spectral counts. To evaluate the statistical significance of observed changes, one-way analysis of variance with log2 transformation was used (26). Prior to normalization, zero values were set to one.
Dynamic Labeling Experiment-13 C labeling experiments were performed as described (13) with adaptations. The cells were pregrown on 5 mM acetate at natural 13 C abundance. Incubation of cells with labeled acetate was performed in 50-ml Falcon tubes containing 2 ml of minimal medium with 5 mM [U-13 C]acetate. After the addition of 1 ml of culture in exponential growth phase, the sample was continuously mixed and incubated for various times. Quenching and metabolite extraction were performed as detailed below. Calculations of 13 C label incorporation were conducted as described (13); the labeling fractions were normalized to the ratio of acetate with a natural 13 C abundance from the culture and uniformly 13 C-labeled acetate from fresh medium.
Sampling, Quenching, and Metabolite Extraction-For the purpose of analyzing label incorporation by mass spectrometry sampling, quenching and extraction of CoA thioesters were performed as described (13) with adaptations. Sample volumes of 3 ml containing 0.6 mg of CDW were added to 13.5 ml of Ϫ20°C 95% acetonitrile containing 25 mM formic acid. After incubation for 10 min on ice with occasional mixing, the samples were frozen in liquid nitrogen and lyophilized. Sampling of keto acids was performed by fast filtration for salt reduction as described by Bolten et al. (27) with some modifications. Cell suspensions of 3 ml (0.6 mg of CDW) were harvested by vacuum filtration and washed with 5 ml of phosphate buffer containing a 90% reduced salt concentration. Subsequently, the filter was transferred to a vessel containing boiling water as described (28) for quenching and metabolite extraction.
HPLC-MS Analysis-All of the HPLC-MS analyses were performed with a Rheos 2200 HPLC system (Flux Instruments, Basel, Switzerland) coupled to an LTQ Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA) equipped with an electrospray ionization probe. CoA thioester samples were resuspended in 100 l of ammonium formate buffer (25 mM, pH 3.5). The supernatant collected after centrifugation was analyzed by HPLC-MS as described (13) with slight modifications. For online desalting, two C 18 analytical columns (Phenomenex, Torrance, CA) were used. The samples were loaded onto a 50 ϫ 2.0-mm C18 column (3-m particle size; Gemini, Phenomenex), and the sample was washed for 5 min with 100% solvent A (50 mM formic acid adjusted to pH 8.1 with NH 4 OH). During desalting, the short column was connected to waste via a six-port valve, and the 100 ϫ 2.0-mm C18 column (3-m particle size; Gemini, Phenomenex) was equilibrated with solvent A with an additional pump. After desalting, both columns were connected in series, and the following methanol gradient was applied to separate CoA thioesters: 5 min, 5%; 15 min, 23%; 25 min, 80%; and 27 min, 80%. The HPLC-MS system was equilibrated for 6 min at initial elution conditions between two successive analyses. Keto acids were analyzed after derivatization. To this end, the samples were resuspended in 100 l of 30 mM pentafluorobenzyl hydroxylamine as the derivatization reagent, followed by incubation for 45 min at 45°C. Subsequently, the samples were centrifuged (4°C, 5 min, 20,000 ϫ g) and analyzed by HPLC-MS. The oximated acids were separated by reversed phase HPLC after online desalting using the same setup described above for CoA thioesters. Solvent A was 2 mM acetic acid with 4 mM NH 4 OH at pH 9.3, and solvent B was methanol. A gradient was applied as follows: after 5 min of online desalting, solvent B was set linearly from 3 to 90% within 33 min. Amino acids were analyzed as described (28).
Sample Preparation and NMR Analysis-To extract proteinogenic amino acids, the cell pellets were resuspended in 20 mM Tris-HCl, pH 7.6, and disrupted by three successive freezethaw cycles (submersion in liquid N 2 for 10 s), three passages through a small French press cell, and three rounds of sonication (30 s at 23 kHz). Cell debris was removed by ultracentrifugation, and proteins were precipitated in 70% ethanol (final concentration) and hydrolyzed in 6 M HCl for 12 h. After deuteriation, two-dimensional [ 1 H-13 C]heteronuclear single quantum coherence spectroscopy and two-dimensional zero quantum filtered total correlation spectroscopy NMR spectra were recorded on a Bruker Avance II 500-MHz spectrometer as described (13,29,30).
Metabolic Flux Analysis-For the purpose of flux analysis, M. extorquens AM1 was grown in the presence of 5 mM 13 C-labeled acetate (80% [2-13 C]acetate and 20% [U-13 C]acetate), and the culture was aerated with synthetic air containing 5% [ 12 C]carbon dioxide. In total, three independent biological replicates were performed. The cells were inoculated at an OD 600 of 0.01 and harvested at OD 600 of 1. Labeling of acetate was designed to resolve fluxes in the TCA cycle, EMC pathway, serine cycle, and C1 metabolism. Flux calculations were performed using the modified reaction network (13,14), in which the half-reaction model of the pentose-phosphate pathway (31) and phosphoacetyl transferase were introduced (see supplemental Table  S2.3). Biosynthetic requirements determined from the quantification of biomass composition were included. Flux distributions were calculated from the positional and mass isotopomers of amino acids obtained by NMR and HPLC-MS (as described above) using the 13 C-Flux software developed by Wiechert et al. (32). Sensitivity analysis was conducted to establish the confidence interval of the calculated fluxes.

RESULTS
Macrokinetic Growth Characterization of M. extorquens AM1 during Growth on Acetate-M. extorquens AM1 was grown in bioreactors in the presence of acetate as its sole source of carbon and energy, whereby the growth rate decreased with increasing substrate concentration (growth rate on 5 mM acetate was 0.068 h Ϫ1 , compared with 0.025 h Ϫ1 on 30 mM acetate). The biomass yield for M. extorquens was 0.37 Ϯ 0.01 g of [C]/ CDW/g of [C] on 5 mM acetate. The one-dimensional 1 H NMR analysis of the supernatant revealed no significant accumulation of cultivation products in the medium (limit of detection: 10 M measured at A 600 of 1). During growth on 5 mM acetate, the specific substrate uptake rate was determined to be 4.0 Ϯ 0.4 mmol⅐g Ϫ1 [CDW]⅐h Ϫ1 , and the proton production rate was 0.53 Ϯ 0.05 mmol⅐g Ϫ1 [CDW]⅐h Ϫ1 . The amount of polyhydroxybutyrate (PHB) was 13 Ϯ 0.4%, which was significantly higher than on methanol (2 Ϯ 0.05%). The biomass composition of M. extorquens AM1 was determined (supplemental Tables S2.1 and S2.2) to calculate accurate biomass precursor requirements for the calculation of 13 C metabolic flux (see below).
Identification of Proteins Present in Cells Grown on Acetate-To determine the enzymatic framework underlying the metabolic network topology and to reveal proteins specific for growth on acetate, we performed a proteome analysis of M. extorquens AM1 grown on acetate. We used cultures grown in the presence of methanol as the reference because metabolism under methylotrophic conditions is well characterized (3,13,14,33), as is the proteome (34,35). Cell extracts from cells grown on acetate and methanol were loaded onto a one-dimensional SDS-PAGE and were analyzed by HPLC-MS/MS after tryptic digestion. In total, ϳ2600 proteins were identified (supplemental Table S1.1). Although fold changes in spectral counts are not linearly correlated with fold changes in protein amounts, this approach is suitable for determining significant variations in protein levels among different sample sets (36). To this end, fold changes in normalized spectral counts were determined, and statistical significance was tested by one-way analysis of variance. The majority of proteins detected under both growth conditions, i.e. acetate and methanol, did not change, thus indicating an overall consistency in the protein repertoire. Among all proteins with a p value smaller than 0.05, 53 proteins from cells grown in the presence of acetate showed a fold change higher than two relative to growth on methanol, and nine of these exhibited fold changes higher than five. Fifty-four proteins were less abundant on acetate (fold change Ͻ0.5) com-pared with methanol, and nine of these exhibited fold changes smaller than 0.2. In total, 53 proteins were detected only in samples from cells grown on acetate, and 20 were detected only in samples from cells grown on methanol; however, most of these were of low abundance (detected with less than 0.3 ‰ of total spectral counts). All of these proteins are listed in supplemental Table S1.2; enzymes of central metabolism and their putative roles are highlighted in Table 1. Acetate must first be activated before it can enter central metabolism. Among the proteins specifically detected or enriched during growth in the presence of acetate, an acetate transporter and acetyl-CoA synthetase were found. Because we detected higher expression of acetyl-CoA synthetase (fold change, 2.9) in cells grown on acetate compared with methanol and we could barely detect acetate kinase and phosphoacetyl-CoA transferase (both less than 0.01‰), we predict acetyl-CoA synthesis by AMP-forming acetyl-CoA synthetase. Accordingly, a proton-translocating pyrophosphate synthase was found only in acetate-grown cells; this enzyme would allow for the effective conservation of energy by coupling proton pumping across the membrane to the cleavage of pyrophosphate, the product of acetyl-CoA formation via acetyl-CoA synthetase.
All of the TCA cycle enzymes were detected during growth on acetate; all of them (with the exception of aconitate hydratase and malate dehydrogenase) exhibited a fold change greater than two relative to methanol-grown cells. The apparent increase in these proteins is consistent with the enhanced operation of the TCA cycle, which we would predict in the case of acetyl-CoA oxidation for energy generation. In principle, oxidation of C1 units to carbon dioxide could be an alternative to catabolic oxidation via the TCA cycle to generate reducing equivalents. In such a scenario, acetate could be converted to glyoxylate and subsequently to glycine, followed by decarboxylation by a glycine cleavage complex to m-H 4 F, which is oxidized to formate via tetrahydrofolate (H 4 F) intermediates (though this would generate NADPH rather than NADH (37)) or, indirectly, via tetrahydromethanotperin intermediates (generating NAD(P)H (38)) and subsequent oxidation of formate to carbon dioxide (39). The glycine cleavage complex was more abundant during growth on acetate, whereas it was weakly detected in methanol-grown cells. The operation of the enzyme complex can be expected to contribute to the synthesis of m-H 4 F from glycine (40). Enzymes involved in one-carbon metabolism linked to H 4 F and tetrahydromethanopterin (H 4 MPT) were detected in acetate-grown cells. Whereas enzymes dependent on H 4 MPT were present at the same order of magnitude in both acetate-and methanol-grown cells, enzymes involved in C1 unit conversion linked to H 4 F were less abundant on acetate (fold change Ϸ 0.2). Tungsten-containing NAD ϩ -linked formate dehydrogenase 1 (41) was more abundant (fold change Ϸ 3) during growth on acetate than on methanol. Otherwise, pyruvate dehydrogenase could contribute to the greater degree of NADH synthesis in cells grown on acetate than on methanol (fold change Ϸ 5-11). These alternative pathways may balance NADH/NADPH production, whereas the TCA cycle produces a fixed ratio of NADH to NADPH of 1:1. Membrane-bound NAD(P) ϩ transhydrogenase was more abundant (fold change Ϸ 10) in acetate-grown cells than methanol, indicating that balancing redox equivalents is critical during growth on acetate.
All of the enzymes of the EMC pathway for acetyl-CoA assimilation and conversion to glyoxylate and succinyl-CoA were detected in cells grown in the presence of acetate. Most enzymes of the EMC pathway were detected at a similar abundance during growth on acetate relative to methanol, with the exception of mesaconyl-CoA hydratase (fold change 0.4) and malyl-CoA/␤-methylmalyl-CoA lyase (fold change 0.2). The latter is also part of the serine cycle, for which most proteins were less abundant during growth on acetate, suggesting a minor role for serine cycle enzymes during growth on acetate. All of the enzymes involved in the phosphoserine pathway (42) were detected in both acetate-and methanol-grown cells; only phosphoserine aminotransferase was less abundant during growth on acetate (fold change, 0.4).
The storage compound PHB was increased more than 6-fold during growth on acetate compared with methanol. Accordingly, PHB polymerase and granule-associated 11-kDa protein, which are required for PHB granule formation (43), were more abundant in cells grown on acetate.
PEP carboxykinase, which catalyzes the decarboxylation of oxaloacetate into PEP, i.e. operating in reverse compared with the serine cycle PEP carboxylase, was detected in acetate-grown cells. Pyruvate kinase, which is required for the conversion of PEP to pyruvate (3), was less abundant during growth on acetate than on methanol (fold change, 0.3), whereas pyruvate phosphate dikinase, which catalyzes the conversion of pyruvate to PEP (3), was more abundant (fold change, 1.8). Pyruvate may be synthesized by malic enzyme, which was more abundant during growth on acetate than on methanol (fold change, 3.5).
Dynamic Acetate Label Incorporation-A central question regarding growth on C2 substrates concerns the involvement of the EMC pathway, which converts acetyl-CoA to glyoxylate and succinyl-CoA. To demonstrate the presence of the enzymes involved (see above), as well as their operation, dynamic 13 C labeling experiments were performed. To this end, [U-13 C]acetate was mixed with a culture of cells grown on naturally labeled acetate, and 13 C incorporation into EMC pathway CoA thioesters was measured by HPLC-MS. In addition, incorporation of the label into the central intermediate 2-oxoglutarate was followed to monitor operation of the TCA cycle. The results are presented in Fig. 1 as percentages of 13 C-labeled atoms incorporated into the metabolite, normalized to the maximal number of [ 13 C]carbon atoms that can be incorporated from [U-13 C]acetate according to the network topology.
As expected, acetyl-CoA was the first metabolite in which the label was detected; subsequently, label incorporation into CoA thioesters of the EMC pathway was observed, in the following sequence: hydroxybutyryl-CoA, ethylmalonyl-CoA, methylsuccinyl-CoA, mesaconyl-CoA, propionyl-CoA, and methylmalonyl-CoA. After methylmalonyl-CoA, succinyl-CoA (Fig.  1A) was labeled, followed by 2-oxoglutarate (Fig. 1B). From these results, we can conclude that the EMC pathway represents an entry point for acetate. Labeling of 2-oxoglutarate was also observed, thus implicating TCA cycle activity for acetyl-CoA conversion. The slower increase in labeled succinyl-CoA relative to methylmalonyl-CoA (Fig. 1B) might be indicative of a complete TCA cycle and the formation of succinyl-CoA from the EMC pathway and TCA. However, exact pool sizes of intermediates or steady-state labeling experiments (see below) are required to confirm this statement.
Approximately 10 s after the addition of [ 13 C]acetate, EMC pathway intermediates reached an apparent isotopic steady state (with the exception of methylmalonyl-CoA and succinyl-CoA). Labeling did not increase further during the next 60 s without reaching enrichment similar to acetyl-CoA, which could be due to the recycling of PHB and/or the permanent hydrolysis and re-esterification of CoA intermediates.
Steady-state Labeling during Growth on Acetate-To quantify fluxes through the EMC pathway, the TCA cycle and more distant steps relative to the entry of acetyl-CoA, such as the operation of the serine cycle/phosphoserine pathway and the potential oxidation of glyoxylate, a 13 C steady-state labeling experiment and 13 C metabolic flux analysis were performed. Isotopomer analysis was conducted by two-dimensional heteronuclear single quantum coherence spectroscopy and twodimensional total correlation spectroscopy NMR measurement combined with HPLC-MS using a retro-biosynthetic approach and was based on three independent biological replicates. The absolute fluxes were calculated using 13 C-Flux software with consideration of the acetate uptake flux and the biomass exit fluxes determined from substrate uptake, biomass composition, and growth rate. The metabolic network topology including in vivo flux values is presented in Fig. 2, and fluxes of the three biological replicates are provided in supplemental Table  S2.4.
Flux calculations revealed that the majority of acetyl-CoA (68%) enters the TCA cycle. Notably, 98% of the acetyl-CoA entering the TCA cycle was found to be completely oxidized to CO 2 ; thus, the TCA cycle represents the main pathway for the generation of reducing equivalents and operates in an almost purely catabolic mode. Only 2% of acetyl-CoA entering the TCA cycle was oxidized to 2-oxoglutarate as a precursor for biomass synthesis. The second largest conversion of acetyl-CoA occurred via the EMC pathway, which converts acetyl-CoA to glyoxylate and succinyl-CoA (21%); thus, 0.8 mmol⅐g Ϫ1 ⅐h Ϫ1 of CO 2 was assimilated, which corresponds to the recycling of 16% of the completely oxidized acetate. A smaller proportion of acetyl-CoA (5%) was condensed with glyoxylate to generate malate. 3% of acetyl-CoA was used directly for biosynthesis, and 5% was used for PHB production. Oxidation of glyoxylate via glycine and m-H 4 F to CO 2 could not be detected with a threshold of detection of 1%, similar to pyruvate dehydrogenase (maximum of 2%). Thus, the main pathway to generate redox equivalents upon acetate oxidation is the TCA cycle.
As mentioned above, acetyl-CoA was condensed with glyoxylate to generate malate; approximately half of the glyoxylate was converted via this route, and the remaining glyoxylate was converted to glycine by serine-glycine aminotransferase. Glycine was used for biomass synthesis (52%), cleaved by the glycine cleavage system to form m-H 4 F (32%), or condensed with m-H 4 F (16%) by serine hydroxymethyltransferase to generate serine. No m-H 4 F was produced by serine hydroxymethyltransferase. Glycine production from glyoxylate by serine glyoxylate aminotransferase requires recycling of hydroxypyruvate, the product of serine deamination; all of the hydroxypyruvate produced was converted to 2-phosphoglycerate by serine cycle enzymes and subsequently to serine via the phosphoserine pathway. The labeling patterns of PEP and pyruvate were identical; both were derived from C4 compounds (oxaloacetate/malate). The substrate cycle of PEP carboxylase/PEP carboxykinase described during C1 assimilation (14) also exists during C2 growth and contributes to a significant loss of energy during growth on acetate (0.69 mmol⅐g Ϫ1 ⅐h Ϫ1 of ATP). No evidence for substrate cycling via malate thiokinase/malyl-CoA thioesterase was observed during growth on acetate, which is in contrast to growth in the presence of methanol.

DISCUSSION
M. extorquens AM1 can grow in the presence of acetate as its sole source of carbon and energy. To understand the catabolic and anabolic strategy of this isocitrate lyase-negative organism, proteome analysis and 13 C labeling experiments, including metabolic flux analysis, were conducted. The glyoxylate cycle allows for complete oxidation by the TCA cycle and replenishes metabolites that leave the cycle for biosynthetic purposes. Here, we show that the EMC pathway functionally replaces the glyoxylate cycle during growth on acetate. This recently discovered pathway operates to refill the TCA cycle; in addition, it supplies glyoxylate for glycine and serine biosynthesis. A more detailed comparison of the operation of the metabolic network revealed 20 and 30% higher rates of production of carbon dioxide and redox equivalents by isocitrate dehydrogenase and 2-oxoglutarate dehydrogenase in M. extorquens AM1 growing on acetate, compared with C. glutamicum (21) and E. coli (18), respectively ( Table 2). This offers M. extorquens AM1 the addi- extorquens AM1 grown on naturally labeled acetate was mixed with fresh medium containing [U-13 C]acetate, and label incorporation was measured in CoA thioesters and 2-oxoglutarate at different times; the average and standard deviation of three technical replicates are given in the graph. The total labeled fraction represents the percentage of 13 C carbon atoms that can be assimilated from [U-13 C]acetate. The total labeled fraction was corrected for CO 2 incorporation for intermediates of the EMC pathway (A).
tional gain of redox equivalents; however, these additional redox equivalents are required (by acetoacetyl-CoA reductase and crotonyl-CoA carboxylase/reductase) to operate the EMC pathway. This anaplerotic reaction sequence contains two car-boxylation steps (crotonyl-CoA carboxylase and propionyl-CoA carboxylase) and recycles 20% of the carbon dioxide. The glyoxylate cycle used by C. glutamicum and E. coli neither requires redox equivalents nor contains carboxylation steps. Interestingly, despite the long and complicated reaction sequence of the EMC pathway, which operates during growth on acetate, the pathway is not disadvantageous to biomass yield.
As a consequence of the overall metabolic network topology of the EMC pathway-positive organism determined in this study, the entry points into metabolism from acetyl-CoA are clearly distinct from those in organisms with a glyoxylate cycle. For isocitrate lyase-positive organisms, two entry points for acetyl-CoA exist: the TCA cycle and malate synthase, which are stoichiometrically coupled. In the case of M. extorquens AM1, three independent entry points were identified in this study, in   (21). b According to Zhao and Shimizu (18). c According to this study. order of their contribution: (i) the TCA cycle, (ii) the EMC pathway, and (iii) condensation with the EMC pathway ("side") product glyoxylate. Consequently, the network topology is more complex in the EMC pathway-positive bacterium, with the convergence of multiple pathways at different metabolites, i.e. succinyl-CoA fueled by the TCA cycle and the EMC pathway, and malate. As a result, fluxes must be tightly regulated to avoid metabolite imbalance at central metabolic nodes, and further investigation will be required to elucidate the underlying mechanism.
M. extorquens AM1 was the first EMC pathway-positive bacterium for which a genome scale metabolic network reconstruction was performed and for which metabolic flux analysis was accomplished under methanol growth conditions (14). The comparison performed in this study of the metabolic network under methylotrophic conditions with the network present during growth on acetate revealed a high degree of overlap in the protein repertoire. Semi-quantitative proteome analysis suggests that EMC pathway enzymes are present in quantities at the same order of magnitude or lower during growth on acetate than on methanol, which is consistent with the enzymatic activities measured by Smejkalova et al. (44). The latter activities were found to be higher than the flux determined via this pathway. Only the activity of ethylmalonyl-CoA mutase (0.55 mmol⅐g Ϫ1 [CDW]⅐h Ϫ1 ) was in the range of the calculated flux value and could represent a growth-limiting step during acetate assimilation, which is in accordance with the earlier identification of the critical B 12 -dependent enzyme reaction (23). However, fluxes through the EMC pathway, normalized to their respective carbon uptake rates, were identical in cells grown on methanol and acetate. Thus, it can be concluded that the EMC pathway is required for the assimilation of both substrates to the same extent relative to their carbon amount. A striking difference was observed in the operation of the TCA cycle under both conditions. The metabolic cycle operates on acetate with a high flux through 2-oxoglutarate dehydrogenase (Fig. 3), whereas it is incomplete under methylotrophic growth conditions (14). The higher flux of carbon through the catabolic TCA cycle during growth on acetate relative to the assimilatory flux in the presence of methanol is consistent with TCA cycle enzymes being more abundant on acetate than on methanol. Alternative routes for substrate oxidation during growth on acetate were determined to play only a minor role, i.e. the oxidation of one-carbon units and pyruvate. Thus, like nonmethylotrophic organisms, methylotrophs also use the TCA cycle as major pathway for the oxidation of acetyl-CoA.
Although a high turnover of the serine cycle is required for assimilation of one-carbon compounds to initiate the formation of multi-carbon compounds during growth on methanol, the cycle plays only a minor role during growth on acetate. This is evident from the observation that most enzymes were less abundant during growth on acetate and is consistent with the results from the flux analysis (Fig. 3). However, in the case of serine hydroxymethyltransferase, only a minor fold change in normalized spectral counts was observed (Ϸ1.5), but flux decreased ϳ70-fold. The integration of data obtained from the flux analysis and proteomics suggests a post-translational flux control of serine hydroxymethyltransferase, for example, by substrate concentrations below the K m value or by allosteric/ competitive enzyme inhibition. Not only was the flux through the serine cycle significantly lower on acetate, but also, at some steps, the net flux direction changed, resulting in a bidirectional operation of the serine "cycle" relative to its classical operation under methylotrophic growth conditions. The flux from glyoxylate to 2-phosphoglycerate via glycine and serine operated in the same direction during growth on acetate and methanol. However, the subsequent steps of the serine cycle, the conversion of 2-phosphogycerate to malate and its subsequent cleavage to glyoxylate and acetyl-CoA, exhibited a reversed flux direction during growth on acetate, despite the irreversible enzymatic reactions catalyzed by PEP carboxylase and malate thiokinase. This is achieved by their substitution with PEP carboxykinase and malyl-CoA thioesterase, respectively. Consequently, all C3 compounds (PEP, pyruvate, and phosphoglycerate) are synthesized by decarboxylation of the C4 intermediates malate and oxaloacetate. However, glycine and serine are synthesized from glyoxylate.
Integrating proteome data and 13 C metabolic flux analysis allowed us to correlate the abundance of proteins and their in vivo activities in M. extorquens AM1 during growth on acetate. A high degree of overlap in the proteome between acetate-and methanol-grown cells was identified, but the connectivity of the metabolic network changed, and fluxes were redirected. This feature should allow the organism to quickly adapt to changes in carbon source availability.