Cloning and sequencing of rat liver carboxylesterase ES-4 (microsomal palmitoyl-CoA hydrolase).

A cDNA which encodes a carboxylesterase of 561 amino acid residues including a cleavable signal peptide is described. The enzyme expressed in COS cells migrates during PAGE (SDS-, and non-denaturing) as a single prominent band in the region of liver ES-4. It ends in the C-terminal cell-retention signal -HNEL, which, in COS cells overexpressing the enzyme, appears to be slightly less efficient than the signals -HTEL and -HVEL of ES-3 and ES-10 respectively. Glycosylation is not essential for intracellular retention, but leads to a higher activity. As do many carboxylesterases, the enzyme expressed in COS cells hydrolyses omicron-nitrophenyl acetate and alpha-naphthyl acetate. It also hydrolyses acetanilide, although less efficiently than ES-3, and, distinctively, palmitoyl-CoA. In addition to the four canonical Cys residues of the carboxylesterases, it contains a fifth, unpaired Cys336, which apparently is not essential for the catalytic properties. Indeed, treatment with iodoacetamide or substitution of Cys336 by Phe does not markedly alter the activity of the enzyme on the various substrates. The predicted structure of ES-4 is highly homologous to that of two other recently cloned esterases which also end in -HNEL [Yan, Yang, Brady and Parkinson (1994) J. Biol. Chem. 269, 29688-29696; Yan, Yang, and Parkinson (1995) Arch. Biochem. Biophys. 317, 222-234]. Together, these isoenzymes probably account for the closely spaced bands observed in the region of ES-4 in non-denaturing PAGE.

The tumor protein MAGE-1 is located in the cytosol of human melanoma cells.

The MAGE-1 protein has been localized to the cytosol of human melanoma culture cells (MZ2-MEL.43) by subcellular fractionation. Its distribution between nuclear, mitochondrial, microsomal and cytosolic fractions was established by SDS-PAGE and immunoblotting with a rabbit antiserum and compared to that of markers for the main cell compartments. The immunoreactive material was recovered in the cytosolic fraction as a polypeptide migrating at 42 kDa, which has been further identified as the MAGE-1 protein because it comigrates with the product of cell-free transcription-translation of the MAGE-1 cDNA. Location to the cytosol was confirmed by immunofluorescence microscopy.

Cloning and sequencing of rat liver carboxylesterase ES-3 (egasyn).

The cDNA of the rat carboxylesterase ES-3 encodes a polypeptide with 561 amino acid residues including a cleavable signal peptide at the N-terminus. The processed polypeptide shows over 90% sequence identity to mouse egasyn (ES-22); its calculated pI (5.5) matches the value determined for purified liver ES-3. The product expressed in COS cells migrates in native gels in the region of ES-3 and is similarly active on acetanilide. It is retained in the cells, as predicted from its C-terminus HTEL, and bears a single endo-H sensitive oligosaccharide chain. The nonglycosylated form expressed in the presence of tunicamycin is also intracellular, but substantially less active.

Identification and characterization of the tumor-specific P1A gene product.

In murine mastocytoma P815, gene P1A directs the expression of antigens P815A and B which are the target of a T cell-mediated rejection response in syngeneic animals. This gene is expressed at a high level in various tumors, but is silent in normal tissues except testis and placenta; its activation is thus possibly related to malignant transformation. An anti-synthetic peptide rabbit antiserum reacted by immunoblotting with a cellular protein migrating near 40 kDa on SDS-PAGE. The immunoreactive protein was detected only in lysates from cells which express antigen P815A: P1.HTR mastocytoma cells and, after transfection with cosmids carrying the P1A gene, the antigen-loss variant P0.HTR cells and DAP-3 H-2Ld fibroblasts. The identity of this protein as the P1A gene product was confirmed by cell-free transcription-translation of the P1A cDNA, the product of which also migrated near 40 kDa in SDS-PAGE and was captured by protein A-Sepharose in the presence of the antiserum. Subcellular fractionation by differential and isopycnic centrifugation indicated that the P1A protein is associated with cytoplasmic membranes demonstrating a broad distribution with respect to size and density. Immunofluorescence microscopy also revealed a cytoplasmic signal, particularly intense in small vesicles, which coincides with that produced by an anti-mouse type I collagen guinea pig antiserum except near the cell periphery where the P1A signal is weaker. We conclude that the P1A protein is bound to membranes of the secretory pathway, at a concentration which goes increasing from the endoplasmic reticulum to secretion vesicles.(ABSTRACT TRUNCATED AT 250 WORDS)

The tum- antigens P91A and P198 derive from proteins located in the cytosolic compartment of cells.

To characterize the proteins P91Ap and P198p, of which mutants generate the tum- antigens P91A and P198, respectively, rabbit antisera were raised with ovalbumin-coupled synthetic peptides that correspond to their respective C terminus. In immunoadsorption tests using immobilized protein A the antisera recognized the translation products synthesized by rabbit reticulocyte lysates programmed with the SP6 polymerase transcripts of the P91A and P198 cDNA. The presence of the two proteins was demonstrated by SDS-PAGE and immunoblotting in all the mouse cells and organs examined. P91Ap is a constituent of the cytosol; despite a remarkable homology to the Drosophila diphenol oxidase DOX-A2, it separates from murine catechol oxidase activity in rate zonal sedimentation analysis. P198p is a ribosomal constituent, or a factor firmly linked to both the free and membrane-bound ribosomes. These subcellular localizations strengthen other evidence that the antigens presented to T lymphocytes by class I products of the major histocompatibility complex derive from proteins of the cytosol, or in direct contact with it.

Presentation of mouse tum- P91A antigen from chimeric proteins with different subcellular localizations by class I molecules of the major histocompatibility complex.

Like many antigens presented by class I molecules the mouse Ld-binding tum- antigen P91A derives from a cytosolic protein. To decide how stringent this localization is for presentation to cytotoxic lymphocytes (CTL) the P91A template has been inserted in the cDNA of rat esterase ES-10, a protein located in the endoplasmic reticulum (ER), and in the cDNA of mouse interleukin-9, a secretory product of lymphocytes. The esterase construct was also engineered to replace the C-terminal leucine by arginine, which causes secretion of the protein, or to delete the N-terminal presequence, which prevents transfer of the nascent chain to the ER. After cell-free transcription-translation, or transfection in COS cells, the products of the chimeric cDNA had the expected size and localization; however, the truncated form of esterase remained undetected in COS cells. The various chimeric templates were transfected in P1.HTR cells (H-2d); upon challenge with Ld-restricted anti-P91A CTL the cells were lyzed almost as efficiently as cells transfected with the full-length P91A cDNA. We conclude that peptide fragments that bind to class I molecules of the major histocompatibility complex can be generated in the ER.

Topogenesis of carboxylesterases: a rat liver isoenzyme ending in -HTEHT-COOH is a secreted protein.

We have cloned a rat liver cDNA that encodes a carboxylesterase isoenzyme, as revealed by immunoprecipitation, cytochemical staining and inhibition by bis-p-nitrophenylphosphate of the product expressed in transfected COS cells. The predicted polypeptide ends in -HTEHT-COOH. The product is secreted by COS cells with a half-time of about 1 hour, after maturation of oligosaccharide chains in the Golgi complex. A variant ending in -HTEL-COOH is stable in the cells. This strengthens the existing evidence that the HXEL-COOH end signals proteins for retrieval from the secretory traffic in animal cells. The encoded enzyme still remains to be identified. It shows 98% homology to an esterase sequenced earlier (Takagi et al. 1988, J. Biochem. 104, 801-806; Long et al. 1988, Biochem. Biophys. Res. Commun. 156, 866-873); however it must be an enzyme from the serum, not from the microsomes.

The COOH terminus of several liver carboxylesterases targets these enzymes to the lumen of the endoplasmic reticulum.

To investigate the potential role of the COOH-terminal peptides in retaining a family of soluble carboxylesterases in the lumen of the endoplasmic reticulum, the pI 6.1 esterase cDNA has been cloned into the pKCR3 vector for transient expression in COS cells. The plasmid-encoded product appeared to be identical to the authentic enzyme: it was active on alpha-naphthyl acetate and behaved as a homotrimer of noncovalently bound 60-kDa subunits which contain a single, endo-beta-N-acetylglucosaminidase H-sensitive oligosaccharide chain. This enzyme was retained in the transgenic COS cells. In contrast, a mutated form ending in HVER-COOH was secreted, indicating that the natural terminus HVEL-COOH contains topogenic information, with the ultimate Leu residue as an essential part. Variants of pI 6.1 esterase ending in HIEL-COOH, or HTEL-COOH were retained in cells to the same extent as the wild-type protein. Therefore, the sequences HIEL and HTEL present at the COOH termini of several liver esterases (rabbit forms 1 and 2, human esterase, mouse egasyn, and rat pI 6.4 esterase) most likely have a function in their localization in the endoplasmic reticulum. Finally, an HDEL-COOH variant of pI 6.1 esterase was also normally retained, demonstrating that this signal can be correctly decoded by the sorting machinery of mammalian cells. Cell retention signals of the type HXEL-COOH appear to be common in higher eukaryotes and tolerate considerable variation at the antepenultimate X residue.

Nucleotide sequence of cDNA coding for rat liver pI 6.1 esterase (ES-10), a carboxylesterase located in the lumen of the endoplasmic reticulum.

A commercial rat liver cDNA library in lambda gt11 was screened with a rabbit antiserum to native pI 6.1 esterase (ES-10). The inserts of the immunoreactive clones were short (0.9-1.1 kbp). One of these was used as a probe to rescreen the library, yielding 30 clones, two of which contained relatively long (approx. 1.9 kbp) and widely overlapping cDNA inserts. They did not contain the first two nucleotide residues of the initiator codon, nor the 5'-end untranslated portion of the mRNA. These were derived from a home-made rat liver cDNA library in lambda gt11, screened with an oligonucleotide corresponding to the 5'-end of the already known cDNA sequence. The nucleotide sequence consists of 48 bp of 5'-end non-coding region, 1695 bp of coding region and 212 bp of 3'-end non-coding region including a 20 bp poly(A) tail. The signal peptide and the mature protein subunit are 18 and 547 residues long respectively. Tyr is confirmed as N-terminal residue. The predicted amino acid sequence is highly similar to those of rabbit liver esterase forms 1 (77% identity) and 2 (56% identity), determined by protein sequencing [Korza & Ozols (1988) J. Biol. Chem. 263, 3486-3495; Ozols (1989) J. Biol. Chem. 264, 12533-12545]. The three enzymes share the Ser and His residues presumed to be part of the active site, four Cys residues and a high proportion of charged side chains at their C-terminus. The C-terminal tetrapeptides of the three esterases (-HVEL, -HIEL and -HTEL for pI 6.1 and forms 1 and 2 esterases respectively) are reminiscent of, but not identical with, the localization signal identified in other proteins of the endoplasmic-reticulum lumen (-KDEL in animal cells [Munro & Pelham (1987) Cell 48, 899-907]; -HDEL in yeast [Pelham, Hardwick & Lewis (1988) EMBO J. 7, 1757-1762]). We still lack direct evidence to decide whether or not these C-terminal tetrapeptides commit esterases to reside in the endoplasmic reticulum. In that case the antepenultimate residue (D, V, I or T) would be only weakly stringent, and some sequences primed by H instead of K would be recognized in animal as well as in yeast cells.

Subcellular fractionation of epithelial cells from toad urinary bladder. 1. Assay of marker enzymes and differential centrifugation.

We have undertaken the analytical fractionation of epithelial cells from toad urinary bladder, a tissue extensively used to study epithelial transport of ions and water. In an attempt to establish markers for the main subcellular organelles, a number of enzymes were assayed in cell homogenates. The nearly ubiquitous plasma membrane marker 5'-nucleotidase, and the transferases that donate N-acetylglucosaminyl, galactosyl, and sialyl residues to glycoproteins and glycolipids in the Golgi complex were not detectable. Glucose-6-phosphatase activity was low in relation to that of nonspecific phosphatases and, therefore, not suitable for identifying the endoplasmic reticulum. Like the cytosolic enzyme lactate, dehydrogenase, catalase was essentially found in the high-speed supernatant, with a noteworthy part of aminopeptidase (substrate, leucyl-beta-naphthylamide) and NAD glycohydrolase. Other enzymes, including cytochrome c oxidase, acid phosphatase, acid N-acetyl-beta-glucosaminidase, alkaline phosphatase, alkaline phosphodiesterase I, nucleoside diphosphatase (substrate ADP), oligomycin-resistant Mg++-ATPase, and mannosyltransferase (acceptor, dolichylphosphate) were fairly active and largely sedimentable. After differential centrifugation, cytochrome oxidase, acid phosphatase, and acid N-acetyl-beta-glucosaminidase were typically associated with the large granule fraction, whereas the other sedimentable enzymes exhibited a broad distribution profile overlapping the nuclear, large granule, and microsome fractions. Their behavior in density equilibrium centrifugation is examined in a companion paper.

Subcellular fractionation of epithelial cells from toad urinary bladder. 2. Isopycnic centrifugation and effect of density perturbants.

Cytoplasmic granules obtained from toad urinary bladder epithelial cells were brought to buoyancy in a linear sucrose gradient. The gradient was loaded either with untreated cytoplasmic granules, or with granules treated with Na pyrophosphate (PPi), with digitonin, or with PPi and digitonin in succession. The following enzymes were assayed in the gradient subfractions: oligomycin-insensitive Mg++-ATPase, alkaline phosphodiesterase I, alkaline phosphatase, acid N-acetyl-beta-glucosaminidase, cytochrome oxidase, nucleoside diphosphatase (substrate, ADP), aminopeptidase (substrate, leucyl-beta-naphthylamide), and mannosyltransferase (acceptor, dolichylphosphate). Comparison of the density distributions of enzymes in untreated and treated preparations led to the characterization of 4 distinct subcellular entities. In agreement with the properties of mitochondria from other cell types, cytochrome oxidase buoys at 1.18 within a narrow density range and its behavior is not significantly altered by PPi or digitonin. Under all conditions, acid N-acetyl-beta-glucosaminidase is recovered over a broad density range in the lower part of the gradient and appears as a qualified lysosomal marker. Mg++-ATPase, alkaline phosphodiesterase I, and alkaline phosphatase belong to a group with the distinguishing features of a low equilibrium density in native cytoplasmic granules and a marked shift (+0.03 density units) after digitonin treatment. Such properties are typical of the plasma membranes. Part of the aminopeptidase activity probably also belongs to plasma membrane-derived elements. Minor differences between alkaline phosphatase and the other 2 members of that group make it possible that their distribution domains in the membrane do not overlap or coincide.(ABSTRACT TRUNCATED AT 250 WORDS)

Comparative study of the effect of chrysotile, quartz and rutile on the release of lymphocyte-activating factor (interleukin 1) by murine peritoneal macrophages in vitro.

The release of lymphocyte-activating factor (LAF) into the medium of cultured mouse resident peritoneal macrophages was estimated from the effect of this medium on the proliferation of mouse thymocytes in the presence of phytohaemagglutinin. Untriggered macrophages released little LAF into their culture medium. Upon addition of UICC chrysotile A (25-50 micrograms/10(6) macrophages), LAF appeared in the medium, and release continued for at least 20 h. DQ12 quartz was a more potent inducer of LAF production, while rutile was nearly inactive. Although chrysotile and quartz caused cell damage (as estimated by the release of lactate dehydrogenase), it was established that LAF release was not attributable to leakage of preformed intracellular LAF. The lymphoproliferative activity detected in macrophage media was stable at 56 degrees C and nondialysable, which corresponds to the properties of interleukin 1. These biochemical observations are consistent with the non-specific stimulation of the immune system found in asbestotic and silicotic patients.

Translocation and proteolytic processing of nascent secretory polypeptide chains: two functions associated with the ribosomal domain of the endoplasmic reticulum.

Rat liver microsomes were subfractionated by isopycnic centrifugation in sucrose gradient. The subfractions were assayed for translocation and proteolytic processing of nascent polypeptides in a rabbit reticulocyte lysate programmed with total RNA from human term placenta. The distribution of the translocation and processing of prelactogen through the gradient correlated with that of the microsomal RNA (ribosomes). Microsomes became inactive upon incubation with elastase, but the proteolyzed membranes recovered their activity by recombination with the soluble and active fragment of the docking protein (SRP-receptor) from dog pancreas. When this fragment was combined with the gradient subfractions, or with the subfractions inactivated by incubation with elastase, the density profile of the translocation activity remained similar to that of RNA. Thus, its distribution cannot be accounted for merely by that of the docking protein; another membrane constituent, still unidentified, is both necessary for translocation of polypeptides and restricted to the rough portions of the endosplamic reticulum. Signal peptidase was assayed in the absence of protein synthesis, by use of preformed prelactogen and detergent-disrupted microsomes. Its density distribution was also similar to that of RNA. Several components of the endosplamic reticulum now appear to be segregated within restricted areas on either side of the membrane, and to make up a biochemically distinct domain. We propose to call it the ribosomal domain in consideration of its contribution to protein biosynthesis by bound ribosomes. This domain probably accounts for a greater part of the membrane area at the cytoplasmic than at the luminal surface, as postulated earlier to explain how enzymes of the cytoplasmic surface are relatively less abundant in the rough microsomes than those of the luminal surface [Amar-Costesec A. & Beaufay H. (1981) J. Theor. Biol. 89, 217-230].

Immunochemical characterization and biosynthesis of pI-6.4 esterase, a carboxylesterase of rat liver microsomal extracts.

Rat liver pI-6.4 esterase was purified from microsomes (microsomal extracts) and used to generate antibodies in the rabbit. Two active enzyme forms, similarly sensitive to endo-H (endo-beta-N-acetylglucosaminidase H (EC 3.2.1.96), but differing slightly in polypeptide chain length, were present in the preparation. In microsomes, immunoblots revealed a single form, with Mr congruent to 62,000, identical with the large component of the purified enzyme, indicating that the second component is an artefact. Rabbit reticulocyte lysates and wheat germ extracts programmed with RNA extracted from total or bound polysomes synthesized a single immunoreactive 61 kDa polypeptide, which was not formed with RNA extracted from free polysomes. The immunoreactive product synthesized in the presence of dog pancreas microsomes was slightly larger (62 kDa); like the authentic enzyme, it bound to concanavalin A and was decreased in molecular size to 60 kDa by the action of endo-H. Thus the enzyme is synthesized with a short cleavable sequence and bears at least one high-mannose oligosaccharide chain. Metabolic labelling in hepatocytes cultured with [35S]methionine also generated a single immunoreactive polypeptide of 62 kDa, which was decreased to 60 kDa in size by treatment with endo-H or addition of tunicamycin to the culture medium. This confirms the molecular homogeneity and the glycosylation of the enzyme in the intact cell. Culture media contained no pI-6.4-esterase-related protein, whether tunicamycin was present or not. The processing steps in the synthesis of pI-6.4 esterase are thus, as for other esterases of the endoplasmic reticulum [Robbi & Beaufay (1986) Eur. J. Biochem. 158, 187-194; (1987) Biochem. J. 248, 545-550] indistinguishable from those occurring early in the synthesis of secretory proteins. Glycosylation is apparently not the sorting signal responsible for their retention in the endoplasmic reticulum.

Biosynthesis of rat liver pI-6.1 esterase, a carboxylesterase of the cisternal space of the endoplasmic reticulum.

Biosynthesis of the rat liver microsomal esterase with pI 6.1 was investigated in cell-free systems and in cultured hepatocytes, by using a rabbit antiserum. Protein synthesis directed by total rat liver RNA in wheatgerm extract or reticulocyte lysate generated a single immunoprecipitable product, also found with the RNA extracted from bound, but not from free, polysomes. When dog pancreas microsomal fractions were included, reticulocyte lysates gave two processed products, a prominent one slightly larger, and another slightly smaller, than the precursor, both resistant to exogenous proteinases and, hence, segregated within vesicles. The processing was co-translational; it consisted of the removal of a peptide fragment and, for the large component, the addition of a single oligosaccharide chain. Indeed, this component bound to concanavalin A-Sepharose and gave the small one (approximately 2000 Mr loss) by cleavage with endo-beta-N-acetylglucosaminidase H (endo-H). A single labelled peptide was precipitated from hepatocytes incubated with [35S]methionine. Its apparent Mr was decreased by approximately 2000 after treatment with endo-H; it was then identical with that of an unglycosylated form produced in hepatocytes poisoned with tunicamycin. Even in that case, immunoreactive peptides were not detected in the culture medium. Whether synthesized in reticulocyte lysate or in hepatocytes, the glycosylated forms migrated in SDS/polyacrylamide-gel electrophoresis as the purified enzyme labelled with [3H]di-isopropyl fluorophosphate. Thus, although pI-6.1 esterase is not secreted, its biosynthesis is, as yet, indistinguishable from that of secretory proteins. Its oligosaccharide moiety is apparently not the structural element that retains it in the endoplasmic reticulum.

The membrane of the rough endoplasmic reticulum contains cytoplasmically exposed high affinity GTP-binding sites.

Binding experiments using [14C]GTP and rat liver rough microsomes gave Scatchard plots with Kd congruent to 0.1 microM and a binding capacity congruent to 40 pmol/mg microsomal protein. Removal of the ribosomes did not modify the binding parameters. GTP was competed out by GTP analogues but not by ATP. Limited proteolysis of rough microsomes decreased the GTP-binding capacity and prevented GTP from suppressing the latency of mannose-6-phosphatase and of the synthesis of dolichol-linked chitobiose. The GTP-binding protein is probably involved in these effects of GTP. Its function could be to act in the association-dissociation of membrane components at the ribosome-membrane junction.

Biosynthesis of rat-liver pI-5.0 esterases in cell-free systems and in cultured hepatocytes.

Rat liver esterases focusing at pH 5.0 (referred to below as pI-5.0 esterases) are structurally related glycoproteins which differ slightly in their mobility in sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS-PAGE). They reside in the lumen of the endoplasmic reticulum. We have studied their biosynthesis in cell-free systems programmed by total liver RNA, using sheep and rabbit antibodies to isolate the translation products related to these enzymes. Our results show that they are assembled as a precursor polypeptide chain (62 kDa) larger than the mature proteins. The pI-5.0 esterase mRNA could be extracted from bound but not free polysomes. Reticulocyte lysates supplemented with dog pancreas microsomes produced four esterase-related components in segregated form (61, 60, 58 and 56 kDa). The largest three correspond in electrophoretic mobility to the mature enzymes. They are glycoproteins that bind to concanavalin A, and can be reduced to the size of the shortest component by endo-beta-N-acetylglucosaminidase H (endo-H). Immunoprecipitation after biosynthetic labeling of the proteins in cultured hepatocytes also gave three glycosylated components that had the same mobility in SDS-PAGE as the mature enzymes. When tunicamycin was present in the culture medium, a single immunoprecipitable form was observed. Its apparent Mr was similar to that of the unglycosylated pI 5.0 esterase form synthesized in vitro in the presence of dog pancreas microsomes. Thus the biosynthesis of these esterases has characteristics in common with that of numerous secretory proteins, except for the rather large difference in size (approximately equal to 6 kDa) resulting from the proteolytic processing of their in-vitro-synthesized precursor.

Analytical study of microsomes and isolated subcellular membranes from rat liver. IX. Nicotinamide adenine dinucleotide glycohydrolase: a plasma membrane enzyme prominently found in Kupffer cells.

The distribution of nicotinamide adenine dinucleotide (NAD) glycohydrolase in rat liver was investigated by subcellular fractionation and by isolation of hepatocytes and sinusoidal cells. The behavior of NAD glycohydrolase in subcellular fractionation was peculiar because, although the enzyme was mainly microsomal, plasma membrane preparations contained distinctly more NAD glycohydrolase than could be accounted for by their content in elements derived from the endoplasmic reticulum or the Golgi complex identified by glucose-6-phosphatase and galactosyltransferase, respectively. When microsomal and plasmalemmal preparations were brought to equilibrium in a linear-density gradient, NAD glycohydrolase differed from these enzymes and behaved like 5'-nucleotidase and alkaline phosphodiesterase I. NAD glycohydrolase was markedly displaced towards higher densities after treatment with digitonin. This behavior in density-gradient centrifugation strongly suggests that NAD glycohydrolase is an exclusive enzyme of the plasma membrane. NAD glycohydrolase differed clearly from other plasmalemmal enzymes when the liver was fractionated into hepatocytes and sinusoidal cells; its specific activity was considerably greater in sinusoidal cell than in hepatocyte preparations. Further subfractionation of sinusoidal cell preparations into endothelial and Kupffer cells by counterflow elutriation showed that NAD glycohydrolase is more active in Kupffer cells. We estimate that the specific activity of NAD glycohydrolase activity is at least 65-fold higher at the periphery of Kupffer cells than at the periphery of hepatocytes. As the enzyme shows not structure-linked latency and is an exclusive constituent of the plasma membranes, we conclude that it is an ectoenzyme that cannot lead to a rapid turnover of the cytosolic pyridine nucleotides.