INTRODUCTION

Top Abstract Introduction Procedures Results Discussion References

Oxidative damage to macromolecules is an inescapable price for the evolution of aerobic metabolism. The co-evolution of protective agents, both catalytic (enzymes such as superoxide dismutases and catalases) and stoichiometric (antioxidants such as glutathione and -tocopherol), can at best reduce the magnitude of damage. Evolution of active mechanisms of repair apparently is limited only for DNA, probably because of chemical feasibility and strong selective pressure. For other kinds of damaged macromolecules clearing by turnover seems to be the only option. A possible exception is the repair of oxidized methionine residues on the surface of proteins by a specific sulfoxide reductase (1). The replacement strategy seems satisfactory for the perpetuation of unicellular organisms, although it is not always available or adequate for multicellular organisms. For instance, accumulation of oxidatively damaged proteins is often associated with senescence and various disease states (2-4).

A significant fraction of protein damages is thought to result from the metal-catalyzed oxidation system (MCO)¹ mediated by reactive species such as hydrogen peroxide, as outlined in Reactions 1 (2, 4, 5). H₂O₂ is formed routinely by monooxygenation reactions and occasionally by autoxidation of flavo-dehydrogenases, when an element in the electron transport chain is rate-limiting. Cationic iron is maintained in the ferrous state by reducing compounds, such as NADH or NADPH. The presence of H₂O₂ and Fe²⁺ generates, by the Fenton reaction, a highly reactive HO^· (hydroxyl radical) which can covalently attack an amino acid residue.

When the iron is bound to a protein, the H₂O₂-dependent redox cycling of Fe²⁺ to Fe³⁺ is thought to proceed in a "cage," thus allowing the hydroxyl radical to extract an H atom from a local amino acid residue, before diffusing into the surrounding medium (6-9). Such a model would account for the limited number of amino acid residues that are susceptible to the damage, with each protein exhibiting a distinctive target signature of residues. The model is also supported by the evidence of substrate protection against oxidative damage (6, 8, 10). Although Arg, Cys, His, Lys, Met, and Pro residues are most susceptible to metal-catalyzed destruction, only the oxidation of Arg, Pro, His, and Lys has been reported to result in the formation of a carbonyl derivative which provides a means for monitoring the protein oxidation process (2, 11).

L-1,2-Propanediol:NAD⁺ 1-oxidoreductase of Escherichia coli is an Fe²⁺-dependent enzyme that normally functions as a reductase in a fermentation pathway for the dissimilation of L-fucose or L-rhamnose (12). This enzyme, inducible by either of the methyl pentoses, is inactivated during aerobic growth (9, 13, 14). In this study we tried to find out whether repeated selection of mutants that utilize this enzyme exclusively as a dehydrogenase for aerobic growth, on propanediol as the sole carbon and energy source, would result in an altered protein resistant to oxidative inactivation.

	EXPERIMENTAL PROCEDURES

Top Abstract Introduction Procedures Results Discussion References

Sequencing the fucO Gene of Strains ECL1, ECL3, ECL56, ECL421, ECL430, and ECL459-- The oligonucleotide primers 5'-CGGATCCGGCATTATCACATCAG and 5'-CGAATTCAAGAGTAATTTCGTAAAGC flanking the coding region of fucO (15, 16) were synthesized and used to amplify by polymerase chain reaction the gene of each strain. A sample of each amplified product was digested with restriction enzymes to give five overlapping fragments that were then subcloned into pBluescript vectors (Stratagene, La Jolla, CA). Each of these fragments was sequenced for both strands with the T3 and T7 primers (Stratagene, La Jolla, CA) by the dideoxy method.

Selecting an E. coli Strain with Complete Deletion of the Fuc Sequence-- Cells with deletions in the fuc locus were selected from strain ECL330 bearing a fuc::Mu d1-amp^r insertion by growth at a nonpermissive temperature (42 °C) that allows the growth of only Mu dl cells. Clones with a Mu dl Amp^s Fuc phenotype were identified by growth at 42 °C on MacConkey/fucose medium and then purified by growing at 37 °C on the same medium. The deletion ranges in these cells were screened by Southern blots probed with a set of fuc fragments spanning the entire region of the fuc regulon (17). A strain completely deleted in the fuc regulon was thus identified and designated as ECL733.

Generating Mutant Alleles of fucO-- The fucO^Ile-7Leu and fucO^Leu-8Val mutations were regenerated by site-directed mutagenesis (18) of the wild-type sequence. A fucO allele with the above two mutations combined, fucO^{Ile-7Leu, Leu-8Val}, was created in parallel. A 5-kb SalI^o-BamHI fragment cut from pfuc16 (19) containing the wild-type fucO allele was subcloned into an M13 mp18 vector (Stratagene, La Jolla, CA), and the DNA was used to transfect CJ236 cells (Bio-Rad). Single-stranded M13 DNA was then prepared from the cells, hybridized to the oligonucleotide primers containing the desired point mutations (ordered from Oligos Etc., Inc., Guilford, CT), and used as template for synthesizing the second DNA strand. The duplex products were used to transfect XL1 cells (Stratagene, La Jolla, CA). A plaque from each transfection was purified, and a fucO fragment in the phage DNA was sequenced to confirm the nucleotide substitution(s). The phage DNA containing the mutated fucO alleles was respectively designated FM5/O^Ile-7Leu, FM5/O^Leu-8Val, and FM5/O^{Ile-7Leu, Leu-8Val}.

Cloning the Entire Wild-type Fuc Regulon into a Shuffling Vector-- As shown in Fig. 1, a modified pBluescript plasmid was used as an intermediary shuffling vector for carrying fuc sequences onto  vector. The modification was made by inserting a cam^r gene cut from a pBC KS⁺ plasmid (Stratagene, La Jolla, CA) into the pBluescript polylinker region, to place the cloned fuc sequence close to a selectable marker for subsequent insertion into the chromosome via  vectors (see below). The resulting plasmid was designated pBB. The wild-type fuc regulon (about 9 kb) was cloned into pBB in a two-step procedure as follows: step 1, an approximately 8-kb fuc fragment PvuII-SalI¹ was cut from pfuc1 containing the entire fuc regulon (19) and inserted into pBB; and step 2, the remaining 1-kb fuc fragment Sa1I¹-Sa1I² was appended to the insertion by substituting the EcoRI-SalI¹ fragment in the first insertion with the EcoRI-SalI² fragment cut from pfuc1. The resulting plasmid was designated pFB1. For the propagation of various plasmids, XL1 cells were used.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Cloning of the fuc+ regulon into the shuffling vector. V, vector fragment; I, insert fragment. Arrows indicate the digestion sites used in generating the fragments for ligations. See "Experimental Procedures."

Cloning Fuc Regulons with the Mutated fucO Alleles into the Shuffling Vector-- The mutant fuc regulons were cloned into the shuffling vector by substituting the wild-type fucO sequence in pFB1 with the mutated counterparts. This procedure was carried out in eight steps, illustrated in a condensed manner in Fig. 2. Step 1: a pBC KS⁺ plasmid was modified by destroying the single PvuI restriction site with the Klenow enzyme and ligase, resulting in a second shuffling vector pBC^* (not shown). Step 2: a 7-kb fuc fragment SalI⁰-EcoRI was cut from pfuc16 and inserted into pBC^*, resulting in pFC1. Step 3: a 2.2-kb fragment BamHI-BamHI^linker was deleted from pFC1 to eliminate a PstI site in the pBC polylinker region, resulting in pFC2 (not shown). Step 4: the 0.5-kb fuc0 fragment PstI-PvuI in pFC2 was cut off and substituted with a corresponding fragment bearing one of the fucO mutations cut from FM5/O^Ile-7Leu, FM5/O^Leu-8Val, or FM5/ O^{Ile-7Leu, Leu-8Val}, resulting in pFC2/O^Ile-7Leu, pFC2/O^Leu-8Val, or pFC2/O^{Ile-7Leu, Leu-8Val} (not shown). Step 5: the PstI-PvuI fragments in pFC2/O^Ile-7Leu, pFC2/O^Leu-8Val, and pFC2/O^{Ile-7Leu, Leu-8Val} were sequenced to confirm the correct substitutions. Step 6: the BamHI-BamHI^linker fragment from pFC1 was inserted back into pFC2/O^Ile-7Leu, pFC2/O^Leu-8Val, and pFC2/O^{Ile-7Leu, Leu-8Val}, resulting in plasmids pFC1/O^Ile-7Leu, pFC1/O^Leu-8Val, and pFC1/O^{Ile-7Leu, Leu-8Val}. Step 7: the 5-kb fuc fragment PstI-EcoRI in pFB1 was cut off and substituted with a corresponding fragment cut from pFC1/O^Ile-7Leu, pFC1/ O^Leu-8Val, or pFC1/O^{Ile-7Leu, Leu-8Val}, resulting in pFB1/O^Ile-7Leu, pFB1/O^Leu-8Val, or pFB1/O^{Ile7-Leu, Leu-8Val}, plasmid bearing full-length fuc regulon. Step 8: the PstI-PvuI fragments in pFB1/O^Ile-7Leu, pFB1/O^Leu-8Val, and pFB1/O^{Ile-7Leu, Leu-8Val} were sequenced to confirm the correct fucO mutation status.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Cloning fuc regulons with wild-type and mutated fucO alleles into shuffling vectors. Various fuc regulons bearing different fucO alleles were cloned into pFB1 vectors. Om represents OIle-7Leu, OLeu-8Val, or OIle-7Leu, Leu-8Val. V, vector segment; I, insert fragment. Arrows indicate the digestion sites used in generating the fragments for ligations. See "Experimental Procedures."

Inserting Wild-type and Mutant Fuc Regulons into Host Chromosomes via  Vectors-- The plasmids pFB1, pFB1/O^Ile-7Leu, pFB1/O^Leu-8Val, and pFB1/O^{Ile-7Leu, Leu8Val} were digested with restriction enzymes ApaI and XbaI, yielding 11-kb fragments each containing a full-length fuc regulon and the cam^r gene. These fragments were then ligated with wild-type  DNA cut with the same enzymes. The ligation mixtures were packaged with the Gigapack Gold Lambda packaging extract (Stratagene, La Jolla, CA) and used to infect the fuc strain ECL733. Cells bearing the foreign sequences on the chromosome were selected as  lysogens growing on chloramphenicol and MacConkey/fucose plates. Single copy fuc regulon insertions were confirmed by Southern blots probed with fuc fragments at both ends of the fuc sequence. The strains bearing the wild-type and mutant fuc regulons at the att site are designated ECL734, ECL735, ECL736, and ECL737.

Growth Conditions and Preparation of Cell Extracts-- Cells were grown aerobically as described previously (20) on Luria broth or 0.5% casein acid hydrolysate. Anaerobic cultures were grown as described previously (20) in 1% casein acid hydrolysate supplemented with 1 mM pyruvate. Where indicated, L-fucose was added as inducer at 10 mM concentration for aerobic growth and 20 mM for anaerobic growth. For enzyme assays, cells were harvested at the end of the exponential phase, and cell extracts were prepared as described previously (20) in 10 mM Tris-HCl buffer, pH 7.5. For enzyme purification, the extracts were prepared using a 50 mM Tris-HCl buffer, pH 7.5, containing 2.5 mM NAD.

Enzyme Purification-- Propanediol oxidoreductase was purified from extracts of cells grown anaerobically in Luria broth plus L-fucose by the method of Cabiscol et al. (9). Enzyme purity was assessed by electrophoresis performed according to Laemmli (21) using 10% acrylamide as resolving gel. Proteins were stained with Coomassie Blue R-250.

Enzyme Activity Assays-- Propanediol oxidoreductase was routinely assayed by its NADH-dependent glycolaldehyde reduction to ethylene glycol (22). Glycolaldehyde, readily available commercially, was shown to be an alternative substrate for the enzyme (23). In experiments testing enzyme protection by NAD or propanediol, the enzyme was assayed by NAD-dependent propanediol dehydrogenation (20). Protein concentration was determined by the Bradford method (24) using bovine serum albumin as standard. Immuno-quantification of propanediol oxidoreductase protein was carried out by Laurell rocket immunoelectrophoresis (25) according to a calibration curve (not shown) derived by using the propanediol oxidoreductase purified from strain ECL1. Antibodies were obtained as described (26).

Testing Thermal Stability-- Stock solutions of purified propanediol oxidoreductases (0.5 mg/ml) in 50 mM Tris-HCl buffer at pH 7.5 were kept at 4 °C. At time 0, 0.5 ml of each solution was transferred to tubes preincubated at different temperatures, with or without supplementation with 50 mM DL-1,2-propanediol or 1 mM NAD. Enzyme activities were assayed at various intervals. Thermal stability of the enzyme in crude extract (0.5-ml samples containing 8 mg/ml of total protein) was tested under similar conditions.

Testing Oxidative Inactivation by NADH-- Purified propanediol oxidoreductases (0.5 mg/ml) were incubated at 20 °C in 50 mM Tris-HCl buffer, pH 7.5, in the presence of 0.5 mM NADH. Enzyme activities were assayed at various intervals.

Testing Oxidative Inactivation by Ascorbate Plus Iron-- Purified propanediol oxidoreductases (0.5 mg/ml) were preincubated at 20 °C for 5 min with 50 mM 1,2-propanediol, followed by addition of 15 µM FeCl₃ and 30 mM ascorbate according to Levine et al. (27). Enzyme activities were assayed at various intervals.

Immunodetection of Protein Carbonyl Groups-- Dinitrophenylhydrazine derivatization of protein carbonyl groups resulting from oxidation was performed according to Levine et al. (28). Electrophoresis was performed using 10% acrylamide as resolving gel. After electrophoresis the proteins were transferred to polyvinylidene difluoride membranes by semi-dry blotting. Immuno-detection of protein-bound dinitrophenylhydrazones was performed according to Levine et al. (27). The primary antibody was a polyclonal rabbit preparation (Dako, V0401, Denmark). The secondary antibody was goat anti-rabbit conjugated with alkaline phosphatase (Tropix, Bedford, MA).

Metal Analyses-- Atomic absorption spectroscopy measurements were conducted with a Jobin-Yvon spectrometer, JY-38. Samples were submitted to high performance liquid chromatography gel filtration in a Protein Pak 125 (Waters) prior to metal analysis to eliminate reagents and metals not bound to the enzyme. The eluent used was MilliQ water (resistivity greater that 18 M). Fractions were collected in metal-free polypropylene tubes.

Fourth Derivative Spectra-- Absorption spectra and their fourth derivative were taken on a Shimadzu UV-160A spectrophotometer, using a derivative interval of 2.4 nm, a slit width of 2 nm, and a scan rate of 80 nm/min. Purified enzymes were dissolved in 50 mM Tris-HCl buffer, pH 7.5.

	RESULTS

Top Abstract Introduction Procedures Results Discussion References

Wild-type and Propanediol-positive Strains and Comparison of Their fucO DNA Sequences-- Propanediol oxidoreductase is encoded by the fucO gene, a member of the fuc regulon specifying the utilization of L-fucose. The genes of the fuc regulon are organized in two divergent operons, fucPIK and fucAO (Fig. 3). Induction of the operons requires the activator FucR (17, 29) and its effector, L-fuculose-1-phosphate (30). L-1,2-Propanediol, a product of L-fucose fermentation, cannot serve aerobically as a sole carbon and energy source, because the 3-carbon compound fails to induce the fuc regulon expression. If fucO is expressed constitutively at an adequate level, the cell should be able to grow on propanediol by converting it to pyruvate via L-lactaldehyde and L-lactate (Fig. 3).

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Metabolic pathway and genetic organization of the L-fucose system. Utilization of the methylpentose depends on L-fucose permease (encoded by fucP), L-fucose isomerase (encoded by fucI), L-fuculose kinase (encoded by fucP), L-fuculose-1-phosphate aldolase (encoded by fucA), and L-1,2-propanediol:NAD+ oxidoreductase. Under fermentative conditions, the oxidoreductase serves to reduce L-lactaldehyde to propanediol which appears to be excreted via a facilitator (35, 56). Under aerobic or other respiratory conditions, L-lactaldehyde is oxidized to L-lactate by an NAD+-dependent oxidoreductase (41) encoded by the ald gene at min 31.2 (68-70). L-Lactate is further oxidized to pyruvate by a flavo-dehydrogenase encoded by lctD (71, 72). Determination of the metabolic flow of L-lactaldehyde by respiratory growth conditions (heavy arrows) depends on transcriptional control of ald (73, 74) and post-translational inactivation of the fucO gene products (15, 42, 43, 71). The fuc genes, located at min 60 of the chromosome are organized in two divergent operons, FucAO, fucPIK, under the positive control of fucR (17, 29). Propanediol oxidoreductase and L-lactaldehyde dehydrogenase are also required for the dissimilation of L-rhamnose (12, 20, 75).

Reconstruction of the fucO Mutations in an Otherwise Wild-type Background-- To find out whether the two different single fucO mutations conferred resistance of the enzyme to aerobic inactivation, we tested plasmid-borne wild-type and mutant fucO alleles in a standard wild-type background. Preliminary activity assays of the oxidoreductase were not satisfactorily reproducible. The difficulty seems to be attributable to variations in the plasmid copy number.

Expression of fucO Genes-- The  lysogens ECL734 (fucO⁺), ECL735 (fucO^Ile-7Leu), ECL736 (fucO^Leu-8Val), and ECL737 (fucO^{Ile-7Leu, Leu-8Val}) and the wild-type nonlysogen, ECL1, were grown aerobically or anaerobically on casein acid hydrolysate in the presence of L-fucose as inducer of the fuc regulon. Extracts from each culture were assayed for propanediol oxidoreductase activity (Table I). The specific activities of propanediol oxidoreductase (units/mg enzyme protein) in extracts of anaerobically grown cells were all about the same, irrespective of the nature of the fuc allele. In contrast, the specific activity of the enzyme in extracts of aerobically grown cells was dependent on the fucO allele: FucO⁺ < FucO^Ile-7Leu < FucO^Leu-8Val < FucO^{Ile-7Leu, Leu-8Val}. It should be mentioned that if the inducer was not added, there was no detectable oxidoreductase activity in extracts prepared from the different strains, even when the cells were grown anaerobically (not shown), indicating that transcriptions were from the same fuc promoter. The data taken together indicate that a single amino acid substitution was sufficient to confer significant resistance of the protein to oxidative damage and that the double amino acid substitutions have an additive protective effect. On the other hand, the amino acid substitutions do not appear to have a significant effect on the catalytic property of the enzymes, at least when assayed by the reduction of the substrate analog glycolaldehyde (Table II).

Thermal Inactivation of Wild-type and Mutant Propanediol Oxidoreductase Enzymes-- Additional evidence of structural alteration of the mutant propanediol oxidoreductase was their decreased thermal stability. When purified wild-type and mutant propanediol oxidoreductases were incubated at two different temperatures at pH 7, the activity decay rate of the enzymes followed the same order: at 50 °C, FucO⁺ (t1/2 = 6.6 min), FucO^Ile-7Leu (t1/2 = 2.3 min), FucO^Leu-8Val (t1/2 = 1.5 min), and FucO^{Ile-7Leu, Leu-8Val} (t1/2 = 0.9 min) (Fig. 4A); at 20 °C, FucO⁺ (t1/2 = 840 min), FucO^Ile-7Leu (t1/2 = 340 min), FucO^Leu-8Val (t1/2 = 140 min), and FucO^{Ile-7Leu, Leu-8Val} (t1/2 = 48 min) (Fig. 4B). It is, however, not clear from these experiments whether the loss of activity resulted from irreversible denaturation of the protein or the loss of the cofactor Fe²⁺.

View larger version (30K): [in this window] [in a new window]   Fig. 4.   Thermal inactivation of mutant and wild-type propanediol oxidoreductases. Purified FucO+) (), FucOIle-7Leu (), FucOLeu-8Val) (), and FucOIle-7Leu, Leu-8Val) () were tested for stability under various conditions. A, 50 °C; B, 20 °C; C, 20 °C in the presence of 50 mM DL-1,2-propanediol; and D, 20 °C in the presence of 1 mM NAD. At the indicated times samples were withdrawn for enzyme activity assay by the rate of glycolaldehyde-dependent oxidation of NADH (A and B) or by the rate of 1,2-propanediol-dependent reduction of NAD (C and D). In each case, the specific activity of the enzyme at 0 time was expressed as 100%.

Oxidative Inactivation of Purified Propanediol Oxidoreductase Enzymes by NADH-- Since propanediol oxidoreductase is an iron enzyme, the presence of both molecular oxygen and NADH is expected to result in an intrinsically catalyzed Fenton reaction that damages the protein. Purified wild-type and mutant enzymes were incubated for 120 min in the presence of 0.5 mM NADH at 20 °C under air. As shown in Fig. 5A, FucO⁺ was rapidly inactivated (t1/2 = 15 min). In contrast, the mutant proteins were substantially more stable: FucO^Ile-7Leu (t1/2 = 28 min), FucO^Leu-8Val (t1/2 = 54 min), and FucO^{Ile-7Leu, Leu-8Val} (t1/2 = 110 min). The possibility that differences in sensitivity of the enzymes are attributable to the disparities in the amount of iron bound was excluded by the finding that the metal contents of the four different purified enzymes were equal, as determined by atomic absorption spectroscopy. The iron-dependent Fenton reaction as the cause of the observed enzyme inactivation was suggested by the NADH dependence of oxidative protein damage, as revealed by immunochemical assays for the carbonyl groups generated (Fig. 5B).

View larger version (25K): [in this window] [in a new window]   Fig. 5.   Oxidative inactivation of purified wild-type and mutant propanediol oxidoreductases. A, Fuc+ (), FucOIle-7Leu (), FucOLeu-8Val (), and FucOIle-7Leu, Leu-8Val () were incubated at 20 °C with 0.5 mM NADH. At the indicated times samples were withdrawn for assay of enzyme activity which was expressed as percent of the 0 time value. B, Western blot of oxidized enzymes. After 40 min of incubation with NADH, samples were taken and processed for immunodetection of protein carbonyl groups produced during oxidation. Samples of unoxidized control proteins were taken before NADH was added for incubation. C, control protein; Ox, NADH-treated protein.

Oxidative Inactivation of Purified Propanediol Oxidoreductase Enzymes by Ascorbate and Fe³⁺-- Other evidence of the susceptibility of propanediol oxidoreductase to damage by Fenton reaction was demonstrated by incubation of the protein in the presence of ferric chloride and ascorbate, instead of NADH, as the reductant (28). All four enzymes were incubated at 20 °C for 120 min in the presence of 50 mM DL-1,2-propanediol to stabilize the proteins against simple thermal inactivation. The relative rates of inactivation of the four enzymes were consistent with those observed during incubation with NADH (Fig. 5A): FucO⁺ (t1/2 = 70 min), FucO^Ile-7Leu (t1/2 = 110 min), FucO^Leu-8Val (t1/2 = 260 min), and FucO^{Ile-7Leu, Leu-8Val} (t1/2 = 430 min) (Fig. 6).

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Inactivation of pure propanediol oxidoreductase by ascorbate and iron. Purified FucO+) (), FucOIle-7Leu (), FucOLeu-8Val) (), and FucOIle-7Leu, Leu-8Val () were oxidized at 20 °C with ascorbate and iron in the presence of 1,2-propanediol. At the indicated times samples were taken to assay enzyme activity which was expressed as percent of the initial activity before incubation with ascorbate and iron was started.

Analysis of Fourth Derivative Ultraviolet Spectra of Wild-type and Mutant Enzymes-- On the basis of crystallographic data on the highly conserved folding motifs of NAD-dependent oxidoreductases, it can be predicted that the amino acid substitutions found in the mutant FucO proteins are close to the A-2 turn of the mononucleotide-binding motif (or Rossmann fold) where Tyr³¹ is located (17, 37).

View larger version (39K): [in this window] [in a new window]   Fig. 7.   Fourth derivative spectra representation of the wild-type and mutant FucO proteins. The plot taken from a Shimadzu UV 160A spectrophotometer shows the spectra of purified propanediol oxidoreductase from strains: A, Fuc+ (0.7 mg/m); B, FucOIle-7Leu (0.9 mg/ml); C, FucOLeu-8Val (0.75 mg/ml); and D, FucOIle-7Leu, Leu-8Val (0.4 mg/ml). The spectra were taken between 240 and 320 nm. Arrows indicate the main differences observed between mutant and wild-type proteins.

	DISCUSSION

Top Abstract Introduction Procedures Results Discussion References

Although early work on propanediol oxidoreductase showed that Fe²⁺ activates the apoenzyme (41) and that the protein induced during aerobic growth lacked catalytic activity (42, 43), MCO of the enzyme protein was not suspected until a hint was provided by the work on Fe²⁺-catalyzed inactivation of E. coli glutamine synthetase (6, 7, 44-46). In the case of FucO, the Fe²⁺ is at the catalytic center which binds the protein to NADH for fermentative reduction of L-lactaldehyde. During aerobic metabolism when the enzyme serves no physiological role, the generation of H₂O₂ allows the Fe²⁺ to catalyze a Fenton reaction, and the highly reactive OH^· formed is likely to diffuse a distance of the order of its mean free path, which is only a few times of the free radical's radius, before hitting a target amino acid residue (i.e. diffusion-controlled encounter). A frequent occurrence is the destruction of the side chain of a conserved His²⁷⁷, 10 residues away from the proposed metal-binding site, His²⁶³-X-X-X-His²⁶⁷, causing a decrease in the apparent affinity of the protein for NAD (13).

It is remarkable that the relatively conservative hydrophobic amino acid substitutions, Ile⁷  Leu and/or Leu⁸  Val, near the NAD-binding sequence Gly¹⁵-Arg-Gly-Ala-Val-Gly²⁰ of FucO (15, 16, 47) could confer significant protective effects against MCO damage. This resistance, however, is achieved at a price of decrease in protein stability. The loss of stability might be attributable to "cavity creation" associated with diminished hydrophobic interactions. In the case of T4 lysozyme a Leu  Ala substitution raised the G of the folded form of the enzyme by 1.9 kcal/mol; increases as high as 6.2 kcal/mol have been observed in hydrophobic amino acid substitutions in other proteins (48).

Interestingly, the alcohol oxidoreductase II of Zymomonas mobilis, an enzyme highly homologous to FucO (16), is damaged by MCO in a similar way (9). Similar cases of enzyme inactivation by MCO were observed in studies of Klebsiella pneumoniae. Glycerol:NAD 2-oxidoreductase, which serves for the utilization of glycerol under fermentative conditions, is inactivated during aerobic metabolism (49). The initial step involves the MCO-caused loss of apparent affinity for NAD (50, 51). 1,3-Propanediol:NAD 1-oxidoreductase (disposing of NADH by reduction of 3-hydroxypropionaldehyde during fermentative growth on glycerol) and ethanol:NAD oxidoreductase (disposing of NADH by reduction of acetyl-CoA during sugar fermentation) seem to be likewise inactivated (52, 53).

Although protein turnover necessitated by MCO is metabolically costly (44, 54, 55), inactivation of certain enzymes during aerobic respiration can be beneficial in the balance. In the case of FucO, the continued presence of a catalytically active protein during aerobic utilization of L-fucose or L-rhamnose would wastefully deplete both L-lactaldehyde and NADH (56, 57). Viewed from this angle, the bound Fe⁺² might be regarded as an adaptive self-destruct mechanism for facilitating the transition from fermentative to aerobic metabolism. The same reasoning should apply to ethanol oxidoreductase. In the case of glycerol oxidoreductase, the rapid inactivation of the enzyme would facilitate the shift from the relatively ineffective anaerobic substrate capturing pathway initiated by the NAD-coupled oxidoreductase to the more avid aerobic substrate scavenging pathway initiated by the ATP-driven kinase (58), a kinetic advantage predictable by the Haldane equation relating K_eq to the k_cat/K_m.

In contrast to NAD(P) enzymes that are involved in fermentative metabolism, those that are needed for both aerobic and anaerobic metabolism, such as glucose-6-phosphate dehydrogenase, isocitric dehydrogenase, and malate dehydrogenase of K. pneumoniae, are resistant to inactivation by oxidative metabolism (53). The same is true for the E. coli NAD-linked L-lactaldehyde dehydrogenase that is required only for aerobic substrate utilization (14). These resistant enzymes are probably dependent on Zn²⁺ instead of Fe²⁺ or are metal-independent (see below). An interesting case is the glucose-6-phosphate dehydrogenase in Leuconostoc mesenteroides, which is thought to mediate only anaerobic glucose catabolism by a pentose pathway (59); it is Fe²⁺-dependent and subject to MCO inactivation (8).

On the basis of amino acid sequence homology, NAD(P)-dependent alcohol dehydrogenases (oxidoreductases) fall into the following three families: (i) long chain Zn²⁺-dependent, (ii) short chain Zn²⁺-independent, and (iii) Fe²⁺-activated (60, 61). Zymomonas mobilis, an O₂-tolerant and obligately ethanologenic anaerobe (62), has two alcohol oxidoreductases, one is Fe²⁺-dependent and the other dependent on Zn²⁺ (16, 63-65). As one would expect, the Zn²⁺ enzyme is MCO-resistant, but the Fe²⁺ enzyme is susceptible to MCO damage (66). It was suggested that having two enzymes with different metal ion requirement would be a nutritional insurance (67). The question of whether the Zn²⁺ enzyme plays the major role in aerobic ethanologenesis and the Fe²⁺ enzyme in anaerobic ethanologenesis has not been addressed. The yeast, Saccharomyces cerevisiae, also possesses both Zn²⁺ and Mn²⁺ alcohol dehydrogenases (60). The physiological and/or evolutionary basis for employing both metal ions has yet to be explored.

In light of the fact that no Fe²⁺-dependent oxidoreductase is known to have an aerobic function, it is tempting to suggest that such enzymes evolved early when the global environment was highly reducing and the supply of ferrous iron was abundant. With the emergence of photosynthesis and attendent accumulation of O₂, aerobic metabolism developed, and iron became mostly sequestered as insoluble Fe³⁺ compounds. Fe²⁺-dependent oxidoreductases were gradually supplanted by Zn²⁺-dependent ones. Those oxidoreductases that persisted as Fe²⁺ enzymes did so either because there was a lack of selective pressure to switch to Zn²⁺ or because retention of Fe²⁺ actually provided the cell with an adaptive mechanism for thrifty shift from anaerobic to aerobic metabolism.