Structural comparison of highly similar nucleoside-diphosphate kinases: Molecular explanation of distinct membrane-binding behavior.

NDPK-A, NDPK-B and NDPK-D are three enzymes which belong to the NDPK group I isoforms and are not only involved in metabolism process but also in transcriptional regulation, DNA cleavage, histidine protein kinase activity and metastasis development. Those enzymes were reported to bind to membranes either in mitochondria where NDPK-D influences cardiolipin lateral organization and is thought to be involved in apoptotic pathway or in cytosol where NDPK-A and NDPK-B membrane association was shown to influence several cellular processes like endocytosis, cellular adhesion, ion transport, etc. However, despite numerous studies, the role of NDPK-membrane association and the molecular details of the binding process are still elusive. In the present work, a comparative study of the three NDPK isoforms allowed us to show that although membrane binding is a common feature of these enzymes, mechanisms differ at the molecular scale. NDPK-A was not able to bind to model membranes mimicking the inner leaflet of plasma membrane, suggesting that its in vivo membrane association is mediated by a non-lipidic partner or other partners than the studied phospholipids. On the contrary, NDPK-B and NDPK-D were shown to bind efficiently to liposomes mimicking plasma membrane and mitochondrial inner membrane respectively but details of the binding mechanism differ between the two enzymes as NDPK-B binding necessarily involved an anionic phospholipid partner while NDPK-D can bind either zwitterionic or anionic phospholipids. Although sharing similar secondary structure and homohexameric quaternary arrangement, tryptophan fluorescence revealed fine disparities in NDPK tertiary structures. Interfacial behavior as well as ANS fluorescence showed further dissimilarities between NDPK isoforms, notably the presence of distinct accessible hydrophobic areas as well as different capacity to form Gibbs monolayers related to their surface activity properties. Those distinct features may contribute to explain the differences in the protein behavior towards membrane binding.

New insights into lipid-Nucleoside Diphosphate Kinase-D interaction mechanism: protein structural changes and membrane reorganisation.

Nucleoside Diphosphate Kinases (NDPKs) have long been considered merely as housekeeping enzymes. The discovery of the NME1 gene, an anti-metastatic gene coding for NDPK-A, led the scientific community to re-evaluate their role in the cell. It is now well established that the NDPK family is more complex than what was first thought, and despite the increasing amount of evidence suggesting the multifunctional role of nm23/NDPKs, the specific functions of each family member are still elusive. Among these isoforms, NDPK-D is the only one to present a mitochondria-targeting sequence. It has recently been shown that this protein is able to bind and cross-link with mitochondrial membranes, suggesting that NDPK-D can mediate contact sites and contributes to the mitochondrial intermembrane space structuring. To better understand the influence of NDPK-D on mitochondrial lipid organisation, we analysed its behaviour in different lipid environments. We found that NDPK-D not only interacts with CL or anionic lipids, but is also able to bind in a non negligible manner to zwitterionic PC. NDPK-D alters membrane organisation in terms of fluidity, hydration and lipid clustering, effects which depend on lipid structure. Changes in the protein structure after lipid binding were evidenced, both by fluorescence and infrared spectroscopy, regardless of membrane composition. Taking into account all these elements, a putative mechanism of interaction between NDPK-D and zwitterionic or anionic lipids was proposed.

Integrity characterization of myoglobin released from poly(ε-caprolactone) microspheres using two analytical methods: UV/Vis spectrometry and conductometric bi-enzymatic biosensor.

Myoglobin (Mb)-loaded poly(ε-caprolactone) (PCL) microparticles were prepared by multiple emulsion with solvent extraction/evaporation method under more or less deleterious operating conditions. The protein integrity was monitored using both UV/Vis absorbance ratio method at specific wavelengths and a conductometric bi-enzymatic biosensor based on proteinase K and pronase. Under standard operating conditions, Mb remained in native conformation, while different degrees of protein denaturation were observed by changing the encapsulation conditions. It was shown that solvent elimination under reduced pressure and in a lower extent addition of a higher molecular weight PCL led to protein alteration. In the first case, the loss of protein integrity can be attributed to residual solvent entrapped in particles whose solidification was accelerated. In the second case, denaturation may be explained by an increase in the protein exposure time at water/organic solvent interface due to an increase in organic phase viscosity.

Magnesium-adenosine diphosphate binding sites in wild-type creatine kinase and in mutants: role of aromatic residues probed by Raman and infrared spectroscopies.

Two distinct methods were used to investigate the role of Trp residues during Mg-ADP binding to cytosolic creatine kinase (CK) from rabbit muscle: (1) Raman spectroscopy, which is very sensitive to the environment of aromatic side-chain residues, and (2) reaction-induced infrared difference spectroscopy (RIDS) and photolabile substrate (ADP[Et(PhNO(2))]), combined with site-directed mutagenesis on the four Trp residues of CK. Our Raman results indicated that the environment of Trp and of Tyr were not affected during Mg-ADP binding to CK. Analysis of RIDS of wild-type CK, inactive W227Y, and active W210,217,272Y mutants suggested that Trp227 was not involved in the stacking interactions. Results are consistent with Trp227 being essential to prevent water molecules from entering in the active site [as suggested by Gross, M., Furter-Graves, E. M., Wallimann, T., Eppenberger, H. M., and Furter, R. (1994) Protein Sci. 3, 1058-1068] and that another Trp could in addition help to steer the nucleotide in the binding site, although it is not essential for the activity of CK. Raman and infrared spectra indicated that Mg-ADP binding does not involve large secondary structure changes. Only 3-4 residues absorbing in the amide I region are directly implicated in the Mg-ADP binding (corresponding to secondary structure changes less than 1%), suggesting that movement of protein domains due to Mg-nucleotide binding do not promote large secondary structure changes.

Cloning, Escherichia coli expression, and phase-transition chromatography-based purification of recombinant rabbit heart mitochondrial creatine kinase.

A cDNA clone of the mitochondrial sarcomeric creatine kinase cDNA was obtained by screening a rabbit heart library. This cDNA is characterized by a 1257-nucleotide open reading frame encoding a 419-amino-acid protein with a cleavable 39-amino-acid mitochondrial presequence (Accession No. AJ011334). This new member of the guanidino kinase family shows a high degree of sequence similarity with the other phosphagen kinases sequenced so far. The mature enzyme was efficiently expressed in Escherichia coli BL21(DE3) cells as a soluble octameric protein using the pET21 plasmid and purified by a three-step improved method including a final phase-transition chromatography.

Evidence for kinetic intermediate states during the refolding of GdnHCl-denatured MM-creatine kinase. Characterization of a trapped monomeric species.

The kinetics of refolding of guanidinium chloride-denatured rabbit MM-creatine kinase was investigated. Recovery of enzymatic activity is biphasic, depending on the temperature but not on the protein or DTT concentration. Only 45% of the original, active dimeric form is recovered even after several hours of refolding. The reactivation yield is limited by the accumulation of a highly stable but nonproductive monomeric species. The ratio of "correct" to "incorrect" forms depends on the duration of exposure to the denaturant, which may be consistent with the existence of a heterogeneous population of unfolded states with regard to proline isomerization. The first fast reaction observed during renaturation results in the appearance of collapsed monomeric states, displaying features of a pre-molten globule state. These burst species are rapidly transformed into more structured monomers resembling a molten globule state possessing a partially folded C-terminal domain. A proportion of these latter transient intermediates (45%) associates into an active dimer, while the remainder (55%) is trapped by reshuffling in a monomeric dead-end product. Our results strongly indicate that (i) the dimeric state is a prerequisite for the expression of catalytic activity, (ii) the kinetic intermediates of refolding are very similar to those observed during equilibrium unfolding, and (iii) refolding of creatine kinase in these conditions is limited by the accumulation of inactive misfolded nondimerizable monomer.

Role of quaternary structure in muscle creatine kinase stability: tryptophan 210 is important for dimer cohesion.

A mutant of the dimeric rabbit muscle creatine kinase (MM-CK) in which tryptophan 210 was replaced has been studied to assess the role of this residue in dimer cohesion and the importance of the dimeric state for the native enzyme stability. Wild-type protein equilibrium unfolding induced by guanidine hydrochloride occurs through intermediate states with formation of a molten globule and a premolten globule. Unlike the wild-type enzyme, the mutant inactivates at lower denaturant concentration and the loss of enzymatic activity is accompanied by the dissociation of the dimer into two apparently compact monomers. However, the Stokes radius of the monomer increases with denaturant concentration as determined by size exclusion chromatography, indicating that, upon monomerization, the protein structure is destabilized. Binding of 8-anilinonaphthalene-1-sulfonate shows that the dissociated monomer exposes hydrophobic patches at its surface, suggesting that it could be a molten globule. At higher denaturant concentrations, both wild-type and mutant follow similar denaturation pathways with formation of a premolten globule around 1.5-M guanidine, indicating that tryptophan 210 does not contribute to a large extent to the monomer conformational stability, which may be ensured in the dimeric state through quaternary interactions.

Involvement of a tyrosine residue in the ADP binding site of creatine kinase. A second-derivative UV-spectroscopy study.

Second-derivative spectroscopy was used to determine the percentage of tyrosine residues that are exposed to solvent in rabbit MM-creatine kinase. Six residues, among the ten present per monomer, are solvent-exposed. The presence of creatine in the incubation medium does not modify this value. However, this number is decreased by one when the enzyme is incubated with saturating concentrations of MgADP. A dissociation constant for MgADP can be estimated and the obtained value (0.085 mM) is comparable to the Km for this substrate. Thus, a tyrosine residue is located near the MgADP binding site or is masked during protein conformational change induced by adenyl nucleotide binding.

Denaturation by guanidinium chloride of dimeric MM-creatine kinase and its proteinase K-nicked form: evidence for a multiple-step process.

Cytosolic MM-creatine kinase is a homodimeric protein. Each monomer can be cleaved by proteinase K at an exposed surface loop, into two fragments K1 and K2, which remain associated. The nicked protein is thus a heterotetrameric protein, named (K1K2)2, made up of two heterodimers K1K2 linked together by their K1 subunit. In non-denaturing conditions, the cleaved protein does not present any measurable difference compared with uncleaved MM-creatine kinase, except for the loss of enzymatic activity. Comparative equilibrium denaturation of the two oligomeric proteins by guanidinium chloride indicates a multistep process with formation of either compact monomer or compact K1K2 dimer, a molten globule and a pre-molten globule state. In the case of the nicked-enzyme, the molten globule is composed of the two peptides K1 and K2, whereas in the pre-molten globule the interactions between K1 and K2 are too weak to maintain their cohesion. At low guanidinium chloride concentration, the proteinase K-nicked protein exhibits a higher accessibility of one of its tryptophan accompanied by a small decrease in its molar ellipticity suggesting a secondary structure loosening of the K1 peptide. Our results suggest that K1 and K2 are not strictly autonomous unfolding units and thus cannot be considered as independent domains.

Proteinase K processing of rabbit muscle creatine kinase.

Proteinase K cleaves selectively both cytosolic and mitochondrial isoforms of creatine kinase leading to the appearance of two fragments, a large N-terminal one (K1) and a small C-terminal peptide (K2) which remain associated together. The loss of enzymatic activity correlates with the extent of monomer cleavage. N-terminal sequencing of the K2 fragments from rabbit cytosolic and pig mitochondrial creatine kinase shows that these peptides begin with A328 and A324, respectively. Electrospray ionization mass spectrometry demonstrates that K2 peptide is composed of 53 residues (A328-K380). However, the C-terminal end of the K1 fragment is not A327 as expected, but D325. Thus, the amino acids residues T326 and A327 have been eliminated by the protease.

Nucleotide binding sites in wild-type creatine kinase and in W227Y mutant probed by photochemical release of nucleotides and infrared difference spectroscopy.

Structural changes induced by nucleotide binding to the wild-type rabbit muscle creatine kinase (CK) and to its W227Y mutant were compared and probed by reaction-induced difference spectroscopy (RIDS). The reaction was induced by the photorelease of nucleotide from the caged nucleotides ADP[Et(PhNO2)] or ATP[Et(PhNO2)], producing the RIDS of CK. The concomitant addition of a saturated concentration of nucleotide and caged nucleotide modified the RIDS of CK, permitting structural changes caused by nucleotide binding in the wild-type creatine kinase to be identified. The W227Y mutant was inactive and its nucleotide binding site was partially impaired as shown by the disappearance or decrease of several nucleotide-sensitive bands in the RIDS of W227Y mutant. The magnitude of the decrease was not the same for each band, suggesting that distinct groups of W227Y mutant were affected differently during nucleotide binding. More precisely, the binding sites for gamma-phosphate and beta-phosphate of the nucleotide were not accessible in W227Y mutant as shown by the absence of the phosphate-sensitive 1666-1667-cm(-1) and 1625-cm(-1) bands in the RIDS of W227Y mutant. However the binding site of other parts of the nucleotide was partially accessible, since the 1638-1639-cm(-1) phosphate-insensitive band did not completely vanish in the RIDS of W227Y mutant. The RIDS of W227Y mutant with ADP[Et(PhNO2)] and creatine lacked the 1613-cm(-1) and 1581-cm(-1) bands, associated with vibrational modes of creatine, suggesting that coupling between the binding sites of the nucleotide and of creatine was altered in W227Y mutant. These results are in accordance with the earlier suggestions that residue W227 in CK is essential for preventing water molecules from penetrating into the active site and for orienting nucleotide in the binding site, by forming stacking interactions between its indole group and purine of the nucleotide and its indole group.

Glucose-dependent and -independent effect of insulin on gene expression.

Insulin action on gene expression can be glucose-dependent or -independent. Accumulation of aldolase B mRNA and of an unidentified 5.4-kilobase mRNA as well as accumulation of L-type pyruvate kinase mRNAs (Decaux, J.F., Antoine, B., and Kahn, A. (1989) J. Biol. Chem. 264, 11584-11590) in cultured hepatocytes isolated from fasted rats require the presence of both glucose and insulin, these agents not being effective individually. In contrast, maintaining the amount of albumin and transferrin mRNAs in these hepatocytes requires the presence of insulin alone, glucose having no effect by itself. Transcription of the albumin gene, investigated by run-on assay, is active in the presence of insulin alone, with or without glucose, whereas transcription of the aldolase B gene is stimulated by glucose and insulin together, but not by insulin or glucose alone. In addition, the stability of the albumin and aldolase B mRNAs in cultured hepatocytes is lowered in the absence of glucose and insulin together as compared to the stability in the presence of one or both agents. These results confirm that transduction of the insulin signal occurs via distinct pathways; one of these pathways could involve a secondary insulin-dependent modification of metabolite concentration, whereas other pathways could be more directly related to the activity(ies) of the occupied insulin receptor.

The oxidative inactivation of mitochondrial electron transport chain components and ATPase.

Bovine heart submitochondrial particles (SMP) were exposed to continuous fluxes of hydroxyl radical (.OH) alone, superoxide anion radical (O2-) alone, or mixtures of .OH and O2-, by gamma radiolysis in the presence of 100% N2O (.OH exposure), 100% O2 + formate (O2- exposure), or 100% O2 alone (.OH + O2- exposure). Hydrogen peroxide effects were studied by addition of pure H2O2. NADH dehydrogenase, NADH oxidase, succinate dehydrogenase, succinate oxidase, and ATPase activities (Vmax) were rapidly inactivated by .OH (10% inactivation at 15-40 nmol of .OH/mg of SMP protein, 50-90% inactivation at 600 nmol of .OH/mg of SMP protein) and by .OH + O2- (10% inactivation at 20-80 nmol of .OH + O2-/mg of SMP protein, 45-75% inactivation at 600 nmol of .OH + O2-/mg of SMP protein). Importantly, O2- was a highly efficient inactivator of NADH dehydrogenase, NADH oxidase, and ATPase (10% inactivation at 20-50 nmol of O2-/mg of SMP protein, 40% inactivation at 600 nmol of O2-/mg of SMP protein), a mildly efficient inactivator of succinate dehydrogenase (10% inactivation at 150 nmol of O2-/mg of SMP protein, 30% inactivation at 600 nmol of O2-/mg of SMP protein), and a poor inactivator of succinate oxidase (less than 10% inactivation at 600 nmol of O2-/mg of SMP protein). H2O2 partially inactivated NADH dehydrogenase, NADH oxidase, and cytochrome oxidase, but even 10% loss of these activities required at least 500-600 nmol of H2O2/mg of SMP protein. Cytochrome oxidase activity (oxygen consumption supported by ascorbate + N,N,N',N'-tetramethyl-p-phenylenediamine) was remarkably resistant to oxidative inactivation, with less than 20% loss of activity evident even at .OH, O2-, OH + O2-, or H2O2 concentrations of 600 nmol/mg of SMP protein. Cytochrome c oxidase activity, however (oxidation of, added, ferrocytochrome c), exhibited more than a 40% inactivation at 600 nmol of .OH/mg of SMP protein. The .OH-dependent inactivations reported above were largely inhibitable by the .OH scavenger mannitol. In contrast, the O2(-)-dependent inactivations were inhibited by active superoxide dismutase, but not by denatured superoxide dismutase or catalase. Membrane lipid peroxidation was evident with .OH exposure but could be prevented by various lipid-soluble antioxidants which did not protect enzymatic activities at all.(ABSTRACT TRUNCATED AT 400 WORDS)

Oxidative and non-oxidative mechanisms in the inactivation of cardiac mitochondrial electron transport chain components by doxorubicin.

The quinonoid anthracycline, doxorubicin (Adriamycin) is a potent anti-neoplastic agent whose clinical use is limited by severe cardiotoxicity. Mitochondrial damage is a major component of this cardiotoxicity, and rival oxidative and non-oxidative mechanisms for inactivation of the electron transport chain have been proposed. Using bovine heart submitochondrial preparations (SMP) we have now found that both oxidative and non-oxidative mechanisms occur in vitro, depending solely on the concentration of doxorubicin employed. Redox cycling of doxorubicin by Complex I of the respiratory chain (which generates doxorubicin semiquinone radicals, O2-, H2O2, and .OH) caused a 70% decrease in the Vmax. for NADH dehydrogenase during 15 min incubation of SMP, and an 80% decrease in NADH oxidase activity after 2 h incubation. This inactivation required only 25-50 microM-doxorubicin and represents true oxidative damage, since both NADH (for doxorubicin redox cycling) and oxygen were obligatory participants. The damage appears localized between the NADH dehydrogenase flavin (site of doxorubicin reduction) and iron-sulphur centre N-1. Succinate dehydrogenase, succinate oxidase, and cytochrome c oxidase activities were strongly inhibited by higher doxorubicin concentrations, but this phenomenon did not involve doxorubicin redox cycling (no NADH or oxygen requirement). Doxorubicin concentrations of 0.5 mM were required for 50% decreases in these activities, except for cytochrome c oxidase which was only 30% inhibited following incubation with even 1.0 mM-doxorubicin. Our results indicate that low concentrations of doxorubicin (50 microM or less) can catalyse a site-specific oxidative damage to the NADH oxidation pathway. In contrast, ten-fold higher doxorubicin concentrations (or more) are required for non-oxidative inactivation of the electron transport chain; probably via binding to cardiolipin and/or generalized membrane chaotropic effects. The development of agents to block doxorubicin toxicity in vivo will clearly require detailed clinical studies of doxorubicin uptake in the heart.

Isoelectric point heterogeneity of the two oligomeric forms of heart mitochondrial creatine kinase.

Rabbit heart mitochondrial creatine kinase has been recently shown to exist in two oligomeric forms: a dimer and an octamer, the latter being the form associated with the inner mitochondrial membrane [(1988) Biochem.Biophys. Res. Commun. 153,1310.]. We report here on the determination of the isoelectric points (pI) of the two purified forms by thin layer isoelectric focusing. The pI of the dimer is 8.2 and that of the octamer is 8.8; the former is higher by more than one pH unit than that of the cytoplasmic form MM-CK. It is proposed that the higher pI of the octamer is responsible for its binding to the inner membrane.

Mitochondria contain a proteolytic system which can recognize and degrade oxidatively-denatured proteins.

When incubated with mitochondria in an air atmosphere, menadione and doxorubicin (which redox cycle with the respiratory chain to produce oxygen radicals), as well as xanthine oxidase plus xanthine (which generate superoxide and H2O2), stimulated the degradation of newly-synthesized [( 3H]leucine-labelled) mitochondrial polypeptides. No stimulation was observed in an N2 atmosphere, ATP was not required, and xanthine oxidase was not effective without xanthine. Various forms of oxidative stress induced varying degrees of protein cross-linking, protein fragmentation and proteolysis, as judged by gel electrophoresis and amino acid analysis. To learn more about the proteolytic enzymes involved in degradation, we undertook studies with purified protein substrates which had been exposed to oxidative stress (OH or H2O2) in vitro. Despite mitochondrial contamination with acid proteases of lysosomal (and other) origin, pH profiles revealed distinct proteolytic activities at both pH 4 and pH 8. The pH 8 activity preferentially degraded the oxidatively-denatured forms of haemoglobin, albumin and superoxide dismutase; was unaffected by digitonin; and exhibited a several-fold increase in activity upon mitochondrial disruption (highest activity being found in the matrix). In contrast, the pH 4 activity was dramatically decreased by digitonin treatment (to reduce lysosomal contamination); was unaffected by mitochondrial disruption; and showed no preference for oxidatively-denatured proteins. The pH 8 activity was not stimulated by ATP, but was inhibited by EDTA, haemin and phenylmethylsulphonyl fluoride. In contrast, the contaminating pH 4 activity was only inhibited by pepstatin and leupeptin. Thus, our experiments reveal a distinct mitochondrial (matrix) proteolytic pathway which can preferentially degrade oxidatively-denatured proteins.

Only one of the two interconvertible forms of mitochondrial creatine kinase binds to heart mitoplasts.

When analyzed by cellulose acetate electrophoresis, solubilized pig or rabbit heart mitochondrial creatine kinase is shown to exist under two distinct forms. The less cathodic one (form 1) is a dimer and the other having a higher cathodic mobility (form 2) has a molecular weight of about 350,000. The latter form can be converted into the former by incubation at alkaline pH or when the enzyme forms a reactive or an abortive complex with its substrates. This conversion is a reversible phenomenon and is not due to proteolysis. When rabbit heart mitoplasts are treated with the creatine kinase releasing agents, the enzyme is always solubilized as its form 2 and conversion to form 1, when it occurs, always take place after solubilization. Form 2 is also the only form which can be bound to pig or rabbit mitoplasts. Thus form 2 may be the actual form associated with heart mitochondria in vivo.

Interaction of creatine kinase with phosphorylating rabbit heart mitochondria and mitoplasts.

This paper demonstrates that the mitochondrial isoenzyme of creatine kinase (CKm) can be solubilized from rabbit heart mitochondria, the outer membrane of which has been removed or at least broken by a digitonin treatment or a short hypotonic exposure, but which has retained an important part of the capacity to phosphorylate ADP. Phosphate, ADP, or ATP, at concentrations which are used to study oxidative phosphorylation and creatine phosphate synthesis, solubilize CKm; the same is true with MgCl2 and KCl. The effect of adenine nucleotides does not seem to be due to their interaction with the adenine nucleotide translocase. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis shows that CKm is the main protein released in the described conditions; however, it does not amount to more than 1% of the total protein content of the mitoplasts. When the apparent Km for ATP of CKm was estimated by measuring creatine phosphate synthesis, the values obtained using water-treated mitochondria (0.21 mM) were slightly higher than those of intact mitochondria (0.12 mM) but the difference was not significant. In the former preparation 77% of CKm was in a soluble state. If we can extrapolate these results to intact mitochondria and suppose that in this case a fraction of CKm is also soluble in the intermembrane space, this does not support the theory of functional association between CKm and the adenine nucleotide translocase.