ADP-ribosylation of phosphorylase kinase and block of phosphate incorporation into the enzyme

Phosphorylase kinase purified from rabbit skeletal muscle was ADP-ribosylated by hen liver nuclear ADP-ribosyltransferase. This modification, as was seen in cAMP-dependent phosphorylation, was observed only in alpha and beta subunits of the phosphorylase kinase and the latter was more rapidly modified. Analysis of the ADP-ribosylated amino acid residue sequenced in alpha and beta subunits showed that both subunits were modified at the area of the arginine residue. The Km for NAD was 0.10 mM and the pH optimum was 9.0. When the ADP-ribosylated phosphorylase kinase was phosphorylated by cAMP-dependent protein kinase, a reduction in phosphate incorporation occurred with increase in the ADP-ribosylation. ADP-ribosylation also suppressed autophosphorylation, to a lesser degree than observed with cAMP-dependent phosphorylation. The ADP-ribosylation-dependent reduction of phosphorylation resulted in a suppression of the phosphorylation-dependent activation of the phosphorylase kinase. These results together with findings of ADP-ribosyltransferase activity in the rabbit skeletal muscle [Soman, G. et al. (1984) Biochem. Biophys. Res. Commun. 120, 973-980] suggest that ADP-ribosylation participates in the regulation of the phosphorylase kinase activity through changes in the rate of phosphorylation.     

Metabolism of bradykinin analogs by angiotensin I converting enzyme and carboxypeptidase N.

Bradykinin (BK) analogs such as Lys-Lys-BK, des-Arg9-BK and [Leu8]des-Arg9-BK were poor substrates for angiotensin I converting enzyme (ACE), and analogs containing D-Phe7 residues, or a pseudopeptide C-terminal bond, were completely resistant. However, many of these analogs were metabolized by carboxypeptidase N (CPN) including Lys-Lys-BK, [Tyr8(OMe)]BK and D-Phe7-containing analogs, with Km and Vmax values comparable to those for BK. The only analogs completely resistant to both ACE and CPN were the B2 agonist [Phe8 psi(CH2NH)Arg9]BK, the B2 agonist D-Arg[Hyp3,D-Phe7,Phe8 psi(CH2NH)Arg9]BK, and the B1 agonist [D-Phe8]des-Arg9-BK. These data indicate an important role for plasma CPN and vascular CPN-like activity in the metabolism of the widely used ACE-resistant/D-Phe7-containing antagonists of B2 kinin receptors.     

Conformation-activity relations of des-Arg9-bradykinin. I. Conformation of molecules in solution

The spatial structure of [des-Arg9]bradykinin, a selective agonist of the B1 type kinin receptors, was studied by means of theoretical conformational analysis. In order to restrict the number of conformations under consideration, we used available data on 1H NMR in (CD3)2SO solution indicating the C-terminal carboxyl group to interact with the guanidine group of Arg1 and the hydroxyl group of Ser6. The calculated set of [des-Arg9]bradykinin low-energy conformations was compared with the 1H NMR data. Four types of conformations were selected, which are consistent with experimental data and serve as sterically reliable models for the preferable three-dimensional structure of [des-Arg9]bradykinin in solution.     

Purification and characterization of bioactive peptides from skin extracts of Rana esculenta

The peptide fraction extracted by methanol from the skin of Rana esculenta, a species widely distributed in Western Europe, was investigated. The pharmacological activity found in the extract is attributable to the presence of authentic bradykinin, together with a shorter, partially active version of this molecule, des-Arg9-bradykinin. Also the bradykinin fragment 1-7 has been isolated, but it was inactive in our bioassay system. Moreover, a family of hydrophobic peptides has been purified and characterized, which appeared devoid of pharmacological activities when tested on smooth muscle preparations, but were provided with hemolytic activities.     

Uncoupling Crk signal transduction by Pseudomonas exoenzyme T

Exoenzyme T (ExoT) is a bifunctional type III cytotoxin of Pseudomonas aeruginosa that possesses both Rho GTPase-activating protein and ADP-ribosyltransferase activities. The ADP-ribosyltransferase activity of ExoT stimulated depolymerization of the actin cytoskeleton independent of Rho GTPase-activating protein function, and ExoT was subsequently shown to ADP-ribosylate Crk (CT10 regulator of kinase)-I and Crk-II. Crk proteins are eukaryotic adaptor proteins comprising SH2 and SH3 domains that are components of the integrin signaling pathway leading to Rac1 and Rap1 functions. Mass spectroscopic analysis identified Arg20 as the site of ADP-ribosylation by ExoT. Arg20 is a conserved residue located within the SH2 domain that is required for interactions with upstream signaling molecules such as paxillin and p130cas. Glutathione S-transferase pull-down and far Western assays showed that ADP-ribosylated Crk-I or Crk-I(R20K) failed to bind p130cas or paxillin. This indicates that ADP-ribosylation inhibited the direct interaction of Crk with these focal adhesion proteins. Overexpression of wild-type Crk-I reduced cell rounding by ExoT, whereas expression of dominant-active Rac1 interfered with the ability of ExoT to round cells. Thus, the ADP-ribosylation of Crk uncouples integrin signaling by direct inhibition of the binding of Crk to focal adhesion proteins.     

Purification, structural characterization, and myotropic activity of a peptide related to des-Arg(9)-bradykinin from an elasmobranch fish, the little skate, Leucoraja erinacea

A bradykinin (BK)-related peptide was isolated from heat-denaturated plasma from an elasmobranch fish, the little skate, Leucoraja erinacea after incubation with porcine pancreatic kallikrein. The primary structure of the peptide (H-Gly-Ile-Thr-Ser-Trp-Leu-Pro-Phe-OH; skate BK) shows limited structural similarity to the mammalian B1 receptor agonist, des-Arg(9)-BK. The myotropic activities of synthetic skate BK, and the analog skate [Arg(9)]BK, were examined in isolated skate vascular and intestinal smooth muscle preparations. Skate BK produced a concentration-dependent constriction of the mesenteric artery (EC(50)=4.37x10(-8)M; maximum response=103.4+/-10.23% of the response to 60mM KCl) but the response to skate [Arg(9)]BK was appreciably weaker (response to 10(-6)M=73.0+/-23.4% of the response to 60mM KCl). Neither the first branchial gill arch nor the ventral aorta responded to either purified peptide. Skate BK also produced a concentration-dependent constriction of intestinal smooth muscle preparations (EC(50)=2.74x10(-7)M; maximum response 31.0+/-12.2% of the response to 10(-5)M acetylcholine). Skate [Arg(9)]BK was without effect on the intestinal preparation. The data provide evidence for the existence of the kallikrein-kinin system in a phylogenetically ancient vertebrate group and the greater potency of skate BK compared with the analog skate [Arg(9)]BK suggests that the receptor mediating vascular responses resembles the mammalian B1 receptor more closely than the B2 receptor.     

ADP-ribosylation of Ca2+-dependent ATPase in vitro suppresses the enzyme activity

We investigated the effect on the Ca2+-dependent ATPase activity of ADP-ribosylation of the enzyme from the rabbit skeletal muscle sarcoplasmic reticulum. A reconstituted ADP-ribosylation system of Ca2+-dependent ATPase in which the enzyme and ADP-ribosyltransferase, both were partially purified from the vesicles, and poly L-lysine were contained, was preincubated with 1 mM NAD, and the Ca2+-dependent ATPase activity was assayed. The NAD-dependent suppression of the enzyme activity depended on both the concentration of NAD and preincubation-time for the ADP-ribosylation, and was reversed by adding 20 mM arginine during the preincubation. These results taken together with the findings that Ca2+-dependent ATPase is a major acceptor protein for the modification in rabbit skeletal muscle sarcoplasmic reticulum [Hara et al. (1987) Biochem. Biophys. Res. Commun. 144; 856-862] suggest that Ca2+-transport in the sarcoplasmic reticulum may be regulated through changes in the rate of ADP-ribosylation of Ca2+-dependent ATPase.     

Botulinum ADP-ribosyltransferase C3 but not botulinum neurotoxins C1 and D ADP-ribosylates low molecular mass GTP-binding proteins

Botulinum ADP-ribosyltransferase C3 modified 21-24 kDa proteins in a guanine nucleotide-dependent manner similar to that described for botulinum neurotoxin C1 and D. Whereas GTP and GTP gamma S stimulated C3-catalyzed ADP-ribosylation in the absence of Mg2+, in the presence of added Mg2+ ADP-ribosylation was impaired by GTP gamma S. C3 was about 1000-fold more potent than botulinum C1 neurotoxin in ADP-ribosylation of the 21-24 kDa protein(s) in human platelet membranes. Antibodies raised against C3 blocked ADP-ribosylation of the 21-24 kDa protein by C3 and neurotoxin C1 but neither cross reacted with neurotoxin C1 immunoblots nor neutralized the toxicity of neurotoxin C1 in mice. The data indicate that the ADP-ribosylation of low molecular mass GTP-binding proteins in various eukaryotic cells is not caused by botulinum neurotoxins but is due to the action of botulinum ADP-ribosyltransferase C3. The weak enzymatic activities described for botulinum neurotoxins appear to be due to the contamination of C1 and D preparations with ADP-ribosyltransferase C3.     

Monoclonal antibodies to bradykinin inhibit smooth muscle contractile action of bradykinin

Four hybridoma cell lines have been established that secrete monoclonal antibodies to nonapeptide bradykinin. Bradykinin coupled to ovalbumin, using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide as coupling agent, was used to immunize BALB/c mice. Spleen cells from the immunized animals were fused to P3-X63-AG8-653 mouse myeloma cells. The resultant hybrid cells were screened by enzyme-linked immunoassay for production of antibodies to bradykinin. Hybrids from four positive wells were subcloned by limiting dilution and expanded as ascites tumor into pristane-primed mice. All the four hybrids secreted monoclonal antibodies of IgG1 (k) isotype. Unlabeled peptides bradykinin, lysyl-bradykinin (kallidin) and methionyl-lysyl-bradykinin competed with the radiolabeled [Tyr1]kallidin for monoclonal antibody binding sites. These antibodies recognized preferentially either NH2- or COOH-terminals of the nonapeptide bradykinin and can distinguish between des-Arg1-bradykinin and des-Arg9-bradykinin. Bradykinin fragments smaller than eight residues were not recognized by these antibodies. Monoclonal antibodies BK-D6A5, BK-B6C9 and BK-A3D9 neutralized the smooth muscle contractile activity of bradykinin. An enzyme-linked immunoassay developed using these monoclonal antibodies showed the effective range of bradykinin determination between 5 and 150 ng.     

Pseudomonas aeruginosa ExoS ADP-ribosyltransferase inhibits ERM phosphorylation

Pseudomonas aeruginosa causes life-threatening infections in compromised and cystic fibrosis patients. Pathogenesis stems from a number of virulence factors, including four type III translocated cytotoxins: ExoS, ExoT, ExoY and ExoU. ExoS is a bifunctional toxin: the N terminus (amino acids 96-219) encodes a Rho GTPase Activating Protein (GAP) domain. The C terminus (amino acids 234-453) encodes a 14-3-3-dependent ADP-ribosyltransferase domain which transfers ADP-ribose from NAD onto substrates such as the Ras GTPases and vimentin. Ezrin/radixin/moesin (ERM) proteins have recently been identified as high-affinity substrates for ADP-ribosylation by ExoS. Expression of ExoS in HeLa cells led to a loss of phosphorylation of ERM proteins that was dependent upon the expression of ADP-ribosyltransferase activity. MALDI-MS and site-directed mutagenesis studies determined that ExoS ADP-ribosylated moesin at three C-terminal arginines (Arg553, Arg560 and Arg563), which cluster Thr558, the site of phosphorylation by protein kinase C and Rho kinase. ADP-ribosylated-moesin was a poor target for phosphorylation by protein kinase C and Rho kinase, which showed that ADP-ribosylation directly inhibited ERM phosphorylation. Expression of dominant active-moesin inhibited cell rounding elicited by ExoS, indicating that moesin is a physiological target in cultured cells. This is the first demonstration that a bacterial toxin inhibits the phosphorylation of a mammalian protein through ADP-ribosylation. These data explain how the expression of the ADP-ribosylation of ExoS modifies the actin cytoskeleton and indicate that ExoS possesses redundant enzymatic activities to depolymerize the actin cytoskeleton.     

Suppression of GTP/T alpha-dependent activation of cGMP phosphodiesterase by ADP-ribosylation by its gamma subunit in amphibian rod photoreceptor membranes.

Our previous study has shown that P gamma, the regulatory subunit of cGMP phosphodiesterase (PDE), is ADP-ribosylated by endogenous ADP-ribosyltransferase when P gamma is free or complexed with the catalytic subunits of PDE in amphibian rod photoreceptor membranes. The P gamma domain containing ADP-ribosylated arginines was shown to be involved in its interaction with T alpha, a key interaction for PDE activation. In this study, we describe a possible function of the P gamma ADP-ribosylation in the GTP/T alpha-dependent PDE activation. When rod membranes were preincubated with or without NAD and washed with a buffer containing GTP, the PDE activity of NAD-preincubated membranes was increased by the GTP-washing only to approximately 50% of that of membranes preincubated without NAD. The P gamma release by the GTP-washing from these NAD-preincubated membranes was also suppressed to approximately 50% of that preincubated without NAD. Taking into consideration that approximately 50% of P gamma is ADP-ribosylated under these conditions, these observations suggest that the ADP-ribosylated P gamma cannot interact with GTP/T alpha. We have also shown that a soluble fraction of ROS contains an enzyme(s) to release the radioactivity of [32P]ADP-ribosylated P gamma in concentration- and time-dependent manners, suggesting that the P gamma ADP-ribosylation is reversible. Rod ADP-ribosyltransferase solubilized from membranes by phosphatidylinositol-specific phospholipase C was separated into two fractions by ion-exchange columns. Biochemical characterization of these two fractions, including measurement of the Km for NAD and P gamma, estimation of their molecular masses, ADP-ribosylation of P gamma arginine mutants, effects of ADP-ribosyltransferase inhibitors on the P gamma ADP-ribosylation, and effects of salts and pH on the P gamma ADP-ribosylation, indicates that rod ADP-ribosyltransferase contains two isozymes, and that these two isozymes have similar properties for the P gamma ADP-ribosylation. Our observations strongly suggest that the negative regulation of PDE through the reversible P gamma ADP-ribosylation may function in the phototransduction mechanism.     

Kinin metabolism in human nasal secretions during experimentally induced allergic rhinitis

We have previously shown that both bradykinin and lysylbradykinin are generated in nasal secretions upon nasal challenge of allergic individuals with appropriate allergen and have suggested that these potent pro-inflammatory peptides may contribute to the pathogenesis of the allergic response. In this study we used a variety of synthetic substrates together with both thin layer and high performance liquid chromatography systems to examine the metabolism of these peptides in nasal secretions obtained by lavage. We now demonstrate that in addition to low levels of angiotensin-converting enzyme, nasal lavages contain an aminopeptidase activity that converts lysylbradykinin to bradykinin, and a carboxypeptidase that removes the C-terminal arginine from bradykinin and lysylbradykinin. The levels of all these activities are significantly increased after allergen challenge of allergic, but not nonallergic, individuals. The aminopeptidase and carboxypeptidase activities present in post-challenge lavages from allergic individuals convert lysylbradykinin to intermediate products (bradykinin and des (Arg10) lysylbradykinin) and eventually to des (Arg9) bradykinin. The nasal carboxypeptidase was activated 475% by 0.1 mM CoCl2 and was inhibited by the carboxypeptidase N inhibitor, MERGETPA (D-L-mercaptomethyl-3-guanidino-ethylthiopropanoic acid) (IC50 = 10 microM). The aminopeptidase activity was not affected by MERGETPA but was potently inhibited by amastatin and bestatin (IC50 = 0.05 microM and 3.0 microM, respectively). The activity of the aminopeptidase against its synthetic substrate was also inhibited by lysylbradykinin (IC50 = 50 microM). Both the carboxypeptidase and aminopeptidase activities had neutral pH optima and were inhibited by o-phenanthroline, but were unaffected by inhibitors of neutral endopeptidases (phosphoramidon) or angiotensin-converting enzyme (Captopril). The Km of bradykinin for the nasal carboxypeptidase was 139 +/- 14 microM (n = 3). We conclude that during the allergic response, nasal secretions contain aminopeptidase and carboxypeptidase activities that convert lysylbradykinin and bradykinin (B2 agonists) to des (Arg9) bradykinin (a B1 agonist). Because the nature of the kinin receptors in the nasal mucosa are currently unknown, it remains to be determined whether this metabolism results in the termination of biologic activity or the production of a biologically active moiety.     

Cytoplasmic inhibitor of eEF-2 ADP-ribosylation catalyzed by diphtheria toxin or endogenous transferase in rat liver cells

eEF-2 (100 kDa) isolated from rat liver cells undergo ADP-ribosylation in the presence of diphtheria toxin or endogenous ADP-ribosyltransferase, which was co-purified with the factor. We separated the fraction free of elongation factor and endogenous transferase, which strongly inhibited the ADP-ribosylation of eEF-2. This fraction did not affect the activity of the elongation factor. The lack of endogenous transferase activity (which is potentially lethal for the cell) in the postribosomal supernatant could be the result of its inhibition. eEF-2 (65 kDa) which is probably responsible for the process of translocation (Gajko, A. et al. (1999) Biochem. Biophys. Res. Commun. 255, 535-538) was protected from ADP-ribosylation and its irreversible inactivation in the presence of the rat liver extract fraction.     

ADP-ribosyltransferase is highly conserved: purification and characterization of ADP-ribosyltransferase from a fish and its comparison with the human enzyme

Summary Covalent modification of proteins by ADP-ribosylation is a major mode of protein regulation in eukaryotic cells. ADP-ribosyltransferases have been characterized from mammals but little is known about these enzymes in lower vertebrates. We purified an ADP-ribosyltransferase (E.C. 2.4.2.30) from trout (Salmo trutta faris) by affinity chromatography and characterized it. The 11700-fold purified activity shows a major protein band at a molecular mass of 75000 kDa in a SDS-polyacrylamide gel.In situ reactivation of SDS gels showed the 75000 kDa protein to be enzymatically active, and additional enzymatically active bands at molecular masses of 115000, 90000 and 87000 kDa, respectively. The enzyme is capable of poly-ADP-ribosylation. It crossreacts with affinity purified antibodies raised against human poly(ADP-ribose)synthetase and, except for the temperature optimum, its properties strongly resemble the mammalian enzymes, indicating the conserved character of nuclear ADP-ribosyltransferases. The trout enzyme is DNA- and histone-dependent, has an optimal pH between 8 and 9 and an apparentK _m for NAD⁺ of 24 M. The temperature optimum is 10°C compared with 25°C for the human enzyme. Known ADP-ribosyltransferase inhibitors also inhibit the enzyme from trout. ribosyl-)residues from NAD⁺. It plays a central role in processes affecting DNA function such as DNA repair, recombination, differentiation, tumorigenic cell transformation and cell proliferation (Pekala and Moss 1983; Ueda and Hayaishi 1985; Creissen and Shall 1982; Borek et al. 1984; Lunec 1984; Cleaver et al. 1985; Tseng et al. 1987). It seems to be a central controller of cell physiology (Loetscher et al. 1987; Schweiger et al. 1987). ADP-ribosylation is high in proliferating cells whereas it is low in differentiated cells (Surowy and Berger 1983). Due to the important functions of ADP-ribosyltransferase this enzyme has been studied extensively but almost exclusively in mammals (Ueda and Hayaishi 1985). Only little is known about ADP-ribosyltransferases of lower vertebrates and of species of lower phylogenetic families. Hence it appeared of interest to us to study this enzyme in a fish. Here we report the purification to homogeneity and the characterization of ADP-ribosyltransferase from trout liver. We compared the fish enzyme with the human one and found very similar enzymatic properties and immunological crossreactivity. This indicates that ADP-ribosyltransferases are highly conserved.     

Eukaryotic mono(ADP-ribosyl)transferase that ADP-ribosylates GTP-binding regulatory Gi protein

An NAD:cysteine ADP-ribosyltransferase designated ADP-ribosyltransferase C was purified approximately 35,000-fold from human erythrocytes with an 11% yield. The purified ADP-ribosyltransferase C exhibited one predominant protein band on sodium dodecyl sulfate-polyacrylamide gels with an estimated molecular weight (Mr) of 28,500. The Km values for NAD and cysteine methyl ester were determined to be 65 and 4,400 microM, respectively. By using human erythrocyte inside-out membrane vesicles, the transferase C was found to ADP-ribosylate the alpha subunit (Mr = 41,000) of Gi, which is a substrate for pertussis toxin. The ADP-ribosylation of Gi alpha catalyzed by ADP-ribosyltransferase C was inhibited by pre-ADP-ribosylation with pertussis toxin. The linkage of ADP-ribose-Gi alpha in the membranes formed by ADP-ribosyltransferase C was as stable to hydroxylamine as that formed by pertussis toxin. These data represent the first demonstration that eukaryotic cells contain an ADP-ribosyltransferase which can catalyze the ADP-ribosylation of a cysteine residue in Gi alpha.     

Low-molecular-weight GTP-binding proteins serving as ADP-ribosylation substrate for ADP-ribosyltransferase from Clostridium botulinum and their relation to phosphoinositides metabolism in thymocytes

ADP-ribosyltransferase from Clostridium botulinum type C strain was found to induce an increase of inositol phosphates (IPs) formation in murine thymocytes membranes. Incubation of electropermeabilized murine thymocytes with the enzyme also caused an increase of IPs formation in the cells. This increase of IPs formation in the enzyme-treated membranes and electropermeabilized cells was dependent on the amount of both NAD and the enzyme, suggesting that the stimulation of phosphoinositide-specific phospholipase C (PLC) was related to ADP-ribosylation of membrane proteins by the enzyme. On the other hand, in calf and murine thymocytes two proteins with the same molecular weight of 21,000 were found to be ADP-ribosylated by the botulinum ADP-ribosyltransferase. A minor ADP-ribosylation substrate was shown by two-dimensional polyacrylamide gel electrophoresis to be G21k, a low-molecular-weight GTP-binding protein (G protein) suggested previously by us to be involved in PLC regulation [Wang, P. et al. (1987) J. Biochem. 102, 1275-1287; (1988) 103, 137-142; and (1989) 105, 461-466], and the other major ADP-ribosylation substrate was identified as a rho A protein. Under the experimental conditions of the IPs formation study, ADP-ribosylation of both G21k and rho A proteins by botulinum ADP-ribosyltransferase in membranes and permeabilized cells was observed. These results suggest that botulinum ADP-ribosyltransferase-induced PLC stimulation in thymocytes is closely correlated with ADP-ribosylation of the low-molecular-weight G proteins.     

Bradykinin receptors in isolated rat duodenum

Pharmacological properties of the bradykinin receptors in the isolated rat duodenum were investigated by examining the relaxant and contractile responses to bradykinin and [des-Arg9]-bradykinin, an agonist of B1 receptors. A specific desensitization and de novo formation for B1 receptors were observed. Changes in medium pH caused a decrease in the responses to bradykinin and [des-Arg9]-bradykinin of rat duodenum. Urea incubation in test tube inhibited the responses to bradykinin and [des-Arg9]-bradykinin of rat duodenum, while urea in bathing medium was ineffective. These findings strongly suggested that (a) ionic bonds are important in the interaction between bradykinin and its receptors, and (b) B2 receptors in rat duodenum are different from those in guinea pig ileum.     

Vanadate, epidermal growth factor and the stimulation of DNA synthesis

We present here what we believe to be the first report of the stimulation of NAD⁺-dependent ADP-ribosyltransferase activity by a hormone. Isoproterenol stimulated the ADP-ribosylation of RL-PR-C hepatocyte membranes in a concentration-dependent fashion; the effect was abolished by the β-adrenergic antagonist, propranolol. Although hepatocyte plasma membrane ADP-ribosyltransferase and adenylate cyclase activities differed in their sensitivity to isoproterenol, the kinetics of both effects were quite similar. PAGE separation of membrane proteins after ADP-ribosylation from [2,8-³H-adenine]NAD⁺ identified the acceptor for isoproterenol-enhanced ADP-ribosylation as the same 55,000 dalton guanyl nucleotide regulatory protein serving for both endogenous and cholera toxin-stimulated processes in these cells.     

3 alpha-hydroxysteroid dehydrogenase activity catalyzed by purified pig adrenal 20 alpha-hydroxysteroid dehydrogenase.

In earlier studies, two distinct molecules, 20 alpha-HSD-I and 20 alpha-HSD-II, responsible for 20 alpha-HSD activity of pig adrenal cytosol were purified to homogeneity and characterized [S. Nakajin et al., J. Steroid Biochem. 33 (1989) 1181-1189]. We report here that the purified 20 alpha-HSD-I, which mainly catalyzes the reduction of 17 alpha-hydroxyprogesterone to 17 alpha,20 alpha-dihydroxy-4-pregnen-3-one, catalyzes 3 alpha-hydroxysteroid oxidoreductase activity for 5 alpha (or 5 beta)-androstanes (C19), 5 alpha (or 5 beta)-pregnanes (C21) in the presence of NADPH as the preferred cofactor. The purified enzyme has a preference for the 5 alpha (or 5 beta)-androstane substrates rather than 5 alpha (or 5 beta)-pregnane substrates, and the 5 beta-isomers rather than 5 alpha-isomers, respectively. Kinetic constants in the reduction for 5 alpha-androstanedione (Km; 3.3 microM, Vmax; 69.7 nmol/min/mg) and 5 beta-androstanedione (Km; 7.7 microM, Vmax; 135.7 nmol/min/mg) were demonstrated for comparison with those for 17 alpha-hydroxyprogesterone (Km; 26.2 microM, Vmax; 1.3 nmol/min/mg) which is a substrate for 20 alpha-HSD activity. Regarding oxidation, the apparent Km and Vmax values for 3 alpha-hydroxy-5 alpha-androstan-17-one were 1.7 microM and 43.2 nmol/min/mg, and 1.2 microM and 32.1 nmol/min/mg for 3 alpha-hydroxy-5 beta-androstan-17-one, respectively. 20 alpha-HSD activity in the reduction of 17 alpha-hydroxyprogesterone catalyzed by the purified enzyme was inhibited competitively by addition of 5 alpha-DHT with a Ki value of 2.0 microM. Furthermore, 17 alpha-hydroxyprogesterone inhibited competitively 3 alpha-HSD activity with a Ki value of 150 microM.