The Reversible Antiport-Uniport Conversion of the Phosphate Carrier from Yeast Mitochondria Depends on the Presence of a Single Cysteine*

Wild type and mutant phosphate carriers (PIC) from Saccharomyces cerevisiae mitochondria were expressed in Escherichia coli as inclusion bodies, solubilized, pu- rified, and optimally reconstituted into liposomal mem-branes. This PIC can function as coupled antiport (P i (cid:50) / P i (cid:50) antiport and P i (cid:50) net transport, i.e. P i (cid:50) /OH (cid:50) antiport) and uncoupled uniport (mercuric chloride-induced P i (cid:50) efflux). The basic kinetic properties of these three trans- port modes were analyzed. The kinetic properties closely resemble those of the reconstituted PIC from beef heart mitochondria. A competitive inhibitor of phosphate transport by the PIC, phosphonoformic acid, was used to establish functional overlap between the the physiological transport modes and the induced efflux mode. Replacement mutants were used to relate the reversible switch from antiport to uniport to a specific residue of the carrier. There are only three cysteines in the yeast PIC. They are at positions 28, 134, and 300 and were replaced by serine, both individually and in com-binations. Cysteine 300 near the C-terminal loop and cysteine 134 located within the third transmembrane segment are accessible to bulky hydrophilic reagents from the cytosolic side, whereas cysteine 28 within the first transmembrane segment is not. None of the three cysteines is relevant to the two antiport

The mitochondrial phosphate carrier (PIC) 1 catalyzes transport of inorganic phosphate into the mitochondrial matrix where the phosphate is utilized for phosphorylating ADP to ATP (LaNoue and Schoolwerth, 1984;Wohlrab, 1986;Wehrle and Pedersen, 1989;Krä mer andPalmieri, 1989, 1992). The function of the PIC was described as P i Ϫ /H ϩ symport, respectively, P i Ϫ /OH Ϫ antiport. The primary structure of the beef heart PIC was elucidated by protein (Aquila et al., 1987) and DNA sequencing (Runswick et al., 1987), and the PIC gene was cloned and sequenced from Saccharomyces cerevisiae (Phelps and Wohlrab, 1991). The PIC is a typical member of the structural family of mitochondrial carriers with six transmembrane segments (Aquila et al., 1985;Kuan and Saier, 1993). Recently the yeast PIC has been expressed as inclusion bodies in Escherichia coli (Murakami et al., 1993;Wohlrab and Briggs, 1994). Procedures have been described to solubilize the PIC from inclusion bodies in a functionally active state (Wohlrab and Briggs, 1994).
The function of the PIC was studied after purification from various kinds of mitochondria (for reviews see Wohlrab (1986), Wehrle and Pedersen (1989), and Krä mer and Palmieri (1989)) and reconstitution into proteoliposomes (Wohlrab, 1980;De Pinto et al., 1982;Wehrle and Pedersen, 1982). PIC catalyzes both homologous P i Ϫ /P i Ϫ as well as heterologous P i Ϫ /OH Ϫ antiport with high activity (Wohlrab and Flowers, 1982;Stappen andKrä mer, 1993, 1994). Transport kinetics using bireactant initial velocity studies identify the PIC as a member of the mitochondrial carrier family also in functional terms. Its mechanism is of the simultaneous (sequential) type, involving a ternary complex in transport catalysis that requires the binding of two ligands at the same time (Stappen and Krä mer, 1994). An additional property that places the PIC into this functional family is its ability to switch to uniport (efflux) activity after chemical modification with some mercurial reagents (Stappen and Krä mer, 1993). It was previously shown that the aspartate/glutamate carrier (Dierks et al., 1990a), the ADP/ATP carrier (Dierks et al., 1990b), and the carnitine carrier (Indiveri et al., 1991) can reversibly be converted by mercurial reagents from coupled antiport to uncoupled uniport, a function that comprises both carrier-like and channel-like properties. An analysis of this conversion, specifically with the aspartate/glutamate carrier, identified the involvement of at least two cysteines and was interpreted to reveal an intrinsic preformed channel structure as a common element in those carriers (Dierks et al., 1990b). These kinds of structural domains had already been postulated on the basis of functional considerations (Klingenberg, 1981) as well as of transport experiments (Brustovetsky and Klingenberg, 1996).
The aim of the present work was to relate the reversible switch from coupled antiport to uncoupled uniport to a specific residue of the carrier protein by using replacement mutagenesis of the yeast PIC expressed in E. coli. This technique has already been used to identify a number of other residues, which are important for the carrier's physiological function with re-spect to substrate recognition and inhibitor interaction (Wohlrab and Briggs, 1994;Phelps et al., 1996). We demonstrate now that the conversion between antiport and uniport of the PIC exclusively depends on the presence of cysteine 28 in the PIC monomer, i.e. two cysteine 28 residues in the functionally active dimer.

EXPERIMENTAL PROCEDURES
Materials and Their Sources-[ 33 P]Phosphate was obtained from Amersham-Buchler. Sigma supplied Triton X-114, mersalylic acid, pC-MBS, DTT, phosphonoformic acid, HEPES, PIPES, and turkey egg yolk phosphatidylcholine. Dowex 2-X10 and SLS were from Fluka, Bio-Beads SM-2 from Bio-Rad, Sephadex G-75 from Pharmacia, and pyridoxalphosphate and HgCl 2 from Merck. All SH reagents (HgCl 2 , pC-MBS, mersalylic acid) were prepared from frozen stock solutions and were diluted with water or the respective gel filtration buffer. Pyridoxalphosphate was dissolved in 1 M imidazole (pH 6.5). All other chemicals were of analytical grade.
Generation of Mutants, Expression, Isolation, and Purification of the PIC-The gene coding for the PIC was cloned from a yeast genomic library as described earlier (Phelps and Wohlrab, 1991) and the PIC was expressed in the E. coli strain BL21 (DE3) (Murakami et al., 1993;Wohlrab and Briggs, 1994). Mutants were generated as described earlier (Phelps and Wohlrab, 1993;Wohlrab and Briggs, 1994). The expression strain Bl21 (DE3) was transformed with the plasmid pNYHM131 either coding for the wild type PIC or a mutant. A total of 1 liter of 2 ϫ YT medium (plus 100 mg of carbenicillin) was inoculated with an overnight colony of transformed BL21 (DE3) and grown to an A 600 of 0.6 (about 5 h) under vigorous shaking at 37°C. PIC expression was initiated by adding 1 mM isopropyl-1-thio-␤-D-galactopyranoside plus 100 mg of carbenicillin. Growth was continued for 3 h, cells were harvested and stored at Ϫ20°C, not longer than 48 h before continuing the procedure. All the following steps were carried out at 0 -4°C. The pellet from 125 ml of culture was suspended in TE (10 mM Tris base, 0.1 mM EDTA, 1 mM DTT, pH 7.0) and passed twice through a French pressure cell, followed by 10 min of centrifugation at 12,000 ϫ g. The pellet was homogenized with 10 ml of TE and centrifuged at 1100 ϫ g for 5 min. 8.8 ml of the supernatant was centrifuged at 12,000 ϫ g for 3 min, and the pellet was stored at Ϫ70°C. Isolation of the PIC from inclusion bodies was carried out applying a previously described procedure for the oxoglutarate carrier  with some modifications. The pellet was washed three times in TE-buffer containing 2% Triton X-114, followed by centrifugation at 12,000 ϫ g for 2.5 min. The pellet was solubilized in 400 l of TE containing 1.67% SLS, followed by addition of 800 l of water. The resulting solution was used for reconstitution. Fig. 1 shows a silver-stained SDS-PAGE of the PIC after purification.
Reconstitution Procedure-The solubilized PIC was reconstituted into preformed liposomes by the Amberlite method as described for the bovine heart PIC (Stappen and Krä mer, 1993), including addition of Triton X-114 in excess over SLS. The reconstitution procedure was modified with respect to the phospholipid/protein and phospholipid/ detergent ratio. Optimal transport activity was obtained at a phospholipid concentration of 16 mg/ml, a phospholipid/protein ratio of 140 mg/mg, and a detergent/phospholipid ratio of 0.62 mg/mg. This means that 70 l of Triton X-114 (10%, w/w), 112 l of liposomes (10% EYPC (w/w) in 50 mM KCl, 20 mM HEPES, 20 mM potassium P i , pH 6.5), and 20 l of protein solution were mixed with HEPES/potassium P i buffer (50 mM HEPES, 30 mM potassium P i , pH 6.5) to yield a final volume of 700 l. A detergent/Amberlite ratio of 12 mg/g and 15 column passages were used for detergent removal. The amount of PIC recovered in the proteoliposomes after reconstitution was found to vary between 27 and 43% of the protein in the SLS-solubilized fraction.
Measurement of Transport Activity-The reconstituted transport activities (P i /P i antiport, P i net transport, and P i efflux) were determined by measuring the flux of [ 33 P]phosphate. The applied methods resemble those described previously for the analysis of the aspartate/glutamate carrier (Dierks and Krä mer, 1988;Dierks et al., 1990a). In most experiments, antiport activity was determined by the forward exchange procedure. The assay was started by adding labeled phosphate to the proteoliposomes containing unlabeled substrate inside and the increase in internal label was followed. In some experiments, the P i /P i antiport mode was measured by the backward exchange method, which is similar to the procedure used for determination of the net transport mode (P i Ϫ /OH Ϫ ) or the HgCl 2 -induced efflux mode. For this, the internal pool was prelabeled by incubating the proteoliposomes with [ 33 P]phosphate of high specific radioactivity at 21°C for 10 min. The external substrate (and label) was then removed by size exclusion chromatography on Sephadex G-75 columns at 4°C. Since the PIC also catalyzes net transport, the use of this method depends on the availability of reversible inhibition during chromatography. Routinely, 200 M mersalylic acid was used for this purpose. After removal of external phosphate, transport was started by adding 10 mM DTT (net transport) or 10 mM DTT together with external phosphate (P i /P i antiport). The mercurial-induced uniport was assayed by adding 0.5 mM HgCl 2 to proteoliposomes after the size exclusion chromatography. After the desired period of time carrier-mediated transport was stopped by adding 25 mM pyridoxal phosphate with or without 10 mM DTT. After application of the stop mix, each sample was passed through an anion exchange column (Dowex 1-X10, Cl Ϫ form) to remove the external label. Further details were as described previously (Dierks and Krä mer, 1988;Dierks et al., 1990a).
For PIC mutants lacking Cys-134, an alternative inhibition technique had to be developed since these mutants cannot be inhibited by mersalyl. In this case, the internal pH of the proteoliposomes was adjusted to pH 6.8 during formation by high internal buffer (HEPES, 50 mM). The external pH in the size exclusion chromatography was set to 5.5 (PIPES, 5 mM). Due to the H ϩ -coupled transport mechanism, the pH gradient prevented loss of internal substrate. Before starting net transport, the external pH was adjusted to 6.8 by adding 50 mM HEPES, pH 7.0. The data of Table I indicate that retention of internal phosphate by this method was comparable with the generally applied procedure involving inhibitor addition.
Forward exchange rates were determined by fitting the time course of isotope equilibration to a single exponential y ϭ A ⅐ (1 Ϫ e Ϫkt ) ϩ B, leading to the first order rate constant k (in min Ϫ1 ). The specific activity (mol/min ⅐ mg Prot ) was calculated from k (min Ϫ1 ), the final value of isotope equilibration (dpm), the specific radioactivity (dmp/nmol), the volume of the proteoliposome fraction (ml), and the protein concentration (g/ml) as published previously (Dierks and Krä mer, 1988). Backward exchange rates were calculated by the equation Ϫkt ) (␣, percentage of isotope equilibration; S in , internal substrate concentration; V in , internal volume of liposomes catalyzing transport) (Dierks and Krä mer, 1988).
Protein Determination-Protein concentrations were determined by the modified Lowry method after precipitation with deoxycholate and trichloroacetic acid and extraction of detergent and lipid by organic solvents (Peterson, 1977;Dulley and Grieve, 1975).

RESULTS
Comparison of Beef Heart and Yeast PIC-So far, for functional characterization of the mitochondrial PIC, mainly the protein from beef heart or pig heart was studied after isolation using Triton X-100 or X-114 (Wohlrab, 1986;Krä mer and Palmieri, 1989). A detailed functional analysis after reconstitution by chromatography on Amberlite was carried out with the beef heart protein (Stappen andKrä mer, 1993, 1994). In the present work, we use a different protein (yeast instead of beef heart PIC), a different source (bacterial inclusion bodies instead of intact mitochondria), a different detergent (SLS instead of Triton X-114), and an altered reconstitution procedure (optimized for SLS-solubilized protein). Consequently, it was essential to compare the properties of the yeast PIC isolated from E. coli inclusion bodies with the data previously obtained in experiments using the beef heart protein.
Since for the PIC, in contrast to the ADP/ATP carrier, no side-specific inhibitors are available, the orientation of the carrier protein in proteoliposomes was previously established by analyzing side-specific substrate interaction (Stappen and Krä mer, 1993). Such a kinetic analysis of the P i Ϫ /P i Ϫ antiport of the yeast PIC, both for the wild type and the Cys-28 3 Ser mutant as representative examples for the recombinant proteins, is shown in Fig. 2. For the P i Ϫ /P i Ϫ and the P i Ϫ /OH Ϫ antiport (net transport), the K m values (transport affinity) at both sides of the proteoliposomes are shown for the wild type and several mutants (Table II). The following conclusions could be drawn. (i) Only one single kinetic component is observed for interaction of phosphate at the inside and the outside, respectively. (ii) The K m values for phosphate in the wild type and the Cys-28 3 Ser mutant are identical. (iii) A comparison shows that the corresponding values of the beef heart and yeast PIC for interaction with phosphate at the two sides of the protein are more or less identical. (iv) These data also prove that the PIC of beef heart and yeast are oriented in the same direction after reconstitution, i.e. right side out (Stappen and Krä mer, 1993). These results indicate that the methods previously derived for the PIC from beef heart mitochondria can be applied to the reconstituted yeast carrier protein.
The reconstituted beef heart PIC was shown to catalyze three different functions, i.e. homologous P i Ϫ /P i Ϫ antiport, heterologous P i Ϫ /OH Ϫ antiport (net transport), as well as sub-strate-unspecific uniport after treatment with HgCl 2 (Stappen and Krä mer, 1993). The same was observed for the PIC in intact yeast mitochondria (Stappen, 1994), as well as for the wild type yeast PIC isolated from inclusion bodies after expression in E. coli (Table II, see also Fig. 5). The transport affinities (K m ) turned out to be very similar, whereas the V max values were in general higher for the yeast PIC. Since the efficiency of reconstitution (the share of successfully incorporated protein) was not quantitated in these experiments in detail, the some-  Ϫ /P i Ϫ antiport and P i Ϫ /OH Ϫ antiport modes for ligand interaction at the external and the internal side of proteoliposomes The experiments were carried out at pH 6.5 (beef heart PIC) and pH 6.8 (S. cerevisiae PIC). The data of the beef heart protein are taken from Stappen and Krä mer (1993); those of the protein from S. cerevisia mitochondria from Guerin et al. (1990 (Wohlrab and Briggs, 1994).
what different V max values do not necessarily mean that the molecular activity of the PIC from the different sources is in fact different. Irrespective of this restriction, it is interesting to note that the ratio of P i Ϫ /P i Ϫ antiport to P i Ϫ /OH Ϫ antiport (net transport) is also very similar for the two proteins (Stappen and Krä mer, 1994).
Characterization of Uniport Activity of the Yeast PIC Expressed in E. coli Cells-To unequivocally correlate the observed uniport activity to the reconstituted yeast PIC and to define its properties in the mutants studied, we used various inhibitors. The unspecific inhibitors pyridoxal phosphate and mersalylic acid were previously described as effective reagents stopping transport (Stappen and Krä mer, 1993). The data in Table III show that pyridoxal phosphate works in all cases. Mersalylic acid, on the other hand, is ineffective when applied to mutants in which Cys-134 has been replaced by Ser. Consequently, we had to develop another method for reversibly inhibiting net transport during size exclusion chromatography, namely application of an inverse pH gradient (see "Experimental Procedures"). The SH reagent pCMBS also reacts with Cys-134 and/or Cys-300 (Table III), as shown above for mersalyl. Besides interacting with Cys-28, HgCl 2 blocks the transport reaction, at least in part, by reacting also with Cys-134 and/or Cys-300.
The availability of phosphonoformic acid (PFA) (Kempson, 1988), originally discovered to block the kidney phosphate carrier, prompted us to characterize the properties of the mercurial-induced efflux by the yeast PIC in more detail. The only evidence that the HgCl 2 -induced uniport activity of the PIC involves the phosphate binding site is its inhibition by external phosphate (Stappen and Krä mer, 1993). PFA significantly inhibits P i Ϫ /P i Ϫ antiport, but has no effect on P i Ϫ net transport (P i Ϫ /OH Ϫ antiport) (Table III). PFA blocks the mercuric chloride-induced phosphate efflux (Fig. 3A) to about the same extent as phosphate (Stappen and Krä mer, 1993). Similar concentrations of sulfate, which is not a ligand of the PIC, shown no effect. We analyzed the effect of PFA on P i Ϫ /P i Ϫ antiport in more detail. Fig. 3B reveals its competitive inhibition of P i Ϫ /P i Ϫ antiport with a K i of 0.75 mM. Antiport and Uniport Activity of Wild type and Mutant Yeast PIC-We measured the V max for all three transport modes of several reconstituted mutant PICs. The results were compared with those obtained with proteoliposomes reconstituted either with the PIC from beef heart or that from S. cerevisiae mitochondria. It should be noted that the results for Cys-134 3 Ser mutants were obtained using an inverse pH gradient during size exclusion chromatography (see "Experimental Procedures"). In Fig. 4, the P i Ϫ /P i Ϫ antiport activities of the various mutants are compared with that of the wild type. In all cases the transport rates were lower than that of the parent strain.
The transport rates, however, are still significant, i.e. more than 20% of that the wild type. Most importantly, the ratio of P i Ϫ /P i Ϫ antiport to P i Ϫ /OH Ϫ antiport (net transport) remained more or less constant, irrespective of the altered absolute values of the two physiological transport modes.
Whereas both the homologous and the heterologous antiport activities were retained in the PIC proteins, this was not the case with the mercuric chloride-induced uniport (Fig. 4). This induced phosphate efflux catalyzed by the wild type as well as the Cys-28 3 Ser and Cys-300 3 Ser mutant proteins is shown in Fig. 5. The relative efflux rate by the Cys-300 3 Ser mutant a Data as k values in % of control (P i Ϫ /P i Ϫ antiport without added inhibitor). b Not measurable since HgCl 2 induces efflux under these conditions. c At 2 mM external phosphate. d In the presence of mersalyl (efflux conditions).
is similar to that of the wild type, while the Cys-28 3 Ser mutant, even when treated with high mercurial concentrations, is not able to undergo the functional switch from coupled antiport to uncoupled uniport (efflux). The induced uniport activities of all PICs are compared in Fig. 4. In all mutants in which Cys-28 was replaced by Ser, independent of a replacement of the other two cysteine residues, Cys-234 and Cys-300), the functional switch to uniport could not be induced.

DISCUSSION
The mitochondrial PIC can reversibly be switched from coupled antiport to uncoupled uniport. We have used replacement mutants of the yeast PIC expressed in E. coli to relate this switch to specific residues of this protein. The involvement of cysteine residues in this switch can most appropriately be studied with the yeast PIC, since 1) it has only three cysteines versus eight in the beef heart PIC used in earlier studies, and 2) its preparation is facilitated with E. coli inclusion bodies. We have shown that the yeast PIC, isolated and purified from inclusion bodies, solubilized by SLS and reconstituted into liposomes resembles the beef heart PIC in all the relevant functional aspects.
The mercuric chloride-induced uniport has now been definitely identified with the reconstituted PIC. Contaminating mitochondrial channel proteins cannot be responsible for this activity since the PICs are heterologously expressed. Furthermore, we found that PFA, a competitive inhibitor of P i Ϫ /P i Ϫ antiport, effectively inhibits also the induced uniport mode. Since PFA stimulates P i Ϫ /OH Ϫ (P i Ϫ net transport), we conclude that it is a transport substrate of the PIC. PFA is not available in labeled form, and thus its true substrate properties could not be directly investigated. Nevertheless, its effect on the mercuric chloride-induced P i Ϫ efflux can be taken as a further indication that the external binding site of the reconstituted PIC has retained its properties. This site, as shown earlier, is the cytosolic site of the PIC (Stappen and Krä mer, 1993).
It has been shown for many other carrier proteins that cysteines are not essential for basic transport functions (van Iwaarden et al., 1991(van Iwaarden et al., , 1992. The present results permit us to characterize the functional significance of the yeast PIC cysteines (Fig. 6). Cys-300 does not seem to be relevant for transport activity or for uniport induction. We did, however, detect a 50% reduction of transport after its replacement by serine. No additional changes were observed in combination with the Cys-134 3 Ser replacement. Cys-134 is responsible for inhibition of all transport modes by mersalylic acid and also by pCMBS. A comparison with published effects on the PIC (Stappen and Krä mer, 1993) makes it likely that Cys-134 is also responsible for inhibition by 5,5-dithiobis(2-nitrobenzoic acid). Although a reaction of mersalyl with Cys-300 seems to be responsible for a slight reduction in antiport activity by the Cys-134 3 Ser mutant relative to the Cys-134 3 Ser/Cys-300 3 Ser and the Cys-28 3 Ser/Cys-134 3 Ser/Cys-300 3 Ser mutants, it is obvious that the major target of mersalyl is Cys-134. The reactivities of the cysteines indicate that Cys-134 (and presumably also Cys-300) is accessible to large, hydrophilic ligands, whereas Cys-28 is accessible only to the small Hg 2ϩ ion. This was documented by the observation that mercurials other than HgCl 2 did not affect the reaction of Hg 2ϩ with Cys-28. Whereas the accessibility of Cys-300 from the external (cytosolic) side seems obvious, this is not so for Cys-134. Our results suggest that Cys-134 is in an aqueous environment with connected to the cytosolic side of the protein, whereas Cys-28 is not. Interestingly, Cys-28 has previously been found to be the target for inhibition of phosphate transport by oxygen (Phelps and Wohlrab, 1993). In a recent paper, furthermore, His-32, Glu-126, and Glu-137 were found to be essential for a coupled phosphate/H ϩ pathway in the PIC (Phelps et al., 1996). In fact, these residues line up at the same side of helices A and C, in which Cys-28 and Cys-134 are located. It may be questioned whether a reaction of HgCl 2 with other residues besides cysteine should be considered. This has in fact been analyzed in detail with respect to the aspartate/glutamate carrier, where a functionally significant reaction was found to be confined to cysteine residues (Dierks et al., 1990a(Dierks et al., , 1990b. In any case, the specific action on Cys-28 and Cys-134 of the PIC was proven here by the absence of these effects in the mutants lacking these cysteines. Most interesting, however, is the observation that the reversible switch from the physiological transport modes to the unphysiological and mercuric chloride-induced uniport depends on the presence of Cys-28. This functional shift from antiport to uniport, which is correlated with the appearance of some channel-type functions, has been observed in several mitochondrial carriers, namely the aspartate/glutamate, the ADP/ATP (Dierks et al., 1990a;Dierks et al., 1990b), and the carnitine carrier (Indiveri et al., 1991). The present results, however, for FIG. 4. Relative activity of the three different modes of phosphate transport in proteoliposomes, reconstituted with various mutant PIC species. P i Ϫ /P i Ϫ antiport (hatched bars), P i Ϫ /OH Ϫ antiport (open bars), and phosphate efflux (solid bars) were measured as described under "Experimental Procedures." In the case of mutants in which cysteine 134 was exchanged for serine, an inverse pH gradient was used to block phosphate efflux during size-exclusion chromatography. For uniport (efflux) induction 0.5 mM HgCl 2 was used. All values are means of at least three determinations; the bars indicate the standard deviation. The P i Ϫ /P i Ϫ antiport activity of the wild type was set to 100%, and this measurement was included in all other experiments for determination of transport activities to provide an appropriate basis for the normalization procedure. The value of the three transport activities were compared on the basis of the derived first order rate constants (k) of the respective transport modes (see "Experimental Procedures"). Basic phosphate efflux from proteoliposomes containing inhibited wild type PIC was Ͻ2% of the mercurial-induced efflux activity.
FIG. 5. Kinetics of phosphate efflux in the mercurial-induced efflux mode from proteoliposomes carrying various PIC proteins, namely wild type (squares), Cys-28 3 Ser (triangles), and Cys-300 3 Ser mutant strain (circles). For induction of the efflux mode, either 0.5 mM (open symbols) or 1 mM HgCl 2 (solid symbols) was added. Internal phosphate was 30 mM. the first time correlate a specific residue with this phenomenon. The observation that mitochondrial carriers retain their original activation energy of transport after this functional switch was interpreted to indicate that a similar conformational change occurs during solute transfer both in the antiport and the mercuric chloride-induced uniport (Herick and Krä mer, 1995). Modification of Cys-28 by HgCl 2 yields a loss of specificity of ligand interaction at the internal PIC binding site. No such loss of the external binding site is observed (Dierks et al., 1990b;Stappen and Krä mer, 1993). In conclusion, our findings suggest that Cys-28 may possess a gating function on the matrix side of the PIC.
There is evidence from a large number of investigations that the structural basis for differences in the mechanisms of coupling among antiport, symport and uniport in secondary systems and differences between carrier and channel-type of functions may be very subtle (Nikaido and Saier, 1992;Krä mer, 1994). Thus, for example, a single amino acid replacement in the bacterial lactose carrier (Lac-permease) shifts its transport from a coupled to an uncoupled mechanism (Eelkema et al., 1991;King and Wilson, 1990;Kaback, 1992). Furthermore, a number of studies using electrophysiological techniques provide evidence for channel properties in carrier proteins, e.g. neurotransmitter transporters (Cammack and Schwartz, 1996;DeFelice and Blakely, 1996), the chloroplast triose phosphate carrier (Schwarz et al., 1994) as well as the mitochondrial ADP/ATP carrier (Tikhonova et al., 1994;Brustovetsky and Klingenberg, 1996). This concept is particularly attractive in view of the fact that conditions are known where large pores appear in the inner mitochondrial membrane (permeability transition pore, "megachannel") (Zoratti and Szabo, 1995;Bernardi and Petronilli, 1996). Taken together, the distinction between "carrier-type" and "channel-type" of transport mechanisms seems to become minimal, when analyzed in terms of functional elements in both types of solute transport systems (Nikaido and Saier, 1992;Krä mer, 1994;DeFelice and Blakely, 1996).
The data presented here prove that the reversible shift between the coupled and the uncoupled transport mode of the mitochondrial PIC only depends on the modification of a single cysteine. It has to be pointed out, however, that mitochondrial carrier proteins are functional dimers, i.e. there are two cysteines at position 28. Interestingly, earlier investigations of the aspartate/glutamate carrier indicated that the modification of two cysteine residues is necessary for the reversible shift from antiport to uniport (Dierks et al., 1990a;Dierks et al., 1990b). Although the primary structure of the aspartate/glutamate carrier is not yet known and thus the two cysteines have not been identified, it is obvious that, on the basis of the apparently common mechanism of this antiport/uniport conversion in mitochondrial carriers, the two results are related. Consequently, studying the functional involvement of the two cysteines lo- FIG. 6. Topological representation of the PIC inserted in the mitochondrial membrane according to structure predictions and experimental data Wohlrab and Briggs, 1994). The three cysteine residues which were replaced by serine are indicated. cated in the two homologous monomers of mitochondrial carrier proteins may be an interesting clue for understanding the crosstalk of the individual subunits during transport catalysis.