Nucleoside Diphosphate-Sugar 4-Epimerases: II. Uridine Diphosphate Arabinose 4-Epimerase of Wheat Germ 1

Uridine diphosphate (UDP)-arabinose 4-epimerase (EC 5.1.3.5) has been purified at least 20-fold from wheat germ by MnCl(2) treatment, (NH(4))(2)SO(4) fractionation, dialysis, and Sephadex and diethylaminoethyl cellulose column chromatography. The enzyme has no action on UDP-d-glucose, UDP-d-glucuronic acid, or TDP-d-glucose. The pH optimum is 8.0. Km values are 1.5 mM for UDP-d-xylose and 0.5 mm for UDP-l-arabinose. The equilibrium constant, K, for the reaction UDP-l-arabinose left arrow over right arrow UDP-d-xylose is 1.25. The enzyme is neither activated by nicotinamide adenine dinucleotide nor inhibited by reduced nicotinamide adenine dinucleotide. It is completely inhibited by p-chloromercuri-phenylsulfonate; the inhibition is reversed by cysteine.     

Characterization of particulate D-apiosyl-and D-xylosyltransferase from Lemna minor.

A particulate enzyme preparation capable of catalyzing the transfer of d-[U-¹⁴C]apiose and d-[U-¹⁴C]xylose from uridine 5′-(α-d-[U-¹⁴C]apio-d-furanosyl pyrophosphate) (UDP[U-¹⁴C]Api) and uridine 5′-(α-d-[U-¹⁴C]xylopyranosyl pyrophosphate) (UDP[U-¹⁴C]Xyl) to endogenous acceptor molecules was isolated from Lemna minor. The two enzymes were named UDP-d-apiose:acceptor d-apiosyltransferase and UDP-d-xylose:acceptor d-xylosyltransferase and were associated with particulate material sedimenting between 480 and 34,800g. The rate of d-[U-¹⁴C]apiose or d-[U-¹⁴C]xylose incorporation was proportional to the quantity of enzyme preparation used and was constant with time to 1.5 min. Both enzymes showed a pH optimum of 5.7 in citrate-phosphate buffer. The d-apiosyltransferase has a K_m for UDP[U-¹⁴C]Api of 4.9 μm. Bovine serum albumin and sucrose stimulated the rate of incorporation of both pentoses. Both enzymes rapidly lost activity; with our best conditions, approximately 50% of each enzyme activity was lost in 6 min at 25 °C or in 3 h at 4 °C. Incorporation of d-[U-¹⁴C]apiose was obtained in the absence of added uridine 5′-(α-d-galactopyranosyluronic acid pyrophosphate) (UDPGalUA); however, the addition of UDPGalUA not only almost doubled the rate of incorporation, but also increased the total incorporation of d-[U-^l4C]apiose and extended the proportional range of incorporation at 25 °C from 1.5 to 2 min.     

Solubilization of a mannose-polymerizing enzyme from Phaseolus aureus

A soluble enzyme preparation, which catalyses the polymerization of mannose, was obtained by Triton X-100 extraction of a particulate fraction derived from Phaseolus aureus hypocotyls. The product that resulted when GDP-alpha-d-mannose was used as a substrate was a beta-(1-->4)-linked mannan, about three-quarters of which was alkali-insoluble. The mannose-polymerizing enzyme activity was at least as great in the soluble preparation as in the particulate preparation, and the specific activity of the solubilized enzyme was greater by a factor of at least 3.5. Kinetic studies of the soluble enzyme indicate that the apparent K(m) is 55-62mum, and a disproportionate increase in rate is observed at high concentrations. GDP-alpha-d-glucose is a strong competitive inhibitor of the mannose-polymerizing reaction, with an apparent K(i) of 6.2mum. The soluble enzyme is relatively unstable, losing about two-thirds of its original activity in 5h at 0 degrees C or in 24h at -20 degrees C. A solvent (acetone, butanol, diethyl ether)-extracted particulate preparation, which also exhibits the same enzyme activity, is more stable, retaining full activity for at least 5 days at -20 degrees C. There was no polymerizing-enzyme activity in the soluble enzyme preparation when UDP-d-glucose, UDP-d-galactose, UDP-d-xylose, UDP-l-arabinose or UDP-d-glucuronic acid were used as substrates. However, the soluble enzyme preparation would catalyse the polymerization of glucose, with GDP-d-glucose as substrate.     

Studies of coenzyme binding to rabbit muscle glyceraldehyde-3-phoshate dehydrogenase.

The binding of NAD+, NADH and adenosine diphosphoribose (Ado-PP-Rib) to a stable, highly active and nucleotide-free preparation of rabbit muscle glyceraldehyde-3-phosphate dehydrogenase (D-glyceraldehyde-3-phosphate: NAD+ oxidoreductase (phosphorylating), EC 1.2.1.12) has been studied. All three nucleotides quench the protein fluorescence to the same extent when they bind to the enzyme, and this property has been used to measure the dissociation constants for the two high-affinity binding sites for the nucleotides. The results indicate negative interactions between, or non-identify of, these two binding sites, to which NAD+ and NADH bind with similar affinity. The binding of NAD+ to the enzyme has been studied by spectrophotometric titrations at 360 nm. It appears that the binding of NAD+ to each of the four subunits of the enzyme contributes equally to the intensity of this 'Racker' band. The dissociation constants associated with the binding of the third and fourth molecules of NAD+ estimated from such titrations confirm some previous estimates. The binding of NADH to the enzyme causes a decrease of intensity of the absorbance of the coenzyme at 340 nm, and the dissociation constants for binding of the third and fourth molecules of NADH have been estimated from spectrophotometric titrations. They are the same as those for NAD+. Judging by the apparent dissociation constants, negative interactions on binding the third molecule of NAD+ or NADH are more marked than those associated with the binding of the second and fourth molecules, suggesting that a major conformational change occurs at half-saturation of the tetramer with coenzyme.     

Cloning and expression of a UDP-glucuronic acid decarboxylase gene in rice

A cDNA fragment was cloned from rice immature seeds by the RT-PCR method. The deduced amino acid sequence of the cDNA showed a high degree of identity with UDP-d-glucuronic acid decarboxylase (UXS) from other plants and was most similar to the soluble UXS from Arabidopsis. The recombinant protein, expressed in an Escherichia coli system, catalysed the conversion of UDP-d-glucuronic acid to UDP-d-xylose, confirming that the gene encoded UXS. The uxs gene was expressed in mature, harvested rice seeds as well as in immature seeds 14 d post-anthesis, suggesting that the uxs gene is necessary at the beginning of the germination period. This is the first report of the cloning of the uxs gene from monocots.     

Relationship between hydroxypyruvate and the production of oxalate in vitro.

Chicken liver lactate dehydrogenase (L-lactate:NAD+ oxidoreductase, EC1.1.1.27) catalyses the reversible reduction reaction of hydroxypyruvate to L-glycerate. It also catalyses the oxidation reaction of the hydrated form of glyoxylate to oxalate and the reduction of the non-hydrated form of glyoxylate to oxalate and the reduction of the non-hydrated form to glycolate. At pH 8, these latter two reactions are coupled. The coupled system equilibrium is attained when the NAD+/NADH ratio is greater than unity. Hydroxypyruvate binds to the enzyme at the same site as the pyruvate. When there are substances with greater affinity to this site in the reaction medium and their concentration is very high, hydroxypyruvate binds to the enzyme at the L-lactate site. In vitro and with purified preparation of lactate dehydrogenase, hydroxypyruvate stimulates the production of oxalate from glyoxylate-hydrated form and from NAD; the effect is due to the fact that hydroxypyruvate prevents the binding of non-hydrated form of glyoxylate to the lactate dehydrogenase in the pyruvate binding site. At pH 8, THE L-glycerate stimulates the production of glycolate from glyoxylate-non-hydrated form and NADH since hydroxypyruvate prevents the binding of glyoxylate-hydrated form to the enzyme     

Effect of benzodiazepines on NAD binding to synaptic membrane in the rat brain

The receptor protein solubilized from synaptic membranes specifically binds [14C] NAD (dissociation constant--0.75 microM, capacity of binding sites--0.0125 nmol of metaid per 1 mg of protein). All the studied benzodiazepines (phenazepam, nitrazepam, clonazepam, flunitrazepam) are able to displace [14C] NAD from its receptor sites, the mixed type of inhibition being manifested. An inhibition constant for flunitrazepam, a ligand of benzodiazepine receptors, equals 10 microM. GABA promotes an inhibiting effect of benzodiazepines. It is supposed that neurotropic action of NAD is realized through the GABA-benzodiazepine complex of neuronal membranes.     

The biosynthesis of the branched-chain sugar d-apiose in plants: functional cloning and characterization of a UDP-d-apiose/UDP-d-xylose synthase from Arabidopsis

d-Apiose is a plant-specific branched-chain monosaccharide found in rhamnogalacturonan II (RG-II), apiogalacturonan, and several apioglycosides. Within RG-II, d-apiose serves as the binding site for borate, which leads to the formation of cross-links within the wall. Biochemical studies in duckweed and parsley have established that uridine 5'-diphospho-d-apiose (UDP-d-apiose) is formed from UDP-d-glucuronate by decarboxylation and re-arrangement of the carbon skeleton, leading to ring contraction and branch formation. The enzyme catalyzing this reaction also forms UDP-d-xylose by decarboxylation of UDP-d-glucuronate, and has therefore been named UDP-d-apiose/UDP-d-xylose synthase. Using a bioinformatics approach, we identified a candidate gene (AXS1) for this enzyme in Arabidopsis and functionally expressed its cDNA in Escherichia coli. The recombinant enzyme catalyzed the conversion of UDP-d-glucuronate to a mixture of UDP-d-apiose and UDP-d-xylose with a turnover number of 0.3 min-1. AXS1 required NAD+ for enzymatic activity, and was strongly inhibited by UDP-d-galacturonate. It was highly expressed in all plant organs consistent with a function in synthesizing an essential cell wall precursor. Database searches indicated the presence of closely related sequences in a variety of crop plants. The cloning of the AXS1 gene will help to investigate the biosynthesis of RG-II, and permit insights into the mechanism by which d-apiose and other branched monosaccharides are formed.     

Inactivation of rabbit muscle glyceraldehyde-3-phosphate dehydrogenase by koningic acid

Koningic acid, a sesquiterpene antibiotic, is a specific inhibitor of the enzyme glyceraldehyde-3-phosphate dehydrogenase (D-glyceraldehyde-3-phosphate:NAD+ oxidoreductase (phosphorylating), EC 1.2.1.12). In the presence of 3 mM of NAD+, koningic acid irreversibly inactivated the enzyme in a time-dependent manner. The pseudo-first-order rate constant for inactivation (kapp) was dependent on koningic acid concentration in saturate manner, indicating koningic acid and enzyme formed a reversible complex prior to the formation of an inactive, irreversible complex; the inactivation rate (k 3) was 5.5.10(-2) s-1, with a dissociation constant for inactivation (Kinact) of 1.6 microM. The inhibition was competitive against glyceraldehyde 3-phosphate with a Ki of 1.1 microM, where the Km for glyceraldehyde 3-phosphate was 90 microM. Koningic acid inhibition was uncompetitive with respect to NAD+. The presence of NAD+ accelerated the inactivation. In its absence, the charcoal-treated NAD+-free enzyme showed a 220-fold decrease in apparent rate constant for inactivation, indicating that koningic acid sequentially binds to the enzyme next to NAD+. The enzyme, a tetramer, was inactivated when maximum two sulfhydryl groups, possibly cysteine residues at the active sites of the enzyme, were modified by the binding of koningic acid. These observations demonstrate that koningic acid is an active-site-directed inhibitor which reacts predominantly with the NAD+-enzyme complex.     

Purification and properties of a nad: 3α-hydroxy-5α-pregnan-20-one-oxidoreductase from rat liver microsomes

From rat liver microsorties a NAD: 3α-hydroxy-5α-pregnan-20-one oxidoreductase was isolated and purified up to a specific activity of 73 nmol/min.mg by affinity chromatography and DEAE-cellulose chromatography. Various Km-values have been determined. The enzyme exhibits highest affinity for 5α-pregnane-3,20-dione and NADH. The 3-oxo group of 5α-dihydrocortisone (17, 21-dihydroxy-5α-pregnane-3,11,20-trione) was not reduced by the purified enzyme preparation and NADH and no dehydrogenation with NAD was observed of 3α, 11β, 17, 21-tetrahydroxy-5α-pregnan-20-one. The optimal pH for the hydrogenation of the 3-oxo group was at pH 5.3 and for the dehydrogenation at pH 8.9. Disc gel electrophoresis in presence of 0.1% sodium dodecylsulfate yielded a homogeneous preparation.     

Purification and characterization of Haemophilus influenzae D-lactate dehydrogenase

Haemophilus influenzae D(-)-lactate dehydrogenase (D(-)-lactate:NAD oxidoreductase; EC 1.1.1.28) was purified to electrophoretic homogeneity using salt fractionation, hydrophobic and dye affinity chromatography. The enzyme was purified 2100-fold with a 14% recovery and a final specific activity of 300 units/mg protein. The enzyme was demonstrated to be a tetramer of Mr 135,000. The enzyme catalyzed the reduction of pyruvate to give exclusively D(-)-lactate using NADH as coenzyme. The reaction catalyzed was essentially unidirectional, with the oxidation of D-lactate in the presence of NAD proceeding at less than 0.2% the rate of pyruvate reduction. Kinetic parameters for the reduction of pyruvate were determined for NADH and four structural analogs of the coenzyme. Coenzyme-competitive inhibition by adenosine derivatives indicated the presence of regions in the coenzyme binding site interacting with the adenosine and pyrophosphate moieties of the coenzyme. The purified enzyme was sensitive to oxidation and was effectively inactivated by sulfhydryl reagents. Conversion of D-lactate to pyruvate catalyzed by a membrane-bound D-lactate oxidase was demonstrated in cell-free extracts of H. influenzae.     

Insights into the evolution of allosteric properties. The NADH binding site of hexameric type II citrate synthases

Study of the hexameric and allosterically regulated citrate synthases (type II CS) provides a rare opportunity to gain not only an understanding of a novel allosteric mechanism but also insight into how such properties can evolve from an unregulated structural platform (the dimeric type I CS). To address both of these issues, we have determined the structure of the complex of NADH (a negative allosteric effector) with the F383A variant of type II Escherichia coli CS. This variant was chosen because its kinetics indicate it is primarily in the T or inactive allosteric conformation, the state that strongly binds to NADH. Our structural analyses show that the six NADH binding sites in the hexameric CS complex are located at the interfaces between dimer units such that most of each site is formed by one subunit, but a number of key residues are drawn from the adjacent dimer. This arrangement of interactions serves to explain why NADH allosteric regulation is a feature only of hexameric type II CS. Surprisingly, in both the wild-type enzyme and the NADH complex, the two subunits of each dimer within the hexameric conformation are similar but not identical in structure, and therefore, while the general characteristics of NADH binding interactions are similar in each subunit, the details of these are somewhat different between subunits. Detailed examination of the observed NADH binding sites indicates that both direct charged interactions and the overall cationic nature of the sites are likely responsible for the ability of these sites to discriminate between NADH and NAD(+). A particularly novel characteristic of the complex is the horseshoe conformation assumed by NADH, which is strikingly different from the extended conformation found in its complexes with most proteins. Sequence homology studies suggest that this approach to binding NADH may arise out of the evolutionary need to add an allosteric regulatory function to the base CS structure. Comparisons of the amino acid sequences of known type II CS enzymes, from different Gram-negative bacteria taxonomic groups, show that the NADH-binding residues identified in our structure are strongly conserved, while hexameric CS molecules that are insensitive to NADH have undergone key changes in the sequence of this part of the protein.     

Purification and substrate specificity of UDP-D-xylose:beta-D-glucoside alpha-1,3-D-xylosyltransferase involved in the biosynthesis of the Xyl alpha1-3Xyl alpha1-3Glc beta1-O-Ser on epidermal growth factor-like domains

A unique O-glycan structure, Xylalpha1-3Xylalpha1-3Glcbeta1-O-Ser is found on the consensus sequence C-X-S-X-P-C (X denotes any amino acid) in epidermal growth factor (EGF)-like domains of plasma proteins such as clotting factor VII and IX. One of the enzymes involved in the biosynthesis of this trisaccharide, UDP-d-xylose:beta-d-glucoside 1,3-d-xylosyltransferase has been identified in HepG2 cells (Omichi, K., Aoki, K., Minamida, S., and Hase, S. Eur. J. Biochem. 245, 143-146 [1997]). Here, we report that this enzyme activity can be detected in bovine liver and that the enzyme has been purified from the microsomal fraction. The enzyme was purified 6200-fold in terms of specific activity and ran as a single band on native-PAGE and isoelectric focusing gel electrophoresis. The best acceptor substrate of those tested was the EGF-like domain of bovine factor IX carrying beta-glucoside at Ser53. The Km value for this substrate was 34 muM. Comparison of initial velocity with various acceptor substrates shows that this xylosyltransferase recognizes not only the glucose moiety to which xylose is transferred but also the tertiary structure of the EGF-like domain. With regard to the donor substrate, the enzyme does not recognize UDP-d-glucose but does recognize UDP-d-xylose.     

Reduction of nicotinamide adenine dinucleotide by pyruvate:lipoate oxidoreductase in anaerobic, dark-grown Rhodospirillum rubrum mutant C.

Cell extracts from fermentatively grown Rhodospirillum rubrum reduced about 80 nmol of nicotinamide adenine dinucleotide (NAD) per mg of protein per min under anaerobic conditions with sodium pyruvate. The reaction was specific for pyruvate and NAD; NAD phosphate was not reduced. Results indicated that pyruvate-linked NAD reduction occurred via pyruvate:lipoate oxidoreductase. The reaction required catalytic amounts of both coenzyme A and thiamine pyrophosphate. Addition of sodium arsenite inhibited enzyme activity by 90%. Pyruvate:lipoate oxidoreductase was the only system detected in anaerobic, dark-grown R. rubrum cell extracts which operated to produce reduced NAD. The low activity of the enzyme system suggested that it was not quantitatively important in ATP formation.     

A novel adenylate binding site confers phosphopantetheine adenylyltransferase interactions with coenzyme A

Phosphopantetheine adenylyltransferase (PPAT) regulates the key penultimate step in the essential coenzyme A (CoA) biosynthetic pathway. PPAT catalyzes the reversible transfer of an adenylyl group from Mg(2+):ATP to 4'-phosphopantetheine to form 3'-dephospho-CoA (dPCoA) and pyrophosphate. The high-resolution crystal structure of PPAT complexed with CoA has been determined. Remarkably, CoA and the product dPCoA bind to the active site in distinct ways. Although the phosphate moiety within the phosphopantetheine arm overlaps, the pantetheine arm binds to the same pocket in two distinct conformations, and the adenylyl moieties of these two ligands have distinct binding sites. Moreover, the PPAT:CoA crystal structure confirms the asymmetry of binding to the two trimers within the hexameric enzyme. Specifically, the pantetheine arm of CoA bound to one protomer within the asymmetric unit displays the dPCoA-like conformation with the adenylyl moiety disordered, whereas CoA binds the twofold-related protomer in an ordered and unique fashion.     

Enzymatic synthesis and reactions of uridine 5'-(5-thio-alphaD-glucopyranosyl pyrophosphate).

Uridine 5′-(5-thio-α-d-glucopyranosyl pyrophosphate), UDPTG, is an efficient substrate for yeast uridine 5′-(d-glucopyranosyl pyrophosphate), UDPG, pyrophosphorylase. K_m for UDPTG with the pyrophosphorylase is 0.2 mm and the analog reacts with a maximal velocity 96% that of UDPG. UDPTG is also a substrate for yeast UDP-galactose 4-epimerase. Although not a substrate for bovine liver UDPG dehydrogenase, UDPTG is a potent, mixed-type inhibitor with respect to both UDPG and nicotinamide adenine dinucleotide (NAD). UDPTG is synthesized in 30% yield from 5-thio-d-glucopyranose and in 85% yield from 5-thio-α-d-glucopyranose 1-phosphate by using mixtures of commercially available enzymes. The pK_a of the uracil moiety in UDPTG is the same as that in UDPG, and UDPTG appears to be similar to UDPG in the extent of secondary structural order. UDPTG, however, is more highly acid-labile than UDPG.     

A transfer nuclear Overhauser effect study of coenzyme binding to distinct sites in binary and ternary complexes in glutamate dehydrogenase

The oxidized coenzyme NAD binds to two sites per subunit of bovine liver glutamate dehydrogenase with equal affinity in the absence of dicarboxylic acid coligands. In the presence of glutarate or 2-oxoglutarate, the affinity to one site is unchanged, but the affinity to the other (presumed to be the active site) is considerably increased and now requires two dissociation constants to describe its saturation. A combination of transfer nuclear Overhauser effects (TRNOE) together with an examination of the slopes of TRNOE time dependence indicates that while NAD is bound in a syn conformation at both binding sites, NADP (which binds only to the active site) is bound in a syn-anti mixture. The existence of N6 to N3' and N6 and N2' and N1' to N3' NOE's with NAD suggests that the two coenzyme binding sites are located near enough to allow intermolecular NOE's. In the presence of 2-oxoglutarate where only binding to the active site is effectively observed, the conformation of either coenzyme is syn. Modeling studies using the distance estimates from the TRNOE results suggest that the nicotinamide ribose approximates a 3'-endo conformation. The absence of evidence for intermolecular NOE's under these conditions indicates that while the active and regulatory NAD sites per subunit are in close proximity, the six active sites per hexamer are located greater than 5 A apart.     

Location of lysyl residues at the allosteric site of fructose 1,6-bisphosphatase

Various flavin analogs were used as alternate substrates or competitive inhibitors to characterize the FMN binding sites of the NADH- and NADPH-specific FMN oxidoreductases from Beneckea harveyi. Several polyhydroxyl compounds were found to be poor competitive inhibitors for the FMN sites of these enzymes. The FMN binding sites of the two enzymes were found to be quite similar. The NADH:FMN oxidoreductase binds FMN exclusively through the isoalloxazine ring. The methyl groups at positions 7 and 8 contribute significantly to this binding. Utilizing lumichrome as a competitive inhibitor of the FMN binding site and AMP as a competitive inhibitor of the NADH binding site, we were able to determine that the NADH:FMN oxidoreductase forms an active ternary complex with NADH binding first in an ordered mechanism. The NADPH oxidoreductase also binds FMN primarily through the isoalloxazine ring. Unlike their participation in reaction with the NADH-specfic enzyme, the methyl groups at positions 7 and 8 are not involved in binding. There was no significant binding of the ribityl phosphate moiety with either enzyme. Both enzymes have lower K_m values for lumiflavin than FMN.     

Purification and properties of alanine dehydrogenase from Bacillus sphaericus.

1. The bacterial distribution of alanine dehydrogenase (L-alanine:NAD+ oxidoreductase, deaminating, EC 1.4.1.1) was investigated, and high activity was found in Bacillus species. The enzyme has been purified to homogeneity and crystallized from B. sphaericus (IFO 3525), in which the highest activity occurs. 2. The enzyme has a molecular weight of about 230 000, and is composed of six identical subunits (Mr 38 000). 3. The enzyme acts almost specifically on L-alanine, but shows low amino-acceptor specificity; pyruvate and 2-oxobutyrate are the most preferable substrates, and 2-oxovalerate is also animated. The enzyme requires NAD+ as a cofactor, which cannot be replaced by NADP+. 4. The enzyme is stable over a wide pH range (pH 6.0--10.0), and shows maximum reactivity at approximately pH 10.5 and 9.0 for the deamination and amination reactions, respectively. 5. Alanine dehydrogenase is inhibited significantly by HgCl2, p-chloromercuribenzoate and other metals, but none of purine and pyrimidine bases, nucleosides, nucleotides, flavine compounds and pyridoxal 5'-phosphate influence the activity. 6. The reductive amination proceeds through a sequential ordered ternary-binary mechanism. NADH binds first to the enzyme followed by ammonia and pyruvate, and the products are released in the order of L-ALANINE AND NAD+. The Michaelis constants are as follows: NADH (10 microM), ammonia (28.2 mM), pyruvate (1.7 mM), L-alanine (18.9 mM) and NAD+ (0.23 mM). 7. The pro-R hydrogen at C-4 of the reduced nicotinamide ring of NADH is exclusively transferred to pyruvate; the enzyme is A-stereospecific.     

Purification and characterization of osmoregulatory betaine aldehyde dehydrogenase of Escherichiacoli

The osmoregulatory NAD-dependent betaine aldehyde dehydrogenase (betaine aldehyde: NAD oxidoreductase, EC 1.2.1.8), of Escherichia coli, was purified to apparent homogeneity from an over-producing strain carrying the structural gene for the enzyme (betB) on the plasmid vector pBR322. Purification was achieved by ammonium sulfate fractionation of disrupted cells, followed by affinity chromatography on 5′-AMP Sepharose, gel-filtration and ion-exchange chromatography. The amino acid composition was determined. The dehydrogenase was found to be a tetramer with identical 55 kDa subunits. Both NAD and NADP could be used as cofactor for the dehydrogenase, but NAD was preferred. The dehydrogenase was highly specific for betaine aldehyde. None of the analogs tested functioned as a substrate, but several inhibited the enzyme competitively. The enzyme was not activated by salts at concentrations encountered during osmotic upshock, but it was salt tolerant, retaining 50% of maximal activity at 1.2 M K⁺. It is inferred that salt tolerance is an essential property for an enzyme participating in the cellular synthesis of an osmoprotectant.