Demonstration of Physical Interactions Between Consecutive Enzymes Of

Fur J Blochein. 117. 527-535 (1981) c FEBS 1981

Demonstration of Physical Interactions between Consecutive Enzymes of the Citric Acid Cycle and of the Aspartate-Malate Shuttle A Study Involving Fumarase, Malate Dehydrogenase, Citrate Synthase and Aspartate Aminotransferase

Sonia BEECKMANS and Louis KANAREK Laboratorium voor Chemie der Proteinen, Vrije Universiteit Brussel

(Received September 15, 198O/Fehruary 26, 1981)

By means of covalently immobilized fumarase and mitochondrial or cytoplasmic malate dehydrogenase we were able to detect physical interactions between different enzymes of the citric acid cycle (fumarase with malate dehydrogenase, malate dehydrogenase with citrate synthase and fumarase with citrate synthase) and between the enzymes of both mitochondrial and cytoplasmic halves of the aspartate-malate shuttle (aspartate aminotransferase and malate dehydrogenase). The interactions between fumarase and malate dehydrogenase were also investigated by immobilizing one enzyme indirectly through antibodies bound to Sepharose - protein A. Our results are consistent with a model in which maximally four molecules of malate dehydrogenase are bound to one fumarase molecule. This complex is able to bind either citrate synthase or aspartate aminotransferase. We propose that these enzymes bind alternatively, in order to allow the cell to perform citric acid cycle or shuttle reactions, according to its needs. The physiological meaning and implications on the regulation of metabolism of the existence of a large citric acid cycle/malate-aspartate shuttle multienzyme complex are discussed.

Whereas it was generally assumed for many years that the Several publications have appeared lately pointing indeed citric acid cycle enzymes are randomly dispersed in the mito- to the existence of physical interactions between enzymes of chondrial matrix, there are in recent literature various publi- the citric acid cycle and aspartate-malate shuttle [7- 141. cations predicting the organization of these enzymes as a Furthermore, a certain compartmentation of different mito- multienzyme complex [I -41. Such predictions were mainly chondria] matrix enzymes has been observed [15] pointing based on the observation that the concentration of free oxalo- to a loose association of certain of the citric acid cycle enzymes acetate is so low that the calculated rate of the citrate syn- with the inner membrane. thase reaction is much slower than the experimentally deter- In this publication we present further indications of real mined citric acid cycle rate as measured by mitochondrial physical interactions between the consecutive citric acid cycle oxygen consumption. Moreover, oxaloacetate further appears enzymes, fumarase, malate dehydrogenase, citrate synthase to be a key metabolite in other important metabolic routes and between the shuttle enzymes malate dehydrogenase and in mitochondria : besides its direct participation in the citric aspartate aminotransferase of both compartments. Fumarase acid cycle and aspartate-malate shuttle it also strongly regu- was included in our experiments as it could be the anchor: lates succinate dehydrogenase activity, whereas it is mainly it links the next enzymes of the citric acid cycle to succinate synthesized by pyruvate carboxylase. It is also important in dehydrogenase, the only enzyme which is located within the cytoplasm: besides operating in the other half of the aspar- mitochondrial inner membrane [I 61 and which connects the tate-malate shuttle, it is the starting product of gluconeo- citric acid cycle with the respiratory chain. Malate dehydro- genesis as the substrate of phosphoenolpyruvate carboxy- genase catalyzes the only citric acid cycle reaction with a kinase [5,6]. Thus channeling on oxaloacetate metabolism highly unfavorable equilibrium malat late]/ [oxaloacetate] and physical association of previously called ‘soluble en- 2 lo4 [17,18]; this forces the cell to make sure that the end zymes’ might be suggested in both cellular compartments. product, oxaloacetate, is removed extremely rapidly in order The advantage of the organization of the citric acid cycle to secure proper functioning of the cycle. Moreover, citrate enzymes as a complex would be the creation of a special synthase, the enzyme next to malate dehydrogenase, is con- microcnvironment around the cycle : the cell would acquire sidered to be the main control point of the cycle [19,20]. One the possibility to maintain a high flux of substrate through could imagine that, besides extensive control by different the cycle with a moderate number of intermediate molecules. substances on this enzyme itself, the fact of switching on and

__ ~~ off a physical interaction between malate dehydrogenase and Enzymes. Fumarase or fumarate hydratase (EC 4.2.1.2); malate citrate synthase would be an extra way of regulating the dehydrogenase (EC 1.1.1.37); aspartate aminotransferase (EC 2.6.1.1); cycle activity, especially with the equilibrium of the former citrate synthase (EC 4.1.3.7); aldolase or fructose-bisphosphate aldolase (EC 4.1.2.13); lysozyme (EC 3.2.2.17); succinate dehydrogenase (EC reaction lying far to the left. Moreover, many metabolites 1.3.99.1); pyruvate carhoxylase (EC 6.4.1.1); phosphoenolpyruvate car- which regulate the activity of citrate synthase in vitro (ATP, boxykinase (EC 4.1.1.32); argininosuccinatc lyase (EC 4.3.2.1); fumaryl- pyridine nucleotides and tricarboxylate compounds) have acetoacetase (EC 3.7.1.2); adenylosuccinate iyase (EC 4.3.2.2); glutamate been proven to be much less or even not at all effective dehydrogenase (EC 1.4.1.3). in vivo [21-241. Another question is how the cell manages 528 to separate the amount of oxaloacetate provided to flux enzyme which causes a rate of change of absorbance of through the cycle, from the oxaloacetate which has to operate 0.1 min-l at 25 "C under the experimental conditions. Specific in the shuttle. One possibility is the existence of two pools activities are 1300 units/mg for both mitochondrial aspartate of mitochondrial malate dehydrogenase, and consequently at aminotransferase from chicken and the cytoplasmic isoenzyme least two pools of oxaloacetate in the mitochondria. This from pig hearts; A:%",,,, = 14 1351. supposition is less probable, since the relative amounts of Citrate synthase activity was determined by titration of aspartate aminotransferase and cycle enzymes are constant the released sulphydryl groups of coenzyme A with 5,5'-di- in the mitochondria of all tissues in an organism [25,26]. thiobis(2-nitrobenzoate) as described by Moriyama and Srere Another possibility is regulation by promoting or inhibiting, [37]. One unit of citrate synthase is defined as the amount according to the needs of the cell, a direct interaction of of enzyme that catalyzes the formation of 1 pmol coenzyme A/ aspartate aminotransferase with the malate dehydrogenase min. The specific activity was 160 units/mg; A;% nm, cm = 16. involved in the cycle, thus forming a large combined citric acid cycle/aspartate-malate shuttle complex. Immobilization of Enzymes on Sepharose 48 Four different immobilized enzyme systems were designed MATERIALS AND METHODS with respectively pig fumarase, chicken fumarase, pig mitochondrial malate dehydrogenase and pig cytoplasmic malate Purifications of Enzymes dehydrogenase covalently coupled to Sepharose 4B. The en- Fumarase from pig hearts [27] and from chicken hearts zymes to be coupled (20 mg of each) were dialyzed first against [28] was prepared as described previously. Mitochondria1 0.01 M potassium phosphate pH 8.2 and afterwards against malate dehydrogenase and mitochondrial aspartate amino- 0.1 M sodium bicarbonate pH 8.2 containing 0.02 M L-malic transferase from chicken hearts were prepared by affinity acid in order to protect the enzymes during the coupling step. chromatography on Sepharose-pyromellitic acid and blue The enzyme solutions were finally adjusted to a concentration Sepharose as will be described elsewhere. Mitochondrial of 1 mg/ml. For each enzyme, 10 g (wet weight) of Sepharose malate dehydrogenase from pig hearts was obtained from 4B (which corresponds to about 10 ml of packed gel), exten- Sigma Chem. Comp. and freed of minor impurities by per- sively washed with distilled water, was suspended in 20ml forming the same affinity chromatographic steps as used for distilled water and activated with 2 g of cyanogen bromide, the chicken heart preparation. Cytoplasmic malate dehydro- suspended in a minimal volume of freshly distilled dimethyl- genase and cytoplasmic aspartate aniinotransferase from pig formamide, during 8 min according to the classical method hearts were from Sigma Chem. Comp. Citrate synthase from of Cuatrecasas [38]. The activated Sepharose was washed chicken hearts was purified as described elsewhere [28]. The quickly with 1 I of cold 0.1 M NaHC03 pH 8.2 and resus- enzyme from pig hearts was obtained from Sigma Chem. pended in the 20 ml enzyme solution. Coupling of the enzymes Comp. All these enzymes were stored at 4°C as ammonium was achieved by overnight shaking of the suspension at 4 "C. sulphate precipitate in 0.01 M potassium phosphate buffer The gels were well washed with 1 1 of 0.1 M NaHC03 and pH 7.3. stored at 4 'C in the presence of sodium azide (200 mg/l All enzymes were tested for purity by electrophoresis in buffer) in order to prevent bacterial growth. the presence of sodium dodecylsulphate [29], by electrophoresis at pH 8.3 according to Davis [30], by isoelectric Immunological Techniques focusing in Ampholine gradients pH 3.5 - 10 and by electrophoresis in 6 M urea at pH 3.2 [31] and in 6 M urea at Antibodies against pig heart fumarase and against pig pH 8.0 [32]. Moreover, in neither of these enzyme prepara- heart mitochondrial malate dehydrogenase were raised in tions could we detect any contaminating activity of other rabbits as follows. A solution of 2 mg/ml enzyme was prepared citric acid cycle or aspartate-malate shuttle enzymes. in 0.01 M potassium phosphate buffer pH 8.0 and an emulsion Lysozyme and aldolase were obtained from Boehringer. was made with an equal volume of complete Freund's adjuvant. The emulsified protein (1 or 2 ml) was injected intramus- Enzyme Activities cularly into the back of the rabbits. Booster injections were given subcutaneously 5 weeks later in incomplete Freund's Fumarase activity was determined spectrophotometrically adjuvant (1 mg/ml of antigen); each rabbit was injected with with L-malate as substrate [27]. One unit of activity is that 1 mg of enzyme. Antiserum was collected at weekly intervals, amount of enzyme which catalyzes the formation of 1 pmol starting five days after the booster injection and for three fumarate/min at 25°C. The specific activity of both chicken consecutive weeks. Booster injections were repeated monthly. and pig funiarase was 550 units/mg; A:i4nm,lcm = 5.1. Antisera were collected and the immunoglobulins were Malate dehydrogenase activity was determined according precipitated by addition of half the volume of saturated to Thorne et al. [33]. One unit activity was defined as the ammonium sulphate. The precipitate was washed with 0.33 amount of enzyme catalyzing the reduction of 1 pmol NAD/ saturated ammonium sulphate, dissolved in phosphate-buf- min at 25°C under the experimental conditions. Specific fered saline (0.01 M potassium phosphate pH 7.2, containing activities were 155 units/mg for mitochondrial malate de- 9 g NaCl/I) and dialyzed against the same buffer. hydrogenase; A:$",, Icm = 2.9, and 88 units/mg for the cyto- Specific antibodies to fumarase and malate dehydrogenase plasmic isoenzyme, At$nm,, Cm = 13.1 [34]. were isolated with the use of the immobilized enzymes. A Activity of aspartate aminotransferase was determined column of 4 g (wet weight) Sepharose-antigen, previously by direct measurement of oxaloacetate formation according equilibrated with phosphate-buffered saline, was loaded with to Banks et al. [35], except that 0.1 M Tris/HCl buffer pH 7.5 dialyzed IgG fractions and washed with phosphate-buffered was used instead of triethanolamine and 50 pM pyridoxal saline until no more protein emerged from the column. Anti- phosphate was added [36]; activity was measured in 1 ml of bodies were eluted with 0.1 M succinic acid (pH 2.5) and substrate solution. One unit of activity is that amount of directly mixed with cold 0.5 M potassium phosphate buffer pH 7.4 containing 9 g NaCl/l (no more than 1 vol. eluent possible to purify an equimolar amount of antibody. This was added to 1 vol. of this buffer) in order to immediately can be calculated from absorbance measurements at 280 nm stabilize the pH of the eluent above 6. The antibodies were and is based on a specific absorption value of A ;$ nm, Fm = 14 dialyzed against phosphate-buffered saline and finally con- for antibodies [43]. The addition of 9 g NaCl/I in the centrated to about I mg/ml by dialysis in Spectrapore mem- eluent was found to be of great importance for stabilization branes (exclusion limit of 3500 molecular weight) against 20 of the antibodies in such extreme pH conditions (pH 2.5). polyethylene glycol (10000) in phosphate-buffered saline. When using the procedure as described under Materials and Purified antibodies were stored at - 20 "C. The Sepharose- Methods, no precipitate could ever be detected upon un- antigen columns were regenerated by slowly passing 0.05 M freezing the antibodies, even several times. potassium phosphate buffer containing 0.05 M L-nialic acid Antibodies raised against pig fumarase crossreact with pH 7.3. They can be used at least four times without notable chicken fumarase as can be demonstrated by classical Ouchter- decrease in capacity. lony double-immunodiffusion techniques. Our anti-(pig mitochondrial malate dehydrogenase) antibodies crossreact readily Other Reagenls with chicken mitochondrial malate dehydrogenase, but no crossreactivity is seen with pig cytoplasmic malate dehydro- All reagents used were 'pro analysis' grade. genase. This observation is consistent with analoguous experi- Sepharose 4B and protein-A- Sepharose were obtained ments previously published by Kitto and Kaplan [34]. from Pharmacia Fine Chemicals. In all our immunodiffusion experiments one single preci- pitin band is observed, which is a supplementary proof of the homogeneity of the antigens used. No precipitation line RESULTS is observed when testing anti-fumarase or anti-(malate de- Enzymes Immobilized on Sepharose hydrogenase) antibodies against the other enzymes used in our study (pig and chicken citrate synthase, mitochondrial In the preparation of all four immobilized enzyme systems and cytoplasmic aspartate aminotransferase), again confirm- (pig and chicken fumarase, pig mitochondrial and cytoplasmic ing that these enzymes are completely devoid of contaminating malate dehydrogenase) no more than 0.5% of the enzyme fumarase or mitochondrial malate dehydrogenase. remained unbound and could be detected in the washing buffer, by protein absorption measurement at 280 nm and by enzyme activity measurement. We thus assume that our gels Enzyme-Enzyme Interactions Studied ivith Directly Immobilized Enzymes contain 2 mg enzyme bound/g (wet weight) of Sepharose. The specific activity of the immobilized enzymes was measured Small columns were made, containing 2 g of Sepharose spectrophotometrically under the same conditions as used for with immobilized enzyme (2 mg/g gel). The columns were the native enzymes by suspending and carefully mixing 5 mg equilibrated with 1 mM potassium phosphate buffer pH 7.3 of gel in 3 ml substrate solution. Immobilized pig and chicken containing 14 mM 2-mercaptoethanol. Other enzymes from fumarase both have specific activities of 150 units/mg which the citric acid cycle or aspartate-malate shuttle which were corresponds to 25 of the native enzyme activity. The much tested for binding on these columns, were dialyzed against lower specific activity of immobilized pig mitochondrial and the same phosphate buffer and mixed with the immobilized cytoplasmic malate dehydrogenase (4.7 units/mg, which means enzyme. After 30 min of incubation, during which the slurry only 3 of the activity of the native mitochondrial and 5 was gently shaken from time to time in order to allow an of the native cytoplasmic enzyme) was proven recently [39] efficient contact between bound and free enzymes, the ap- to be due to diffusional limitations acting upon the coenzyme plied solution was slowly passed through the gel in the NADH and is not a concequence of structural modifications column and the gel was washed with 1 mM potassium phos- in the enzyme. phate buffer pH 7.3 containing 14 mM 2-mercaptoethanol In our experimental conditions of immobilization, the until no more activity emerged from the column. The necessary enzymes are coupled to the Sepharose matrix via multipoint washing volumes ranged from 12 ml to 60 ml, depending on attachment. Indeed, it is known that fumarase [40] as well as the enzyme system under consideration. Bound enzyme was malate dehydrogenase [41,42] are readily denatured and finally eluted with 0.01 M potassium phosphate, 0.01 M dissociated by the action of 6 M urea, with concomitant loss L-malate pH 7.3 and containing 14 mM 2-mercaptoethanol. of enzyme activity. 1 g of the Sepharose gels, each containing Results of these studies are given in Tables 1 and 2. As an 2mg of enzyme, was incubated for 90min in 1 mM po- example, chromatography of chicken mitochondrial malate tassium phosphate buffer pH 7.3 in the presence of 6 M urea dehydrogenase on immobilized chicken fumarase is given in at 25°C. From absorption measurements of the eluate we Fig. 1. It can be seen that (a) the amount of malate dehydro- calculated that about 0.1 ing fumarase (i.e. 5 was released genase bound on fumarase is independent of the amount of and less than 0.1 mg of both mitochondria1 and cytoplasmic dehydrogenase which is loaded on the column, (b) the bound inalate dehydrogenases. malate dehydrogenase remains on the column even aftcr We may conclude that in our immobilized enzyme prepa- extensive washing (150 column volumes!) and (c) equal rations an average of only one subunit per five fuinarase amounts of malate dehydrogenase are eluted when L-malate tetramers remained free of direct covalent binding to the in the elution buffer is replaced by L-aspartate, a substance matrix, and less than one subunit per ten malate dehydro- which is not inhibitory for fumarase [44], or when 0.1 M genase dimers. potassium phosphate buffer pH 7.3 or phosphate-buffered saline both containing 14 mM 2-mercaptoethanol, is used without addition of any other ligand. From (a) and (b) it can Purification of Antibodies be concluded that the interactions between the enzymes are On 4 g of Sepharose 4B, containing 8 mg of immobilized specific. This is further confirmed by the absence of intcr- pig fumarase or mitochondrial malate dehydrogenase, it was action of aldolase and lysozynie with the immobilized enzymes. 5 30

Table 1. Binding studies of different enzymes on immobilizedpigfumarase and chicken fumarase In all cases, saturating amounts of enzyme (ranging over 0.5-3 mg) were applied on the columns. The amount of enzyme bound on immobilized fumarase was determined as activity found in the elution fraction (see text for details). The sum of unbound and eluted enzyme activity was always equal to the amount of enzyme applied on the columns. n.d. = not determined

Enzyme applied Source of enzyme Amount of enzyme on column bound on immobilized fumarase from

Pig chicken p 201 mg

______.~~ Malate pig mitochondria 0.25 0.22 dehydrogenase chicken mitochondria 0.22 0.27 0 l0ILJbI I I I I I I I I I I I t8 IIi pig cytoplasm 0 0 0 4 8 12 300 304 308 Volume (ml) Citrate synthase Pig 0.27 0.30 chicken 0.23 0.28 Fig. 1. Binding of chicken mitochondrial malate dehydrogenase on im- Aspartate chicken mitochondria 0 0 mobilized chicken fumarase. A column containing 2 g Sepharose-fumarase aminotransferase pig cytoplasm 0 0 was used (containing 4 mg immobilized fumarasej. (A) Chromatography of 1 mg chicken mitochondrial malate dehydrogenase. The column was Malate chicken mitochondria n.d. 0.18 washed with buffer until no more activity emerged. (B) Chromatography dehydrogenase 0.13 of 0.25 mg chicken mitochondrial malate dehydrogenase. (Cj Chro- + aspartate matography of 0.50 mg chicken mitochondrial malate dehydrogenase. aminotransferase The column was extensively washed with 150 column volumes of buffer. Elution was started as indicated by the arrow. Elution buffer was either Aldolase 0 0 0.01 M potassium phosphate buffer containing 0.01 M L-malate or Lysozyme 0 0 L-aspartate, or 0.1 M potassium phosphate buffer, or phosphate-buffered saline; all contained 14 mM 2-mercaptoethanol

Table 2. Binding studies of different enzymes on immobilized pig mitochondrial and cytoplasmic matale dehydrogenase absence of interaction of all the different enzymes used in Enzyme applied Source of enzyme Amount of enzyme our study with Sepharose which has undergone the activation on column bound on step with cyanogen bromide but which was further washed immobilized with and left overnight in 0.1 M NaHC03 buffer pH 8.2 pig malate without coupling any ligand. This rules out aspecific adsorp- dehydrogenase tion of the enzymes on eventually residual active groups re- maining on the Sepharose matrix after the coupling step. mito- cyto- From (c) we may conclude that elution of bound enzymes by chondrial plasmic L-malate is a consequence of raising the ionic strength of the elution buffer, rather than this substance playing a more mg - specific role in the elution as an intermediate product of the Fumarase Pig 0 0 citric acid cycle. chicken 0 0 It is also important to notice from Table 1 that, although chicken mitochondrial aspartate aminotransferase does not Citrate synthase pig 0.28 0.23 chicken 0.26 0.18 interact directly with immobilized chicken fumarase, it can be bound on the column when it is applied as a mixture with Aspartate amino- chicken mitochondria 0.20 0.10 chicken mitochondrial malate dehydrogenase. In this experi- transferase pig cytoplasm 0.14 0.46 ment, equal amounts of 0.2 mg malate dehydrogenase and Aldolase 0 0 0.2 mg aspartate aminotransferase were incubated with the gel. Lysozyme 0 0 Further studies of complex formation involving more than two enzymes were performed by adding, one after another, consecutive enzymes from the citric acid cycle and finally the shuttle enzyme aspartate aminotransferase. One experiment Aldolase, as a member of the glycolytic pathway and itself is summarized in Table 3. The column with immobilized pig involved in a larger multienzyme complex 145,461 (and refer- fumarase was loaded first with pig mitochondrial malate ences mentioned herein), should not be expected to have dehydrogenase under the same conditions as described above affinity for one of the citric acid cycle or shuttle enzymes. and rinsed with 1 mM potassium phosphate buffer pH 7.3 Lysozyme was chosen as blank because of its extremely high containing 14 mM 2-mercaptoethanol until no more activity PI value (ll), which approximates to the PI values of mito- could be detected in the washings. Instead of eluting malate chondrial malate dehydrogenase (PI 9) and aspartate amino- dehydrogenase, we now applied pig citrate synthase onto the transferase (PI 9.5). Specificity was also confirmed by the column. Again, the column was washed with buffer until no 531

Table 3. Consecutive binding of different enzymes on immobilized pig.fumamse The enzymes were applied in the order given in the first column of the table. The amounts of enzymes as expressed in nanomolar quantities were calculated assuming molecular weights of 72000 for malate dehydrogenase (dimer of 36000), 92000 for citrate synthase (dimer of 46000), 86000 for aspartate aminotransferase (dimer of 43 000) and 194000 for fumarase (tetramer of 48 500); subunit molecular weights were determined by electrophoresis in the presence of sodium dodecylsulphate [20]

Enzyme applied on column Amount bound on immobilized pig fuamarase

-~ -~ ~ -~----~ ~__ -~ -~ ~- number name maqs of enzyme molar quantity of enzymc

-~ ~-__ -~- __ ~ ~ 1 2 3 1 2 3

mg nmol

-~--~ ~ __ ~ 1 Pig mitochondrial malate dehydrogenase 0 250 - - 1 47 - - -7 Pig cilrate synthdse 0 250 0 150 - 3 47 163 - 3 Chicken mitochondrial aspartate aminotransferase 0 170 0 105 0 345 2 36 114 175

Table 4 Consec utive hinding of different enzyme7 on immobilized pig mitochondrial malute dehjdrogennte Details as in Table 3

Enzyme applied on column Amount bound on immobilized pig mitochondrial malate dehydrogenase

~~ _____ -~____ ~_-~-~__~~ number name mass of enzyme molar quantity of enzyme

-~ - - __ __ - ~ - 1 2 3 1 2 3

mg nmol - -~__- , 1 Pig cltrdte synthdse 0 275 - - 2 99 - - 2 Pig fumarase 0 275 0 440 - 2 99 2 27 - 3 Chicken mitochondrial aspartate ammotiansferase 0 170 0 145 0 125 185 0 75 145

Table 5. Consecutive hinding of difjerent enzymes on immobilized chicken.fumarase Details as in Table 3

Enzyme applied on column Amount bound on immobilized chicken fumarase

- ~-~- -~ __ _- -- ~- -- ~~ .~~~ number name inass of enzyme molar quantity of enzyme _-__ 1 2 3 1 2 3

mg nmol ____~___- -~ - 1 Chicken mitochondrial malate dehydrogenase 0 290 - - 4 03 - - 2 Chicken mitochondria1 aspartate aminotransferase 0 200 0 240 - 2 78 2 79 - 3 Chicken citrate synthase 0 200 0 110 0 210 2 78 128 2 28 2 0 200 0 145 0 170 2 78 1 69 185

more activity emerged. No bound malate dehydrogenase was it was seen from Table 2 that pig fumarase can not be bound lost from the column by the addition of citrate synthase. directly on immobilized mitochondrial malate dehydrogenase, Finally we applied mitochondrial aspartate aminotransferase as much as 0.44 mg is adsorbed when citrate synthase is first on this column. It can be noticed from Table 3 that addition applied on the column. Part of the bound citrate synthase of the transaminase causes loss of part of the bound malate and fumarase are displaced when mitochondrial aspartate dehydrogenase and citrate synthase from the column. Here aminotransferase is applied as last enzyme on the column. again, mitochondrial aspartate aminotransferase can be bound Here too, the initially bound citrate synthase and aspartate on the pig fumarase column by complex formation with aminotransferase are in good agreement with the data from adsorbed mitochondrial malate dehydrogenase. Finally all Table 2. enzymes are eluted together with 0.01 M potassium phos- Table 5 summarizes the results of consecutive binding of phate, 0.01 M L-malate, pH 7.3 and containing 14 mM different enzymes on immobilized chicken fumarase, but 2-mercaptoethanol. The sums of eluted and displaced enzyme applied in a different order to the previously described experi- are in good agreement with the data given in Table 1. ment of Table 3. After adsorption of mitochondrial malate An analoguous experiment on immobilized pig mitochon- dehydrogenase, we first loaded mitochondrial aspartate amino- drial malate dehydrogenase is summarized in Table 4. Whereas transferase onto this column. This results again in a loss of 532

Table 6. (A) Binding of’ mitochondrial malate dehydrogenase on fumarase, itself immobilized via anti- (pig fumarase) antibodies on Sepharose - protein A. (B) Binding of fumarase on mitochondrial mulute dehydrogenase, itself immobilized via anti-(pig mitochondria1 malute dehydrogenase) anitbodies See text for experimental details. Nanomolar quantities were calculated as in Table 3

Antibodies loaded on Antigen immobilized Enzyme adsorbed protein-A - Sepharose

(nmol) mg nmol mg ninol A. Anti-(pig fumarase) (6) fumarase from mitochondrkal malate dehydrogenase from Pig 1.22 6.29 Pig 1.20 16.67 chicken 0.32 1.65 chicken 0.37 5.14

B. Anti-(pig mitochondrial mitochondrial malate fumarase from malate dehydrogenase) ‘6) dehydrogenase from Pig 0.44 6.11 Pig 0.52 2.68 chicken 0.24 3.33 chicken 0.27 1.39

part of the bound malate dehydrogenase. When citrate syn- chondrial malate dehydrogenase is more important (55 %) thase was subsequently applied onto the gel, the transaminase than between pig and chicken fumarase (26%). The pig was in turn partly displaced. When we finally incubated the antigens bind to the antibodies with a 1 : 1 molar ratio. gel with the transaminase, no more malate dehydrogenase The columns were quickly equilibrated with 0.01 M Tris came off, while a fraction of the adsorbed citrate synthase buffer adjusted to pH 7.3 with acetic acid, and tested for appeared to be replaced by transaminase. When expressing binding of free enzyme (free mitochondria1 malate dehydro- these results in molar quantities, it may be noticed that the genase on immobilized fumarase and vice versa), dialyzed sum of adsorbed aspartate aminotransferase and citrate syn- against the same Tris/acetdte buffer. After 15 min of incuba- thase is a constant value. tion, during which the slurry was gently shaken, the columns were washed with Tris/acetate buffer until no more activity emerged, and finally eluted .with phosphate-buffered saline. Enzyme- Enzyme Interactions Studied Results of these experiments are summarized in the right side with Enzymes lmmobilized Indirectly by Antibodies of Table 6. Citrate synthase could not be tested in this system Interactions between fumarase and mitochondrial malate because of its instability in the low-ionic-strength Tris/acetate dehydrogenase were also established by a slightly different buffer. experimental approach. Instead of covalently linking the enzymes to Sepharose, we immobilized them indirectly via Tentative Molecular Organization antibodies, which were in turn immobilized by interaction the Multienzyme Complex with protein A coupled on a Sepharose matrix. It is known of that binding of IgG molecules on protein A occurs through From the experiments with fumarase and mitochondrial their Fc part, thus leaving the antibody combining site malate dehydrogenase immobilized indirectly through anti- available for interaction with antigen [47,48]. The obvious bodies, we may attempt to predict the stoichiometry of part advantage of such a system lies in the fact that this indirect of the multienzyme complex. Indeed, in these experiments and gentle method excludes any modification of the enzyme we largely overloaded the antibody gels with antigen. So we brought about by its immobilization. A disadvantage is the can assume that every enzyme molecule is immobilized on much greater restriction in buffer systems that can be used the gel via only one very limited part, thus leaving all the in this approach : 2-1nercaptoethanol, always added in the rest of the molecule available for binding other enzymes. In former experiments to stabilize the enzymes against thiol contrast, the enzymes directly attached via cyanogen bromide modifications, has to be omitted here in order to avoid activation are linked to the Sepharose matrix through multiple destruction of the antibodies; furthermore, higher salt con- covalent bonds, consequently probably masking a large centrations which svabilize the antibodies can not be used number of binding sites for the other enzymes. When fumarase because they prevent formation of the enzyme complexes. is immobilized via anti-(pig fumarase) antibodies, we see that We found an acceptable compromise by using 0.01 M Tris/ about 3 mol mitochondrial malate dehydrogenase are bound/ acetate buffer pH 7.3 in these studies. mol immobilized fumarase. Assuming that one of the four For each column, 100 mg of dry protein-A-Sepharose fumarase monomers is used for binding to the antibody, we was allowed to swell in phosphate-buffered saline and was predict that fumarase has four binding regions available to loaded with 1 mg pure anti-fumarase or anti-(mitochondria1 malate dehydrogenase (schematically shown in Fig.2A). On malate dehydrogenase) antibodies. These antibody gels were the other side, when mitochondrial malate dehydrogenase is then loaded with respectively pig or chicken fumarase and attached via anti-(pig malate dehydrogenase) antibodies, one with pig or chicken mitochondrial malate dehydrogenase. fumarase molecule can be bound per two immobilized malate Table 6 gives the amounts of antigen that can he bound on dehydrogenase molecules. As one molecule of protein A is these antibody gels. They were calculated from the amount able to bind two antibody molecules [49,50], we may assume of unbound protein, which was estimated from absorption that two malate dehydrogenase molecules are close enough and enzyme activity measurements of the washings. It can together on the column to allow an average binding of one be noticed that crossreactivity between pig and chicken mito- fumarase molecule (see Fig. 2B). 533

A that no malate dehydrogenase is displaced when citrate syn- Protein A Antibody thase is introduced before aspartate aminotransferase suggests that a situation is possible where one molecule fumarase binds four molecules malate dehydrogenase plus a certain, as yet undetermined, amount of citrate synthase, whereas ggkarase this docs not seem to be possible with the transaminase. Mitochondrial malate OO dehydrogenase

0 DISCUSSION II Mitochondria1 malate In our experiments, we clearly demonstrate the existencc of physical interactions between enzymes from the citric acid cycle and the aspartate-malate shuttle. Our studies with directly immobilized fumarasc show the existence of interactions of this enzyme with mitochondrial malate dehydrogenase and also with the next enzyme citrate synthase. No recognition is observed between fumarase and Fig. 2. (A) Hypothetical arrangement qf a complex of furnurase with mitochondrial malate dehydrogenase when fumarase is immobilized in- cytoplasmic malate dehydrogenase, although fumarase is directly on Sepharose -protein A via anti-fumarase antibodies. (Bj Hvpo- also partly located outside the mitochondria [51 J, However, thetical arrangement of the same complex with the dehydrogenase im- its role in cytoplasm is mainly the conversion of fumarate, inohilized indirectly on Sepharo.Fr -protein A via anti- (malate dehydro- which is generated by argininosuccinate lyase in the urea genase) antibodies. These representations are exclusively schematical cycle [52], by fumarylacetoacetase in the catabolism of tyrosine [53,54] and by adenylosuccinate lyase in the purine nucleotide synthesis [55]. The reaction product L-malate can then be Mitochondrial aspartate ammotransferase transported into the mitochondria via specific carriers, in contrast with fumarate for which the inner membrane is a%O Mitochondrial malate deh yd rogenase impermeable [56], or it may serve as start product for glu- 00 coneogenesis. Cytoplasmic malate dehydrogenase, on the other hand, forms part of the extramitochondrial half of the aspartate-malate shuttle and no direct benefit for the cell of complex formation between fumarase and cytoplasmic malate dehydrogenase can be expected. Fig. 3. Hypothetical and .schematical arrangement of a ternary complex Although no direct interactions are seen between fumarase oj ,furnurase, mitochondria1 malate ciehydrogenase and mitochondrial and mitochondrial aspartate aminotransferase, a ternary usparlate amino1 ransf erase complex of fumarase, malate dehydrogenase and aspartate aminotransferase can be formed. Thus the possibility exists of binding both fumarase and aspartate aminotransferase on Further stoichiometric prognoses can be made from the malate dehydrogenase at the same time, which makes the binding of mitochondrial aspartate aminotransferase on existence of a combined citric acid cyde/shuttle complex malate dehydrogenase, which is itself adsorbed on covalently indeed more probable. immobilized fumarase. Indeed, as the aminotransferase has When mitochondrial malate dehydrogenase is covalently no direct affinity for fumarase, every molecule of the trans- immobilized, interactions between this enzyme and citrate aminase must be bound on a molecule malate dehydrogenase, synthase are readily observed. Interactions with fumarase can whose surface is completely free except for the part already not be revealed in this system. The most obvious explanation used in the complex formation with fumarase. It can be is that an amino group on the surface of mitochondrial malate seen from the experiments that addition of aspartate amino- dehydrogenase, being localized in a domain essential for transferase on such a column results in desorption of part of binding of fumarase, is used by the immobilization on the the bound malate dehydrogenase and leaves finally 1 mol Sepharose matrix. The interactions between the different aminotransferase/mol malate dehydrogenase. Our hypothesis enzymes seem indeed to have mostly an electrostatic nature, is that binding of mitochondrial aspartate aminotransferase since they all can be broken by higher ionic strength. The destabilizes the interaction of fumarase with two of the four involvement of electrostatic interactions in the loose binding originally bound malate dchydrogenase molecules, possibly of some mitochondrial matrix enzymes to the inner membranc as a consequence of steric hindrance, so that we now can has already been proposed by Addink et al. [57]. However, represent the situation schematically as in Fig. 3. When intro- when immobilized mitochondrial malate dehydrogenase was ducing citrate synthase upon this complex, part of the trans- first loaded with citrate synthase, a ternary complex can aminase is displaced. When introducing the transaminase readily be formed with fumarase. again, citrate synthase is in turn partly displaced, whereas The ionic strength, effective in dissociating the enzymes no more malate dehydrogenase is lost this time. The sum in our experiments, at first sight does not exceed ‘physio- of bound transaminase and citrate synthase is constant. This logical’ values. However, it was calculated by Srere [4] that suggests that the niultienzyme complex can ‘shuttle’ between protein molecules in mitochondria occupy at least half the more aspartate aminotransferase bound and more citrate matrix space. As a direct consequence of such high enzyme synthase bound. This alternative binding of these two enzymes concentrations one can expect that most metabolites will be may have a physiological meaning in changing the direction bound by different enzymes, water molecules will be ordered of oxaloacetate towards the citric acid cycle or towards the in layers and free metabolites will be divided into compart- aspartate-malate shuttle upon the needs of the cell. The fact ments throughout the matrix. Moreover, citric acid cycle 534 multienzyme complexes are expected to occur in the immediate carboxylic acids or palmitoyl-coenzyme A, either the inter- vicinity of the hydrophobic mitochondrial inner membrane, action of glutamate dehydrogenase with malate dehydro- succinate dehydrogenase being well burried within it. So it genase or glutamate dehydrogenase with aspartate amino- is only possible to measure overallmitochondrial ionic strength transferase would be favored [68]. Especially in liver tissue, and metabolite concentrations, but it is very difficult to the possibility of forming these alternative enzyme-enzyme predict conditions at one specific location. Also, as a con- complexes might determine, depending on the levels of sequence of the unusual cellular protein concentrations, ‘even different ligands, whether a cell will perform mainly the citric small associative forces between protein molecules would be acid cycle reactions, which implies that malate dehydro- favored so that weak interactions, not apparent in dilute genase interacts with citrate synthase, or whether it will mainly solutions, might become important’, as was stated by Srere [4]. perform amino acid catabolism and carbohydrate anabolism, Concerning the enzymes of the aspartate-malate shuttle, which implies that complexes between glutamate dehydro- important binding of mitochondrial aspartate aminotrans- genase and malate dehydrogenase or transaminase should be ferase and of the cytoplasmic aminotransferase on immobilized formed. This situation interrupts the circularity of the citric mitochondrial and cytoplasmic malate dehydrogenase respec- acid cycle : L-malate leaves the mitochondria for gluconeo- tively is observed. However, a certain, although much smaller, genesis instead of being converted further to oxaloacetate amount of interaction is also seen between one cytoplasmic and picked up by citrate synthase. However, at this time citric and one mitochondrial isoenzyme. Such interaction was not acid cycle intermediates are amply replenished by the de- observed by Backman and Johansson using the counter- gradation of amino acids. current distribution technique [9]. This supports the idea that In all our studies, it appears that interactions are observed mitochondrial and cytoplasmic malate dehydrogenase origi- between a chicken and a pig enzyme, whenever there is an nate from one single gene, as well as mitochondrial and cyto- interaction between the corresponding two enzymes prepared plasmic aspartate aminotransferase [58]. Although no im- from the same species. Once again, we may conclude that munological crossreactivity is seen between the cytoplasmic the determinants on the surface of these enzymes involved and the mitochondrial forms of both these enzymes [34,58, in the interactions have been very resistant to mutation 591, it seems that some parts of their surface were not affected throughout the course of evolution. In their study about by mutation. The same surface residues in the cytoplasmic enzyme complexes involving glutamate dehydrogenase, Fahien and mitochondrial isoenzymes seem to be maintained in the and coworkers also observed that the interactions remain interaction between malate dehydrogenase and aspartate detectable when the enzymes are extracted from different aminotransferase. The same arguments can be used to explain species (pig, rat and bovine) [69]. It was even observed by interactions seen between citrate synthase and cytoplasmic Srere that citrate synthase from Escherichia coli can replace malate dehydrogenase. These interactions have evidently no the enzyme from pig heart in the interaction with pig heart physiological meaning since citrate synthase is an exclusively mitochondrial malate dehydrogenase [7]. It seems that the mitochondrial enzyme [60]. Lack of crossreactivity between outer surfaces of the aspartate-malate shuttle and citric acid mitochondrial and cytoplasmic isoenzymes of aspartate ami- cycle enzymes, at least those used in our study, are as con- notransferase was explained by Sonderegger et al. on the basis servative as the inner contact surfaces of the subunits of of too little sequence homology between these two proteins oligomeric enzymes, which can very often be exchanged be- (less than 60%) [59]. No sequence data are available to tween species [70 - 731. estimate the degree of homology between mitochondrial and It is not impossible that the Sepharose matrix itself plays cytoplasmic malate dehydrogenase. Interesting is the fact that a certain role in stabilizing the enzyme-enzyme interactions the evolution rate of the cytosolic isoenzyme of the trans- seen in our experiments. It was quoted by Laurent and co- aminase is higher than the rate of the mitochondrial iso- workers [74,75] that polymers like polyethylene glycol and enzyme, as was reported in the literature [58,59]. It was sug- dextran affect the solubility of certain proteins, which was gested that more evolutionary constraints operate on the explained by steric exclusion of the protein from part of the mitochondrial isoenzyme. This would certainly be the case solvent. It was already noted by Backman et al. [9] that the if in mitochondria aspartate aminotransferase is superposed specific interactions of pig mitochondrial malate dehydro- on a larger multienzyme citric acid cycle complex, as we genase with aspartate aminotransferase and of pig cyto- suggest. Besides, it is striking to note throughout the literature plasmic malate dehydrogenase with aspartate aminotrans- that, in general, physicochemical properties of all vertebrate ferase could be observed only in a biphasic water/dextran/ citric acid cycle and aspartate-malate shuttle enzymes are so trimethylamino-polyethylene glycol system and not in buffer. much a like from one species to another (as exemplified in Interactions between citrate synthase and mitochondrial our unpublished study on chicken and pig fumarase. malate dehydrogenase were also revealed by Halper and Srere According to a series of different and very diverse [7] through specific precipitation in polyethylene glycol. experiments [61- 691, Fahien and coworkers studied the Finally it was stated [8] that associations between citric acid existence of multienzyme complexes between glutamate de- cycle enzymes could not be detected either by analytical ultra- hydrogenase, on the one hand, and mitochondrial malate de- centrifugation, nor by simple gel filtration on Bio-Gel. Yet, hydrogenase or aspartate aminotransferase (and other amino- we made the interactions visible by immobilizing one of the transfemses, both in mitochondria and cytoplasm) on the enzymes on Sepharose, which is a polydextran. other. In consequence, even more evolutionary restrictions In summary, our results point to the existence of one large can be expected on both mitochondrial malate dehydrogenase citric acid cycle/aspartate-malate shuttle complex. We suggest and aspartate aminotransferase. No interaction of glutamate that the stoichiometry of the different enzymes in this complex dehydrogenase was observed with citrate synthase [@I. It might be regulated by local concentrations of different me- was stated that in mitochondria one hexamer of glutamate tabolites, so channeling oxaloacetate towards the citric acid dehydrogenase can bind either three dimers of malate dehydro- cycle or towards the aspartate-malate shuttle. However, genase or three dimers of aspartate aminotransferase [69]. further experiments with different metabolites have to be Depending on the presence of different nucleotides, di- performed in order to test this last hypothesis. 535

We thank L. Holderbeke for tcchnical assistance. This work was 33. Thorne, C. J. R., Grossman, C. I. & Kaplan. N. 0. (1963) Biochini. supported by a grant of Belgian Government (Ondwling Ovcrlegde Bi~ph)',~.Ac~u, 73, 193 - 203. Onderzoeksacties) and Fonds voor Kollelciicf Fundnrncnterl 0nder;oc.k 34. Kitlo, G. B. & Kaplan, N. 0. (1966) Bioc17mzi,~try,5, 3966-3980. (contract 2.0021.79). 35. Banks, B. E. C., Doonan, S., Lawrence, A. J. & Vernon, L. A. (1968) Eur. J. Biochem. 5, 528 - 539. 36. Bergmeyer, H. U. (1974) Meth0d.s of Enzyncitic. Anul,v.si.s,2nd edn, vol. 2, p. 727, Academic Press, Ncw York. REFERENCES 37. Moriyama, T. & Srere, P. A. (1971) J. Bid. Chem. 246, 321 7 3223. 1. Srere, P. A. (1972) in Energy Metabolism and the Regulation qfkfeta- 38. Cuatrecasas, P. (1970) J. Biol. Chem. 245, 3059-3065. holic Processes in Mitochondria (Mehlman, M. A. & Hanson. R. 39. Miiller, J. & Pfleiderer, G. (1980) Hoppe-Seyler's Z. Phy.viol. Chrin. W., cds) pp. 79-91, Academic Press, New York. 361,675-680, 2. Srere, P. A. & Mosbach, K. (1974) Annu. Rev. Microbiol. 28,61-83. 40. Hill, R. L. & Kanarek, L. (1964) Brookhaven Symp. Biol. 17, XO-97. 3. Srere, P. A. & Henslee, J. G. (1980) in Cell Compartmenlatiort and 41. Jaenicke, R., Rudolph, R. & Heider, I. (1979) Biochemi.stry, 18, Metabolic Channeling (Nover, L., Lyncn, F. & Mothes, K., eds) 1217-1223. pp. 159 - 168, F.lsevier/North-Holland Biomedical Press, Amster- 42. Chilson, 0. P., Kitto, G. B. & Kaplan, N. 0. (1965) Pro(.. Null dam. Acad. Sci. USA, 53, 1006-1014. 4. Srere, P. A. (19x0) Trends Biochem. Sci. 5, 120-121. 43. Fuchs, S. & Givol, D. (1968) FEES Lett. 2, 45-48. 5. Sols, A. ~9:Marco, R. (1 970) Curr. Top. Cell. Re,&. 2, 227 - 273. 44. Massey, V. (1953) Biochem. J. 55, 172-177. 6. Ackrell, B. A. C. (1974) Horizons in Biochemistry and Biophysic,s, 45. Grazi, E. & Troinbetta, G. (1980) Eur. J. Biochcm. 107, 369-373. 1,175-219. 46. Batke, J., Asboth, G., Lakatos, S., Schmitt, B. & Cohen, R. (1980) 7. Halper, L. A. & Srere, P. A. (1977) Arch. Biuchem. Biopl7.p.s. 154, Eur. J. Biochem. 107,389- 394. 529 - 534. 47. Forsgren, A. & Sjoquist, J. (1966) J. Imniunol. Y7, 822-827. 8. Srere, P. A., Halper, L. A. & Finkelstein, M. B. (1978) in Micro- 48. Kronvall, G. & Frommel, D. (1970) Inirnunochernistr~J,7, 124- 127. environnzents rind Metabolic Compartmentation (Srere, P. A. & 49. Sjoquist, J., Meloun, €3. & Hjclm, H. (1972) Lur. J. Uiochern. 29, Estabrook, R. W., eds) pp. 419-432, Academic Press, New York. 572- 578. 9. Backman, L. & Johansson, G. (1976) FEBS Lett. 65, 39-43. SO. Hjelm, H., Sjodahl, J. & Sjoquist, J. (1975) Eur. J. Biorhrm. 57. 10. Backman, L., Shanbag, V. & Johansson, G. (1977) Biochem. Soc. 395-403. TYUFIS.5,748-750. 51. Kuff, E. L. (1954) J. Bid. Chem. 207, 361 -365. 11. Bryce, C. F. A., Williams, D. C., John, R. A. & Fasclla, P. (1976) 52. Ratncr, S., Anslow, W. P. & Petrack, B. (1953) J. Biol. C'hrm. 204, Biochem. .I. 153, 571 -577. 115-125. 12. Srere, P. A., Mattiasson, B. & Mosbach, K. (1973) Proc. Nail Acad. 53. Ravdin, R. G. & Crandall, D. I. (1951) J. Bid. Chcw. 189, 137- 149. Sci. USA, 70, 2534-2538. 54. Knox, W. E. & Leinay-Knox, M. (1951) Biochern. J. 49, 686-693. 13. Koch-Schmidt, A.-C., Mattiasson, B. & Mosbach, K. (1977) Eur. 55. IIartman, S. C. (1970) in Metabolic Pathways (Greenberg, D. M., J. Biochem. 81, 71 -78. ed.) voI. IV, pp. 1 - 58, Academic Press, New York. 14. Erekin, N. & Friedman, M. E. (1979) J. Solid-Phase Biochem. 4, 56. Passarella, S. & Quagliariello, E. (1976) Biorhirnir iPori.si 58, 123-130. 989-1001. 15. Matlib, M. A. & O'Brien, P. J. (1975) Arch. Biochem. Biophys. 167, 57. Addink, A. D. F., Boer, P., Wakabayashi, T. & Green, D. E. (1972)

193 - 202. Eur. J. Biochem. 29. 47-59. 16. Ernster, L. & Kuylenstierna, B. (1970) in M~~mhranesof Mitochon- 58. Sonderegger, P. &Christen, P. (1978) Nature (I-ond.) 275, 157- 159. dria and Chloroplasts (Racker, E., ed.) pp. 172- 212, Van Nostrand 59. Sonderegger, P., Gehring, H. & Christcn, P. (1977) .J. Bid. C'hcm. Reinhold Comp, Princeton, NJ. 252, 609-612. 17. Lehninger, A. L. (1975) Biochemistry, 2nd edn, pp. 460-461, Worth 60. Srere, P. A. (1972) Curr. Top. Cell. Regul. 5, 229-283. Publ. Inc., New York. 61. Fahien, L. A. & Smith, S. E. (1969) Arch. Biochem. Brophys. 135, 18. Newsholme, E. A. & Start, C. (1973) Regulation in Metabolism, 136-151. pp. 94 and 319, John Wiley, New York. 62. Shemisa, 0. A,, Happy, J. M. & Fahien, L. A. (1972) J. Biol. C'hcnz. 19. Newsholme, E. A. & Start, C. (1973) Regulation in MPtaholism, 247, 7556-7569. pp. 132-136 and pp. 316-323, John Wiley, New York. 63. Fahien, L. A. & Smith, S. E. (1974) J. Bid. C'hcm. 249, 2696-2703. 20. Srere, P. A. (1974) Ljfi Sci. 45, 1695-1710. 64. Fahien, L. A. & Van Engelen, D. L. (1976) Arch. Brocheni. Biop1t.v.s. 21. Weitzinan, P. D. J. & Hewson, J. K. (1973) FEBS Let/. 36, 227- 231. 176,298-305. 22. Matlib, M. A,, Shannon, W. A. & Srere, P. A. (1977) Arch. Bio- 65. Fahien, L. A., Hsu, S. L. & Kmiotek, E. (1977) J. Bhl. C'hcm. 252, chenz. Biophvs. 178, 396-407. 1250 - 1256. 23. Matlib, M. A,, Boesman-Finkelstein, M. & Srere, P. A. (1978) Arch. 66. Fahien, L. A,, Ruoho, A. E. & Kmiotek, E. (1978) J. BifJl. C'hem. Biochem. Bioph-vs. 191, 426-430. 253, 5745-5751. 24. Matlib, M. A., Shannon, W. A. & Srere, P. A. (1979) Method.s 67. Fahien, L. A. & Kmiotek, E. (1978) in Microenvironments and Enzvniol. 56, 544 - 550. Metabolic Compnrtmmiation (Srere, P. A. & Eslabrook, R. W., 25. Pette, D.. Klingenberg, M. & Biicher, T. (1962) Biochem. Biophy~. eds) pp. 355-368, Academic Press, New York. Res. Comtnurz. 7, 425 - 429. 68. Fahien, L. A,, Kiniotek, E. & Smith, I,. (1979) Awh. Biochcn7. 26. Pettc, D. (1966) in Regulation of Metabolic Processes in Mito- Biophys. 192, 33-46. c,/7ondria (Tager, J. M., Papa, S.. Quagliariello, E. & Slater, E. C., 69. Fahien, L. A. & Kmiotek, E. (1979) J. Bid. Chem 254, 5983-5990.

eds) pp. 28 - 50, Elsevier, Amsterdam. 70. Levinthal, C., Signer, E. 8: Fetherolf, K. (1962) Proc. Nu// Accrd. 27. Beeckmans, S. & Kanarek, L. (1977) Eur. J. Biochem. 78, 437-4444, Sci. USA, 48, 1230- 1237. 28. Beeckmans, S. & Kanarek, L. (1979) Arch. Int. Physiol. Biochitn. 71. Zabin, I. & Villarejo, M. (1975) Annu. Rev. Bioc.hem. 44, 295-3313, 88, B13-BI5. 72. Feldmann, K., Zeisel, H. J. & Helmreich, E. J. M. (1976) Eur. J. 29. Laemrnli, U. K. (1970) Muturr (L-ond.) 227, 680-685. Biochem. 65, 285- 291. 30. Davis, B. J. (1964) Ann. N.Y. Acad. Sci. 121, 404-427. 73. IIeib, J. A. & Lebherz, H. G. (1978) J. Bid. C'hcm. 253, 6599-6605. 31. Panyim, S. 81 Chalklcy, R. (1969) Arch. Biochem. Bioph-vs. 130, 74. Laurent, T. C. (1963) Biochem. J. 89, 253-257. 337 - 346. 75. Iverius, P. H. & Laurcnt, T. C. (1967) Biochim. Biophys. Acta, 133, 32. Creighton, T. E. (1979) J. Mol. Biol. 129, 235-264. 371 -373.

S. Beeckinans and L. Kanarek, Laboratorium voor Chemie der Proteinen, Instituut voor Molcculairc Biologie, Vrije Universitcit (tc) Brusscl, Paardestraat 65, 8-1640 Sint-Genesius-Rode, Belgium