Adenine Nucleotide and Phosphate Transport Systems of Mitochondria RELATIVE LOCATION OF SULFHYDRYL GROUPS BASED ON THE USE OF THE NOVEL FLUORESCENT PROBE EOSIN-5-MALEIMIDE*

Eosin-5-maleimide is impermeable to the inner mi- tochondrial membrane, exhibiting essentially no reactivity with matrix glutathione or with j3-hydroxybu-tyrate dehydrogenase located on the matrix surface of the inner membrane. In intact mitochondria, eosin-5-maleimide is unreac- tive with the ADP/ATP antiporter even under conditions which promote maximal labeling by N-[3H]ethyl- maleimide (Le., ADP present). However, eosin-5-mal-eimide readily labels the ADP/ATP antiporter in “in- ver,ted” inner membrane vesicles even in the presence of N-ethylmaleimide. Labeling is prevented if the ves- ides are prepared from mitochondria pretreated with carboxyatractyloside. In contrast to the ADP/ATP antiporter, essential sulfhydryl groups of the Pi/H+ symporter are accessible to eosin-5-maleimide in intact mitochondria with op-timal inhibition of phosphate transport being observed at 25 “C. Eosin-5-maleimide also prevents labeling of the Pi/H+ symporter by N-[3H]ethylmaleimide. These results show that essential sulfhydryl groups of the ADP/ATP antiporter and the Pi/H+ symporter have differing reactivities and locations in functionally intact

Eosin-5-maleimide is impermeable to the inner mitochondrial membrane, exhibiting essentially no reactivity with matrix glutathione or with j3-hydroxybutyrate dehydrogenase located on the matrix surface of the inner membrane.
In intact mitochondria, eosin-5-maleimide is unreactive with the ADP/ATP antiporter even under conditions which promote maximal labeling by N-[3H]ethylmaleimide (Le., ADP present). However, eosin-5-maleimide readily labels the ADP/ATP antiporter in "inver,ted" inner membrane vesicles even in the presence of N-ethylmaleimide. Labeling is prevented if the vesides are prepared from mitochondria pretreated with carboxyatractyloside.
In contrast to the ADP/ATP antiporter, essential sulfhydryl groups of the Pi/H+ symporter are accessible to eosin-5-maleimide in intact mitochondria with optimal inhibition of phosphate transport being observed at 25 "C. Eosin-5-maleimide also prevents labeling of the Pi/H+ symporter by N-[3H]ethylmaleimide.
These results show that essential sulfhydryl groups of the ADP/ATP antiporter and the Pi/H+ symporter have differing reactivities and locations in functionally intact mitochondria. With respect to eosin-5-maleimide, sulfhydryl groups of the ADP/ATP carrier occur in two distinct classes, both of which are inaccessible in intact mitochondria. Only one class, depending on conditions, can be exposed in submitochondrial particles. In contrast, sulfhydryl group(s) of the Pi/H+ symporter behave as a single reactive class which is readily accessible in mitochondria at 2 5 "C.
The ADP/ATP antiport and Pi/H' symport systems are essential components of the mitochondrial phosphorylating assembly. They facilitate the transport of ADP and inorganic phosphate into mitochondria and thus enable ADP to be phosphorylated (1-3). These two transport systems are clearly the most studied of the mitochondrial carriers. Both have been extensively characterized (1-7) and successfully isolated in a reconstitutively active state (8-12).
Although the ADP/ATP antiporter and PJH' symporter are different proteins (9, 11, 13), they possess several similar properties. Both are hydrophobic, integral (8, 11) membrane *This work was supported by Grant PCM 8300772 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Present address: Czechoslovak Academy of Sciences, Institute of Physiology, 142 20 Praha 4-KRC Budgjovicka 1083, Czechoslovakia. proteins consisting of subunits with similar molecular weights. The subunit monomer of the ADP/ATP antiporter is close to 30,000 (9, 14), with two monomers in dimeric form constituting the functional carrier unit (15). The subunit of the PJH' symporter has a molecular weight somewhat higher, 33,000-35,000 (11-13, 16, 17), and a dimeric functional unit has been suggested (13). Both carriers can be solubilized by the nonionic detergent Triton X-100, and in both cases the major step of purification is adsorption chromatography on hydroxylapatite (8,9,11).
An important feature of the ADP/ATP antiporter and Pi/ H' symporter is that both are sulfhydryl-containing proteins with one or more of these sulfhydryl groups being required for function. The Pi/H" symporter was one of the first carrier proteins found to be highly sensitive to chemical modification of sulfhydryl groups (18, 19). Phosphate transport can be inhibited by a number of sulfhydryl group reagents, e.g., mercurials, disulfides, and maleimides (4). Their inhibitory effect is relatively specific since sulfhydryl groups of the Pi/ H' symporter represent perhaps the most reactive class of sulfhydryl groups in intact mitochondria (4, 20). Although the total number of sulfhydryl groups associated with the Pi/H' symporter is still unknown, two types have been proposed (21)(22)(23). It is suggested that these reactive sulfhydryl groups may be localized near the outer surface of the inner membrane in intact mitochondria (21-23).
The ADP/ATP antiporter was originally thought to be insensitive to sulfhydryl-reactive reagents. Sensitivity, however, can be demonstrated if mitochondria are preincubated with a low concentration of either ADP or ATP (24)(25)(26). Under these conditions, the sulfhydryl groups are unmasked and become reactive toward N-ethylmaleimide. As a result, the function of the carrier is inhibited (24-26). The location of these sulfhydryl groups is unclear, but they do not seem to change orientation from one side of the membrane to the other (27).
In the characterization of protein sulfhydryl groups, maleimides have proven to be especially useful. Because the reaction of maleimides with sulfhydryl groups of proteins is not readily reversed, proteins containing sulfhydryl groups can be radioactively labeled and monitored by gel electrophoresis. Unfortunately, the most commonly used radioactive maleimide, N-ethylmaleimide, is highly permeable (4). Maleimide analogs of low permeability, like carboxylated maleimides (22) or N-acetyl-4-sulfamoylphenylmaleimide (28), are either not available in radioactive form, or they are not completely impermeable. However, another maleimide analog, eosin-Ei-maleimide, is fluorescent, relatively large in size, and bears several charged groups. Hence, a limited permeability might be expected. Significantly, this compound has been reported recently to react readily with the ADP/ATP antiporter in submitochondrial particles with the protein-bound eosin-5-maleimide being detected easily by fluorography (29).
With the above thoughts in mind, it was of interest to establish whether eosin-5-maleimide was, in fact, impermeable to the mitochondrial inner membrane and, if so, whether the use of this probe could provide insight into the relative topographies of the PJH' symporter and ADP/ATP antiporter in intact mitochondria.
Mitoplasts and Membrane Vesicles--Rat liver mitoplasts were prepared according to the procedure of Schnaitman and Greenawalt (31). Rat liver inner membrane vesicles were prepared according to the procedure of Wehrle et al. (32).
Bovine Heart Mitochondria-Bovine heart mitochondria were prepared according to the procedure of Smith (33) in STE medium containing 250 mM sucrose, 10 mM Tris, 1 mM EDTA, pH 7.4 (HCI). Such mitochondria routinely exhibited acceptor control ratios of 4.2-5.3 with P-hydroxybutyrate as substrate.
Submitochondrial Particles-Bovine heart submitochondrial particles were prepared according to the procedure of Hansen and Smith (34).
Glutathione was measured according to Tietze (35). The assay contained: 0.1 M NaPi, 5 mM EDTA, 0.6 mM DTNB, 10 pg, of glutathione reductase/ml, 0.2 mM NADPH, and an aliquot of mltochondrial extract. The change in absorbance at 412 nm was measured for 4 min after the reaction was initiated by NADPH and the AA of control (without extract) was subtracted. The AA uersus glutathione concentration plot was linear up to 0.4 nmol of glutathione. @-Hydroxybutyrate Dehydrogenase Actiuity-This activity was assayed polarographically in 1 ml of H medium (without bovine albumin), supplemented with 10 mM P-hydroxybutyrate, 2.5 mM KPi, 1 CIM FCCP, 1 mg of mitochondrial or submitochondrial particle protein, and, in the case of submitochondrial particles, 75 p~ NAD+. Incubations with Eosin-5-maleimide-Incubations of mitochondria, submitochondrial particles, or mitoplasts with eosin-5-maleimide were performed routinely at 0 "C in H medium without bovine albumin (rat liver) or in STE medium (bovine heart) at a protein concentration of 10 mg/ml with 150 pM eosin-5-maleimide for 20 min in the dark. (Eosin-5-maleimide was dissolved in dimethylsulfoxide to give a 10-15 mM stock solution.) The reaction was terminated by addition of sufficient 2 M dithiothreitol to give a final concentration of 50 mM. After an additional 10 min the labeled membranes (3-4 mg) were washed twice with 10-12 ml of H or STE medium.
Labeling of the ADPIATP Antiporter by N-fHlEth~lmaleimide-Labeling was performed in a manner similar to that previously described (26). Mitochondria (1 mg/ml) in H medium without bovine albumin, containing 5 mM succinate, 5 r g of oligomycin/m~, were incubated at 25 "C. After 2 min, 100 p~ N-ethylmaleimide was added, followed after 5 min by N-[3H]ethylmaleimide with or without 100 p~ ADP. The reaction was stopped after 6 min with dithiothreitol. Where indicated, 100 p~ carboxyatractyloside was added 1 min before N-[3H]ethylmaleimide.
Labeling of the PJH+ Symporter by N-rHlEthylmaleimide in Mitochondria-This labeling was performed in a manner similar to that described previously (36). Mitochondria (5 mg/ml) were incubated a t 0 "C for 1 min with 10 nmol of mersalyl/mg of protein, then for 5 min with 100 nmol of N-ethylmaleimidelmg of protein. The reaction was stopped with dithiothreitol (500 nmol/mg of protein) and mitochondria were washed twice and resuspended in H medium without bovine albumin. The labeling with N-[3H]ethylmaleimide (20 nmol/mg of protein) was performed for 2 min with or without preincubation with various SH reagents.
Phosphate Transport-Phosphate transport in isolated mitochondria was measured spectrophotometrically a t 25 "C by monitoring thedecrease in absorbance a t 540 nm in a 3-ml volume containing 1 mg of mitochondrial protein, 120 mM NH,Pi, 1 mM EDTA, 0.05 p M rotenone, pH 7.2.
SDS-Polyacrylamide Gel Electrophoresis-This was carried out according to the procedure of Laemmli (37) in slabs containing a gradient of acrylamide. Samples (<5 mg of protein/ml) were dissociated in the presence of 2.5% SDS, 5% 2-mercaptoethanol, 10% glycerol, 62.5 mM Tris-HCI, pH 6.8, by boiling for 2 min. Electrophoresis was performed at 2500 V. h overnight in a water-cooled apparatus (Proteus, Bio-Rad).
Eosin-5-maleimide Labeling of Separated Proteins-This was visualized by exposing the gel immediately after electrophoresis to UV light. Fluorographs were obtained by photographing the fluorescent emission with the help of cutoff filters. N-fH1Ethylmaleimide Labeling of Separated Proteins-Labeling was detected by fluorography (38) in Coomassie Brilliant Blue-stained gels with the aid of Kodak X-OMAT films.
Protein concentration was determined by the Biuret method (39) with bovine albumin as standard.

Eosin-5-maleimide Is Impermeable to the Mitochondrial
Inner Membrane-To determine whether or not eosin-5-maleimide penetrates the inner mitochondrial membrane, two different methods were used, both of which were employed previously to measure the permeability of sulfhydryl reagents (28, 32). The first method determines the ability of the compound to react with mitochondrial glutathione which is localized exclusively in the mitochondrial matrix (28, 40), where it is present mostly in a reduced form (40). The second method is based on the sensitivity of P-hydroxybutyrate dehydrogenase to sulfhydryl-reactive reagents (32). Since the dehydrogenase is localized on the matrix side of the inner membrane, only permeable sulfhydryl reagents are capable of inhibiting the activity in intact mitochondria.
Data presented in Table I and Fig. 1 clearly show that, unlike N-ethylmaleimide, which is readily permeable to the inner mitochondrial membrane, the fluorescent analog, eosin-5-maleimide, is highly impermeable. Thus, a 10-20-min incubation of freshly isolated rat liver mitochondria with 300 FM N-ethylmaleimide at 0 "C reduces the content of glutathione completely, whereas the same concentration of eosin-5maleimide is without effect, Le., the amount of glutathione recovered was identical with the untreated control mitochondria (Table I).
As shown in Fig. lB, eosin-5-maleimide is as effective an inhibitor of 8-hydroxybutyrate dehydrogenase as N-ethylmaleimide. Incubation of inner membrane vesicles which are more than 90% inverted (32) with 10 PM eosin-5-maleimide or N-ethylmaleimide for 10 min at 25 "C results in 95% inhibition of enzyme activity. In freshly isolated mitochondria Eosin-5-maleimide does not affect the glutathione content of mitochondria Rat liver mitochondria were suspended in H medium without bovine albumin at a protein concentration of 10 mg/ml. To each sample consisting of a final volume of 0.35 ml, the N-ethylmaleimide or eosin-5-maleimide at concentrations indicated below were added.
Samples were left on ice for 10 (Experiment I) or 20 min (Experiment 11). Samples were then diluted with 13 ml of H medium and sedimented (10 min, 19,000 X g), washed once more with 13 ml of H medium, and again sedimented (10 min, 19,000 x g). Each pellet was suspended in 0.2 ml of H medium and 1 ml of 5% trichloroacetic acid, 0.01 N HCl. After 5 min at 0 "C, precipitated proteins were sedimented (20 min, 19,000 X g). Supernatants were extracted 5 times with 1 ml of HPO-saturated diethyl ether. The remaining ether was evaporated under nitrogen. Aliquots (25-40 pl) were assayed for glutathione.

Concentration
Glutathione content FIG. 1. Impermeability of intact mitochondria to eosin-5maleimide. Mitochondria ( A ) or inner membrane vesicles ( B ) isolated from rat liver were incubated for 1-2 min with stirring at 25 "C in 1 ml of H medium (without bovine albumin), containing 2. 5 mM KPi, 1 p~ FCCP, and 1 mg/ml of protein. Ten min after the addition of indicated concentrations of N-ethylmaleimide or eosin-5-maleimide, the activity of P-hydroxybutyrate dehydrogenase was determined as P-hydroxybutyrate oxidase activity by measuring the respiration initiated by the addition of 10 mM P-hydroxybutyrate. Control values represented as 100% in the figure were 24 and 35 ng atoms of oxygen/ min/mg of protein for mitochondria and submitochondrial particles, respectively.
( Fig. lA), 100 p~ eosin-5-maleimide is without effect, and 150 p~ of this reagent decreases the enzyme activity by only 7%. The same concentration of N-ethylmaleimide inhibits phydroxybutyrate dehydrogenase activity by 85%.
It seems clear from these studies that eosin-5-maleimide does not penetrate the mitochondrial inner membrane at 0 "C or 25 "C, and can, therefore, be used as a highly specific probe for sulfhydryl groups which lie on the cytoplasmic surface of the mitochondrial inner membrane.

Eosin-5-maleimide Preferentially Labels Sulfhydryl Groups of the ADPIATP Antiporter in Submitochondrial Particles
Rather Than in Intact Mitochondria-Eosin-5-maleimide was recently shown to react selectively with sulfhydryl groups of the ADP/ATP antiporter (29). Although labeling was performed in bovine heart submitochondrial particles, it was suggested that eosind-maleimide penetrates the inner mitochondrial membrane, reacting with the cytosolic side of the carrier (29). Since the data summarized above seem to exclude this possibility, experiments were carried out to establish on which side of the membrane the eosin-5-maleimide-reactive sulfhydryl group is located. For this purpose, mitochondria and submitochondrial particles were prepared from bovine heart and rat liver. In all cases membranes were incubated with 150 PM eosind-maleimide at 0 "C for 20 min. After quenching the reaction with excess dithiothreitol (50 mM), samples were analyzed by SDS-PAGE. The distribution of protein-bound eosin-5-maleimide was visualized by fluorography (29).
As shown in Fig. 2, the amount of the bound label and its distribution among individual mitochondrial proteins differs greatly in mitochondria and submitochondrial particles. Both types of submitochondrial particle preparations (Fig. 2, B and   D), in contrast to the mitochondria from which they were derived ( Fig. 2, C, 289, are highly labeled by eosin-5-maleimide. The majority of the label is associated with a 30-kDa protein in bovine heart and a 29.3-kDa protein in rat liver; for simplicity this band will be referred to below as having a molecular mass of 30 kDa. Significantly, no labeling of the A m. Relative capacity of eosin-5-maleimide to label sulfhydryl groups associated with mitochondria, mitoplasts, and submitochondrial particles. Membranes were incubated for 20 min at 0 "C with 150 p~ eosin-5-maleimide in either 0.3-0.5 ml of H medium without bovine albumin (rat liver) or 0.3-0.5 ml of STE medium (bovine heart) at a protein concentration of 10 mg/ml. The reaction was terminated with 7.5-10 pl of 2 M dithiothreitol and aliquots (10-20 pl) of washed membranes (10 mg/ml) were analyzed by SDS-PAGE. Labeling of proteins was visualized by fluorography. Fluorographs shown here represent the labeling of the following samples. B, bovine heart submitochondrial particles (100 pg); C, bovine heart mitochondria (100 pg); D, rat liver inner membrane vesicles (110 pg); E, rat liver mitoplasts (160 pg); F, rat liver mitochondria (160 pg). A is a partially purified preparation of the ADP/ ATP transporter (20 pg) obtained from eosin-5-maleimide-labeled bovine heart submitochondrial particles (B), exactly as described by Klingenberg et al. (14). The carrier was purified up to the Ultrogel step.
30-kDa band is observed in freshly isolated rat liver mitochondria (Fig. 2F) nor rat liver mitoplasts (Fig. 2E), which are "outer membrane-free'' mitochondria (32). Increasing the temperature to 25 "C also failed to induce labeling of the 30-kDa band, whereas bands with molecular weights corresponding to the Pi/H+ carrier were labeled (see Fig. 7).
In freshly isolated bovine heart mitochondria, only very weak labeling of the 30-kDa band is observed (Fig. 2C). The respiratory control ratio of bovine heart mitochondria is always somewhat lower than that of rat liver mitochondria (4.2-5.3 versus 6.7-7.5 with 6-hydroxybutyrate as substrate). Also, about 10% of the total P-hydroxybutyrate dehydrogenase activity is inhibited by 10 p~ eosind-maleimide in the bovine heart mitochondrial preparation, the remaining 90% being insensitive to 150 p~ eosin-5-maleimide. Thus, the low but visible labeling of the 30-kDa protein in bovine heart mitochondria is most likely related to the presence of a certain percentage of damaged mitochondria, rather than to a difference between bovine heart and rat liver.
When the time of the labeling is increased from 5 to 30 min and the concentration of eosin-5-maleimide increased from 50 to 200 p~, the same basic pattern of the labeling is obtained for both rat liver and bovine heart preparations. The intensity of all labeled bands is decreased when lower concentrations and shorter incubation times are used (not shown).
To ascertain whether the 30-kDa band labeled by eosin-5maleimide in submitochondrial particles is, in fact, the ADP/ ATP antiporter, the eosind-maleimide-labeled submitochondrial particles were used to isolate the ADP/ATP carrier. Isolation according to Klingenberg et al. (14) was carried up to the Ultrogel step. The resultant preparation contains more than 75% of the total protein as a 30-kDa band with the major contaminant being a protein of 34-35 kDa. As shown in Fig. 2A, the isolated transporter is highly labeled by eosin-5-maleimide with all of the label being associated with the 30-kDa band.
Results presented in Fig. 2 demonstrate that the ADP/ATP antiporter is intensively labeled in both bovine heart and rat liver submitochondrial particles, while the corresponding mitochondria are labeled very little (bovine heart) or not at all (rat liver). Prior incubation with ADP or ATP had no effect on these labeling patterns. Because submitochondrial particles prepared by the methods described here (see "Methods") have a predominantly "inverted" inner membrane orientation relative to intact mitochondria (32), these results indicate that the eosin-5-maleimide-reactive sulfhydryl groups of the ADP/ATP antiporter are expressed exclusively on the matrix side of the inner mitochondrial membrane.
Submitochondrial Particles Prepared from Carboxyatractyloside-treated Mitochondria Are Not Readily Labeled with Eosin-5-maleimide-When submitochondrial particles are prepared from mitochondria which have been pretreated with carboxyatractyloside, there is very little labeling by eosin-5maleimide (Fig. 3, C and F) relative to control submitochondrial particles (Fig. 3, A and D) or to submitochondrial particles to which carboxyatractyloside is added directly (Fig.   3, B and E). Thus, carboxyatractyloside, when added to mitochondria, "freezes" the carrier in a form in which eosin-5-maleimide-reactive sulfhydryl groups remain inaccessible even after preparation of submitochondrial particles. It is known that carboxyatractyloside is a nonpenetrant inhibitor of the ADP/ATP antiporter and stabilizes the so-called "Cstate" of the carrier, Le., that state of the carrier in which the nucleotide-binding site faces the cytoplasmic surface of the inner membrane (41, 42). Thus, data presented here suggest that sulfhydryl groups reactive with eosin-5-maleimide at the was added directly to submitochondrial particles 2 min before addition of eosin-5-maleimide. B, bovine heart; E, rat liver, or it was added to mitochondria from which the submitochondrial particles were then isolated and labeled by eosin-5-maleimide. C, bovine heart, F, rat liver. Fluorographs were obtained from 100-pg protein samples after SDS-PAGE. matrix surface of the inner membrane may be expressed only after the transporter has undergone a transition from the Cstate to the so-called ""state" (41,42), a state of the carrier in which the nucleotide-binding site faces the matrix space.
Eosin-5-maleimide and N-Ethylmaleimide Appear to Label Different Sulfhydryl Groups of the ADPIATP Antiporter-Previous studies have shown that when the freely permeable sulfhydryl agent N-ethylmaleimide is added to bovine heart mitochondria, it is able to react with sulfhydryl groups of the ADP/ATP antiporter (24)(25)(26). Although reactivity is slow, it can be markedly stimulated by adding ADP to the mitochondria. Pretreatment of the mitochondria with carboxyatractyloside prevents labeling by N-ethylmaleimide. Results presented in Fig. 4 confirm these observations in rat liver mitochondria by showing that N-[3H]ethylmaleimide moderately labels the 30-kDa protein in the absence of ADP (Fig. 4A), more so in the presence of ADP (Fig. 4B), and not at all when carboxyatractyloside is present (Fig. 4, C and D).
Significantly, sulfhydryl groups of the ADP/ATP antiporter labeled by N-ethylmaleimide appear to differ from those labeled by eosin-5-maleimide. Thus, data presented in Fig. 5 show that, in the presence of ADP, neither bovine heart nor rat liver mitochondria are labeled with eosind-maleimide ( Fig. 5, C and G). Moreover, as shown in Fig. 6, pretreatment of either bovine heart or rat liver submitochondrial particles with N-ethylmaleimide (Fig. 6, B and E ) or N-ethylmaleimide + ADP (Fig. 6, C and F) fails to alter labeling of the ADP/ ATP antiporter by eosin-5-maleimide.
These results indicate that, in intact mitochondria, the Cstate of the ADP/ATP antiporter contains two classes of sulfhydryl groups, one of which has only limited accessibility to N-ethylmaleimide and both of which are inaccessible to

30K-
FIG. 5. Inability of ADP to induce reactivity of eosin-5maleimide with the ADP/ATP antiporter in intact mitochondria. Control mitochondria: B, bovine heart; F, rat liver. Submitochondrial particles: A, bovine heart; E, rat liver, were labeled by eosin-5-maleimide as in Fig. 2. Labeling of mitochondria was also performed with 100 p~ ADP: C, bovine heart; G, rat liver. The following protein aliquots were subjected to SDS-PAGE: bovine heart submitochondrial particles, 100 pg; rat liver submitochondrial particles, 100 pg; bovine heart mitochondria, 150 pg. 6. Inability of N-ethylmaleimide and ADP to alter labeling of the ADP/ATP antiporter in submitochondrial particles by eosin-5-maleimide. Control submitochondrial particles were labeled by eosin-5-maleimide as in Fig. 2 ( A , bovine heart; D, rat liver). Labeling was also performed after 5 min of prior incubation with 1.5 mM N-ethylmaleimide ( B , bovine heart, E, rat liver) or with 1.5 mM N-ethylmaleimide and 100 p~ ADP (C, bovine heart; F, rat liver). Fluorographs were obtained from 100-pg protein samples (bovine heart) or 130-pg proteins samples (rat liver) after SDS-PAGE. eosin-5-maleimide. Preparation of submitochondrial particles (i.e., inversion of the inner membrane) is able to "unmask" the eosind-maleimide-reactive groups in the absence of ADP. However, ADP must be present to fully unmask the Nethylmaleimide-reactive groups.

Eosin-5-maleimide Inhibits the Pi/H' Symporter in Intact
Mitochondria-In functionally intact mitochondria, the ADP/ATP antiporter must work in synchrony with the Pi/ H+ symporter in order to supply the proton ATPase with ADP and Pi for ATP synthesis. It was, therefore, of interest to examine the reactivity of the Pi/H+ symporter with eosin-5-maleimide to establish whether the sulfhydryl group(s) of this transporter exhibit a membrane orientation similar to or different from that of the ADP/ATP antiporter. Results presented in Fig. 7A show that, at 0 "C, eosin-5-maleimide has only a slight inhibitory effect on phosphate transport in mitochondria as measured by the classical ammonium phosphate swelling technique (43). This finding is in contrast to that observed with the well-established phosphate transport inhibitor N-ethylmaleimide, which inhibits phosphate transport nearly completely (Fig. 7A). At 25 "C, however, it can be seen in Fig. 7B that eosin-5-maleimide becomes almost as effective as N-ethylmaleimide in inhibiting phosphate transport. Significantly, dithiothreitol prevents inhibition by both reagents, indicating that sulfhydryl groups are involved (data not shown). Moreover, the inner mitochondrial membrane remains impermeable to eosin-5-maleimide at 25 "C ( Fig. 1).
It seems clear, therefore, that essential sulfhydryl groups of the Pi/H+ symporter reactive with eosin-5-maleimide are oriented near the cytoplasmic surface of the inner membrane in intact mitochondria.
Eosin-5-maleimide Prevents Labeling of the PJH' Symporter by N-[3H]Ethylmaleimide in Intact Mitochondria-Work carried out in this laboratory (44) first showed that the P,/H+ symporter could be labeled rather selectively by N-[3H] ethylmaleimide in rat liver mitochondria by (a) first protect- ing the carrier with p-chloromercuribenzoate, (b) blocking the many other mitochondrial sulfhydryl groups with "cold" Nethylmaleimide, (c) removing the protecting mercurial with dithiothreitol, and ( d ) adding labeled N-ethylmaleimide. By using a slightly modified version of this procedure (36) and then subjecting the mitochondria to SDS-PAGE, two bands of 32-33.5 kDa are seen to be intensely labeled (Fig. 8A). The mobility of these bands corresponds closely to those recently reported for several highly purified preparations of the Pi/H+ symporter (11,12,16,17). Significantly, when "cold" eosin-5-maleimide is added prior to the N-[3H]ethylmaleimide, labeling of these two bands is abolished (Fig. 8, F and G). That the sulfhydryl groups involved are, in fact, those associated with the Pi/H+ symporter is further supported by experiments which show that the addition of either cold mersalyl or cold N-ethylmaleimide prior to N-[3H]ethylmaleimide also abolishes labeling of the 32-33.5-kDa proteins (Figs. 8, B, C, D , and E ) .
Eosin-5-maleimide Labels Proteins of 34-35 kDa in Intact Mitochondria-Results presented in the previous section indicate that the same sulfhydryl group of the Pi/H+ symporter oriented toward the cytoplasmic surface of the mitochondrial inner membrane is reactive with both eosin-5-maleimide and N-ethylmaleimide. Therefore, it should be possible to label the Pi/H+ symporter by direct addition of eosin-5-maleimide to freshly isolated mitochondria. As shown in Fig. 9, B and C, eosin-5-maleimide does, in fact, label several proteins in mitochondria in the 33.5-38-kDa region which are not labeled in submitochondrial particles (Fig. 9A). Of these proteins, those banding with molecular masses of 33.5 and 34.7 kDa are significantly more intense when labeling is done at 25 "C ( Fig. 9C), i.e., where phosphate transport is maximally inhibited (see Fig. 7). Under these conditions (0 "C or 25 "C), the ADP/ATP antiporter is not labeled with eosin-5-maleimide. The finding that the intensity of the 33.5-34.7-kDa bands is much less than that of the 30-kDa ADP/ATP antiporter band is consistent with earlier reports indicating that the Pi/H+ symporter is present at much lower levels in mitochondria than is the ADP/ATP antiporter (44,45).  Fig. 4).

DISCUSSION
Results of experiments described here provide new information about the fluorescent probe eosin-5-maleimide, showing clearly that this agent is impermeable to the mitochondrial inner membrane (Table I and Fig. 1). Thus, eosin-5-maleimide was found to be unreactive with both matrix glutathione and /3-hydroxybutyrate dehydrogenase, an enzyme localized on the matrix surface of the inner membrane. These findings should prove valuable in extending the use of this agent to other membrane proteins where knowledge of the topography of the protein is a prerequisite to understanding the mechanism.
Significantly, in the present study the use of eosin-5-maleimide together with the permeable sulfhydryl reagent, Nethylmaleimide, provides new insight into the relative locations of sulfhydryl groups of the Pi/H+ and ADP/ATP carriers of both rat liver and bovine heart mitochondria. These results and our interpretations of them can best be depicted as shown in Fig. 10. To the left in the figure, the Pi/H+ carrier is shown localized in intact mitochondria, such that its essential sulfhydryl group(s) face the cytosolic surface of the inner membrane. Only one class of sulfhydryl groups is shown. This is consistent with the findings that both eosin-5-maleimide and Nethylmaleimide inhibit phosphate transport in mitochondria at 25 "C (Fig. 7), and that eosin-5-maleimide and mersalyl protect the PJH+ carrier from labeling by N-[3H]ethylmaleimide (Fig. 8). It is consistent also with the earlier findings from this laboratory showing that the impermeant phosphate transport inhibitor, p-chloromercuribenzoate (46), protects A B C n FIG. 9. Eosin-5-maleimide labels a (34-kDa) protein in freshly isolated rat liver mitochondria. Rat liver mitochondria were incubated with eosin-5-maleimide (20 nmol/mg of protein) as described in Fig. 2, except that incubation was performed either at 25 "C for 10 min (C) or at 0 "C for 30 min ( B ) . For comparison, rat liver submitochondrial particles, labeled for 30 min at 0 "C ( A ) to reveal the 30-kDa ADP/ATP antiporter band, are shown. Arrows 1 and 2 indicate the position of bands with molecular masses of 34.7 kDa and 33.5 kDa that become labeled by eosin-5-maleimide in mitochondria at higher temperature.
the Pi/H+ carrier from both inhibition and labeling by Nethylmaleimide (44). Although several studies suggest that two "types" of sulfhydryl groups may be an intrinsic feature of the Pi/H+ carrier (21)(22)(23), it is possible that essential sulfhydryl group(s) within monomeric units of a "dimeric" carrier (13) may be in different environments depending on the conditions selected for study. In any event, eosin-5-maleimide at 25 "C is able to reach those sulfhydryl groups essential for phosphate transport.
In contrast to the Pi/H+ carrier, the ADP/ATP carrier is depicted in Fig. 10 as having two distinct classes of sulfhydryl groups (SH, and SHB). Sulfhydryl groups in class A are shown to become reactive with N-ethylmaleimide but not eosindmaleimide in intact mitochondria, provided ADP is present (Fig. 10, top right). This is consistent with results presented here which show that eosin-5-maleimide will not label sulfhydryl groups of the ADP/ATP carrier in intact mitochondria either in the presence or absence of ADP (Figs. 2 and 5). It is consistent also with the earlier observations of others (24)(25)(26), confirmed in this study (Fig. 4), that N-ethylmaleimide maximally labels the ADP/ATP carrier in intact mitochondria when ADP is present. Sulfhydryl groups in class B are depicted as becoming reactive to eosin-5-maleimide but not N-ethylmaleimide when mitochondria are sonicated (Fig. 10, middle right). Support for this conclusion is derived from the observations that eosin-5-maleimide labels the ADP/ATP carrier in submitochondrial particles, but not in intact mitochondria (Fig. 2), and that N-ethylmaleimide cannot prevent eosin-5-maleimide from labeling the carrier in submitochondrial particles (Fig. 6). Finally, when carboxyatractyloside is added to intact mitochondria, it is depicted as "freezing" the ADP/ATP carrier in a state in which neither N-ethylmaleimide nor eosin-5-maleimide can label the ADP/ATP carrier (Fig. 10, middle left and bottom right). This conclusion is supported by the observations that carboxyatractyloside FIG. 10. Model depicting the relative location of "essential" sulfhydryl groups in the PJH+ and ADP/ATP carriers of mitochondria. One reactive class is proposed for the P,/H+ symporter which is oriented toward the cytoplasmic surface of the inner membrane, whereas two reactive classes localized at (SHA) or near (SHB) the matrix surface are proposed for the ADP/ATP antiporter. The effect of ADP addition, sonication, or carboxyatractyloside on the relative locations of SHa and SHB as deduced from this study are also depicted in the figure. See "Discussion" for more detail. added to intact mitochondria prevents labeling of the ADP/ ATP carrier, both by N-ethylmaleimide (Fig. 4) and by eosin-5-maleimide (Fig. 3).
With regard to the studies reported here, it should be pointed out that the amino acid sequence of the ADP/ATP antiporter is now known (47) and that cysteine residues are located at positions 56, 129, 159 and 256 in the monomeric 30-kDa protein. Two models for the folding of the carrier polypeptide in the mitochondrial inner membrane have been proposed, one by Bogner et al. (48), on the basis of the positions of lysine residues, and one by Saraste and Walker (49), on the basis of a Hydroplot computer program. The latter computer-derived model places two out of the four cysteine residues (cysteines 159 and 256) in a hydrophilic environment, projecting away from the membrane, another within the membrane (cysteine 129), and the fourth (cysteine 56) undesignated but in a polypeptide fold projecting into a hydrophilic space opposite to that occupied by cysteines 159 and 256. Clearly, the observations in this study and other studies (24)(25)(26) cannot be easily accommodated by the Saraste and Walker model. Thus, the latter model predicts that, even in the absence of ADP, the carrier should be labeled readily by sulfhydryl reagents from both sides of the inner mitochondrial membrane. In contrast, the model proposed by Bogner et al. (48) can readily accommodate the observations described in this study because two cysteine residues (residues 159 and 256) reside within the inner membrane and two others (residues 56 and 129) lie near the matrix surface. The "folding" model of Bogner et al. (48) is redrawn in Fig. 10 (inset) to emphasize the predicted locations of the four cysteine residues.
Finally, it should be noted that Muller et al. (50) have recently employed eosin-5-maleimide to study the rotational diffusion of the ADP/ATP translocator in bovine heart mitochondria. In sharp contrast to the results presented here, they conclude that eosin-5-maleimide readily labels the ADP/ ATP carrier in mitochondria. However, the mitochondria used in their studies were stored frozen for an undefined time period, which is well-known to impair mitochondrial intactness. Significantly, Muller et al. provide no information about the intactness of their preparation in the way of acceptor control ratios. In the studies reDorted here we have alwavs