Vol. 174, No. 23

JOURNAL OF BACrERIOLOGY, Dec. 1992, p. 7635-7641

0021-9193/92/237635-07$02.00/0 Copyright © 1992, American Society for Microbiology

Carbon and Energy Metabolism of atp Mutants of Escherichia coli PETER R. JENSEN' 2* AND OLE MICHELSEN' Department ofMicrobiology, Technical University of Denmark, Lyngby, Denmark 1 and Division of Molecular Biology, The Netherlands Cancer Institute, Amsterdam, The Netherlands2

Received 4 March 1992/Accepted 25 September 1992 The membrane-bound H+-ATPase plays a key role in free-energy transduction of biological systems. We report how the carbon and energy metabolism of Escherichia coli changes in response to deletion of the atp operon that encodes this enzyme. Compared with the isogenic wild-type strain, the growth rate and growth yield were decreased less than expected for a shift from oxidative phosphorylation to glycolysis alone as a source of ATP. Moreover, the respiration rate of a atp deletion strain was increased by 40% compared with the wild-type strain. This result is surprising, since the atp deletion strain is not able to utilize the resulting proton motive force for ATP synthesis. Indeed, the ratio of ATP concentration to ADP concentration was decreased from 19 in the wild type to 7 in the atp mutant, and the membrane potential of the atp deletion strain was increased by 20%o, confirming that the respiration rate was not controlled by the magnitude of the opposing membrane potential. The level of type b cytochromes in the mutant cells was 80%o higher than the level in the wild-type cells, suggesting that the increased respiration was caused by an increase in the expression of the respiratory genes. The atp deletion strain produced twice as much by-product (acetate) and exhibited increased flow through the tricarboxylic acid cycle and the glycolytic pathway. These three changes all lead to an increase in substrate level phosphorylation; the first two changes also lead to increased production of reducing equivalents. We interpret these data as indicating that E. colh makes use of its ability to respire even if it cannot directly couple this ability to ATP synthesis; by respiring away excess reducing equivalents E. coli enhances substrate level ATP synthesis. The membrane-bound ATP synthase of Escherichia coli has a central role in free-energy transduction under aerobic and anaerobic growth conditions. Under aerobic conditions, the proton motive force generated by the respiratory chain is utilized by the ATP synthase to drive ATP synthesis and to maintain solute gradients. Under anaerobic conditions, the proton motive force can be generated by the ATP synthase through hydrolysis of ATP. The atp operon, which encodes the ATP synthase, consists of nine genes in the following order: atpIBEFHAGDC (12, 19, 20). atpI encodes a 14-kDa protein of unknown function and has been shown to be dispensable (28). atpBEF encodes the three subunits that form the membrane-integrated Fo part (the proton channel), whereas atpHAGDC encodes the five subunits that form the cytoplasmic F, part (the ATPase) (12, 20). von Meyenburg et al. (27) analyzed the growth of strains that lack an intact ATP synthase. These strains are not able to grow on nonfermentative carbon sources, such as acetate or succinate. The growth rate and growth yield in aerobic minimal glucose medium were reduced to 75 and 55% of the wild-type levels, respectively. Mutant strains with Tn1O insertions in the promoter region of the atp operon also exhibited a reduced growth yield that paralleled reduced expression of the ATP synthase. In theory, an E. coli strain deprived of oxidative phosphorylation could increase its glycolytic rate and/or decrease its growth rate in order to maintain a balanced free-energy metabolism. However, such a metabolic shift should lead to a more significant reduction in growth yield than that observed by von Meyenburg and colleagues (3, 27). Alternatively, the cells might redirect their glycolytic flux from

lactate to products which permit more substrate level phosphorylation. Although solving the free-energy balance, such a shift might lead to increased production of redox equivalents. Because of the absence of respiration coupled to ATP synthesis, only uncoupled respiration or secretion of highly reduced products could then restore the redox balance. In this study we examined how E. coli manages its carbon and free-energy metabolism so as to allow for a fairly high growth yield in the absence of the ATP synthase. We found that E. coli strains that carry mutations in the atp operon have greater respiration (140% of the wild-type level) and enhanced substrate level phosphorylation pathways. Our results suggest that the cells support uncoupled respiration in order to be able to profit from increased substrate level phosphorylation.

MATE:RIALS AND METHODS List of parameters. The following parameters were used in this study: specific respiration rate (Q02) (millimoles of 02 per hour per gram of dry weight), total oxygen consumption (O) (millimoles of 02 per millimole of glucose), growth yield (Y) (grams of dry weight per millimole of glucose), specific growth rate (>) (hour-'), amount of cell carbon (Cc) (millimoles of carbon per millimole of glucose), amount of byproduct carbon (B.) (millimoles of carbon per millimole of glucose), and amount of substrate carbon (input) (Sc) (millimoles of carbon per millimole of glucose). Bacterial strains. The strains used in this study are listed in Table 1. Construction of strains carring deletions in the atpI gene. The deletions were constructed on plasmids by cloning the 867-bp ClaI-HindIII fragment from pFH248 (8), which carries the C-terminal end of the gidB gene and the N-terminal

* Corresponding author. 7635

7636

J. BACT1ERIOL.

JENSEN AND MICHELSEN

TABLE 1. E. coli strains used in this studya Strain

Type

AS19 ER CM789 CM845 CM927 CM987 CM1471 CM1559 CM2080 CM2630 CM2673 CM2681 CM2684 CM2687 DG7/1 LM408 LM827 LM1167 LM1178 LM1237 LM1316 LM1830 LM2800 LM3058 LM3115 LM3118 MA1079 RH50

B K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 K-12 B K-12 K-12 K-12 K-12

Characteristic, genotype, and/or plasmid

Permeable outer membrane F+ asnA31 asnB32 thi-1 reL41 spoTi As LM1237, but asnA31 (Xasn2OIXhb) As LM1237, but asnA31 (XasnlOSIXh) As LM1237, but asnA31 recAlI(Xh) As LM1237, but F- asnA31 recAl As LM1237, but AatpIBEFHI4706 asnA31 asnB32 reLA1 spoTi lysA As LM1237, but AatpIB768 As CM987, but (Xh) pXasn3-727, isolated as pAasn3-6 As LM1237, but atpI::TnlO-219 As LM1237, but atpI::TnlO-137 As LM1237, but atpI::TnlO-501 As LM1237, but atpI::TnlO-442 lacIfadR but-12 rha ilv metE As CM987, but pBJC408 (pBR322 Tetr AmpS) As LM1237, but atp-727 (gidAB-atpl) recAl As LM1237, but TnlO-112/pBJC917 As LM1237, but TnlO-112/pFHC350 F+ asnB32 thi-1 reLA1 spoT1 As LM1237, but AatpI705 As LM1237, but AatpI707 As LM1237, but AatpIBEFHAGDC750 As AS19, but AatpIBEFHA4706 As LM1237, but AatpIBEFIIAGDC750 lacUV5 lacY As LM1237, but lacUV5 lacY Hfr KL96 recA1 serA thi Hfr P01 ilvS215 bglR thi-1 relAl

Reference, source, or construction 26 9 9 9 25 9 9 L. Boe 9 9 27 27 27 27 6 This study This study 28 28 CM1471 x XasnlO5 This study This study This study This study 11 11 J. Davies via N. Fiil 10

a We used strain LM3118 (= LM1237 lacWS lacY) as the wild-type strain and strain LM3115 (= LM2800 lacUV5 lacY) as the deletion strain in the analysis of ATP/ADP ratios and type b cytochromes. As expected, the lacUVS and lacY mutations had no effect on the growth of the strains (i.e., on the growth rates, growth yields, and respiration rates of the strains). b

Xh, AI85757.

end of the atpI gene together with the atpIp promoter, into plasmid pOMC11 (HindIII in the C-terminal end of the atpI gene to Aval in the atpE gene [18]). The resulting plasmid, pOMC28, carries the atpIp promoter, the atpI gene (except the 199-bp internal HindIII fragment), and the atpB gene. Plasmid pOMC28 was restricted with HindIII, and the protruding ends were filled in with the Klenow fragment of DNA polymerase I before ligation. The resulting plasmid, pOMC32, has a 195-bp in-frame deletion in the atpI gene. Strain CM2080, carrying the atpIB768 deletion, was transformed with pOMC28 and pOMC32. The transformants were transferred to minimal medium supplemented with succinate as a carbon source, on which the high expression of the atpB gene from the plasmids causes severe inhibition of growth. Recombinants between plasmid and chromosome were picked as fast-growing colonies. The atpI deletions were then transduced by phage P1 into strain CM2080. The resulting strains, strains LM1316 and LM1830, have a 199-bp deletion (atpI705) and a 195-bp deletion (atpI707), respectively. Construction of a strain carrying a deletion of the entire atp operon (del atpIBEFIIAGDC750). The deletion was first constructed on a plasmid. Plasmid pBJC260, a derivative of pBR325, which carries the 4,646-bp EcoRI fragment from atpD to gimS in the orientation opposite to the transcription of the cam gene, was restricted with HindIII and BalI. The gidB gene was then cloned on the 819-bp HindIII-HpaI fragment from plasmid pFH248 (8) into plasmid pBJC260. The ligation mixture was transformed into strain LM827 (relevant genotype, recA gidB). The transformants were

plated onto Luria-Bertani medium supplemented with ampicillin (100 mg/liter) and were screened for complementation of thegidB mutation, which causes very slow growth in recA strain backgrounds (14). Fast-growing colonies were isolated, and they all harbored a plasmid with the intact gidB gene. The resulting plasmid, pOMC45, carries chromosomal DNA flanking a 7.4-kb deletion of the atp operon. The deletion was then transferred to the chromosome by transforming strain CM789 with plasmid pOMC45. Strain CM789 is lysogenic for a specialized transducing lambda phage, lambda asn2O (25), which carries chromosomal DNA from the asnA gene to the atpB gene. The lambda phage was induced, and the lysate was used to transduce strain LM408 to ampicillin resistance (50 mg/liter) at 30°C. Ampicillin resistance can be transferred only if the plasmid has integrated into the phage genome. Strain LM408 harbors a Tetr Amp' pBR322 derivative in order to stabilize the lambda asn2O::pOMC45 cointegrant by suppressing replication from the integrated pBR322 replicon. The lambda asn2O::pOMC45 phage was induced, and the lysate was used to transduce strain ER to Asn+ at 42°C. The resulting strain, strain LM2800, has the chromosomal DNA from the HpaI site 444 bp upstream of the atpIp promoter to the BalI site in the C-terminal part of the atpC gene deleted. The putative atp deletion strains were lysogenized with lambda asnlOS-oriC1221 (24), which has the DNA from within the asnA gene to within the atpD gene replaced with TnlO (14). A lysate was prepared from these strains and was used to transduce strain CM927 (Table 1) to Asn+ at 30°C. Only lambda phages that have the TnlO DNA replaced with

VOL. 174, 1992

CARBON AND ENERGY METABOLISM OF atp MUTANTS

chromosomal DNA are able to transduce strain CM927 to Asn+. The resulting lambda asniOS derivatives, which carried the deletions, were isolated, and the phage DNA was subjected to restriction enzyme analysis. All of the putative atpIBEFI-L4GDC750 deletion strains were shown to carry the deletion, while the putative atpI deletion strains were shown either to carry the desired deletion or to be strains which carried the initial plasmids in the chromosomes. The constructions were further verified by Southern blot analysis (21), after the chromosomal DNA was restricted with either HindIII, EcoRI, PvuII, or HpaI plus PstI. The probe was prepared by restricting pPRJ2 with EcoRI and labelling the preparation with 35S-dATP, using the Klenow polymerase. Transfer of atp mutations between strains. Deletion atpIBEFA706 was transduced by P1 into strain AS19 after ilvC::TnS had been inserted into strain AS19. Deletion atp727 was transduced into strain CM1559 from lambda asn3-727 at 42°C, with selection for Asn+. The transductant was then conjugated with Hfr strain MA1079, with selection for LysA+ and screening for sensitivity to UV light (recA). Preparation of DNA, transformation, and phage transduction were accomplished by using standard methods (13, 15). Restriction endonucleases, exonucleases, and T4 DNA ligase were obtained from and used as recommended by New England Biolabs. Growth of bacterial cultures. Bacterial cultures were grown essentially as described by von Meyenburg et al. (27). Experiments were performed at 30°C in AB minimal medium (4) (pH 7.0) supplemented with 0.0400% glucose (molecular weight, 198.2) and thiamine (2.5 mg/liter). The cell dry weights of cultures were determined to be 186 and 189 mg/liter at an optical density at 450 nm of 1.00 for wild-type strain LM1237 and atp deletion strain LM2800, respectively. Membrane potential. Membrane potential was determined by measuring the uptake of 3H-labeled tetraphenylphosphonium+, as described by von Meyenburg et al. (26). Respiration rate. Respiration rate was determined by transferring 15 ml of a culture into a reaction tube. The sample was aerated heavily for 2 min, and the decrease in dissolved oxygen was then recorded with a Clark type of electrode (26). Moderate stirring of the sample was necessary in order to exchange the liquid layer next to the electrode, thereby minimizing the influence of the oxygen consumed by the electrode. By-product measurement. The by-product excreted by the strains was identified as acetate by gas chromatography with mass spectroscopic detection. When the cultures entered stationary phase, they were cooled on ice and centrifuged, and the media were filtered through 0.45-,um-pore-size membrane filters (Millipore). The amount of acetate was determined from the growth yield obtained after filtered growth medium was inoculated with wild-type strain LM1237 pregrown in minimal acetate medium, by using a standard curve for known concentrations of acetate. Measurement of ATP synthase. ATP synthase was measured as described by von Meyenburg et al. (27). Briefly, the total protein content of cells was labelled with 35S-methionine and was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The content of ATP synthase subunit c was quantified relative to lipoprotein (which is assumed to be constitutively expressed) after optical scanning of the autoradiograms. Although the expression of subunit c is not necessarily equivalent to the activity of ATP synthase in the cells, it can be taken as a rough estimate, as the ATP synthase subunits are expressed in stoichiometric ratios (12).

7637

ATP/ADP ratios. Samples (1 ml) were withdrawn from cultures and were immediately mixed with 1 ml of phenol (room temperature, equilibrated with TE buffer [pH 8] containing 0.1% 8-hydroxyquinoline) by vortexing the preparation for 10 s. After 5 min at room temperature, the samples were cooled on ice. The samples were centrifuged for 5 min at 14,000 rpm, and the water phase was removed and extracted with chloroform-isoamyl alcohol (24:1, vol/ vol). The ATP assay was carried out at room temperature by using a luciferase ATP monitoring kit that was obtained from and used as recommended by LKB. The ADP concentration was determined in the same sample after the ATP concentration was determined by adding phosphoenolpyruvate (3 mM) and pyruvate kinase and recording the increase in the ATP concentration. Type b cytochromes. Portions (200 ml) of cells were harvested from exponential-phase cultures at an optical density at 450 nm of 0.8 and were concentrated in 2 ml of phosphate buffer (pH 7.5). The cells were broken by subjecting them to a 60-s ultrasonic treatment, which left less than 5% of the cells intact. We then used a 0.1 dilution of this preparation to make a reduced sample by adding dithionite and an oxidized sample by adding ferricyanide. The difference absorbance spectrum between these two samples was then recorded and analyzed by using a chopped, dualwavelength Aminco model DW-2000 spectrophotometer. The type b cytochromes gave rise to a peak at 562 nm in both strains.

RESULTS Respiration rate of atp deletion strains. We examined the growth rates, yields, and respiration rates of two E. coli strains which have deletions of essentially the entire atp operon (Table 2), strains LM2800 (del atpIBEFIIAGDC750) and CM1471 (atpIBEFHA706). When these strains were grown in glucose minimal medium at 30°C, their respiration rates increased by approximately 40% compared with the wild-type rate, although these strains are not able to synthesize ATP through oxidative phosphorylation. The growth rates of strains LM2800 and CM1471 were 79 and 74% of the wild-type level, respectively, and the growth yields were 58 and 55% of the wild-type levels, respectively. There was no significant difference between strain LM2800 and CM1471, although strain CM1471 carries the atpIBEFHA706 deletion that fuses the atpI gene in frame with the atpA gene (18) and thus expresses the atpGDC genes at a low level. These two strains are missing coding information for both the Fo and the F1 parts of the ATP synthase. We also examined strain CM2080, which expresses the F1 part of the ATP synthase. This strain exhibited a strongly reduced yield and growth rate, although its respiration rate was increased (Table 2). The increased respiration rates of the atp mutant strains (approximately 40%) were also observed at 37°C. Membrane potential of atp mutants. In most E. coli strains, including wild-type strain LM1237, measurement of membrane potential requires pretreatment of the cells with TrisEDTA. We found that this treatment greatly affected respiratory rates in strain LM1237, and therefore, we used strain AS19 cells, which do allow measurement of membrane potential without Tris-EDTA pretreatment. The atpBE FII4706 deletion was transduced into strain AS19, resulting in strain LM3058. The transmembrane electric potentials in isogenic strains AS19 and LM3058 were found to be 154 + 1.1 and 184 + 1.4 mV, respectively. Thus, deletion of the atp

7638

JENSEN AND MICHELSEN

J. BACTERIOL.

TABLE 2. Growth parameters of atp mutants in glucose minimal medium Strain

LM1237C LM2800 CM1471 LM1316 LM1830 LM1178 CM2684 CM2687 CM2673 CM2681 LM1167 CM2080

Genotype of chromosome/genotype of plasmida

atp+ AatpIBEFHAGDC750 AatpIBEFHA 706 AatpI705 AatpI707 atp+/atpI+ atpI719::TnlO atpI717::TnlO atpI716::TnlO atp-gid::TnlO atp+/atp+ AatpIB

% of wild-type levels

Subunit cL

100 0 0 96 93 103 23C 37C 47C 127C 250c

NDd

100 79 74 100 101 101 83 91 98 102 100 57

Q02 100 139 137 102 101 103 125 122 120 92 90 141

Y

°t

100 58 55 100 99 99 72C 80C 90C 99C 95C 45

100 102 102

102 99 101

108 107 110

89 86 111

a See Table 1 for complete genotypes. b Values determined previously by von Meyenburg et al. (27). c The actual values of the growth variables for strain LM1237 were as follows: ,, 0.53 h-i; Y, 0.080 g (dry weight) per mmol of glucose; and Q02, 12.4 mmol ofd°2 per h per g of dry weight. ND, not determined. The subunit c level and the amount of ATPase in this strain were found to be 30% of the wild-type levels by Friedl et al. (6).

operon is associated with a 20% increase in membrane potential. Respiration rates of strains with decreased or increased levels of ATP synthase. To make sure that the changes observed in the deletion mutant are also relevant for cases in which ATP synthase activity is merely modulated, we also studied mutants with TnlO insertions in the atp promoter region (27). When the expression of the atp operon was 23 to 47% of the expression in the wild-type strain, the growth parameters had values between the deletion strain values and the wild-type strain values (Table 2). When the atp operon was overexpressed (strains CM2681 and LM1167), the respiration rates were lower than the wild-type rate, while the growth rates and the growth yields remained close to the wild-type levels (Table 2). Changes in the atpI gene only. The deletion strains, as well as the strains with Tn1O insertions, were all defective in the atpI gene, which has been reported to affect growth (7). We examined strains that had deletions in the atpI gene, either in frame (strain LM1830) or out of frame (strain LM1316), and found that the deletions had no effect on the growth parameters examined (Table 2). The expression of subunit c of the ATP synthase in the atpI deletion strains was found to be the same as the expression in the wild-type strain (Table 2). Strain LM1178 carries a plasmid from which the atpI gene is expressed from the atpIp promoter. The growth parameters observed for this strain were normal (Table 2), which is in contrast to previously published results (28). Oxygen consumption of atp mutants. The total amount of oxygen consumed per millimole of glucose input (0t) can be calculated from the following equation:

Ot=(Qo2

Y)4j.

(1)

The Ot values were 1.86 mmol of 02 per mmol of glucose for the wild-type strain and 1.90 mmol of 02 per mmol of glucose for deletion strain LM2800. Thus, the deletion strain consumed the same amount of oxygen as the wild type per mole of glucose metabolized, although the yield was only one-half of the wild-type yield. Where does the carbon flow? The decrease in growth yield in the atp deletion mutants implies that there is an increase in the formation of carbon products other than biomass. We

used gas chromatography coupled to mass spectrometry to identify these products. Both in the wild type (strain LM1237) and in the mutant (strain LM2800) more than 90% of the by-product produced was in the form of acetate. We then measured the amounts of by-product carbon and found that in the wild type and in the deletion strain the amounts of by-product carbon were 1.0 and 1.8 mmol/mmol of glucose consumed, respectively. This implies that upon deletion of the atp operon, E. coli does not just shift toward fermentative metabolism with lactate as the end product. ATP/ADP ratios. The decreased growth rate of the atp deletion strain indicated that the rate of ATP synthesis was decreased in this strain. Therefore, we measured the concentrations of ATP and ADP in extracts from wild-type cells (strain LM3118) and from cells depleted of ATP synthase (cells carrying the atpIBEFHAGDC750 deletion [strain LM3115]). We found that the ratio of ATP concentration to ADP concentration was decreased from 19 ± 4 in the wild-type strain to 7 ± 1 in the deletion strain, whereas the sum of ATP and ADP concentrations was not changed (approximately 8 ,umol/g of dry weight). Thus, it seems that deletion of the atp operon did indeed affect the free-energy state of the cells. This effect is large compared with the relatively small effect that we observed on the growth rate (a decrease of 25%). Type b cytochromes. The increased respiration rates of atp mutants indicated that these strains might have increased levels of respiratory enzymes. Therefore, we analyzed the type b cytochrome contents in membranes prepared from wild-type cells and cells depleted of ATP synthase (cells carrying the atpIBEFHAGDC750 deletion). This was done by recording the reduced-minus-oxidized spectrum in the region around 562 nm. We found that the amount of type b cytochromes in the membranes from the deletion strain was increased by approximately 80% compared with the wildtype strain. If the respiratory enzymes control respiration even in part, then this result provides an explanation for the increased respiratory rates of atp mutants. Redox and carbon balances. By using the redox and carbon balance we could calculate the amount of CO2 produced in two different ways and perform a consistency check. When E. coli cells are grown in minimal media supplemented with

CARBON AND ENERGY METABOLISM OF atp MUTANTS

VOL. 174, 1992

7639

TABLE 3. Carbon balance (millimoles of carbon per millimole of glucose input) for wild-type strain LM1237 and atp deletion strain LM2800 Carbon balance (mmol of C/mmol of glucose input)a

Strain

SSC

LM1237 LM2800 a b

c

6 6

ci'

Bc

C02balance) (electron

(carbon C02 balance)

C02 produced

3.2 1.9

1.0 1.8

2.0 2.2

1.8 2.3

0.5 0.3

C

in anabolic reactionsc

Carbon utilized in pure

Carbon catabolized in the TCA cycle'

2.3 3.7

0.5 0.7

catabolisme

All values are the output from 1 mmol of glucose (6 mmol of Sc). Calculated from the growth yield. Calculated as follows: 0.17 x Cc. -

dCalculated as follows: Sc Cc 0.17 (Cc). ¢ Calculated as follows: 2/[(3 x carbon utilized in pure catabolism)

-

Bc] (see text).

glucose as the carbon and energy source, the following reaction occurs: 1 mmol of C6H1206 + x mmol of 02 -ymmol of Cc + z mmol of CO2 + w mmol of Bc where x, y, z, and w indicate the number of millimoles consumed or evolved in the reaction. This reaction can be considered a redox process, in which the substrate carbon, glucose (Sc), is both oxidized and reduced and molecular oxygen is reduced. For cells grown in glucose minimal media, we calculated that the cellular carbon content was 39.8 mmol of carbon per g of cell dry weight, based on the cell composition given by Neidhardt (16). The average oxidation state for carbon in the cell material can be calculated to be -0.20, which means that the cell carbon is reduced compared with glucose. This average oxidation state of the cell carbon includes the reduction of sulfur from +6 to -2 and the oxidation of nitrogen (e.g., in purines and pyrimidines) that takes place when cells are grown in minimal medium containing S042- and NH4+. Therefore, the electron balance for the redox process in E. coli grown on glucose with acetate as a by-product can be expressed as follows: (0) 6. 1 + (-4) .x + (-0.20) .y + (4) .z+(0) w = 0 or

-4x + (-0.20)y + 4z = 0 (2) Equation 2 was used to calculate the amount of CO2 (z) produced during the growth of the cultures. The amount of oxygen consumed (x) was calculated from equation 1, and the amount of cell carbon (y) was calculated from the growth yield (grams of dry weight per millimole of glucose) by multiplication with the carbon content of the cells. The carbon balances for the wild-type strain, strain LM1237, and the atp deletion strain, strain LM2800, in glucose minimal medium, are shown in Table 3. The levels of carbon recovery for both strains were within 4% of the level of carbon input. The main difference between deletion strain LM2800 and wild-type strain LM1237 was the amount of by-product produced, which almost accounted for the difference in yields. The calculated amounts of CO2 produced by the two strains were almost identical, which was a consistency check in our analysis. From the cell composition of glucose-grown cells (16) we calculated that the net production of CO2 carbon from the anabolic reactions is 17% of the cell carbon. When this value was used with the results from the carbon balance determinations (Table 3), it was found that when the wild-type strain (strain LM1237) and the deletion strain (strain LM2800) were

grown on medium containing 1 mmol of glucose (6 mmol of substrate carbon), they had 2.3 and 3.7 mmol of substrate carbon, respectively, left over for mobilization of free energy (fuelling reactions). This resulted in 1.5 and 2.5 mmol of acetyl coenzyme A (acetyl-CoA) carbon after decarboxylation of pyruvate. The by-products excreted by the two strains were 1.0 and 1.8 mmol of acetate carbon, respectively, which implies that in both strains, a small but significant amount of acetyl-CoA was metabolized, presumably through the tricarboxylic acid (TCA) cycle. In these calculations of metabolism parameters (more specifically, for the value 17%), it is necessary to account for the generation of NADPH, which is used in the biosynthetic reactions and the different reactions that produce C02, as well as the reactions that consume CO2. NADP is reduced by specific dehydrogenases (e.g., isocitrate dehydrogenase) and by the NADH-NADP transhydrogenase. Csonka and Fraenkel (5) have found that the NADH-NADP transhydrogenase has only a minor role in generating NADPH. CO2 is produced in the glycolytic pathway and in the TCA cycle, but a significant amount of CO2 is also produced in the pentose pathway in normal cells together with NADPH. In order to calculate the amounts of CO2 produced through this pathway, we have assumed on the basis of the radioactive incorporation data of Csonka and Fraenkel (5) that 30% of the total NADPH requirement is also generated there in the atp mutant. However, because the membrane potential is higher in the atp mutant, one might expect increased activity of the NADH-NADP transhydrogenase. This would lead to a decreased requirement in the atp mutant for NADPH production by the pentose phosphate pathway and thus decreased production of anabolic CO2 compared with the wild-type cells. The net effect would be that the calculated increase in the TCA cycle described above would be even higher. DISCUSSION

The regeneration of ATP from ADP can occur in two different ways (either by substrate level phosphorylation or by oxidative phosphorylation) which are catalyzed by the ATP synthase and energized by the proton motive force. Wild-type strains can use both pathways, while atp mutants are dependent solely on substrate level phosphorylation. The growth rates of strains which have deletions in the atp operon were reduced to 74 and 79% of the wild-type levels, while the growth yields were reduced to 55 and 58% of the wild-type levels. Although the growth of wild-type cells is not limited by the synthesis of ATP, as shown below, these cells do not synthesize more ATP than they need unless the

7640

JENSEN AND MICHELSEN

J. BACTE RIOL.

TABLE 4. Carbon flow (millimoles of carbon per hour per gram of dry weight) for wild-type strain LM1237 and atp deletion strain LM2800 Carbon flow (mmol of C/h/g [dry wt])0 CO2 (electron

CO2 (carbon

Sc

Cc

Bc

balance)

balance)

C02 produced reactions

Carbon utilized catabolism

40 51

21 17

7 15

14 19

12 19

4 3

15 32

Strain

LM1237 LM2800

Carbon catabolized

in the TCA cycle

4 6

a Values were calculated from the data in Table 3 by multiplying by the growth rate and dividing by the growth yield. See text and Table 3.

growth conditions have induced a futile cycle. Thus, the growth rate and the growth yield depend on the amount of ATP needed for polymerization of the cellular macromolecules and are relative measures of the rate of synthesis of ATP and the amount of ATP synthesized per unit of substrate consumed, respectively. Thus, the lower growth rates and growth yields of atp deletion strains indicate that these cells are affected by their deficiency in oxidative phosphorylation of ADP (about 75% of the normal rate) and in the amount of ATP synthesized (about 55% of the normal amount). The growth rates and growth yields were further reduced in strain CM2080, which has an uncoupled ATPase attached to the membrane (6, 7). This indicates that free energy is lost in strain CM2080 cells, probably because of uncoupled hydrolysis of ATP by F1. Strain CM2673 has one-half of the wild-type amount of ATP synthase, but the growth rate of this strain was close to the wild-type rate. Thus, the rate of synthesis of ATP was sufficient to support growth at a normal rate, but the reduced growth yield shows that the total amount of ATP synthesized from the substrate was reduced in this strain. The growth rates of strains CM2684 and CM2687 (25 and 40% of the wild-type amount of ATP synthase) were significantly lower than the wild-type growth rate, which indicates that the rates of ATP synthesis in these strains were too low to support growth at a normal rate. Together, these results indicate that the growth rate of E. coli cells which had more than one-half of the wild-type amount of ATP synthase were not limited by the availability of ATP when the cells were grown with glucose as carbon source. However, the fraction of the substrate that was catabolized to provide ATP increased as the amount of ATP synthase decreased. The main difference in the metabolism of the wild-type and atp deletion strains was revealed by the carbon balances. Especially when the amounts of substrate carbon converted in the different pathways are considered as fluxes, the difference between the wild-type and atp deletion strains becomes clear (Table 4). If we assume that the cells without an atp operon have the same composition as the wild-type cells, then the carbon fluxes through the anabolic pathways in growing cells are fixed in proportion to each other. These can then be represented by the carbon flow into cell mass (Cc), which can be derived from the growth yield and the growth rate. In contrast, the flow through the fuelling reactions might vary in response to changes in the energy metabolism of the cells. The growth yield of the azp deletion strains is reduced compared with the wild type, yet not by as much as might have been expected (3). If, for instance, the cells had carried out the same shift as they do upon anaerobiosis as a route to produce ATP, their growth yield should have been decreased to approximately 20% of the wild-type yield (22), rather than the observed 55 to 58%. Clearly, the cells have a

more profitable way to produce ATP, enhanced substrate level phosphorylation by directing carbon toward acetate (ATP produced from the acetyl-CoA produced by the pyruvate dehydrogenase complex). Indeed, the calculations show that (i) the flow of carbon through the glycolytic pathway for fuelling purposes in the atp deletion strain is twice the flow in the wild-type strain and (ii) the flow through the TCA cycle is higher in the atp deletion strain than in the wild-type strain. The amount of ATP generated by the deletion strain is the same whether acetyl-CoA is converted to acetate (by the phosphotransacetylase pathway) or oxidized to carbondioxide (through the TCA cycle), as both pathways involve one substrate level phosphorylation, but the latter is followed by production of reduced pyrimidine nucleotides (3 mol of NADH and 1 mol of FADH2 per mol of acetyl-CoA). In the deletion strain the increased carbon flow through the glycolytic pathway and the TCA cycle results in increased generation of NADH. If oxidative phosphorylation were strongly coupled, this would cause a redox problem; the elimination of the ATP synthase would force respiration down. However, the cells seem to make use of the tendency of oxidative phosphorylation to be uncompletely coupled, and they continue to respire. Amplification of their type b cytochromes even gives them the ability to further get rid of their excess reducing equivalents by respiration, which causes the observed 40% increase in the respiration rate. When the respiration level was recalculated as oxygen consumption per generation, we found that the deletion mutants consumed twice as much oxygen per generation as the wild type. The 20% increase in the membrane potential which we found in an atp deletion strain indicates that the respiration rate of wild-type E. coli is not limited by the magnitude of the opposing proton motive force. This might explain why it has been impossible to show an increase in the respiration rate of exponentially growing E. coli cells after addition of an

uncoupler (respiratory control) (2, 23). The increased respiration rate of strain CM2673 (50% ATP synthase compared with that of the wild type), which synthesized ATP at a normal rate (see above), shows that the rate of NADH generation and thus the rate of ATP synthesis by substrate level phosphorylation were increased in this strain. If the total rate of ATP synthesis in cells of strain CM2673 was the same as or less than the rate in the wild-type cells, then the rate of ATP synthesis by oxidative phosphorylation was decreased despite increased respiration. This suggests that the P/O ratio of oxidative phosphorylation in this strain was less than the P/O ratio in wild-type cells. Of course, the deletion mutants are the ultimate examples of this situation. On the other hand, the reduced respiration rates of strains CM2681 and LM1167, which both overproduce the ATP synthase and whose growth rates and

VOL. 174, 1992

CARBON AND ENERGY METABOLISM OF atp MUTANTS

yields were close to the normal levels, suggest that these strains had a decreased catabolic flow and that they thus had increased P/O ratios. These changes in the P/O ratio of oxidative phosphorylation extend the results of Andersen and von Meyenburg (1), who found that the P/O ratios of oxidative phosphorylation in E. coli depended on the carbon substrate during growth.

13. 14. 15.

16.

ACKNOWLEDGMENTS We thank J0rgen 0gaard Madsen (Technical University of Denmark) for gas chromatography and mass spectroscopic analysis and Klaas Hellingwerf and Wim Crielaard (University of Amsterdam) for helping with the cytochrome analysis. We are very grateful to Hans V. Westerhoff, Tove Atlung, and Knud V. Rasmussen for comments on the manuscript. This work was supported by the Danish Centre for Microbiology. REFERENCES 1. Andersen, K. B., and K. von Meyenburg. 1980. Are growth rates of Escherichia coli in batch cultures limited by respiration? J. Bacteriol. 144:114-123. 2. Burstein, C., L. Tiankova, and A. Kepes. 1979. Respiratory control in Escherichia coli K12. Eur. J. Biochem. 94:387-392. 3. Butlin, J. D., G. B. Cox, and F. Gibson. 1971. Oxidative phosphorylation in Escherichia coli K12. Biochem. J. 124:7581. 4. Clark, D. J., and 0. Maal0e. 1967. DNA replication and the division cycle in Escherichia coli. J. Mol. Biol. 23:99-112. 5. Csonka, L. N., and D. G. Fraenkel. 1977. Pathways of NADPH formation in Escherichia coli. J. Biol. Chem. 252:3382-3391. 6. Friedl, P., J. Hoppe, R. P. Gunsalus, 0. Michelsen, and H. U. Schairer. 1983. Membrane integration and function of the three Fo subunits of the ATP synthase of Escherichia coli K-12. EMBO J. 2:99-103. 7. Gay, N. J. 1984. Construction and characterization of an Escherichia coli strain with an uncI mutation. J. Bacteriol. 158:820825. 8. Hansen, F. G., S. Koefoed, K. von Meyenburg, and T. Atlung. 1981. Transcription and translation events in the oriC region of the E. coli chromosome. ICN-UCLA Symp. Mol. Cell. Biol. 21:37-55. 9. Hansen, F. G., J. Nielsen, E. Riise, and K. von Meyenburg. 1981. The genes for the eight subunits of the membrane bound ATP synthase of Escherichia coli. Mol. Gen. Genet. 183:463-472. 10. Humbert, R., W. S. A. Brusilow, R. P. Gunsalus, D. J. Klionsky, and R. D. Simoni. 1983. Escherichia coli mutants defective in the uncH gene. J. Bacteriol. 153:416-422. 11. Jensen, P. R., H. V. Westerhoff, and 0. Michelsen. On the use of laq-type promoters in control analysis. Eur. J. Biochem., in press. 12. Maloney, P. C. 1987. Coupling to an energized membrane: role of ion-motive gradients in the transduction of metabolic energy, p. 222-243. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Esche-

17. 18.

19. 20. 21. 22.

23.

24.

25.

26.

27.

28.

7641

richia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D.C. Maniatis, T., E. F. Fritsch, and J. Sambrook 1982. Molecular cloning. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Michelsen, 0. Unpublished data. Miller, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. Neidhardt, F. C. 1987. Chemical composition of Escherichia coli, p. 3-6. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D.C. Nielsen, J. 1984. Structure and expression of the ATP synthase operon of Escherichia coli. Ph.D. thesis. Technical University of Denmark, Lyngby, Denmark. Nielsen, J., F. G. Hansen, J. Hoppe, P. Friedl, and K. von Meyenburg. 1981. The nucleotide sequence of the atp genes coding for the Fo subunits a, b, c and the F1 subunit of the membrane bound ATP synthase of Eschenichia coli. Mol. Gen. Genet. 184:33-39. Senior, A. 1988. Oxidative phosphorylation. Physiol. Rev. 68: 177-231. Senior, A. 1990. The proton-translocating ATPase of Escherichia coli. Annu. Rev. Biophys. Biophys. Chem. 19:7-41. Southern, E. M. 1975. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol. 98:503-517. Tempest, D. W., and 0. M. Neissel. 1987. Growth yield and energy distribution, p. 797-806. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D.C. Tsuchiya, T., and B. P. Rosen. 1980. Respiratory control in Escherichia coli. FEBS Lett. 120:128-130. von Meyenburg, K., and F. G. Hansen. 1980. The origin of replication, oriC, of the Escherichia coli chromosome: genes near oriC and construction of oriC deletion mutants. ICNUCLA Symp. Mol. Cell. Biol. 19:137-159. von Meyenburg, K., F. G. Hansen, L. D. Nielsen, and E. Riise. 1978. Origin of replication, oriC, of the Eschenichia coli chromosome on specialized transducing phages )asn. Mol. Gen. Genet. 160:287-295. von Meyenburg, K., B. B. J0rgensen, 0. Michelsen, L. Sorensen, and J. E. G. McCarthy. 1985. Proton conduction by subunit a of the membrane bound ATP synthase of Escherichia coli revealed after induced over production. EMBO J. 4:2357-2363. von Meyenburg, K., B. B. J0rgensen, J. Nielsen, and F. G. Hansen. 1982. Promoters of the atp operon coding for the membrane bound ATP synthase of Escherichia coli mapped by TnlO insertion mutations. Mol. Gen. Genet. 188:240-248. von Meyenburg, K., B. B. J0rgensen, and B. van Deurs. 1984. Physiological and morphological effects of over production of membrane-bound ATP synthase in Escherichia coli K-12. EMBO J. 3:1791-1797.