The Interaction of Phospholipid Membranes and Detergents with Glutamate Dehydrogenase

Transcription

1 Eur. J. Biochem. 74, (1977) The Interaction of Phospholipid Membranes and Detergents with Glutamate Dehydrogenase 2. Fluorescence and Stopped-Flow Studies Mohsen NEMAT-GORGANI and George DODD Department of Molecular Sciences, University of Warwick (Received October 22, 1975 / November 23, 1976) 1. Both the anionic detergent sodium dodecylsulphate and the cationic detergent cetyltrimethylammonium bromide quenched the protein fluorescence of glutamate dehydrogenase. The anionic compound was more effective and brought about 50% quenching at a detergent concentration of 0.4 mm. The zwitterionic amphiphile, lysolecithin, did not quench the protein fluorescence and neither did the short-chain detergent n-hexylsulphonate, which under the range of concentrations examined (< 1 mm) does not form micelles. 2. The zwitterionic phospholipid, phosphatidylcholine, did not quench the protein fluorescence but the anionic phospholipids, phosphatidylserine and cardiolipin, induced a reversible quenching of the enzyme fluorescence. These observations confirm the specificity of the phospholipid-enzyme interactions as deduced from the kinetic studies of the preceding paper. The degree of quenching brought about by the phospholipids decreased with increasing ionic strength and increasing ph and could be substantially reduced by basic proteins. An electrostatic contribution to the interaction is inferred from these results. 3. The binding of the anionic phospholipids to the enzyme is manifested in a further enhancement of the fluorescence of a l-anilinonaphthalene-8-sulphonate. enzyme complex. The presence of substrates and allosteric effectors affect the interaction of the lipids with the enzyme as indicated by the magnitude of this increase in fluorescence. The enhancement of fluorescence of NADH when bound to the enzyme was not affected by the binding of the lipids. 4. The complex formed between the enzyme and phosphatidylserine/phosphatidylcholine can be solubilized in isooctane. The photolability of the aqueous protein when subjected to irradiation at 280 nm is suppressed in the isooctane-soluble complex. 5. Phosphatidylserine brings about a rapid (ti is about 150 ms at a lipid concentration of 5 mm) dissociation of the linear aggregates formed between the enzyme oligomers. 6. A model of the enzyme. lipid-membrane complex, consistent with these results, is proposed. It is suggested that the enzyme is an allotopic protein and that the dissociation of the enzyme in vitro may involve binding sites on the protein which are designed for interaction with the cardiolipin of the inner mitochondrial membrane, when the enzyme is in the mitochondrial matrix. The interaction between glutamate dehydrogenase and anionic detergents and phospholipids has been studied by steady-state kinetic methods as reported in the preceding paper. Since kinetic methods do not permit us to observe the binding of the amphiphiles to the protein in the absence of substrates, we have also studied the enzyme - amphiphile interaction using spectroscopic methods. In this paper we report our work using both intrinsic protein fluorescence and the extrinsic fluorescence of 1-anilinonaphthalene-8-sul- Enzyme. Glutamate dehydrogenase or L-glutamate: NAD(P) oxidoreductase (deaminating) (EC ). phonate. Studies on the effect of phospholipids on the aggregation of the enzyme and on the extraction of the enzyme -lipid complex into liquid hydrocarbons are also reported. The results provide evidence for the allotopic behaviour of the enzyme. EXPERIMENTAL PROCEDURE Materials These were as described in the preceding paper.

2 140 Glutamate-Dehydrogenase-Phospholipid Interactions Methods The general methods were as described in the preceding paper. Fluorescence Measurements Fluorescence studies were carried out on a Perkin- Elmer spectrofluorimeter MPF-3. The enzyme in glycerol solution was dialysed in 0.1 M phosphate buffer, ph 8 (+ 0.1 mm EDTA) overnight. For the protein fluorescence, excitation and emission wavelengths of 292 nm and 334 nm respectively were chosen. Under these conditions there was little photodecomposition of the enzyme. A filter with a cutoff below 310 nm was used to reduce the scatter contribution from phospholipids to 2-8% (depending on phospholipid concentration) of the fluorescence due to free enzyme. Extrinsic fluorescence studies were carried out using (I-anilinonaphthalene- 8-sulphonate (magnesium salt) as the fluorescence probe. The fluorescence studies, unless otherwise stated, were carried out in a buffer containing 0.02 M Hepes, 0.02 M Pipes, 0.02 M Tris and 8 mm phosphate, containing 0.5 mm EDTA, at a final ph of In all cases the required quantity of the enzyme was added to a l-anilinonaphthalene-8-sulphonate solution in the absence or presence of metabolites and the phospholipids were added last. Control measurements were taken and the enhancement of 1-anilinonaphthalene-8-sulphonate on the binding to the enzyme and further fluorescence increase on addition of the anionic phospholipids were recorded. When the effect of NADH on lipid. enzyme complex formation was investigated using the fluorescence of 1-anilinonaphthalene-8-sulphonate, excitation and emission wavelengths of 410 nm and 550 nm were chosen. No NADH fluorescence could be detected in these conditions and the fluorescence due to bound 1 -anilinonaphthalene-8-sulphonate was detec ted. RESULTS Effect of Phospholipids and Amphiphiles on the Protein Fluorescence Studies on the intrinsic fluorescence of the enzyme showed that those amphiphiles which caused inhibition of enzyme activity also quenched the enzyme fluorescence. Thus sodium dodecylsulphate and cetyltrimethylammonium bromide caused strong quenching but Iysolecithin and hexylsulphonate showed none (Fig. 1). The anionic phospholipids cardiolipin and phosphatidylserine also quenched the intrinsic fluorescence of the enzyme (Fig. 2). Phosphatidylcholine, which does not affect the activity of the enzyme, showed only a small effect (Fig. 2) O 1.2 [Detergent] (mm) '' O [Detergent] (mm) Fig. 1. Change in fluorescence ofglutamate dehydrogenuse on titrating with amphiphiles. The titration was performed at 18 "C in a buffer mixture of 0.02 M Hepes, 0.02 M Pipes, 0.02 M Tris and 6-8 mm phosphate, at a final ph of 7.3. The protein concentration was 40 pg/ml. The excitation and emission wavelengths were 291 nm and 334 nm respectively. (A) (0) Lysolecithin, (0) sodium dodecylsulphate; (B) (0) n-hexylsulphonate, (0) cetyltrimethylammonium bromide ud OO o 1.5 [Phospholipid] (mm) Fig. 2. Chunge in fltiorescmce ofglutamate dehydrogenase on titrating withphospholipids. Conditions as for Fig. 1. The phospholipids were sonicated and clarified as described in the preceding paper. (0) Phosphatidylcholine, (0) phosphatidylserine, (A) cardiolipin Quenching of the enzyme fluorescence could be due to either a chromophore -lipid interaction or to a lipid-induced conformational change in the enzyme accompanied by the accessibility of fluorescent residues to the surrounding aqueous medium. A glutamate dehydrogenase oligomer consists of six identical polypeptide chains. There are 500 amino acid residues

3 M. Nemat-Gorgani and G. Dodd 141 in each polypeptide chain with 18 tyrosine, 23 phenylalanine and 3 tryptophan residues [I]. The maximum fluorescence emission is in the region of 335 nm and is due to tryptophan residues. To explore the possibility of a direct interaction between the tryptophan residues and the amphiphiles the effect of these compounds on the fluorescence behaviour of N-acetyltryptophanamide was studied. This compound has been used as a model for tryptophan residues incorporated in a polypeptide chain [2]. Generally additives which decrease the dielectric constant of water result in enhancement of fluorescence; those which increase the dielectric constant have the opposite effect. The results showed that neither sodium dodecylsulphate nor lysolecithin quenched its fluorescence. In the presence of cetyltrimethylammonium bromide a 2-4% increase was found with the detergent at 1 mm and N-acetyltrytophanamide at 0.4 mm. The anionic phospholipids phosphatidylserine and cardiolipin quenched N-acetyltryptophanamide (0.4 mm) fluorescence by 12% and 8% at 0.5 mm and 0.25 mm respectively. Lecithin, at a concentration of 0.5 mm, caused only 2-3% quenching. The extent of quenching by the anionic phospholipids was independent of ph in the range From these results it may be concluded that direct interaction of the detergents with tryptophan residues exposed on the surface of the enzyme is not responsible for the quenching of the protein fluorescence. In the case of quenching of the enzyme fluorescence by detergents the process takes place at concentrations of the detergent (above 0.1 mm) at which irreversible denaturation takes place. At these concentrations extensive unfolding in the structure of the polypeptides can take place and the tryptophan residues present in the interior of the enzyme structure may be displaced and positioned in an aqueous environment with a higher dielectric constant, so that the residues fluorescence to a smaller extent. In the case of reversible enzyme - phospholipid interactions, fluorescence quenching can be partly due to direct interaction with the tryptophan residues exposed at the surface of the enzyme and partly due to changes in the accessibility of the tryptophan residues buried in the anhydrous interior to the surrounding aqueous medium. The extent of fluorescence quenching of the enzyme by the anionic phospholipids diminished with increasing ionic strength (Fig. 3A) and increasing ph (Fig. 3 B), supporting our kinetic results. Thus, conditions of low ph and ionic strength, which are favourable for complex formation, may increase the accessibility of the tryptophan residues to the aqueous environment with the result that the fluorescence quenching of the enzyme is increased. The extent of fluorescence quenching was also affected by the presence of basic proteins, such as cytochrome c and ribonuclease. These proteins interact d g I 30' ' [NH&I] (M) 4 75' :5 8:5 PH Fig. 3. Effect of NH,Cl and of ph on the cardiolipin quenching of glutamate dehydrogenase fluorescence. The protein concentration was 40 pg/ml, the cardiolipin concentration was 0.5 mm for (A) and 0.2 mm for (B), and other conditions were as for Fig. 3. (A) (0) Glutamate dehydrogenase; (0) glutamate dehydrogenase and cardiolipin. (B) (0) Glutamate dehydrogenase; (0) glutamate dehydrogenase and cardiolipin with phosphatidylserine vesicles and neutralise or even reverse the zeta potential of these vesicles [3]. Ribonuclease at a concentration of 25 pm did not show any effect on the intrinsic fluorescence of the enzyme but diminished the quenching of enzyme fluorescence by phosphatidylserine (0.4 mm) from 75% to 40%. Cytochrome c had a similar effect. Ribonuclease was especially useful because it contains no tryptophan residues and did not interfere with the measurements of the tryptophan emission of the enzyme. The effect of substrates and effectors of the enzyme on its interaction with phosphatidylserine was studied. Both L-glutamate and 2-oxoglutarate diminished the quenching of the enzyme fluorescence by phosphatidylserine. At final protein and lipid concentrations 0.2

4 142 Glutamate-Dehydrogenase-Phospholipid Interactions of 40 pg/ml and 0.5 mm respectively, the protein fluorescence was quenched to 64% of the control value. 2-oxoglutarate (3 mm) decreased the quenching to 57% of the control amount. Godinot and Lardy [4] obtained 34% solubilisation of the enzyme activity from microsomal membranes (which contain phosphatidylserine) in the presence of 20 mm L-glutamate and 2mM NAD, and this may be due to the substrate effects observed by us. GTP, ADP, NADH and NAD + quenched the intrinsic fluorescence of the enzyme and hence it was not possible to study their effects on the binding of the enzyme to phospholipid membranes. Effect of An?plziphiles on the Fluorescence of NADH Bound to the Enzyme The enzyme undergoes allosteric conformational changes in the presence of NADH and GTP IS]. Thus, GTP further increases the NADH fluorescence enhancement in the presence of the enzyme. Phosphatidylserine (0.53 mm) did not affect NADH (2 pm) fluorescence when the nucleotide was partially or fully bound to the enzyme, both in the absence and presence of GTP (0.2 mm). Thus, in the presence of the phospholipid membranes the enzyme can bind NADH and can undergo the heterotropic allosteric conformation changes induced by GTP. The anionic detergent sodium dodecylsulphate did not affect the fluorescence of NADH bound to the enzyme at detergent concentrations up to 0.1 mm. Above this concentration (at which irreversible complex formation occurs) the fluorescence of bound NADH increased with increasing detergent concentration, reaching a 150% value at a detergent concentration of 0.3 mm. The increase was accompanied by a blue-shift in the P 1.01 I [Phospholipid] (rnm) Fig. 4. Further mhuncement ofjluorescence of I-anilinonaphthalene- X-sulphonate bound to glutamate dehydrogenase on titrution with phospholipids. The protein concentration was 0.67 mg/ml and the l-anilinonaphthalene-8-sulphonate concentration was 25 pm. The excitation and emission wavelengths were 370 nm and 470 nm respectively. (0) Phosphatidylserine: (0) cardiolipin. Fluvresccncc ratio is of (anilinonaphthalenesulphonate + enzyme + phosphatidylserine)/(analinonaphthalenesulphonate + enzyme) emission maximum of about 10nm. Above a concentration of 0.3 mm the system showed a timedependent effect with NADH fluorescence rapidly decreasing with time. This effect of sodium dodecylsulphate confirms the dependence of structural changes, brought about in the enzyme, on the concentration of detergent. At concentrations below 0.1 mm reversible complex formation takes place and NADH binding to the enzyme is not affected. At concentrations above 0.1 mm the detergent brings about unfolding of the enzyme structure and presumably binds the coenzyme more tightly without creating new binding sites. Above a concentration of 0.3 mm, total disruption of the polypeptide regions with respect to NADH binding sites takes place and the time-dependent effect is observed. The Interaction between the Enzyme and Phospholipids as Studied Using the Fluorescence Probe 1 -Anilinonaphthalene-8-sulphonate Extrinsic fluorescence studies were carried out using anilinonaphthalene-8-sulphonate as a fluorescence probe. The probe showed a weak affinity for the enzyme and stronger binding to the zwitterionic lipids, lysolecithin and phosphatidylcholine, and to the cationic amphiphile cetyltrimethylammonium bromide. Anilinonaphthalene-8-sulphonate at a concentration of 1 pm showed maximum enhancement with the phospholipid and the cationic detergent at concentrations of 0.53 mm and 0.05 mm respectively. It showed no fluorescence enhancement with the anionic amphiphiles. 1 -Anilinonaphthalene-8-sulphonate fluorescence when the probe was bound to lysolecithin (0.54 mm lysolecithin, 10 pm l-anilinonaphthalene-8-sulphonate) or phosphatidylcholine (0.53 mm phosphatidylcholine, 1 pm l-anilinonaphthalene-8-sulphonate) was not affected by the enzyme, confirming that no interaction between these lipids and the enzyme was occurring. It has been shown that the probe is located in the polar head-group region of the membrane with the non-polar fluorescent moiety of the probe penetrating to a short distance between the fatty acid chains of the phospholipid bilayer [6]. The fluorescence of the probe bound to the enzyme was increased by phosphatidylserine and cardiolipin (Fig. 4). This further enhancement of fluorescence could occur in three ways: (a) anionic phospholipids might bring about conformational changes in the enzyme thereby creating new or more accessible hydrophobic binding sites in the protein which could bind the probe; (b) charge neutralisation of some of the negatively charged head-groups in the phospholipid bilayer may take place in the lipid. enzyme complex, this would facilitate penetration of the negatively charged probe molecule into the hydrophobic region of the bilayer; (c) at positions in the

5 ~~ M. Nemat-Gorgani and G. Dodd 143 Table 1. Effect of substrates and allosteric effectors on the phosphatidylserine(ps)-induced enhancement of the fluorescence of l-unilino- ~ ~ ~ i ~ ~ h t h ~ i l ~ ~ ~ i i ~ A -NS) K - shound i i l ~ 10 ~ l ghitainotc~ ~ o i ~ ~ ~ clc.h~.h.ofii'i~~i.~' ~ ~ ~ icdhl Conditions as described for Fig. 4 Conditions Fluorescence enzyme phosphatidylserine added compound (concn) (ANS + GDH) (ANS + GDH + PS) (ANS + GDH + PS)/ concentration concentration (ANS + GDII) mgw mm GTP NADH GTP NADH - 2-oxoglutarate NAD+ 2-oxoglutarate NAD+ - ADP (mm) 8.0 (1) 8.5 (0.1) 10.0 (0.1) 20.5 (0.1) 12.5 (5) 13.0 (0.5) I :!5) 23.5 (0.1) bilayer where slight insertion of some of the hydrophobic residues of the enzyme into the phospholipid occurs, 1-anilinonaphthalene-8-sulphonate penetration into the hydrophobic bilayer region may take place. To help distinguish between these possibilities, the binding of the probe to the enzyme was studied by looking at the quenching of the intrinsic fluorescence of the enzyme brought about by binding of l-anilinonaphthalene-8-sulphonate. An average dissociation constant of 0.12 mm was obtained for the interaction between 1-anilinonaphthalene-8-sulphonate and the enzyme. The binding plots indicated the appearance of new binding sites for the probe in the lipid. enzyme complex. The effect of some of the metabolites involved in the regulation of the enzyme activity on the lipid induced further enhancement of 1 -anilinonaphthalene- 8-sulphonate bound to the enzyme was studied. Typical results are shown in Table 1. As mentioned above, some of these metabolites, such as ADP, GTP and NADH, were shown to quench the intrinsic fluorescence of the enzyme and it was not possible to examine their effect on the lipid. enzyme complex formation by intrinsic fluorescence studies. Extrinsic fluorescence studies made this possible. ADP and NAD further increased the 1 -anilinonaphthalene-8-sulphonate fluorescence enhancement by phosphatidylserine. NADH, especially in the presence of GTP, caused a decrease of the probe fluorescence enhancement by phosphatidylserine. GTP alone showed no appreciable effect. The activity of the enzyme is subject to allosteric regulation by ADP and GTP; GTP is a strong inhibitor and ADP is an activator. Both these effectors caused conformational changes in the enzyme ~ xx) Time (rns) Fig. 5. Turbidity changes on mixing glutamate dehydrogenase with phosphatidylserine. The enzyme in 0.1 M phosphate ph 8.0 containing 50 pm EDTA, was mixed with lipid sonicates prepared in the buffer mixture described in Fig. 1. The final concentrations of the reactants were: protein, 1.1 mg,ml; lipid, 0,75 mm: NADH, 30pM: and GTP, 50 pm. The experimental points were obtained from photographs of oscilloscope tracings. (0) Enzyme mixed with lipid; (0) enzyme mixed with NADH and GTP Effect of Phospholipids on the State of Association of Glutamate Dehydrogenase The enzyme oligomers associate at high concentrations (> 0.2 mg/ml) to form linear aggregates [7]. Huang and Frieden [8] followed the rate of dissociation of the enzyme by measuring the absorbance (turbidity) changes at 310 nm using stopped-flow methods. We investigated the possible effect of phospholipids on the state of association of the enzyme following the procedure of Huang and Frieden. A final enzyme concentration above 1 mg/ml was used, at which concentration the enzyme is known to be mainly in an associated form. It was found that on mixing the enzyme (in 0.1 M phosphate buffer) with phosphatidylserine (in a mixture of Hepes, Pipes and Tris) there was a rapid increase in turbidity followed by a slow decrease (Fig.5). The time taken for the second phase of the

6 144 Glutamate-Dehydrogenase-Phospholipid Interactions e- I/ Time (5) Fig. 6. Turbidity changes on mixing gluutamate dehydrogenase with lipids. Conditions and concentrations of protein, NADH and GTP as for Fig. 5. The final concentrations of phosphatidylcholine and phosphatidylserine were 0.5 mm and 0.25 mm respectively. The experimental points were obtained from photographs of oscilloscope tracings. (A) enzyme mixed with phosphatidylcholine; (0) enzyme, NADH, GTP, mixed with phosphatidylserine; (0) enzyme mixed with phosphatidylserine reaction was found to be about 350 m s with final concentrations of the enzyme and phospholipid at 1.1 mg/ml and 5 mm respectively. The time taken for completion of the second phase was found to increase to 70 s with the same enzyme concentration but a lipid concentration of 0.25 mm (Fig. 6). The turbidity decrease was shown to indicate dissociation of the associated form of the enzyme in the following three ways (a) when the enzyme was mixed with GTP (50 ym) and NADH (30 ym), which conditions are known to dissociate the enzyme, there was a decrease in turbidity (Fig. 5) ; (b) when the enzyme, in the presence of the above effectors at the same concentration, was mixed with phosphatidylserine, there was no further turbidity decrease (Fig. 6) ; (c) phosphatidylcholine, which does not interact with the protein, did not cause a turbidity decrease when it was mixed with the enzyme (Fig. 6). The initial rise in turbidity was found to be, at least partly, due to interaction of the phospholipids with phosphate buffer. Though the phospholipid sonicates were prepared in a buffer mixture, the enzyme was left in 0.1 M phosphate buffer to make sure that no denaturation would take place. As shown in the preceding paper, phosphatidylserine, when sonicated in phosphate buffer, gradually lost its capacity to complex with glutamate dehydrogenase. Extraction of Glutamate-Dehydrogenase. Phospholipid Complexes into Isooctane Protein phospholipid complexes, in which the main stabilizing forces are electrostatic interactions, can be solubilised in liquid hydrocarbons under certain conditions [9,10]. A co-sonicate of a zwitterionic and an anionic phospholipid is required and there are a limited number of anionic lipids which interact with the protein in the extracted complex. We have extracted several single polypeptide enzymes into hydrocarbon solvents [lo]. The activity of the enzymes is retained, and active enzymes can be recovered from the organic medium. When the complexes formed between glutamate dehydrogenase and co-sonicates of phosphatidylcholine and phosphatidylserine in the aqueous phase were extracted with isooctane, part of the enzyme. lipid complex passed into the hydrocarbon phase. The protein in the isooctane was identified from both its absorption and emission spectra. Up to 90% of the protein could be extracted into the organic phase using a phosphatidylserine/phosphatidylcholine ratio of 1.5 in the aqueous medium. The method of extraction which we used is essentially the same as that used for the smaller proteins and enzymes. It is a relatively harsh procedure which involves considerable agitation of the protein solution. The small single polypeptide extracellular enzymes, such as trypsin and a-chymotrypsin, are not denatured by the extraction procedure, but variable results were obtained with the more complex enzyme glutamate dehydrogenase. A much gentler extraction procedure would be desirable for subunit enzymes. The re-extraction procedure, which permits re-extraction of an active trypsin into the aqueous phase [lo] is also harsh and glutamate dehydrogenase was denatured under these conditions. The fluorescence of the protein in isooctane differed from that of the enzyme in the aqueous medium in two respects: first, a blue-shift in the emission maximum by about 15 nm and secondly, the photodecomposition which the protein in aqueous systems undergoes when irradiated with a high intensity light source at 280 nm did not occur in the hydrocarbon solvent. DISCUSSION The results of the fluorescence experiments confirm the specificity of the phospholipid -protein inter-

7 M. Nemat-Gorgani and G. Dodd 145 action which was inferred from the kinetic studies, in that the zwitterionic lipid phosphatidylcholine does not bind to the enzyme but the acidic lipids, phos- ~ phatidylserine and cardiolipin, both bind to the enzyme and quench the protein fluorescence. A quantitative interpretation of the quenching data (Fig. 2) presents difficulties which highlight some of the unusual features of phospholipids as ligands. In these experiments the phospholipid molecules are aggregated into vesicles of a size comparable to the enzyme and at the concentrations of lipid used the only significant interaction takes place between the protein oligomer and the lipid vesicle. When the quenching curves of the type shown in Fig. 2 are analysed using the Scatchard equation non-linear plots are obtained. The average number of lipids bound, calculated from such plots, varies between The large number of bound ligands arises if no account is taken of the aggregation of the lipids. The particle size distribution of the lipid vesicles which we used in these experiments was not measured but if we assume that the average vesicle particle weight is about 3 x lo6 daltons, then we can conclude that in the quenching experiments we are observing 3-4 vesicles of phosphatidylserine binding to the enzyme under the conditions used. The dissociation constant calculated from the fluorescence quenching is about lop4 M, some 100 times magnitude weaker than the Ki(,,,, calculated from the kinetic studies in the preceding paper. This difference may arise from the fact that in the fluorescence experiments we are probably observing only the interaction between the lipids and tryptophan residues in the protein whereas in the kinetic experiments we are observing the effect of the lipid - protein complexation on the active site residues. 1-Anilinonaphthalene-8-sulphonate does not give a fluorescence enhancement in the presence of anionic phospholipids. This is probably due to the fact that the negatively charged probe cannot penetrate into the hydrophobic regions of the bilayer. However, after formation of a lipid. enzyme complex, charge neutralisation of the phospholipid head-groups (by interaction with the positively charged groups in the enzyme structure) may take place and the lipid may acquire the capacity to accommodate l-anilinonaphthalene-8-sulphonate in its hydrophobic regions. At the ph values at which the kinetic and fluorescence experiments were carried out (ph 6-9) the anionic phospholipids and the enzyme (pl 4-5) have the same charge type. Formation of enzyme. lipid complexes increased with decreasing ionic strength and ph, demonstrating an electrostatic contribution to the binding energy. Involvement of hydrophobic interaction was indicated by an increase in the extent of complex formation with temperature. However, the interaction must be different from that between basic proteins and anionic phospholipids such as cytochrome c and cardiolipin [ll]. The net charges of these basic proteins (pl values > 10) are opposite to the anionic phospholipid bilayer surface. In the case of cytochrome c it has been shown that the basic lysine residues as well as the hydrophobic amino acid residues tend to occur in distinct clusters along the protein chain [12]. Similarly the presence of a specific region in bovine prothrombin for binding to phospholipids has been demonstrated [13]. In the case of glutamate dehydrogenase, 33 out of 500 amino acid residues of each polypeptide chain are lysine residues. It is possible that they may have an asymmetrical distribution which would make them suitable for binding to an anionic surface. Because the charge type of the phospholipid membranes which bind to glutamate dehydrogenase are the same as that of the protein, we can postulate that there must be specific binding sites for anionic phospholipids on the protein, as in the case of prothrombin. Electrostatic interaction between a negatively charged membrane and positively charged residues in the enzyme may be followed by conformational changes in the enzyme and structural changes in phospholipid membranes. Hydrophobic interactions may then take place between the hydrophobic amino acids of the enzyme exposed to the surface as a result of the primary electrostatic interaction, and the hydrophobic regions in the phospholipid bilayer. This is possible in view of the fluid state of the phospholipids. Thus, after complex formation, some hydrophobic resides of the enzyme may penetrate into the bilayer structure of the phospholipid membranes. Only a slight penetration would be expected in view of the low proportion of hydrophobic amino acid residues in the enzyme, its high solubility and the fact that an enzyme. phospholipid complex may be partially dissociated by increasing the ionic strength. The association of the enzyme at high (> 0.2 mg/ml) protein concentrations is a well known property of the enzyme [14]. The degree of association varies with the source of enzyme and has been most intensively studied for the beef liver enzyme which we have used in this work. This enzyme exhibits a concentration-dependent linear association to form chains. The consequence of this association for the behaviour of the enzyme in the mitochondrial matrix is not known. The concentration of the enzyme within the matrix is at least 2 mg/ml and the associated form of the enzyme would be found at this concentration. Our results suggest that the association of the enzyme may be an artifact in vitro, which involves the utilization of binding sites on the enzyme which are designed for binding to cardiolipin in the inner mitochondrial membrane. The kinetic and intrinsic fluorescence studies were carried out at very low protein concentration (< 50 pg/ml) at which the enzymic species is the oligomer

8 146 Glutamate-Dehydrogenase-Phospholipid-Interactions comprised of six subunits. The six polypeptide chains of each enzyme oligomer appear to be arranged in two layers, each composed of three elongated ellipsoidal subunits (Fig. 7) the oligomer having dimensions of about 14nm by 8.6nm by 9nm [14]. The phospholipid vesicles are spherical in shape and have an average diameter of 23 nm. Thus, in the case of the interaction between glutamate dehydrogenase and phospholipids, the process takes place between two aggregates of similar size. Oligomer association takes place along the major axis of the ellipsoid units[14]. Amino acid residues which are involved in the formation of a complex between the enzyme and phospholipid membranes may be present along either the longitudinal axis of the oligomer or be at the same side at which association takes place. These two possibilities are indicated in Fig. 7. Arrangement I would leave the sites involved in the association intact and any effect of a phospholipid membrane would be due to secondary conformational changes in the enzyme. Arrangement 11, however, would interfere with association directly and would result in competition between the two processes of association and lipid. protein complex formation. The stopped-flow experiments indicate that the associated form of the enzyme interacts with anionic phospholipid membranes and that this leads to dissociation of the enzyme aggregates into oligomer. lipid complexes. This it may be concluded that the oligomers in the enzyme. membrane complex are arranged as in scheme I1 (Fig. 7). The conformational changes induced in the oligomer by binding to the phospholipid membranes were detected by the extrinsic fluorescence studies. These conformational changes may prevent binding of an oligomer to the free side of an oligomer involved in an oligomer. membrane complex. In the case of dipalmitoyl-lecithin the area of each phospholipid head-group is about 4.8 nm with a constant width for each head-group layer of approximately 1 nm [15]. From the above data it may be calculated that the maximum area involving 4 polypeptide chains in arrangement I (Fig.7) is approximately equal to 120 nm2, which is equivalent to the surface area on a phospholipid vesicle occupied by 120 phospholipid molecules. In arrangement I1 the total surface area is 77 nm2 corresponding to 77 phospholipid molecules. If the enzyme just touches the surface of the membrane, the total area of contact would be very small and would, as a rough estimation, be equivalent to the area occupied by 6 or 4 phospholipid head groups for arrangements I and I1 respectively. With slight penetration of the polypeptide chain into the hydrophobic interior of phospholipid bilayers, these values increase to 36 and 11 phospholipid molecules. Besides its interactions with phospholipid membranes, the enzyme can also bind to the cardiolipin of the inner mitochondria1 membrane (unpublished results), thus making it a peripheral membrane protein. The affinity of the protein for the membrane is only moderately strong as measured in our conditions, with the consequence that in the matrix compartment, the enzyme would be distributed between the membrane and the soluble matrix fluid. Since we have demonstrated that several properties of the enzyme change as a result of the reversible association with membranes, we may consider glutamate dehydrogenase as an allotopic enzyme according to the original definition of Racker [16]. The enzyme interacts with the anionic phospholipids over a range of conditions and the affinity of the enzyme for the membranes is affected by substrates and regulators. These conclusions are in line with those of Godinot, who found that the association of IT 1 Fig. 7. Two possible urrungements of the glutamate dehydrogenuse oligomer when it is bound to a phospholipid membrane. The terms (a) and (b) are the axial ratios of the protomer [14]. The relative sizes of the protein and phospholipids are not to scale

9 M. Nemat-Gorgani and G. Dodd 141 the enzyme with both cardiolipin membranes and with inner-membrane-matrix mitochondrial fractions was affected by substrates and regulators [17]. Our results are also consistent with the hypothesis of Godinot and Lardy [4] that the enzyme, which is synthesised in the endoplasmic reticulum, may utilize specific interactions with both phosphatidylserine and cardiolipin during passage from the cytoplasmic compartment to the mitochondrial matrix. A comprehensive range of interactions can be expected on the basis of the studies on the interactions which have been performed so far. It is clear that only limited aspects of the allotopic behaviour of the enzyme may have been revealed, as instanced by the activation of the enzyme by a total mitochondrial phospholipid extract [18] and the inhibition induced by the pure anionic phospholipids. Thus the behaviour of the phospholipids investigated in this study may be expected to be modified when the lipids are in situ in the mitochondrial membranes. This work was supported by the Science Research Council. We thank Dr P. Moore for helpful advice and the use of his stoppedflow apparatus. REFERENCES 1. Moon,K. & Smith,E.L. (1973) J. Biol. Chem. 248, Steiner,R. E., Lippoldt,R.E., Edelhoch, H. & Frattali,V. (1964) Bipolymers Symp. no. 1, Kimelberg,H.K. & Papahadjopoulos,D. (1971) J. Biol. Chern. 246, Godinot,C. & Lardy,H. (1973) Biochemistry, 12, Bayley,P. & Radda, G. K. (1966) Biochem. J. 98, Lesslauer, W., Cain, J. E. & Blasie, J.K. (1972) Proc. Nut1 Acad. Sci. U.S.A. 69, Reisler,E., Pouyet, J. & Eisenberg,H. (1970) Biochemistry, 9, Huang,C. & Frieden,C. (1969) Proc. Nut1 Acad. Sci. U.S.A. 64, Das,M. L. & Crane,F. L. (1964) Biochemistry, 3, Austin,P., Dodd, G. H., Davis, M. & Leslie, R. B. (1974) Trans. Biochem. Soc. 2, Green,D.E. & Fleischer,S. (1963) Biochim. Biophys. Acta, 70, Margoliash,E. (1962) J. Bid. Chem. 237, Gitel,S.N., Owen, W. G., Esmon,C.T. & Jackson,C. M. (1973) Proc. Nut1 Acad. Sci. U.S.A. 70, Sund, H., Pilz, I. & Herbst, M. (1969) Eur. J. Biocheut. 7, Engelman,D.M. (1970) J. Mol. Biol. 47, Racker,E. (1967) Fed. Proc. 26, Godinot,C. (1973) Biochemistry, 12, Julliard,J. & Gautheron,D. (1972) FEBS Lett. 25, M. Nemat-Gorgani, Department of Biochemistry, National University of Iran, Tehran, Iran G. Dodd*, Department of Molecular Sciences, University of Warwick, Coventry, Warwickshire, Great Britain, CV4 7AL * To whom correspondence should be addressed