atp Mutants of Escherichia coli Fail To Grow on Succinate Due to a Transport Deficiency

Abstract

Escherichia coli atp mutants, which lack a functional H⁺-ATPase complex, are capable of growth on glucose but not on succinate or other C₄-dicarboxylates (Suc⁻ phenotype). Suc⁺ revertants of an atp deletion strain were isolated which were capable of growth on succinate even though they lack the entire H⁺-ATPase complex. Complementation in trans with the yhiF gene suppressed the growth of the Suc⁺ mutants on succinate, which implicates the yhiF gene product in the regulation of C₄-dicarboxylate metabolism. Indeed, when the E. coli C₄-dicarboxylate transporter (encoded by the dctA gene) was expressed in trans, the Suc⁻ phenotype of the atp deletion strain reverted to Suc⁺, which shows that the reason why the E. coli atp mutant is unable to grow aerobically on C₄-dicarboxylates is insufficient transport capacity for these substrates.

The membrane-bound H⁺-ATPase plays a central role in free energy transduction in Escherichia coli. Under aerobic conditions, the H⁺-ATPase catalyzes the phosphorylation of ADP to ATP by use of proton motive force; under fermentative conditions, it energizes the inner membrane by catalyzing the extrusion of protons at the expense of ATP hydrolysis (7).

The H⁺-ATPase complex in E. coli is encoded by nine genes, located in the atp operon and transcribed into a polycistronic messenger (7, 21, 22). Deletion of the entire atp operon results in an E. coli mutant that is completely devoid of the H⁺-ATPase and that therefore lacks any of its associated activities (10). Such strains have to rely solely on substrate-level phosphorylation to produce ATP but grow relatively quickly under aerobic conditions when supplemented with glucose, although the growth rate and growth yield are somewhat decreased compared with those of the wild-type strain. Under these conditions, the respiration rate is increased compared with a normal E. coli strain, and it was suggested that the atp deletion strain supported uncoupled respiration in order to be able to profit from the increased rate of substrate-level phosphorylation (10).

The C₄-dicarboxylates, succinate, fumarate, and malate, sustain growth of wild-type E. coli strains under aerobic conditions. When these substrates are completely dissimilated to carbon dioxide via the tricarboxylic acid (TCA) cycle, ATP is produced mainly via oxidative phosphorylation, while only one ATP arises in the TCA cycle via substrate-level phosphorylation. It is a well-established fact that atp mutants of E. coli are unable to grow on nonfermentable C₄-dicarboxylates. Indeed, the Suc⁻ phenotype of atp mutants has traditionally been used to distinguish an atp mutant from a wild-type strain (3). There is, however, a difference between a substrate being nonfermentable, i.e., unable to support growth under anaerobic conditions, and a substrate which does not support the aerobic growth of an atp deletion mutant. This is because the aerobic atp mutant has the option to respire away any surplus of reducing equivalents that may be formed in catabolism, whereas the anaerobic E. coli cell must rely on the formation of reduced byproducts to get rid of any surplus of reducing equivalents. The atp deletion mutants do indeed respire away their excess of redox equivalents produced in the catabolic reactions, even though these mutants are unable to use the generated proton motive force for driving ATP synthesis by the H⁺-ATPase (10).

Therefore, the fact that atp mutants do not grow on succinate or malate in the first place is somewhat more surprising: malate can be converted into pyruvate, which can in turn be converted either through the TCA cycle to carbon dioxide, yielding one ATP, or to acetate and ATP. In fact, pyruvate is quite a good substrate for growth of the atp deletion mutant.

In this paper we describe the isolation and characterization of mutants of an E. coli atp deletion strain that are capable of growth on the nonfermentable C₄-dicarboxylates. We demonstrate that the expression of the dctA gene, the structural gene encoding the transporter for C₄-dicarboxylates, enables an atp deletion strain to grow on the C₄-dicarboxylates. Furthermore, our data suggest that the product of the yhiF gene may act as a transcriptional regulator of the dctA gene.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The strains and plasmids used in this study are listed in Table 1. The E. coli K-12 strain BOE270 is highly competent with respect to transformation and was derived from strain MT102, which in turn is an hsdR derivative of strain MC1000 (4). BOE270 was used for cloning purposes and for propagation of plasmid DNA in E. coli. Plasmid pFH2106 is a pBR322-derived expression vector that harbors a synthetic lacUV5 promoter, two lacO operator sites, a poly-linker region, and a strong transcriptional terminator. Furthermore, it carries the lacI gene, encoding the lac repressor protein that binds to both lacO operators, conferring a tight uninduced repression of the cloned gene of interest in addition to the bla gene, conferring ampicillin resistance to the transformants. pFH2106 was kindly provided by F. G. Hansen (Technical University of Denmark) and will be described in detail elsewhere. pUN121 (18) was used for preparing gene libraries of genomic DNA.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Genotype or description	Reference
N43	ara-14 Δlac-85 acrA1 supE44 galK2 rpsL197 malA1 xyl-5 mtc-1	17
LM1237	F+asnB32 thi-1 relA1 spoT1	10
LM2800	LM1237, but ΔatpIBEFHAGDC750	10
LM3115	LM1237, but ΔatpIBEFHAGDC750 lacUV5 lacY	11
LM3305	Spontaneous Suc+ mutant of LM2800	This study
LM3559	LM2800, but yhiD::mini-Tn5	This study
BOE270	MC1000, but hsdR highly competent for Ca2+-competent cells	This study
MC1000	araD139 Δ(ara-leu)7697 galU galK Δ(lacIPOZY)X74 rpsL thi-1	4
pOMC3694	pBR322, carrying a mini-Tn5 transposon plus flanking DNA cloned from LM3559 chromosomal DNA	This study
pFH2106	Expression vector derived from pBR322, carrying the lacUV5 promoter, the lacI gene, and a multiple cloning site	F. G. Hansen, personal communication
pLAC-yhiD	pFH2106, carrying the yhiD gene from E. coli	This study
pLAC-yhiF	pFH2106, carrying the yhiF gene from E. coli	This study
pLAC-dctA	pFH2106, carrying the dctA gene from pKAT204	This study
pKAT204	pBR322, carrying the dctA gene from E. coli	1

Growth of bacteria.

Luria-Bertani broth (LB) (15) was used as a rich medium, supplemented with antibiotics as required for the cloning experiments. Complementation tests were performed on AB minimal medium (5), pH 7.0, supplemented with thiamine (2.5 mg/liter) and the indicated carbon source. Plates were incubated at 37°C and contained 2% agarose in place of the usual (impure) agar.

Enzymes.

Restriction enzymes and T4 DNA ligase were obtained from and used as recommended by Pharmacia and New England Biolabs.

Oligonucleotides.

Oligonucleotides were obtained from Hobolth DNA synthesis (Hillerød, Denmark).

PCR amplification.

A 1 μM concentration of each primer was combined with approximately 30 ng of genomic DNA isolated from LM1237 in a 100-μl PCR mixture. Thirty cycles, each consisting of 30 s of denaturation at 94°C, 30 s of annealing at 58°C, and 60 s of elongation at 72°C, were carried out with the AmpliTaq DNA polymerase, obtained from and used as recommended by Perkin Elmer.

Cloning of the yhiD and yhiF genes into pFH2106.

Primers complementary to DNA sequences upstream and downstream of the yhiD gene and the yhiF gene were designed on the basis of the genomic sequence of E. coli (2) in order to amplify the two genes individually, including their ribosome binding site, from chromosomal DNA isolated from strain LM1237. The primers used were 5′-GCTCTAGACTTGCCGAATTAATGAGGTGC and 5′-CGGAATTCGTGTGAATTTCAGGCTTACGG for amplification of the yhiD gene and 5′-GCTCTAGAGTCCTGTTAATTACCTTTGGC and 5′-CGGAATTCGTCGATAGAAGACCTGTTGCG for amplification of the yhiF gene. In both cases the forward primer was extended at the 5′ end with an EcoRI site and the reverse primer was extended with an XbaI site in order to allow for proper insertion into the multiple cloning site of pFH2106. The fragments were ligated and transformed into E. coli by standard ligation and transformation procedures (15) and were plated with selection for ampicillin resistance, resulting in the plasmids pLAC-yhiD and pLAC-yhiF, in which the expression of the yhiD and yhiF genes has been placed under the control of the lacUV5 promoter.

Cloning of the E. coli dctA gene in the expression vector pFH2106.

Plasmid pKAT204 carries a 2.5-kb PvuI fragment of E. coli chromosomal DNA encoding the E. coli dctA gene (1). pKAT204 was digested with KpnI and FspI, resulting in a 1.4-kb fragment which harbors the intact dctA gene without its native promoter. This fragment was then cloned into pUC19, digested with KpnI and HincII, resulting in the plasmid pUC-dctA. Subsequently, the dctA gene was cut out with HindIII and SacI and inserted into pFH2106, which was also digested with HindIII and SacI. This resulted in the plasmid pLAC-dctA, in which the expression of the dctA gene has been placed under the control of the lacUV5 promoter.

Preparation of gene libraries.

A preparation of chromosomal DNA from strain CSH11 (16) was digested with EcoRI, BamHI, or HindIII. The resulting fragments were then cloned into the EcoRI, BclI, or HindIII sites, respectively, all located within the repressor cI gene of plasmid pUN121 carrying a tetracycline resistance gene and an ampicillin resistance gene. Successful insertions in pUN121 were selected after transformation into strain LM1237 by growth on LB plates containing ampicillin (100 μg/ml) and tetracycline (4 μg/ml). Each of the three gene libraries consisted of more than 10,000 transformants.

Transposon mutagenesis.

A library of transposon-induced mutations was prepared in strain N43 (17). The method uses a Tn5-derived mini-transposon that specifies resistance to kanamycin as a selection marker (6); the transposon is located on a suicide delivery plasmid, pUT (9), that provides the transposase gene in cis orientation, but external to the mobile element, so as to generate stable insertion mutants after transposition. The library of transposon insertions was transferred into the recipient strain LM2800 by P1 transduction.

P1 transduction.

Generalized P1 transduction was carried out by the standard protocol of Miller (16).

RESULTS AND DISCUSSION

Isolation of a spontaneous mutant of an E. coli atp deletion strain that grows on C₄-dicarboxylates.

The E. coli atp deletion strain LM2800 is unable to grow on minimal medium supplemented with succinate as the sole carbon and energy source. However, when this strain was incubated on this medium, colonies began to appear after one week of incubation at 37°C and turned out to be mutants able to grow on succinate (Suc⁺ phenotype), although at a much lower growth rate than an atp⁺ strain. One of these colonies was restreaked to obtain a pure culture (LM3305) and used to investigate what kind of mutation had occurred to enable the atp deletion strain to grow on succinate.

In principle, the Suc⁺ mutants could have activated an enzyme activity that would somehow allow these strains to benefit from oxidative phosphorylation in the absence of the H⁺-ATPase complex (although this is not very likely in view of the complexity of this huge enzyme complex). If this was the explanation, then the Suc⁺ mutants should grow also with acetate as the sole energy source. Table 2 shows the relative aerobic growth of LM3305 (Suc⁺) and LM2800 (Suc⁻) on agarose plates with various substrates. The growth of the two strains on the glycolytic substrates glucose, pyruvate, and lactate is similar, whereas neither strain could grow on acetate. These data show that the Suc⁺ phenotype does not result from oxidative phosphorylation being reestablished. The Suc⁺ mutant grew on all the C₄-dicarboxylates tested, whereas the Suc⁻ mutant did not. Thus, the Suc⁺ phenotype includes growth on all of the C₄-dicarboxylates that can easily serve as carbon and energy sources for aerobic growth of wild-type E. coli strains. Interestingly, the data revealed quantitative differences in the growth of the Suc⁺ mutant on the various substrates: strain LM3305 grew on succinate but grew even better on the two other C₄-dicarboxylates, fumarate and malate.

TABLE 2.

Growth of an atp deletion mutant, a Suc⁺ mutant, and the wild-type strain of E. coli on solid medium and various carbon and energy sources

Strain	Relevant genotype or phenotype	Colony diama (mm) on:
		Glycolytic substratesc				TCA substratesc
		Glucose	Pyruvate	Lactate	Acetate	Succinate	Fumarate	Malate
LM1237	atp+Suc+	1	1.5	0.7	0.7	0.7	0.4	0.8
LM2800	atp Suc−	1	1	0.5	—b	—	—	—
LM3305	atp Suc+	1	1	0.5	—	0.1	0.2	0.3

^aDiameter of single colonies, measured with a microscope and scored after 4 days of incubation at 37°C. The standard error on the determination of colony sizes is less than 15%. 

^b—, colony diameter smaller than 0.03 mm. 

^cCarbon sources: pyruvate, lactate, acetate, succinate, fumarate, and malate (5 g/liter) and glucose (2 g/liter). 

The metabolism of malate will generate one additional molecule of NADH compared with the metabolism of pyruvate, but it is not likely that this excess of redox equivalents makes up the whole difference between those two substrates, since lactate, which produces a similar amount of redox energy, does support the growth of the atp deletion mutant (although at a somewhat lower growth rate compared with pyruvate).

The growth rate of LM3305, in liquid minimal medium supplemented with succinate as the sole carbon and energy source, was 33% relative to the wild-type (atp⁺) control strain, while the yield of biomass on succinate was 29% of the wild-type yield. The rate of succinate consumption and the rate of oxygen consumption were slightly increased, to 115% and 132%, respectively, compared with the wild type, and the Suc⁺ mutant excreted 15% of the succinate input as acetate into the growth medium, which amounts to more than three times as much as the amount excreted by the wild-type (atp⁺) strain.

Isolation of a transposon-induced Suc⁺ mutant.

The relatively high frequency at which the spontaneous Suc⁺ mutant occurred indicated that the Suc⁺ phenotype could be the result of a gene inactivation; it should be possible, then, to obtain the Suc⁺ mutant through transposon mutagenesis. Indeed, the screening of a mini-Tn5 transposon library in strain LM2800 resulted in a transposon-induced Suc⁺ mutant, LM3559. The phenotype of transposon-induced Suc⁺ mutant LM3559 was similar to that of the spontaneous mutant, LM3305, except that LM3305 yielded somewhat larger colonies on plates supplemented with succinate.

Mapping the mutation that leads to the Suc⁺ phenotype.

First, in order to ascertain that the Suc⁺ phenotype of LM3559 was not due to multiple mutations, we performed generalized transduction of the Suc⁺ mutation with phage P1. Indeed, it was possible to transfer the Suc⁺ phenotype together with kanamycin resistance (encoded by the transposon) from LM3559 to LM2800, suggesting that only a single mutation (or two closely linked mutations) is involved in the Suc⁺ phenotype of LM3559.

We then tried, unsuccessfully, to convert the Suc⁺ phenotype of LM3559 to a Suc⁻ phenotype by complementation in trans with libraries of wild-type E. coli chromosomal genes. Also unsuccessful was the reverse complementation test, i.e., conversion of the atp deletion strain, LM2800 (Suc⁻), into a Suc⁺ strain by complementation in trans with E. coli genomic libraries.

The next part of our strategy was to determine the insertion point of the mini-Tn5 transposon in the E. coli chromosome in strain LM3559. Chromosomal DNA from strain LM3559 was digested with EcoRI, which will release a DNA fragment carrying the mini-transposon plus one of the regions flanking the insertion point on the E. coli chromosome. The DNA fragments were cloned into the EcoRI site on pBR322 with selection for kanamycin resistance encoded by the transposon. This resulted in a plasmid, pSUC-Tn5, carrying an insert of 1.8 kb. DNA sequencing revealed the insertion point of the mini-Tn5 transposon in strain LM3559: the transposon is integrated in the C-terminal part of the yhiD gene, in the slp-hdeB intergenic region (section 317, complement 3652655 to 3653302 bp on the E. coli chromosome [2]). The function of the 23.2-kDa polypeptide encoded by the yhiD gene is unknown, but the polypeptide shows a relatively weak homology to the MgtC helper proteins involved in high-affinity Mg²⁺ transport in other bacteria (19).

Complementation with the yhiD gene in trans in the Suc⁺ mutants does not restore the Suc⁻ phenotype.

The insertion of the transposon probably leads to inactivation of the yhiD gene product, and it was therefore of interest to see if we could suppress the Suc⁺ phenotype by complementation with the wild-type yhiD gene in trans. We therefore cloned the yhiD gene into an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible expression vector, resulting in plasmid pLAC-yhiD. The growth of the Suc⁺ mutants, LM3305 and LM3559, harboring pLAC-yhiD was then compared on succinate plates in the presence and absence of IPTG, but the plasmid had no suppressive effect on growth on succinate medium.

Complementation with the yhiF gene in trans in the Suc⁺ mutants restores the Suc⁻ phenotype.

The negative outcome of the yhiD complementation test (see above) prompted us to look more closely at the genes flanking the yhiD gene on the E. coli chromosome. The open reading frames next to the yhiD gene, yhiF and hdeB, encode a hypothetical transcriptional regulator and a protein with unknown function, respectively. The yhiD and yhiF genes are convergently transcribed, and the transposon insertion in the yhiD gene in strain LM3559 may then somehow affect the expression of the yhiF gene. Among the proteins which are homologous to the yhiF gene product was DctR, the transcriptional regulator of dicarboxylate transport in Rhodobacter capsulatus (8).

We therefore cloned the yhiF gene into the expression vector pFH2106, yielding plasmid pLAC-yhiF, and tested for suppression of the Suc⁺ phenotype in strain LM3305 and strain LM3559. A typical experiment is shown for LM3305 in Table 3. Clearly, a strong decrease in colony size was observed on the IPTG-containing plates. With malate as the substrate a similar result was obtained, but with pyruvate no suppression of growth occurred on IPTG-containing plates. We therefore conclude that expression of the yhiF gene in trans can suppress the Suc⁺ phenotype.

TABLE 3.

Complementation with the yhiF gene in trans in the Suc⁺ mutant LM3305

Strain	Plasmid	Colony diama (mm)		Carbon sourcec
		−IPTG	+IPTGb
LM3305	pFH2106	0.12	0.12	Malate
LM3305	pLAC-yhiF	0.12	0.03	Malate
LM3305	pFH2106	0.06	0.06	Succinate
LM3305	pLAC-yhiF	0.06	0.01	Succinate

^aDiameter of single colonies, measured with a microscope and scored after 4 days of incubation at 37°C. The standard error on the determination of colony sizes is less than 15%. 

^b100 μM IPTG. 

^cCarbon sources at 15 mM. 

The suppression of the Suc⁺ phenotype found upon introduction of the wild-type yhiF gene in trans in the Suc⁺ mutants strongly suggests that this gene is no longer expressed in strains with the Suc⁺ phenotype. But it is unclear how the insertion of the mini-Tn5 transposon in the yhiD gene might have affected the expression of the adjacent yhiF gene. One possibility is that the insertion of the transposon affects the degree of negative supercoiling in the vicinity of the yhiF gene and thereby the expression of this gene.

Recently, it was demonstrated that changes in the cellular energy state (the ATP/ADP ratio) do affect the level of DNA supercoiling in vivo (13, 20), and with succinate as the substrate it is conceivable that the intracellular ATP/ADP ratio reaches such a low level in the atp deletion strain (12) that the supercoiling in the yhiF locus is affected. One could then speculate that the following scenario is taking place: the expression of the yhiF gene is enhanced by the low level of negative DNA supercoiling in the atp mutant, and the yhiF gene product then represses the expression of the dicarboxylate transporter and prevents the atp mutant from growing on these substrates.

Complementation with the dctA gene in trans transforms the atp deletion mutant to a Suc⁺ strain.

The fact that the expression of the yhiF gene from the expression vector inhibited the growth of the Suc⁺ mutants on C₄-dicarboxylates indicated that the yhiF gene product might function as a negative transcriptional regulator involved in C₄-dicarboxylate metabolism. Furthermore, since the atp deletion mutant grows well on pyruvate but not on malate plates, it is likely that growth of the atp deletion mutant on malate is limited either by transport of the C₄-dicarboxylates into the cell or by the conversion of malate into pyruvate. Together, these indications prompted us to analyze how the expression of the dctA gene (encoding the C₄-dicarboxylate transporter) might affect the growth of the atp deletion strain, LM2800. For this purpose, we inserted the dctA gene into the expression vector pFH2106, yielding pLAC-dctA (see Materials and Methods). As was noticed by Baker et al. (1), massive overexpression of the C₄-dicarboxylate transporter is detrimental to the cells. This was also the case in our experimental system, and we therefore also used another E. coli atp deletion mutant, LM3115, lacking the lactose carrier (lacY) in order to allow us to fine-tune the expression of the dctA gene, as was previously reported for other systems (11). Indeed, at intermediate concentrations of IPTG, the Suc⁻ phenotype of strains LM2800 and LM3115 was converted to the Suc⁺ phenotype (Table 4), whereas concentrations of IPTG that were too low or too high resulted in the Suc⁻ phenotype. This demonstrates that the reason why the atp deletion strains fail to grow on the C₄-dicarboxylates is a lack of transport capacity for these compounds.

TABLE 4.

Complementation with the dctA gene in trans in the atp mutants, LM2800 and LM3115

Strain	Plasmid	Colony diama (mm)		Carbon sourcec
		−IPTG	+IPTGb
LM2800	pFH2106	—d	—	Malate
LM2800	pLAC-dctA	—	0.04	Malate
LM3115	pFH2106	—	—	Malate
LM3115	pLAC-dctA	—	0.06	Malate

^aDiameter of single colonies, measured with a microscope and scored after 60 h of incubation at 37°C. The standard error on the determination of colony sizes is less than 15%. 

^b3 μM for LM2800 and 6 μM for LM3115. 

^cCarbon sources at 15 mM. 

^d—, colonies smaller than 0.02 mm. 

The expression in trans of the dctA gene, the structural gene encoding the C₄-dicarboxylate transporter, enabled the atp deletion strain to grow on the C₄-dicarboxylates, which suggests that the transport of succinate across the cytoplasmic membrane has a significant control on the growth (the biomass flux) of the atp deletion mutant on these substrates. Metabolic control analysis (14) predicts that the sum of the control coefficients, with respect to a flux, for all the enzymes in a system should add up to one, and experimental determination of control (particularly the control on growth rate) has indicated that the control is probably distributed among the many enzymes in the living cell. Therefore, the finding of a controlling step on the growth of E. coli is indeed an interesting result, but one should keep in mind that the atp deletion mutant is crippled by a lack of H⁺-ATPase, which is a very unusual situation for an E. coli cell.

In summary, we conclude that growth of E. coli atp deletion mutants on C₄-dicarboxylates is limited by the activity of the C₄-dicarboxylate transporter. The results also point towards the yhiF gene product as a negative transcriptional regulator of the dctA gene in E. coli, and our current working hypothesis is that the atp deletion strain cannot grow on C₄-dicarboxylates, because the negative transcriptional regulator prevents sufficient expression of the C₄-dicarboxylate transporter. We therefore suggest that the yhiF gene be renamed dctR, for dicarboxylate transport regulator.