Reversibility of nucleoside diphosphate kinase solubilization from the surface of the outer mitochondrial membrane

It was found that in medium with low ionic strength nucleoside diphosphate kinase (NDPK) solubilization from the outer membrane of liver mitochondria could be partially reversed by the addition of 3.3 mM MgCl₂. Complete rebinding of the enzyme after the addition of MgCl₂ was observed when the mitochondrial washing and storage medium contained leupeptin, an inhibitor of cathepsins. It was demonstrated that leupeptin and another inhibitor of cysteine proteinases, E-64, do not influence the rate of NDPK solubilization as well as its solubilized and membrane-associated activity. We conclude that NDPK becomes sensitive to proteolysis only after its solubilization; proteolysis does not affect the part of the enzyme molecule that is responsible for catalysis. After solubilization of NDPK in the absence of leupeptin, cathepsins damage sites of its binding on the membranes. The rate of the enzyme solubilization is dependent on the pH of the storage medium (pH 6.0–8.0); it decreases with increase in pH. It was shown that in the medium with high ionic strength, MgCl₂ does not reverse pH-dependent NDPK solubilization, but solubilization could be reversed by increase in medium pH in the presence of E-64 and BSA. The physiological importance of these results is discussed. Published in Russian in Biokhimiya, 2009, Vol. 74, No. 5, pp. 710–720.     

Studies on the nature of adenosine diphosphatase activity from rat liver mitochondria

Adenosine diphosphatase (ADPase) activity was studied in rat liver with [beta-32P]ADP as a substrate. Mitochondria and outer mitochondrial membrane fractions were isolated and assayed for ADPase and various marker enzymes. ADPase activity was strikingly reduced when the outer membranes were removed from the mitochondria whether by digitonin treatment or osmotic shock. Addition of the inter-membrane space subfraction to the purified outer membranes resulted in enhanced ADPase activity. Addition of the inter-mitochondrial membrane enzyme adenylate kinase to outer membranes also produced a large stimulation of activity. The ADPase activity could also be reconstituted in vitro with adenylate kinase and either mitoplast ATPase or ouabain-sensitive (Na+ + K+ + Mg2+)-ATPase. Chloroform-released ATPase, however, was not capable of producing an ADPase activity when combined with adenylate kinase. Gel permeation chromatography of Triton-solubilised outer mitochondrial membranes was unable to resolve ADPase activity from contaminating ATPase. These results suggest that the majority of ADPase activity in rat liver mitochondria consists of the coupled activity of adenylate kinase and ATPase.     

Mitochondrial nucleoside diphosphate kinase: Mode of interaction with the outer mitochondrial membrane and proportion of catalytic activity functionally coupled to oxidative phosphorylation

In the present study, we found that ionic interactions are not essential for the binding of nucleoside diphosphate kinase of liver mitochondria outer compartment to outer mitochondrial membrane and that the proportion of the enzyme activity involved in functional coupling with oxidative phosphorylation (we demonstrated the existence of functional coupling earlier) is only 17%. Additional evidence was obtained that functionally coupled activity of nucleoside diphosphate kinase is associated with the outer surface of mitochondria. Dextran (10%) did not increase functional coupling. The physological importance of these effects is discussed. Published in Russian in Biokhimiya, 2008, Vol. 73, No. 3, pp. 395–407.     

Mitochondrial boundary membrane contact sites in brain: points of hexokinase and creatine kinase location, and control of Ca2+ transport

The location of hexokinase at the surface of brain mitochondria was investigated by electron microscopy using immuno-gold labelling techniques. The enzyme was located where the two mitochondrial limiting membranes were opposed and contact sites were possible. Disruption of the outer membrane by digitonin did not remove bound hexokinase and creatine kinase from brain mitochondria, although the activity of outer membrane markers and adenylate kinase decreased, suggesting a preferential location of both enzymes in the contact sites. In agreement with that, a membrane fraction was isolated from osmotically lysed rat brain mitochondria in which hexokinase and creatine kinase were concentrated. The density of this kinase-rich fraction was specifically increased by immuno-gold labelling of hexokinase, allowing a further purification by density gradient centrifugation. The fraction was composed of inner and outer limiting membrane components as shown by the specific marker enzymes, succinate dehydrogenase and NADH-cytochrome-c-oxidase (rotenone insensitive). As reported earlier for the enriched contact site fraction of liver mitochondria the fraction from brain mitochondria contained a high activity of glutathione transferase and a low cholesterol concentration. Moreover, the contacts showed a higher Ca2+ binding capacity in comparison to outer and inner membrane fractions. This finding may have regulatory implications because glucose phosphorylation via hexokinase activated the active Ca2+ uptake system and inhibited the passive efflux, resulting in an increase of intramitochondrial Ca2+.     

Activity of nucleoside diphosphate kinase α (NDPK α) capable of binding to outer mitochondrial membrane accounts for less than 10% of total NDPK activity present in cytoplasm of liver cells

During incubation of a constant volume of rat liver cytosol with an increasing quantity of mitochondrial protein in the presence of 3.3 mM MgCl₂, the binding of nucleoside diphosphate kinase (NDPK) from the cytosol to mitochondrial membranes is described by a saturation curve. The highest bound NDPK activity accounts for less than 9% of the added activity. Analysis of the results suggests that only one NDPK isozyme is bound to the membranes. Western blotting showed it to be NDPK α, a homolog of human NDPK-B. Substrates of NDPK, hexokinase, and glycerol kinase, as well as N,N’-dicyclohexylcarbodiimide and palmitate, did not influence the association of NDPK with mitochondrial membranes. We conclude that the sites of NDPK binding to the outer mitochondrial membrane are not identical to those of hexokinase and glycerol kinase.     

Modified properties of hexokinase from heart mitochondria prepared using proteolytic enzyme

Summary Isolation of muscle mitochondria is made easier by using proteolytic treatment of the tissue before homogenization. Normally, the proteolytic enzyme is discarded with the supernatant of the first centrifugation. However, our results show that a fraction of enzyme activity remains associated with mitochondria. As shown in experiments described in this paper, mitochondrial hexokinase from tissue treated or not with the proteolytic enzyme exhibits similar properties except that the solubilized enzyme from protease treated tissue is no longer able to rebind to mitochondrial membrane. This modification of the binding ability of the enzyme results from a partial hydrolysis of hexokinase during solubilization experiments by the proteolytic enzyme. Since, as pointed out here, proteolytic enzyme can remain associated with mitochondria, [either adsorbed on mitochondrial membrane or included in the mitochondrial pellet] its use for the isolation of muscle mitochondria should be avoided.     

Effects of SH group reagents on creatine kinase interaction with the mitochondrial membrane

Solubilization of the specific mitochondrial isoenzyme of creatine kinase (CKm) from rabbit heart mitochondria by treatment with SH group reagents has been studied. From the various compounds tested only the negatively charged organomercurials are able to induce an extensive solubilization of the enzyme. This effect is fully reversible since the solubilized enzyme readily reassociates with the membrane when the bound organomercurial is removed by treatment of the homogenate by an excess of dithiothreitol. Solubilization by negatively charged organomercurials can be partly prevented by pretreatment of mitochondria with either disulfide or uncharged organomercurials. No clear-cut relationship has been pointed out when the amount of SH titrated by various reagents has been compared with the extent of CKm solubilization. More detailed studies with para-chloromercuribenzoate (pCMB) show that extensive CKm solubilization (about 75%) occurs for pCMB concentration as low as 25 microM, whereas pronounced inhibition of the enzyme is observed only for concentrations greater than 200 microM. By cross-reassociation of enzyme solubilized either by para-hydroxymercuribenzoate (pHMB) or by 20 mM sodium phosphate (NaPi) with mitochondria depleted of CKm by pHMB or by NaPi treatment, SH groups whose titration impedes CKm reassociation with the mitochondrial membrane have been tentatively located on the enzyme. Thus, negatively charged organomercurials, could induce a reversible conformational modification of the enzyme which is no longer able to bind on the inner mitochondrial membrane. Furthermore, our results show that the binding of an excess of mitochondrial CK, which has been previously reported, could reflect unspecific binding since it occurs only on mitoplasts incubated in very hypotonic medium, but not in isotonic medium.     

Subcellular localization of 3-methylcrotonyl-coenzyme A carboxylase in bovine kidney

The intracellular localization of 3-methylcrotonyl-CoA carboxylase (MCase), an enzyme of the leucine oxidative pathway, was studied in bovine kidney. Differential centrifugation of kidney homogenates demonstrated that the majority of the enzyme was associated with the mitochondrial and cytosolic fractions. Isopycnic centrifugation of the mitochondrial fraction demonstrated cofractionation of MCase with mitochondrial markers, but not with lysosomal markers, consistent with a mitochondrial location for the enzyme. Using different homogenization techniques and comparing the fractional extraction of MCase and mitochondrial and cytosolic marker enzymes, the appearance of MCase in the “cytosolic” fraction was shown to be due to mitochondrial damage. The intramitochondrial distribution of MCase was determined using a digitonin procedure, and indicated that the enzyme is associated with the inner mitochondrial membrane. Although a fraction of MCase (30–40%) was “solubilized” by homogenization of whole mitochondria, the remaining MCase (60–70%) was tightly associated with the mitochondrial membrane fraction. Release and “solubilization” of this latter fraction was achieved by polyethylene glycol treatment. The “solubilized” MCase was stabilized in a glycerol-containing buffer.     

A Ca2+, Calmodulin-dependent NAD kinase from corn is located in the outer mitochondrial membrane

NAD kinase activity from dark grown corn coleoptiles is shown to be almost totally dependent on Ca2+ and calmodulin. Nearly all of the enzyme activity is found in a particulate fraction. Upon differential and density gradient centrifugation the NAD kinase activity co-migrates with the mitochondrial cytochrome c oxidase whereas marker activities for nuclei, etioplasts, endoplasmic reticulum, and microbodies could well be separated, indicating that the NAD kinase is associated with mitochondria. This NAD kinase, associated with intact mitochondria, can be activated by exogenously added Ca2+ and calmodulin. In order to investigate the submitochondrial localization of the NAD kinase, the organelles were ruptured by osmotic treatment and sonication and the submitochondrial fractions were separated by density gradient centrifugation. The NAD kinase activity exhibits the same density pattern as the antimycin A-insensitive NADH-dependent cytochrome c reductase, a marker enzyme of the outer mitochondrial membrane. Marker enzymes for the mitochondrial matrix and the inner mitochondrial membrane reveal different density profiles. These results indicate that the Ca2+, calmodulin-dependent NAD kinase from coleoptiles of dark grown corn seedlings is located at the outer mitochondrial membrane. The physiological relevance of the location and the Ca2+, calmodulin-dependence of the NAD kinase will be discussed.     

The intracellular distribution of adenylate kinase in the leaves of spinach,wheat and barley

Of the total adenylate-kinase activity in 10-d-old barley and wheat leaves, 40–50% is localised in the chloroplasts, while in mature spinach leaves 50–70% of the enzyme is chloroplastic. The extra-chloroplastic adenylate-kinase activity is associated with the mitochondria, very little, if any, is freely soluble in the cytoplasm. The adenylate pool of the cytoplasm could have access to adenylate-kinase activity in the intermitochondrial space because of the free permeation of adenylates across the outer mitochondrial membrane. Thus the adenylate pool of the cytoplasm could be subject to adenylate-kinase equilibrium. The mitochondrial adenylate kinase appeared to the localised exclusively in the intermembrane space.     

Functional Coupling between Nucleoside Diphosphate Kinase of the Outer Mitochondrial Compartment and Oxidative Phosphorylation

In rat liver mitochondria all nucleoside diphosphate kinase of the outer compartment is associated with the outer surface of the outer membrane (Lipskaya, T. Yu., and Plakida, K. N. (2003) Biochemistry (Moscow), 68, 1136-1144). In the present study, three systems operating as ADP donors for oxidative phosphorylation have been investigated. The outer membrane bound nucleoside diphosphate kinase was the first system tested. Two others employed yeast hexokinase and yeast nucleoside diphosphate kinase. The two enzymes exhibited the same activity but could not bind to mitochondrial membranes. In all three systems, muscle creatine phosphokinase was the external agent competing with the oxidative phosphorylation system for ADP. Determination of mitochondrial respiration rate in the presence of increasing quantities of creatine phosphokinase revealed that at large excess of creatine phosphokinase activity over other kinase activities (of the three systems tested) and oxidative phosphorylation the creatine phosphokinase reaction reached a quasi-equilibrium state. Under these conditions equilibrium concentrations of all creatine phosphokinase substrates were determined and K(eq)app of this reaction was calculated for the system with yeast hexokinase. In samples containing active mitochondrial nucleoside diphosphate kinase the concentrations of ATP, creatine, and phosphocreatine were determined and the quasi-equilibrium concentration of ADP was calculated using the K(eq)app value. At balance of quasi-equilibrium concentrations of ADP and ATP/ADP ratio the mitochondrial respiration rate in the system containing nucleoside diphosphate kinase was 21% of the respiration rate assayed in the absence of creatine phosphokinase; in the system containing yeast hexokinase this parameter was only 7% of the respiration rate assayed in the absence of creatine phosphokinase. Substitution of mitochondrial nucleoside diphosphate kinase with yeast nucleoside diphosphate kinase abolished this difference. It is concluded that oxidative phosphorylation is accompanied by appearance of functional coupling between mitochondrial nucleoside diphosphate kinase and the oxidative phosphorylation system. Possible mechanisms of this coupling are discussed.     

Identification and characterization of tyrosine kinase activity associated with mitochondrial outer membrane in sarcoma 180 cells

Tyrosine protein kinase activity has been detected in the mitochondrial fraction purified from sarcoma 180 tumor cells. Following hypotonic disruption of mitochondria, tyrosine kinase activity appeared to cosediment with monamine oxidase, marker enzyme of mitochondrial outer membrane; meanwhile, serine and threonine kinases were found to be associated with the inner membrane and matrix of mitochondria. Mitochondrial tyrosine kinase(s) showed thermosensitivity and Mn2+ dependence, useful properties for its characterization and separation from tyrosine kinases associated with other particulate fraction and from serine and threonine kinases associated with mitochondria. Following in vitro incubation of mitochondria with labelled ATP as substrate and analysis by PAGE, a complex pattern of phosphotyrosine containing proteins with a major band of 50-55 kilodaltons resulted.     

Solubilization and reassembly of the mitochondrial benzodiazepine receptor

We have solubilized and reassembled the peripheral-type benzodiazepine receptor, a component of the mitochondrial outer membrane, from rat adrenal gland mitochondria. The ligand binding site of this receptor undergoes denaturation during solubilization in digitonin, Triton X-100, or 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate at detergent concentrations above 0.1%, which is evident from the loss of high-affinity binding of [3H]PK11195, a ligand selective for the mitochondrial benzodiazepine receptor. The conformation of the binding site for PK11195 can be stabilized during solubilization in sodium cholate by relatively low concentrations of supplementary soybean lipid. Drug displacement studies demonstrate that the pharmacological properties of the receptor are preserved under these conditions. Electron micrographs of the solubilized preparation show a heterogeneous population of many small particles (less than 100 A) and some larger membranous aggregates (up to 500 A). Sucrose gradient centrifugation indicates that these lipoprotein complexes are of high buoyant density. They can be incorporated in liposomes via cholate dialysis in the presence of additional supplementary lipid. The behavior of the mitochondrial benzodiazepine receptor during solubilization and reassembly suggests that it is an integral protein of the outer membrane.     

Biochemical studies on sub-fractions of rat liver mitochondria

Highly purified mitochondria from rat liver were separated into six sub-fractions by differential centrifugation. The sub-fractions represent a spectrum from “heavy” to “very light” mitochondria. Enzymes representative of mitochondrial compartments were assayed to see whether functional differences occurred among the various mitochondrial sub-fractions. Respiratory control and NADH oxidase activity, both of which are indicators of mitochondrial structural integrity, were also measured. An enzyme marker for endoplasmic reticulum (glucose-6-phosphatase, G-6-Pase) was also assayed. Specific activities for monoamine oxidase (outer membrane marker), cytochrome oxidase (inner membrane marker) and malate-cytochrome c reductase did not vary within experimental error in all sub-fractions; similarly, for respiratory control and NADH oxidase activity. Malate dehydrogenase, a component of malate-cytochrome c reductase is located within the matrix surrounded by the inner membrane. Specific activity of adenylate kinase (located between the outer and inner membrane) decreased markedly from the “heavy” mitochondria to the “very light” fractions. Specific activity for G-6-Pase, very low in the “heavy” fractions, increased markedly in the “light” to “very light” fractions. Isopycnic density centrifugation on a linear sucrose density gradient of each of the fractions indicated that the correlation coefficient for the sucrose concentrations at which cytochrome oxidase and G-6-Pase activities peaked was 0.995. Thus the “light” to “very light” mitochondria may represent mitochondria whose outer membrane is still contiguous with the endoplasmic reticulum. Microsomes containing the endoplasmic reticulum peaked on the gradient at a significantly lower sucrose concentration than any of the mitochondrial sub-fractions. A buoyant effect of endoplasmic reticulum still attached to any of the mitochondrial sub-fractions would be expected to lower the density of attached mitochondria and thus give rise to “light” and “very light” mitochondria.     

The effect of lactoperoxidase-catalyzed iodination on the integrity of mitochondrial membranes.

The effect of lactoperoxidase-catalyzed iodination on rat liver mitochondria was investigated. A change from the condensed to the swollen conformation is observed by electron microscopy after extensive iodination of the mitochondria. The outer membrane breaks after incorporation of 0.2 nmol or more iodine atoms per mg of mitochondrial protein releasing adenylate kinase, a soluble enzyme located in the intermembrane space. Further iodination of the mitochondria ruptures the inner membrane, releasing proteins such as glutamic dehydrogenase from the matrix space. Lipid peroxides and I₂ are not intermediates in the disruptive effect of extensive lactoperoxidase-catalyzed iodination on the membranes. During iodination at pH 6.5 almost no release of protein or glutamic dehydrogenase activity is detectable and the loss of adenylate kinase activity from the particulate is diminished. The effect of extensive iodination on mitochondrial membranes limits the amount of iodide which can be incorporated with the lactoperoxidase membrane-labeling procedure when this technique is used as a surface probe of mitochondrial membranes.     

Structural changes of isolated hepatocytes during treatment with digitonin

The structural changes accompanying digitonin-induced release of enzymes and metabolites from isolated hepatocytes have been studied by scanning and transmission electron microscopy. In the initial phase, characterized by total release of the cytosolic marker enzyme, lactate dehydrogenase, the plasma membrane was immediately damaged, rapidly followed by extensive damage to the endoplasmic reticulum. The shape of the cell, however, was maintained, and the mitochondria and nucleus remained tightly held together by the cytoskeleton. Mitochondria remained intact initially, whereas the cytosol became less electron dense and the nuclear chromatin was more dispersed. An intermediate phase was characterized by total release of adenylate kinase and most of the glucose-6-phosphatase, marker enzymes for the mitochondrial intermembrane space and the endoplasmic reticulum, respectively. The outer mitochondrial membrane was ruptured, but mitochondria maintained their normal matrix electron density. In the final phase, characterized by the beginning of citrate synthase release from the mitochondrial matrix space, the mitochondria became swollen, and only the nucleus, inner and outer mitochondrial membranes, and the cytoskeleton could be clearly distinguished. Although the plasma membrane could not be readily discerned in electron micrographs after the initial phase, the plasma membrane marker enzyme 5′-nucleotidase remained associated with digitonin-treated hepatocytes. Acetyl-CoA carboxylase was released much more slowly than lactate dehydrogenase, indicating some severe restriction on its release. The release of acetyl-CoA carboxylase closely paralleled the release of glucose-6-phosphatase. The controlled exposure of hepatocytes to digitonin, therefore, leads to the sequential release of soluble, compartmentalized cellular components and some membrane-bound components, but the mitochondrial membrane, cytoskeleton and the nucleoskeleton survive even long-term digitonin treatment.     

Structural changes of isolated hepatocytes during treatment with digitonin

The structural changes accompanying digitonin-induced release of enzymes and metabolites from isolated hepatocytes have been studied by scanning and transmission electron microscopy. In the initial phase, characterized by total release of the cytosolic marker enzyme, lactate dehydrogenase, the plasma membrane was immediately damaged, rapidly followed by extensive damage to the endoplasmic reticulum. The shape of the cell, however, was maintained, and the mitochondria and nucleus remained tightly held together by the cytoskeleton. Mitochondria remained intact initially, whereas the cytosol became less electron dense and the nuclear chromatin was more dispersed. An intermediate phase was characterized by total release of adenylate kinase and most of the glucose-6-phosphatase, marker enzymes for the mitochondrial intermembrane space and the endoplasmic reticulum, respectively. The outer mitochondrial membrane was ruptured, but mitochondria maintained their normal matrix electron density. In the final phase, characterized by the beginning of citrate synthase release from the mitochondrial matrix space, the mitochondria became swollen, and only the nucleus, inner and outer mitochondrial membranes, and the cytoskeleton could be clearly distinguished. Although the plasma membrane could not be readily discerned in electron micrographs after the initial phase, the plasma membrane marker enzyme 5'-nucleotidase remained associated with digitonin-treated hepatocytes. Acetyl-CoA carboxylase was released much more slowly than lactate dehydrogenase, indicating some severe restriction on its release. The release of acetyl-CoA carboxylase closely paralleled the release of glucose-6-phosphatase. The controlled exposure of hepatocytes to digitonin, therefore, leads to the sequential release of soluble, compartmentalized cellular components and some membrane-bound components, but the mitochondrial membrane, cytoskeleton and the nucleoskeleton survive even long-term digitonin treatment.     

Effects of advanced glycation end-products on the proliferation and differentiation of osteoblast-like cells

Dextran M20 was added to isolated rat liver mitochondria to mimic cytosolic macromolecules. Under these conditions, the morphological changes in the mitochondrial periphery that occur upon isolation of the organelle are restored, i.e. the volume of the intermembrane space decreases and the contact site frequency increases. The ADP routing from mitochondrial kinases at various locations was investigated by using the activities of oxidative phosphorylation and externally added pyruvate kinase as sensors for ADP transport into the matrix and extramitochondrial compartment, respectively. The studies reveal that a significant fraction of the ADP generated by either adenylate kinase in the intermembrane space or by outer membrane bound hexokinase isozyme I, is not accessible to extramitochondrial pyruvate kinase. Quantitative information on the ADP compartmentation in rat liver mitochondria was obtained by comparing the ADP supply from mitochondrial kinases to oxidative phosphorylation with that of non-bound, extramitochondrially located kinases. This approach allowed us to estimate the ADP diffusion gradients which were present across the outer membrane and between the compartment formed by bound hexokinase and the extramitochondrial compartment. In the presence of 10% dextran M20 these ADP gradients amounted to approximately 12 µM. The possible role of mitochondrial kinases in ADP transport into mitochondria in vivo is discussed. (Mol Cell Biochem 174: 43–51, 1997)     

Effects of amphipathic peptides,including presequences,on the functional integrity of rat liver mitochondrial membranes

A number of amphipathic peptides were tested for their effects on structural and functional properties of isolated rat liver mitochondria. The peptides included the matrix targeting sequence of subunit IV of (yeast) cytochromec oxidase. Titration experiments in which the mitochondria were incubated with increasing concentrations of the peptides revealed two major stages in the interaction. First, at low peptide/mitochondria ratios, peptide binding to the outer membrane occurred which was accompanied by gradual lysis of the outer membrane at higher ratios. The latter was deduced from the release of adenylate kinase, the classical marker enzyme of the intermembrane space. Secondly, at still higher peptide/mitochondria ratios, the permeability of the inner membrane progressively increased, as evidenced by measurements of respiratory control and of the membrane potential. Complete uncoupling of respiration seemed to precede dissipation of the membrane potential.     

Preparation of an inner membrane-matrix fraction from yeast yeast mitochondria.

Treatment of yeast mitochondria with digitonin was used in order to prepare an inner membrane-matrix fraction preserving its permeability properties. The incubation time of mitochondria with digitonin was an essential parameter for the selective solubilization of the outer membrane. The incubation of mitochondria for l min at different concentrations of digitonin led to a three-step release of mitochondrial enzymes: (a) at low concentrations of digitonin, adenylate kinase was released; (b) higher concentrations were required to solubilize kynurenine hydroxylase, an outer membrane marker; (c) inner membrane markers (succinate dehydrogenase and oligomycin-sensitive adenosine triphosphatase) and matrix markers (fumarase and isocitrate dehydrogenase) were significantly released at concentrations of digitonin higher than 0.4 mg/mg of protein. The electron microscopic aspects of yeast mitoplasts (inner membrane-matrix fraction obtained by treatment with 0.4 mg of digitonin) showed an orthodox and a twisted configuration. These new organelles retained respiratory control when assayed with ethanol as the substrate. Their selective permeability properties were preserved as shown by isoosmotic swelling in potassium or ammonium salt solutions.