Methanol esterification reactions catalyzed by snake venom and bovine intestinal 5'-nucleotide phosphodiesterases. Formation of nucleoside 5'-monophosphate methyl esters from guanosine 5'-triphosphate and other nucleoside 5'-polyphosphates.

It is not known whether the enzymes 5'-nucleotide phosphodiesterase/nucleotide pyrophosphatase (EC 3.1.4.1/EC 3.6.1.9) catalyze the transfer of nucleotides to acceptors other than water. We have investigated the action of snake venom and bovine intestinal mucosa phosphodiesterases on nucleoside 5'-polyphosphates in the presence of methanol. In those conditions, GTP was converted by snake venom phosphodiesterase to a mixture of GMP and another compound with a different retention time in reverse-phase high-performance liquid chromatography. That compound, by ultraviolet, 1H- and 13C-nuclear magnetic resonance spectroscopic analysis, and by enzyme analysis, was characterized as the methyl ester of GMP (GMP-OMe). The molar fraction [GMP-OMe]/[GMP + GMP-OMe] formed was higher than the molar fraction of methanol as a solvent in reaction mixtures. Similar reactions took place at comparable rates with snake venom and bovine intestinal mucosa phosphodiesterases using several nucleoside 5'-polyphosphates as substrates. The ability of 5'-nucleotide phosphodiesterases to catalyze transfer reactions to a non-water acceptor is relevant to the mechanism of the enzymes, to their use as analytical tools, and to their possible use/role in the preparative/in vivo synthesis of nucleotide esters.     

Hydrolysis of phosphonate esters catalyzed by 5'-nucleotide phosphodiesterase.

4-Nitrophenyl and 2-napthyl monoesters of phenylphosphonic acid have been synthesized, and an enzyme catalyzing their hydrolysis was resolved from alkaline phosphatase of a commerical calf intestinal alkaline phosphatase preparation by extensive ion-exchange chromatography, chromatography on L-phenylalanyl-Sepharose with a decreasing gradient of (NH4) 2SO4, and gel filtration. Detergent-solubilized enzyme from fresh bovine intestine was purified after (NH4)2SO4 fractionation by the same technique. The purified enzyme is homogeneous by polyacrylamide gel electrophoresis and sedimentation equilibrium centrifugation. It has a molecular weight of 108,000, contains approximately 21% carbohydrate, and has an amino acid composition considerably different from that reported from alkaline phosphatase from the same tissue. The homogeneous intestinal enzyme, an efficient catalyst of phosphonate ester hydoolysis but not of phosphate monoester hydrolysis, was identified as a 5'-nucleotide phosphodiesterase by its ability to hydrolyze 4-nitrophenyl esters of 5'-TMP but not of 3'-TMP. Also consistent with this identification was the ability of the enzyme to hydrolyze 5'-ATP to 5'-AMP and PPi, NAD+ to 5'-AMP and NMN, TpT to 5'-TMP and thymidine, pApApApA to 5'-AMP, and only the single-stranded portion of tRNA from the 3'-OH end. Snake venom 5'-nucleotide phosphodiesterase also hydrolyzes phosphonate esters, but 3'-nucleotide phosphodiesterase of spleen and cyclic 3',5'-AMP phosphodiesterase do not. Thus, types of phosphodiesterases can be conveniently distinguished by their ability to hydrolyze phosphonate esters. As substrates for 5'-nucleotide phosphodiesterases, phosphonate esters are preferable to the more conventional esters of nucleotides and bis(4-nitrophenyl) phosphate because of their superior stability and ease of synthesis. Furthermore, the rate of hydrolysis of phosphonate esters under saturating conditions is greater than that of the conventional substrates. At substrate concentrations of 1 mM the rates of hydrolysis of phosphonate esters and of nucleotide esters are comparable and both superior to that of bis(4-nitrophenyl) phosphate.     

Distribution of phosphodiesterase I in normal human tissues

Phosphodiesterase I (PDE I) is an exonuclease capable of hydrolyzing a variety of phosphate ester and pyrophosphate bonds. Cell fractionation and histochemical studies in animal tissues have localized PDE I in the plasma membrane of various epithelia. This suggests a role for the enzyme in active transport. Distribution of PDE I in human tissues has not previously been studied. We have produced a polyclonal antiserum to bovine intestinal PDE I and have demonstrated crossreactivity with the human intestinal enzyme. This polyclonal antiserum was used in PAP immunocytochemistry to localize immunoreactive PDE I in a variety of human tissues. Localization was prominent in the gastrointestinal tract, including the cytoplasm of gastric mucosa parietal cells, cytoplasm of surface epithelium and isolated crypt cells in small intestine, and the colonic epithelial cytoplasm and brush border. Parotid gland acinar cells and scattered ductal cells showed positive cytoplasmic staining. Acinar and scattered pancreatic islet cells contained immunoreactive PDE I, as did Kupffer cells of the liver sinusoids. Immunoreactive PDE I was found in all vascular endothelia. The epithelium of the urinary tract showed extensive immunoreactivity. This included the distal convoluted and collecting tubules of the kidney, and ureteral and bladder urothelium. In previous histochemical studies of animal tissues, no evidence of PDE I activity was noted in male or female reproductive tract. In this study, immunoreactive PDE I was localized to human Sertoli cells and to basal epithelium of the epididymis and prostate acini. Fallopian tube epithelium of female reproductive tract also demonstrated immunoreactive PDI I, as did several cell types in term placenta. Our immunocytochemical results with human tissues differ significantly from previous histochemical studies in animal tissues, principally in the genitourinary system. This may be due in part to the different detection systems employed as well as the higher sensitivity of the immunoperoxidase technique. This underscores the importance of adjunct techniques in tissue surveys. The widespread epithelial distribution of immunoreactive PDE I detected by this polyclonal antibody implies an integral role in cell function, probably in active transport.     

Synthesis of two diastereoisomeric p-nitrophenyl phosphodiesters of 2'',3''-secouridine and their affinity for phosphodiesterases.

The synthesis of the p-nitrophenyl esters of the 5'- and 3'-phosphates of the nucleoside analogue 2',3'-secouridine are described. Unlike the corresponding diesters of thymidine, these two compounds are diastereoisomers. Their affinity for phosphodiesterases types I and II were investigated. Both analogues were hydrolysed very slowly by snake venom phosphodiesterase but their affinity for the enzyme was similar to that of the p-nitrophenyl ester of thymidine 5'-monophosphate of which they were both competitive inhibitors with Ki approximately Km. Neither compound was hydrolysed by spleen phosphodiesterase but both competitively inhibited the p-nitrophenyl ester of thymidine 3'-monophosphate, with Ki's slightly higher than the Km. Although for each enzyme the Ki of the correct analogue phosphodiester (i.e. the 5'-derivative for snake venom and the 3'-derivative for spleen) was the lower, the absolute specificity seen for the normal substrates had been lost.     

Enzymic hydrolysis of phosphonate esters. Reaction mechanism of intestinal 5'-nucleotide phosphodiesterase.

The mechanism of bovine intestinal 5'-nucleotide phosphodiesterase was investigated by determining kinetic constants of systematically varied substrates, with emphasis on esters of phosphonic acids (which have much higer Vmax values than conventional phosphodiester substrates), and by pre-steady-state kinetics using bis(4-nitrophenyl) phosphate as substrate. The results suggest a ping-pong type mechanism, with participation of a covalent enzyme intermediate.     

Characterization of phosphohydrolase activity in bovine spleen extracts: identification of a bis(p-nitrophenyl)phosphate-hydrolyzing activity (phosphodiesterase IV) which also acts on adenosine triphosphate

Several bovine spleen enzymes with acid pH optima, some of which hydrolyze bis(p-nitrophenyl)phosphate and therefore fit the definition of "phosphodiesterase IV," were partially separated by isoelectric focusing and ion-exchange techniques. The activities were characterized by zymogram analysis with the aid of p-nitrophenyl and 4-methylumbelliferyl phosphate and phosphonate substrates. A number of these enzymes meet the criteria for phosphodiesterase I or other phosphodiesterases. However, the predominant phosphodiesterase I hydrolyzes the bis(p-nitrophenyl)-and 4-methylumbelliferyl phosphates, p-nitrophenyl and 4-methylumbelliferyl phenylphosphonate, and ATP at the beta-gamma bond, but not p-nitrophenyl or 4-methylumbelliferyl 5'-thymidylate (the usual PDE I substrates). These properties, as well as the pH optimum, distinguish the activity from the previously described, alkaline pH optimum PDE I. A second phosphodiesterase hydrolyzes only the phenylphosphonates. Several other activities, less well described, are apparent on zymograms. None of the phosphodiesterases IV was also a phosphodiesterase II (no hydrolysis of 4-methylumbelliferyl 3'-thymidylate).     

Interaction of Dictyostelium discoideum lysosomal enzymes with the mammalian phosphomannosyl receptor. The importance of oligosaccharides which contain phosphodiesters

Mammalian cell lysosomal enzymes or phosphorylated oligosaccharides derived from them are endocytosed by a phosphomannosyl receptor (PMR) found on the surface of fibroblasts. Various studies suggest that 2 residues of Man-6-P in phosphomonoester linkage but not diester linkage (PDE) are essential for a high rate of uptake. The lysosomal enzymes of the slime mold Dictyostelium discoideum are also recognized by the PMR on these cells; however, none of the oligosaccharides from these enzymes contain 2 phosphomonoesters. Instead, most contain multiple sulfate esters and 2 residues of Man-6-P in an unusual PDE linkage. In this study I have tried to account for the unexpected highly efficient uptake of the slime mold enzymes. The results show that nearly all of the alpha-mannosidase molecules contain the oligosaccharides required for uptake, and that each tetrameric, holoenzyme molecule has sufficient carbohydrate for an average of 10 Man8GlcNAc2 oligosaccharides. None of the oligosaccharides or glycopeptides from the lysosomal enzymes bind to an immobilized PMR, but those with 2 PDE show slight interaction. Competition of 125I-beta-glucosidase uptake by various carbohydrate-containing fractions indicates that the best inhibitors are those with 2 PDE, either with or without sulfate esters. Furthermore, the uptake of a lysosomal enzyme isolated from a mutant strain (modA), which produces oligosaccharides with only 1 but not 2 PDE, is about 10-fold less than the uptake of wild-type enzyme which has predominantly 2 PDE. Complete denaturation of 125I-labeled wild-type beta-glucosidase in sodium dodecyl sulfate/dithiothreitol also reduces its uptake by about 10-fold. Taken together, these results suggest that the interactions of multiple, weakly binding oligosaccharides, especially those with 2 PDE, are important for the high rate of uptake of the slime mold enzymes. The conformation of the protein may be important in orienting the oligosaccharides in a favorable position for binding to the PMR.     

The action of venom and proteolytic enzymes on the non-specific inhibitor of hyaluronidase

1. It has been shown that a number of proteolytic enzymes and snake venom, in relatively small amounts, and within a wide range of pH variation, will restore hyaluronidase activity after its inhibition by serum. 2. The known properties of the venom protease are found to be identical with those of Haas' "proinvasin I." It is concluded that the protease of the venom offers adequate explanation for the effects previously attributed to "proinvasin I." 3. Proteolytic activity is found in hyaluronidase preparations of bovine origin and is considered to be responsible for the reversal of inhibition of hyaluronidase by serum.     

Selective protection of in vitro synthesized cDNA against nucleases by incorporation of phosphorothioate-analogues.

The conditions for the stepwise synthesis of single- (ss) and double-stranded (ds) cDNA using thio-analogues instead of dNTPs are described in this paper. RNA of paramyxovirus Sendai (strain 6/94) serves as template in these experiments. The increased resistance of this alpha S-modified cDNA against several nucleases, like S1-Nuclease, DNase I, Exonuclease III, snake venom Phosphodiesterase (PDE) and the combination of DNase I and PDE is demonstrated.     

Structural and functional characterization of a prothrombin activator from the venom of Bothrops neuwiedi

A prothrombin activator from the venom of Bothrops neuwiedi was purified by gel filtration on Sephadex G-100, ion-exchange chromatography on DEAE-Sephacel and affinity chromatography on a Zn2+-chelate column. The overall purification was about 200-fold, which indicates that the prothrombin activator comprises about 0.5% of the crude venom. The venom activator is a single-chain protein with an apparent molecular weight of 60 kDa. It readily activated bovine prothrombin with a Km of 38 microM and a Vmax of 120 mumol prothrombin activated per min per mg of venom activator. Venom-catalyzed prothrombin activation was not accelerated by the so-called accessory components of the prothrombinase complex, phospholipids plus Ca2+ and Factor Va. Gel-electrophoretic analysis of prothrombin activation indicated that the venom activator only cleaved the Arg-323-Ile-324 bond of bovine prothrombin, since meizothrombin was the only product of prothrombin activation. The activator did not hydrolyze commercially available p-nitroanilide substrates and its prothrombin-converting activity was not inhibited by benzamidine, phenylmethylsulfonyl fluoride, dansyl-Glu-Gly-Arg-chloromethyl ketone and soy-bean trypsin inhibitor. However, chelating agents such as EDTA, EGTA and o-phenanthroline rapidly destroyed the enzymatic activity of the venom activator. The activity of chelator-treated venom activator could be partially restored by the addition of an excess CaCl2. These results indicate that the venom activator remarkably differs from Factor Xa and that the enzyme is not a serine proteinase, but likely belongs to the metalloproteinases. The structural and functional properties of the venom prothrombin activator from B. neuwiedi are similar to those reported for the venom activator from Echis carinatus.     

A FYVE-containing unusual cyclic nucleotide phosphodiesterase from Trypanosoma cruzi

Cyclic-nucleotide-specific phosphodiesterases (PDEs) are key players in the intracellular signaling pathways of the important human pathogen Trypanosoma cruzi. We report herein the identification of an unusual PDE from this protozoal organism. This enzyme, TcrPDEC, is a member of the class I PDEs, as determined from the presence of a characteristic signature sequence and from the conservation of a number of functionally important amino acid residues within its catalytic domain. Class I PDEs include a large number of PDEs from eukaryotes, among them all 11 human PDE families. Unusually for an enzyme of this class, TcrPDEC contains a FYVE-type domain in its N-terminal region, followed by two closely spaced coiled-coil domains. Its catalytic domain is located in the middle of the polypeptide chain, whereas all other class I enzymes contain their catalytic domains in their C-terminal parts. TcrPDEC can complement a PDE-deficient yeast strain. Unexpectedly for a kinetoplastid PDE, TcrPDEC is a dual-specificity PDE that accepts both cAMP and cGMP as its substrates.     

A protein inhibitor of calmodulin-regulated cyclic nucleotide phosphodiesterase in amphibian ovaries

The cytosol fraction of an extract of Xenopus laevis ovaries contains a protein inhibitor that can specifically block the activation of calmodulin-sensitive cyclic nucleotide phosphodiesterase (PDE I) found in that tissue. This inhibitor was purified by DEAE-cellulose chromatography, gel filtration on Sephacryl S-200, and affinity chromatography on calmodulin-Sepharose. It has a molecular weight of approximately 90,000, and is heat-labile and susceptible to inactivation by chymotrypsin. The inhibitor blocks calmodulin activation of cyclic nucleotide phosphodiesterases from amphibian ovary and bovine brain and of the myosin light chain kinase from rabbit smooth muscle, but does not affect the activity of a calmodulin-insensitive cyclic nucleotide phosphodiesterase. The inhibitor not only affects the activation of Xenopus PDE I and of the bovine brain phosphodiesterase by calmodulin, but also inhibits the stimulation of these enzymes by lysophosphatidylcholine. The inhibitor also acts on PDE I activated by partial tryptic proteolysis, but the enzyme fully activated by trypsin is only slightly susceptible to inhibition by this protein. The inhibition of PDE I activation caused by this ovarian factor can be reversed by adding excess amounts of calmodulin or lysophosphatidylcholine. The presence of this inhibitor provides a possible explanation for the previously observed inactivity of PDE I in vivo.     

3',5'-cyclic nucleotide phosphodiesterase from human brain

3':5'-Cyclic nucleotide phosphodiesterase was isolated from human brain and characterized. After the first stage of purification on phenyl-Sepharose, the enzyme activity was stimulated by Ca2+ and micromolar concentrations of cGMP. High pressure liquid chromatography on a DEAE-TSK-3SW column permitted to identify three ranges of enzymatic activity designated as PDE I, PDE II and PDE III. Neither of the three enzymes possessed a high selectivity for cAMP and cGMP substrates. The catalytic activity of PDE I and PDE II increased in the presence of Ca2+-calmodulin (up to 6-fold); the degradation of cAMP was decreased by cGMP. The Ca2+-calmodulin stimulated PDE I and PDE II activity was decreased by W-7. PDE I and PDE II can thus be classified as Ca2+-calmodulin-dependent phosphodiesterases. With cAMP as substrate, the PDE III activity increased in the presence of micromolar concentrations of cGMP (up to 10-fold), Ca2+ and endogenous calmodulin (up to 2-3-fold). No additivity in the effects of saturating concentrations of these compounds on PDE III was observed. Ca2+ did not influence the rate of cGMP hydrolysis catalyzed by PDE III. In comparison with PDE I and PDE II, the inhibition of PDE III was observed at higher concentrations of W-7 and was not limited by the basal level of the enzyme. These results do not provide any evidence in favour of the existence of several forms of the enzyme in the PDE III fraction. The double regulation of PDE III creates some difficulties for its classification.     

Purification and characterization of calmodulin from porcine renal medulla

A calcium sensitive phosphodiesterase (PDE) activated by an endogenous calmodulin was identified in the cytosolic fraction of porcine renal medulla. The PDE and calmodulin were separated from each other by DEAE-cellulose column chromatography. Calmodulin was purified from a heat-treated supernatant by column chromatography with DEAE-cellulose and hydroxylapatite. The purified renal calmodulin has a molecular weight of 17,500, is heatstable, and has a pI of 4.2. Activation of the renal PDE by calmodulin was immediate and stoichiometric. The renal calmodulin and PDE cross react with bovine brain calmodulin and PDE, indicating a lack of tissue and species specificity. Thus, renal calmodulin is very similar to bovine brain calmodulin. However, renal calmodulin did not affect detergent-solubilized or membrane-bound renal adenylate cyclase or the antidiuretic hormone-stimulated activity of the enzyme. These results suggest that calmodulin may function in the renal medulla to regulate cAMP levels by stimulation of PDE but not adenylate cyclase. However, the ubiquitous distribution of calmodulin in eukaryotic cells and its effects on a number of other enzymes allow the possibility that calmodulin may have a role in renal function other than cAMP metabolism.     

Epiphyseal cartilage cyclic nucleotide phosphodiesterase: partial purification and kinetic characteristics of multiple enzymes.

Partial purification of chicken epiphyseal PDE activity by centrifugation and column chromatography has defined two distinct peaks of PDE activity. The faster eluting peak (I) has a higher apparent Km for cyclic AMP than the slower eluting major peak (IIs and II). Peak I has greater activity towards cyclic AMP as a substrate than towards cyclic GMP but does use both substrates. Peak I is not inhibited by T-3 or indomethacin at physiological concentrations. Substrates studies demonstrate the presence of at least two overlapping PDE species in the major peak(IIs and II). There is suggestive evidence that indomethacin is a more potent inhibitor of peak IIs which can use either cyclic AMP or cyclic GMP as substrates, whereas T-3 is a more potent inhibitor of fractions eluting where the enzyme only has activity with cyclic AMP.     

Three-dimensional structure of non-activated cGMP phosphodiesterase 6 and comparison of its image with those of activated forms

Cyclic GMP phosphodiesterase (PDE6) in rod photoreceptors, a key enzyme in vertebrate phototransduction, consists of two homologous catalytic subunits (Palpha and Pbeta) and two identical regulatory subunits (Pgammas). Pgamma regulates the PDE activity through its direct interaction with transducin. Here, using electron microscopy and image analysis of single particles, we show the three-dimensional organization of the basic form of bovine PDE, Palphabetagammagamma, and compare its average image with those of Pgamma-released PDE. The structure of Palphabetagammagamma appears to be a flattened bell-shape, with dimensions of 150 x 108 x 60A, and with a handle-like protrusion attached to the top of the structure. Except for the protrusion, the organization consists of two homologous structures arranged side by side, with each structure having three distinct regions, showing pseudo twofold symmetry. These characteristics are consistent with a model in which the overall structure of Palphabetagammagamma is determined by hetero-dimerization of Palpha and Pbeta, with each subunit consisting of one catalytic and two GAF regions. A comparison of the average image of Palphabetagammagamma with those of Pgamma-released PDE suggests that Pgamma release does not affect the overall structure of Palphabeta, and that the Palphabeta C-terminus, but not Pgamma, is a determinant for the Palphabeta orientation on carbon-coated grids. These observations suggest that the basic structure of PDE does not change during its regulation, which implies that Palphabeta is regulated by its regional interaction with Pgamma.     

Prothrombin activation by an activator from the venom of Oxyuranus scutellatus (Taipan snake)

The prothrombin activator from the venom of Oxyuranus scutellatus (Taipan snake) was purified by gel filtration on Sephadex G-200 and ion-exchange chromatography on QAE-Sephadex. The activator is a large protein with a molecular weight of approximately 300,000, which is composed of subunits of Mr 110,000 and 80,000 and two disulfide-linked polypeptides of Mr 30,000. One or both of these Mr 30,000 subunits contain the active site. The venom activator readily converts Factor Xa-specific chromogenic substrates and is also able to activate prothrombin (Km = 166 microM, Vmax = 2.5 mumol of prothrombin activated per min/mg of venom). Gel electrophoretic analysis of prothrombin activation indicates that the venom activator randomly cleaves the Arg274-Thr275 and Arg323-Ile324 bonds of prothrombin since both thrombin and meizothrombin are formed as reaction products. Venom-catalyzed prothrombin activation is not affected by bovine Factor Va but is greatly stimulated by phospholipids plus Ca2+ ions. This stimulatory effect is explained by a decrease of the Km for prothrombin. In the presence of 50 microM phospholipid vesicles (25% phosphatidylserine/75% phosphatidylcholine; mole/mole), the Km is 0.34 microM and the Vmax is 7.1 mumol of prothrombin activated per min/mg of venom. The purified venom activator contains gamma-carboxyglutamic acid residues which presumably function in the interaction between the venom activator and phospholipids. Treatment of the activator with 0.8 M NaSCN strongly reduces its ability to activate prothrombin but has no effect on its amidolytic activity. The prothrombin-converting activity of the NaSCN-treated activator can be restored with bovine Factor Va. During prolonged gradient gel electrophoresis, the Mr 300,000 activator dissociates into smaller subunits. This causes a loss of the prothrombin-converting activity, while the amidolytic activity is recovered in a protein with an apparent molecular weight of 57,000. This protein can, however, rapidly activate prothrombin in the presence of Factor Va or in the presence of a protein component of Mr 220,000 that also migrates on the gel. These results suggest that the prothrombin activator from the O. scutellatus venom is a multimeric protein complex consisting of a Factor Xa-like enzyme and a Factor Va-like cofactor.     

Isolation and properties of carboxycathepsin (peptidyl-dipeptidase) from bovine lung tissue]

2200-fold purified homogenous preparation of carboxycathepsin (peptidyl-dipeptidase) is isolated from bovine lung. The enzyme isolated converts angiotensine I into angiotensine II and distroys bradikinin. It is active in neutral medium, is activated by chloride ion and is inhibited by EDTA and Middle Asian snakes venom. The molecular weight of the enzyme is 180 000-190 000 as estimated by means of polyacrylamide gel electrophoresis, its isoelectric point is 4.48-4.53. The comparison of properties and specificity of carboxycathepsin from bovine lung and kidney draws to the conclusion that both enzymes are identical.     

Purification and characterization of an organ specific haemorrhagic toxin from Vipera russelli russelli (Russell's viper) venom

A haemorrhagic toxin (VRR-12) from Vipera russelli russelli (Russell's viper) venom has been purified by ion-exchange chromatography on CM-Sephadex C-50 followed by size-exclusion HPLC to electrophoretically homogeneous state. It is a 12 kDa single polypeptide having 1 mole of Zn+2 ion. This toxin induces intense intestinal haemorrhage and to a lesser extent skeletal muscle haemorrhage in mice. It does not show detectable proteolytic and esterolytic activity with selected substrates under specified conditions, haemolytic and phospholipase activity. When VRR-12, preincubated with bivalent antiserum against Saw-scaled and Russell's viper venom or EDTA was injected, haemorrhagic activity was not reduced, on the other hand preincubation with phenylmethyl sulphonyl fluoride reduced the activity markedly. Biodistribution studies with 125I VRR-12 show that haemorrhagic manifestation by this toxin is not a direct function of the fraction of the totally administered toxin distributed to that tissue.     

Molecular organization of bovine rod cGMP-phosphodiesterase 6.

Phosphodiesterase 6 (PDE6), a multisubunit (alphabetagamma(2)delta) enzyme, plays a major role in visual function by hydrolysing cGMP in response to a light stimulus. Solubilized bovine rod PDE6 molecules depleted of their gamma subunits were purified to homogeneity from bovine retinal rods and their molecular organization was investigated by electron microscopy. Image analysis of single particles revealed the three-dimensional dimeric arrangement of the purified alphabetadelta complex, and the internal organization of each catalytic subunit into three distinct domains at a resolution of 2.8 nm. The relative volume of each domain is consistent with sequence analysis and functional data, which suggest that these domains correspond to the catalytic and two GAF domains. This hypothesis was confirmed by immunolabelling experiments, which located the N-terminal part of the catalytic subunit where the major interaction between the two alphabeta subunits was found to occur. The 3D molecular organization of human platelet PDE5 appears highly homologous to that of bovine rod PDE6, as predicted by similarities in their primary sequences. These observations describe the quaternary organization of the catalytic PDE6 alphabeta complex, and place the catalytic and regulatory domains on a structural model.