Does Escherichia coli possess a second citrate synthase gene?

Eur. J. Biochem. 214,75-81 (1993) 0 FEBS 1993

Does Escherichia coli possess a second citrate synthase gene ? Amanda J. PATTON, David W. HOUGH, Paul TOWNER and Michael J. DANSON Department of Biochemistry, University of Bath, England (Received January 25, 1993) - EJB 93 012813

Escherichia coli possesses a hexameric citrate synthase that exhibits allosteric kinetics and regulatory sensitivity, and for which the gene (&A) has previously been cloned and sequenced. A citratesynthase-deficient strain of E. coli (K114) has been mutated to generate a revertant (K114r4) that produces a dimeric citrate synthase with altered kinetic and regulatory properties. On cloning and sequencing the gltA gene from both K114 and K114r4, a single mutation was found that caused the replacement of Asp362 with Asn. Asp362 has been previously shown to be a catalytically essential residue in E. coli citrate synthase, and we demonstrate that the hexameric enzyme produced on expression of the gltA gene from K114 and K114r4 is inactive. The dimeric citrate synthase from K114r4 has been purified and shown to be immunologically distinct from the wild-type hexameric enzyme. Determination of its N-terminal amino acid sequence demonstrates that the mutant citrate synthase is encoded by a gene distinct from the E. coli gltA gene. The N-terminal sequence is compared with those of other eukaryotic, eubacterial and archaebacterial citrate synthases.

Citrate synthase catalyses the condensation of acetylCoA with oxaloacetate to produce citrate and coenzyme A. It is often regarded as the first enzyme in the citric acid cycle, as it is by the formation of citrate that carbon atoms enter the cycle. The central role of citrate synthase in metabolism has, therefore, generated much interest and consequently citrate synthase has been studied from a variety of organisms, though the pig heart and Escherichia coli forms of the enzyme have been the most extensively investigated [l -31. The amino acid sequences of citrate synthase have been determined, either directly from the protein or inferred from DNA sequences, from the eukaryotes pig [4-61, Tetrahymenu thermophila [7], Arabidopsis thaliana [8] and Saccharomyces cerevisiae [9], the eubacteria E. coli [10-21], Pseudomonas aeruginosa [12], Acinetobacter anitratum [13], Acetobacter aceti [14], Bacillus sp c4 [15], Mycobacterium smegmatis [161, Coxiella burnetti [17] and Rickettsia prowazekii [181, and the archaebacterium Thermoplasma acidophilum [19]. Multiple alignment of these sequences [ 191 has shown high sequence identities within the eukaryotic enzymes (46-92%) and within members of the Gramnegative bacteria (56-75%); however, between the three Kingdoms, eukaryotes, eubacteria and archaebacteria, identities are far lower (20-26%). In addition, high-resolution X-ray crystallographic structures are available for pig and chicken citrate synthases [20-231. This has led to the identification of 12 residues important for the activity of citrate synthase, three of which, His274, His320 and Asp375, participate directly in catalysis and have been shown by multiple Correspondence to M. J. Danson, Department of Biochemistry, University of Bath, Claverton Down, Bath, England BA2 7AY Fax: +44 225 826449. Abbreviations. Me,SO,Et, ethyl methanesulphonate; PCR, polymerase chain reaction. Enzyme. Citrate synthase (EC 4.3.1.7).

alignment studies to be conserved throughout all citrate synthase sequences analysed so far. Comparative studies on citrates synthases have shown that a pattern of diversity exists with respect to oligomeric nature and regulatory properties of the enzyme (reviewed in [l -31). Eukaryotes, Gram-positive eubacteria and archaebacteria produce a small dimeric form of citrate synthase that is regulated in an isosteric manner by ATP, whilst Gramnegative eubacteria produce a large hexameric enzyme that is regulated allosterically by NADH or 2-oxoglutarate. Both forms are composed of a single type of subunit of M,=50000, suggesting that the quaternary structure of the enzyme plays an important role in determining its regulatory properties. In-vivo mutagenesis has been used to investigate this structure/function relationship [24, 2.51 and using such an approach, a mutant of E. coli has been isolated that produces a catalytically active, dimeric form of citrate synthase that is insensitive to NADH or 2-oxoglutarate. This form of the enzyme therefore resembles a Gram-positive/eukaryotic enzyme more closely than it does the Gram-negative parent enzyme. The present paper reports the analysis of the nature of this mutation and explores the relationship of the wildtype hexameric and the mutant dimeric citrate synthases.

MATERIALS AND METHODS Bacterial strains and plasmids The citrate-synthase-deficient strain of E. coli K12 [K114] (Hfr pps met thy gltA st?) was kindly donated by Professor Sir Hans Kornberg (University of Cambridge). The routine transformation host was E. coli strain TGl but, when a citrate-synthase-deficient host was required, E. coli W620 was used. HBlOl was used as a wild-type K12 strain. Plasmid pDB2 refers to pBR322 that contains the E. coli citrate

76 synthase gene gZtA (3.2 kb HindIII-EcoR1 fragment) [26]. pUCl8 was purchased from Northumbria Biologicals Ltd.

tions and DNA modifications were performed as recommended by the enzyme manufacturers.

Media

DNA amplification

Either Luria-Bertani medium or minimal medium were used. The latter was based on the basal salts medium of Ashworth and Kornberg [27] supplemented with glucose (10 mM) or succinate (10 mM) and thiamine (5 mg/l), uracil (35 mgfl), methionine (10 mgA) or thymine (10 mgA) as required.

Amplification of the gltA gene encoding E. coli citrate synthase was performed in a final volume of 100 p1 containing 10 mM Tris/HCl, pH 8.3, 50 mM KCI, 3 mM MgCL 200 nM each dNTP, 1 pM oligonucleotide primers, 2.5 U Tuq DNA polymerase and 100 ng purified chromosomal DNA or 10 ng purified plasmid DNA. Oligodeoxynucleotides 5 ’ ATCATTAGAATTCACCTACAT and 5’-ATAAAAATCAAGCTTGCCATAT were used as the forward and reverse primers, respectively. The forward primer is complementary to a region of the DNA 441-461 nucleotides upstream of the gltA initiation codon, and includes the putative citratesynthase promoter regions ;the reverse primer is complementary to DNA 45-66 nucleotides downstream of the gltA coding region. Amplication is therefore expected to result in a 1.8-kb fragment containing the &A gene. Sequencing was performed by the dideoxynucleotide chain-termination sequencing method [30] using [a-35S]dATP and denatured double-stranded DNA. Immediately prior to sequencing, 5 pg double-stranded DNA in 20 p1 was denatured by incubation with 5 pl 1 M NaOH, 1 mM EDTA, for 5 min at room temperature. The alkali was removed by spun column chromatography using Sepharose CL6B.

Enzymes and reagents CoA and NADH were from Boehringer Mannheim; Rabbit IgG conjugated to horseradish peroxidase was from Sigma. All restriction endonucleases and DNA-modifying enzymes were from Northumbria Biologicals Ltd; Tuq polymerase was from Perkin Elmer Cetus. Sequenase sequencing kit was from United States Biochemical Corporation, Cambridge Bioscience. Protein standards were purchased from Pharmacia LKB. All other enzymes were from Boehringer. Multiprime DNA labelling kit was from Amersham International. DEAE-Sephadex and Sepharose CL6B were from Pharmacia LKB. All radioactive isotopes were from Amersham. All other reagents were of the highest grade available. DNA amplification primers and sequencing primers were synthesised on an Applied Biosystems 381A DNA synthesiser.

Enzyme assays Citrate synthase was assayed spectrophotometrically at 412 nm and 25°C by the method of Srere et al. [28].

Purification of wild-type citrate synthase Wild-type E. coli citrate synthase was purified from the citrate-synthase-deficient strain of E. coli W620, transformed with the plasmid pDB2 [26], using the method of Else [31].

Preparation of antiserum against E. coli citrate synthase Production and selection of revertants Strain K114 was grown overnight in 200 ml minimal medium. The cells were harvested by centrifugation (10 min, 10000 g) and resuspended in 20 ml 100 mM Na+/K+ phosphate, pH 7.0. 200 pl ethyl methanesulphonate (Me,SO,Et) were added and the suspension incubated at 37 “C for 60 min. Cells from 1 ml suspension were collected by centrifugation and resuspended in 100 pl 100 mM phosphate buffer. Revertants that had regained citrate-synthase activity were then selected by virtue of their glutamate auxotrophy, by growth on glucose (10 mM) minimal-medium plates.

Estimation of citrate synthase M , values The molecular sizes of revertant citrate synthases were determined by gel filtration using a Superose 12 column connected to a Phamiacia-LKB FPLC system. 2 0 0 4 samples of cell extracts were loaded onto the column and elution was performed in 0.1 M TrisMCl, pH 8.0, 1 mM EDTA and 20% (by vol.) glycerol, at a flow rate of 0.3 mumin. 0.2 ml fractions were collected on ice and assayed for citrate synthase using 20% (by vol.) glycerol in the assay mixture.

A 10-ml test bleed was taken from a rabbit prior to immunisation with 300 pg purified wild-type E. coli citrate synthase in 50% Freunds adjuvant. Two booster injections of 200 pg protein in Freunds complete adjuvant were given at days 17 and 28. Bleedings were performed on day 35.

SDS/PAGE and immunoblotting analysis Cells were disrupted by sonication and the supernatant, after centrifugation, was analysed by SDS/PAGE with 10% acrylamide using the Laemmli system [32]. Protein bands were visualised by staining with Coomassie Blue. The bands were transferred to a nitrocellulose membrane using a semidry transfer cell (BioRad) as recommended by the manufacturer. After blocking with 3 % (mass/vol.) bovine serum albumin, the membrane was incubated with the polyclonal antiserum directed against wild-type E. coli citrate synthase. The second antibody was horseradish-peroxidase-conjugated to anti-Rabbit IgG, and immunoreactivity was detected with 3,3’-diaminobenzidine (0.7 mg/ml 50 mM Tris/HCl, pH 7.6) and 0.08% (by vol.) H,O,.

Protein microsequencing

I

DNA manipulations Plasmid DNA preparations, transformations of E. coli and agarose gel ele%phoretic analysis were performed by conventional methods [291. Restriction endonuclease diges-

20 pg protein were run on a 12.5% (mass/vol.) SDS/polyacrvlamide gel. The urotein was then transferred to Immobilo; poly(vit&lidene hifluoride) (Millipore) by electrophoresis. Microsequencing was performed on an Applied Biosys-

77 tems 470A gas-phase sequencer coupled to an Applied Biosystems 120 phenylthiohydantoin analyser.

RESULTS AND DISCUSSION Selection for revertants Following Me,SO,Et exposure of K114,41 glutamate revertants were obtained by growth on glucose (10 mM) minimal medium plates. Of these, 38 were shown to have regained citrate-synthase activity that was inhibited by 2-0x0glutarate (5 mM) in a manner characteristic of the wild-type enzyme. The citrate-synthase activity in the remaining three revertants was not inhibited by 2-oxoglutarate ; of these, the enzyme from one revertant, K114r4, was more stable than the other two and this was therefore selected for further investigation. The properties of K114r4 citrate synthase (hereafter referred to as the mutant citrate synthase, to distinguish it from the enzyme produced by the wild-type organism) are given in Table 1. The enzyme is clearly of the small type citrate synthases, both in size and kinetic properties. It should be noted that the levels of citrate synthase in K114r4 are greatly enhanced by growth on succinate as carbon source as opposed to glucose, although in the latter medium glutamate auxotrophy is still observed. K114 failed to grow in minimal medium with succinate as the energy source. Table 1. Comparison of wild-type (HB101) and mutant (K114r4) E. coli citrate synthases. Property

Molecular size (gel filtration) Substrate dependence Acetyl CoA K,,, or [SI,, ( W ) Oxaloacetate K,, or [SI,, ( W ) Inhibition by 2 mM 2-oxoglutarate (%) Inhibition by 0.2 mM NADH (%)

Citrate synthase wild type

mutant

=280,000 sigmoid

=110,000 hyperbolic

=200

11 ( 5 4)

-100

7 ( 2 2)

90

0

70

0

Southern blotting Southern blots of HindIIIIEcoRl restriction digests of chromosomal DNA from K114r4 and HBlOl were probed with the whole citrate synthase gltA gene probe. Autoradiographs showed that the probe bound to a single 3.2-kb fragment, corresponding to the citrate synthase gltA gene, on both K114r4 and HBlOl digests. This suggests that the gltA gene is still intact in K114r4 and, therefore, that the dramatic changes in characteristics of the enzyme have not been brought about by major alterations in that gene. Thus, a mutant citrate synthase has been produced from E. coli, possibly by minor genetic changes, that exhibits characteristics associated with a small-type enzyme. Cloning of the citrate synthase gene from K114r4 A 1.8-kb fragment of the gltA gene was amplified by the polymerase chain reaction (PCR) from K114r4 genomic

Fig.1. SDSPAGE of cell extracts of E. coli TG1 transformed with cloned gltA genes. SDSRAGE was performed using 10% acrylamide by the method of Laemmli [32]. Protein bands were visualised by staining in Coomassie Blue. The relative molecular masses (X and positions of standard proteins are shown to the left of the gel. The expression of a protein of approximate subunit M,= 45000 is indicated by the arrow. Cell extracts electrophoresed: E. coli TG1 (track 1); E. coli containing gltA from K114r4 (track 2); E. coli TG1 containing gltA from K114 (track 3); E. coli TG1 containing gltA from pDB2 (track 4).

DNA. A single product was produced which was purified, digested with Hind111 and EcoRl and cloned into HindIIV EcoRl digested pUC 18. The DNA amplification and cloning procedures were then repeated using DNA from the plasmid pDB2 and genomic DNA from E. coli Kll4. Amplified fragments were then sequenced. The sequence of the wild-type clone matched exactly the published sequence of the gltA gene [ l l ] and included the T to G correction at position 1170 [33]. The K114r4 gltA clone was found to contain a single point mutation, G to A at position 1392, resulting in the replacement of aspartate 362 with asparagine. The same base change was seen in a further two clones from the K114r4 amplification and in three clones sequenced from a second amplification of K114r4. Sequence analysis of the gltA gene from K114, the parent of K114r4, showed this same base substitution.

Expression of the cloned gltA genes Cell extracts from overnight cultures of TG1 and the TG1 clones containing the gltA gene from K114, K114r4 and wild-type E. coli were analysed by SDSPAGE (Fig. 1) and assayed for citrate-synthase activity. The levels of citratesynthase activity in the three transformants were compared with those obtained for TG1 alone. Clones of the 1.8-kb fragment from K114 and K114r4 gave specific activities of citrate synthase of 0.220 U/mg protein and 0.228 U/mg protein, respectively, both of which are comparable with that from TG1 alone, 0.243 U/mg protein. However, citrate-synthase activity in the wild-type clone was 0.657 U/mg protein, which represents a threefold amplification compared with the citrate-synthase activity of TG1. SDSPAGE shows that TG1 clones containing the gltA gene from K114r4 and K114 produce citrate synthase protein in amounts comparable to that seen in wild-type gltA clones and greater than that seen in TG1 alone; that is, as judged by densitometry of the protein migrating with a subunit M, of approximately 45 000. Therefore, it would appear that the mutation in the gltA gene has no effect on the production of

78 citrate synthase protein. However, this single base change causing the replacement of Asp362 with Asn has a profound effect on the activity of the enzyme as TG1 clones containing the gZtA gene from either K114 or K114r4 display citratesynthase-activity levels comparable with that observed from TG1 alone, despite the greater levels of expression. Multiple sequence alignment has shown that Asp362 of the E. coli citrate synthase is equivalent to Asp375 of pig heart citrate synthase and is highly conserved throughout all known citrate-synthase sequences. Moreover, this Asp residue has been implicated, from both crystallographic [20-23, 341 and site-directed mutagenesis [35-381 studies, as a key residue involved in the catalytic mechanism of citrate synthases. The course of catalysis by citrate synthase can be divided into three consecutive partial reactions : enoli.zation of oxaloacetate, condensation of oxaloacetate and acetylCoA to give citrylCoA, and hydrolysis of the citrylCoA to citrate and CoA. It is proposed that Asp362 is involved in both the enolization and hydrolysis steps of the reaction sequence, where, in concert with His274, it acts in general acid-base catalysis. However, K114r4 does display citrate-synthase activity, although the kinetic, regulatory and oligomeric properties of this enzyme are markedly different from the wild-type protein. As the gltA gene product appears to have been inactivated, it was proposed that the protein displaying citratesynthase activity is encoded by a gene distinct from the gltA gene and that this enzyme may be structurally dissimilar to wild-type citrate synthase. To test this hypothesis, a polyclonal antibody was raised against wild-type E. coli citrate synthase and used in immunological comparisons of E. coli strains K114, K114r4 and TG1.

A

0.5 10.4

- 0.3 Dot Blot Score

- 0.2

CS activity (0) U/mg protein

- 0.1 - 0.0 Fraction Number

B

6l 5

Dot Blot Score

37 39 41 43 45 47 49 51 53 55 57 Fraction Number

r 0.06 0.05

0.04

Immunoblotting An investigation of the immunoreactivity of the mutant citrate synthase in cell extracts by Western blotting would be hampered by the presence of co-synthesised wild-type enzyme. Therefore, antibody binding of the mutant protein was investigated after separation from the large wild-type enzyme by gel filtration. Gel filtration of 200 pl cell extract of K114r4 was performed using a Superose 12 column. Aliquots from each fraction were applied to duplicate nitrocellulose filters by the dot-blot method. The filters were then reacted with either pre-immune or post-immune serum and immuno-recognition was determined using a horseradish-peroxidase-conjugated anti-rabbit IgG and chromogenic development. The procedure was then repeated using cell extracts from TG1 and K114. The TG1 citrate-synthase elution profile and dot-blot score are shown in Fig. 2. Peak citrate-synthase activity was observed in fraction 43 (0.48 U/mg protein) and on analysis, immunoreactivity of the fractions mirrored citrate-synthase activity as expected. No reaction was seen when the filters were reacted with the pre-immune serum. Predictably, K114 did not display citrate-synthase activity in fractions eluted from the Sepharose 12 column. However, the dot blot gave positive results in a distribution that matched that of TG1 (Fig. 2). Thus, although K114 does not display citrate-synthase activity, it still produces the citrate synthase protein. This is consistent with the sequence data which showed that the promoter sequence of the K114 gltA gene was unaltered and that the only change involved the active-site residue Asp362 which is essential for activity. The

Blot Score

CS activity (0) U/mg protein

0.03 0.02 0.01 37 39

11 43 45 47 49 51 53 Fraction Number

55 57

Fig. 2. Gel filtration of wild-type and mutant E. coli cell extracts. Cell extracts of E. coli TG1 (A), K114 (B) and K114r4 (C) were gel filtered on Superose 12. Fractions were assayed for citrate-synthase activity (m) and were analysed by the dot-blot method for immunoreactivity with antibodies raised to the wild-type enzyme. The intensity of the immunoreactivity of each dot was scored on a relative scale of 1-5, 5 being the strongest reaction; the data are shown in the form of a histogram. CS, citrate synthase.

elution of citrate synthase from K114r4 was also as expected in that the peak of citrate-synthase activity (fraction 51) eluted after that from wild-type E. coli (Fig. 2). However, the peak of antibody binding (fraction 45) did not coincide with the peak of enzyme activity but was more consistent with the results obtained from the wild type. It would appear that, as in K114, the normal wild-type hexameric enzyme is being produced in K114r4 but is inactive due to the nucleotide substitution. Additionally, the data suggest quite clearly that K114r4 produces two citrate synthases and that it is therefore necessary to purify the active mutant enzyme from the inactive hexameric protein. There is little indication from these data of reactivity of the mutant citrate synthase with antiserum to wild-type, hexameric E. coli citrate synthase, although this cannot be stated with confidence without purified protein. However,

79 40

Pig 5

Bacillus Sp c4

40

5 Thermopl a sma a ci dophi 1 um

E.coli mutant

Fig. 3. N-terminal amino-acid sequence of the mutant citrate synthase and its alignment to other citrate-synthase sequences. The Nterminal amino-acid sequence of the mutant citrate synthase was aligned with known sequences of eukaryotic, eubacterial and archaebacterial citrate synthases using the University of Wisconsin GCG package [43] on a MicroVAX 3300 computer. Only sequence identities to the mutant enzyme are boxed. The numbers above the first residues shown indicate the position of that amino acid in the sequence of each particular enzyme.

immunoreactivity is clearly distinct from the wild-type protein in that the post-immune serum had no effect on the activity of the mutant enzyme (130 pg se rudml assay mixture) whereas, at the same antibody concentration, the wild-type citrate synthase was inhibited >95%.

Purification of the mutant citrate synthase from K114r4 The mutant citrate synthase was purified by a modification of the method used for the wild-type enzyme [31]. Glycerol (20%, by vol.) was included in all buffers to stabilise the enzymic activity. After precipitation of nucleic acids with protamine sulphate, the cell extract of K114r4 was chromatographed on DEAE-Sephadex with a salt gradient of 00.5 M NaCl in 20 mM Tris/HCl, pH 8, 1 mM EDTA. After desalting fractions containing citrate-synthase activity on a PD 10 column, the enzyme was chromatographed on a Pharmacia-LKB Mono Q column under the same conditions as used for DEAE-Sephadex. Finally, the enzyme was applied to a Pharmacia-LKB Mono S column in 50 mM Mes/ NaOH, pH 5.5, 1 mM EDTA, and eluted in the same buffer with a salt gradient of 0-0.5 M NaC1. It should be noted that, while this procedure is similar to that used to purify the wild-type enzyme, a higher pH (pH 8) was necessary to bind the mutant enzyme to the Mono Q column than is required for the wild-type citrate synthase (pH 7.3), emphasising the difference between the two enzymes. The enzyme recovered from the Mono S column had a specific activity of 1.94 U/mg protein, which represents a 62fold purification of the mutant citrate synthase. 10% SDS/ PAGE revealed two bands: a major band of Mr=46000 and a minor band of Mr=50000. By comparison with the data for the M , of the native enzyme (Table l),this would suggest that the mutant citrate synthase is a dimeric enzyme. Immunoreactivity of the mutant citrate synthase As the mutant citrate synthase had been purified, the immunoreactivity of the protein by Western blotting was investigated. 500 ng was loaded onto an SDS/polyacrylamide gel, electrophoresed and Western blotted. No immunoreactivity was observed with either band of the mutant citrate synthase whereas strong reactivity was seen with the wild-type citrate synthase (20 ng).

N-terminal sequencing Sequencing of the purified mutant citrate synthase was performed after transfer from SDS/polyacrylamide gels to Immobilon Poly(viny1idene difluoride). Both protein bands gave the same N-terminal sequence and it is assumed that they are products of the same protein, with the band of lower molecular mass arising due to degradation of the protein. The N-terminal sequence was aligned with the known citratesynthase sequences, and the alignment with that from pig (a eukaryote), Bacillus sp C4 (a Gram-positive eubacterium), E. coli (a Gram-negative eubacterium with a hexameric enzyme), R. prowasekii (a Gram-negative eubacterium with a dimeric enzyme) and Tp. acidophilum (an archaebacterium) is shown in Fig. 3. Several points can be made from the alignment; the first 18 amino acids of the mutant citrate synthase show very low similarities with the other sequences, although the identity is significant with the dimeric enzyme from Tp. acidophilum (33% identity); amino acids 19-37 of the mutant enzyme show significant similarity, particularly to the eubacterial and the archaebacterial sequences (33 60%), the greatest identity being with the Gram-negative, dimeric citrate synthase; if the alignment is correct, the mutant enzyme will be approximately 40 residues shorter at its N-terminus than the typical eukaryotic and eubacterial citrate synthases, as are the dimeric citrate synthases from Bacillus sp. C4 and Tp. acidophilum; consistent with the other data reported in this study, the mutant citrate synthase is clearly not a product of the gltA gene. A search of protein data banks did not reveal any other proteins homologous to the mutant citrate synthase N-terminal sequence.

Concluding remarks The results show that a mutant of E. coli, capable of growth in the absence of glutamate, has been created from a previously citrate-synthase-deficient strain that was unable to grow without glutamate supplementation. This mutant produces a protein that is capable of catalysing the same reaction as citrate synthase but displays different kinetic and molecular properties compared to the wild-type enzyme, differs immunologically from the native citrate synthase and is not encoded by the gltA gene. The question then arises as to the

80

nature of this mutant protein. There exist several possible explanations. The first of these is that the mutant protein is a second species of citrate synthase and the sequence alignment data would support this possibility. There are many instances of pairs or groups of isoenzymes in E. coli that catalyse the same reaction but can be distinguished from each other by their kinetic and molecular properties. For example, E. coli possesses three fumarase genes :fumA, expressed under aerobic conditions, fumB expressed during anaerobiosis, and fumC which is expressed constitutively [39]. Fumarases A and B are 90% identical in sequence, but neither show significant similarity to fumarase C. However, fumarase C is 60% identical to mammalian fumarases. Thus, multiple forms of an enzyme may be required to catalyse the same reaction but under different metabolic conditions. As yet, there is no evidence that two forms of citrate synthase co-exist in wild-type E. coli. However, the mutant citrate synthase has low activity which, if present, would be masked by the higher activity of the wild-type enzyme. Relatively little is known about the mechanism controlling the transcription of the gltA gene, although it is known that it is induced by aerobic conditions and growth on acetate, and repressed by glucose and anaerobiosis; also its synthesis is inversely proportional to growth rate [40]. It could be that a second gene exists that is induced under conditions different from those for the gltA gene and that has been mutated (e.g. in the promotor) such that it is now constituitively produced. The mutant protein appears to be under different control as its activity is severely depressed in the presence of glucose (this study and [31]). A second explanation is that the mutant citrate synthase may somehow be related to another enzyme. Citrate synthase catalyses the condensation of acetyl-CoA with oxaloacetate, a 2-0x0 acid. Several other enzymes, involved in amino-acid biosynthesis, catalyse a similar reaction and it has been proposed that these may all share a common ancestor [41]. As the catalytic mechanism may be similar in these cases, one could envisage that in their native state, or by minor modifications, this group of enzymes may be able to utilise different 2-0x0 acids, albeit to a lesser extent. Indeed, the activity of the mutant citrate synthase is considerably less than that of the wild-type enzyme. A third possibility is that the mutant citrate synthase is a product of a normally silent citrate synthase-type gene that has now been activated in the mutagenesis experiments. Of the three possibilities, the presence of a second citrate synthase gene, active or silent, would appear to be the most likely explanation of our data. In addition to the sequence alignments, this conclusion is supported by the fact that a dimeric citrate synthase can also be generated by in-vivo mutagenesis of another Gram-negative eubacterium, A. calcoaceticus l2.51, and that two distinct citrate synthases (a large and a small form) have been found in a mutant of P aeruginosa [42]. The determination of the N-terminal sequence of the mutant E. coli citrate synthase now enables us to isolate and sequence the coding gene and thereby distinguish the possible explanations forwarded. Furthermore, if the gene in question has indeed arisen by duplication and divergence from the gZtA gene (or vice-versa), then a comparison of the wild-type and mutant enzyme sequences may provide important information on the structural features of the citrate synthase subunit that determine whether it associates into a dimer or a hexamer.

We thank the Science and Engineering Research Council for a studentship (to AJP) and for a research grant (GRD93940; to MJD, DWH and Dr D J Osguthorpe) through the Molecular Recognition Initiative. We also thank Mrs J. Young and Dr M. Davison (ICI Pharmaceuticals Division, Macclesfield, U.K.) for carrying out the N-terminal sequencing, Dr A.J. Else and Dr A.J. Wolstenholme for valuable and constructive discussions, and Professor Sir Hans Kornberg (University of Cambridge) for the gift of the citrate-synthase-deficient strain of E. coli K12 [K114].

REFERENCES 1. Weitzman, P. D. J. & Danson, M. J. (1976) Curr. Top. Cell. Regul. 10, 161-204. 2. Weitzman, P. D. J. (1981) Adv. Microb. Physiol. 22, 185-244. 3. Danson, M. J. (1988) Adv. Microb. Physiol. 29, 165-231. 4. Evans, C. T., Owens, D. D., Simegi, B., Kispal, G. & Srere, P. A. (1988) Biochemistry 27, 4680-4686. 5. Bloxham, D. P., Parmlee, D. C., Kumar, S., Walsh, K. A. & Titani, K. (1982) Biochemistry 21, 2028-2036. 6. Bloxham, D. P., Parmlee, D. C., Kumar, S., Wade, R. D., Ericsson, L.H., Neurath, H., Walsh, K. A. & Titani, K. (1981) Proc. Natl Acad. Sci. USA 78, 5381 -5385. 7. Numata, O., Takemasa, T., Takagi, I., Hirono, M., Hirono, H., Chiba, J. & Watanabe, Y. (1991) Biochem. Biophys. Res. Commun. 174, 1028-1034. 8. Unger, E. A,, Hand, J. M., Cashmore, A. R. & Vasconcelos, A. C. (1989) Plant Mol. Biol. 13, 411 -418. 9. Suissa, M., Suda, K. & Schatz, G. (1984) EM60 J. 3, 17731781. 10. Bhayana, V. & Duckworth, H. W. (1984) Biochemistry 23, 2900-2905. 11. Ner, S. S., Bhayana, V., Bell, A. W., Giles, I. G., Duckworth, H. W. & Bloxham, D. P. (1983) Biochemistry 22,5243-5249. 12. Donald, L. J., Molgat, G. F. & Duckworth, H. W. (1989) J. Bacteriol. 171, 5542-5550. 13. Donald, L. J. & Duckworth, H. W. (1986) Biochem. Biophys. Res. Commun. 141, 797-803. 14. Fukaya, M., Takemura, H., Okumura, H., Kawamura, Y., Horinouchi, S. & Beppu, T. (1990) J. Bacteriol. 172, 2096-2104. 15. Schendel, F. J., August, P. R., Anderson, C. R., Hanson, R. S. & Flickinger, M. C. (1992) Appl. Environ. Microbiol. 58, 335345. 16. David, M., Lubinsky-Mink, S., Ben-Zui, A,, Suissa, M., Ultizur, S. & Kuhn, J. (1991) Biochem. J. 278, 225-234. 17. Heinzen, R. A., Frazier, M. E., Mallavia L. P. (1991) Gene 109, 63-69. 18. Wood, D. O., Williamson, L. R., Winkler, H. H. & Krause, D. C. (1987) J. Bacteriol. 169, 3564-3572. 19. Sutherland, K. J., Henneke, C. M., Towner, P., Hough, D.W. & Danson, M.J. (1990) Eur. J. Biochem. 194, 839-844. 20. Remington, S., Wiegand, G. & Huber, R. (1982) J. Mol. Biol. 158, 111-152. 21. Wiegand, G., Remington, S., Deisenhofer, J. & Huber, R. (1984) J. Mol. Bid. 174, 205-219. 22. Karpusas, M., Holland, D. & Remington, S. J. (1991) Biochemistry 30, 6024-6031. 23. Liao, D.-I., Karpusas, M. & Remington, S. J. (1991) Biochemistry 30, 6031-6036. 24. Danson, M. J., Harford, S. & Weitzman, P. D. J. (1979) Eus J . Biochem. 101, 515-521. 25. Weitzman, P. D. J., Kinghom, H. A., Beecroft, L. J. & Harford, S . (1978) Biochem. Soc. Trans. 6, 436-438. 26. Bloxham, D. P., Herbert, C. J., Ner, S. S. & Drabble, W. T. (1983) J. Gen. Microbiol. 129, 1889-1897. 27. Ashworth, J. M. & Komberg, H. L. (1966) Proc. Nntl Acad. Sci. USA 165, 179-188. 28. Srere, P. A., Brazil, H. & Gonen, L. (1963) Actu Chem. Scand. 17, 129-134. 29. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) in Molecular cloning: a laboratory manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor NY.

81 30. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl Acad. Sci. USA 74, 5463-5467. 31. Else, A. J. (1986) Ph.D. Thesis, University of Bath, U.K. 32. Laemmli, U. (1970) Nature 227, 680-685. 33. Anderson, D. H. & Duckworth, H. W. (1988) J. Biol. Chem. 263, 2163-2169. 34. Karpusas, M., Branchaud, B. & Remington, S. J. (1990) Biochemistrv 29. 2213-2219. 35. Handford,'P. A,, Ner, S. S., Bloxham, D. P. & Wilton, D. C. (1988) Biochim. Biophys. Acta 953, 232-240. 36. Man, W.-J., Li, Y., O'Connor, D. & Wilton, D. C. (1991) Biochem. J. 280, 521- 526. 37. Alter, G. M., Casazza, J. P., Zhi, W., Nemeth, P., Srere, P. A. & Evans, C. T. (1990) Biochemistry 29,7557-7563.

38. Kurz, L. C., Drysdale, G. R., Riley, M. C., Evans, C. T. & Srere, P. A. (1992) Biochemistry 31, 7908-7914. 39. Woods, S. A,, Schartzbach, S. D. & Guest, J. R. (1988) Biochim. Biophys. Acta 954, 14-26. 40 Smith, M. W. & Neidhart, F. C. (1983) J.Bacterio1. 154, 344350. 41. Jensen, R. A. (1976) Annu. Rev. Microbiol. 30, 409-425. 42. Solomon, M. & Weitzman, P. D. J. (1983) FEBS Lett. 155, 157-160. 43. Devereux, J., Haeberli, P. & Smithies, 0. (1984) Nucleic. Acids Res. 12, 387-395.