Binding of Citrate Synthase to Mitochondrial Inner Membranes*

Citrate synthase and other mitochondrial matrix proteins bind to the inner surface of the mitochondrial inner membrane. No binding was observed to the outer membrane or to the outer surface of the inner membrane. When mitochondria are disrupted, and the membranes are removed by centrifugation, it is usually observed that all the citrate synthase, aconitase, isocitrate dehydrogenase, fumarase, and malate dehydrogenase remain in the supernatant fraction (matrix) as soluble enzymes. Pyruvate dehydrogenase and a-ketoglutarate dehydrogenase complexes remain par-tially bound’ (1,2) to the inner membrane, but can be removed easily (1, 2). Succinate dehydrogenase remains firmly bound to the inner membrane (3) and is sometimes considered an intrinsic membrane component, although some workers have shown its removal from the membrane by more drastic (but nondetergent) treatment (4).

Citrate synthase and other mitochondrial matrix proteins bind to the inner surface of the mitochondrial inner membrane. No binding was observed to the outer membrane or to the outer surface of the inner membrane.
When mitochondria are disrupted, and the membranes are removed by centrifugation, it is usually observed that all the citrate synthase, aconitase, isocitrate dehydrogenase, fumarase, and malate dehydrogenase remain in the supernatant fraction (matrix) as soluble enzymes. Pyruvate dehydrogenase and a-ketoglutarate dehydrogenase complexes remain partially bound' (1,2) to the inner membrane, but can be removed easily (1, 2). Succinate dehydrogenase remains firmly bound to the inner membrane (3) and is sometimes considered an intrinsic membrane component, although some workers have shown its removal from the membrane by more drastic (but nondetergent) treatment (4).
There have been a number of conflicting reports concerning the association of matrix Krebs cycle enzymes with the mitochondrial inner membranes. Several groups have studied this problem using either differential digitonin extraction of mitochondria, or measuring the enzyme activity remaining with the isolated inner membrane fraction. Some malate dehydrogenase of pig heart (5) and of chicken liver (6) has been reported to be bound to mitochondrial inner membranes based on isolation techniques. These latter workers also reported that fumarase, glutamate dehydrogenase, aspartate aminotransferase, and lactate dehydrogenase were also bound to the inner membrane, but to a lesser extent than the malate dehydrogenase. Aspartate aminotransferase has been reported to be f i i l y bound to guinea pig heart inner membranes (7-9). By measuring the loss of latency of rat liver mitochondrial enzyme activities at different digitonin concentrations, Matlib and O'Brien (2) concluded that fumarase was located close to the inner membrane and malate dehydrogenase was located away from the inner membrane. Among the mitochondrial enzymes similar to fumarase were NAD-isocitrate dehydrogenase, a-ketoglutarate dehydrogenase, and pyruvate dehydrogenase. Citrate synthase, NADP-isocitrate dehydrogenase, * This work was supported by Grant PCM 7904007 from the National Science Foundation and funding by the Veterans Administration Research Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' T. C. Linn, personal communication. and aspartate aminotransferase behaved like malate dehydrogenase.
Wit-Peeters et UI!. (7) concluded that for a number of matrix enzymes, "no clear distinction can be made between membrane-bound and soluble matrix enzymes" but the enzymes only differed in the tightness of their binding to the inner membrane. On the other hand, Landriscina et al. (10) have reported that NADP-isocitrate dehydrogenase, aspartate aminotransferase, glutamate dehydrogenase, and NAD-isocitrate dehydrogenase are isolated in a membrane-free fraction. Thus, studies reported so far have not provided unequivocal evidence for or against specific interactions of these enzymes with the mitochondrial inner membrane.
Our studies of the inner membrane-matrix compartment of mitochondria have led us to postulate that an organization of Krebs cycle enzymes exists (11, 12). Histochemical studies (13) and cross-linking experiments (14) indicated that citrate synthase was located near the inner surface of the inner membrane. These results have prompted us to look for specific interactions between some of the so-called "soluble" Krebs cycle enzymes and the mitochondrial inner membrane. This communication provides evidence for a specific interaction between citrate synthase and two other Krebs cycle enzymes and the matrix surface of the inner membrane of rat liver mitochondria.

EXPERIMENTAL PROCEDURES
Materials-The following chemicals were obtained from the indicated sources: NADH, NADP, L-malic acid, DTNB? bovine serum albumin, yeast glucose-6-phosphate dehydrogenase, and yeast alcohol dehydrogenase from Sigma; oxalacetic acid from Calbiochem; CoA from P-L Biochemicals; pig heart mitochondrial malate dehydrogenase, pig heart citrate synthase from Boehringer-Mannheim, pig heart cytosolic malate dehydrogenase from Miles Laboratories; and trypsin from Worthington. Yeast citrate synthase was purified in this laboratory (15). CoSAc was made from CoA and acetic anhydride (16). The enzymes, usually obtained as ammonium sulfate suspension were dialyzed against 2 mM Hepes buffer, pH 7.0.
Methods-Citrate (si)-synthase (EC 4. Mitochondria were isolated from livers of fasted (24 h) male Sprague Dawley rats (200-250 g, body weight) by the method described earlier (24, 25). Mitoplasts were prepared using digitonin as described by Greenawalt (26). Cytosolic protein was obtained by centrifugation of the liver homogenate at 144,OOO X g for 60 min. The inverted inner membrane vesicles were obtained by sonication as described by . The amounts of membranes or organelles are given in terms of pg or mg of proteins. Matrix proteins were obtained by lysis of mitoplasts by freeze-thawing in an acetone-solid COZ mixture. Lysate was centrifuged at 144,OOO x g for 60 min and the supernatant solution obtained was used as the matrix fraction. The inner membranes were washed extensively and cytosol and matrix fractions were dialyzed against 2 mM Hepes buffer, pH 7.0. For the binding studies, enzyme samples and membranes (700 pg) were incubated for 15 min at 0 "C in (400 pl) 2 mM Hepes buffer, 0.5 mM dithiothreitol, pH 7.0. The membranes were sedimented by centrifuging for 30 min in an Airfuge centrifuge at 30 p.s.i. and were washed with the same buffer. The final pellet was then resuspended in the original volume of the above buffer. Enzyme activities in the different fractions were measured. Over 95% recovery of added enzyme activity was always found. The protein bound in pg was calculated from the specific activities of the added enzymes.

RESULTS AND DISCUSSION
When inner membrane vesicles are isolated by the sonication procedure used here (27), one obtains inside out vesicles whose sidedness can be estimated by measuring the activities of cytochrome c oxidase (EC 1.9.3.1) and cytochrome c reductase (EC 1.6.99.3) (28). Vesicles prepared in this laboratory were about 70% inside out. These vesicles contain no bound citrate synthase, malate dehydrogenase, fumarase, or aspartate amino transferase activity. When pure citrate synthase is added to this preparation, binding to these vesicles occurs and the process is saturable (Fig. 1). Binding of malate dehydrogenase occurs to a greater extent (Table I), and this process is also saturable (Fig. 1). In addition, it can be seen (Fig. 1) that malate dehydrogenase reduces the amount of citrate synthase that can be bound.
The binding of proteins to this preparation is specific in that a series of pure non-matrix proteins, including cytosolic malate dehydrogenase, does not bind to these vesicles (Table  I). In addition, it has been noted that cytosolic rat liver proteins, as a group, bind less well than does matrix proteins (Table 11). Our results also show that, whereas the isozymes of malate dehydrogenase, fumarase, and aspartate aminotransferase present in the added matrix proteins bind quite well to the vesicles, the isozymes of these enzymes present in the cytosol fraction do not bind as well (Table 11).
The binding is decreased by an increase in ionic strength (Fig. 2). In addition to KC1 shown here, we have found that NaCl, potassium phosphate, Tris-acetate, and CaCh also remove bound enzyme from the membranes with approximately the same efficiency. Binding is also decreased by an increase in the pH of the medium (Fig. 3). The same effect of pH change is observed when 2 m~ Tris-HC1 is used as a buffer in place of 2 mM Hepes. Since the ionic strength change with change of pH is opposite for these two buffers, then the effect is due to a change in pH and not due to change in ionic strength. A variety of treatments of the membrane were carried out in an attempt to characterize the membrane component(s) responsible for the binding of the Krebs cycle enzymes (Table 111). Trypsin treatment decreased the binding somewhat, (30%), but further treatment of the membranes with trypsin caused a dissolution of the membranes. Extraction of the membranes with 0.5 M KCl, which removes some membrane-associated proteins as shown by gel electrophoresis of this fraction (data not shown), or with toluene, which removes some membrane lipids (29), did not affect the binding capacity of the membranes. Acetic acid and urea are solvents which have been employed to remove extrinsic but firmly held inner membrane proteins (4,30). Treatment of the membranes with these solvents markedly reduced the binding capacity of the membranes. However, it is possible that these solvents irreversibly denatured the binding sites instead of extracting them.
Finally, we were able to demonstrate the membrane specificity and orientation of the binding site in the inner mem-

TABLE I1
Interaction of rat liver mitochondrial inner membranes with enzymespresent in matrix and cytosolic fractions Rat liver mitochondrial inner membrane (700 pg) and 1 mg of matrix ( m ) or cytosolic ( c ) proteins in 400 pl of 2 mM Hepes, pH 7.0, and 0.5 mM dithiothreitol were incubated at 0 "C for 15 min. The membranes were reisolated by centrifugation, washed, and enzyme activities were estimated as described in the text. The values represent mean f S.E. of three experiments.

TABLE I Interaction ofpure enzymes with rat liver mitochondrial inner membranes
Rat liver mitochondrial inner membranes (700 pg) and pure enzymes (100 pg) in 400 pl of 2 mM Hepes (pH 7.0) and 0.5 mM dithiothreitol were incubated at 0 "C for 15 min. The membranes were reisolated by centrifugation, washed, and enzyme activities were estimated as described in the text. Over 95% of added enzyme activities were brane by testing the binding of citrate synthase to intact mitochondria, mitoplasts, and the inner membrane. No binding to mitochondria (outer membranes) or to mitoplasts ("outside out" inner membranes) was found. Thus, the binding sites are located on the matrix side (inside) of the inner membranes (Fig. 4). Treatment of the inner membranes with digitonin (0.12 mg/mg of protein) did not effect their ability to bind citrate synthase. One can calculate, based on the citrate synthase content of the mitoplasts (about 6 pg/700 pg Rat liver mitochondrial inner membranes (3 mg) were incubated in 200 pl of 2 mM Hepes buffer, pH 7.0, containing 45 pg of trypsin for 10 min at room temperature. The proteolysis was stopped by the addition of trypsin inhibitor.
Rat liver mitochondrial inner membranes (3 mg) were treated with 200 p. l of 2% cold toluene and incubated in ice for 5 min. pg) was incubated with either mitochondria (7.3 mg), mitoplast (2.4 mg), or inner membrane (700 pg) in a medium (400 pl) containing 220 mM mannitol, 70 mM sucrose, and 2 mM Hepes, pH 7.0, for 15 min at 0 "C. The mixture was centrifuged and washed, and citrate synthase activity in different preparations was estimated. In this experiment, the quantity of mitochondria used was chosen so that the total amount of outer membrane was about 700 pg to be equivalent with the quantity of inner membranes used. Similarly, the amount of mitoplasts used contained about 700 pg of inner membrane.
of inner membrane), that the binding capacity of the inner membrane for that enzyme is about 7 times as great (about 36 pg/700 pg of inner membrane) (Fig. 4). At the normal concentration of citrate synthase then, the membranes bind about 30-50% of the total citrate synthase. The evidence for, and the metabolic implications of, enzyme binding to membranes has been thoroughly reviewed (31). There are similarities between certain aspects of the binding of the Krebs cycle enzymes to the inner membrane and the binding of glycolytic enzymes to red blood cell membranes (32-34). In the latter case, sensitivity to high ionic strength is observed as well as competition between enzymes (32). Kliman and Steck (35) have pointed out that a consideration of mass action makes it likely that the poor binding of glyceraldehyde 3-phosphate to red cell ghosts at isotonic ionic strength is an