Purification and characterization of putative endothelin converting enzyme in bovine adrenal medulla: evidence for a cathepsin D-like enzyme

A specific and sensitive assay has been established for measurement of endothelin converting activity in a tissue extract. This assay is based on measuring endothelin-1 generated from big endothelin-1 by endothelin converting enzyme (ECE) with radioimmunoassay using an endothelin C-terminal specific antibody. By using this assay, we purified and characterized ECE in bovine adrenomedullary chromaffin granules ECE was purified over 3,000 times by a combination of DEAE, hydrophobic and gel filtration chromatography. A molecular weight of ECE was estimated to be approximately 30,000 by gel filtration. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed that ECE had three major components with estimated molecular weights of 45,000, 30,000 and 15,000 like bovine spleen cathepsin D. ECE had a pH optimum at 3.5 and was inhibited by pepstatin. These results strongly suggest that ECE is a cathepsin D-like aspartic protease.     

The pH dependence of breakdown of various purified brain proteins by cathepsin D preparations

In a continuing study of control processes of cerebral protein catabolism we compared the activity of cathepsin D from three sources (rat brain, bovine brain, and bovine spleen) on purified CNS proteins (tubulin, actin, calmodulin, S-100 and glial fibrillary acidic protein). The pH optimum was 5 for hydrolysis with tubulin as substrate for all three enzyme preparations, and it was pH 4 with the other substrates. The pH dependence curve was somewhat variable, with S-100 breakdown relatively more active at an acidic pH range. The formation of initial breakdown products and the further catabolism of the breakdown products was dependent on pH; hence the pattern of peptides formed from glial fibrillary acidic protein was different in incubations at different pH's. The relative activity of the enzyme preparations differed, depending on the substrate: with tubulin and S-100 as substrates, rat brain cathepsin D was the most active and the bovine spleen enzyme was the least active. With calmodulin and glial fibrillary acidic protein as substrates, rat brain and spleen cathepsin D activities were similar, and bovine brain cathepsin D showed the lowest activity. Actin breakdown fell between these two patterns.The rates of breakdown of the substrates were different; expressed as μg of substrate split per unit enzyme per h, with rat brain cathepsin D activity was 8–9 with calmodulin and S-100, 4 with glial fibrillary acidic protein, 1.8 with actin, and 0.9 with tubulin. The results show that there are differences in the properties of a protease like cathepsin D, depending on its source; furthermore, the rate of breakdown and the characteristics of breakdown are also dependent on the substrate.We recently measured the breakdown of brain tubulin by cerebral cathepsin D in a continuing study of the mechanisms and controls of cerebral protein catabolism (Bracco et al., 1982a). We found that tubulin breakdown is heterogeneous, that membrane-bound tubulin is resistant to cathepsin D but susceptible to thrombin (Bracco et al., 1982b), and that cytoplasmic tubulin was in at least two pools, one with a higher, another with a lower, rate of breakdown. The pH optimum of tubulin breakdown by cerebral cathepsin D differed significantly from the pH optimum of hemoglobin breakdown by the same enzyme.These findings showed that the properties of breakdown by a cerebral protease depend on the substrate. To further examine this dependence of properties of breakdown on the substrate, we now report measurements of pH dependence of breakdown of several purified proteins (tubulin, actin, calmodulin, S-100, glial fibrillary acidic protein [GFA]) from brain by cathepsin D preparations from three sources, rat brain, bovine brain, and bovine spleen. We also compare the rate of breakdown of the various proteins with the rate of hemoglobin breakdown.     

Cathepsin E from rat neutrophils: its properties and possible relations to cathepsin D-like and cathepsin E-like acid proteinases

An extract of rat neutrophils was found to contain a high hemoglobin-hydrolyzing activity at pH 3.2, about 70% of which does not cross-react with anti-rat liver cathepsin D antibody. A neutrophil non-cathepsin D acid proteinase was successfully isolated from cathepsin D and characterized in comparison with the properties of rat liver cathepsin D. The neutrophil enzyme differed from cathepsin D in chromatographic and electrophoretic behaviors as well as immunological cross-reactivity, and its molecular weight was estimated to be 98,000 by gel filtration on Toyopearl HW 55. These findings strongly suggest that the neutrophil enzyme could be classified as cathepsin E. The enzyme, now designated rat cathepsin E, had an optimal pH at 3.0-3.2, preferred hemoglobin to albumin as substrate, and was markedly resistant to urea denaturation. Rat cathepsins D and E cleaved the insulin B-chain at six and eight sites, respectively; five sites were common for both enzymes. Possible relations among cathepsin E and cathepsin D-like or E-like acid proteinases reported so far were discussed.     

Isolation and characterization of cathepsin B from bovine brain

Cathepsin B (EC 3.4.22.1) was purified 746-fold with a 21% recovery from bovine brain by autolysis, fractional precipitation with acetone, carboxy-methyl-Sephadex chromatography, affinity chromatography on a cystamine containing column and gel filtration chromatography. The purified cathepsin B eluted on gel filtration with an apparent molecular weight of 27,000 but was resolved into three bands of 30,000, 25,000 and 5,000 molecular weight by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (SDS-PAGE). Antibodies to cathepsin B, raised against the 30,000 dalton band, were shown by immunoblots to react with both the 30,000 and 25,000 dalton proteins with results suggesting that the former predominated as the immunoreactive form in bovine brain homogenates. Isoelectric focusing demonstrated multiple bands, ranging from pH 4.75–5.2 with the major band at pH 5.1–5.2, all of which were capable of degrading N-carbobenzoxy-l-arginyl-l-arginine 4-methoxy--naphthylamide. The cathepsin B activity against N-benzoyl-dl-arginine -naphthylamide (BANA) and bovine myelin basic protein (MBP) had a pH optimum of pH 6.0. The Km for the degradation of BANA was 1.0 mM and 5.1 mM when assayed in the presence of 1% and 2.5% dimethylsulfoxide, respectively. Cathepsin B from bovine brain has many properties similar to cathepsin B isolated from other organs. The degradative effect of cathepsin B on MBP suggests a role for this proteinase in inflammatory demyelination.     

Purification of an endopeptidase from bovine adrenal medulla granules which cleaves in vitro at paired but not at single basic residues

An endopeptidase was isolated from bovine adrenomedullary granules by fast protein liquid chromatography, including two ion exchange, one hydrophobic interaction, and one gel filtration step. The enzyme assay was based on the HPLC detection of the degradation of the dodecapeptide BAM 12P from the sequence of proenkephalin. After a 1200-fold purification, the enzyme fraction was homogeneous on polyacrylamide gel electrophoresis. The apparent molecular weight of the monomeric protein was 68,000 and its pH optimum was 5.6, in agreement with the internal pH of the granules. The specificity of the protease was determined initially by analysis of the degradation fragments of BAM 12P which showed that cleavage occurred at the double but not at the single Arg site of BAM 12P. The cleavage pattern of other substrates showed that the enzyme also recognized other pairs of basic amino acids. The behavior of the enzyme toward specific inhibitors showed that it was an acid thiol protease different from serine proteases and from lysosomal cathepsin B. The endopeptidase may act as a maturation enzyme in vivo.     

Unique cathepsin D-type proteases in rat thoracic duct lymphocytes and in rat lymphoid tissues.

Two unique cathepsin D-type proteases apparently present only in rat thoracic duct lymphocytes and in rat lymphoid tissues are described. One, termed H enzyme, has an apparent molecular weight of similar to95,000; the other, termed L enzyme, has an apparent molecular weight of similar to45,000, in common with that of most cathepsins D from other tissues and species. Both enzymes differ from cathepsin D, however, by a considerably greater sensitivity to inhibition by pepstatin and by a smaller degree of inhibition by an antiserum which inhibits rat liver cathepsin D. H enzyme is converted to L enzyme by treatment with beta-mercaptoethanol; the relationship between the two enzymes remains unknown. H and L enzyme have been detected in rat lymphoid tissues and in mouse spleen, but they are not present in other rat tissues (liver, kidney, adrenals), rabbit tissues, calf thymus, bovine spleen, or human tonsils. As measured on acid-denatured bovine hemoglobin as substrate, both enzymes have pH activity curves identical with that of rat liver cathepsin D, with optimal activity at pH 3.6. Activity on human serum albumin is much less and also shows an optimum at pH 3.6; hence, neither enzyme has the properties of cathepsin E. Thiol-reactive inhibitiors have no effect on the activity of H and L enzyme; thus they do not belong to the B group of cathepsins. Additional information, discussed in this paper, leads us to conclude that partially purified H and L enzymes are cathepsin D-type proteases.     

Characterization of endothelin converting enzyme activities in soluble fraction of bovine cultured endothelial cells

Endothelin converting enzyme activities in the soluble fraction of cultured bovine aortic endothelial cells were characterized. The two major endothelin converting enzyme activities were eluted from a hydrophobic chromatography column and the elution profile of the endothelin converting enzyme activities was the same as that of cathepsin D activities. These activities had a same pH optimum at pH 3.5 and were effectively inhibited by pepstatin A. Furthermore, anti-cathepsin D antiserum absorbed these activities as well as cathepsin D activity. Immunoblotting analysis using the antiserum showed the major active fractions have immunostainable components of identical molecular weights with cathepsin D. From these results, we concluded that the major endothelin converting activities in the soluble fraction of endothelial cells are due to cathepsin D. In addition to these cathepsin D activities, a minor endothelin converting enzyme activity with an optimum pH at 3.5 was found, which does not have angiotensin I generating (cathepsin D) activity from renin substrate and needs much higher concentrations of pepstatin A to inhibit the activity than cathepsin D.     

Purification and characterization of a lysosomal aspartic protease with cathepsin D activity from the mosquito

A lysosomal aspartic protease with cathepsin D activity, from the mosquito, Aedes aegypti, was purified and characterized. Its isolation involved ammonium sulfate (30–50%) and acid (pH 2.5) precipitations of protein extracts from whole previtellogenic mosquitoes followed by cation exchange chromatography. Purity of the enzyme was monitored by SDS-PAGE and silver staining of the gels. The native molecular weight of the purified enzyme as determined by polyacrylamide gel electrophoresis under nondenaturing conditions was 80,000. SDS-PAGE resolved the enzyme into a single polypeptide with M_r = 40,000 suggesting that it exists as a homodimer in its non-denatured state. The pI of the purified enzyme was 5.4 as determined by isoelectric focusing gel electrophoresis. The purified enzyme exhibits properties characteristic of cathepsin D. It utilizes hemoglobin as a substrate and its activity is completely inhibited by pepstatin-A and 6M urea but not by 10 mM KCN. Optimal activity of the purified mosquito aspartic protease was obtained at pH 3.0 and 45°C. With hemoglobin as a substrate the enzyme had an apparent K_m of 4.2 μ M. Polyclonal antibodies to the purified enzyme were raised in rabbits. The specificity of the antibodies to the enzyme was verified by immunoblot analysis of crude mosquito extracts and the enzyme separated by both non-denaturing and SDS-PAGE. Density gradient centrifugation of organelles followed by enzymatic and immunoblot analyses demonstrated the lysosomal nature of the purified enzyme. The N-terminal amino acid sequence of the purified mosquito lysosomal protease (19 amino acids) has 74% identity with N-terminal amino acid sequence of porcine and human cathepsins D.     

Purification and partial characterization of rat brain acid proteinase (isorenin).

1. Isorenin was purified 2000-fold from rat brain by a simple 3-step procedure involving affinity chromatography on pepstatinyl-Sepharose, The preparation appears as a homogenous protein in analytical polyacrylamide gel electrophoresis. Sodium dodecyl sulfate gel electrophoresis indicated an apparent molecular weight of 45 000. Isoelectric focusing separated isoenzymes with isoelectric points at pH 5.45, 5.87, 6.16 and 7.05. 2. The enzyme generates antiotensin I from tetradecapeptide (pH optimum 4.7) and from sheep angiotensinogen (pH optima 3.9 and 5.5). The rate of angiotensin I formation from tetradecapeptide was 30 000 times higher than that from sheep angiotensinogen. The enzyme has acid protease activity at pH 3.2 with hemoglobin as the substrate and pepstatin is a potent inhibitor of the enzyme with a Ki of less than 10(-9) M. 3. The properties of the enzyme strongly suggest that it is identical with cathepsin D.     

Isolation, and catalytic and immunochemical properties of cathepsin D-like acid proteinase from rat erythrocytes

An erythrocyte membrane-associated cathepsin D-like acid proteinase, termed "EMAP," was purified to homogeneity from freshly collected rat blood in a yield of 60-65%. The molecular weight of the enzyme was determined to be 80,000-82,000 by Sephadex G-100 gel filtration. The enzyme was inhibited strongly by pepstatin and partially by HgCl2, Pb(NO3)2, and iodoacetic acid. The preferred substrate for the enzyme was hemoglobin. The enzyme also hydrolyzed serum albumin and casein, but to lesser extents, with an optimum pH of 3.5-4.0. However, it could not hydrolyze leucyl-2-naphthylamide, benzyloxycarbonyl-Phe-Arg-4-methyl-7-coumarylamide or other synthetic substrates at pH values ranging from 3.5 to 9.5. The enzyme was very similar to human EMAP in a number of enzymatic properties, whereas it differed from rat cathepsin D in several respects, such as pH stability, molecular weight, isoelectric point, and chromatographic properties. Immunologically, the enzyme cross-reacted with the rabbit antibody prepared against human EMAP. The patterns of immunoelectrophoresis, immunoblotting, and immunoprecipitation of the enzyme were remarkably similar, if not identical, to those of human EMAP. In contrast, rat EMAP showed no reaction with the rabbit antibody raised to rat spleen cathepsin D. These results indicate that EMAP is a unique cathepsin D-like acid proteinase different from ordinary cathepsin D.     

Purification and properties of cathepsin D from porcine spleen.

Cathepsin D was purified from porcine spleen to near homogeneity as determined by gel electrophoresis. The isolation scheme involved an acid precipitation of tissue extract, DEAE-cellulose and Sephadex G-200 chromatography, and isoelectric focusing. The end product represented about a 1000-fold purification and about a 10% recovery. The purified enzyme was the major isoenzyme, which represented 60% of cathepsin D present in porcine spleen. Two minor isoenzymes of cathepsin D were present in small amounts. The purified enzyme resembled porcine pepsin in molecular weight (35,000), amino acid composition, and inactivation by specific pepsin inactivators. The pH activity curve of the purified enzyme showed two optima near pH 3 and 4. The relative activities at these optimal pH values were affected by salt concentration. Experimental evidence indicated that the two-optima phenomenon is a property of a single enzyme species.     

Affinity purification and properties of cathepsin-E-like acid proteinase from rat spleen.

A unique acid proteinase different from cathepsin D was purified from rat spleen by a method involving precipitation at pH 3.5, affinity chromatography on pepstatin-Sepharose 4B and concanavalin A-Sepharose 4B, chromatography on Sephadex G-100 and DEAE-Sephacel, and isoelectric focusing. A purification of 4200-fold over the homogenate was achieved and the yield was 11%. The purified enzyme appeared to be homogeneous on electrophoresis in polyacrylamide gels. The isoelectric point of the enzyme was determined to be 4.1-4.4. The enzyme hydrolyzed hemoglobin with a pH optimum of about 3.1. The molecular weight of the enzyme was estimated to be about 90000 by gel filtration on Sephadex G-100. In sodium dodecylsulfate polyacrylamide gel electrophoresis, the purified enzyme showed a single protein band corresponding to a molecular weight of about 45000. The hydrolysis of bovine hemoglobin by the enzyme was much higher than that of serum albumin. Various synthetic and natural inhibitors of the enzyme were tested. The enzyme was inhbited by Zn2+, Fe3+, Pb2+, cyanide, p-chloromercuribenzoate, iodoacetic acid and pepstatin, whereas 2-mercaptoethanol, phenylmethyl-sulfonyl fluoride and leupeptin showed no effect.     

Purification and characterization of collagenolytic property of renal cathepsin L from arthritic rat.

1. This paper describes the purification and characterization of collagenolytic property of renal cathepsin L isolated from kidney of rats rendered adjuvant arthritis. The enzyme was isolated by acid extraction, ammonium sulfate fractionation, Sephadex gel filtration, CM-Sephadex chromatography and Sephacryl S-300 chromatography. 2. The enzyme preparation was found to be homogeneous by gel filtration and SDS-polyacrylamide gel electrophoresis. The molecular weight of the enzyme was estimated to be 29,000. 3. Incubation of rat tail tendon collagen with purified cathepsin L resulted a conversion of cross-linked beta-chain dimers into uncross-linked alpha-chain monomers. The pH optimum for collagen degradation by purified cathepsin L was found to be 3.5. This optimal pH is shifted to 4.5 when haemoglobin was used as a substrate for the enzyme. 4. Various activators and inhibitors were tested for their influence on the activity of cathepsin L. The purified enzyme showed a maximal activity in the presence of EDTA. Cysteine was also found to increase the activity of cathepsin L. This enzyme was strongly inhibited by iodoacetate, p-chloromercurobenzoate, mercuric chloride but not inhibited by pepstatin or PMSF. E-64 and leupeptin were also found to be strong inhibitors for cathepsin L. The degradation of rat tail tendon collagen by cathepsin L was completely inhibited by E-64. 5. The results presented in this investigation suggest that cathepsin L play a crucial role in the pathogenesis of adjuvant arthritis.     

Cathepsin D from Atlantic cod (Gadus morhua L.) liver. Isolation and comparative studies

The isolated cathepsin D-like enzyme from Atlantic cod (Gadus morhua L.) liver was shown to be a monomer with a molecular mass of approximately 40 kDa. It was inhibited by Pepstatin A and had an optimum for degradation of haemoglobin at pH 3.0. The purified enzyme had lower temperature stability than bovine cathepsin D. Antibodies raised against the purified enzyme and against two C-terminal peptides of cod cathepsin D recognized a 40 kDa protein in immunoblotting of the samples from the purification process. Both antisera showed cross reactivity with a similar sized protein in liver from cod, saithe (Pollachius virens L.), Atlantic herring (Clupea harengus L.) and Atlantic salmon (Salmo salar L.). A protein of same size was detected in wolffish (Anarhichas lupus L.) liver with the antibody directed against the purified enzyme. This antibody also recognized the native enzyme and detected the presence of cathepsin D in muscle of cod, saithe, herring and salmon. These antibodies may be useful in understanding the mechanisms of post mortem muscle degradation in fish by comparing immunohistochemical localization and enzyme activity, in particular in cod with different rate of muscle degradation. They may also be used for comparing muscle degradation in different fish species.     

Cathepsin D. Purification of isoenzymes from human and chicken liver

1. The Barrett (1967) assay for cathepsin D was slightly modified. 2. The enzyme was purified from liver of man and chicken by a procedure involving autolysis, acetone fractionation, ion-exchange chromatography and isoelectric focusing. 3. Several isoenzymes of cathepsin D were resolved in the isoelectric-focusing step, and three major forms, alpha,beta and gamma, were distinguished for each species. 4. A modified analytical method of isoelectric focusing in polyacrylamide gel indicated a high degree of homogeneity of the purified beta and gamma isoenzymes from each species, and this was supported by their constant high specific activities. 5. Gel filtration of the isoenzymes in a calibrated column of Sephadex G-100 showed that each had a molecular weight of 45000. 6. Human cathepsin D had a pH optimum of 3.5, and that of chicken enzyme was 3.0, haemoglobin being used as substrate. In each species, the three isoenzymes have the same pH-dependence curve. 7. The purified cathepsin D samples showed very little action on acid-denatured albumin.     

Purification and certain properties of pyruvate decarboxylase from bovine brain]

A procedure of isolation and purification of pyruvate decarboxylase (PDC) from bovine brain is worked out. 350-fold purified enzyme preparation was homogenous under polyacrylamide gel electrophoresis. Molecular weight of PDC from bovine brain was estimated to be 180 000 by means of gel chromatography through Sephadex G-200. The protein was eluted in two peaks (with molecular weight of 180 000 and 90 000 respectively). After the treatment of the enzyme preparation with 6 M guanidine chloride. Probably, partial dissociation of the enzyme molecule into two subunits takes place in this case. Data on paper chromatography confirmed that highly purified PDC preparations from bovine brain were isolated as apoenzyme, since they were almost free of TPP.     

Cold-Adapted Digestive Aspartic Protease of the Clawed Lobsters Homarus americanus and Homarus gammarus: Biochemical Characterization

Aspartic proteinases in the gastric fluid of clawed lobsters Homarus americanus and Homarus gammarus were isolated to homogeneity by single-step pepstatin-A affinity chromatography; such enzymes have been previously identified as cathepsin D-like enzymes based on their deduced amino acid sequence. Here, we describe their biochemical characteristics; the properties of the lobster enzymes were compared with those of its homolog, bovine cathepsin D, and found to be unique in a number of ways. The lobster enzymes demonstrated hydrolytic activity against synthetic and natural substrates at a wider range of pH; they were more temperature-sensitive, showed no changes in the K _M value at 4°C, 10°C, and 25°C, and had 20-fold higher k _cat /K _M values than bovine enzyme. The bovine enzyme was temperature-dependent. We propose that both properties arose from an increase in molecular flexibility required to compensate for the reduction of reaction rates at low habitat temperatures. This is supported by the fast denaturation rates induced by temperature.     

Structures at the proteolytic processing region of cathepsin D

The amino acid sequences at the "proteolytic processing regions" of cathepsin Ds have been determined for the enzymes from cows, pigs, and rats in order to deduce the sites of cleavage as well as the function of the proteolytic processing of cathepsin D. For bovine cathepsin D, the "processing region" sequence was determined from a peptide isolated from the single-chain enzyme. The COOH-terminal sequence of the light chain and the NH2-terminal sequence of the heavy chain were also determined. The processing region sequence of porcine cathepsin D was determined from its cDNA structure, and the same structure from rat cathepsin D was determined from the peptide sequence of the single-chain rat enzyme. From sequence homology to other aspartic proteases whose x-ray crystallographic structures are known, such as pepsinogen and penicillopepsin, it is clear that the processing regions are insertions to form an extended beta-hairpin loop between residues 91 and 92 (porcine pepsin numbers). However, the sizes of the processing regions of cathepsin Ds from different species are considerably different. For the enzymes from rats, cows, pigs, and human, the sizes of the processing regions are 6, 9, 9, and 11 amino acid residues, respectively. The amino acid sequences within the processing regions are considerably different. In addition, the proteolytic processing sites were found to be completely different in the bovine and porcine cathepsin Ds. While in the porcine enzyme, an Asn-Ser bond and a Gly-Val bond are cleaved to release 5 residues as a consequence of the processing; in the bovine enzyme, two Ser-Ser bonds are cleaved to release 2 serine residues. These findings would argue that the in vivo proteolytic processing of the cathepsin D single chain is probably not carried out by a specific "processing protease." Model building of the cathepsin D processing region conformation was conducted utilizing the homology between procathepsin D and porcine pepsinogen. The beta-hairpin structure of the processing region was found to (i) interact with the activation peptide of the procathepsin D in a beta-structure and (ii) place the Cys residue in the processing region within disulfide linkage distance to Cys-27 of cathepsin D light chain. These observations support the view that the processing region of cathepsin D may function to stabilize the conformation of procathepsin D and may play a role in its activation.     

Breakdown of brain tubulin by cerebral cathepsin D

The breakdown of cytoplasmic tubulin from brain (purified by ammonium sulfate fractionation and DEAE cellulose chromatography) by cathepsin D from brain (purified by ammonium sulfate fractionation and pepstatin Sepharose chromatography) was studied; changes in the intensity of tubulin gel bands were determined. The pH optimum of hemoglobin breakdown by cathepsin D was 3.2; the pH optimum for tubulin breakdown was 5.8; at pH 5.8 there was no significant hemoglobin breakdown by the enzyme. Tubulin breakdown had an apparent K_m of 1.8 × 10⁻⁵ M and a V_max of 0.56 μg tubulin (μg enzyme per min). The rate of breakdown was heterogeneous and studied on length of incubation; the major portion of tubulin was rapidly broken down and a smaller portion was more stable. The rate under our experimental conditions was 18%/h in the 1–4 h period and 2%/h after 4 h. This was not due to enzyme instability: after 4 h of inhibition freshly added tubulin was rapidly broken down, whereas freshly added enzyme did not increase the rate of breakdown. Thus breakdown heterogeneity was due to substrate (tubulin) heterogeneity. Pepstatin inhibited cathepsin D breakdown of tubulin at acid pH; at pH 7.6 it had no effect. Leupeptin was not inhibitory. We calculated that the cathepsin D content in brain, if fully active, could break down cytoplasmic tubulin with a half-life of 24 h, but it is likely that under in vivo conditions enzyme activity is greatly modified.     

The microtubule-associated nucleoside diphosphate kinase

Microtubule protein prepared by cycles of assembly-disassembly contains a nucleoside diphosphate kinase (NDP kinase) activity. We have isolated the NDP kinase responsible for this activity from twice-polymerized bovine brain microtubule protein by a five-step chromatographic procedure. The molecular weight of this enzyme was 103,000 +/- 7,000 daltons as determined by sedimentation equilibrium experiments performed with a Beckman Airfuge. A doublet of subunit bands with molecular masses of about 18,000 daltons was detected by silver staining after gradient sodium dodecyl sulfate-polyacrylamide gel electrophoresis of this preparation. We conclude that the enzyme is a hexamer, although we cannot identify the mix of subunits. We were able to isolate only nanogram quantities of this enzyme, too little for extensive studies, so we isolated the enzyme directly from bovine brain without a preliminary microtubule protein isolation. The whole-brain NDP kinase was isolated by the same chromatographic steps as the enzyme from microtubule protein preparations. Both enzymes had a doublet of subunits at the same molecular weights and both were the same isozyme, chromatofocusing at a pH of 8.0. Both enzymes had similar kinetic properties and similar thermal inactivation profiles. These similar properties of the two enzymes suggest that they are identical. Both subunits of NDP kinase could be reversibly phosphorylated by ATP. Phosphorylation of the native enzyme created multiple, more acidic forms that retained activity. The isolation of this NDP kinase, which can copurify with microtubule protein through cycles of assembly-disassembly, will facilitate future studies on the role of this enzyme in the mechanism and regulation of microtubule assembly.