INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Under anaerobic conditions and in the absence of exogenous electron acceptors, Escherichia coli ferments glucose to a mixture of products consisting primarily of acetate, formate, and ethanol, as well as smaller amounts of lactate and succinate (2). In this process, NADH generated during glycolysis is reoxidized through the reduction of organic intermediates derived from glucose. The relative proportions of the end products vary with the strain and growth conditions, but in all cases the distribution of products is balanced so that the reducing equivalents generated are fully consumed (7). Normally, no more than 0.2 mol of succinate is formed per mol of glucose consumed by E. coli (2, 3, 7).

Succinate is of commercial interest as a potential precursor of industrial chemicals. Purified, biologically produced succinic acid can be converted chemically to numerous commercial chemicals ranging from low-volume, high-value products such as malic acid to commodity chemicals such as 1,4-butanediol (18, 34). Two bacteria that naturally produce succinate as their major fermentation product, Anaerobiospirillum succiniciproducens and Actinobacillus succinogenes, generate up to 1 mol of succinate per mol of glucose and have been developed for commercial production of succinic acid (11, 14). The possibility of increasing succinate production in E. coli through genetic engineering has been investigated recently. Expression of plasmid-encoded phosphoenolpyruvate (PEP) carboxylase (12, 20) or pyruvate carboxylase (12) resulted in increased succinate formation, but the highest yield of succinate was less than 0.5 mol per mol of glucose (20). Expression of the malic enzyme in a nonfermenting mutant of E. coli, NZN111, resulted in very slow formation of succinate and a higher yield (29).

Recently, we described a mutant strain of E. coli, designated AFP111, that ferments glucose to an unusual mixture of products consisting of succinate, acetate, and ethanol (9). AFP111 arose by spontaneous chromosomal mutation in strain NZN111, which is unable to ferment glucose due to inactivation of the genes encoding pyruvate:formate lyase and the fermentative lactate dehydrogenase (5). In this study we identified the mutation that restored the ability of NZN111 to grow fermentatively on glucose as a lesion in the ptsG gene. When a null mutation of the ptsG gene was introduced into various strains of E. coli not blocked in the ability to ferment glucose, the resulting strains also produced more succinate.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains, media, and growth conditions. All E. coli strains used in this study (Table 1) were routinely cultured in Luria-Bertani (LB) medium (27) at 37°C. Antibiotics were included as necessary at the following concentrations: carbenicillin, 100 µg per ml; kanamycin, 30 µg per ml; tetracycline, 10 µg per ml; and chloramphenicol, 30 µg per ml. Rich broth contained (per liter) 10 g of tryptone, 5 g of NaCl, and 1 g of yeast extract. Solid media for plates contained 1.5% (wt/vol) Difco Bacto Agar. Minimal medium E was prepared as described by Vogel and Bonner (31).

Genetic methods. Hfr conjugations for mapping of the mutation were performed by using the set of Hfr donor strains constructed in the laboratory of B. Wanner (32) with modifications of methods described previously (21, 33). AFP111 was used as the recipient, and 15- to 20-min mating periods were typically allowed. Exconjugants were selected for resistance to tetracycline (to select for transfer of Tn10 into the recipient) and kanamycin (to select against the donor) and were screened for formation of white colonies on LB medium-glucose/lactose/X-Gal plates, which indicated that transfer of a locus that restored normal catabolite repression had occurred. Exconjugant colonies were purified by restreaking on the same selective medium; well-isolated white colonies were analyzed for fermentation of glucose.

Construction and introduction of an insertionally inactivated ptsG gene. The native ptsG gene of E. coli was cloned by PCR from genomic DNA prepared from W1485 by using primers targeting the N and C termini of the protein, and no additional genomic sequences were amplified. The gene was cloned into the vector pFJ118EH (10) to give pJFptsG. The gene was disrupted by insertion of the kanamycin resistance cassette of pUC-4K (Pharmacia), excised with EcoRI, into the MfeI site of the ptsG gene in pJFptsG to give plasmid pPTSGK. Because NZN111 already included a kanamycin resistance marker, an equivalent strain was constructed by transducing a Tn10-inactivated ldhA gene from strain SE1752 (Table 1) into FMJ123. The physiology of the resulting strain, DC1327, was indistinguishable from that of NZN111. The disrupted ptsG gene was transferred into DC1327 by transforming the cells with pPTSGK, growing the cells for approximately 30 generations in the presence of kanamycin and absence of ampicillin, and then plating the culture on LB agar plates containing glucose and incubating the plates anaerobically. Colonies that were able to grow fermentatively were purified and screened for sensitivity to the two antibiotics. Strain AFP400 was isolated as a stable kanamycin-resistant, ampicillin-sensitive strain that fermented glucose to succinate, acetate, and ethanol. Proper integration of the disrupted ptsG gene was confirmed by PCR. The disrupted gene was amplified from AFP400 DNA by using primers that matched flanking sequences approximately 110 bp outside the coding region of the gene. These sequences were not present in the integration vector. The resulting product was 3.0 kb long, as predicted from the known sequences of ptsG, its flanking regions, and the kanamycin insert. The product was digested with ClaI (site in the kanamycin cassette) and AgeI (site in ptsG), and this generated the fragments expected for insertion of the cassette into the MfeI site of ptsG (1.95 and 1.05 kb for ClaI and 2.3 and 0.7 kb for AgeI).

Analytical methods and enzyme assays. Substrate consumption and product formation during fermentation were quantified by high-performance liquid chromatography using a Bio-Rad Aminex HPX-87H ion-exchange column (7.8 by 300 mm) and a Shimadzu LC-10A chromatography system equipped with UV absorbance and refractive index detectors. Samples of the anaerobic culture were removed anoxically and centrifuged at 7,000 × g for 1 min. Each supernatant was diluted with 2 volumes of 5 mM H₂SO₄, and 200 µl of the diluted sample was injected. The column was eluted isocratically at a rate of 0.5 ml/min with 5 mM H₂SO₄, and data were analyzed with an EZChrom data system (Scientific Software, Inc.). The approximate retention times were as follows: glucose, 11.8 min; mannose, 12.6 min; succinic acid, 15.0 min; lactic acid, 16.5 min; formic acid, 17.9 min; acetic acid, 19.3 min; and ethanol, 27 min. Glucose levels were also monitored enzymatically by using a commercial kit obtained from Stanbio, Inc. Pyruvate levels were determined enzymatically by the lactate dehydrogenase reaction (17).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Physiological characteristics of AFP111. Strain AFP111 arose by spontaneous chromosomal mutation of strain NZN111 that restored the ability to grow fermentatively on glucose (9). The fermentation carried out by AFP111 was well balanced in terms of accounting for both carbon atoms and electrons and generated an unusual mixture of products consisting of approximately 1 mol of succinate, 0.5 mol of acetate, and 0.5 mol of ethanol per mol of glucose consumed (Table 2). AFP111, however, did not grow as well on glucose as its ancestral strain, W1485, did (Fig. 1). Whereas W1485 grew with a generation time of less than 1 h, AFP111 grew with a generation time of approximately 7 h and reached a lower final cell density. After a small amount of growth after inoculation, NZN111 grew with a generation time of more than 25 h (in the interval between 12 and 25 h) (Fig. 1).

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Anaerobic growth of strain AFP111 (), its immediate parent, NZN111 (), and wild-type ancestor W1485 () on LB medium supplemented with 10 g of glucose per liter.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Aerobic growth and consumption of a mixture of glucose and mannose by NZN111 (A) and mutant AFP111 (B). LB medium was supplemented with approximately 2 g of glucose per liter () or 2 g of mannose per liter (). The cell density values (×) are culture OD600 values divided by 10.

Mapping the causative mutation. The causative mutation was mapped by first screening for restoration of normal glucose repression of -galactosidase (see above) and then screening for fermentation of glucose. Exconjugants and transductants were evaluated based on the color of colonies produced on aerobic LB medium-glucose/lactose/X-Gal plates and subsequently were evaluated based on growth on glucose under anaerobic conditions. In every case throughout the mapping experiments, these two traits were found to be linked; blue-green colonies were able to grow fermentatively on glucose, but yellow colonies were not.

Effects of various alleles of ptsG. The mapping results indicated that the causative gene very likely lay between 24.6 and 25.5 min on the chromosome. The ptsG gene lies squarely in this region at 25.0 min and could contribute to the observed phenotypic changes. Its product, protein EIICB^glc, is the glucose-specific permease of the PTS. Mutation of ptsG could affect both the growth rate and the repression of other operons in the presence of glucose (25). To evaluate this possibility, we transferred various mutant and functional forms of ptsG into appropriate strains and observed the physiological effects of their presence (Table 3). Introduction of a mutant allele, ptsG21, into NZN111 eliminated repression of -galactosidase and restored the ability to grow fermentatively on glucose. Analysis of the fermentation products of representative transductants showed that all of them converted glucose to the succinate-acetate-ethanol mixture observed in AFP111. Introduction of a wild-type, plasmid-borne ptsG gene under control of its own promoter (4) into AFP111 eliminated the ability of the organism to ferment glucose (Table 3). These results strongly suggest that mutation of ptsG was responsible for phenotypic changes observed in AFP111.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Aerobic growth and consumption of a mixture of glucose and mannose by AFP400. LB medium was supplemented with approximately 2 g of glucose per liter () or 2 g of mannose per liter (). The cell density values (×) are culture OD600 values divided by 10.

Enzymatic analysis of the EIICB^glc permease activities of strains. The glucose-specific phosphotransferase activities of various strains were assayed by using permeabilized cells grown in the presence of glucose (see Materials and Methods). In this assay, cells are mildly permeabilized to allow exchange of small molecules while retaining intracellular enzymes and are washed to remove endogenous small molecules. In these cells, glucose-dependent conversion of PEP to pyruvate, adjusted for appropriate controls, measures the glucose-dependent phosphotransferase activity and depends on the activity of the EIICB^glc permease, the product of the ptsG gene. Controls lacking PEP or glucose or controls in which cells grown on fructose were used were included to ensure that the activity reported reflected EIICB^glc activity. A comparison of exponentially growing cells of representative strains showed that all of the succinate-producing mutants had approximately 10-fold less EIICB^glc activity than their parental strains (Table 4). The reductions in EIICB^glc activity were consistently associated with the lack of repression of the -galactosidase by glucose. The fructose-dependent phosphotransferase activities of fructose-grown NZN111 and AFP111 were 5.4 and 5.9 nmol/min/OD₆₀₀ unit, respectively, indicating that the mutation in AFP111 did not affect the PTS in general but was specific for the glucose PTS.

Effect of the null ptsG allele in other strains. Strains NZN111 and DC1327 are blocked in normal fermentation of glucose. To evaluate possible effects of ptsG in nonblocked strains, we transduced insertionally inactivated ptsG into ancestral strains of NZN111 and into unrelated E. coli lineages (Table 5). In all cases, inactivation of ptsG shifted the fermentative metabolism strongly toward succinate. Even in the presence of a functional pyruvate:formate lyase, succinate production increased from about 0.2 to about 0.6 to 0.8 mol per mol of glucose in all strains tested. This result occurred not only in K-12 lineages but also in E. coli B. In the absence of a functional pyruvate:formate lyase, an even larger shift occurred. The immediate precursor of NZN111, FMJ123, produces lactic acid almost exclusively. Upon introduction of the ptsG null mutation, the resulting strain, AFP402, produced a mixture of products very similar to that produced by AFP111 but containing more lactate and less ethanol (Tables 2 and 5).

Metabolism of other sugars by NZN111. Ancestral strain W1485 can ferment a number of sugars and sugar derivatives. Anaerobic plate tests revealed that NZN111 was also able to grow fermentatively on sugars other than glucose. The only sugar that failed to support comparable growth of both AFP111 and NZN111 was glucose. Sugars that supported growth of both strains were mannose, lactose, fructose, and trehalose. When tested in liquid medium, NZN111 converted these sugars primarily to succinate, acetate, and ethanol with approximately the same yields as those obtained with glucose (data not shown). The sugar derivatives glucuronate and sorbitol also supported growth of NZN111, giving alternative distributions of products that were consistent with the redox balance required because of the different oxidation states of the substrates. Glucuronate was converted primarily to the more oxidized product acetate (and presumably carbon dioxide) and to smaller amounts of succinate and ethanol, whereas sorbitol generated higher yields of the reduced products succinate and ethanol. Substrates at the same oxidation state as glucose (mannose, lactose, fructose, and trehalose) yielded approximately 1 mol of succinate per mol of glucose.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

E. coli normally carries out mixed-acid fermentation of glucose to give primarily acetate, formate, and ethanol, all of which are derived from pyruvate in a series of reactions initiated by pyruvate:formate lyase (Fig. 4). Succinate, derived via carboxylation and subsequent reduction of PEP, is a minor product. The ancestral precursor of the mutants investigated here produces approximately 0.2 mol of succinate per mol of glucose. The naturally occurring succinate-producing bacterium A. succiniciproducens produces only succinate and acetate from glucose and theoretically can generate 1.3 mol of succinate per mol of glucose (28). In this case, two-thirds of the PEP derived from glucose is converted via the reductive arm of the tricarboxylic acid cycle to succinate, and one-third is oxidized to acetate via pyruvate:formate lyase to establish the necessary redox balance. The results described here indicate that E. coli possesses a latent ability to carry out a similar fermentation when the genes encoding pyruvate:formate lyase (pflB), the fermentative lactate dehydrogenase (ldhA), and EIICB^glc (ptsG) are inactivated. Under these conditions, 1 mol of glucose is converted to approximately 1 mol of succinate, 0.5 mol of acetate, and 0.5 mol of ethanol. The difference between these two fermentations may result from the established coupling of acetate formation and ethanol formation in E. coli (15); mutation of either pta or adhE, the two genes whose products partition acetyl coenzyme A to acetate and ethanol, eliminates the ability to grow fermentatively in wild-type strains. Mutation of adhE in AFP111 similarly eliminated the ability to ferment glucose (Champion, unpublished observation), indicating that the same restraint applies to the partitioning of acetyl coenzyme A in the new succinate fermentation pathway. If ethanol must be formed in equimolar yields with acetate, reducing equivalents that could be used in succinate production are consumed, lowering the theoretical maximum yield of succinate to 1 mol of succinate per mol of glucose.

View larger version (20K): [in this window] [in a new window]   FIG. 4.   Key intermediates, genes, and products relevant to the fermentation of glucose by AFP111. In parental strain NZN111, inactivation of pflB and ldhA eliminated the ability to grow fermentatively on glucose. Mutation of ptsG restored growth and resulted in the formation of succinate, acetate, and ethanol. See text for details. FDP, fructose diphosphate; T3P, triose phosphates; OAA, oxaloacetic acid; Acetyl CoA, acetyl coenzyme A.

The potential of E. coli to generate this alternative distribution of fermentation products is widespread, occurring in different K-12 lineages, in strains that are not blocked in the fermentation of glucose, and in E. coli B (Table 5). In all of these examples, inactivation of the ptsG gene shifts metabolism toward succinate production, resulting in approximately 0.5 to 1 mol of succinate formed per mol of glucose. The differences in succinate yield in part reflect the presence of competing pathways; strains containing pyruvate:formate lyase make less succinate because they continue to produce large amounts of acetate and ethanol, but strain-dependent variation in the relative amounts of the products was observed. The most dramatic shift occurs in strain FMJ123. This strain lacks a functional pflB gene and carries out nearly homolactic fermentation of glucose with a yield approaching the theoretical maximum of 2 mol of lactate per mol of glucose. Inactivation of ptsG almost completely replaces this fermentation with the succinate-acetate-ethanol fermentation first observed in AFP111. Only a small amount of lactate is still produced. In strains which still possess a functional pflB gene, disruption of ptsG results in superimposition of the original pathway (which generates primarily acetate, formate, and ethanol) and the succinate-acetate-ethanol pathway (Table 5).

The evidence presented here establishes that a null mutation of ptsG alters the fermentative metabolism of E. coli and results in increased succinate formation, but the mechanism of this effect is not obvious. The product of ptsG, the glucose-specific PTS permease EIICB^glc, catalyzes efficient transport and phosphorylation of glucose using phosphate derived originally from PEP. Inactivation of EIICB^glc in principle could cause an increase in the PEP pool, favoring formation of succinate (Fig. 4). However, under aerobic conditions, a null mutation of ptsG in a different strain of E. coli had exactly the opposite consequence; the sizes of the PEP pools decreased, and the amounts of the products derived from PEP decreased (6). Alternatively, rapid glucose uptake by a functional glucose PTS in the absence of efficient conversion to products could lead to detrimental accumulation of intermediates. Fructose diphosphate, PEP, and NADH have regulatory roles at both the enzyme level and the gene level (13, 26), and the triose phosphates can give rise to toxic methylglyoxal (23). Perhaps the slower metabolism of the ptsG mutants prevents such metabolic imbalances.

Inactivation of EIICB^glc could also modify metabolism indirectly by influencing the phosphorylation state of the immediate phosphate donor in the PTS, EIIA^glc (25, 26). During normal glucose uptake, EIIA^glc is largely dephosphorylated and represses uptake of other sugars by a process known as inducer exclusion (16). In addition, under these conditions the phosphorylated form of EIIA^glc is depleted, which interrupts a regulatory cascade that activates many operons (26). This loss of activation is known as catabolite repression. Both mechanisms act to reduce expression of the lac operon in the presence of glucose and contribute to the preferential use of glucose in mixtures of sugars. The expression of -galactosidase in the presence of glucose by the ptsG mutants described here reflects the loss of these regulatory controls, as does the simultaneous consumption of glucose and mannose by AFP111 and AFP400 (Fig. 2 and 3). With respect to succinate production, loss of a functional EIICB^glc might similarly allow expression of enzymes normally repressed in the presence of glucose that could be crucial to the fermentation. EIICB^glc has recently been proposed to have regulatory functions itself as well, including a direct effect on the expression of several genes, including its own regulatory gene, and possibly a role as a global sensor of glucose (8, 24, 30).

Regardless of the mechanism of the shift in metabolism, the results reveal that many strains of E. coli have a latent ability to produce good yields of succinate, which provides a foundation for developing E. coli as an organism for producing succinic acid. AFP111 can reproducibly generate up to 50 g of succinate per liter at a scale of 500 liters by using inexpensive agricultural feedstocks (22). Given the widespread distribution of the capacity to make succinate and the numerous tools available for manipulating E. coli genetically, it should be possible to improve succinate production significantly. Identification of a mutation of ptsG as the cause that unmasks this fermentation constitutes the first step in understanding how the novel distribution of products is generated. The homolactic fermenting strain FMJ123 and its derivative with the ptsG null mutation, AFP402, both ferment glucose well, but these organisms make drastically different fermentation products. Thus, these strains provide an ideal isogenic pair for further study.