A Novel Subtype of Class II Alcohol Dehydrogenase in Rodents

Mice and rats were found to possess class II alcohol dehydrogenases with novel enzymatic and structural properties. A cDNA was isolated from mouse liver and the encoded alcohol dehydrogenase showed high identity (93.1%) with the rat class II alcohol dehydrogenase which stands in contrast to the pronounced overall variability of the class II line. The two heterologously expressed rodent class II enzymes exhibited over 100-fold lower catalytic efficiency (k cat/K m ) for oxidation of alcohols as compared with other alcohol dehydrogenases and were not saturated with ethanol. Hydride transfer limited the rate of octanol oxidation as indicated by a deuterium isotope effect of 4.8. The mutation P47H improved hydride transfer and turnover rates were increased to the same level as for the human class II enzyme. Michaelis constants for alcohols and aldehydes were decreased while they were increased for the coenzyme. The rodent class II enzymes catalyzed reduction of p-benzoquinone with about the same maximal turnover as for the human form. This activity was not affected by the P47H mutation while a S182T mutation increased theK m value for benzoquinone 10-fold. ω-Hydroxy fatty acids were catalyzed extremely slow but functioned as potent inhibitors by binding to the enzyme-NAD+ complex. All these data indicate that the mammalian class II alcohol dehydrogenase line is divided into two structurally and functionally distinct subgroups.

The family of alcohol dehydrogenases (ADH) 1 has a well documented ability to metabolize various alcohols and aldehydes and on the basis of these enzymatic properties one can assign potential functions for these enzymes in the metabolism of steroids, biogenic amines, lipid peroxidation products, retinoids, as well as xenobiotics (1)(2)(3)(4)(5)(6). For assessment of the physiological role of ADHs, knowledge of the whole ADH system is important since the different classes and isozymes display considerable overlap in metabolic activities and inhibitor reper-toires. The system is of an old origin with a split into at least seven vertebrate classes of which six have been identified in mammals (6,7). Within the murine ADH system, three different enzymes were early identified, and their corresponding genes were named Adh-1, -2, and -3 (8) encoding classes I, III, and IV, respectively (9 -11). Neither class II nor class V/VI have, as yet, been isolated or cloned and it has been suggested that mice either lack these genes or possess forms with very low identity relative to the human variants (11).
The human form was the first class II ADH to be identified. This enzyme is predominantly found in liver and contributes to the metabolism of ethanol (12)(13)(14). It has a preference for unsaturated hydrophobic aldehydes and has been suggested a redox specific role in the noradrenaline metabolism (2,13). Furthermore, this form is particularly effective in the reduction of the lipid peroxidation derived 4-hydroxyalkenals (4). Although the physiological implication is not clear, class II ADH also catalyzes the reduction of some benzoquinones and benzoquinone imines (15). At the structural level an extreme variability of the class II line is well established from characterization of species variants (16 -18).
This paper reports on the existence of an ADH of class II type in mouse which together with the rat counterpart form a conserved subgroup of class II ADHs that exhibits low catalytic efficiency as a consequence of slow hydride transfer. The enzymatic characteristics and the consequences of the unique coenzyme binding residues Pro 47 and Ser 182 are investigated.

EXPERIMENTAL PROCEDURES
Isolation of a cDNA Coding for Mouse Class II ADH-A cDNA coding for an ADH of class II type was isolated from an adaptor-ligated mouse cDNA library (Marathon-Ready TM cDNA, CLONTECH), by PCR amplification utilizing Pfu polymerase (Stratagene). Multiple sequence alignments of characterized class II ADH sequences were used to localize regions with a high degree of positional identity and two primers were designed: 5Ј-TTGGGCCAGGAGTGA(A/C)CAA-3Ј and 5Ј-AGTCAGTG-GCTCCCAGGGC-3Ј (Fig. 1). Cycling conditions were: 30 cycles of 95°C, 45 s; 68°C, 1 min; 72°C, 1.5 min. PCR amplification yielded a 500-bp fragment which was subsequently ligated into the vector pCRII (Invitrogen) according to the TA-cloning protocol (19) for sequence analysis.
primers introducing restriction sites NdeI and BamHI, respectively, which facilitated subcloning into the unique restriction sites NdeI/ BamHI of the pET29 expression vector (Novagen). In the same manner, the coding region of rat class II cDNA (17) was subcloned into the NcoI and BamHI restriction sites of pET3d (Novagen). Both plasmid constructs were verified by sequence analysis throughout the entire coding region.
Double-stranded plasmid was prepared with the flexiprep kit (Amersham Pharmacia Biotech) for mutagenesis of mouse class II cDNA. Reagents in the U.S.E. mutagenesis kit (Amersham Pharmacia Biotech) and mutagenesis primers: ACGTGTGTGTGCC(A/C/G)TACTGA-CATC(A/C)ATGCCACCGATCC and 5Ј-GATGTGGGTTCTCAACCG-GCTACGGGGCTG-3Ј were used to alter the codons and subsequently replacing Pro 47 for His, Asn 51 for His, and Ser 182 for Thr (nucleotides corresponding to the changed codons are underlined). Selection was based on the elimination of the unique XbaI site of the pET29 vector according to the method described by Deng and Nickoloff (24). Sequence analysis was performed to confirm the presence of the correct mutations and the absence of any unexpected mutations in the cDNA.
Expression and Isolation of Class II ADH-Recombinant protein was expressed in 1-liter LB cultures of Escherichia coli strain BL21(DE3) at 29°C and induced with a burst of 0.8 mM isopropyl-thio-␤-D-galactosidase at an OD 595 of about 1. Cells were harvested 4 h later and disrupted in 10 mM potassium phosphate, pH 7.5, with 0.3 mM dithiothreitol and 10 M ZnSO 4 which was used throughout the purification unless otherwise stated. Cells were lysed by sonication before centrifugation for 60 min at 48,000 ϫ g. Human class II ADH was isolated as described previously (18) while the rodent ADHs were isolated by DEAE-cellulose (DE-52, Whatman, 150 ml) chromatography followed by AMP-Sepharose (Amersham Pharmacia Biotech, 10 ml) chromatography. The void volume from the first purification step containing rodent ADH was applied to the AMP-Sepharose column, washed with 40 ml of 0.1 M potassium phosphate, 0.2 M NaCl, pH 7.5, before elution with 2.5 mM NAD ϩ in 10 mM potassium phosphate, pH 7.5, with 0.3 mM dithiothreitol and 10 M ZnSO 4 . Isolated class II ADH was subjected to buffer change to 10 mM Hepes, pH 7.5, by gel filtration on Sephadex G-25 columns (Amersham Pharmacia Biotech), concentration (Microsep 30K, Pall Filtron) and purity analysis by SDS-polyacrylamide gel electrophoresis. Protein concentration was determined with the Bio-Rad protein assay (Bio-Rad) with bovine serum albumin as standard complemented with amino acid analysis on a Amersham Pharmacia Biotech AlphaPlus analyzer. The zinc content of mouse wild-type class II ADH was analyzed by atomic absorption spectroscopy (Perkin Elmer 5000 Zeeman) using a flame at standard conditions with deuterium background correction.
Enzyme Assays-Enzyme activity was monitored with a Hitachi U-3000 spectrophotometer by following the conversion of NAD ϩ (⑀ 340 6.22 mM Ϫ1 cm Ϫ1 ) with exception for oxidation of all-trans-retinol (⑀ 400 29.5 mM Ϫ1 cm Ϫ1 ) (25) and menadione where activity was monitored with 3-[4,5-dimethylthiazo-2-yl-]2,5-diphenyltetrazolium bromide (⑀ 610 11.3 mM Ϫ1 cm Ϫ1 ) as described previously for DT-diaphorase (26). Alcohol oxidation activity was assayed in 0.1 M potassium phosphate, pH 7.5, and in 0.1 M glycine/NaOH, pH 10.0, at 25°C with 2.4 mM NAD ϩ while reductase activity was assayed in 0.1 M potassium phosphate, pH 7.5, with NADH concentrations of 0.2 mM. For quinone reduction, 0.1 mM NADH was used instead. If not else stated, reagents were from Sigma and of the highest purity readily available. Octanol and [1,1-2 H 2 ]octanol were synthesized by lithium aluminum hydride and lithium aluminum deuteride (Merck) reduction, respectively, of octanoic acid in ether. After ether/1 M NaOH extraction and subsequent distillation, the synthesis was confirmed by GC/MS analysis and the deuterium content of [1,1-2 H 2 ]octanol was determined to 98%. Benzaldehyde and acetaldehyde were distilled before use. 4-Hydroxyoctenal, kindly supplied by the late Professor H. Esterbauer, was stored at Ϫ20°C in chloroform which was evaporated under nitrogen immediately before use and the concentration determined spectrophotometrically at 224 nm (⑀ 224 13.75 mM Ϫ1 cm Ϫ1 ). Stock solutions of substrates and inhibitors were prepared in methanol yielding a final concentration of 2% except for ethanol, acetaldehyde, and pyrazole derivatives which were dissolved in water and all-trans-retinol which was dissolved in acetone. Class II ADHs were neither active toward nor inhibited by methanol while acetone gave a partial inhibition. Activity measurements with 4-hydroxybenzyl alcohol and 4-hydroxy-3-methoxybenzyl alcohol were corrected for absorbance of the corresponding aldehydes (⑀ 340 22.0 mM Ϫ1 cm Ϫ1 and ⑀ 340 23.3 mM Ϫ1 cm Ϫ1 , respectively). The extent of non-enzymatic p-benzoquinone reduction was determined for each concentration and was subtracted from the values obtained for the enzymatic reaction.
The dependence of the activity of mouse class II ADH on the concen-tration of ammonia (1-20 mM) was studied with benzyl alcohol and octanol in 0.1 M glycine/NaOH, pH 9.4 (27). The ammonia stock solution was titrated to pH 9.4 with hydrochloric acid.
To fit lines to data points and to calculate kinetic parameters a weighted nonlinear regression analysis program was used (Fig.P for Windows; Biosoft) while inhibition data was analyzed with the programs of Cleland (28). All kinetic parameters are based on measurements performed with 2-4 preparations of enzyme. The k cat values are per subunit (40 kDa) and standard errors were less than 10%. Standard errors for K m values were less than 15%.
Immunoblot Analysis-Frozen organs from mouse strain C57BL/6 (B6) and Harlan-Sprague Dawley rat were homogenized in 10 mM Tris/Cl, pH 8.0, and centrifuged for 15 min at 20,000 ϫ g. 5 g of total protein from the homogenates were separated by SDS-polyacrylamide gel electrophoresis together with recombinantly expressed human class I-III ADH (100 ng) and mouse class II ADH as controls. Proteins were transferred by electroblotting to poly(vinylidene difluoride) transfer membranes (Bio-Rad) which were subsequently blocked with 5% nonfat dry milk (Semper) in 10 mM Tris/Cl, 150 mM NaCl, 0.05% Tween 20, pH 8.0. The membranes were incubated with a 1:2000 dilution of rabbit antiserum raised against human class II ADH (18) in 10 mM Tris/Cl, 150 mM NaCl, 1% non-fat dry milk, 0.05% Tween 20, pH 8.0, washed, and incubated with a 1:2000 dilution of Protein A-horseradish peroxidase conjugate (Bio-Rad). Immunodetection was performed according to the ECL TM method protocol (Amersham Pharmacia Biotech).

Structural and Evolutionary Characteristics of Rodent Class
II ADHs-A cloning strategy based on the high sequence identity of earlier characterized class II ADHs in the regions around amino acid residues 79 and 241 was used to PCR amplify a 500-bp fragment of the mouse class II ADH cDNA. The fulllength cDNA was thereafter isolated from an adaptor ligated mouse liver cDNA library using the rapid amplification of cDNA ends technique. The entire cDNA sequence, covering 1354 bp, included a 1131-bp coding region, a 23-bp 5Ј noncoding region, a 200-bp 3Ј noncoding region, a poly(A) signal, and a poly(A) tail (Fig. 1). The region around the translation start codon was similar to that of the corresponding human form but lacked similarities with the consensus sequence, CCRC-CATGR, except for ATG and the purines at position Ϫ3 and ϩ4 (30). The cDNA translated into a 376-amino acid polypeptide that had a sequence identity of 93.1% with rat class II ADH. A relative rate test of divergence versus ostrich class II ADH showed higher positional identities with ostrich for the human and rabbit isoforms (68 -70%) than for the mouse and rat forms (65-66%). Still, it could not be excluded that differences in identities might be explained by the natural higher rates of nucleotide substitutions in rodents than in man (31). The phylogenetic tree of class I-IV ADHs showed generally longer branches in the class II line indicating faster divergence than for class I, III, and IV (Fig. 2). The short separation distance between rat and mouse class II as compared with the other class II forms is not compatible with the general assumption of a uniform divergence rate for the same enzyme in different species.
Alignments of all structurally characterized class II ADHs revealed three variable regions around positions 60, 120, and 300 with insertions and deletions as compared with the ADH consensus sequence. Within these regions there are two deletions specific for the rodent class II ADHs and a 4-residue insertion around position 119, that is common to all characterized class II ADHs. A single residue is deleted at either position 57 or 59, equal alignment scores, and two residues are deleted at positions 299 -300. Deviations from the consensus sequence of ADHs were found at three coenzyme binding positions, 47, 51, and 182. The latter corresponds to 178 in the class I ADH numbering system, a residue positioned on the opposite side of the nicotinamide ring as compared with the substrate. The otherwise conserved Thr was Ser in both rodent class II ADHs while the His at either position 47 or 51, common in most ADHs, were lacking. The effects of these replacements were studied by site-directed mutagenesis and three mutants were created, denoted P47H, N51H, and S182T.
Tissue Distribution-The distribution of class II ADH in 13 mouse and rat tissues were studied by immunoblot analysis with antiserum raised against human class II ADH. A strong immunoreactive signal was found in liver and a weaker signal in kidney (Fig. 3). However, a more specific Northern blot analysis of mouse liver and kidney with a mouse class II cDNA probe revealed that detectable amounts of class II ADH mRNA was only expressed in liver (Fig. 4). This indicates that the immunoreactive signal from kidney probably is of another origin.
Enzymatic Properties of Class II ADH Species Variants-Class II ADHs were expressed recombinantly and isolated by ion exchange and affinity chromatography. Expression yields were 15-20 mg/liter culture medium. The same isolation protocol with pET29 expression plasmid without cDNA insert yielded roughly 100 g/liter of protein exhibiting no acetaldehyde or p-benzoquinone reducing activity, which indicates a purity over 99% which was also confirmed by Coomassie Blue staining of SDS-polyacrylamide gels. The zinc content was determined for the mouse wild-type enzyme to 2.2 mol/mol subunit indicating one active and one structural zinc per subunit. The alcohol oxidation, aldehyde reduction, and quinone reduction activities of rodent class II ADHs were investigated and compared with those of the human class II enzyme. Ethanol, octanol, benzyl alcohol, and their corresponding aldehydes together with benzoquinone were used as model substrates for these activities (Tables I and II). The alcohol dehydrogenase reaction was catalyzed dramatically less efficient by the rodent forms as compared with the human form. Turnover numbers were more than 10-fold lower and saturation was reached at far higher concentrations. In the case of ethanol, none of the rodent class II forms were saturated with substrate. The reduction of aldehydes was more efficiently catalyzed by the rodent forms than alcohol oxidation, still turnover numbers were 5-10-fold lower than for the human form. In contrast, benzoquinone reduction proceeded with about the same turnover rate for all species forms, although the K m values were lower for the human form. The substrate and inhibitor repertoire of mouse class II ADH was more thoroughly investigated (Table III). The activity for class II ADH characteristic substrates such as benzyl alcohol derivatives and 4-hydroxyoctenal were more than 100-fold lower than has been reported for human class II ADH. Neither activity nor inhibition of benzyl alcohol oxidation was observed for the retinoids, steroids, and endogenous quinones tested. Pyrazole and 4-methylpyrazole were poor inhibitors while the halogenated derivative, 4-bromopyrazole, gave stronger inhibition, a pattern resembling that of the human form of the enzyme (32). However, mouse class II ADH showed extremely low activity for the three -hydroxy fatty acids investigated and was potently inhibited by these compounds at both pH 7.5 and 10.0 (Fig. 5). Inhibition constants were determined at pH 10.0 by varying the octanol and inhibitor concentration at saturating NAD ϩ . Competitive inhibition patterns were obtained and K is values were determined to 26 Ϯ 7 M, 6.0 Ϯ 0.7 M, and 6.9 Ϯ 0.7 M for 10-HDA, 12-HDA, and 16-HHA, respectively (Fig. 5). The importance of the carboxylic acid group for efficient binding was evident from the high affinity of octanoic acid (Fig. 5). Moreover, uncompetitive inhibition patterns were obtained when NAD ϩ and 12-HDA were varied at subsaturating concentrations of 4-hydroxybenzyl alcohol or octanol and gave K ii values of 5.0 Ϯ 0.4 M and 14 Ϯ 1.8 M, respectively (Fig. 5). This result is consistent with an ordered mechanism for alcohol oxidation with NAD ϩ binding as first reactant (33), a mechanism also observed for human class II ADH (32).
Mouse class II mutants were expressed and purified according to the same protocol as for the wild-type enzyme. The P47H mutant showed increased k cat values and decreased K m values for alcohol oxidation and aldehyde reduction while benzoquinone reduction was unaffected (Tables I and II). Catalytic efficiency for alcohol oxidations were increased about 50-fold at FIG. 1. Nucleotide and deduced amino acid sequence of mouse class II type ADH. The amino acid residues are given in one-letter code and numbered on the right-hand side. The arrows indicate the primers used for the initial PCR reaction and bold amino acid residues show positions subjected to mutagenesis. The initiation codon, ATG, is underlined and the stop codon, TGA, is indicated with an asterisk. Also underlined is a poly(A) signal in the 3Ј noncoding region. pH 10.0 and at least 200-fold at pH 7.5 as compared with the wild-type enzyme. An increase was seen for aldehyde reduction as well, but less pronounced (10 -50-fold). The N51H mutation did not significantly effect the catalytic activity of the enzyme with the exception for a 3-fold decrease in octanol and octanal K m values. The S182T mutation increased k cat /K m values for alcohol oxidation (5-10-fold) while the corresponding values for aldehyde reduction decreased 2-fold. The most striking characteristic of this mutant was a 10-fold increased K m value for benzoquinone (Table II).
Coenzyme saturation was studied for human class II, mouse class II, and the P47H mutant and the Michaelis constants were determined (Table IV). The wild-type mouse enzyme had lower K m values for NAD ϩ and NADH as compared with the human form, 10-and Ͼ30-fold, respectively. Introduction of the P47H mutation gave an increase in the K m values to a level comparable with that of the human form. Furthermore, NADPH could not serve as a coenzyme for mouse class II ADH.
Isotope effects were determined using octanol and [1,1-2 H 2 ]octanol as substrates (Table V). A large isotope effect was seen on k cat and k cat /K m for mouse wild-type enzyme showing that the hydride transfer step was rate-limiting for the oxidation of octanol. Isotope effects were seen for the P47H and S182T enzymes as well, indicating that, although catalytic efficiency was increased, hydride transfer was at least partially rate-limiting. The lack of isotope effect on k cat for human class II ADH is compatible with coenzyme release being rate-limiting which has been proposed previously (13).
The effect of ammonia on the benzyl alcohol and octanol activity of mouse class II ADH was studied at pH 9.4. Addition of 1-20 mM ammonia did not significantly effect the specificity constants for oxidation of these alcohols suggesting that introduction of exogenous amines as proton acceptors do not increase the activity of the enzyme. have been identified which share about 65% positional identity. In the mouse only three of these classes have been described earlier: class I, III, and IV. With the aim to further investigate the ADH repertoire we found an ADH of class II type with novel structural and functional characteristics. The class II branch of the ADH family show extreme divergence, and the rate for introduction of nonsynonymous substitutions are 2-and 6-fold higher as compared with the class I and III ADHs, respectively. The low identity between the human and mouse class II cDNAs probably explains why cross-hybridization at the mRNA level has been unsuccessful for this class of ADH (11). With this in mind, the high identity between mouse and rat class II ADH (93.1%, at the protein level) is surprising. The corresponding value for the overall more conserved class I ADH is 89.6% and for the highly conserved class III ADH is 96.5%. The deviation from a uniform divergence in species variants could be indicative of a different function for this enzyme in rodents or even the existence of two subtypes of class II ADH that are related by gene duplication, i.e. they are paralogous rather than orthologous. Thus far, gene duplications of class II ADH have only been detected in rabbit, and although one variant shares some enzymatic characteristics with the rodent class II ADHs, structural identities indicate a more recent gene duplication for the two rabbit isoforms (18).
By immunoblotting of tissue homogenates, rodent class II were found to be predominantly expressed in liver, a pattern resembling that of the human variant. Immunoreactive signals were also found for kidney homogenates from both rat and mouse. Class II mRNA has previously been found absent in rat kidney (34) which evidently also was the case for mouse kidney (Fig. 3). We conclude that our immune serum cross-reacts with a protein in kidney with a subunit mass of the same size as mouse class II ADH (40 kDa) which possibly could be an uncharacterized ADH expressed in the kidney (35).
Most enzymes catalyze reactions with turnover numbers between 1 and 1000 s Ϫ1 (36) and the human class II ADH oxidizes alcohols at a maximal rate of 4 -9 s Ϫ1 at pH 10.0 and more than 10-fold lower at physiological pH (13,37). The kinetic constants determined for the human class II ADH in this study were in agreement with these previous results. In comparison, the rodent forms had over 10-fold lower turnover for alcohol oxidation and this difference was observed at both pH 7.5 and 10.0. In addition, K m values were higher for the alcohol/aldehyde pairs used for comparison of species variants (Tables I and II). While the human form contributes to the metabolism of ethanol in the liver, the rodent forms were not possible to saturate with this alcohol. Taken into account, the rodent forms of class II ADH can be considered as low activity ADHs, with presumably no significance in general alcohol detoxification. The low activity can further explain why this form of ADH has not previously been described in studies on liver ADH activity in these species (38,39). The characterization of the mouse class II ADH indicated no overlap in activity for endogenous substrates with other classes within the ADH family (Table III). Moreover, class II characteristic substrates such as 4-hydroxyoctenal and benzaldehyde derivatives were metabolized with low efficiency. In consistence, the reductive metabolism of 4-hydroxynonenal in rat liver homogenates, attributed to class I ADH only, is low The enzyme was not possible to saturate with ethanol. The activity was directly proportional to ethanol concentrations up to 1.7 M and the k cat /K m value was derived from the slope in the velocity versus ethanol graph.

TABLE II
Steady-state kinetic constants for aldehyde and p-benzoquinone reduction at pH 7.5 catalyzed by mouse, rat, human, and mutant forms of mouse class II ADHs Michaelis constants (K m ) and turnover numbers (k cat ) were determined from initial velocity experiments at 25°C. Aldehyde and p-benzoquinone concentrations were varied over a 32-fold range and the NADH concentration was fixed at 0.2 mM for aldehyde reduction and 0.1 mM for p-benzoquinone reduction. compared with the bioconversion via the glutathione S-transferase pathway (40). Since human class II ADH is far more efficient than class I for this reaction, it is possible that reductive metabolism of 4-hydroxynonenal is more pronounced in homogenates of human liver. Furthermore, mouse class II ADH oxidized -hydroxyfatty acids at an extremely slow rate. Still they were potent inhibitors against octanol oxidation (Fig. 5).
Since octanoic acid also inhibited octanol oxidation competitively, it is likely that the carboxylic group coordinates to the catalytic zinc in the enzyme-NAD ϩ complex where the charged nicotinamide ring also can contribute with electrostatic interactions. The special substrate repertoire can be explained by a number of exchanges at residue positions suggested to be important for substrate binding (Table VI). Notably, while the residues are identical for mouse and rat class II ADH at all these positions, the residue identity between mouse and human class II ADH is only 54% which is even lower than the positional identity between the entire sequences (72%). The Role of Pro 47 and Ser 182 for Alcohol Dehydrogenase Activity-The ADH activity of the rodent class II forms could be drastically improved by replacing Pro with His at position 47. The effects of residue exchanges at this position have been studied extensively and explains, e.g. the activity differences between the allelic variants of the human ADH class I ␤␤isoform and the resistance to allyl alcohol poisoning of mutant yeast strains (41)(42)(43). The guanidino group of Arg 47 stabilizes the enzyme-coenzyme complex by the formation of a salt bridge with the pyrophosphoryl moiety of the coenzyme (44 -46) and subsequently, coenzyme dissociation rates are in general    slower with stronger bases than with neutral or mild bases at this position (43). Substitutions at position 47 also affects hydride transfer (43). In addition to His 47 and Arg 47 , Gly 47 are found in a few ADHs whereas Pro 47 is unique for the rodent forms of class II ADH. Local structural rearrangements and/or alternative coenzyme binding seems to compensate for the lack of salt bridge formation with Gly 47 and results in coenzyme dissociation constants comparable to that of Arg 47 variants (47). In analogy, the P47H replacement could result in weaker coenzyme binding and increased turnover rates. Still, the pronounced isotope effect found for mouse class II ADH showed that hydride transfer is rate-limiting for alcohol oxidation. This implies that the two mutations P47H and S182T, both increasing the catalytic efficiency more than 1 order of magnitude, primarily acts by modulation of the hydride transfer step. It has been observed that ADHs lacking His at position 51 often harbors a His at position 47 (37). The conserved His 51 in class I ADHs acts as a general base catalyst through a proton relay connecting the alcohol substrate and His 51 via the hydroxyl group of Thr/Ser 48 and the 2Ј-hydroxyl of the nicotinamide ribose (48 -52). Structures of binary complexes of both cod ADH and human class III ADH show that His 47 is at a position where hydrogen binding to the 2Ј-hydroxyl is possible and could therefore potentially act in the same manner as His 51 (53,54). In addition, the drop in activity seen for a His for Gln mutant at a position analogous to 47 of Pseudomonas putida benzyl alcohol dehydrogenase could partially be restored by introduction of either exogenous amines as proton acceptors or His at a position analogous to 51 (27). However, this alternative route of proton transfer cannot be valid for the few examples of ADH forms where no base is present at these positions, e.g. for the rodent class II ADHs. For mouse class II ADH, absence of a His 47 could not be compensated for by exogenous amines as in the case for the P. putida benzyl alcohol dehydrogenase. Furthermore, the N51H mutant showed only slightly higher activity as compared with the wild-type enzyme (Tables I and II). Among the structurally characterized species variants of class II ADH both the class I theme with His 51 and the class III theme with His 47 are present. However, the B isoform of rabbit class II ADH (previously denoted II-2) has no His at either of these positions (18). Proton transfer directly to the solvent is in accordance with the strong pH dependence on class II ADHcatalyzed alcohol oxidation.
The nicotinamide ring of the coenzyme is held in position by van der Waals contacts with the side chains of conserved Thr 178 , Val 203 , and Val 294 . The rodent class II ADHs together with the gene product of the rice ADH2 are the only ADHs where Thr 178 , analogous to 182 in the class II structure, is replaced by Ser. Replacements of Val 203 results in changes of hydrogen transfer distances and dramatically influence the role of hydrogen tunneling in the hydride transfer of ADH (55). Structural changes caused by the mutations described in this report are likely to result in similar effects.
The Role of Pro 47 and Ser 182 for Quinone Reductase Activity-The human class II, in contrast to class I ADH, efficiently catalyze the reduction of benzoquinones and benzoquinone immines (15), compounds that also undergo spontaneous reduction with NADH in solution. This reaction was fairly efficiently catalyzed by mouse class II and the catalytic efficiency (k cat / K m ) is only 4.5-fold lower than for the human enzyme. Notably, benzoquinone reduction was not affected by the P47H mutation. Furthermore, while the S182T mutation was beneficial for alcohol oxidation it increased the K m value for benzoquinone 10-fold (Table II). Evaluation of these two mutations indicates that the structural requirements for benzoquinone reduction are slightly different than for aldehyde reduction. This could be a consequence of the longer distance between hydride transfer and proton donation for quinone reduction since the hydride is not transferred to the ␣-carbon in this case, but to the oxygen at the opposite side of the quinone ring, yielding hydroquinones.
The residue replacements discussed are shared by both rodent class II ADHs and make them biased for quinone reduction rather than alcohol oxidation and thus makes us suggest that this enzyme is primarily involved in reductive metabolism. Although derivatives of benzoquinone are naturally occurring, the physiological significance of this activity is not clear and must be studied further. Mouse class II did not catalyze reduction of the naphtoquinone menadione. Neither was Q 0 , the functional part of the ubiquitous coenzyme Q 10 , a substrate for the enzyme. For the human enzyme, lack of activity for bulkier quinones have been suggested to be the result of steric hindrance due to an assumed narrow substrate pocket (15). In conclusion, the special functional characteristics together with the structural conservation between the mouse and rat class II enzymes as opposed to the divergence in the class II line in general, strongly indicates that the rodent enzymes form a class II subgroup within the ADH family.