Acetate accumulation through alternative metabolic pathways in ackA⁻ pta⁻ poxB⁻ triple mutant in E. coli B (BL21)

Abstract

Individual deletions of acs and aceA genes in E. coli B (BL21) showed little difference in the metabolite accumulation patterns but deletion of the ackA gene alone or together with pta showed acetic acid gradually accumulated to 3.1 and 1.7 g/l, respectively, with a minimal extended lag in bacterial growth and a higher pyruvate formation. Single poxB deletion in E. coli B (BL21) or additional poxB deletion in the ackA-pta mutants did not change the acetate accumulation pattern. When the acetate production genes (ackA-pta-poxB) were deleted in E. coli B (BL21) acetate still accumulated. This may be an indication that perhaps acetate is not only a by-product of carbon metabolism; it is possible that acetate plays also a role in other cellular metabolite pathways. It is likely that there are alternative acetate production pathways.

Introduction

Acetate accumulation is one of the major concerns during high cell density culture of E. coli since acetate above 2.4 g/l retards growth and recombinant protein production (Dittrich et al. 2005). Several metabolic pathways in E. coli are directly related to acetate production and utilization: (a) acetate production by AckA-Pta; this pathway contains two enzymes—phosphotransacetylase (pta) that reversibly converts acetyl-CoA; and inorganic phosphate to acetyl phosphate and CoA, and acetate kinase (ackA) that reversibly converts acetyl phosphate and ADP to acetate and ATP (De May et al. 2007). (b) Acetate utilization by acetyl-CoA synthetase (acs) that anabolically converts acetate to acetyl-CoA. The acs is expressed in the stationary phase under the control of the sigma factors RpoS and RpoD and, in order to be fully expressed the cAMP receptor protein (CRP) and the oxygen regulator (Fnr) are also required (Kumari et al. 2000). (c) Acetate production by pyruvate oxidase B (poxB) which catalyzes the oxidative decarboxylation of pyruvate to acetate and CO₂ (Abdel-Hamid et al. 2001). The ackA-pta pathway is more active during the growth phase while the poxB during late growth and stationary phases (Dittrich et al. 2005).

Acetate formation is also affected by the cellular availability of acetyl-CoA (El-Mansi 2005). Acetyl-CoA concentration is affected by the TCA cycle, the glyoxylate shunt and several anaplerotic pathways. Acetyl-CoA accumulation can cause acetate production, while an active glyoxylate shunt and anaplerotic pathways can decrease the acetyl-CoA concentration and consequently decrease acetate production (Noronha et al. 2000; Phue et al. 2005). The glyoxylate shunt operon (aceBAK) contains aceK (isocitrate dehydrogenase kinase/phosphatase), aceA (isocitrate lyase) and aceB (malate synthase) genes. The anaplerotic pathways include the conversion of oxaloacetate to phosphoenolpyruvate through phosphoenolpyruvate carboxykinase (pckA), the conversion of pyruvate to phosphoenolpyruvate through phosphoenolpyruvate synthase (ppsA), the conversion of phosphoenolpyruvate to oxaloacetate through phosphoenopyruvate carboxylase (ppc), and the conversion of malate to pyruvate through NAD-linked malate dehydrogenase (sfcA).

Considerable amount of work has been done on reducing acetate accumulation in E. coli K-12 by both growth strategies and genetic approaches (Shin et al. 2009; Kang et al. 2009; De May et al. 2007). The growth strategies include glucose feeding approaches and the use of less acetate-producing carbon sources such as glycerol or fructose. The genetic approaches include the development of mutant strains with altered acetic acid metabolic flux (Lee 1996). The deletion of acetate producing pathways ackA/pta and poxB were the first targets. When this was accomplished in E. coli K-12 (Dittrich et al. 2005; Yang et al. 1999), the deletion of ackA and/or pta decreased growth rate and acetate accumulation. Other genetic approaches have been based on altering either the central carbon metabolism by altering the phosphotransferase system (PTS), glycolysis or the pyruvate branch point (De May et al. 2007).

Acetate production pattern in E. coli B (BL21) is different from E. coli K-12 (JM109). When the cells grow at high glucose concentrations, E. coli K-12 accumulates acetate up to 11 g/l and its growth rate slows, while E. coli B (BL21) does not accumulate acetate and its growth is not affected. Extensive evaluation of these two strains by NMR/MS and their genomic patterns revealed that E. coli B (BL21) has more active glyoxylate shunt, gluconeogenesis, anaplerotic pathway, and TCA cycle than E. coli K-12 (Noronha et al. 2000; Phue et al. 2005).

A systematic study on acetate metabolism in E. coli B (BL21), which has better performance in high glucose condition, has not been carried out. The purpose of the present study was twofold: first, to try and convert E. coli B (BL21) to a high acetate producer like E. coli K-12 by modifying specific pathways or genes; and second, to try reducing further its acetate production. These two approaches may provide information on specific genes that should be considered for modification in E. coli K-12 and perhaps will assist in converting E. coli K-12 to a low acetate producer. To perform this study, several mutant E. coli B (BL21) strains were constructed and time-course profiles of growth, glucose consumption and acetate and other metabolites production associated genes were monitored during high glucose batch fermentation.

Materials and methods

Bacterial strains

Bacterial strains and phage used in this study are listed in Supplementary Table 1. Primers used for gene disruption are listed in Supplementary Table 2. Gene knock-outs were carried out by recombination using plasmid pKD46 (Datsenko and Wanner 2000). pKD46 is not stably maintained in E. coli BL21 strain. For this reason, genes were disrupted in E. coli MG1655 carrying pKD46 and then transferred to E. coli BL21 by P1 phage transduction. PCR primers for gene disruption were designed to amplify chloramphenicol and kanamycin markers in plasmids pACYC184 and pACYC177 (New England Biolab, Inc. Beverly, MA), respectively, so that pta-ackA and poxB genes could be replaced with the antibiotic markers. PCR products carrying an appropriate antibiotic marker and homologous flanking sequences were electroporated into E. coli MG1655 carrying pKD46 where lambda Red recombinase was fully induced by l-arabinose during culture at 30°C. Two hours after electroporation and incubation at 37°C, cells were selected on LB agar plates containing either 25 μg kanamycin/ml or 10 μg chloamphenicol/ml.

P1 lysates of AJW1939, AJW1707, and AJW1781 strains were used to transduce ackA⁻, aceA⁻, and acs⁻ mutations into E. coli BL21, respectively, as previously described (Miller 1992). All types of mutant strains in this study were confirmed by PCR with primers that can hybridize to antibiotic genes and the upstream or downstream of deleted genes.

Fermentation

E. coli strains were grown in 4 l modified LB medium using a Braun fermentor as previously described (Phue et al. 2005). Samples were centrifuged at 14,000×g for 10 min at 4°C, the supernatant was kept at −20°C for metabolite analysis, and the cell pellets were quickly frozen by dry ice and stored at −80°C for RNA extraction.

Analytical methods

Glucose and lactate in culture supernatants were quantified using a YSI 2700 select biochemistry analyzer (YSI Inc.). Ethanol was measured by GC/MS. Pyruvate was determined on a Bio-Rad Aminex HPX-87H ion exclusion column eluted isocratically with 0.005 M H₂SO₄ at 60°C at 0.6 ml/min. Other organic acids were determined using commercially available enzymatic kits (Darmstadt).

Northern blot analysis

The amplification of DNA fragments of aceA, aceE, acs, ackA, gltA, pckA, pflB, poxB, ppc, ppsA, pta, and sfcA genes were performed using primers listed in Supplementary Table 2 and E. coli B (BL21) genomic DNA as a template. The PCR products were purified and labeled with ³²P using Ready-To-Go DNA Labeled Beads (Amersham Pharmacia Biotech) and were purified by Probe Quant G-50 Micro column (Amersham Pharmacia Biotech).

Total RNA was isolated with the MasterPure RNA Purification Kit (Epicentre Technologies). The concentration and purity of RNA was determined as previously described (Phue et al. 2005). The isolated RNA (5 μg/well) was separated using a 1% agarose/formaldehyde denaturing gel at 75 V. The gels were blotted on Nytran Super Charge membrane (11 × 14 cm) (Schleicher and Schuell) at room temperature in 20× SSC. The membranes were fixed by UV-induced cross-linking. The hybridization with ³²P-labeled DNA probes was performed with Quickhyb solution (Stratagene) as recommended by the manufacturer. Northern blots were quantified by phosphor image scanning.

Results

Growth, glucose consumption and acetate accumulation patterns in E. coli mutant strains

Growth kinetics, glucose consumption and acetate production profile of the different mutant strains are shown in Figs. 1, 2 and 3. The gene deletions did not affect or minimally affected the growth rate, the final cell yield (Fig. 1) and the glucose consumption pattern (Fig. 2). Both the individual gene-deleted strains (poxB, ackA, pta, acs, and aceA) and the double and triple mutant strains (ackA⁻/pta⁻, and ackA⁻/pta⁻/poxB⁻) reached final OD (600 nm) of around 70 and growth rate of around 0.3/h. But, the ackA related mutant’s strains: the ackA⁻, ackA⁻/pta⁻, and ackA⁻/pta⁻/poxB) showed slightly longer lag period and somewhat lower glucose consumption than the other tested strains.

Fig. 1.

Cell growth curves during batch fermentations of E. coli B (BL21) and mutant strains at high glucose condition. (Filled squares) E. coli BL21 wild type, (open squares) poxB⁻, (filled triangles) ackA⁻, (open triangles) aceA⁻, (filled circles) acs⁻, (open circles) ackA⁻/pta⁻, (filled diamonds) ackA⁻/pta⁻/poxB⁻

Fig. 2.

Glucose consumption during batch fermentations of E. coli BL21 and mutant strains at high glucose condition. (Filled squares) E. coli B (BL21) wild type, (open squares) poxB⁻, (filled triangles) ackA⁻, (open triangles) aceA⁻, (filled circles) acs⁻, (open circles) ackA⁻/pta⁻, (filled diamonds) ackA⁻/pta⁻/poxB⁻

Fig. 3.

Acetic acid production during batch fermentations of E. coli B (BL21) and mutant strains at high glucose condition. (Filled squares) E. coli BL21 wild type, (open squares) poxB⁻, (filled triangles) ackA⁻, (open triangles) aceA⁻, (filled circles) acs⁻, (open circles) ackA⁻/pta⁻, (filled diamonds) ackA⁻/pta⁻/poxB⁻

A larger difference was observed in the acetate accumulation patterns of the various mutant strains, as shown in Fig. 3. Acetate accumulation profiles of the aceA⁻, acs⁻ and poxB⁻ mutant strains were similar to the parent E. coli B (BL21) strain; its concentration reached 3.5 g/l before it decreased. In the single mutant ackA⁻ strain, acetate increased to g/l and stayed at this value until the end of the growth. In the double mutant (ackA⁻/pta⁻) and in the triple mutant (ackA⁻/pta⁻/poxB⁻) acetate production reached 1.7 g/l. In summary, ackA deletion caused a slight increase in the lag phase associated with slight reduction in the glucose consumption rate and a decrease in acetate accumulation. A dual ackA and pta deletion had a greater effect on acetate accumulation; the added deletion of poxB did not have effect on the acetate accumulation pattern.

Metabolite accumulation

The effect of gene deletion on the production and accumulation of citrate, succinate, malate, lactate, ethanol and pyruvate was evaluated during growth. The accumulation patterns of lactate, malate, succinate and citrate in all the mutant strains were similar to the parent E. coli BL21 strain; the gene deletion did not affect the accumulation (data not shown). There were some differences compared to the parent strain when the concentrations of pyruvate and ethanol were measured (Fig. 4). In the ackA⁻ mutant strain, pyruvate reached 0.9 g/l before disappearing. The pattern in the double and the triple mutant strains (ackA⁻/pta⁻, and ackA⁻/pta⁻/poxB⁻) was similar but pyruvate reached 0.6 and 0.4 g/l respectively, while there was no accumulation at all in the parent E. coli BL21 strain. Ethanol was highest in the acs⁻ and the aceA⁻ mutant strains at the late growth phase (Fig. 4).

Fig. 4.

Pyruvate and ethanol accumulation patterns of E. coli B (BL21) and its mutants strains during high glucose batch fermentations. (Filled squares) E. coli BL21 wild type, (open squares) poxB⁻, (filled triangles) ackA⁻, (open triangles) aceA⁻, (filled circles) acs⁻, (open circles) ackA⁻/pta⁻, (filled diamonds) ackA⁻/pta⁻/poxB⁻

Time course northern blot analysis

Several mutant cells (ackA⁻, ackA⁻/pta⁻, and ackA⁻/ pta⁻/poxB⁻) showed a lag in their growth. We, therefore, considered overall transcription patterns with cellular growth of mutants by comparing normalized median values (Table 1). Transcriptions of 11 genes related to acetate metabolism were monitored and analyzed. In the poxB⁻ strain, acetate production-associated genes (ackA and pta) were up-regulated and the transcriptions of acetate utilization-associated genes (acs and aceA) were slightly affected. In the ackA⁻ strain, the pta gene was affected due to polar effect of inactivation of upstream ackA, and acetate utilization-associated genes were substantially up-regulated. The ackA⁻/pta⁻ strain showed higher expression of poxB and aceE genes and down regulation of most other genes (Table 1). Further, deletion of the poxB gene in the ackA⁻/pta⁻ strain resulted in down-regulation of acs and aceA genes. Both the acs and aceA mutants showed almost similar gene expression patterns when compared with wild type E. coli B (BL21).

Table 1.

The summary of gene expression associated in acetate metabolism at log phase

Function	Gene name	Gene product	Normalized median ratio (mutant strain/BL21)
			poxB−/BL21	ackA−/BL21	ackA− pta−/BL21	ackA− pta− poxB−/BL21	acs−/BL21	aceA−/BL21
Acetate production	ackA	Acetate kinase	1.54	0	0	0	0.73	0.74
	pta	Phosphotransacetylase	1.28	0.31	0	0	0.66	0.86
	poxB	Pyruvate oxidase B	0	1.13	1.89	0	1.27	1.42
Acetate utilization	acs	Acetyl-CoA synthetase	0.68	1.67	0.95	0.22	0	0.5
	aceA	Isocitrate lyase	0.90	1.58	0.97	0.65	0.84	0
Acetyl-CoA production	aceE	Pyruvate dehydrogenase	1.23	0.85	2.30	2.00	1.14	1.30
Gluconeogenesis	pckA	Phosphoenolpyruvate carboxykinase	1.28	0.97	0.85	0.89	0.54	0.61
	ppsA	Phosphoenolpyruvate synthase	1.16	0.91	0.68	0.48	0.79	0.79
Anaplerotic pathway	ppc	Phosphoenolpyruvate carboxylase	1.31	1.08	0.99	0.70	0.54	0.58
	sfcA	NAD-linked malate dehydrogenase	1.03	1.28	0.93	0.83	0.78	0.82
TCA cycle	gltA	Citrate synthase	1.15	1.45	0.83	0.65	1.42	1.15

Median values were gained from the data of signal intensity of each gene during log phase (n = 4). Each median value was normalized with respect to the value of E. coli BL21 for that particular gene

Discussion

A great deal of work has been done on attempting to decrease acetate excretion from E. coli K-12 strains by modifying different acetate production and consumption pathways. However, no attempt has been made to investigate how these pathways affect the acetate excretion pattern and concentration in E. coli B (BL21), which is known to be a low acetate-producing strain when growing to high densities on glucose. Modifying acetate-producing and acetate-consuming genes in E. coli B (BL21) may offer some answers to the two following questions: a) Is it possible to convert E. coli B (BL21) to behave like E. coli K-12 when the genes responsible for acetate consumption (the glyoxylate cycle gene isocitrate lyase and the acetate uptake gene acetyl CoA synthetase) are deleted? b) Is it possible to decrease further the acetate excretion if the genes responsible for acetate production (the phosphotransacetylase, acetate kinase, and pyruvate oxidase B) are deleted? The results showed that the deletion of the acetate utilization genes did not affect the growth nor the acetate excretion levels and pattern of E. coli B (BL21) when growing at high glucose; and that the deletion of the genes responsible for acetate production had slight effects on the lag period and the glucose consumption, but more significantly, decreased further the acetate excretion level and modified its pattern.

The glyoxylate shunt is not active in E. coli K-12 (JM109) but is active in E. coli B (BL21) and, therefore, is likely responsible for the low acetate production. When E. coli B (BL21) was grown at a high glucose concentration, acetyl CoA synthetase transcription was higher than in E. coli K-12 (JM109) (Phue et al. 2005). The deletion of acs and aceA did not affect the acetate excretion level and profile. This seems to contradict previous results indicating that an active glyoxylate shunt and acetyl CoA synthetase are responsible for the lower acetate concentration. But, at the same time, it is also a confirmation of the previous observation that in E. coli B (BL21) there are several pathways that keep the acetate concentration low (Phue et al. 2005). In addition to the glyoxylate shunt and acetyl CoA synthetase, the gluconeogensis pathway is active and the TCA cycle has a higher flux. When the glyoxylate shunt is not functioning, another pathway probably takes its place. This may explain why it was impossible to increase the acetate accumulation by eliminating the glyoxylate shunt, or the Acs activity. An active gluconeogensis pathway, higher flux through the TCA cycle or possible reverse Pta-AckA activity can compensate for the loss of the knocked-out pathways. It is possible that an operational glyoxylate shunt in E. coli B (BL21) is not essential for reducing acetate concentration as we previously suggested. The behavior of the E. coli B (BL21) strain is another indication that E. coli K-12 is limited in its acetate consumption ability; not only it does not contain an operational glyoxylate shunt, as was thought, it does not have other mechanisms that exist in E. coli B (BL21).

Concerning the concentration of metabolites, the deletion of the acetate-consuming genes acs and aceA was associated with higher ethanol concentration at the late log phase (Fig. 4); this might be due to the conversion of surplus acetyl-CoA to ethanol. The deletion of the acetate producing genes ackA and pta had an effect on the acetate and pyruvate levels and production patterns. Acetate in E. coli B (BL21) reached 3.5 g/l before it declined toward the end of growth (Fig. 3). During the growth of the ackA/pta double deletion mutant, acetate gradually increased to 1.7 g/l and during the growth of the ackA⁻ single mutant, it reached 3 g/l; in both mutants the concentration did not come down as it did in the parental E. coli B (BL21) strain. The increase in acetate was also followed by an increase in the pyruvate concentration (0.4 g/l in the ackA⁻ single deletion mutant, 0.6 g/l in the ackA⁻/pta⁻ double deletion mutant and 0.9 g/l in the ackA⁻/pta⁻/poxB⁻ triple deletion mutant) but, unlike the acetate, the pyruvate concentration went down towards the end of the growth (Fig. 4).

With respect to acetate formation, the ackA⁻ mutant produced acetate almost twice as much as the ackA⁻/pta⁻ mutant (Fig. 3). Pta can convert acetyl-CoA to acetyl-phosphate which could be subsequently be used as a phosphate donor, resulting in acetic acid formation even in the absence of acetate kinase. The ackA⁻ and ackA⁻/pta⁻ mutants did not decrease the acetate until the end of the fermentation, while other mutants decreased acetate from the late exponential phase (Fig. 3).

As for the role of the PoxB, pyruvate oxidase plays a significant role in the aerobic growth of E. coli K-12 strain W3110 (Abdel-Hamid et al. 2001). Both Pta-AckA and PoxB pathways are responsible for acetate production in E. coli K-12 (MG1655). The Pta-AckA pathway is thus operational during the growth phase and the PoxB pathway is operational during the stationary phase (Dittrich et al. 2005). However, in the present study, there was no significant difference in specific growth rates and the patterns of acetate accumulation between E. coli B (BL21) poxB⁻ and E. coli B (BL21) strains (Fig. 1 and 3). Pyruvate concentration did not increase during the fermentation of the E. coli B (BL21) poxB⁻ cells (Fig. 4). Also, acetate profile was not affected by an additional a poxB mutation in E. coli B (BL21) ackA⁻/pta⁻ strain. We suggest that poxB transcription is a metabolic burden on the E. coli K-12 strain, but does not affect the E. coli B (BL21) strain at standard growth conditions. Therefore, we conclude that the pyruvate oxidase pathway does not play significant role in the acetate production in the aerobic fermentation of E. coli B (BL21) at high glucose condition.

In spite of the deletion of ackA, pta and poxB in E. coli B (BL21), acetate accumulated up to 1.7 g/l (Fig. 3). This is an indication that acetate is not only a by-product of carbon metabolism; it also plays a role in cellular metabolism. Acetic acid may be formed through anabolic pathways in addition to the central carbon metabolism. Several metabolic reactions related to acetate formation such as N-acetylglucosamine-6-phosphate deacetylase (NagA), UDP-3-O-acyl-N-acetylglucosamine deacetylase (LpxC), acetylornithine deacetylase (ArgE), cysteine synthases (CysM and CysK), and acetoacetyl-CoA transferases (AtoA and AtoD) might contribute to acetic acid accumulation. This suggests that acetic acid accumulation might come from both central carbon metabolism-related acetate fermentation and/or alternative anabolic pathways.