Cooperation between Enzyme and Transporter in the Inner Mitochondrial Membrane of Yeast REQUIREMENT FOR MITOCHONDRIAL CITRATE SYNTHASE FOR CITRATE AND MALATE TRANSPORT IN SACCHAROMYCES CEREVISIAE*

We have characterized acid-sensitive, mersalyl-insensitive citrate uptake from two Saccharomyces the course, and dependence, response to inhibitors. In unloaded mitochondria from PSY142 CS1-cells, a mutant that lacks mitochondrial citrate synthase, both citrate and efflux were reduced 7- and &fold, respectively, compared with the parental strain. No malate uptake cells, while in the strain, In mutations in peroxisomal citrate (CS2-) not in in mitochondrial citrate transport, for mitochon- citrate synthase in transport. protein of

Physical ordering of multienzyme complexes within metabolic pathways that catalyze sequential reactions has recently been recognized as a form of compartmentation of "soluble" enzymes that provides the pathway with the organization necessary for the most efficient use of substrates (for a review, see Ref. 1). Demonstration of discreet substrate labeling patterns (2) predict that the mitochondrial tricarboxylic acid cycle enzymes are organized into a complex. The term metabolon has been used to describe this form of compartmentation. Evidence has been presented that several of the tricarboxylic acid cycle enzymes are bound to the inner mitochondrial membrane (3). Grigorenko et al. (4) provided preliminary evidence that the loss of mitochondrial citrate synthase results in a reduction of citrate transport activity without a loss of citrate transporter ans Affairs, NIDDK Grant DK11313, National Science Foundation protein. Subsequent searches for mitochondrial membranebinding proteins using affinity chromatography have identified the citrate transporter (5) and the dicarboxylate carrier ( 6 ) as being capable of being bound to immobilized citrate synthase and malate dehydrogenase, respectively. These results suggest that the tricarboxylic acid cycle has a vectorial organization that includes transport of substrates across the mitochondrial membrane. Similar observations have been made for the transport and metabolism of glucose (71, carnitine (8), and ornithine (9).
We have begun a series of structure/function studies to probe further into the interactions of the tricarboxylic acid cycle metabolon and the citrate transporter. To accomplish this, we have measured the effects of mutations in tricarboxylic acid enzymes of Saccharomyces cerevisiae mitochondria on the uptake of citrate and malate into both intact mitochondria and mitochondrial protein extracts reconstituted into liposomes.
Media and Growing Conditions-Cells were grown overnight in 20 ml of YP (1% yeast extract, 2% Bacto-peptone) or selective medium containing 2% glucose with shaking at 30 "C. These cultures were transferred into 500 ml of YP with 2% galactose, and growth was allowed to continue for 24 h. All media were supplied with 20 m M potassium phosphate, pH 7.0, in order to buffer accumulating acetate when acetate-strains were grown.
Preparation ofMitochondria-The procedure for isolating mitochondria was essentially that of Daum et al. (11). Cells were collected and The abbreviation used is: BTC, 1,2,3-benzenetricarboxylic acid. washed with distilled water by centrifugation at 2800 x g for 5 min (4000 rpm; Beckman J-10 rotor). Cells were incubated in 0.1 M Tris, pH 9.4 (0.15-0.2 g (wet weight)/ml in this and subsequent steps), in the presence of 10 mM dithiothreitol for 15 min at 30 "C. Cells were then washed twice in 1.2 M sorbitol in 20 mM potassium phosphate, pH 7.4 (buffer A), and incubated at 30 "C in buffer A with 3 mg of lyticase/g of cells for 1 h. After 1 h, an additional 1 mg of lyticase/g of cells was added, and the digestion was continued for 1 additional h. Spheroplasts were collected by centrifugation; washed twice with 1.2 M sorbitol; and suspended in 0.6 M mannitol, 0.1% bovine serum albumin, 20 mM potassium phosphate, pH 7.2, supplemented with protease inhibitors (1 m~ benzamidine, 1 mM phenylmethylsulfonyl fluoride, 2 units of aprotinid100 ml) at 4 "C. Spheroplasts were homogenized twice with three strokes each in a tight fitting motor-driven Potter homogenizer (after the first three strokes, the solution was chilled on ice for 2 min). The homogenate was centrifuged at 7000 x g for 5 min (4000 rpm; Beckman 5-20 rotor). The supernatant solution was decanted and centrifuged in the same rotor at 17,500 x g for 5 min (12,000 rpm) to sediment yeast mitochondria.
Dunsport Studies on Intact Mitochondria-The mitochondrial pellet was resuspended in 0.55 M mannitol, 20 mM potassium phosphate, pH 6.8, 0.1% bovine serum albumin, 8 p~ rotenone, and 5 PM antimycin (buffer B). This suspension of mitochondria (at 10-20 mg of proteidml) was placed on ice, and transport was assayed immediately.
For uptake studies, 200 pl of buffer B supplemented with 1.5 mM [1,5-l4C1citrate (2000 cpdnmol) were equilibrated a t 30 "C in 1.5-ml Eppendorf tubes for 2 min. To this solution were added 100 pl of an ice-cold mitochondrial suspension, resulting in a 1 mM final citrate concentration. At the indicated times, or in the zero-time tubes before the addition of mitochondria, 100 p1 from the 160 mM BTC stock solution (40 mM final concentration) were added. Suspensions were diluted with 900 pl of ice-cold 40 mM BTC, and the tubes were gently vortexed and centrifuged in a n Eppendorf centrifuge (Model 5415) for 1 min at 14,000 rpm (8000 x g ) . The supernatant solution was decanted, the wall of the tube was wiped to remove residual radioactivity, and the pellet was dissolved in 100 pl of concentrated formic acid and transferred into vials for determination of radioactivity. Uptake was found to be linear for at least 30 s and was relatively insensitive to pH between pH 6 and 8. For malate uptake studies, the same procedure was employed with the exception that [U-l4C1malate was used as substrate and butyl malonate was the specific inhibitor.
In studies of citrate efflux, mitochondria were preloaded with 5 mM [1,5-14C]citrate (specific activity of 2000 cpdnmol) at 5 "C. To monitor loading, 100-p1 aliquots were taken, added to 900 pl of 40 mM ice-cold BTC, and centrifuged, and the radioactivity in the pellets was measured as described above. Mitochondria from the PSY142 parental strain were usually loaded for 10 min, while those from the CS1-mutant were loaded for 15 min. To measure citrate efflux, 100 pl of loaded mitochondria were added to 200 pl of buffer B pre-equilibrated at 30 "C for 2 min with or without 1 mM unlabeled citrate as specified. After 30 s, or in the zero-time tubes before the addition of mitochondria, 100 pl from the 160 mM BTC stock solution were added (40 mM final concentration). The tubes were centrifuged as described above, and 350 p1 of the supernatant were used to determine radioactivity. Extramitochondrial space was determined with [U-'4Clsucrose as a marker. Extraction of Mitochondria-Mitochondrial protein was solubilized with 50 mM NaCl, 20 mM HEPES, pH 7.2,5 m~ citrate, 4% Triton X-114, and 3 mg/ml cardiolipin a t a final protein concentration of -20-30 mg/ml. The suspension was incubated for 20 min on ice and centrifuged in a Beckman Ti-55 rotor at 40,000 rpm (110,000 x g) for 20 min, and the supernatant was saved. The protein yield of extraction was -25%, resulting in an extract with 6 7 mg of proteidml.
Reconstitution of Proteoliposomes-Proteoliposomes were prepared using a modification of the method of Kaplan et al. (12). Phospholipid (95% phosphatidylcholine, 5% cardiolipin) was suspended by vortexing at 80 mg/ml in buffer C (50 mM NaCl, 120 mM HEPES, pH 7.2, 1 mM EDTA) and was sonicated under a stream of N, in the center of a round bath sonicator (Model GllSSPIT, Laboratory Supplies Co. Inc.) until the suspension cleared. To 300 pl of this solution were added 150 pl of protein extract and 10 mM sodium citrate. The suspension was vortexed, frozen in liquid N,, and stored at -70 "C for no more than 24 h.
Citrate Dunsport into Proteoliposomes-Proteoliposomes (460 pV tube) were thawed in a water bath a t room temperature, sonicated twice for 2 s each, and applied to Dowex 1-X8 columns (100 mesh, 0.5 x 3.5 cm in a Pasteur pipette) that had been prewashed with 5 ml of buffer C. After elution with buffer C, a 0.8-ml aliquot of the cloudy, proteoliposome-containing fraction was divided into 2 0 0 4 aliquots and incubated for 2 min a t 25 "C. To these tubes were added 20 p1 of 0.1 mM [1,5-l4C1citrate (10,000 cpdnmol) to start transport. The reaction was quenched with 10 mM BTC (final concentration) after 5 min in the two sample tubes or by adding the BTC before the [1,5-14C]citrate to the two zero-time tubes. A 2 0 0 4 aliquot of each reaction mixture was then added to a Dowex 1-X8 column as described above and eluted with 1.3 ml of buffer C into scintillation vials. Protein Determination-Mitochondrial extracts were collected by centrifugation in a n Eppendorf centrifuge (14,000 rpm for 1 rnin), precipitated with 5 volumes of cold acetone (-20 "C), and recentrifuged. The pellets were suspended in 0.1 N NaOH, and protein was determined with the bicinchoninic acid assay (Pierce) using bovine serum albumin as a standard.

RESULTS
Citrate Uptake by Intact Mitochondria from the PSY142 Strain- Fig. 1 shows the citrate concentration dependence of citrate uptake by mitochondria isolated from the PSY142 strain of S. cereuisiae. From the data in Fig. L 4 , it is apparent that there are two kinetic components to this uptake. One shows a linear concentration dependence throughout the concentrations studied, while the other displays saturation kinetics. The nonsaturable uptake is similar to that seen with pyruvate uptake into yeast mitochondria (13) and is thought to be either a second, low affinity citrate transporter or nonspecific uptake of citrate. The saturable component of the curve has an apparent K, for citrate of 240 VM and a V, , , of 21.8 nmoV midmg of protein. This K, is an order of magnitude higher than that reported for rat liver mitochondria, and the V, , exceeds the rat liver value by a factor of 2 (14). Table I compares the effects of different inhibitors on citrate uptake into these mitochondria, and differences in the inhibitor sensitivity of yeast and rat liver mitochondrial citrate transport are noted. The concentrations of cy-cyano-3-hydroxycinnamic acid required to inhibit the yeast transporter are higher than those required for the rat liver transporter. This is con- BuMal, n-butylmalonic acid; PhSuc, phenylsuccinic acid; Cycin, a-cyano-3-hydroxycinnamic acid; Mars, mersalyl. sistent with the above finding that the yeast transporter has a lower affinity for citrate and thus should display a lower sensitivity for its competitor, a-cyano-3-hydroxycinnamic acid. Also, the yeast citrate transporter appears to be insensitive to mersalyl, in contrast to the citrate transporter from rat liver, which is sensitive to mersalyl both in the intact mitochondrion (15) and after reconstitution (12,16). These observations suggest that while the yeast citrate transporter may use the same mechanism as the mammalian transporter, there are kinetic differences that could reflect differences in the proteins from the two sources. Table I1 shows the citrate and malate transport effects of mutating several tricarboxylic acid cycle enzymes. Mitochondria from two different strains of S. cerevisiae show both citrate and malate transport inhibition when mitochondrial citrate synthase is deleted. The inhibition of citrate uptake cannot be due to the lack of intramitochondrial substrate for exchange by the transporter as a similar level of inhibition is observed when citrate efflux is measured in preloaded mitochondria from these CS1-cells (Table 111). More important, when the peroxisomal form of the enzyme is mutated, no effects are observed on mitochondrial citrate or malate uptake (PSY142 CS2-cells; Table 11). These findings suggest that it is exclusively the mitochondrial matrix form of citrate synthase that affects the function of the citrate transporter.

Citrate and Malate Dansport by Mitochondria from Cells with Mutations in Dicarboxylic Acid Cycle Enzymes-
Mutations in mitochondrial malate dehydrogenase, isocitrate dehydrogenase, and fumarase had little or no effect on citrate uptake, but, as might be expected from the work of LanGar-Benba et al. (61, the malate dehydrogenase mutant shows reduced malate transport (Table 11). Results of malate uptake experiments in these mutants reveal that disruption of either fumarase or citrate synthase, the immediately proximal or distal steps in the tricarboxylic acid cycle, seriously compromises malate transport ( Table IIj, while disruption of mitochondrial isocitrate dehydrogenase has no effect. Since the measurement made in Table I1    tracts were prepared from mitochondria of parental cells, CS1cells deficient in mitochondrial citrate synthase, and CS2-cells deficient in the peroxisomal form of citrate synthase. Reconstitution of these extracts into liposomes by freeze-thaw sonication shows that the transport activity of proteoliposomes from CS1-cells is indistinguishable from the rates obtained with the parental line or CS2-cells (Table IV). Malate uptake was also found to be independent of the cell type from which the extract was prepared (Table N). These data suggest that although transport is compromised in the intact mitochondrion, similar amounts of citrate transport protein with similar transport activity are extracted from the membranes of mitochondria with the CS1-mutation compared with the wild type and CS2cells. These data also suggest that disruption of the metabolic complex results in an alteration in transporter function without an alteration in protein levels.
Proof of this regulatory mechanism would be to observe modulation of citrate transport activity after the readdition of citrate synthase to the proteoliposome preparation. Unfortunately, the addition of mitochondrial citrate synthase had no effect on citrate uptake by proteoliposome preparations from either CS1-or parental cell mitochondria (data not shown). Since the extract from parental cell mitochondria contains 92% of the citrate synthase activity found in the intact organelle, the identical uptake rates found in proteoliposomes from both CS1-and parental cells suggest that loss of a structural requirement for interaction between citrate synthase and the transporter may occur during extraction and reconstitution of mitochondrial protein from the parent strain. In this regard, it should be remembered that the tricarboxylic acid complex has been shown to be exquisitely sensitive to ionic strength (3). Also, protein concentrations that produce volume-excluding effects in the intact mitochondrion, would enhance proteinprotein interactions, but are not possible to achieve with liposomes.
More important, it should be noted that malate uptake into intact mitochondria from CS1-cells was undetectable, while butyl malonate-sensitive malate uptake into proteoliposomes reconstituted from extracts of CS1mitochondria is virtually identical to the uptake into proteoliposomes from parental cell extracts (Table IV). This finding strongly supports the conclusion that the transporters are under different regulatory constraints in the intact organelle.

DISCUSSION
The mitochondrial citrate transporter catalyzes the exchange of citrate for citrate, malate, isocitrate, succinate, and phosphoenolpyruvate (17-19). The mammalian mitochondrial citrate transporter has been well studied (20, 211, and the rat liver transporter has been recently cloned (22). Here, we have characterized further the mitochondrial citrate transporter from S. cereuisiae, and subtle kinetic differences and inhibitor sensitivity differences suggest that the yeast protein may not be identical to the mammalian protein. Using this characterization of a system amenable to genetic manipulation, we have asked fundamental questions about the organization of mitochondrial membrane transport proteins and mitochondrial matrix enzymes.
We (23, 27-30) and others (24-26) have reported physical interactions between mitochondrial matrix enzymes in both mammalian and yeast cells. We (3) and others (26) have shown that these enzymes are associated with mitochondrial membrane-binding proteins and that these proteins could be the respective transport proteins for the enzyme substrates or products. Demonstration of these complexes would make metabolic pathways like the tricarboxylic acid cycle a closed loop system for substrates from the mitochondrial membrane surface through the pathway.
In this study, we have probed for functional associations of the mitochondrial citrate and malate transporters with tricarboxylic acid cycle enzymes by measuring citrate and malate transport in mitochondria with deletions in the tricarboxylic acid cycle enzymes. We have found that deficiencies in mitochondrial, but not cytosolic, citrate synthase result in severely impaired citrate uptake and efflux as well as malate uptake. Other mutations in tricarboxylic acid cycle enzymes had little or no effect on citrate uptake. Mutation of malate dehydrogenase did impair malate uptake, as would be expected from the work of LanGar-Benba et al. (61, who clearly demonstrated interaction of malate dehydrogenase with the dicarboxylate carrier of mitochondria when they were able to purify the dicarboxylate carrier with a malate dehydrogenase affinity column. Interestingly, disruption of nearest neighbor tricarboxylic acid cycle enzymes to malate dehydrogenase had a large impact on malate uptake, while no effect was observed with disruptions in more distal enzymes in the cycle.
When extracts of mitochondria from mutant strains were prepared and reconstituted into liposomes, citrate and malate uptake rates were indistinguishable from the rates observed with parental mitochondrial extracts. This finding indicates that there was no difference in the levels of transport protein in these respective mitochondria, but that there was regulation of the activity of the transporters. More important, it should be noted that while malate uptake was undetectable in the mitochondria from mutants lacking mitochondrial citrate synthase, butyl malonate-sensitive malonate uptake was the same in extracts from the mutant and parental strains in the proteoliposome.
These data support the previous demonstration that citrate synthase and fumarase associate with the malate dehydrogenase enzyme (25) and suggest that this complex associates with the citrate transporter and the dicarboxylate carrier in the mitochondrial membrane. Disruption of any member of the complex results in dissociation of the complex from the dicarboxylate carrier, while disruption of citrate synthase is required for dissociation of the complex from the citrate transporter. We propose that the vectorial organization of the tricarboxylic acid cycle metabolon includes the citrate transporter and the dicarboxylate carrier of the mitochondrial membrane.