The catalytic consequences of experimental evolution. Transition-state structure during catalysis by the evolved beta-galactosidases of Escherichia coli (ebg enzymes) changed by a single mutational event

Hexose/Pentose and Hexitol/Pentitol Metabolism

Escherichia coli and Salmonella enterica serovar Typhimurium exhibit a remarkable versatility in the usage of different sugars as the sole source of carbon and energy, reflecting their ability to make use of the digested meals of mammalia and of the ample offerings in the wild. Degradation of sugars starts with their energy-dependent uptake through the cytoplasmic membrane and is carried on further by specific enzymes in the cytoplasm, destined finally for degradation in central metabolic pathways. As variant as the different sugars are, the biochemical strategies to act on them are few. They include phosphorylation, keto-enol isomerization, oxido/reductions, and aldol cleavage. The catabolic repertoire for using carbohydrate sources is largely the same in E. coli and in serovar Typhimurium. Nonetheless, significant differences are found, even among the strains and substrains of each species. We have grouped the sugars to be discussed according to their first step in metabolism, which is their active transport, and follow their path to glycolysis, catalyzed by the sugar-specific enzymes. We will first discuss the phosphotransferase system (PTS) sugars, then the sugars transported by ATP-binding cassette (ABC) transporters, followed by those that are taken up via proton motive force (PMF)-dependent transporters. We have focused on the catabolism and pathway regulation of hexose and pentose monosaccharides as well as the corresponding sugar alcohols but have also included disaccharides and simple glycosides while excluding polysaccharide catabolism, except for maltodextrins.

The chemical synthesis of 2-deoxy-2-fluorodisaccharide probes of the hen egg white lysozyme mechanism

2,4-Dinitrophenyl 2-acetamido-2-deoxy-beta-d-glucopyranosyl-(1-->4)-2-deoxy-2-fluoro-beta-d-glucopyranoside (GN2FG-DNP) and 2-acetamido-2-deoxy-beta-d-glucopyranosyl-(1-->4)-2-deoxy-2-fluoro-beta-d-glucopyranosyl fluoride (GN2FG-F) were prepared using a divergent synthetic approach involving 10 steps. The key steps involved the preparation of 1-O-acetyl-3,6-di-O-benzyl-2-deoxy-2-fluoro-alpha/beta-d-glucopyranose using Selectfluor(trade mark) in the presence of acetic acid and the subsequent glycosylation of this acceptor to generate the core 2-fluorodisaccharide. After further elaboration, the target molecules were obtained and tested as probes of the mechanism of hen egg white lysozyme (HEWL). Compound GN2FG-DNP is not a substrate for the enzyme while compound GN2FG-F is cleaved slowly with an apparent K(m) greater than 5mM and a second-order rate constant of k(cat)/K(m)=9.6s(-1)M(-1). Comparison of this value to that estimated for the hydrolysis of beta-chitobiosyl fluoride by HEWL (1200s(-1)M(-1)) [Ballardie, F. W.; Capon, B.; Cuthbert, M. W.; Dearie, W. M. Bioorg. Chem.1977, 6, 483-509] revealed a 126-fold rate decrease upon substitution of a fluorine group for the 2-acetamido group of beta-chitobiosyl fluoride. This decrease resulted in the steady-state accumulation of an intermediate as visualized by mass spectrometry and the ultimate crystallographic determination of its structure [Vocadlo, D. J.; Davies, G. J.; Laine, R.; Withers, S. G. Nature2001, 412, 835-838].

Experimental evolution of Ebg enzyme provides clues about the evolution of catalysis and to evolutionary potential

The ebg (evolved beta-galactosidase) operon of Escherichia coli has been used since 1974 as a model system to dynamically study the evolutionary processes which have led to catalytic efficiency and substrate specificity in enzymes. Wild-type ebg beta-galactosidase, encoded by ebgA, is a catalytically feeble enzyme that does not hydrolyze lactose or other beta-galactosidase efficiently enough to permit growth on those substrates. Each of two specific base substitutions at widely separated sites increases catalytic activity sufficiently to permit growth, and the combination of the two mutations further increases catalytic effectiveness and expands the substrate range of the enzyme in a non-additive fashion. Experimental studies suggested that in the 3126 bp coding region those two substitutions were the only mutations capable of increasing activity toward lactose sufficiently to permit growth. Alignment of EbgA with the LacZ beta-galactosidase showed that both mutations were in active site amino acids. Multiple alignment and phylogenetic analysis of EbgA, LacZ, and 12 other related beta-galactosidases showed that EbgA and LacZ diverged from a common ancestor at least 2.2 billion years ago, that they belonged to different subclasses of the family of 14 beta-galactosidases, that the two subclasses differed at 12 of the 15 active site residues, and confirmed that the two previously identified mutations in ebgA are the only ones that can lead to enzyme with sufficient activity on lactose to permit growth. Studies of the catalytic mechanism of Ebg beta-galactosidase have allowed the widely accepted Albery and Knowles model for the evolution of catalysis to be rejected.

Larger increases in sensitivity to paracatalytic inactivation than in catalytic competence during experimental evolution of the second beta-galactosidase of Escherichia coli

Second-order rate constants (M-1.s-1) at 25 degrees C and pH 7.5 for inactivation of first-generation (ebga and ebgb), second-generation (ebgab and ebgabcd) and third-generation (ebgabcde) experimental evolvants of the title enzyme by 2',4'-dinitrophenyl 2-deoxy-2-fluoro-beta-D-galactopyranoside are 0.042, 0.30, 10, 24 and 57 respectively. Only partial inactivation is observed, except for ebgabcde. At a single high inactivator concentration, inactivation of the wild-type ebgo is also seen. The changes in sensitivity to the paracatalytic inactivator (over a range of 10(3.3)) are larger than changes in kcat/Km for lactose (over a range of 10(2.7)) or nitrophenyl galactosides (over a range of only 10(1.3)), or changes in degalactosylation rate (over a range of 10(1.7)). These data raise the possibility that evolution in the reverse sense, towards insensitivity to a paracatalytic inactivator with a proportionally lower effect on transformation of substrate, may become a mechanism for the development of bacterial resistance to antibiotics that act by paracatalytic enzyme inactivation.

Catalytic consequences of experimental evolution: catalysis by a 'third-generation' evolvant of the second beta-galactosidase of Escherichia coli, ebgabcde, and by ebgabcd, a 'second-generation' evolvant containing two supposedly 'kinetically silent' mutations

The kinetics of hydrolysis of a series of synthetic substrates by two experimentally evolved forms ('evolvants'), ebgabcd and ebgabcde, of the second beta-galactosidase of Escherichia coli have been measured. The ebgabcd enzyme differs from the wild-type (ebgo) enzyme by Asp92-->Asn (a) and Trp977-->Cys (b) changes in the large subunit, as well as two changes hitherto considered to have no kinetic effect, Ser979-->Gly in the large subunit (c) and Glu122-->Gly in the small subunit (d). The enzyme ebgabcde contains in addition a Glu93-->Lys change in the large subunit (e). Comparison of ebgabcd with ebgab [Elliott, K, Sinnott, Smith, Bommuswamy, Guo, Hall and Zhang (1992) Biochem. J. 282, 155-164] indicates that the c and d changes in fact accelerate the hydrolysis of the glycosyl-enzyme intermediate by a factor of 2.5, and also decrease the charge on the aglycone oxygen atom at the first transition state; the charge on the glycone, however, is unaltered [see K, Konstantinidis, Sinnott and Hall (1993) Biochem. J. 291, 15-17]. The e mutation causes a fall in the degalactosylation rate of about a factor of 3, and its occurrence only together with c and d mutations [Hall, Betts and Wootton (1989) Genetics 123, 635-648] suggests that degalactosylation of a hypothetical ebgabe enzyme would be so slow that the enzyme would have no biological advantage over the ancestral ebgab. The transfer products from galactosyl-ebgabcd and galactosyl-ebgabcde to high concentrations to glucose have been measured; the predominant product is allolactose, but significant quantities of lactose are also formed; however, at apparent kinetic saturation of the galactosyl-enzyme, hydrolysis rather than transfer is the preponderant pathway. A knowledge of the rates of enzyme-catalysed exchange of 18O from [1-18O]galactose to water permits the construction of the free-energy profiles for hydrolysis of lactose by begabcd and ebgabcde. As with the other evolvants, changes in the profile away from the rate-determining transition state are essentially random, and there is no correlation between the changes in the free energies of intermediates and of their flanking transition states. We consider the aggregate of our kinetic data on the ebg system to be telling experimental support for the theoretical objections of Pettersson [Pettersson (1992) Eur. J. Biochem. 206, 289-295 and previous papers] to the Albery-Knowles theory of the evolution of enzyme kinetic activity.

Catalysis by the large subunit of the second beta-galactosidase of Escherichia coli in the absence of the small subunit

Plasmids containing the ebgAo and ebgAa genes of Escherichia coli under the control of the lac repressor and promoter have been constructed and inserted into Salmonella typhimurium CH3. This system expresses the large subunit of the ebgo and ebga beta-galactosidase in high yield (20-60% of total protein). The large subunits have been purified to homogeneity. As isolated they are tetramers of significant catalytic activity; the N-terminal amino acid residue is Met, but it is not formylated. The kcat. values for a series of aryl galactosides were 6-200-fold reduced from the corresponding values for the holoenzymes. kcat/Km Values for glycosides of acidic aglycones, though, were unchanged, whilst kcat./Km values for galactosides of less acidic aglycones showed a modest (up to 10-fold) decrease. The kcat. values for glycosides of acidic aglycones hydrolysed by ebgo and ebga large subunits were essentially invariant with aglycone pK, suggesting that hydrolysis of the galactosyl-enzyme intermediate had become rate-determining for these substrates. Rate-determining hydrolysis of the glycosyl-enzyme intermediate was confirmed by pre-steady-state measurements and nucleophilic competition with methanol. Absence of the small subunit was thus estimated to cause a 200-fold decrease in degalactosylation rate for ebgo and a 20-fold one for ebga. beta 1g(V/K) values of -0.57 +/- 0.08 for ebgo and -0.54 +/- 0.08 for ebga isolated subunits were significantly more negative than for holoenzymes. It is suggested that the small subunit is associated with the optimal positioning of the electrophilic Mg2+ ions in these enzymes. Use of PCR in the construction of the plasmid also inadvertently led to the production of psi ebgo large subunit in which there was a PCR-introduced Leu9-->His change. Values of kcat. for aryl galactosides, calculated on the assumption that the psi ebgo large subunit, like the ebgo and ebga large subunits, was 100% active as isolated, were about an order of magnitude lower than for true ebgo large subunit, whilst Km values were similar. The very significant kinetic effect of this inadvertant site-undirected mutagenesis indicates that quite large kinetic effects of amino-acid replacements in enzymes may have no obvious mechanistic significance.

Hydrolysis of glycosylpyridinium ions by anomeric-configuration-inverting glycosidases

The hydrolyses of five beta-D-xylopyranosylpyridinium ions by the beta-D-xylosidase of Bacillus pumilus proceed with kcat values 10(8)-10(9)-fold larger than the rates of spontaneous hydrolysis of the same compounds. Log(kcat) values correlate well with aglycon pK(a) [B1g(V) = -0.52, r = 0.99], whereas the correlation of log(kcat/Km) is poor [r = 0.77; beta 1g(V/K) = approximately -0.6]. The (1-->3)-beta-D-glucanase of Sporotrichum dimorphosporum hydrolyses 4-bromo-2-(beta-D-glucopyranosyl)isoquinolinium ion with a rate enhancement of 10(8). The amyloglucosidase II of Aspergillus niger hydrolyses three alpha-D-glucopyranosylpyridinium ions with rate enhancements of 10(5)-10(8). The efficient hydrolysis of glycosylpyridinium ions by these three inverting glycosidases, the catalytic mechanism of which is unlikely to involve a nucleophile from the enzyme, makes it improbable that the hydrolysis of glycosylpyridinium ions by retaining glycosidases, discovered some years ago, is initiated by addition of a catalytic nucleophilic carboxylate group of the enzyme to the pyridinium ring.

Functions of the gene products of Escherichia coli

A list of currently identified gene products of Escherichia coli is given, together with a bibliography that provides pointers to the literature on each gene product. A scheme to categorize cellular functions is used to classify the gene products of E. coli so far identified. A count shows that the numbers of genes concerned with small-molecule metabolism are on the same order as the numbers concerned with macromolecule biosynthesis and degradation. One large category is the category of tRNAs and their synthetases. Another is the category of transport elements. The categories of cell structure and cellular processes other than metabolism are smaller. Other subjects discussed are the occurrence in the E. coli genome of redundant pairs and groups of genes of identical or closely similar function, as well as variation in the degree of density of genetic information in different parts of the genome.

Large changes of transition-state structure during experimental evolution of an enzyme

The question of whether, during the evolution of an enzyme, the transition state of the catalysed reaction is largely unchanged, or whether transition state and protein change together, was examined using the egb beta-galactosidases of Escherichia coli. Charge development at the first chemical state was assumed [Konstantinidis and Sinnott (1991) Biochem. J. 279, 587-593] to be proportional to delta delta G++, the ratio of second-order rate constants for the hydrolysis of beta-D-galactopyranosyl fluoride and 1-fluoro-D-galactopyranosyl fluoride, expressed as a free-energy difference. delta delta G++ (kJ.mol-1) falls from 10.4 for wild-type enzyme to 6.8 and 7.2 as a consequence of two different single amino-acid changes (which arise from single evolutionary events), to 6.3 as a consequence of the two amino-acid changes together, and then increases slightly to 7.3 as a consequence of a third single evolutionary change involving three further amino-acid changes.

The catalytic consequences of experimental evolution. Studies on the subunit structure of the second (ebg) beta-galactosidase of Escherichia coli, and on catalysis by ebgab, an experimental evolvant containing two amino acid substitutions

1. The ratio of ebgA-gene product of ebgC-gene product in the functional aggregate of ebg beta-galactosidases was determined to be 1:1 by isolation of the enzyme from bacteria grown on uniformly radiolabelled amino acids and separation of the subunits by gel-permeation chromatography under denaturing conditions. 2. This datum, taken together with a recalculation of the previous ultracentrifuge data [Hall (1976) J. Mol. Biol. 107, 71-84], analytical gel-permeation chromatography and electron microscopy, strongly suggests an alpha 4 beta 4 quaternary structure for the enzyme. 3. The second chemical step in the enzyme turnover sequence, hydrolysis of the galactosyl-enzyme intermediate, is markedly slower for ebgab, having both Asp-97----Asn and Trp-977----Cys changes in the large subunit, than for ebga (having only the first change) and ebgb (having only the second), and is so slow as to be rate-determining even for an S-glycoside, beta-D-galactopyranosyl thiopicrate, as is shown by nucleophilic competition with methanol. 4. The selectivity of galactosyl-ebgab between water and methanol on a molar basis is 57, similar to the value for galactosyl-ebgb. 5. The equilibrium constant for the hydrolysis of lactose at 37 degrees C is 152 +/- 19 M, that for hydrolysis of allolactose is approx. 44 M and that for hydrolysis of lactulose is approx. 40 M. 6. A comparison of the free-energy profiles for the hydrolyses of lactose catalysed by the double mutant with those for the wild-type and the single mutants reveals that free-energy changes from the two mutations are not in general independently additive, but that the changes generally are in the direction predicted by the theory of Burbaum, Raines, Albery & Knowles [(1989) Biochemistry 28, 9283-9305] for an enzyme catalysing a thermodynamically irreversible reaction. 7. Michaelis-Menten parameters for the hydrolysis of six beta-D-galactopyranosylpyridinium ions and ten aryl beta-galactosides by ebgab were measured. 8. The derived beta 1g values are the same as those for ebgb (which has only the Trp-977----Cys change) and significantly different from those for ebgo (the wild-type enzyme) and ebga. 9. The alpha- and beta-deuterium secondary isotope effects on the hydrolysis of the galactosyl-enzyme of 1.08 and 1.00 are difficult to reconcile with the pyranose ring in this intermediate being in the 4C1 conformation.

beta-deuterium kinetic isotope effects in the purine nucleoside phosphorylase reaction

1. [2'-2H]Inosine was made from inosine by tetraisopropyldisiloxanyl protection of the 3'- and 5'-positions, oxidation with dimethyl sulphoxide and acetic anhydride, immediate NaB2H4 reduction of the oxo sugar product and inversion at C-2' of the resultant protected [2'-2H]arabino-inosine by trifluoromethanesulphonylation and reaction with caesium propionate, followed by deprotection. 2. The equilibrium-perturbation technique was used to measure beta 2H(V/K) for phosphorolysis of this compound by the purine nucleoside phosphorylase of Escherichia coli as a function of pH. 3. The pH variation indicates an intrinsic effect of 1.068 masked by isotopically silent steps near the pH optimum. 4. The similar pH variation of these beta-deuterium effects and the alpha-deuterium effects measured previously [Stein & Cordes (1981) J. Biol. Chem. 256, 767-772; Lehikoinen, Sinnott & Krenitsky (1989) Biochem. J. 257, 355-359] for this reaction provides the first experimental reassurance for the common assumption that pH changes merely mask and unmask the chemical steps in an enzyme-catalysed reaction, and do not detectably alter transition-state structure. 5. The dihedral angle between the C-H-2' bond and the electron-deficient p-orbital at the transition state is in the range 32-48 degrees, in accord with an essentially planar furanose ring.