Stephen A. Kuby

Studies on adenosine triphosphate transphosphorylases: XI. Isolation of the crystalline adenosine triphosphate-creatine transphosphorylases from the muscle and brain of man, calf, and rabbit; and a preparation of their enzymatically active hybrids

Abstract Detailed procedures are described for the isolation of the crystalline isoenzymes of adenosine triphosphate-creatine transphosphorylase, in good yield, from the muscle and brain of man, calf, and rabbit. The original isolation procedure of Kuby et al. (1) for the rabbit muscle enzyme has been modified to include a batchwise adaptation of a chromatographic step to be described herein for the other muscle types, to permit the isolation of large amounts of a relatively pure preparation for chemical studies. In addition, another convenient procedure is given for the crystallization of the rabbit muscle enzyme from aqueous (NH 4 ) 2 SO 4 solutions, which might be amenable to X-ray structural analyses. Practicable procedures are presented for the preparation, in excellent yields, of the enzymatically active hybrids of the ATP-creatine transphosphorylase, from their component polypeptide chains of the muscle- and of the brain-type enzymes from both man and calf. Descriptions are given for the disruption and annealing of each of the component polypeptide chains of their respective isolated isoenzymes; and a chromatographic system is provided for the separation of the reannealed isoenzymes (with the hybrid included) as native species. Finally, the stability properties of each of the several (six) isoenzymes from the muscle and brain are briefly outlined, in particular, with respect to pH and temperature, to provide a future basis for their physicochemical characterization and comparison. In the case of the human isoenzymes (muscle- and brain-type), a comparative study has been conducted on the kinetics of H + -induced inactivation. These studies illustrate the greater acid lability of the brain type.

Studies on adenosine triphosphate transphosphorylases XIV. Equilibrium binding properties of the crystalline rabbit and calf muscle ATP-AMP transphosphorylase (adenylate kinase) and derived peptide fragments☆

Abstract Both a fluorescence-quenching technique and a uv-difference spectral method have been used to study the binding of 1, N 6 -etheno analogs of the adenine nucleotides (ϵATP, ϵADP, ϵAMP) ( J. A. Secrist III, J. R. Barrio, N. J., Leonard, and G. Weber, 1972 , Biochemistry , 11 , 3499–3506) to crystalline rabbit and calf muscle ATP-AMP transphosphorylase in the presence and absence of Mg 2+ , at 0.16 ( Γ 2 ), 25 °C, and pH 7.4. In addition, the binding of the ϵ-analogs of the adenine nucleotides has been studied to two S -[ 14 C]carboxymethylated peptide fragments of the rabbit muscle enzyme (residues 1–44 = MT-I; residues 171–193 = MT-XII), as well as to a synthetic nonapeptide corresponding to residues 32 − 40 of the rabbit muscle enzyme. In the case of the rabbit and calf enzymes: MgϵATP 2− , ϵATP 4− , MgϵADP − , and ϵAMP 2− are bound stoichiometrically ( n ∼- 1), MgϵAMP is insignificantly bound, and n ∼- 2 for ϵADP 3− ( n = maximal number of moles bound per mole of protein). In the case of S -carboxymethylated peptide fragments: MT-I binds stoichiometrically to MgϵATP 2− , ϵATP 4− , MgϵADP − , and ϵADP 3− with values of n ∼- 1; but MT-I does not bind to ϵAMP 2− significantly. MT-XII binds stoichiometrically to uncomplexed ϵAMP 2− or to uncomplexed ϵADP 3− (both with n ∼- 1); whereas, the binding of MgϵADP − , ϵATP 4− , and MgϵAMP to MT-XII are comparatively insignificant. Other peptide fragments in the molecule, viz. fragments MT-IV (residues 77–96) or MT-VI (residues 106–126) did not bind significantly to any of the ethenoanalogs; nor did insulin, nor, e.g., did bo vine serum albumin. The binding of the etheno analogs was also studied to an equimolar mixture of peptides MT-I + MT-XII, which qualitatively duplicated the binding pattern of the entire native molecule, and except for ϵATP 4− or MgϵATP 2− (which are bound more tightly to the entire native molecule), even quantitatively. The synthetic peptide (residues 32 to 40) was found to bind to MgϵATP 2− , ϵATP 4− , and MgϵADP − , with n ∼- 1; but it does not significantly bind to ϵAMP 2− , nor to ϵADP 3− . These binding data support the idea that there are two separate sites for the binding of either (a) the complexed nucleotide substrate (MgATP 2− or MgADP − ) residing in the sequence of MT-I (residues 1 to 44) and in the neighborhood of residues 32 to 40, or (b) the uncomplexed nucleotide substrate (AMP 2− or ADP 3− ) residing in the sequence of MT-XII (residues 171 to 193) of the rabbit muscle enzyme.

Studies on adenosine triphosphate transphosphorylases: Isolation and several properties of the crystalline calf ATM-AMP transphosphorylases (adenylate kinases) from muscle and liver and some observations on the rabbit muscle adenylate kinase☆

Abstract Detailed procedures are described for the large-scale isolation of crystalline calf muscle and calf liver adenylate kinase isoenzymes applicable to frozen tissue. Proposed modifications in the large-scale isolation of the crystalline rabbit muscle adenylate kinase are also described. Although a great deal of homology exists between the muscle types, calf and rabbit, as revealed by their amino acid compositions (differing in only 10 out of 193 residues), the calf muscle adenylate kinase acts as a potent antigen in the rabbit. A description for the purification (ca. 600-fold) from the rabbit antiserum of the anti-enzyme globulin is given. The anti-enzyme shows a remarkable specificity in that it is largely unreactive as an inhibitor toward the calf liver-type enzyme, but is a powerful inhibitor of the calf muscle type (or human muscle type) with a kinetically evaluated dissociation constant of ca. 3 × 10−14 m 2 (for the case of a stoichiometry of 2 Ag to 1 Ab), a second-order k ⋍ 5 × 10 6 M −1 · min −1 at 30 ° C , and an Arrhenius energy of activation of ca. 6.3 kcal/mol. Dodecyl sulfate polyacrylamide gel electrophoresis revealed only a single heavy chain (Mr, ~42,000) and a single light chain (Mr, ~25,000) of the antibody, which together with studies employing 3.5% stacking gels lead to a molecular weight of approximately 134,000 for the native antibody. Similarly, by dodecyl sulfate polyacrylamide gel electrophoreses, only single bands were obtained for the calf liver and rabbit muscle adenylate kinase (Mr, ca. 21,000–23,000) and the calf liver adenylate kinase (Mr, ca. 25,000). Sedimentation velocity of the calf muscle-type and calf liver-type proteins yielded single sedimenting boundaries, with s 0 20,w ⋍ 2.1 6 and 2.37 S, respectively. Estimates of the molecular weights of the native structure of the calf isoenzymes by sedimentation equilibrium gave M w = 21,200 ± 200 for the calf muscle enzyme and 25,600 ± 200 for the calf liver adenylate kinase. Electrophoresis by liquid boundary of the rabbit muscle enzyme at protein concentrations of 5–8 mg/ml resulted in a nonideal system, as a result largely of charge effects and interactions with buffer components. Similarly, isoelectric points determined by isoelectric focusing were very dependent on the protein load, but could be extrapolated to zero protein to approximately pI0 of 10.6. However, electrophoresis on cellulose acetate at microgram to nanogram levels of protein under conditions where electroendosmosis seemed absent did permit an accurate estimate of the pI0 by a plot of pIapp vs ( Γ 2 ) 1 2 to a pI0 = 10.60. This value is now in good agreement with an estimate made earlier from its amino acid composition (T. A. Mahowald et al., 1962, J. Biol. Chem.237, 1138–1145). By this technique, the pIapp values at 0.05 ( Γ 2 ) for the calf isoenzymes were estimated and found to differ only slightly (viz., 9.6 for the liver adenylate kinase vs 10.1 for the muscle myokinase). While inhibition of the rabbit and calf muscle adenylate kinases (which are apparently cytoplasmic enzymes) by Ap5A [p1,p5-di(adenosine-5′)pentaphosphate] was very significant at even 10−8 m and similar to that previously reported for the rabbit muscle adenylate kinase ( Lienhard and Secemski, 1973 , J. Biol. Chem.248, 1121–1123), inhibition of the liver-type adenylate kinase (which is apparently a mitochondrial enzyme) required almost 10−6 m (Ap5A) for similar percentage inhibitions. For the forward reactions (i.e., MgATP2− + AMP2− →), Ap5A acts as a competitive inhibitor with respect to either substrate with K i ⋍ (0.5−1) × 10 −8 M for both of the muscle enzymes, but yields a K i ⋍ (3−8) × 10 −7 M for the liver enzyme. However, in the reverse direction (MgADP− + ADP3− →), inhibition by Ap5A appears to be noncompetitive with respect to either substrate, with a K i ⋍ (0.6−2) × 10 −8 M for the muscle enzymes, but only (3–4) × 10−6 m for the liver enzyme. Thus, although Ap5A may act in part as a transition state analog, the explanation for its “multisubstrate inhibition” (Lienhard and Secemski, see above) may also lie in the structure of its metal chelates, e.g., of Mg2(Ap5A)−.

Studies on adenosine triphosphate transphosphorylases: XIII. Kinetic properties of the crystalline rabbit muscle ATP-AMP transphosphorylase (adenylate kinase) and a comparison with the crystalline calf muscle and liver adenylate kinases

Abstract The steady-state kinetics, at pH 7.4, 25 °C and essentially fixed (Γ/2) of 0.16–0.18, of the rabbit muscle, calf muscle, and calf liver adenylate kinases all seem to be adequately expressed by a random quasi-equilibrium type of mechanism with a rate-limiting step largely at the interconversion of the ternary complexes. However, the data presented, by themselves, will not exclude the possibility of a random mechanism in which product release is rate limiting. In the case of the rabbit muscle enzyme, five substrate pairs for the forward reaction (MgATP 2− /AMP 2− ; MgdATP 2− /dAMP 2− ; MgATP 2− /dAMP 2− ; MgdATP 2− /AMP 2− ; MgϵATP 2− /AMP 2− ) and two substrate pairs for the reverse reaction (MgADP − /ADP 3− ; MgdADP − /DP 3− ) were studied under conditions where a careful and systematic control was exercised over the metal complex nucleotide species. For the forward reactions between MgATP 2− + AMP 2− , and MgdATP 2− + dAMP 2− , the estimated values for K s,1 and K s ,1 (i.e., intrinsic dissociation constants of the substrate from the binary and ternary complexes) differed, pointing to substrate (s) induced conformational changes in the ternary complexes. The fact that similar derived values either for K AMP 2− - and for K AMP 2− could be estimated for each of the above three sets of substrate pairs in the forward reaction, supports the conclusion that a common rate-limiting step exists and is likely in the interconversion of the ternary complexes. ϵAMP 2− was found to be a competitive inhibitor of AMP 2− (at fixed MgATP 2− ) and MgdADP − to be noncompetitive inhibitor of ADP 3− (at fixed MgADP − ), in support of the conclusion of D. G. Rhoads and J. M. Lowenstein ( J. Biol. Chem . (1968) 243 , 3963–3972) that a separate site exists for the magnesium chelates of the nucleotide substrates (MgATP 2− and MgADP − ) and for the uncomplexed nucleotide substrates (ADP 3− and AMP 2− ). Distinguishing kinetic features of the calf liver adenylate kinase over the calf muscle isoenzyme are: 1. (3) inhibition of the liver enzyme by phosphoenolpyruvate; and 2. (4) relatively weak inhibition by P 1 ,P 5 -di-(adenosine-5′) pentaphosphate of the liver enzyme compared to the muscle enzyme (S. A. Kuby et al . (1978) Arch. Biochem. Biophys . 187 , 34–52). A thermodynamic value for K eq = (MgADP − ) (ADP 3− ) ( MgATP 2− ) (AMP 2− ) = 2.7 (±0.6) × 10 −1 has been arrived at, which compared favorably with the average value from its kinetic estimation via four Haldane relations for the three adenylate kinase sets of data.

Glucose 6-phosphate dehydrogenase from brewers' yeast (Zwischenferment): Further observations on the ligand-lnduced macromolecular association phenomenon; kinetic properties of the two-chain protein species; and studies on the enzyme-substrate interactions☆

Abstract In the presence of EDTA3−, the two-subunit species of apoenzyme (Mr 102,000 g/mole) ( Yue, Noltmann, and Kuby (1967) Biochemistry6, 1174) maintains its integrity without further dissociation into component subunits over a broad range of pH values to 9.8 and in very dilute solutions, as demonstrated by sedimentation equilibrium studies. In the absence of EDTA3−, studies on the ligand-induced macromolecular association to the four-subunit species ( Yue, Noltmann, and Kuby (1969) J. Biol. Chem.244, 1353–1364) have been further extended for NADP+ to include ultracentrifugal titrations at pH 7.8 and 3 °C. A significant concentration of EDTA3−, as before, will inhibit these macromolecular association reactions and stabilize the apoenzyme species in solution. NADPH is similar to NADP+ in that it will induce the formation of the four-subunit species; but by way of contrast, glucose 6-phosphate2−will not. In fact, the addition of glucose 6-phosphate2− to NADPH solutions may actually inhibit or reverse the association reaction induced by NADPH, and in this respect, glucose 6-phosphate2− acts in a fashion not unlike the EDTA− ion. Thus, the ratio of ( NADP ) 0 ( Glc -6-P) 0 as well as their absolute concentrations will affect the extent of dimerization of the enzyme in the absence of any other modifiers. The use of EDTA3− has now permitted direct steady-state kinetic measurements on the two-chain species alone, which has proved to be active even in the presence of high concentrations of EDTA3−. The steady-state kinetic mechanism of the “monomeric” enzyme species appears to have reduced to a relatively simple case, viz., to that of a random quasi-equilibrium type with a rate-limiting step at the interconversion of the ternary complexes and with independent binding of the substrates; values for the kinetic parameters have been derived for this mechanism. The product inhibition pattern of NADPH (competitive with respect to both substrates) would point, in addition, to the absence of any kinetically significant concentrations of dead-end complexes with the two-chain species. To confirm the assigned kinetic mechanism for the two-chain enzyme, measurements of the equilibrium binding of the substrates were undertaken by several techniques. By gel filtration through Sephadex G-25, in the presence of EDTA3−, two equilavent binding sites for NADP+ per mole of apoenzyme (or one per subunit)- is observed, with a value for the intrinsic dissociation constant approximating the kinetically determined value. On the other hand, in the absence of EDTA3−, the nonlinear Scatchard plot obtained is likely a reflection of the superimposed association-dissociation reactions involving the protein. Similar quantitative studies for the ligand NADP+, in the presence and absence of EDTA3−, were conducted by equilibrium dialysis, by a uv difference-spectral technique, and by measurement of the protection afforded by either substrate against inactivation by 5,5′-dithiobis(2-nitrobenzoic acid), which apparently reacts with only a single exposed sulfhydryl group per subunit. Coincident with these studies, a total amino acid composition for this protein is reported. In the case of glucose 6-phosphate, by the technique of equilibrium dialysis, there appears to be two major equivalent binding sites (or one per subunit) with an intrinsic dissociation constant approximating that measured kinetically, as well as two additional but very weak sites (or one additional site per subunit). Moreover, EDTA3− exerts no effect on the equilibrium binding behavior of Glc-6-P, in agreement with the observation that Glc6-P does not induce the formation of the tetra-chain species. Glucose 6-phosphate will also protect the enzyme against inactivation by 5,5′dithiobis(2-nitrobenzoic acid), pointing to a close proximity of both binding sites for Glc-6-P and NADP+ to the single exposed sulfhydryl group per subunit. Finally, fluorometric observations have qualitatively confirmed the existence of both NADPH- and NADP+- enzyme binary compounds.

Studies on muscular dystrophy: a comparison by physical and chemical means of the normal human ATP-creatine transphosphorylases (creatine kinases) with those from tissues of Duchenne muscular dystrophy.

Abstract The four human Duchenne dystrophic isoenzymes (M-M, M-B, B-B, from the muscle and B-B from the brain) of ATP-creatine transphosphorylase (S. A. Kuby, H. J. Keutel, K. Okabe, H. K. Jacobs, F. Ziter, D. Gerber, and F. H. Tyler, 1977, J. Biol. Chem.252, 8382–8390) have now been compared physically and chemically with their normal human counterparts (viz., with the three isoenzymes, M-M, M-B, B-B, 2). All isoenzymes proved to be composed of two noncovalently linked polypeptide chains, by sedimentation equilibrium analyses in the presence and absence of disruptive agents. In the presence of 2-mercaptoethanol at 0.16(Γ/2), pH 7.8, the two native muscle types yielded identical values for s20,w, concentration dependencies, and molecular weight, and similarly for the brain types (from the brain). But the human brain type proved to be slightly heavier than the muscle type (viz. 88,400 vs 85,900). All of the isoenzymes showed similar electrophoretic behavior between their several counterparts between pH 5–8, except perhaps between pH 8–10, where small differences appeared. The three native normal human isoenzymes, as well as the dystrophic human isoenzymes (M-M from the muscle and B-B from the brain) all contain 2 reactive sulfhydryl groups per mole or 1 per polypeptide chain of these two-chain proteins, which may be titrated with 5,5′-dithiobis(2-nitrobenzoic acid) (Nbs2); and under acidic conditions, quantitative titrations with 4,4′-dithiodipyridine yield a total of 10 -SH groups per mole of each brain type and 8 -SH groups per mole of muscle type, in the case of man, dystrophic man, calf, and rabbit. The kinetics of reactions between Nbs2 and the sulfhydryl groups of all three normal human isoenzymes and two dystrophic human isoenzymes have been measured under several sets of denaturing conditions. A comparison of their reactive calculated second-order velocity constants reveal significant differences between these three normal human isoenzymes, but the ksecond order values for the reactions of the sulfhydryl groups of the dystrophic M-M and B-B with Nbs2, when compared with their normal counterparts, gave identical values in the presence of 7.3 m urea or 1.8% laurylsulfate, from which it may be inferred that very similar, if not identical, environments surround these two sets of sulfhydryl groups. A comparison of the amino acid compositions of the normal human muscle type and brain type with the human dystrophic M-M and B-B (from the brain) reveal essentially identical values for the muscle types but nearly identical values for the brain types, with a few differences. Their respective tryptic peptide maps have been compared of the S-carboxy-methylated proteins (alkylated with iodo[2-14C]acetic acid at the two exposed -SH groups per mole). Thus, the muscle types, normal and dystrophic, yield identical maps, but the brain types nearly identical maps, with a few significant differences. Isolation of the tryptic tridecapeptide from the S-carboxymethylated normal human and dystrophic human dimeric muscle-type ATP-creatine transphosphorylases, labeled at the single exposed SH group per polypeptide chain with iodo[2-14C]acetate, yielded the following sequence for both proteins: ValLeuThrCys(CH2COOH)ProSerAsnLeuGlyThr GlyLeuArg [where Cys(CH2COOH) is S-carboxymethyl cysteine]. This sequence showed remarkable homology with a few other equivalent peptides reported to be derived from the exposed SH group of other ATP-creatine transphosphorylases. In conclusion, there does not appear to be a mutation in the structural genes for the muscle-type creatine kinases detectable by the analyses presented here. However, the brain types warrant further investigation.

Studies on adenosine triphosphate transphosphorylases. XVII. A physicochemical comparison of the ATP-creatine transphosphorylase (creatine kinase) isozymes from man, calf, and rabbit

All of the creatine kinase isozymes from human, calf, and rabbit brain and muscle are composed of two noncovalently linked polypeptide chains, based upon sedimentation equilibrium analyses in the presence and absence of disruptive agents. The brain-type isozymes of man, calf, and rabbit proved to be slightly heavier than the muscle types. Various physicochemical properties of the isozymes are recorded. Each group of isozymes, i.e., the muscle, hybrid (muscle-brain), and brain isozymes from man, calf, and rabbit, showed similar electrophoretic behavior, although isoelectric points were not precisely identical for the muscle and hybrid types. Theoretical titration curves constructed from amino acid compositions of the calf isozymes showed reasonable agreement between their calculated and measuredpI0 values (isoelectric point extrapolated to zero ionic strength). The three native muscle isozymes and brain isozymes all contain two reactive sulfhydryl groups per mole or one per polypeptide chain of their two-chain proteins, which may be titrated with 5,5′-dithiobis (2-nitrobenzoic acid); and under acidic conditions, quantitative titrations with 4,4′-dithiodipyridine yield a total of ten- SH groups per mole of each brain-type and eight- SH groups per mole of muscle-type isozyme in the case of man, calf, and rabbit. A comparison of their amino acid compositions and tryptic peptide maps shows that there is only a slightly greater degree of homology between the individual isozymes of the same type (muscle type or brain type) than between the muscle- and brain-type isozymes of the same species.

A Steady-State Kinetic Analysis of the Prolyl-4-Hydroxylase Mechanism

Published kinetic data by Kivirikko, et al. on the prolyl-4-hydroxylase reaction have been re-evaluated using the overall steady-state velocity equation in the forward and reverse directions for an ordered ter ter kinetic mechanism. Qualitatively, the published data for prolyl-4-hydroxylase appear to fit the predicted patterns for this kinetic mechanism. More kinetic data are needed to confirm these results and to quantitate the kinetic parameters but, tentatively, the order of substrate addition would appear to be alpha-ketoglutarate, oxygen, and peptide; and the order of product release would be hydroxylated peptide (or collagen), carbon dioxide, and succinate.

Studies on NADPH-cytochrome c reductase I: Isolation and several properties of the crystalline enzyme from ale yeast.

Abstract NADPH-cytochrome c reductase has been isolated from a top-fermenting ale yeast, Saccharomyces cerevisiae (Narragansett strain), after ca. a 240-fold purification over the initial extract of an acetone powder, with a final specific activity (at pH 7.6, 30 °C) of ca. 150 μmol cytochrome c reduced min−1mg−1 protein. The preparation appears to be homogeneous by the criteria of: sedimentation velocity; electrophoresis on cellulose acetate in buffers above neutrality; and by polyacrylamide gel electrophoresis. Although the reductase appeared to partially separate into species “A” and “B” on DEAE-cellulose at pH 8.8, the two species have proven to be indistinguishable electrophoretically (above pH 8) and by sedimentation. By sedimentation equilibrium at 20 °C, a molecular weight of ca. 6.8 (± 0.4) × 104 was obtained with use of a V 20 ° = 0.741 calculated from its amino acid composition. After disruption in 4 m guanidinium chloride- 10 m m dithioerythritol- 1 m m EDTA, pH 6.4 at 20 °C, an M r of 3.4 (± 0.1) × 104 resulted, which points to a subunit structure of two polypeptide chains per mole. Confirmatory evidence of the two-subunit structure with similar, if not identical, polypeptide chains was obtained by polyacrylamide gel electrophoresis in dodecyl-sulfate, after disruption in 4 m urea and 2% sodium dodecyl sulfate, and yielded a subunit molecular weight of ca. 4 × 104. Sulfhydryl group titration with 4,4′-dithiodipyridine under acidic conditions revealed one sulfhydryl group per monomer, which apparently is necessary for the catalytic reduction of cytochrome c. NADPH, as well as FAD, protects this-SH group from reaction with 5,5′-dithiobis (2-nitrobenzoate). The visible absorption spectrum of the oxidized enzyme (as prepared) has absorption maxima at 383 and 455 nm, typical of a flavoprotein. Flavin analysis (after dissociation by thermal denaturation of the “A” protein) conducted fluorometrically, revealed the presence of 2.0 mol of FAD per 70,000 g, in confirmation of the deduced subunit structure. The identity of the FAD dissociated from either “A” or “B” protein was confirmed by recombination with apo- d -amino acid oxidase and by thin-layer chromatography. A kinetic approach was used to estimate the dissociation constant for either FAD or FMN (which also yields a catalytically active enzyme) to the apoprotein reductase at 30 °C and pH 7.6 (0.05 m phosphate) and yielded values of 4.7 × 10−8 m for FAD and 4.4 × 10−8 m for FMN.

Studies on NADH(NADPH)-cytochrome c reductase (FMN-containing) from yeast: Steady-state kinetic properties of the flavoenzyme from top-fermenting ale yeast☆☆☆

A study of the steady-state kinetics of NADH(NADPH)-cytochrome c reductase (FMN-containing) from ale yeast (M. S. Johnson and S. A. Kuby (1985) J. Biol. Chem. 260, 12341-12350) has led to a postulated three-substrate random-ordered hybrid mechanism, where NAD(P)H and FMN add randomly and very likely in a steady-state fashion, followed by an ordered addition of cytochrome c. Kinetic parameters have been derived from this mechanism. Arrhenius plots showed large differences between NADH and NADPH, as the substrate-reductant. Menadione accelerated cytochrome c reduction and also O2 uptake, but vitamin K1 and coenzyme Q10 were ineffective as electron mediators, possibly as a result of their insolubility. With NADPH as the substrate-reductant, the order of the rate of reduction of electron acceptors was ferricyanide greater than DCIP greater than cytochrome c greater than oxygen; with menadione, the specificity sequence was cytochrome c greater than ferricyanide greater than DCIP greater than oxygen. With NADH, the order was ferricyanide greater than cytochrome c greater than oxygen greater than DCIP, which changed to cytochrome c greater than ferricyanide greater than oxygen greater than DCIP on addition of menadione. Cytochrome b5 was also reduced in the absence of oxygen. No transhydrogenase activity was observed, but the reduced thionicotinamide analogs of NADH and NADPH acted as substrates. Superoxide dismutase inhibited cytochrome c reduction in air by 50%, but O2-. was not necessary for cytochrome c reduction, as evidenced by the increase in rate in the absence of O2. The product of the reaction with oxygen appeared to be H2O2.