Simultaneous Separation and Purification of Neurofilament and Glial Filament Proteins from Brain

Neurofilaments (NF) and glial filaments (GF) were purified from bovine brain by the axonal flotation method, followed by hydroxylapatite chromatography in 8 M-urea. The proteins were shown to be competent to reassemble into intermediate filaments with removal of the denaturant, and reassembly was used as the final step in the purification of the filament proteins. The reassembly was found to be dependent on ionic strength and pH. This dependence was greater for neurofilaments than for the glial filaments. The NF and GF preparations were found not to be contaminated with each other by their gel electrophoretic profile and their immunological distinctness. The filament proteins can be obtained in high yield, and remain in solution if the urea is removed by dialysis against a low-ionic-strength buffer. Hence, they can provide a source for further biochemical studies.     

Preparation of Neurofilament Protein from Guinea Pig Peripheral Nerve and Spinal Cord

Abstract: A simple and rapid method for preparation of enriched neurofilament protein from mammalian peripheral nerve or spinal cord is described. Tissue extracts from guinea pig nerve or spinal cord are fractionated by ammonium sulfate fractionation, chromatography on Sepharose 4B, and precipitation with ethanol. Molecular exclusion chromatography on Sepharose 4B, in which the neurofilament protein elutes quantitatively in the exclusion volume of the column, with little contamination by other proteins, is found to be a highly effective purification step. The protein is found to precipitate in ammonium sulfate fractions over a wide range of salt concentration, from 20 to 80% saturation. It is found to be quantitatively precipitated in 40% v/v ethanol-water. The preparative method described yields 0.25 mg of neurofilament protein per gram of nerve or spinal cord, with a purity of approximately 50%. The three principal neurofilament polypeptides, which have molecular weights by SDS-polyacrylamide gel electrophoresis of 200K, 145K, and 68K, are found to be present in the preparation in a molar ratio of 1:2:6. A variant form of neurofilament protein occurring in approximately 20% of Hartley strain guinea pigs is described, which has the polypeptide composition: 200K, 192K, 145K, 68K.     

Effects of ATP on Phosphatidylinositol-Phospholipase C and Inositol 1-Phosphate Accumulation in Rat Brain Synaptosomes

Incubation of rat brain synaptosomes prelabeled with [2-3H]inositol resulted in a time-dependent release of labeled inositol 1-phosphate. This process was Ca2+ dependent, and ATP (1 mM) enhanced the inositol 1-phosphate formation three- to fivefold. Using [1-14C]arachidonoyl-phosphatidylinositol which was introduced into saponin-permeabilized synaptosomes, ATP (1 mM) and free Ca2+ (approximately 20 microM) enhanced the phospholipase C hydrolysis of this substrate to form labeled diacylglycerol. When the same permeabilized synaptosomal preparation was incubated with [2-3H]inositol-phosphatidylinositol, ATP not only enhanced the formation of labeled inositol 1-phosphate, but also inhibited the conversion of inositol 1-phosphate to inositol. Furthermore, ATP appeared to reduce the Ca2+ requirement of the phosphatidylinositol-phospholipase C. Inhibition of the conversion of inositol 1-phosphate to inositol could not be overcome by increasing the Mg2+ concentration in the incubation medium. Although the ATP effect is not viewed as a receptor-mediated event, it is possible that such an event may occur in synaptosomes under conditions in which intrasynaptic Ca2+ concentration becomes elevated.     

Stimulation of Inositol Phospholipid Hydrolysis by Excitatory Amino Acids Is Enhanced in Brain Slices from Vulnerable Regions after Transient Global Ischemia

Stimulation of inositol phospholipid hydrolysis by transmitter receptor agonists was measured in slices from hippocampus, cerebral cortex, and corpus striatum at various intervals after transient global ischemia in rats. Ischemia was induced through the four-vessel occlusion model. Stimulation of [3H]inositol monophosphate formation by excitatory amino acids was greatly enhanced in hippocampal slices prepared from ischemic rats at 24 h or 7 days after reperfusion. This potentiation was more evident using ibotenic acid and was also observed in cerebral cortex, but not in corpus striatum. This regional profile correlated with the pattern of ischemia-induced neuronal damage observed under our experimental conditions. The enhanced responsiveness to excitatory amino acids was always accompanied by an increase in both basal and norepinephrine-stimulated [3H]inositol monophosphate formation. In contrast, stimulation of [3H]inositol monophosphate formation by carbamylcholine was not modified in hippocampal or cortical slices from ischemic animals.     

Simultaneous labelling of microtubules and fibrillar bundles in tobacco BY-2 cells by the anti-intermediate filament antibody,ME 101

Summary In previous studies on plant cells, antibodies directed against intermediate filaments (IFs) have shown that IF antigens are distributed in one of two quite distinct forms. The first co-distributes with each of the four microtubule arrays (cortical, preprophase band, spindle and phragmoplast), while the second form is associated with cytoplasmic paracrystalline fibrillar bundles (FBs) of 10 nm filaments. Conditions allowing one form to be labelled with antibody have generally proved unsuitable for labelling of the other; this has prevented the simultaneous visualization of the two forms of IF antigen in plants and the study of any possible physical relationships between them at the electron microscopic level. In this paper, we show that ME 101, which recognizes an epitope in the N-terminal portion of all classes of intermediate filaments, stains both forms of plant IF antigen simultaneously in tobacco suspension cells using immunofluorescence or immunogold labelling techniques. These cells contain in their cortex short (ca. l m) fibrillar bundles which stain with ME 101. These bundles appear to be independent of the microtubule-associated epitope which stains in a continuous linear manner with ME 101. When protoplasts are either cleaved open on grids or sequentially extracted with detergents prior to critical point drying, the short fibrillar bundles are specifically labelled by ME 101 tagged with colloidal gold. ME 101 also co-distributed with underlying linear filaments, which appeared to be microtubules. In addition to these structures, the cortex also contains a meshwork of variably-sized fine filaments but these are not labelled with ME 101 nor with an antibody raised against the plant cytoskeleton, which recognizes cytokeratin 8. These results confirm that the fibrillar bundles and the microtubule-associated form of plant IF antigens are present simultaneously rather than experimentally-interconvertible, and that they appear to be physically unconnected.Abbreviations DAPI 4,6-diamidino-2-phenylindole - FB fibrillar bundle - FITC fluorescein isothiocyanate - IF intermediate filaments - MTSB microtubule stabilizing buffer - TBS Tris-buffered-saline     

Histamine Stimulation of Cyclic AMP Accumulation in Astrocyte-Enriched and Neuronal Primary Cultures from Rat Brain

Histamine stimulates cyclic AMP accumulation in astrocyte-enriched and neuronal primary cultures from rat brain in the presence of the phosphodiesterase inhibitor isobutylmethylxanthine. The response in the astrocyte cultures (Emax = 304 +/- 44% over basal, EC50 = 43 +/- 5 microM) was much higher than in neuronal cultures (Emax = 24 +/- 2%, EC50 = 14 +/- 7 microM). The histamine effect in astrocytes was competitively inhibited by the H2 antagonists cimetidine (Ki = 1.1 +/- 0.2 microM) and ranitidine (Ki = 46 +/- 10 nM) but was insensitive to the H1 antagonist mepyramine (1 microM). The two selective H2 agonists impromidine and dimaprit behaved as partial agonists and showed relative potencies (139 and 0.5, respectively) consistent with an interaction with H2 receptors. The more selective H1 agonist 2-thiazolylethylamine (0.01-1 mM) did not potentiate the response to impromidine (10 microM). Thus, in contrast to what is generally observed in intact cell preparations from brain, the histamine-induced cyclic AMP accumulation in astroglial cells is mediated solely by H2 receptors. The small effect shown in neuronal cultures also appears to be mediated by H2 receptors.     

A Method to Measure Simultaneously Cyclic AMP and Inositol Phosphate Accumulation in Rat Brain Slices

The simultaneous measurement of the accumulation of cyclic AMP and inositol phosphates in rat cerebral cortical slices is described. After stimulation, the separation of cyclic AMP and inositol phosphates was achieved using ion-exchange chromatography and their concentrations were determined by means of a double-labeling technique, the substrates adenine and inositol being labeled with 14C and 3H, respectively. The recoveries were 70-80% for inositol phosphates and 40-50% for cyclic AMP. To test the applicability of the method, norepinephrine was chosen as an agonist, because it is known to stimulate the production of these two second messengers by interacting with alpha- and beta-adrenergic receptors. This procedure is an improvement over existing methods, because we obtained the simultaneous formation of 3H-inositol phosphates and [14C]cyclic AMP in a concentration-dependent process. EC50 values were similar for the two, 8.5 +/- 3.9 microM for 3H-inositol phosphates and 20.2 +/- 6.3 microM for [14C]cyclic AMP, and close to the values obtained when each process was studied alone. The action of adrenergic antagonists was also tested. Propranolol blocked the norepinephrine stimulation of [14C]cyclic AMP, but did not inhibit the norepinephrine stimulation of 3H-inositol phosphates. The opposite results were observed with prazosin. Our results suggest that this method could be a useful tool to examine the interaction between these two receptor-coupled effectors.     

Effects of Single and Repeated Electroconvulsive Shock Administration on Inositol Phosphate Accumulation in Rat Brain Slices

Accumulation of inositol-1-phosphate after labeling with [3H]inositol and stimulation with noradrenaline, carbachol, and serotonin was measured in rat cortical, caudate nucleus, and hippocampal slices. The response to noradrenaline was significantly increased in cortical slices from animals that had received either a single electroconvulsive shock (ECS) or a series of 10 daily ECS but was unchanged in caudate nucleus or hippocampal slices. The response to carbachol, a muscarinic cholinergic agonist, was unchanged in cortical or caudate nucleus slices but was significantly reduced in hippocampal slices from animals that had received chronic ECS. The response to serotonin in cortical slices was not affected by the treatment. The results are correlated with changes in receptor number, which have been demonstrated to occur after administration of ECS.     

Regional Differences in the Coupling of Muscarinic Receptors to Inositol Phospholipid Hydrolysis in Guinea Pig Brain

The differential effects of muscarinic agents on inositol phospholipid hydrolysis and the role in this process of putative muscarinic receptor subtypes (M1 and M2) were investigated in three regions of guinea pig brain. Addition of the agonist oxotremorine-M to slices of neostriatum, cerebral cortex, or hippocampus incubated in the presence of myo-[2-3H]inositol and Li+ resulted in a large accumulation of labeled inositol phosphates (733, 376, and 330% of control, respectively). In each tissue, the principal product formed was myo-inositol 1-phosphate (59-86%), with smaller amounts of glycerophosphoinositol and inositol bisphosphate. Only trace amounts of inositol trisphosphate could be detected. Regional differences were observed in the capacity of certain partial agonists to evoke inositol lipid hydrolysis, the most notable being that of bethanechol, which was four times more effective in the neostriatum than in either the cerebral cortex or hippocampus. In addition, the full agonists, oxotremorine-M and carbamoylcholine, were more potent stimulators of inositol phosphate release in the neostriatum than in the cerebral cortex. The putative M1 selective agonist 4-m-chlorophenylcarbamoyloxy-2-butynyl trimethyl ammonium chloride had little stimulatory effect in any brain region, whereas the putative M1 selective antagonist pirenzepine blocked the enhanced release of inositol phosphates with high affinity in the cerebral cortex and hippocampus (Ki = 12.1 and 13.9 nM; "M1") but with a lower affinity in the neostriatum (Ki = 160 nM; "M2"). In contrast to its differential effects on stimulated inositol lipid hydrolysis, no regional differences were observed in the capacity of pirenzepine to displace [3H]quinuclidinyl benzilate, a muscarinic antagonist, bound to membrane fractions. Atropine, an antagonist that does not discriminate between receptor subtypes, inhibited the enhanced release of inositol phosphates with similar affinities in the three regions (Ki = 0.40-0.60 nM). The results indicate that by measurement of inositol lipid hydrolysis, regional differences in muscarinic receptor coupling characteristics become evident. These differences, which are not readily detected by radioligand binding techniques, might be accounted for by either the presence of functionally distinct receptor subtypes, or alternatively, by regional variations in the efficiency of muscarinic receptor coupling to inositol lipid hydrolysis.     

Inhibition by Forskolin of Excitatory Amino Acid-Induced Accumulation of Cyclic AMP in Guinea Pig Hippocampal Slices

Left sciatic nerves of adult male Sprague-Dawley rats were crushed and allowed to recover for 0, 1, 2, 4, 7, or 14 days. At each of these times both L-5 dorsal root ganglia were injected with 100 microCi of [3H]glucosamine. Two days later, dorsal root ganglia, lumbosacral trunks, and sciatic nerves were removed bilaterally. The amounts of radiolabelled ganglioside in crushed lumbosacral trunks were consistently higher than in the controls, with the largest difference occurring within 2 days from simultaneous crush and injection to killing (specimens labelled day 0). The largest difference in the amount of radiolabelled ganglioside between crushed and control sciatic nerve (4-9 days from crush to killing) occurred later than that of lumbosacral trunk, but no significant difference occurred within the first 3 days following crush. There was only a slightly higher radioactivity in gangliosides totalled from all three anatomical specimens of crushed than in control nerves. The neutral nonganglioside lipid and acid-precipitable fraction followed patterns of synthesis and accumulation similar to those of the gangliosides. These findings indicate that after nerve crush gangliosides, glucosamine-labelled neutral nonganglioside lipids, and glycoproteins accumulate close to the proximal end of the regenerating axon. This accumulation could serve as a reservoir to increase the ganglioside concentration in the growth cone membrane.     

Lithium Reduqes the Accumulation or Inositol Polyphosphate Second Messengers Following Cholinergic Stimulation of Cerebral Cortex Slices

Abstract: The ability of lithium to interfere with the metabolism of inositol phosphates in brain may underlie its therapeutic action in manic-depressive illness. In these experiments, lithium, at therapeutic concentrations, enhanced the accumulation of [³H]inpsitol monophosphate but suppressed the accumulation of the putative second messengers [³H]inositol 1,4,5-trisphosphate ([³H]Ins(1,4,5)P₃) and f³H]inositol 1,3,4,5-tetrakisphosphate following stimulation of cerebral cortex slices with carbachol. Mass measurements of Ins(1,4,5)P₃showed similar inhibitory effects, which could be prevented by preincubation with myo -inositol. These data may reveal the mechanism by which lithium can reduce polyphosphoinositide-midiated neurotransmission in brain.     

Quantitation of Cholinergic Synaptosomes from Guinea Pig Brain

An antiserum raised to nerve terminal sacs derived from the electric organ and Torpedo marmorata was used to lyse guinea pig brain synaptosomes in the presence of complement. From the release of the cytoplasmic enzymes choline acetyltransferase, lactate dehydrogenase, tyrosine hydroxylase and glutamate decarboxylase it appears that the antiserum binds specifically to cholinergic terminals. The amount of lactate dehydrogenase released was used to estimate the proportion of cholinergic nerve terminals in different synaptosome preparations.     

Cyclic GMP Formation and Inositol Phosphate Accumulation Do Not Share Common Origins in Rat Brain Slices

Cyclic GMP formation and inositol phospholipid hydrolysis were studied in rat brain slices to determine if the two processes have common origins. Muscarinic cholinergic stimulation enhanced [3H]inositol phosphate ([ 3H]IP) accumulation from slices prelabelled with [3H]inositol but did not affect cyclic GMP formation in the cortex, striatum, or cerebellum. An elevated level of extracellular K+ stimulated accumulation of both cyclic GMP and [3H]IP in cortex slices. The former, but not the latter, was reduced by lipoxygenase and phospholipase A2 inhibition. Calcium channel activation enhanced and blockade reduced K+-stimulated [3H]IP formation without affecting the cyclic GMP level, and there were differences in the Ca2+ requirements for the two responses. Thus, there is no support for the concept that guanylate cyclase activation inevitably accompanies inositol phospholipid breakdown, and the evidence presented demonstrates that K+ stimulation promotes cyclic GMP and [3H]IP accumulation by different transducing pathways.     

Two-Step Immunoaffinity Purification of Acetylcholinesterase from Rabbit Brain

Acetylcholinesterase (AChE; EC 3.1.1.7) extracted in 1% Triton X-100 from rabbit brain was purified 2,000-fold by chromatography on agarose conjugated with a monoclonal antibody directed against human red blood cell cholinesterase. After elution from the immunoadsorbent with pH 11 buffer, the preparation was purified further by affinity chromatography on phenyltrimethylammonium-Sepharose 4B with decamethonium elution. Overall yield of purified enzyme was 37% of the AChE originally solubilized, with a specific activity of 2,950 units/mg protein. Electrophoresis under reducing conditions in 7.5% sodium dodecyl sulfate polyacrylamide gels revealed only one silver-staining polypeptide band. A streamlined purification procedure enabled the isolation of electrophoretically homogeneous AChE to be completed in fewer than 7 days, at yields exceeding 50%. Electrophoretic analysis of purified AChE indicated an apparent MW of 71,000 for the monomeric subunit. Gel filtration and sucrose density gradient centrifugation in the presence of Triton X-100 showed little difference between the properties of the native and the purified enzyme. The molecular mass of the main species was estimated from the gel filtration and sedimentation data to be 280,000 daltons. Kinetic parameters of the purified protein (Km = 0.16 +/- 0.01 mM) were close to those of the native enzyme (Km = 0.12 +/- 0.01 mM) when examined with acetylthiocholine iodide as substrate. The two-step immunopurification procedure presented in this communication offers a convenient route to homogeneous neural AChE in quantities useful for detailed biochemical and immunochemical study.     

Properties of a CDP-Diglyceride Hydrolase from Guinea Pig Brain

Enzymatic hydrolysis of the pyrophosphate bond of CDP-diglyceride (CDP-DG), previously shown to occur in bacteria, is demonstrable in mammalian tissues. Activity was enriched in a lysosomal fraction obtained from guinea pig cerebral cortex and was purified 92-fold relative to the homogenate by a combination of XM-300 ultrafiltration and DEAE-cellulose column chromatography. When incubated with CDP-dipalmitin, the purified enzyme produced stoichiometric amounts of CMP and phosphatidate. dCDP-DG served as a substrate, while ADP-DG was an inhibitor, as were 5'-AMP and 5'-dAMP. CDP-DG hydrolysis was not affected by the presence of excess amounts of CDP-choline, CDP-glycerol, sodium pyrophosphate, or cyclic 3',5'-AMP.     

The Cytoskeleton of Primary Astrocytes in Culture Contains Actin, Glial Fibrillary Acidic Protein, and the Fibroblast-Type Filament Protein, Vimentin

Abstract: Primary astrocytes were cultured from the forebrains of 1-day-old rats. Immunofluorescence microscopy showed that approximately 80% of the cells were positive for glial fibrillary acidic protein (GFAP) and >80% were stained with an antiserum to the molecular weight 58,000 fibroblast intermediate filament protein (vimentin). Gel electrophoresis of Triton-insoluble cytoskeleton preparations from these cultures revealed three major bands having molecular weights of 58,000, 51,000, and 42,000, together with some prominent lower-molecular-weight species. The protein of molecular weight 51,000 was not present in preparations from fibroblasts. Each of the three major astrocyte proteins was subjected to limited proteolysis, while two of the proteins were cleaved by cyanogen bromide. The electrophoretic peptide patterns of the 58,000 protein were similar to those of vimentin isolated from NIL-8 fibroblasts, and the patterns of the 51,000 protein were similar to those of GFAP isolated from rat spinal cord. The patterns of the protein of molecular weight 42,000 resembled those of muscle actin. Rocket immunoelectrophoresis showed that the 51,000 astrocyte protein reacted with an antiserum to bovine GFAP, but the 58,000 and 42,000 proteins failed to react. We conclude that the major proteins of cytoskeleton preparations from cultured primary astrocytes are vimentin (58,000), GFAP (51,000), and actin (42,000), and that our data show no obvious structural relationship among them.     

Clonidine Inhibition of Norepinephrine Release from Normal and Morphine-Tolerant Guinea Pig Cortical Slices

Endogenous norepinephrine (NE) release in cerebral cortex slices taken from normal and morphine-tolerant guinea pigs was measured by HPLC. In normal slices, a linear relationship was found between electrically evoked NE release and the log of the frequency of stimulation in the range of 1-20 Hz. The efficiency of the alpha 2-mediated autofeedback was tested by adding the alpha 2-agonist clonidine and the alpha 2 agonist idazoxan. NE release was dose-dependently reduced by clonidine (1 nmol/L-1 mumol/L) and increased by idazoxan (10-100 nmol/L). The inhibition by clonidine was significantly greater at 1 Hz than at 3 Hz, whereas the absolute increase in NE release induced by idazoxan was greater at 3 Hz than at 1 Hz. Morphine at 1 mumol/L (a concentration per se ineffective) shifted to the left the clonidine concentrations able to inhibit NE release at 3 and 1 Hz (1-10 nmol/L), but at both frequencies, the opiate reduced the maximal inhibition induced by clonidine at 1 mumol/L. In slices taken from morphine-tolerant guinea pigs (in the presence of morphine at 1 mumol/L), clonidine (1 nmol/L-1 mumol/L) was ineffective at the stimulation rate of 3 Hz, but it was more active than in normal slices at 1 Hz. Such a response pattern suggests a reduced availability of alpha 2 receptors and an increase in their sensitivity to clonidine. However, chronic morphine treatment did not influence the physiological autoinhibition because the increase in NE release elicited by idazoxan (10-100 nmol/L) at 1 and 3 Hz was the same in normal and in "morphine-tolerant" slices.(ABSTRACT TRUNCATED AT 250 WORDS)     

Receptor-Mediated Phosphorylation of Astroglial Intermediate Filament Proteins in Cultured Astroglia

Primary cultures of purified astroglia have been shown to exhibit a variety of membrane receptors that regulate intracellular cyclic AMP levels. The experiments described in this paper were completed to examine the effect of such receptor agonists on protein phosphorylation in intact astroglia. An analysis of 32P-labelled proteins derived from whole cell extracts and separated via two-dimensional gel electrophoresis indicated that increasing cyclic AMP levels in astroglia stimulated the phosphorylation of two distinct proteins that had apparent molecular weights/isoelectric points (pI) of 51K/6.0 and 57K/5.7. Similar experiments with cultured meningeal cells indicated that only the 57K/5.7 protein was phosphorylated in response to elevated levels of cyclic AMP. The 51K/6.0 protein was never observed in gels derived from meningeal cells. Immunoblot experiments indicated that the 51K/6.0 protein stained with antiserum to glial fibrillary acidic protein (GFAP) and the 57K/5.7 protein stained with antibodies to vimentin. Concentration-effect studies indicate that these proteins are maximally phosphorylated at concentrations of receptor agonists that only slightly elevate cyclic AMP levels. All receptor agonists that have been shown to increase cyclic AMP levels appear similarly efficacious with respect to increasing the phosphorylation of the two proteins. These experiments suggest that the membrane receptors present on astroglia function, in part, to regulate phosphorylation of the intermediate filament proteins GFAP and vimentin.     

Synthesis and Turnover of Cytoskeletal Proteins in Cultured Astrocytes

Abstract: We previously reported that the cytoskeleton of rat astrocytes in primary culture contains vimentin, glial fibrillary acidic protein (GFAP), and actin. These proteins were found in a fraction insoluble in Triton X-100 and thought to be assembled in filamentous structures. We now used primary astrocyte cultures to study the kinetics of synthesis and turnover of these cytoskeletal proteins. The intermediate filament proteins were among the most actively synthesized by astrocytes. High levels of synthesis were detectable by the third day of culture in the early log phase of growth, and the pattern of labeling at day 3 was similar to that at 14 days when the cultures had reached confluency. In short-term incorporation experiments vimentin, GFAP, and actin in the Triton-insoluble fraction were labeled within 5 min after exposure of the cultures to radioactive leucine. We did not detect any saturation of labeling for up to 6 h of incubation. The turnover of filament proteins studied by following the decay of radioactivity from prelabeled vimentin, GFAP, and cytoskeletal actin displayed biphasic decay kinetics for all three proteins. In the initial phase a fast-decaying pool with a half-life of 12–18 h contributed about 40% of the total activity in each protein. A major portion, about 60%, of each protein, however, decayed much more slowly, exhibiting a half-life of about 8 days.     

Inactivation of G1ucose-6-Phosphate Dehydrogenase Isozymes from Human and Pig Brain by Aluminum

Prolonged intake of low levels of aluminum from the drinking water has been found to increase the aluminum content in rat brain homogenates and to reduce the activity of hexokinase and glucose-6-phosphate dehydrogenase (G6PD). To determine the interaction of G6PD with aluminum in the brain, we have recently purified two isozymes of G6PD (isozymes I and II) from human and pig brain. Unlike isozyme I, isozyme II also had 6-phosphogluconate dehydrogenase (6-PGD) activity. We report here that G6PD isozymes I and II from human and pig brain purified to apparent homogeneity are inactivated by aluminum. Aluminum did not affect the 6-PGD activity of isozyme II. The aluminum-inactivated enzyme contained 1 mol of aluminum/mol of enzyme subunit. The protein-bound metal ion was not dissociated by exhaustive dialysis at 4 degrees C against 10 mM Tris-HCl (pH 7.0) containing 0.2 mM EDTA. Preincubation of aluminum with citrate, NADP+, EDTA, NaF, ATP, and apotransferrin protected the G6PD isozymes against aluminum inactivation. However, when the G6PD isozymes were completely inactivated by aluminum, only citrate, NaF, and apotransferrin restored the enzyme activity. The dissociation constants for the enzyme-aluminum complex of the isozymes varied from 2 to 4 microM, as measured by using NaF, a known chelator for aluminum. Inhibition of G6PD by low levels of aluminum further strengthens the suggested role of aluminum toxicity in the energy metabolism of the brain.