Wheat-germ aspartate transcarbamoylase: revised purification, stability and re-evaluation of regulatory kinetics in terms of the Monod-Wyman-Changeux model.

A revised and simplified purification scheme for aspartate transcarbamoylase (ATCase) from wheat-germ is reported, with an eightfold increase in scale (yielding approximately 10 mg of the pure protein from 4 kg of wheat-germ), and improved characteristics of stability and regulatory kinetics. The ATCase obtained is greater than 96% pure, as judged by polyacrylamide gel electrophoresis. The long-term stability (i.e. on a time-scale of several hours to weeks) of the activity of the purified enzyme, under various storage conditions, was investigated. At 4 degreesC and pH 7.5, stability was found to be strongly dependent on protein concentration (increased stability at high concentration), buffer concentration (decreased stability at high buffer concentration) and the inclusion of glycerol (increased stability with increasing glycerol concentration). The enzyme is routinely stored at 4 degreesC, in 0. 05 m Tris/HCl buffer containing 25% glycerol and at high protein concentration (approximately 1 mg.mL-1, or 10 microm in trimers). Under these conditions, the half-life of the enzyme activity is greater than 300 days. Over the time-scale of kinetic experiments (up to 20 min), the diluted activity (at around 1 nm of ATCase, in the presence of ligands) is completely stable. The specific activity remains constant in the range 0.1-10 nm, in the absence and presence of ligands, showing that dissociation of the trimeric enzyme into its subunits is negligible. Steady-state kinetics were examined using the enzyme at a concentration of 1.3 nm. Initial-rate curves for both allosteric ligands, carbamoylphosphate (CP) and uridine 5'-monophosphate (UMP), showed pronounced sigmoidicity, each in the presence of the other. In the absence of UMP, initial-rate curves for CP are hyperbolic. The initial rate data fit reasonably well to a trimeric Monod-Wyman-Changeux model, suggesting a two-state conformational mechanism, greatly favouring the active (R) state when both ligands are absent, in which the R-state binds CP exclusively (dissociation constant = 23.2 microm), and the T-state binds UMP exclusively (dissociation constant = 0.49 microm). This regulatory behaviour was found to be quite stable, and was indistinguishable from that of the enzyme in a freshly made crude extract, even after storage of the pure sample for 5 months. This enzyme preparation is therefore free of the anomalous allosteric kinetics produced by a previous purification scheme, in which the affinity for UMP was markedly reduced, CP rate curves showed no sigmoidicity, while UMP rate curves had sigmoidicity exaggerated by a low maximum.

Effects of lipids on nucleotide inhibition of wheat-germ aspartate transcarbamoylase: evidence of an additional level of control?

Wheat-germ aspartate transcarbamoylase, a monofunctional trimer, is strongly inhibited by uridine 5'-monophosphate (UMP), which shows kinetic interactions with the substrate, carbamoyl phosphate, suggesting a classical allosteric mechanism of regulation. Inhibition of the purified enzyme by UMP was amplified in the presence of a variety of ionic lipids at concentrations low enough to preclude denaturation. In the absence of UMP, most of these compounds had no kinetic effect or were slightly activating. Two phospholipids did not show the effect. In a homologous series of fatty acids (C6-C16), the potentiating effect was only seen with homologues greater than C8, reaching a maximum at C12. The effect of dodecanoate (C12) on kinetic cooperativity (UMP as variable ligand) was studied. At each of several fixed concentrations of carbamoyl phosphate, dodecanoate had a pronounced effect on the half-saturating concentration of UMP, which was reduced by about half in every case, indicating substantially tighter binding of UMP. However, dodecanoate had relatively little effect on the kinetic Hill coefficient for the cooperativity of UMP. The possible metabolic significance of these effects is discussed.

Endogenous polypeptide-chain length and partial sequence of aspartate transcarbamoylase from wheat, characterised by immunochemical and cDNA methods.

Aspartate transcarbamoylase (ATCase) is purified from wheat germ as a monofunctional trimer of 36 kDa chains. The possibility that this may be a proteolytic fragment of a large endogenous complex in which ATCase is covalently fused to other pyrimidine-pathway enzymes (such as exists in animals or fungi) was tested. Examination of a rabbit antiserum raised against the purified enzyme confirmed the presence of anti-(wheat ATCase) antibodies by several independent methods. Extracts of wheat seedlings prepared under non-proteolysing conditions were challenged by the antiserum, and in some cases by purified anti-(36 kDa ATCase) antibodies, using immunoblotting techniques. The 36 kDa species was the dominant immunopositive polypeptide. However, the extract also contained small amounts of two larger (45 and 55 kDa) immunopositive polypeptides, as well as traces of polypeptides smaller than 36 kDa, which were assumed to be minor proteolytic products. Neither of the 45 or 55 kDa polypeptides is large enough to also incorporate a carbamoyl phosphate synthetase or dihydroorotase of the sizes found in other organisms. They may be targeted precursors of ATCase with intact leader sequences. A screen of a wheat cDNA expression library by the anti-(ATCase) serum yielded a single positive clone which was shown, by DNA sequencing, to be a concatemer of two cDNAs, one of which encoded a partial ATCase. Northern analysis using this clone as probe identified two transcripts of about 1.3 and 1.0 kbp, but showed no evidence of a transcript of 2 kbp or greater which would be required to encode a bifunctional polypeptide. These results confirm that wheat ATCase is not translationally fused to dihydroorotase or carbamoylphosphate synthetase, as it is in animals and fungi. The deduced amino-acid sequence of the partial wheat ATCase is compared with the catalytic chain of the ATCase from Escherichia coli and with other ATCases.

Inactivation of wheat-germ aspartate transcarbamoylase by the triazinyl dye, procion red HE3B.

Aspartate transcarbamoylase from wheat germ is irreversibly inactivated by the triazinyl dye Procion Red HE3B. Since triazinyl dyes may mimic nucleotides, and UMP is a known allosteric modifier of this enzyme, the reaction was studied to elucidate whether the dye is an 'affinity label' for the enzyme. The reaction is apparently first order in the first 5-10 min, but is more complex in the longer term and does not go to completion. Kinetic analysis of the initial phase suggests that there are two parallel reactions, one saturable (dye binds reversibly before reaction) and one non-saturable (biomolecular). The apparent rate constant kapp (i.e. the sum of the rate constants for the parallel reactions) varies only slightly over the pH range 7-10. In the presence of a number of active centre ligands, as well as the allosteric ligand UMP, there is a clear increase in kapp. This finding is contrary to the reduction in rate of inactivation (protection) normally provided by ligands against active-site directed reagents, suggesting that in the saturable reaction, there is a conformational change upon dye-binding that increases the exposure of the essential residue(s) with which the dye reacts. These results show that, although it probably inactivates by reaction with specific amino-acid residues, the dye is not bound at the substrate-binding or allosteric sites, i.e. it is not an affinity-labeling reagent in the usual sense.

Carrot cells detoxify N-phosphonoacetyl-L-aspartate by esterification.

Unlike bacterial and mammalian cells, carrot cells are able to tolerate N-phosphonoacetyl-L-aspartate (PALA), a potential inhibitor of pyrimidine biosynthesis, by detoxifying the compound. Anion-exchange chromatography showed that detoxified PALA was less negatively charged than PALA, and allowed detoxified PALA to be isolated. Incubation of detoxified PALA with a low-specificity carboxylic-ester hydrolase fully restored the ability to inhibit aspartate transcarbamoylase, the target enzyme, indicating that the detoxification involves the formation of carboxylic ester. G.1.c. analysis of the alcohol products of enzymic hydrolysis, and of their ratio to PALA, showed that the detoxification produced a mixture of mono- and di-carboxylic esters and of methyl and ethyl esters. The detoxification mechanism showed considerable specificity towards PALA, since the analogous carboxy groups of succinate were not modified in the same way. Succinate was depleted much more slowly, no succinate esters could be detected, and the presence of a 10-fold excess of succinate did not inhibit the esterification rate of PALA. The possible significance of these results is discussed.

Active-site-directed inactivation of wheat-germ aspartate transcarbamoylase by pyridoxal 5'-phosphate.

Treatment of 1 microM wheat-germ aspartate transcarbamoylase with 1 mM-pyridoxal 5'-phosphate caused a rapid loss of activity, concomitant with the formation of a Schiff base. Complete loss of activity occurred within 10 min when the Schiff base was reduced with a 100-fold excess of NaBH4. Concomitantly, one amino group per chain was modified. No further residues were modified in the ensuing 30 min. The kinetics of inactivation were examined under conditions where the Schiff base was reduced before assay. Inactivation was apparently first-order. The pseudo-first-order rate constant, kapp., showed a hyperbolic dependence upon the concentration of pyridoxal 5'-phosphate, suggesting that the enzyme first formed a non-covalent complex with the reagent, modification of a lysine then proceeding within this complex. Inactivation of the enzyme by pyridoxal was 20 times slower than that by pyridoxal 5'-phosphate, indicating that the phosphate group was important in forming the initial complex. Partial protection against pyridoxal phosphate was provided by the leading substrate, carbamoyl phosphate, and nearly complete protection was provided by the bisubstrate analogue, N-phosphonoacetyl-L-aspartate, and the ligand-pair carbamoyl phosphate plus succinate. Steady-state kinetic studies, under conditions that minimized inactivation, showed that pyridoxal 5'-phosphate was also a competitive inhibitor with respect to the leading substrate, carbamoyl phosphate. Pyridoxal 5'-phosphate therefore appears to be an active-site-directed reagent. A sample of the enzyme containing one reduced pyridoxyl group per chain was digested with trypsin, and the labelled peptide was isolated and shown to contain a single pyridoxyl-lysine residue. Partial sequencing around the labelled lysine showed little homology with the sequence surrounding lysine-84, an active-centre residue of the catalytic subunit of aspartate transcarbamoylase from Escherichia coli, whose reaction with pyridoxal 5'-phosphate shows many similarities to the results described in the present paper. Arguably the reactive lysine is conserved between the two enzymes whereas the residues immediately surrounding the lysine are not. The same conclusion has been drawn in a comparison of reactive histidine residues in the two enzymes [Cole & Yon (1986) Biochemistry 25, 7168-7174].

Comparison of aspartate transcarbamoylases from wheat germ and Escherichia coli: functionally identical histidines in nonhomologous local sequences.

Aspartate transcarbamoylase (ATCase) from wheat germ and the catalytic subunit of the enzyme from Escherichia coli are trimers of similar size. The former is a regulatory enzyme in its trimeric state, while the latter is a component of a complex regulatory dodecamer. In a comparison of the two enzymes, reaction with diethyl pyrocarbonate revealed a highly active, essential histidine residue in each case. The two histidines (i.e., one in each enzyme) behaved nearly identically with respect to the following functional properties: kinetics of acylation (ethoxyformylation) and concomitant inactivation; kinetics of deacylation by hydroxylamine and concomitant reactivation; hyperbolic dependence of the apparent first-order rate constant (kapp) on diethyl pyrocarbonate concentration; pH dependence of kapp; failure of active-center ligands to protect the residue against diethyl pyrocarbonate, producing instead near-identical increases in the inactivation rate. These similarities point to an essential, highly conserved histidine in each enzyme, in a functional microenvironment that has changed relatively little since the divergence of plants and bacteria. Ethoxyformylated peptides were isolated from tryptic digests of the two inactivated enzymes. Sequencing of the major labeled peptide in each case showed the wheat and E. coli histidines embedded in nonhomologous primary segments, suggesting that, contrary to expectation, these segments are not part of the conserved microenvironment. In the case of the E. coli enzyme, the essential residue was identified as His-134 in the known sequence, which has a potential catalytic role on crystallographic evidence [Krause, K. L., Volz, K. W., & Lipscomb, W. N. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 1643-1647]. A second, much less reactive histidine was identified as His-64.(ABSTRACT TRUNCATED AT 250 WORDS)

Inactivation of wheat-germ aspartate transcarbamoylase by the arginine-specific reagent phenylglyoxal.

Wheat-germ aspartate transcarbamoylase (EC 2.1.3.2) was inactivated by phenylglyoxal in a first-order process, provided that the inactivation time did not exceed 10 min. Apparent first-order rate constants were linearly dependent on phenylglyoxal concentration, indicating a bimolecular reaction between a single active-centre residue and phenylglyoxal, with second-order constant of 0.023 mM-1 X min-1. A plot of apparent first-order rate constant versus pH showed a steep rise above pH 9.5, indicating that the essential residue has a pKa value of 10.5 or higher, consistent with an arginine residue. Saturating concentrations of the following ligands provided a degree of protection (percentages in parentheses) against 1 mM-phenylglyoxal: N-phosphonoacetyl-L-aspartate, a bisubstrate analogue (94%); carbamoyl phosphate (75%); UMP, an end-product inhibitor (53%). Succinate (an analogue of L-aspartate) alone gave no protection, but in combination with carbamoyl phosphate raised the protection to 92%, in agreement with the known binding order of the two substrates. These results indicate that the essential arginine residue is close to the carbamoyl phosphate site, probably oriented towards the aspartate site. Attempts to desensitize the UMP-binding site by reaction with phenylglyoxal, while protecting the active centre, were unsuccessful. The essential active-centre arginine residue is compared with a similar residue in the Escherichia coli enzyme.

Detoxification of N-(phosphonoacetyl)-L-aspartate by carrot cells in suspension culture.

In bacterial and mammalian cells, N-(phosphonoacetyl)-L-aspartate (PALA) suppresses growth by strongly inhibiting aspartate transcarbamoylase (ATCase; EC 2.1.3.2), a key enzyme of the pyrimidine biosynthetic pathway. At a concentration that would suppress growth in mammalian or bacterial cells, and that is nearly a million-fold greater than the inhibition constant (K i ) for ATCase in carrot (Daucus carota) seedling extracts, PALA does not suppress growth of carrot cells in suspension culture. To study this anomaly an assay based on the inhibition of wheat (Triticum vulgare) ATCase (K i =2 nM) was developed. Using this assay it was found that PALA is detoxified relatively rapidly by low inocula of carrot cells. The detoxification product accumulates in the extracellular fluid although the enzyme(s) responsible is intracellular or in the cell wall. The PALA-detoxifying activity can be detected at all stages of the growth cycle in culture, but reaches a maximum early in the exponential phase of growth. Cells that were repeatedly subcultured into media initially containing 1 mM PALA had the same low level of ATCase activity as control cells; there was no evidence of the amplification of the gene for this enzyme, such as occurs in mammalian cells upon repeated exposure to the drug.

Rapid purification of 2,4-dichlorophenol hydroxylase by biospecific desorption from 10-carboxydecylamino-sepharose.

10-Carboxydecylamino-Sepharose, which bears a mixture of ionic and aliphatic substituent groups, adsorbs 2,4-dichlorophenol hydroxylase from Acinetobacter in a non-biospecific manner. The enzyme has been specifically desorbed by its substrate, 2,4-dichlorophenol, giving a 42-fold purification (to greater than 90% purity) in a single step. The enzyme contained 3.1 moles of FAD per mole and displayed a catalytic constant of 14.7 s(-1). Mixed-function adsorbents probably have wide applicability for biospecific desorption of proteins. The present report indicates that they may be useful in the purification of aromatic hydroxylases bearing flavin prosthetic groups that readily dissociate in conventional purification procedures employing conditions of high ionic strength.

Ligand-mediated conformational changes in wheat-germ aspartate transcarbamoylase indicated by proteolytic susceptibility.

Ligand-mediated effects on the inactivation of pure wheat-germ aspartate transcarbamoylase by trypsin were examined. Inactivation was apparently first-order in all cases, and the effects of ligand concentration on the pseudo-first-order rate constant, k, were studied. Increase in k (labilization) was effected by carbamoyl phosphate, phosphate and the putative transition-state analogue, N-phosphonoacetyl-L-aspartate. Decrease in k (protection) was effected by the end-product inhibitor, UMP, and by the ligand pairs aspartate/phosphate and succinate/carbamoyl phosphate, but not by aspartate or succinate alone up to 10 mM. Except for protection by the latter ligand pairs, all other ligand-mediated effects were also observed on inactivation of the enzyme by Pronase and chymotrypsin. Ligand-mediated effects on the fragmentation of the polypeptide chain by trypsin were examined electrophoretically. Slight labilization of the chain was observed in the presence of carbamoyl phosphate, phosphate and N-phosphonoacetyl-L-aspartate. An extensive protection by UMP was observed, which apparently included all trypsin-sensitive peptide bonds. No significant effect by the ligand pair succinate/carbamoyl phosphate was noted. It is concluded from these observations that UMP triggers an extensive, probably co-operative, transition to a proteinase-resistant conformation, and that carbamoyl phosphate similarly triggers a transition to an alternative, proteinase-sensitive, conformation. These antagonistic conformational changes may account for the regulatory kinetic effects reported elsewhere [Yon (1984) Biochem. J. 221, 281-287]. The protective effect by the ligand pairs aspartate/phosphate and succinate/carbamoyl phosphate, which operates only against trypsin, is concluded to be due to local shielding of essential lysine or arginine residues in the aspartate-binding pocket of the active site, to which aspartate (or its analogue, succinate) can only bind as part of a ternary complex.

Regulatory kinetics of wheat-germ aspartate transcarbamoylase. Adaptation of the concerted model to account for complex kinetic effects of uridine 5'-monophosphate.

The kinetic effects of the end-product inhibitor UMP on aspartate transcarbamoylase (EC 2.1.3.2) purified to homogeneity from wheat germ were studied. In agreement with an earlier study of the relatively crude enzyme [Yon (1972) Biochem. J. 128, 311-320], the half-saturating concentrations of UMP and of the first substrate, carbamoyl phosphate (but not of the second, L-aspartate), were found to be strongly interdependent. However, the kinetic behaviour of the pure enzyme differed from that of the crude enzyme in several important respects, namely: (a) the apparent affinity for UMP was lower with the pure enzyme; (b) sigmoidicity was absent from plots of initial rate versus carbamoyl phosphate concentration, each at a fixed UMP concentration; (c) sigmoidicity was greatly exaggerated in plots of initial rate versus UMP concentration, each at a fixed carbamoyl phosphate concentration, owing to the occurrence of a slight but definite maximum in each plot at low UMP concentration; (d) there was a relative increase in this maximum in the presence of N-phosphonacetyl-L-aspartate, an inhibitor competitive with carbamoyl phosphate. It is shown that a modified two-conformation concerted-transition model can be used to account for most of these features of the pure enzyme. The model treats carbamoyl phosphate and UMP as antagonistic allosteric ligands binding to alternative conformational states [Monod, Wyman & Changeux (1965) J. Mol. Biol. 12, 88-118], carbamoyl phosphate binding non-exclusively (dissociation constants 20 microM and 85 microM respectively) and UMP binding exclusively (dissociation constant 2.5 microM). The model postulates further that the conformation with lower affinity for carbamoyl phosphate has the higher value of kcat., and that it binds UMP in competition with carbamoyl phosphate. Parameters giving the best fit of experimental data to this model were found by a non-linear least-squares search procedure.

A quantitative study of the biospecific desorption of rat liver (M4) lactate dehydrogenase from 10-carboxydecylamino-Sepharose. Determination of the number of ligand-binding sites blocked on adsorption.

1. The theory of Nichol, Ogston, Winzor & Sawyer [(1974) Biochem. J. 143, 435-443] for quantitative affinity chromatography, when adapted for use with a non-specific column from which a multi-site protein can be specifically desorbed by its free ligand, permits determination of the concentration of adsorption sites on the column, their adsorptive affinity (as an association constant) and either the intrinsic (site) constant for ligand-binding to the protein or an 'occlusion coefficient' (defined as the number of ligand-binding sites blocked on adsorption), one of which must be known. 2. The theory has been applied to the NADH-specific desorption of rat liver M4 lactate dehydrogenase from 10-carboxydecylamino-Sepharose. It suggests that most of the enzyme molecules are adsorbed with at least two NADH-binding sites blocked, indicating an extensive adsorption interface in relation to the protein surface. Other chromatographic parameters were also determined for the system. 3. Among topics discussed are (a) factors affecting the experimentally determined value for the number of blocked sites, (b) the nature of the adsorption sites on the column and (c) the similarity of the analysis to that for determining Hill coefficients, and other possible applications.

The quaternary structure of wheat-germ aspartate transcarbamoylase.

1. The molecular mass of aspartate transcarbamoylase purified from wheat germ was found to be 101kDa by sucrose-density-gradient centrifugation, 103kDa by gel-filtration chromatography and 108kDa by polyacrylamide-gel electrophoresis. A mean value of 104 +/- 11kDa was obtained by pooling several replicate results from each method. 2. Polyacrylamide-gel electrophoresis in the presence of sodium dodecyl sulphate indicated a single size of polypeptide chain of mean molecular mass 37 +/- 4kDa. The ratio of the mean molecular masses of the active and denatured enzymes is 2.8.3. When the active enzyme was covalently cross-linked at a low protein concentration by dimethyl suberimidate, and then examined electrophoretically under denaturing conditions, three size species were observed to predominate, of apparent molecular masses 36, 77 and 106kDa respectively. 4. These results indicate that the intact, fully regulatory enzyme is a simple trimer, slightly larger than the trimeric "catalytic subunit' of the aspartate transcarbamoylase from Escherichia coli [Weber (1968) Nature (London) 218, 1116-1118]. The prevalence of trimeric structures amongst carbamoyl-transferase enzymes is discussed.

Sensitivity and other factors affecting biospecific desorption in chromatography of proteins. A study by computer simulation.

Some theoretical aspects of the desorption of a column-bound protein by elution with its biospecific ligand are considered in cases where, in comparison with the unliganded protein, the protein-ligand complex has a diminished but finite affinity for the adsorbent. A quantity termed the biospecific sensitivity, B, is introduced to facilitate comparison between different systems. Biospecific sensitivity may be defined as the fractional change in standard free energy of adsorption on formation of the protein-ligand complex. The effects of a moderate-to-low biospecific sensitivity on theoretical desorption profiles have been examined by using a computer simulation of the classical multiple-plate column model. Desorption was simulated under various boundary conditions involving protein-adsorbent and protein-ligand affinities and the initial concentrations of adsorption sites, protein and ligand. These simulations suggest that, when the biospecific sensitivity is low, desorption is optimized if (a) the unliganded protein is adsorbed as weakly as possible, (b) the column is loaded to near-saturation with the required protein, (c) the free ligand concentration is many times greater than that giving near-saturation of the protein in free solution, and (d) protein contaminants with high affinity for the adsorbent, and present in large amount, are removed in preliminary purification steps.

Wheat-germ aspartate transcarbamoylase. Purification and cold-lability.

1. Aspartate transcarbamoylase was purified approx. 3000-fold from wheat (Triticum vulgare) germ in 15-20% yield. The product has a specific activity of 14 mumol/min per mg of protein and is approx. 90% pure. The purification scheme includes the use of biospecific "imphilyte" chromatography as described by Yon [Biochem.J.(1977) 161, 233-237]. The enzyme was passed successively through columns of CPAD [N-(3-carboxypropionyl)aminodecyl]-Sepharose in the absence and presence respectively of the ligands UMP and L-aspartate. In the second passage the enzyme was specifically displaced away from impurities with which it co-migrated in the first passage. These two steps contributed a factor of 80 to the overall purification. 2. The enzyme is slowly inactivated on dilution at 0 degrees C and pH 7.0, the inactivation being partially reversible. A detailed investigation of the temperature- and pH-dependence of the cold-inactivation suggested that it was initiated by the perturbation of the pKa values of groups with a moderately high and positive heat of ionization, which were tentatively identified as histidine residues. These findings support a new concept of cold-lability proposed by Bock, Gilbert & Frieden [Biochem. Biophys. Res. Commun. (1975) 66, 564-569].

Wheat-germ aspartate transcarbamoylase. Steady-state kinetics and stereochemistry of the binding site for L-aspartate.

1. The steady-state kinetics of the bisubstrate reaction catalysed by aspartate transcarbamoylase purified from wheat (Triticum vulgare)-germ have been studied at 25 degrees C, pH 8.5 AND I 0.10-0.12. Initial-velocity and product-inhibition results are consistent with an ordered sequential mechanism in which carbamoyl phosphate is the first substrate to bind, followed by L-aspartate, and carbamoyl aspartate is the first product to leave, followed by Pi. The order of substrate addition is supported by dead-end inhibition studies using pyrophosphate and maleate as inhibitory analogues of the substrates. Product inhibition permitted a minimum value for the dissociation constant of L-aspartate from the ternary complex to be estimated. This minimum is of the same order as the dissociation constant (Ki) of succinate. 2. A range of dicarboxy analogues of L-aspartate were tested as possible inhibitors of the enzyme. These studies suggested that L-aspartate is bound with its carboxy groups in the eclipsed configuration, and that the stereochemical constraints around the binding site are very similar to those reported for the catalytic subunit of the enzyme from Escherichia coli [Davies, Vanaman & Stark (1970) J. Biol. Chem. 245, 1175-1179].