Flux through Citrate Synthase Limits the Growth of Ethanologenic Escherichia coli KO11 during Xylose Fermentation

Abstract

Previous studies have shown that high levels of complex nutrients (Luria broth or 5% corn steep liquor) were necessary for rapid ethanol production by the ethanologenic strain Escherichia coli KO11. Although this strain is prototrophic, cell density and ethanol production remained low in mineral salts media (10% xylose) unless complex nutrients were added. The basis for this nutrient requirement was identified as a regulatory problem created by metabolic engineering of an ethanol pathway. Cells must partition pyruvate between competing needs for biosynthesis and regeneration of NAD⁺. Expression of low-K_m Zymomonas mobilis pdc (pyruvate decarboxylase) in KO11 reduced the flow of pyruvate carbon into native fermentation pathways as desired, but it also restricted the flow of carbon skeletons into the 2-ketoglutarate arm of the tricarboxylic acid pathway (biosynthesis). In mineral salts medium containing 1% corn steep liquor and 10% xylose, the detrimental effect of metabolic engineering was substantially reduced by addition of pyruvate. A similar benefit was also observed when acetaldehyde, 2-ketoglutarate, or glutamate was added. In E. coli, citrate synthase links the cellular abundance of NADH to the supply of 2-ketoglutarate for glutamate biosynthesis. This enzyme is allosterically regulated and inhibited by high NADH concentrations. In addition, citrate synthase catalyzes the first committed step in 2-ketoglutarate synthesis. Oxidation of NADH by added acetaldehyde (or pyruvate) would be expected to increase the activity of E. coli citrate synthase and direct more carbon into 2-ketoglutarate, and this may explain the stimulation of growth. This hypothesis was tested, in part, by cloning the Bacillus subtilis citZ gene encoding an NADH-insensitive citrate synthase. Expression of recombinant citZ in KO11 was accompanied by increases in cell growth and ethanol production, which substantially reduced the need for complex nutrients.

The engineering of metabolic pathways for production of industrial chemicals from renewable carbohydrates offers an alternative to petroleum-based processes and chemical synthesis (3, 12, 22). Genetically altering metabolic pathways, however, often causes unanticipated changes which may limit utility. Undesirable changes, such as reduced growth, decreased glycolytic flux, and low volumetric productivity, are generally attributed to a lack of ATP (19, 50), creation of futile cycles (10, 11, 37), changes in intracellular metabolite pools, or a metabolic imbalance (2, 7, 9, 13, 28, 51, 53). Often, these detrimental effects are masked by abundant complex nutrients in laboratory media and are apparent only in more minimal media (7, 10, 11, 31).

Salmonella enterica serovar Typhimurium was engineered for succinate production by increasing expression of pyc encoding pyruvate carboxylase (50). Although succinate production increased, the growth rate declined by 18% and the glycolytic flux decreased by 40%. Similar results were reported for an analogous construction in Escherichia coli (19). Donnelly and coworkers (15) isolated E. coli mutants which produced five times more succinate than the parent strain, and again the growth rate was impaired. Growth rate and cell yield were also decreased by engineering E. coli for the production of 3-deoxy-d-arabinoheptulosonate 7-phosphate by overexpression of pps (phosphoenolpyruvate synthase) (36, 37). The inhibition of growth was more pronounced in minimal medium (10).

Acetate production during aerobic growth of E. coli on sugars has been correlated with a decline in metabolic activity and reduced expression of heterologous genes (2, 6, 9, 29). Many approaches have been employed to decrease acetate production and increase the levels of recombinant products (2, 5, 6, 9, 13, 53). Mutations in the primary acetate pathway genes (pta, the gene encoding phosphotransacetylase, and ackA, the gene encoding acetate kinase) increased the yield of recombinant proteins but usually reduced cell growth. The detrimental effect on growth was attributed to accumulation of metabolic intermediates, such as acetyl coenzyme A (acetyl-CoA) or acetyl phosphate. An alternative approach, channeling pyruvate away from acetate by expressing the Bacillus subtilis alsA gene encoding acetolactate synthase, also reduced acetate production by 80% and increased product yields but it also reduced cell growth (51). Other attempts to decrease acetate production by increased expression of ldhA (lactate dehydrogenase) were ineffective in rich medium (53). In mineral salts medium, overexpression of ldhA was accompanied by severe growth limitation (7).

Workers in our laboratory have previously engineered E. coli strain B for production of ethanol from pentose-rich hemicellulose syrups by expressing high levels of Zymomonas mobilis pdc (pyruvate decarboxylase) and adhB (alcohol dehydrogenase) (22, 35). This strain was chosen for metabolic engineering because of its hardiness, wide substrate range, and ability to grow well in mineral salts medium without organic nutrients (1, 29). During xylose fermentation, the ATP yield in E. coli is quite low (∼0.67 ATP per xylose) due to separate energy requirements for uptake and phosphorylation (42). Unlike most genetically engineered strains of E. coli, strain KO11 grew to higher densities than the parent in both mineral salts and complex media (31). Initial studies performed with Luria broth demonstrated that KO11 rapidly and efficiently converted sugars to ethanol, with yields approaching 95% of the theoretical maximum. However, the volumetric productivity and ethanol yields were considerably lower in mineral salts medium without complex nutrients (4, 26, 31, 34, 54, 55).

Supplementing mineral salts medium with complex nutrients significantly increased ethanol production. The least expensive complex nutrient, corn steep liquor, supported growth rates and levels of ethanol production near the values obtained with Luria broth but only when it was provided at a high concentration (5%, wt/vol). Although not prohibitively expensive, addition of high levels of complex nutrients adds to the cost of ethanol production and increases the requirements for waste treatment.

The lower rates of ethanol production (volumetric productivity) in minimal media resulted from low cell densities and reduced expression of recombinant pdc and adhB genes (lower metabolic activity). Inorganic components did not appear to be limiting, and no specific auxotrophic requirements could be identified (31). Reduced expression of heterologous genes was attributed to biosynthetic burden, the competitive reduction in synthesis of heterologous products due to derepression of native genes for biosynthetic enzymes (31). In this study, we reduced the corn steep liquor concentration to 1% and investigated the basis of the requirement for higher levels of nutrients during xylose fermentation. The following four possibilities to explain this requirement were examined: (i) availability of macronutrients, (ii) loss of a biosynthetic pathway due to metabolic engineering, (iii) insufficient ATP during xylose fermentation, and (iv) an imbalance in central metabolism.

MATERIALS AND METHODS

Microorganisms and media.

E. coli B (= ATCC 11303) and an ethanologenic derivative, strain KO11 (35), were used in all fermentation experiments. KO11 contains a deletion in the frd region (anaerobic fumarate reductase) which eliminates succinate production. Genes encoding the Z. mobilis ethanol pathway (pdc, adhB) and chloramphenicol acyltransferase (cat) were integrated into the pfl gene (chromosome) by a single crossover event that resulted in a functional, full-length pfl gene downstream. Both E. coli B and KO11 are prototrophic. Stock cultures were stored in glycerol at −75°C. Working cultures were transferred daily on solid medium containing mineral salts and 1% corn steep liquor. Xylose (2%) and chloramphenicol (alternating between 40 and 600 mg liter⁻¹) were included in the solid media for K011.

A citrate synthase mutant, E. coli W620 (= CGSC 4278) (glnV44 gltA6 galK30 LAM⁻ pyrD36 relA1 rpsL129 thi-1), was obtained from the E. coli Genetic Stock Center and was used to test expression of the B. subtilis citZ gene (citrate synthase). This strain contains a gltA6 (citrate synthase) mutation that prevents growth on M9 medium containing thiamine and glucose (21).

Corn steep liquor medium (CSL medium) contained (per liter of distilled water) 10 g of corn steep liquor (∼50% solids), 1 g of KH₂PO₄, 0.5 g of K₂HPO₄, 3.1 g of (NH₄)₂SO₄, 0.4 g of MgCL₂ · 6H₂O, and 20 mg of FeCl₃ · 6H₂O. A 1-liter stock solution of corn steep liquor was prepared by dilution of 200 g with distilled water, adjustment to pH 7.2 with 50% NaOH, and steam sterilization. Before use, the sterile stock solution of corn steep liquor was aseptically clarified by centrifugation (10,000 × g, 5 min). Mineral solutions were prepared as described previously (31). Broth cultures and fermentations contained 10% xylose (CSL+X medium), unless indicated otherwise. In some experiments, Luria broth containing xylose was included for comparison.

Fermentation.

Seed cultures (100 ml in 250-ml flasks) were grown for 14 to 16 h at 35°C with agitation (120 rpm). Cells were harvested by centrifugation (5,000 × g, 5 min) and used as an inoculum to provide an initial concentration of 33 μg ml⁻¹ (optical density at 550 nm [OD₅₅₀], 0.1). The fermentation vessels each contained a total volume of 350 ml, and the cultures were incubated at 35°C and 100 rpm. The cultures were maintained at pH 6.5 by automatic addition of 2 N KOH (34). For strain B, 6 N KOH was used to maintain the pH after the initial 24 h.

Supplements were added with distilled water (total volume, 10 ml) as necessary. Organic acids and amino acids were neutralized, sterilized by filtration, and added at a final concentration of 2 mg ml⁻¹. Acetaldehyde was added at a final concentration of 0.25 or 0.5 mg ml⁻¹. Cell mass and ethanol, organic acid, and sugar contents were monitored at 24-h intervals.

Aerobic growth studies.

Cells were grown with aeration at 35°C and 220 rpm in 250-ml baffled flasks containing 50 ml of CSL medium. A range of sugar concentrations (0.5 to 5%) was tested to determine maximal cell density under excess-sugar conditions. The flasks were inoculated directly by using cells grown on solid media for 18 to 24 h. Ethanol content and cell mass were measured after 16 h. For comparison, Luria broth containing 5% xylose was also used.

Analytical methods.

Cell mass was estimated by determining OD₅₅₀ with a Bausch & Lomb Spectronic 70 (1 U of OD₅₅₀ = 0.33 mg [dry weight] of cells ml⁻¹). Ethanol content was measured by gas chromatography with a Varian model 3400 CX as described previously (34). Organic acid and sugar contents were measured by high-performance liquid chromatography (HPLC) by using an HP 1090 Series II chromatograph equipped with a Bio-Rad Aminex HPX-87H ion exclusion column (45°C; 4 mM H₂SO₄; flow rate, 0.5 ml min⁻¹; injection volume, 10 μl) and dual detectors (refractive index monitor and UV detector at 210 nm).

Fermentation products were also analyzed by nuclear magnetic resonance to confirm the identities of HPLC peaks. Broth samples were centrifuged to remove cells. Supernatants (0.9 ml) were mixed with deuterium oxide (0.1 ml) and sodium 3-(trimethylsilyl)propionate (10 mM) as an internal standard in 5-mm sample tubes. Proton spectra were obtained with a modified Nicolet NT300 spectrometer operated in the Fourier transform mode (8) as follows: frequency, 300.065 MHz; excitation pulse width, 5 μs; pulse repetition delay, 3 s; and spectral width, 3.6 kHz. A minimum of 100 acquisitions were obtained for each sample.

Genetic methods.

The citZ gene encoding B. subtilis citrate synthase II has been described previously (24). This gene was amplified by PCR (forward primer, 5′-TGTGCTCTTCCATGTTTTTACAACACTGTTAAAG-3′; reverse primer, 5′-TTGCTCTTCGTTAGGCTCTTTCTTCAATCG-3′) using genomic DNA from B. subtilis strain YB886 as the template (5). Primers were added to a Taq PCR Master mixture (Qiagen) as recommended by the manufacturer. The following conditions were used for thermal cycling: (i) two initial cycles consisting of denaturation at 94°C for 60 s, annealing at 50°C for 60 s, and elongation at 68°C for 90 s; (ii) 28 cycles consisting of denaturation at 94°C for 10 s, annealing at 70°C for 60 s, and elongation at 68°C for 90 s; and (iii) a final elongation step consisting of 72°C for 10 min. The PCR product (1.5 kbp) was cloned into pCR2.1-TOPO (Invitrogen) by using ampicillin (50 μg/ml) for selection. Colonies were screened for size and ability to complement the gltA mutation of E. coli W620 on glucose minimal medium (21). The presence of the citZ gene was also confirmed by DNA sequencing with a LI-COR model 4000L sequencer (33).

NAD(P)H/NAD(P)⁺ ratio.

Whole-cell fluorescence was used as a relative measure of reduced nucleotides in situ (43, 45). Since only the reduced form NAD(P)H fluoresces at 460 nm, an immediate decrease in the fluorescence of fermenting cells was interpreted as a decline in the level of NAD(P)H and a decline in the NAD(P)H/NAD(P)⁺ ratio. Cells were grown for 12 h in CSL+X medium, harvested by centrifugation (5,000 × g, 5 min), and washed three times in mineral salts medium (CSL medium lacking corn steep liquor and sugar). The pellet was then suspended in mineral salts solution to an OD₅₅₀ of 1.0. Emission at 460 nm (excitation wavelength, 340 nm) was recorded at 5-s intervals with an Aminco-Bowman Series 2 luminescence spectrometer. Cells were energized by adding 1% xylose, which resulted in an immediate increase in fluorescence, primarily due to an increase in the NADH/NAD⁺ ratio. Test compounds were added at a final concentration of 2 mg ml⁻¹ (organic acids, amino acids) or 0.25 mg ml⁻¹ (acetaldehyde), and distilled water was used as a control. The results obtained with each test compound were expressed relative to the xylose-dependent increase in fluorescence. Control experiments confirmed that quenching of cellular fluorescence did not occur when additives were mixed with energy-deficient cells (without xylose).

Enzyme assays.

Citrate synthase was assayed by using a modification of the method described previously (16, 17). Cultures were grown in 1-liter flasks containing 250 ml of Luria broth for 16 h at 35°C and 150 rpm. Cells were harvested by centrifugation, washed three times in buffer containing 50 mM Tris-Cl (pH 8.0) and 20% glycerol, and suspended in 2 volumes of the same buffer. Cell-free preparations were obtained by two passages through a French pressure cell (20,000 lb/in²), followed by treatment with ∼100 μg of DNase I per ml. Cell debris was removed by centrifugation (15,000 × g, 1 h, 4°C). The supernatant was dialyzed against 20 mM Tris-Cl and 20% glycerol. Each assay mixture (1 ml) contained 20 mM Tris-Cl (pH 8.0), 10 mM KCl, 1 mM 5′,5′-dithio-bis(2-nitrobenzoic acid), 10 mM oxaloacetate, and 0.5 mM acetyl-CoA. Reactions were initiated by adding cell lysate and were monitored for 300 s at 412 nm. Specific activity was expressed in micromoles of reduced CoA produced per minute per milligram of protein.

RESULTS

Macronutrient limitation.

Previous studies have shown that up to 5% corn steep liquor is needed to support anaerobic growth and ethanol production at a rate near the rate observed in Luria broth (31). Based on a comparison of E. coli elemental composition (44), the macronutrient salts in CSL+X medium should provide sufficient nitrogen and phosphorus to support the growth of at least 5 mg (dry weight) ml⁻¹. During anaerobic growth in pH stats with CSL+X medium (Fig. 1), the maximal cell density for KO11 was only 1 mg ml⁻¹, which was 33% lower than the value obtained for the parent strain, E. coli B (1.5 mg ml⁻¹), in CSL+X medium and only 25% of the value obtained for KO11 (4 mg ml⁻¹) in Luria broth plus sugar (31, 54, 55). During aerobic growth with the same nutrients, however, KO11 grew to a maximum density of 2.7 mg ml⁻¹. This value is almost 2-fold higher than the density of E. coli B grown under the same conditions and 2.7-fold higher than the density of KO11 grown anaerobically. Together, these results indicate that anaerobic growth of KO11 in the presence of 1% corn steep liquor and 10% xylose is not limited by the availability of macronutrients (i.e., N, P, etc.) or by inactivation of a biosynthetic pathway due to genetic manipulation. However, metabolic engineering of the ethanol pathway does appear to contribute to the reduced growth of KO11 in this medium under anaerobic conditions.

FIG. 1.

Comparison of maximal cell densities achieved during aerobic and anaerobic growth in mineral salts medium containing 1% corn steep liquor and either xylose (X) or glucose (G). The error bars indicate standard errors of the means for experiments in which three or more replicates were examined.

Energy limitation.

The separate energy requirement for uptake (ATP-dependent transporter or proton symport) and phosphorylation (xylulokinase) results in a low net yield of ATP from xylose fermentation (0.67 ATP per xylose, which is 33% of the yield from glucose [2 ATP per glucose]) (42). An additional ATP can also be produced from pyruvate by the acetate pathway. Since KO11 produced less acetate than strain B (Table 1), the growth of KO11 could be limited by the availability of ATP. To test this hypothesis, we compared the growth of KO11 in the presence of 1% corn steep liquor and 10% xylose to the growth of KO11 in the presence of 1% corn steep liquor and 10% glucose (Fig. 2). Glucose was fermented to ethanol at a faster rate than xylose was fermented to ethanol. However, the growth of KO11 was the same with both sugars, despite the threefold difference in net ATP production. Although the cell densities were low, the cells remained metabolically active for at least 96 h and produced most of the ethanol after growth had ceased.

TABLE 1.

Effects of additives on the composition of the fermentation products after 24 h of incubation in CSL medium containing 1% corn steep liquor and 10% xylose.

Medium or strain	Additive	% of additive remaining	Concn of fermentation products (mM)
			Fumarate	Succinate	Lactate	Formate	Acetate	Ethanol
CSL medium			<0.01	<0.3	9.6	<1.5	<1.0	<1.0
E. coli B (parent)	None		<0.01	53.4	27.7	95.8	91.4	71
E. coli KO11	None		0.03	0.5	12.2	20.1	24.0	150
	Sodium pyruvate (2 mg ml−1)	<5	0.29	0.9	12.1	13.2	27.4	208
	Acetaldehyde (2 × 0.25 mg ml−1)	<5	0.21	0.5	11.1	25.6	26.6	161
	Citric acid (2 mg ml−1)	<5	1.65	0.6	11.8	11.9	27.0	154
	Isocitric acid (2 mg ml−1)	78	<0.01	0.5	11.4	23.5	23.0	140
	Sodium 2-ketoglutarate (2 mg ml−1)	<5	0.28	0.9	12.5	37.8	32.1	193
	Oxaloacetic acid (2 mg ml−1)	10	0.02	0.8	10.8	17.9	20.4	100
	Sodium malate (2 mg ml−1)	6	1.45	0.3	10.9	14.7	17.3	125
	Sodium fumarate (2 mg ml−1)	16	1.22	0.4	10.1	19.7	18.8	110
	Sodium succinate (2 mg ml−1)	100	0.07	13.0	11.3	20.0	22.7	144

FIG. 2.

Comparison of growth (A) and ethanol production (B) by E. coli KO11 during fermentation of 10% xylose or 10% glucose in mineral salts medium containing 1% corn steep liquor. The error bars indicate standard errors of the means for experiments in which three or more replicates were examined.

Metabolic imbalance relieved by addition of pyruvate or acetaldehyde.

Pyruvate serves a dual role during fermentation; it is both a source of carbon skeletons for biosynthesis and a source of electron acceptors (acetaldehyde) that allows continued ATP production by glycolysis. During sugar fermentation to ethanol, one NADH molecule is produced per pyruvate molecule. Each NADH molecule must be oxidized by reducing an electron acceptor, such as acetaldehyde, or by biosynthetic reactions (32). It is presumed that in a growing wild-type E. coli strain, partitioning of pyruvate between biosynthesis and redox needs is balanced for optimal growth. Metabolic engineering of the ethanol pathway contributed to the reduced growth of KO11 only under fermentative conditions, which is consistent with a metabolic imbalance resulting from uncontrolled utilization of pyruvate for ethanol production. This possibility was confirmed by addition of pyruvate to CSL+X medium. Addition of pyruvate resulted in dose-dependent increases in cell growth and ethanol production that were particularly evident after 24 h (Fig. 3A and B). When 2 mg of pyruvate per ml was added, the growth and ethanol production values were twice the values obtained for the control to which pyruvate was not added (Table 2). Adding pyruvate did not result in a buildup of tricarboxylic acid (TCA) intermediates or acidic fermentation products (Table 1). Note that formate was produced in all fermentations, confirming that the pfl gene encoding pyruvate formate-lyase is functional in KO11.

FIG. 3.

Effects of added pyruvate and acetaldehyde on growth and ethanol production by E. coli KO11 in CSL+X medium. (A) Cell growth with added pyruvate; (B) ethanol production with added pyruvate; (C) cell growth with added acetaldehyde; (D) ethanol production with added acetaldehyde. The error bars indicate standard errors of the means.

TABLE 2.

Effects of additives on growth and ethanol production by KO11 in CSL medium containing 1% corn steep liquor and 10% xylose

Additive	No.	Concn		Cell yield		Volumetric production of ethanola		Maximum ethanol production		Yield (%)b
		mg ml−1	mM	Avge	% of control	g liter−1 h−1	% of control	mg ml−1	% of control
None (control)	26			0.94 ± 0.15	100	0.38 ± 0.08	100	31.12 ± 4.64	100	61
Sodium pyruvate	3	0.5	4.6	1.32 ± 0.32	140	0.55 ± 0.17	144	37.60 ± 5.91	121	74
Sodium pyruvate	3	1.0	9.1	1.49 ± 0.30	158	0.66 ± 0.17	173	40.09 ± 5.53	129	79
Sodium pyruvate	15	2.0	18.2	1.99 ± 0.20	212	0.81 ± 0.14	213	44.22 ± 2.69	142	87
Sodium pyruvate	3	4.4	36.4	2.08 ± 0.08	221	0.83 ± 0.01	218	43.23 ± 1.50	139	85
Acetaldehyde (half initially + half after 12 h)	4	0.5 (total)	11.4	1.55 ± 0.07	165	0.60 ± 0.35	158	44.05 ± 1.89	142	86
Citric acid	2	2.0	10.4	0.88, 0.75	86	0.35, 0.33	89	30.00, 27.59	93	56
Isocitric acid	2	2.0	6.6	1.11, 0.88	105	0.51, 0.36	113	36.45, 29.59	106	65
Sodium 2-ketoglutarate	5	2.0	11.9	1.89 ± 0.18	201	0.84 ± 0.08	221	41.54 ± 1.24	133	81
Oxaloacetic acid	2	2.0	15.1	0.75, 0.75	80	0.28, 0.28	74	23.58, 23.64	76	46
Sodium malate	2	2.0	11.2	0.76, 0.84	85	0.27, 0.27	71	23.89, 24.97	79	48
Sodium fumarate	2	2.0	12.5	0.80, 0.96	91	0.37, 0.32	92	27.15, 29.84	92	56
Sodium succinate	3	2.0	7.4	1.00 ± 0.12	106	0.41 ± 0.05	108	35.00 ± 1.31	112	69
Potassium glutamate	3	2.0	10.8	1.66 ± 0.17	177	0.66 ± 0.13	174	43.14 ± 0.13	139	85
Aspartic acid	2	2.0	15.0	0.85, 0.82	88	0.33, 0.32	84	28.40, 27.40	90	55

^aVolumetric production was calculated by determining the average hourly rate of ethanol production between 24 and 48 h after inoculation.

^bYield is expressed as a percentage of the theoretical yield (100% = 0.51 g of ethanol per g of xylose).

Addition of pyruvate to media has been shown to increase the intracellular pyruvate pool in E. coli (52), which increases the ratio of potential electron acceptors for oxidation of NADH (from glycolysis). When added at a level of 2 mg ml⁻¹, pyruvate was metabolized concurrent with cell growth during the first 24 h after inoculation (Table 1). The pyruvate-dependent increase in cell mass (∼1 mg ml⁻¹) was roughly equivalent to one-half of the added pyruvate (Table 2; Fig. 3A). It was presumed that the remaining pyruvate was metabolized to acetaldehyde by the recombinant Z. mobilis pyruvate decarboxylase. Since acetaldehyde has been shown previously to stimulate growth and ethanol production by yeasts (46) and Z. mobilis (40), it seemed possible that the stimulation of cell growth by pyruvate could be mediated in part by an increase in acetaldehyde obtained from pyruvate (Table 2; Fig. 3C and D). Concentrations of acetaldehyde greater than 0.50 mg ml⁻¹ were toxic. In the presence of lower concentrations of acetaldehyde (0.25 and 0.50 mg ml⁻¹), cell growth and ethanol production increased. Like pyruvate, added acetaldehyde was fully metabolized during the initial 24 h after inoculation (Table 1). A nearly optimal level of acetaldehyde was provided by two additions of 0.25 mg ml⁻¹ to CSL+X medium (initially and after 12 h). This treatment was almost as effective as addition of pyruvate (2 mg ml⁻¹) for stimulating ethanol production and also resulted in a 65% increase in cell mass. The basis for the increase in cell growth is not readily explained by the limited routes for acetaldehyde metabolism in E. coli compared to the routes for metabolism of pyruvate, a key central metabolite. These results provide evidence that the beneficial effect of added pyruvate results primarily from an increase in electron acceptors.

Pyruvate as a source of carbon skeletons for biosynthesis.

The pyruvate-stimulated increase in cell growth reflected a twofold increase in the flow of carbon into biosynthesis. Pyruvate and upstream metabolites in glycolysis are used for the biosynthesis of approximately one-half of cellular constituents. The pools of these upstream intermediates may increase when pyruvate is added, increasing the availability for biosynthesis. Pyruvate and phosphoenolpyruvate are also converted to a series of biosynthetic intermediates by the TCA pathway and linking reactions. The TCA pathway provides one-half of the carbon skeletons for cell protein. None of the TCA pathway intermediates can be produced readily from acetaldehyde by biosynthetic reactions. Note that the TCA pathway is not cyclic during fermentation. This pathway is interrupted between 2-ketoglutarate and succinate by ArcAB-mediated repression of genes encoding 2-ketoglutarate dehydrogenase (sucAB) (23). One side of the TCA pathway produces 2-ketoglutarate, the precursor for the glutamic acid family of amino acids, polyamines, among other compounds. Precursors such as oxaloacetate on the other side of the TCA pathway are derived from phosphoenolpyruvate. Oxaloacetate is used for synthesis of the aspartic acid family of amino acids, etc. Addition of pyruvate could potentially increase the flow of carbon to both sides.

TCA pathway intermediates were tested as additives to CSL+X medium. Utilization of these additives was investigated by using HPLC and nuclear magnetic resonance (Table 2; Fig. 4). All but two of the additives, succinate (100% remaining) and isocitrate (78% remaining), were metabolized efficiently during the initial 24 h of fermentation (Table 1). Additions of malate and fumarate resulted in similar small increases in fumarate levels but did not stimulate growth or ethanol production. Despite the potential interconversion of these intermediates, fumarate did not accumulate when oxaloacetate was added. Addition of aspartate, the transamination product of oxaloacetate, was similarly ineffective. Indeed, addition of oxaloacetate, malate, fumarate, and aspartic acid reduced growth and ethanol production. In contrast, 2-ketoglutarate was almost as effective as pyruvate in stimulating growth and ethanol production by KO11. A similar stimulation was also observed with glutamate, the transamination product of 2-ketoglutarate.

FIG. 4.

Initial effects of added TCA pathway intermediates on growth (A) and ethanol production (B) by E. coli KO11 incubated for 24 h. The error bars indicate standard errors of the means.

TCA pathway intermediates that are immediate precursors of 2-ketoglutarate were not beneficial. Isocitrate was not readily metabolized. Citrate was metabolized but had no effect on growth and ethanol production. Growth with added citrate was accompanied by accumulation of fumarate and a high acetate/formate ratio similar to the ratio obtained with pyruvate (Table 1). Note that this ratio is near unity for other fermentations with TCA pathway intermediates, which provides a clue to why citrate is ineffective. Addition of citrate may induce citrate lyase (18, 30, 39), an enzyme that cleaves citrate into an equimolar mixture of oxaloacetate and acetate. Oxaloacetate is readily metabolized to fumarate. The concentrations of both fumarate and acetate were higher in fermentations with added citrate than in fermentations with 2-ketoglutarate and other TCA pathway intermediates, which is consistent with induction of citrate lyase. It is presumed that induction of this enzyme blocks the beneficial effects of this TCA pathway intermediate for biosynthesis.

When considered together, studies in which added TCA pathway intermediates were used provided evidence that the beneficial effect of pyruvate on growth and ethanol production by KO11 in CSL+X medium results in large part from an increase in the flow of carbon skeletons into 2-ketoglutarate and subsequent products of biosynthesis. However, investigations with added pyruvate and acetaldehyde provided evidence that an increase in electron acceptors was arguably of primary importance for the beneficial effect of pyruvate. For both of these effects to be possible, both must be mediated by a common mechanism.

Whole-cell fluorescence.

The ratio of NAD(P)H to NAD(P)⁺ has been shown to alter cellular patterns of metabolic flux (14). NAD(P)H is an allosteric inhibitor of many enzymes, including pyruvate dehydrogenase (20), phosphotransacetylase (41), malate dehydrogenase (38), and citrate synthase (17, 48). In KO11, addition of acetaldehyde or pyruvate (metabolized to acetaldehyde by recombinant pyruvate decarboxylase) would be expected to decrease the level of NAD(P)H and the NAD(P)H/NAD(P)⁺ ratio by increasing the pool of acetaldehyde available for reduction to ethanol. This was investigated in nongrowing cells by examining the effects of these additives on whole-cell fluorescence (Fig. 5).

FIG. 5.

Effects of metabolites on whole-cell fluorescence. (A) Effects of acetaldehyde on the xylose-dependent increase in fluorescence (time course); (B) effects of metabolites on the xylose-dependent increase in fluorescence. The values in panel B are percentages of the xylose-dependent increase in the fluorescence of whole cells observed in the presence of both xylose and the additive indicated. Note that a decrease in xylose-dependent fluorescence is interpreted as a decrease in the NAD(P)H level and a decrease in the NAD(P)H/NAD(P)⁺ ratio.

The fluorescence changes in response to additives were immediate and stable, as shown for acetaldehyde in Fig. 5A. Relative fluorescence increased when fermentation was initiated by addition of xylose and decreased immediately upon addition of acetaldehyde (alcohol dehydrogenase) and pyruvate (pyruvate decarboxylase plus alcohol dehydrogenase), which is consistent with expected changes in the oxidation of NADH. The fluorescence of energized cells also decreased immediately upon addition of 2-ketoglutarate and oxaloacetate (Fig. 5B). The apparent decline in NAD(P)H in response to these two TCA pathway intermediates may have been due to reductive amination (oxaloacetate and 2-ketoglutarate). Addition of the corresponding amino acid products, glutamic acid and aspartic acid, did not cause a similar change. When oxaloacetate was added, malate dehydrogenase provided an additional opportunity for NADH oxidation.

Additions of malate, fumarate, succinate, citrate, and isocitrate did not significantly alter whole-cell fluorescence. Together, these data demonstrate that three compounds which increased the growth and fermentation of KO11 in CSL+X medium (acetaldehyde, pyruvate, and 2-ketoglutarate) also decreased the NAD(P)H/NAD(P)⁺ ratio in cells. Compounds which did not decrease this ratio were not beneficial. Oxaloacetate was an exception. Although this compound decreased the NAD(P)H/NAD(P)⁺ ratio, growth and fermentation were retarded. The negative effects of added oxaloacetate may be attributed to induction of pyruvate carboxykinase. Together with phosphoenolpyruvate carboxylase, this enzyme creates a futile cycle for ATP (10, 11). ATP yields are low with xylose, and ATP wasted by this futile cycle may offset any potential benefit from increased oxidation of NADH.

Citrate synthase, a link between NADH and 2-ketoglutarate.

In E. coli (gltA), as in most gram-negative bacteria, citrate synthase is allosterically inhibited by NADH and activated by acetyl-CoA (49). The activity of this enzyme regulates the flow of carbon into the 2-ketoglurate side of the TCA pathway, linking the cellular abundance of NADH and acetyl-CoA to the production of 2-ketoglutarate for biosynthesis (17, 27, 47). This enzyme integrates both beneficial effects of added pyruvate, increased electron acceptors (acetaldehyde) and increased carbon skeletons in the 2-ketoglutarate arm of the TCA pathway (2-ketoglutarate). The allosteric control of this enzyme by NADH could restrict the flow of carbon into the biosynthesis of 2-ketoglutarate and other products. This hypothesis can be tested easily by expressing an NADH-insensitive recombinant citrate synthase gene in KO11.

The primary citrate synthase in gram-positive bacteria is allosterically regulated by ATP and is relatively insensitive to NADH (25). Since an overabundance of ATP is not anticipated during xylose fermentation (42), expression of B. subtilis citZ in KO11 would be expected to increase carbon flow into the oxidizing arm of the TCA pathway. Primers were used to clone the citZ gene (including a ribosomal binding site) into pCR2.1-TOPO to produce pLOI2514. Plasmid pLOI2514 was found to complement a gltA mutation in E. coli W620 on plates containing M9 minimal medium supplemented with glucose and thymine. We also confirmed that citrate synthase activity (0.08 U mg of protein⁻¹) was present in strain W620(pLOI2514) and absent in strain W620 lacking citZ.

Expression of citZ in KO11(pLOI2514) increased growth and ethanol production by approximately 75% (Fig. 6) compared to the control with the vector alone, KO11(pCR2.1-TOPO). The low level of NADH-insensitive citrate synthase produced from pLOI2514 was almost as effective as pyruvate, acetaldehyde, and 2-ketoglutarate additions in stimulating growth. Thus, the allosteric regulation of the native citrate synthase by high NADH levels appears to limit the flow of carbon skeletons into biosynthesis in CSL+X medium.

FIG. 6.

B. subtilis citZ increases the growth and fermentation of KO11 in CSL+X medium. (A) Growth; (B) ethanol production. The error bars indicate standard errors of the means.

DISCUSSION

The rate of ethanol production and ethanol yield are important factors in determining the cost of large-scale fermentation processes. For KO11, both of these are directly related to the extent of growth of the biocatalyst (Fig. 7). In CSL+X medium, more than one-half of the ethanol was produced after cells entered the stationary phase (Fig. 2). By increasing cell densities, fermentation times can be reduced without sacrificing ethanol yield. However, previous studies with KO11 have shown that high levels of complex nutrients are needed for cell growth and rapid ethanol production (31, 54, 55). This apparent requirement for high levels of complex nutrients now appears to reflect a regulatory error in the partitioning of pyruvate skeletons between competing requirements for the oxidation of NADH and biosynthesis. Our study demonstrated that the growth of KO11 was not limited by nutrients, a lack of biosynthetic enzymes, or insufficient ATP from xylose metabolism (0.67 ATP per xylose). During fermentation of 10% xylose, growth was limited by a lack of carbon skeletons for biosynthesis of products derived from 2-ketoglutarate. Growth and ethanol production were increased by addition of pyruvate or 2-ketoglutarate but not by addition of oxaloacetate, malate, or fumarate.

FIG. 7.

Relationship between cell yield and fermentation performance. Results from fermentations with CSL+X medium alone and fermentations with supplements were combined. A computer-generated polynomial was used to determine approximate cell yields. The results of a linear regression analysis are shown for volumetric productivity. The dashed lines indicate the 95% confidence intervals.

The apparent starvation for carbon skeletons to produce 2-ketoglutarate was also alleviated by addition of acetaldehyde, which is consistent with involvement of NADH or NADH/NAD⁺ ratios. The ratio of NADH to NAD⁺ is typically higher during fermentation than during oxidative metabolism (14). A high level of NADH is an allosteric inhibitor of citrate synthase, the first committed step in the production of 2-ketoglutarate and a likely bottleneck for the biosynthesis of many amino acids (47). Addition of acetaldehyde decreased the NADH/NAD⁺ ratio by increasing the pool of electron acceptors, which potentially increased the function of the native citrate synthase in vivo. This hypothesis was confirmed, in part, by using the B. subtilis citZ gene encoding an NADH-insensitive citrate synthase (24, 25). Expression of citZ in KO11 stimulated growth and ethanol production almost twofold and substantially reduced the need to supply high levels of complex nutrients.

The pattern of carbon flow in KO11 is summarized in Fig. 8. Expression of high levels of Z. mobilis pdc and adhB redirects pyruvate away from native fermentation pathways (pyruvate formate-lyase, lactate dehydrogenase) and into ethanol, even in the presence of competing native enzymes (35). Since the K_m of pyruvate decarboxylase for pyruvate is approximately 10% that of the competing enzyme, pyruvate formate-lyase, production of acetyl-CoA is also limited. Integration of the ethanol production genes into the chromosomal pfl gene may further contribute to this problem by reducing the level of pyruvate formate-lyase activity. Although pyruvate dehydrogenase has a K_m for pyruvate that is equal to that of pyruvate decarboxylase, pyruvate dehydrogenase is expressed at low levels during fermentation and is allosterically inhibited by the high levels of NADH present during fermentation (14, 20). In CSL+X medium, a portion of cellular pyruvate was converted to acetyl-CoA by KO11 during the first 24 h, as shown by the accumulation of acetate as a fermentation product. The acetate levels are presumed to be in excess of biosynthetic needs.

FIG. 8.

Carbon flow through central metabolism in KO11 at the pyruvate node. Unless noted otherwise, enzymes are native to E. coli. Enzymes: 1, pyruvate kinase (pykA, pykF); 2, pyruvate formate-lyase (pflB); 3, pyruvate dehydrogenase (aceEF, lpd); 4, phosphotransacetylase (pta); 5, acetate kinase (ackA); 6, alcohol/aldehyde dehydrogenase (adhE); 7, Z. mobilis pyruvate decarboxylase (pdc); 8, Z. mobilis alcohol dehydrogenase II (adhB) and E. coli alcohol/aldehyde dehydrogenase (adhE); 9, lactate dehydrogenase (ldhA); 10, phosphoenolpyruvate carboxylase (ppc); 11, citrate synthase (gltA); 12, aconitase (acn); 13, isocitrate dehydrogenase (icd); 14, glutamate synthase (gltB, gltD); 15, glutamine synthetase (glnA); 16, malate dehydrogenase (mdh); 17, fumarase (fumB); 18, fumarate reductase (frdABC); 19, aspartate transaminase (aspA); 20, aspartase (aspC). The arrows beneath citrate synthase indicate inhibition of activity by NADH and antagonism of NADH inhibition by acetyl-CoA.

Addition of either pyruvate or acetaldehyde dramatically stimulated the growth of KO11 in CSL+X medium. At least three sites of allosteric regulation may contribute to the increase in growth. Both pyruvate dehydrogenase (14, 20) and citrate synthase (17, 48, 49) are allosterically inhibited by NADH. Oxidation of NADH from glycolysis during reduction of acetaldehyde from added pyruvate or added acetaldehyde tends to decrease the NADH/NAD⁺ ratio. This should reduce the allosteric inhibition of pyruvate dehydrogenase and citrate synthase by NADH. Additional acetyl-CoA from pyruvate dehydrogenase should supplement the acetyl-CoA produced by pyruvate formate-lyase and increase the pool of acetyl-CoA, a substrate for citrate synthase and an allosteric antagonist of NADH inhibition of the native citrate synthase (49). Together, these regulatory circuits should feed forward to promote the flow of additional carbon skeletons into 2-ketoglutarate and subsequent products of biosynthesis.