Eye lens zeta-crystallin relationships to the family of "long-chain" alcohol/polyol dehydrogenases. Protein trimming and conservation of stable parts

Activation of farnesoid X receptor (FXR) induces crystallin zeta expression in mouse medullary collecting duct cells

Crystallin zeta (CRYZ) is a phylogenetically restricted water-soluble protein and provides cytoprotection against oxidative stress via multiple mechanisms. Increasing evidence suggests that CRYZ is high abundantly expressed in the kidney where it acts as a transacting factor in increasing glutaminolysis and the Na⁺/K⁺/2Cl^- cotransporter (BSC1/NKCC2) expression to help maintain acid-base balance and medullary hyperosmotic gradient. However, the mechanism by which CRYZ is regulated in the kidney remains largely uncharacterized. Here, we show that CRYZ is a direct target of farnesoid X receptor (FXR), a nuclear receptor important for renal physiology. We found that CRYZ was ubiquitously expressed in mouse kidney and constitutively expressed in the cytoplasm of medullary collecting duct cells (MCDs). In primary cultured mouse MCDs, CRYZ expression was significantly upregulated by the activation and overexpression of FXR. FXR-induced CRYZ expression was almost completely abolished in the MCD cells with siRNA-mediated FXR knockdown. Consistently, treatment with FXR agonists failed to induce CRYZ expression in the MCDs isolated from mice with global and collecting duct-specific FXR deficiency. We identified a putative FXR response element (FXRE) on the CRYZ gene promoter. The luciferase reporter and ChIP assays revealed that FXR can bind directly to the FXRE site, which was further markedly enhanced by FXR activation. Furthermore, we found CRYZ overexpression in MCDs significantly attenuated hypertonicity-induced cell death possibly via increasing Bcl-2 expression. Collectively, our findings demonstrate that CRYZ is constitutively expressed in renal medullary collecting duct cells, where it is transcriptionally controlled by FXR. Given a critical role of FXR in MCDs, CRYZ may be responsible for protective effect of FXR on the survival of MCDs under hypertonic condition during dehydration.

Zeta-crystallin: a moonlighting player in cancer

Crystallins were firstly found as structural proteins of the eye lens. To this family belong proteins, such as ζ-crystallin, expressed ubiquitously, and endowed with enzyme activity. ζ-crystallin is a moonlighting protein endowed with two main different functions: (1) mRNA binding with stabilizing activity; (2) NADPH:quinone oxidoreductase. ζ-crystallin has been clearly demonstrated to stabilize mRNAs encoding proteins involved in renal glutamine catabolism during metabolic acidosis resulting in ammoniagenesis and bicarbonate ion production that concur to compensate such condition. ζ-crystallin binds also mRNAs encoding for antiapoptotic proteins, such as Bcl-2 in leukemia cells. On the other hand, the physiological role of its enzymatic activity is still elusive. Gathering research evidences and data mined from public databases, we provide a framework where all the known ζ-crystallin properties are called into question, making it a hypothetical pivotal player in cancer, allowing cells to hijack or subjugate the acidity response mechanism to increase their ability to resist oxidative stress and apoptosis, while fueling their glutamine addicted metabolism.

Prediction and experimental validation of enzyme substrate specificity in protein structures

Structural Genomics aims to elucidate protein structures to identify their functions. Unfortunately, the variation of just a few residues can be enough to alter activity or binding specificity and limit the functional resolution of annotations based on sequence and structure; in enzymes, substrates are especially difficult to predict. Here, large-scale controls and direct experiments show that the local similarity of five or six residues selected because they are evolutionarily important and on the protein surface can suffice to identify an enzyme activity and substrate. A motif of five residues predicted that a previously uncharacterized Silicibacter sp. protein was a carboxylesterase for short fatty acyl chains, similar to hormone-sensitive-lipase-like proteins that share less than 20% sequence identity. Assays and directed mutations confirmed this activity and showed that the motif was essential for catalysis and substrate specificity. We conclude that evolutionary and structural information may be combined on a Structural Genomics scale to create motifs of mixed catalytic and noncatalytic residues that identify enzyme activity and substrate specificity.

Crystallin distribution patterns in Litoria infrafrenata and Phyllomedusa sauvagei lenses

The eye lens remains transparent because of soluble lens proteins known as crystallins. For years γ-crystallins have been known as the main lens proteins in lower vertebrates such as fish and amphibians. The unique growth features of the lens render it an ideal structure to study ageing; few studies have examined such changes in anuran lenses. This study aimed to investigate protein distribution patterns in Litoria infrafrenata and Phyllomedusa sauvagei species. Lenses were fractionated into concentric layers by controlled dissolution. Water-soluble proteins were separated into high (HMW), middle (MMW) and low molecular weight (LMW) fractions by size-exclusion HPLC and constituents of each protein class revealed by 1DE and 2DE. Spots were selected from 2DE gels on the basis of known ranges of subunit molecular weights and pH ranges and were identified by MALDI-TOF/TOF MS following trypsin digestion. Comparable lens distribution patterns were found for each species studied. Common crystallins were detected in both species; the most prominent of these was γ-crystallin. Towards the lens centre, there was a decrease in α- and β-crystallin proportions and an increase in γ-crystallins. Subunits representing taxon-specific crystallins demonstrating strong sequence homology with ζ-crystallin/quinone oxidoreductase were found in both L. infrafrenata and P. sauvagei lenses. Further work is needed to determine which amphibians have taxon-specific crystallins, their evolutionary origins, and their function.

Structural and Mutational Analysis of Trypanosoma brucei Prostaglandin H2 Reductase Provides Insight into the Catalytic Mechanism of Aldo-ketoreductases

Trypanosoma brucei prostaglandin F2alpha synthase is an aldo-ketoreductase that catalyzes the reduction of prostaglandin H2 to PGF2alpha in addition to that of 9,10-phenanthrenequinone. We report the crystal structure of TbPGFS.NADP+.citrate at 2.1 angstroms resolution. TbPGFS adopts a parallel (alpha/beta)8-barrel fold lacking the protrudent loops and possesses a hydrophobic core active site that contains a catalytic tetrad of tyrosine, lysine, histidine, and aspartate, which is highly conserved among AKRs. Site-directed mutagenesis of the catalytic tetrad residues revealed that a dyad of Lys77 and His110, and a triad of Tyr52, Lys77, and His110 are essential for the reduction of PGH2 and 9,10-PQ, respectively. Structural and kinetic analysis revealed that His110, acts as the general acid catalyst for PGH2 reduction and that Lys77 facilitates His110 protonation through a water molecule, while exerting an electrostatic repulsion against His110 that maintains the spatial arrangement which allows the formation of a hydrogen bond between His110 and C11 that carbonyl of PGH2. We also show Tyr52 acts as the general acid catalyst for 9,10-PQ reduction, and thus we not only elucidate the catalytic mechanism of a PGH2 reductase but also provide an insight into the catalytic specificity of AKRs.

Pseudomonas fluorescens mannitol 2-dehydrogenase and the family of polyol-specific long-chain dehydrogenases/reductases: sequence-based classification and analysis of structure-function relationships

Multiple sequence alignment and analysis of evolutionary relationships have been used to characterize a family of polyol-specific long-chain dehydrogenases/reductases (PSLDRs). At the present time, 66 known and putative NAD(P)H-dependent oxidoreductases of mainly prokaryotic origin and between 357 and 544 amino acids in length constitute this family. The family is shown to include D-mannitol 2-dehydrogenase, D-mannonate 5-oxidoreductase, D-altronate 5-oxidoreductase, D-arabinitol 4-dehydrogenase, and D-mannitol-1-phosphate 5-dehydrogenase which form individual sub-families (defined by internal sequence identity of >/=30%) having distant origin and divergent substrate specificity but clearly displaying entire-chain relationship. When all forms are aligned, only three residues, Gly-33, Asp-230, and Lys-295 (in the numbering of Pseudomonas fluorescens D-mannitol 2-dehydrogenase (PsM2DH)) are strictly conserved. By combining sequence alignment with the known structure of PsM2DH and results from site-directed mutagenesis, we have developed a structure/function analysis for the family. Gly-33 is in the N-terminal coenzyme-binding domain and part of a nucleotide fingerprint region for the family, and Asp-230 and Lys-295 are at an interdomain segment contributing to the active site in which the lysine likely functions as the catalytic general acid/base. PSLDRs do not require a metal cofactor for activity and are specific for transferring the 4-pro-S hydrogen from NAD(P)H. Comparisons reveal that the core part of the two-domain fold has been conserved throughout all family members, perhaps reflecting the recruitment of a stable oxidoreductase structure and extensive trimming thereof to acquire functional properties specific to each sub-family. They also identify interactions that define the chemical mechanism of oxidoreduction and likely contribute to substrate and co-substrate specificities and are thus relevant for protein engineering.

Zeta-crystallin displays strong selectivity for salicylic acid over aspirin

Interaction of camel lens zeta-crystallin with aspirin was investigated by activity and fluorescence measurements. Aspirin minimally inhibited the oxidoreductase activity of the enzyme and weakly quenched its fluorescence. However, significant fluorescence quenching of zeta-crystallin coincided with the appearance of a fluorescence signal characteristic of salicylic acid thereby raising the possibility that salicylic acid might have been the moiety responsible for inhibition and fluorescence quenching. Direct fluorescence measurements showed that zeta-crystallin had a much higher affinity for salicylic acid than aspirin (K(i) of about 24 microM for salicylic acid versus 630 microM for aspirin). Salicylic acid was also far more effective in inhibiting zeta-crystallin than aspirin (K(i) values were 23 microM versus 820 microM, respectively). Inhibition kinetics suggested that salicylic acid interacted with zeta-crystallin via a binding site that was distinct from that of NADPH. Salicylic acid also interacted with and quenched the fluorescence of camel lens alpha-crystallin suggesting a general mode of interaction with lens proteins. Within the normal therapeutic concentrations of salicylic acid or aspirin, only crystallin-salicylic acid interactions might be significant. These results showed that camel lens zeta- and alpha-crystallin exhibited remarkable selectivity for salicylic acid over aspirin, and thus, could be considered as salicylate-binding proteins.

Inhibition of camel lens zeta-crystallin by aspirin and aspirin-like analgesics

Camel lens zeta-crystallin was reversibly inhibited to various degrees by aspirin (acetyl salicylic acid) and the aspirin-like analgesics: paracetamol (acetaminophen) and ibuprofen (2-(4-isobutyl phenyl)-propionic acid). Among these, aspirin was the most potent inhibitor, causing nearly complete inhibition in a dose-dependent, but time-independent manner. Analysis of inhibition kinetics revealed that aspirin was uncompetitive inhibitor (K(i) 0.64 mM) with respect to NADPH and non-competitive inhibitor (K(i) 1.6 mM) with respect to the substrate, 9,10-phenanthrenequinone (PQ). Multiple-inhibition analysis showed that aspirin and pyridoxal 5' phosphate (PAL-P), a lysine specific reagent, simultaneously bound to a critical lysine residue located towards the NADPH binding region. Consistent with this, NADPH was able to substantially protect zeta-crystallin against aspirin, whereas PQ did not provide any protection. The results suggested that an essential lysine residue was the locus of aspirin binding. The inhibition of zeta-crystallin by aspirin and aspirin-like analgesics was reversible thus eliminating acetylation as a mechanism for inhibition. Reversible binding of aspirin to this lysine may cause steric hindrance resulting in uncompetitive inhibition with respect to NADPH.

Hydrophobicity of the NADPH binding domain of camel lens zeta-crystallin

Interaction of camel lens zeta-crystallin with the hydrophobic probe 1-anilinonaphthalene-8-sulfonic acid (ANS) enhanced the ANS fluorescence and quenched the protein fluorescence. Both of these events were concentration-dependent and showed typical saturation curves suggesting specific ANS-zeta-crystallin binding. Quantitative analysis indicated that 1 mole zeta-crystallin bound at most 1 mole ANS. NADPH but not 9,10-phenanthrenequinone (PQ) was able to displace zeta-crystallin-bound ANS. These results suggested the presence of a hydrophobic domain in zeta-crystallin, possibly at the NADPH binding site. alpha-Crystallin as well as NADPH protected zeta-crystallin against thermal inactivation suggesting the importance of this site for enzyme stability. The NADPH:quinone oxidoreductase activity of zeta-crystallin was inhibited by ANS with NADPH as electron donor and PQ as electron acceptor. Lineweaver-Burk plots indicated mixed-type inhibition with respect to NADPH, with a K(i) of 2.3 microM. Secondary plots of inhibition with respect to NADPH indicated a dissociation constant (K'I) of 12 microM for the zeta-crystallin-NADPH-ANS complex. The K(i) being smaller than K'I suggested that competitive inhibition at the NADPH binding site was predominant over non-competitive inhibition. Like ANS-zeta-crystallin binding, inhibition was dependent on ANS concentration but independent of incubation time.

High-affinity binding of NADPH to camel lens zeta-crystallin

Fluorescence spectrum of camel lens zeta-crystallin, a major protein in the lens of camelids and histicomorph rodents, showed maximum emission at 315 nm. This emission maximum is blue shifted compared to most proteins, including alpha-crystallin, and appeared to be due to tryptophan in highly hydrophobic environment. Interaction of NADPH with zeta-crystallin quenched the protein fluorescence and enhanced the fluorescence of bound NADPH. Analysis of fluorescence quenching suggested high-affinity interaction between NADPH and zeta-crystallin with an apparent Km<0.45 microM. This value is at least an order of magnitude lower than that suggested by activity measurements. Analysis of NADPH fluorescence showed a biphasic curve representing fluorescence of free- and bound-NADPH. The intersection between free- and bound-NADPH closely paralleled the enzyme concentration, suggesting one mole of NADPH was bound per subunit of the enzyme. Phenanthrenequinone (PQ), the substrate of zeta-crystallin, also was able to quench the fluorescence of zeta-crystallin, albeit weaker than NADPH. Quantitative analysis suggested that zeta-crystallin had low affinity for PQ in the absence of NADPH, and PQ binding induced significant conformational changes in zeta-crystallin.

Probing the affinity and specificity of yeast alcohol dehydrogenase I for coenzymes

Yeast (Saccharomyces cerevisiae) alcohol dehydrogenase I (SceADH) binds NAD+ and NADH less tightly and turns over substrates more rapidly than does horse (Equus caballus) liver alcohol dehydrogenase E isoenzyme (EcaADH), and neither enzyme uses NADP efficiently. Amino acid residues in the proposed adenylate binding pocket of SceADH were substituted in attempts to improve affinity for coenzymes or reactivity with NADP. Substitutions in SceADH (Gly202Ile or Ser246Ile) with the corresponding residues in the adenine binding site of the homologous EcaADH have modest effects on coenzyme binding and other kinetic constants, but the Ser246Ile substitution decreases turnover numbers by 350-fold. The Ser176Phe substitution (also near adenine site) significantly decreases affinity for coenzymes and turnover numbers. In the consensus nucleotide-binding betaalphabeta fold sequence, SceADH has two alanine residues (177-GAAGGLG-183) instead of the Leu200 in EcaADH (199-GLGGVG-204); the Ala178-Ala179 to Leu substitution significantly decreases affinity for coenzymes and turnover numbers. Some NADP-dependent enzymes have an Ala corresponding to Gly183 in SceADH; the Gly183Ala substitution significantly decreases affinity for coenzymes and turnover numbers. NADP-dependent enzymes usually have a neutral residue instead of the Asp (Asp201 in SceADH) that interacts with the hydroxyl groups of the adenosine ribose, along with a basic residue (at position 202 or 203) to stabilize the 2'-phosphate of NADP. The Gly203Arg change in SceADH does not significantly affect the kinetics. The Gly183Ala or Gly203Arg substitutions do not enable SceADH to use NADP+ as coenzyme. SceADH with the single Asp201Gly or double Asp201Gly:Gly203Arg substitutions have similar, low activity with NADP+. The results suggest that several of the amino acid residues participate in coenzyme binding and that conversion of specificity for coenzyme requires multiple substitutions.

An ethanol-inducible MDR ethanol dehydrogenase/acetaldehyde reductase in Escherichia coli: structural and enzymatic relationships to the eukaryotic protein forms

An ethanol-active medium-chain dehydrogenase/reductase (MDR) alcohol dehydrogenase was isolated and characterized from Escherichia coli. It is distinct from the fermentative alcohol dehydrogenase and the class III MDR alcohol dehydrogenase, both already known in E. coli. Instead, it is reminiscent of the MDR liver enzyme forms found in vertebrates and has a K(m) for ethanol of 0.7 mM, similar to that of the class I enzyme in humans, however, it has a very high k(cat), 4050 min(-1). It is also inhibited by pyrazole (K(i) = 0.2 microM) and 4-methylpyrazole (K(i)= 44 microM), but in a ratio that is the inverse of the inhibition of the human enzyme. The enzyme is even more efficient in the reverse direction of acetaldehyde reduction (K(m) = 30 microM and k(cat) = 9800 min(-1)), suggesting a physiological function like that seen for the fermentative non-MDR alcohol dehydrogenase. Growth parameters in complex media with and without ethanol show no difference. The structure corresponds to one of 12 new alcohol dehydrogenase homologs present as ORFs in the E. coli genome. Together with the previously known E. coli MDR forms (class III alcohol dehydrogenase, threonine dehydrogenase, zeta-crystallin, galactitol-1-phosphate dehydrogenase, sensor protein rspB) there is now known to be a minimum of 17 MDR enzymes coded for by the E. coli genome. The presence of this bacterial MDR ethanol dehydrogenase, with a structure compatible with an origin separate from that of yeast, plant and animal ethanol-active MDR forms, supports the view of repeated duplicatory origins of alcohol dehydrogenases and of functional convergence to ethanol/acetaldehyde activity. Furthermore, this enzyme is ethanol inducible in at least one E. coli strain, K12 TG1, with apparently maximal induction at an enthanol concentration of approximately 17 mM. Although present in several strains under different conditions, inducibility may constitute an explanation for the fairly late characterization of this E. coli gene product.

Hamster sperm protein, p26h: a member of the short-chain dehydrogenase/reductase superfamily

For successful fertilization to occur, mammalian spermatozoa must undergo a series of modifications in order to reach and penetrate the oocyte. Some of these modifications occur during passage through the epididymis, the site where spermatozoa acquire their fertilizing ability. We have previously described hamster sperm protein, P26h, which is acquired by spermatozoa during epididymal transit, and have proposed that this protein is involved in sperm-egg binding. In the present study, we report the cloning and characterization of the full-length cDNA encoding hamster P26h. A database search using the predicted hamster P26h amino acid sequence revealed 85% identity with mouse AP27 protein and porcine carbonyl reductase, members of the short-chain dehydrogenase/reductase (SDR) family of proteins. Northern blot analysis revealed a major P26h 1-kilobase transcript in the testis. No signal was detected in other somatic tissues of the hamster. In situ hybridization experiments revealed that the P26h gene was predominantly transcribed in seminiferous tubules of the testis and at a lower level in the corpus epididymidis. The identity of the cloned P26h was confirmed by immunoprecipitating in vitro-translated P26h using polyclonal antiserum raised against purified hamster sperm P26h. Taken together, these results identify P26h as a new member of the SDR family of proteins involved in the processes of mammalian gamete interactions.

SDR and MDR: completed genome sequences show these protein families to be large, of old origin, and of complex nature

Short-chain dehydrogenases/reductases (SDR) and medium-chain dehydrogenases/reductases (MDR) are protein families originally distinguished from characterisations of alcohol dehydrogenase of these two types. Screening of completed genome sequences now reveals that both these families are large, wide-spread and complex. In Escherichia coli alone, there are no fewer than 17 MDR forms, identified as open reading frames, considerably extending previously known MDR relationships in prokaryotes and including ethanol-active alcohol dehydrogenase. In entire databanks, 1056 SDR and 537 MDR forms are currently known, extending the multiplicity further. Complexity is also large, with several enzyme activity types, subgroups and evolutionary patterns. Repeated duplications can be traced for the alcohol dehydrogenases, with independent enzymogenesis of ethanol activity, showing a general importance of this enzyme activity.

Molecular basis for thermoprotection in Bemisia: structural differences between whitefly ketose reductase and other medium-chain dehydrogenases/reductases

The silverleaf whitefly (Bemisia argentifolii, Bellows and Perring) accumulates sorbitol as a thermoprotectant in response to elevated temperature. Sorbitol synthesis in this insect is catalyzed by an unconventional ketose reductase (KR) that uses NADPH to reduce fructose. A cDNA encoding the NADPH-KR from adult B. argentifolii was cloned and sequenced to determine the primary structure of this enzyme. The cDNA encoded a protein of 352 amino acids with a calculated molecular mass of 38.2 kDa. The deduced amino acid sequence of the cDNA shared 60% identity with sheep NAD(+)-dependent sorbitol dehydrogenase (SDH). Residues in SDH involved in substrate binding were conserved in the whitefly NADPH-KR. An important structural difference between the whitefly NADPH-KR and NAD(+)-SDHs occurred in the nucleotide-binding site. The Asp residue that coordinates the adenosyl ribose hydroxyls in NAD(+)-dependent dehydrogenases (including NAD(+)-SDH), was replaced by an Ala in the whitefly NADPH-KR. The whitefly NADPH-KR also contained two neutral to Arg substitutions within four residues of the Asp to Ala substitution. Molecular modeling indicated that addition of the Arg residues and loss of the Asp decreased the electric potential of the adenosine ribose-binding pocket, creating an environment favorable for NADPH-binding. Because of the ability to use NADPH, the whitefly NADPH-KR synthesizes sorbitol under physiological conditions, unlike NAD(+)-SDHs, which function in sorbitol catabolism.

Inhibition of camel lens zeta-crystallin/NADPH:quinone oxidoreductase by pyridoxal-5'-phosphate

Camel lens zeta-crystallin was inhibited by pyridoxal-5'-phosphate (PAL-P) and o-phthalaldehyde. PAL-P inactivated zeta-crystallin in a time- and concentration-dependent manner. The initial rate of inactivation followed pseudo-first-order kinetics with the second-order rate constant of 91 M-1 s-1. The modified enzyme showed the characteristic absorption peak at 325 nm indicative of the formation of phosphopyridoxallysine. Quantitative analysis suggested the incorporation of 1 mole of PAL-P/subunit of enzyme. NADPH was able to substantially protect zeta-crystallin against PAL-P inactivation, whereas the substrate 9,10-phenanthrenequinone (PQ) did not provide any protection. Inhibition of zeta-crystallin by PAL-P was uncompetitive with NADPH (Ki=37 microM) and non-competitive with respect to the substrate (Ki=57 microM). Inhibition of zeta-crystallin by o-phthalaldehyde was used to establish the location of an essential lysine residue. Incubation of zeta-crystallin with o-phthalaldehyde resulted in the formation of an isoindole derivative that had a characteristic fluorescence spectrum. This suggested that a lysine residue is located within 3 A of a cysteine residue at the NADPH binding region. SDS-PAGE showed the o-phthalaldehyde-modified enzyme remained largely monomer (approx. 80%), although bands corresponding to dimer and tetramer forms were also present. These results suggested that an essential lysine residue is located in the vicinity of the NADPH binding site. This residue may simply ensure the proper binding of NADPH to the active site of zeta-crystallin.

Inhibition of camel lens zeta-crystallin/NADPH: quinone oxidoreductase activity by chlorophenols

Chlorophenols comprise a major class of environmental contaminants. They are extensively used as insecticides, fungicides, mold inhibitors, antiseptics and disinfectants. We found some of these compounds to be strong inhibitors of zeta-crystallin. This oxidoreductase enzyme was isolated from camel lens and its enzymatic activity was inhibited by the chlorophenols tested in a time-independent but concentration-dependent manner. 2,4,5-Trichlorophenol was the most potent inhibitor (IC50 = 3 microM; Ki = 3.2 microM) whereas 4-chlorophenol was the least potent (IC50 = 4.1 mM). There appeared to be a relationship between the degree of chlorination of the phenols and inhibition of zeta-crystallin activity. The position of the chlorine substituent on the phenol may also influence the potency of these compounds.

Mycothiol-dependent formaldehyde dehydrogenase, a prokaryotic medium-chain dehydrogenase/reductase, phylogenetically links different eukaroytic alcohol dehydrogenases--primary structure, conformational modelling and functional correlations

Prokaryotic mycothiol-dependent formaldehyde dehydrogenase has been structurally characterized by peptide analysis of the 360-residue protein chain and by molecular modelling and functional correlation with the conformational properties of zinc-containing alcohol dehydrogenases. The structure is found to be a divergent medium-chain dehydrogenase/reductase (MDR), at a phylogenetic position intermediate between the cluster of dimeric alcohol dehydrogenases of all classes (including the human forms), and several tetrameric reductases/dehydrogenases. Molecular modelling and functionally important residues suggest a fold of the mycothiol-dependent formaldehyde dehydrogenase related overall to that of MDR alcohol dehydrogenases, with the presence of the catalytic and structural zinc atoms, but otherwise much altered active-site relationships compatible with the different substrate specificity, and an altered loop structure compatible with differences in the quaternary structure. Residues typical of glutathione binding in class-III alcohol dehydrogenase are not present, consistent with that the mycothiol factor is not closely similar to glutathione. The molecular architecture is different from that of the 'constant' alcohol dehydrogenases (of class-III type) and the 'variable' alcohol dehydrogenases (of class-I and class-II types), further supporting the unique structure of mycothiol-dependent formaldehyde dehydrogenase. Borders of internal chain-length differences between this and other MDR enzymes coincide in different combinations, supporting the concept of limited changes in loop regions within this whole family of proteins.

New alcohol dehydrogenases for the synthesis of chiral compounds

The enantioselective reduction of carbonyl groups is of interest for the production of various chiral compounds such as hydroxy acids, amino acids, hydroxy esters, or alcohols. Such products have high economic value and are most interesting as additives for food and feed or as building blocks for organic synthesis. Enzymatic reactions or biotransformations with whole cells (growing or resting) for this purpose are described. Although conversions with whole cells are advantageous with respect to saving expensive isolation of the desired enzymes, the products often lack high enantiomeric excess and the process results in low time-space-yield. For the synthesis of chiral alcohols, only lab-scale syntheses with commercially available alcohol dehydrogenases have been described yet. However, most of these enzymes are of limited use for technical applications because they lack substrate specificity, stability (yeast ADH) or enantioselectivity (Thermoanaerobium brockii ADH). Furthermore, all enzymes so far described are forming (S)-alcohols. Quite recently, we found and characterized several new bacterial alcohol dehydrogenases, which are suited for the preparation of chiral alcohols as well as for hydroxy esters in technical scale. Remarkably, of all these novel ADHs the (R)-specific enzymes were found in strains of the genus Lactobacillus. Meanwhile, these new enzymes were characterized extensively. Protein data (amino acid sequence, bound cations) confirm that these catalysts are novel enzymes. (R)-specific as well as (S)-specific ADHs accept a broad variety of ketones and ketoesters as substrates. The applicability of alcohol dehydrogenases for chiral syntheses as an example for the technical use of coenzyme-dependent enzymes is demonstrated and discussed in this contribution. In particular NAD-dependent enzymes coupled with the coenzyme regeneration by formate dehydrogenase proved to be economically feasible for the production of fine chemicals.

Inhibition kinetics of camel lens zeta-crystallin: multiple inhibition studies

The inhibition of camel lens zeta-crystallin by nitrofurantoin (NF) was uncompetitive with respect to co-factor NADPH, (Ki = 90 microM) and competitive with respect to the substrate 9,10-phenanthrenequinone (PQ), (Ki = 50 microM). Inhibition at micromolar concentrations was also observed with dicoumarol, NADP+ and cibacron blue (CB). Theorell-Yonetani double-inhibition analysis showed that NF and dicoumarol were mutually exclusive inhibitors against PQ. However, analysis of NF and NADP+ by a double-inhibition plot showed that they simultaneously bind to the enzyme molecule. These studies demonstrate that NF and dicoumarol share the same site so that both molecules are prevented from binding at the same time, while NF and NADP+ can bind simultaneously to different sites on the enzyme. Although CB was noncompetitive with respect to PQ, double inhibition analysis showed that CB and dicoumarol or NF were mutually exclusive inhibitors against PQ, implying a distinct mode of inhibition for CB.

Effect of pH on sheep liver sorbitol dehydrogenase steady-state kinetics

The variation with pH of the kinetic parameters for sorbitol oxidation and fructose reduction by sheep liver sorbitol dehydrogenase has been studied over the pH 5-10 range. The reaction is compulsory ordered in both directions with the coenzyme as the leading substrate, and the rate-determining step in either direction is the enzyme-coenzyme product dissociation. Throughout the pH range, the lack of a primary kinetic isotope effect on Vm with (2H8) sorbitol confirms that the ternary complexes are not of rate-determining significance under maximum velocity conditions. The association rate constants for NAD and NADH increase and decrease, respectively, towards high pH. NAD binding to the enzyme is dependent upon pK values of 9.2 and 9.6. Whereas the dissociation rate constant for NAD release from the enzyme shows no pronounced variation with pH, NADH release is dependent upon pK values of 7.2 and 7.7. The kinetic constants that characterize the dependence on substrate concentration of the steady-state rate of catalysis vary with pH in accordance with a single pK of 7.1 for sorbitol oxidation and of 7.7 for fructose reduction. These pK values reflect the ionization properties of a catalytically essential group, which is tentatively considered to be either the H2O/OH- ligand binding to the catalytic zinc atom or a histidine residue. Catalysis by sorbitol dehydrogenase, due to the absence of a second ionization contribution, appears not to involve any obligatory step of proton transfer to solution at the ternary complex level. A mechanism for sorbitol dehydrogenase catalysis is proposed.

Structural organization of the human sorbitol dehydrogenase gene (SORD)

The primary structure of human sorbitol dehydrogenase (SORD) was determined by cDNA and genomic cloning. The nucleotide sequence of the mRNA covers 2471 bp including an open reading frame that yields a protein of 356 amino acid residues. The gene structure of SORD spans approximately 30 kb divided into 9 exons and 8 introns. The gene was localized to chromosome 15q21.1 by in situ hybridization. Two transcription initiation sites were detected. Three Sp1 sites and a repetitive sequence (CAAA)5 were observed in the 5' noncoding region; no classical TATAA or CCAAT elements were found. The related alcohol dehydrogenases and zeta-crystallin have the same gene organization split by 8 introns, but no splice points coincide between SORD and these gene types. The deduced amino acid sequence of the SORD structure differs at a few positions from the directly determined protein sequence, suggesting allelic forms of the enzyme. High levels of SORD transcripts were observed in lens and kidney, as judged from Northern blot analysis.

A super-family of medium-chain dehydrogenases/reductases (MDR). Sub-lines including zeta-crystallin, alcohol and polyol dehydrogenases, quinone oxidoreductase enoyl reductases, VAT-1 and other proteins

The protein super-family of medium-chain alcohol dehydrogenases (and glutathione-dependent formaldehyde dehydrogenase), polyol dehydrogenases, threonine dehydrogenase, archaeon glucose dehydrogenase, and eye lens reductase-active zeta-crystallins also includes Escherichia coli quinone oxidoreductase, Torpedo VAT-1 protein, and enoyl reductases of mammalian fatty acid and yeast erythronolide synthases. In addition, two proteins with hitherto unknown function are shown to belong to this super-family of medium-chain dehydrogenases and reductases (MDR). Alignment of zeta-crystallins/quinone oxidoreductases/VAT-1 reveals 38 strictly conserved residues, of which approximately half are glycine residues, including those at several space-restricted turn positions and critical coenzyme-binding positions in the alcohol dehydrogenases. This indicates a conserved three-dimensional structure at the corresponding parts of these distantly related proteins and a conserved binding of a coenzyme in the two proteins with hitherto unknown function, thus ascribing a likely oxidoreductase function to these proteins. When all forms are aligned, including enoyl reductases, a zeta-crystallin homologue from Leishmania and the two proteins with hitherto unknown function, only three residues are strictly conserved among the 106 proteins characterised within the superfamily, and significantly these residues are all glycines, corresponding to Gly66, Gly86 and Gly201 of mammalian class I alcohol dehydrogenase. Notably, these residues are located in different domains. Hence, a distant origin and divergent functions, but related forms and interactions, appear to apply to the entire chains of the many prokaryotic and eukaryotic members. Additionally, in the zeta-crystallins/quinone oxidoreductases, a highly conserved tyrosine residue is found. This residue, in the three-dimensional structure of the homologous alcohol dehydrogenase, is positioned at the subunit cleft that contains the active site and could therefore be involved in catalysis. If so, this residue and its role may resemble the pattern of a conserved tyrosine residue in the different family of short-chain dehydrogenases/reductases (SDR).

Features of structural zinc in mammalian alcohol dehydrogenase. Site-directed mutagenesis of the zinc ligands

All four cysteine ligands to the structural zinc atom of human class-I and class-III alcohol dehydrogenase have been exchanged by site-directed mutagenesis in order to study the importance of the metal in the mammalian enzymes. The cysteine residues were replaced with Ala and Ser, residues that are not able to ligand zinc. All mutations resulted in inactive, unstable enzymes, in contrast to the non-mutated human alcohol dehydrogenases that are easily isolated. Northern-blot analysis revealed the presence of the expected mRNAs from expression plasmids constructed with the different mutated and non-mutated alcohol dehydrogenases, and Western-blot analysis gave faint signals for the mutated recombinant proteins from crude extracts. This verifies that the plasmid constructs are correct, but that the translated, mutated proteins lacking the zinc-stabilized local fold, are subject to rapid degradation. Hence, the results directly illustrate the importance of the structural zinc atom in mammalian alcohol dehydrogenase and confirm it as a component with 'structural' properties. The results are compatible with those from sensitivities to proteases and from the structures of other proteins within the super-family, indicating that the structural role of the zinc atom may involve conservation of interfaces regulating the enzyme quaternary structure.

cDNA and deduced amino acid sequences of a human colon dihydrodiol dehydrogenase

A 1.24 kbp cDNA encoding a human colon HT29 cell dihydrodiol dehydrogenase of 323 amino acid residues (M(r) 36,735) has been isolated and sequenced. The deduced amino acid sequence was 97%, 81%, and 76% identical to human liver dihydrodiol dehydrogenase (also called bile acid binder), human liver chlordecone reductase, and bovine lung prostaglandin f synthase, respectively. It was 98% identical in 990 bp overlap with the nucleotide sequence of human liver pseudochlordecone reductase. Northern blots revealed that dihydrodiol dehydrogenase(s) were markedly induced in HT29 colon cells by treatment with the Michael acceptor agent ethacrynic acid.

Mammalian class IV alcohol dehydrogenase (stomach alcohol dehydrogenase): structure, origin, and correlation with enzymology

The structure of a mammalian class IV alcohol dehydrogenase has been determined by peptide analysis of the protein isolated from rat stomach. The structure indicates that the enzyme constitutes a separate alcohol dehydrogenase class, in agreement with the distinct enzymatic properties; the class IV enzyme is somewhat closer to class I (the "classical" liver alcohol dehydrogenase; approximately 68% residue identities) than to the other classes (II, III, and V; approximately 60% residue identities), suggesting that class IV might have originated through duplication of an early vertebrate class I gene. The activity of the class IV protein toward ethanol is even higher than that of the classical liver enzyme. Both Km and kcat values are high, the latter being the highest of any class characterized so far. Structurally, these properties are correlated with replacements at the active site, affecting both substrate and coenzyme binding. In particular, Ala-294 (instead of valine) results in increased space in the middle section of the substrate cleft, Gly-47 (instead of a basic residue) results in decreased charge interactions with the coenzyme pyrophosphate, and Tyr-363 (instead of a basic residue) may also affect coenzyme binding. In combination, these exchanges are compatible with a promotion of the off dissociation and an increased turnover rate. In contrast, residues at the inner part of the substrate cleft are bulky, accounting for low activity toward secondary alcohols and cyclohexanol. Exchanges at positions 259-261 involve minor shifts in glycine residues at a reverse turn in the coenzyme-binding fold. Clearly, class IV is distinct in structure, ethanol turnover, stomach expression, and possible emergence from class I.

The alcohol dehydrogenase system

Alcohol dehydrogenases constitute a complex system of enzymes, classes, isozymes, and allelic variants. The zinc containing, well-known liver enzyme is a class I medium-chain alcohol dehydrogenase. Other classes of this family include the class II protein, the glutathione-dependent formaldehyde dehydrogenase (the class III enzyme), the stomach-expressed class IV form, and the recently defined class V protein. Characterized forms suggest that the glutathione-dependent formaldehyde dehydrogenase is the original ancestor, defining a role for the whole protein family in cellular defense mechanisms. The isozyme-multiple class I protein is derived from an early gene duplication, allowing sub-specialization in vertebrates. Class IV is the one most ethanol-active and appears to be derived from the class I line. Allelic variants within class I, in association with aldehyde dehydrogenase variants, correlate with population differences in ethanol metabolism and hence with susceptibility to develop alcohol-related diseases. The structures also correlate with functional properties and define molecular building units for the whole family.

Molecular characterization of microbial alcohol dehydrogenases

There is an astonishing array of microbial alcohol oxidoreductases. They display a wide variety of substrate specificities and they fulfill several vital but quite different physiological functions. Some of these enzymes are involved in the production of alcoholic beverages and of industrial solvents, others are important in the production of vinegar, and still others participate in the degradation of naturally occurring and xenobiotic aromatic compounds as well as in the growth of bacteria and yeasts on methanol. They can be divided into three major categories. (1) The NAD- or NADP-dependent dehydrogenases. These can in turn be divided into the group I long-chain (approximately 350 amino acid residues) zinc-dependent enzymes such as alcohol dehydrogenases I, II, and III of Saccharomyces cerevisiae or the plasmid-encoded benzyl alcohol dehydrogenase of Pseudomonas putida; the group II short-chain (approximately 250 residues) zinc-independent enzymes such as ribitol dehydrogenase of Klebsiella aerogenes; the group III "iron-activated" enzymes that generally contain approximately 385 amino acid residues, such as alcohol dehydrogenase II of Zymomonas mobilis and alcohol dehydrogenase IV of Saccharomyces cerevisiae, but may contain almost 900 residues in the case of the multifunctional alcohol dehydrogenases of Escherichia coli and Clostridium acetobutylicum. The aldehyde/alcohol oxidoreductase of Amycolatopsis methanolica and the methanol dehydrogenases of A. methanolica and Mycobacterium gasti are 4-nitroso-N,N-dimethylaniline-dependent nicotinoproteins. (2) NAD(P)-independent enzymes that use pyrroloquinoline quinone, haem or cofactor F420 as cofactor, exemplified by methanol dehydrogenase of Paracoccus denitrificans, ethanol dehydrogenase of Acetobacter and Gluconobacter spp. and the alcohol dehydrogenases of certain archaebacteria. (3) Oxidases that catalyze an essentially irreversible oxidation of alcohols, such as methanol oxidase of Hansenula polymorpha and probably the veratryl alcohol oxidases of certain fungi involved in lignin degradation. This review deals mainly with those enzymes for which complete amino acid sequences are available. The discussion focuses on a comparison of their primary, secondary, tertiary, and quaternary structures and their catalytic mechanisms. The physiological roles of the enzymes and isoenzymes are also considered, as are their probable evolutionary relationships.

Aldehyde dehydrogenases: widespread structural and functional diversity within a shared framework

Sequences of 16 NAD and/or NADP-linked aldehyde oxidoreductases are aligned, including representative examples of all aldehyde dehydrogenase forms with wide substrate preferences as well as additional types with distinct specificities for certain metabolic aldehyde intermediates, particularly semialdehydes, yielding pairwise identities from 15 to 83%. Eleven of 23 invariant residues are glycine and three are proline, indicating evolutionary restraint against alteration of peptide chain-bending points. Additionally, another 66 positions show high conservation of residue type, mostly hydrophobic residues. Ten of these occur in predicted beta-strands, suggesting important interior-packing interactions. A single invariant cysteine residue is found, further supporting its catalytic role. A previously identified essential glutamic acid residue is conserved in all but methyl malonyl semialdehyde dehydrogenase, which may relate to formation by that enzyme of a CoA ester as a product rather than a free carboxylate species. Earlier, similarity to a GXGXXG segment expected in the NAD-binding site was noted from alignments with fewer sequences. The same region continues to be indicated, although now only the first glycine residue is strictly conserved and the second (usually threonine) is not present at all, suggesting greater variance in coenzyme-binding interactions.

Basic features of class-I alcohol dehydrogenase: variable and constant segments coordinated by inter-class and intra-class variability. Conclusions from characterization of the alligator enzyme

The enzymatic and structural properties of alligator liver alcohol dehydrogenase have been determined. Aliphatic and alicyclic alcohols serve as substrates for this first reptilian form of the enzyme characterized, with Km values decreasing rapidly from methanol to hexanol, as for the human class I enzymes, and a Km of 1.2 mM for ethanol at pH 9.9. The N-terminus of the 374-residue protein chain is acetyl-blocked. The enzyme is related in descending order to class I > III > V > II of the structurally characterized mammalian alcohol dehydrogenases. This observation is compatible with the presence of a I/III ancestral line. Differences of the enzyme classes exceed those of the species, suggesting an early origin of the classes. Within its enzyme class, the reptilian protein is most closely related to the avian form (82% residue identities), and is closer to the human than to the amphibian form (76%, versus 69%, respectively). This establishes class I alcohol dehydrogenase as an enzyme having fairly constant rate of change during much of vertebrate evolution, approximately 10% residue differences/100 million years of separation between pairs compared. Residues interacting with the substrate and coenzyme are largely conserved. In the alligator enzyme, there are only four replacements in the substrate pocket compared with the human class I gamma subunit, and those are not known to have functional roles. These properties account for the kinetic parameters, and suggest distinct metabolic functions for the class I enzyme in vertebrates. Comparisons of the enzymes of the different vertebrate lines reveal that segment patterns are characteristic features of the class I enzymes. Three segments are 'variable', while two are 'constant', and both these types of segment are identical with those of the classes. There is extensive variability in close proximity to the active site of the enzyme and this appears to constitute a fundamental property of class I liver alcohol dehydrogenases in general.

Zinc coordination in mammalian sorbitol dehydrogenase. Replacement of putative zinc ligands by site-directed mutagenesis

Rat sorbitol dehydrogenase was expressed in Escherichia coli and purified to homogeneity, resulting in a protein with a specific activity of 4.7 U/mg, close to that of the enzyme isolated from mammalian liver. A Glu residue has been postulated to replace the Cys of alcohol dehydrogenase as a ligand to the active-site zinc atom of sorbitol dehydrogenase. This Glu (position 155 in the rat enzyme) was mutated both to Cys, in order to mimic the alcohol dehydrogenase relationships, and to Ala, as a control. A third mutation, Cys164 to Ala, was also performed since Cys has also been considered as a possible zinc ligand. With Ala at position 155, an inactive enzyme was obtained, showing that correct active-site relationships have been destroyed. With Cys at position 155, the enzyme is still partly active, but rapidly looses activity unless stabilized by the addition of ZnSO4. The catalytic efficiency in the oxidation of sorbitol is 120-fold less than that of the native form, and reduction of fructose is lost completely. In contrast, the activity of the Cys164Ala mutant is comparable with that of the native enzyme and, in fact, even increased in the oxidation of sorbitol. Combined, the results strongly suggest that Glu155 is a ligand to the active-site zinc atom. Zinc analysis of the different variants of sorbitol dehydrogenase establishes that all contain one atom of zinc/subunit, also when the catalytic function is lost. Apparently, zinc remains coordinated even after replacement with an amino acid residue (Ala) unable to ligand metal atoms.

Dual relationships of xylitol and alcohol dehydrogenases in families of two protein types

Xylitol dehydrogenase encoded by gene XYL2 from Pichia stipitis is a member of the medium-chain alcohol dehydrogenase family, as evidenced by the domain organization and a distant homology (24% residue identity with the human class I gamma 1 alcohol dehydrogenase). Much of a loop structure is missing, like in mammalian sorbitol and prokaryotic threonine dehydrogenases, many additional differences occur, and relationships are closest with the sorbitol dehydrogenase, the equivalence of which in P. stipitis may actually be the xylitol dehydrogenase. A second P. stipitis gene, also cloned and corresponding to a xylitol dehydrogenase, is highly different from XYL2, but encodes an enzyme with structural properties typical of the short-chain dehydrogenase family, which also contains an alcohol dehydrogenase (from Drosophila). Thus, yeast xylitol dehydrogenases, like alcohol and polyol dehydrogenases from other sources, have dual derivations, combining similar enzyme activities in separate protein families. In contrast to the situation with the other enzymes, both forms of xylitol dehydrogenase are present in one organism.

Zeta-crystallin versus other members of the alcohol dehydrogenase super-family. Variability as a functional characteristic

Species variability of the lens protein zeta-crystallin was correlated with those of alcohol dehydrogenases of classes I and III and sorbitol dehydrogenase in the same protein family. The extent of overall variability, nature of residues conserved, and patterns of segment variability, all fall within the limits typical of the 'variable' group of medium-chain alcohol dehydrogenases. This shows that zeta-crystallin is subject to restrictions similar to those of classical liver alcohol dehydrogenase and therefore derived from a metabolically active enzyme like other enzyme crystallins. Special residues at the active site, however, differ substantially, including an apparent lack of a zinc-binding site. This is compatible with altered functional properties and makes the spread within this medium-chain dehydrogenase family resemble the wide spread within the short-chain dehydrogenases. Schematic plotting is useful for illustrating the differences between 'variable' and 'constant' enzymes.

Comparative proteolysis of sorbitol and alcohol dehydrogenases

Mammalian alcohol and sorbitol dehydrogenases, with distantly related subunits but different substrates, quaternary structures and zinc contents, were evaluated by comparison of their sensitivities to proteases. Sorbitol dehydrogenase is more sensitive to proteolysis than alcohol dehydrogenase, but both enzymes show limited cleavage with Lys-specific and Glu-specific proteases. With the former, the major cleavage in both proteins involves Lys-Lys-Pro segments, at positions 247-248 in alcohol dehydrogenase, surface-positioned after the most distal beta-strand in the coenzyme-binding domain, and at 61-62 in sorbitol dehydrogenase, at another surface in the catalytic domain. Further cleavages affect these two and a third surface. A non-surface cleavage was obtained with the Glu-specific protease and sorbitol dehydrogenase, after insufficient protease inhibition before analysis by SDS/PAGE. It probably reflects non-native conditions, and the fact that this protease is active in strong SDS, necessitating pre-analytical use of a specific inhibitor. Differences in cleavage patterns between the two proteins do not involve areas corresponding to the dimer interactions in alcohol dehydrogenase. Hence, these areas are likely to be the same in sorbitol dehydrogenase. However, major differences involve regions that flank the surface which has a 20-residue loop segment in alcohol dehydrogenase that is missing in sorbitol dehydrogenase. This segment, previously highlighted from comparisons, is therefore also emphasized by experimental results with proteolysis. The surface exposed by the missing segment is likely to correlate with the separate quaternary structures and to indicate possible sites for the additional subunit interactions in the sorbitol dehydrogenase tetramer versus the alcohol dehydrogenase dimer.

A guinea-pig hereditary cataract contains a splice-site deletion in a crystallin gene

A congenital cataract present in guinea pigs provided a unique opportunity to study a hereditary lens disease at the molecular level. zeta-Crystallin, one of the most abundant guinea pig lens proteins, was found to be altered in the lens of cataractous animals. Several zeta-crystallin cDNA clones were isolated from a cataractous lens library and found to contain a 102-bp deletion towards the 3' end of the coding region. This deletion does not interfere with the reading frame but results in a protein 34 amino acids shorter. Sequence analysis of a normal genomic zeta-crystallin clone revealed that the missing 102-bp fragment corresponds to an entire exon (exon 7). PCR analysis of the genomic DNA isolated from cataractous animals showed that exon 7, though missing from the mRNA, is intact in the cataractous genome. Further sequence analysis of the zeta-crystallin gene disclosed a dinucleotide deletion of the universal AG at the acceptor splice-site of intron 6 of the mutant gene. The presence of this mutation results in the skipping of exon 7 during the mRNA processing which in turn results in the altered zeta-crystallin protein. This is the first time a genomic mutation in an enzyme/crystallin gene has been directly linked to a congenital cataract.

Preliminary X-ray crystallographic studies on alcohol dehydrogenase from Drosophila

The alcohol dehydrogenase (ADHase) enzyme catalyses the oxidation of alcohols to aldehydes or ketones using NAD+ as a cofactor. Functional ADHase from Drosophila lebanonensis is a dimer, with a monomeric molecular weight of 27,000 and with 254 residues in each polypeptide chain. Crystals of the protein have been grown with and without NAD+. Two crystal forms have been observed. Most crystals are plate-like, 0.05 mm in their shortest dimension and up to 0.4 mm in their longest dimension. These crystals are generally too small to diffract efficiently using conventional X-ray sources, so preliminary studies were carried out using the Synchrotron Radiation Source at the SERC Daresbury Laboratory. Twinning was a severe problem with this crystal form. The second form is grown in the absence of NAD+ but with DL-dithiothreitol present. These crystals grow more evenly and diffract to better than 2 A resolution. They are monoclinic, with cell dimensions, a = 81.24(6) A, b = 55.75(4) A, c = 109.60(7) A and beta = 94.26(9) degrees, space group P2(1). There are two dimers in the asymmetric unit, but at low resolution a rotated cell with one dimer per asymmetric unit can be obtained.

Progressive sequence alignment and molecular evolution of the Zn-containing alcohol dehydrogenase family

Sequences of 47 members of the Zn-containing alcohol dehydrogenase (ADH) family were aligned progressively, and an evolutionary tree with detailed branch order and branch lengths was produced. The alignment shows that only 9 amino acid residues (of 374 in the horse liver ADH sequence) are conserved in this family; these include eight Gly and one Val with structural roles. Three residues that bind the catalytic Zn and modulate its electrostatic environment are conserved in 45 members. Asp 223, which determines specificity for NAD, is found in all but the two NADP-dependent enzymes, which have Gly or Ala. Ser or Thr 48, which makes a hydrogen bond to the substrate, is present in 46 members. The four Cys ligands for the structural zinc are conserved except in zeta-crystallin, the sorbitol dehydrogenases, and two bacterial enzymes. Analysis of the evolutionary tree gives estimates of the times of divergence for different animal ADHs. The human class II (pi) and class III (chi) ADHs probably diverged about 630 million years ago, and the newly identified human ADH6 appeared about 520 million years ago, implying that these classes of enzymes may exist or have existed in all vertebrates. The human class I ADH isoenzymes (alpha, beta, and gamma) diverged about 80 million years ago, suggesting that these isoenzymes may exist or have existed in all primates. Analysis of branch lengths shows that these plant ADHs are more conserved than the animal ones and that class III ADHs are more conserved than class I ADHs. The rate of acceptance of point mutations (PAM units) shows that selection pressure has existed for ADHs, implying that these enzymes play definite metabolic roles.