The Phosphate Carrier from Yeast Mitochondria

Wild type phosphate carrier (PIC) fromSaccharomyces cerevisiae and recombinant PIC proteins with different C-terminal extensions were expressed in Escherichia coli as inclusion bodies. From these, PIC was isolated with the detergent sodium lauroyl sarcosinate in a form, partially monomeric and unfolded. This PIC associates to stable dimers after exchanging the detergent to the polyoxyethylene detergent C12E8 and dialysis. Combining two differently tagged monomers of PIC and following this with affinity chromatography yields defined homo- and heterodimeric forms of PIC, which are all fully active after reconstitution. As a member of the mitochondrial carrier family PIC is supposed to function as a homodimer. We investigated its dimeric nature in the functionally active state after reconstitution. When reconstituting PIC monomers a sigmoidal dependence of transport activity on the amount of inserted protein is observed, whereas insertion of PIC dimers leads to a linear dependence. Heterodimeric PIC constructs consisting of both an active and an inactivated subunit do not catalyze phosphate transport. In contrast, reconstitution of a mixture of active and inactive monomeric subunits led to partially active carrier. These experiments prove (i) that PIC does not function in monomeric form, (ii) that PIC dimers are stable both in the solubilized state and after membrane insertion, and (iii) that transport catalyzed by PIC dimers involves functional cross-talk between the two monomers.

Wild type phosphate carrier (PIC) from Saccharomyces cerevisiae and recombinant PIC proteins with different C-terminal extensions were expressed in Escherichia coli as inclusion bodies. From these, PIC was isolated with the detergent sodium lauroyl sarcosinate in a form, partially monomeric and unfolded. This PIC associates to stable dimers after exchanging the detergent to the polyoxyethylene detergent C 12 E 8 and dialysis. Combining two differently tagged monomers of PIC and following this with affinity chromatography yields defined homo-and heterodimeric forms of PIC, which are all fully active after reconstitution. As a member of the mitochondrial carrier family PIC is supposed to function as a homodimer. We investigated its dimeric nature in the functionally active state after reconstitution. When reconstituting PIC monomers a sigmoidal dependence of transport activity on the amount of inserted protein is observed, whereas insertion of PIC dimers leads to a linear dependence. Heterodimeric PIC constructs consisting of both an active and an inactivated subunit do not catalyze phosphate transport. In contrast, reconstitution of a mixture of active and inactive monomeric subunits led to partially active carrier. These experiments prove (i) that PIC does not function in monomeric form, (ii) that PIC dimers are stable both in the solubilized state and after membrane insertion, and (iii) that transport catalyzed by PIC dimers involves functional cross-talk between the two monomers.
The mitochondrial phosphate carrier (PIC) 1 or phosphate transport protein (PTP) catalyzes transport of phosphate into the mitochondrial matrix where the phosphate is utilized for oxidative phosphorylation (1)(2)(3)(4)(5)(6). The primary structure of the beef heart PIC was elucidated by protein (7) and DNA/protein sequencing (8) and the PIC gene was cloned and sequenced from Saccharomyces cerevisiae (9). The yeast PIC has been expressed as inclusion bodies in Escherichia coli (10,11). Methods have been described to solubilize mitochondrial carriers from inclusion bodies including PIC in a functionally active state (11)(12)(13).
PIC is a typical member of the structural family of mitochondrial carriers with subunits of six transmembrane segments and a molecular mass of 32 kDa (14,15). There are several lines of evidence that mitochondrial carriers do not function as monomers but form dimers in the functional state. The first and still one of the most convincing indications up to now was the observation of a binding stoichiometry of one molecule of the tightly binding ligand carboxyatractylate to two monomeric units of the ADP/ATP carrier (16). An even lower binding stoichiometry was observed for the ligands ADP and ATP (17), which in experiments with fluorescent nucleotide analogs led to the suggestion of a tetrameric functional unit of the ADP/ATP carrier (18). By using cross-linking and analytical ultracentrifuge techniques it was shown that the ADP/ATP carrier as well as the mitochondrial uncoupling protein, at least in the solubilized state, forms a homodimer (19,20). In recent experiments the formation of an intermolecular disulfide bridge between the monomers of the ADP/ATP carrier provided further evidence for the dimeric state of this mitochondrial carrier (21). Studies with PIC in mitochondria demonstrated the requirement of less than one NEM per subunit of PIC (22) and in reconstituted proteoliposomes that a disulfide between Cys-28 of the two monomers reversibly blocks transport (23). Besides these findings, it has been argued also on a theoretical point of view that a dimeric state is favorable for carrier function (24). It is noteworthy that there is a further reason for the acceptance of the dimeric nature of mitochondrial carriers, namely the "consensus minimal unit" of about 12 transmembrane segments which holds true for many carrier proteins (25)(26)(27).
The oligomeric state of secondary carriers has been investigated in a number of cases. Evidence has been provided for both the monomeric and the dimeric form of the E. coli lactose permease to be functional (28). However, the dominating evidence suggests that lactose permease is functional as a monomer with 12 transmembrane segments (28). The situation is not better understood for other well studied carriers. Mammalian facilitative sugar carriers, i.e. uniporters (GLUT-family) and Na ϩ -coupled symporters (SGLT-family), were found to function both as monomers and oligomers, and cooperative interactions were suggested as a regulatory mechanism (29 -31). There is experimental evidence that also the erythrocyte anion transporter (band 3 protein) may exist as a mixture of dimers and tetramers (32)(33)(34), but evidence for a monomeric function of this protein has also been provided (35). Recently, by coexpression and co-reconstitution of functional and nonfunctional monomers of the small secondary carrier EmrE, the oligomeric state of this protein has been demonstrated (36). However, several of the methods used to prove oligomeric associations in these proteins may be questioned (37), and con-sequently their functional oligomeric structures have not yet been definitely established.
The aim of the present work was to prove that the dimeric state is a prerequisite for function of PIC in the membrane. We used differently tagged PIC monomers to prepare defined heterodimers. The monomers were obtained after heterologous expression in E. coli and solubilization. Analyses of these constructs showed both an inability of monomers to function in phosphate transport and cross-talk between the subunits when integrated into stable dimers.

EXPERIMENTAL PROCEDURES
Materials-[ 33 P]Phosphate was obtained from Amersham-Buchler (Braunschweig, Germany). Sigma (Deisenhofen, Germany) supplied the following chemicals: mersalylic acid, dithiothreitol, HEPES, PIPES, anti-mouse IgG, alkaline phosphatase conjugate, turkey egg yolk phospholipid, Sigma Fast BCIP/NBT tablets. Dowex 2-X10, sodium lauroyl sarcosinate (SLS), and C 12 E 8 were purchased from Fluka (Deisenhofen, Germany), C 8 E n from Bachem (Bubendorf, Switzerland), Bio-Beads SM-2 from Bio-Rad (Munich, Germany), Sephadex G-75 from Pharmacia (Freiburg, Germany), and N-ethylmaleimide and pyridoxal phosphate were from Merck (Darmstadt, Germany). The micro BCA protein assay was used for protein determination and was purchased from Pierce. Ni-NTA-agarose and Ni-NTA alkaline phosphatase conjugate was purchased from Qiagen (Hilden, Germany). The FLAG system for protein tagging was purchased from Kodak (Rochester, MN). All further chemicals were of analytical grade. All sulfhydryl reagents used were prepared freshly. The reagents were diluted with water or the respective gel filtration buffer. Pyridoxal phosphate in high concentrations was dissolved in 1 M imidazole (pH 6.5).
Generation of Tagged Carrier Proteins-Cloning of DNA and subsequent transformation steps were carried out using standard techniques (38,39). The 3Ј part of the mir gene coding for the phosphate carrier was amplified by polymerase chain reaction using two oligonucleotide primers annealing upstream (5Ј-GACTGCTGGTTTGGC-3Ј) of the KpnI site and downstream of the 3Ј part thereby introducing the FLAG tag (5Ј-GGTGGTGGTGGTCATGACTACAAGGACGACGATGACAAGTA-GGGATCC-3Ј) or the His tag (5Ј-GGTGGTGGTGGTCATCATCATCA-TCATCATTAGGGATCC-3Ј) (tags underlined), respectively, and a BamHI restriction site downstream from the stop codon. Polymerase chain reaction (30 s 94°C, 30 s 50°C, 60 s 72°C, 30 cycles) was carried out using Taq polymerase (Boehringer, Mannheim) and a Thermo-Cycler 480 (Perkin-Elmer). Plasmid pNYHM131 (13,40) was used as template. The polymerase chain reaction products were cut with BamHI and KpnI and cloned into the plasmid pUC18 for sequencing. Sequencing was carried out using a Pharmacia (Freiburg, Germany) A.L.F. DNA sequencer and the AutoRead sequencing kit (Pharmacia, Freiburg, Germany) as recommended by the supplier. Appropriate fragments were subsequently cloned into plasmid pNYHM131 via BamHI and KpnI restriction and ligation. All cloning steps were carried out in the E. coli strain DH5␣. The expression of the different proteins was carried out in E. coli strain BL21 (DE3) as described below.
Isolation and Purification of the PIC-Expression strain BL21 (DE3) carrying plasmids coding for the wild type PIC or a mutant, respectively, was transformed. A total of 1 liter of 2 ϫ YT medium (plus 100 mg of carbenicillin) was inoculated with a fresh overnight colony of transformed BL21 (DE3) and grown to an OD 600 of 0.6 (about 5 h) under vigorous shaking at 37°C. Expression of PiC was initiated by 1 mM isopropyl-␤-D-thiogalactopyranoside, and 100 mg of carbenicillin was added. Growth was continued for 3 h, and the cells were harvested and stored at Ϫ20°C. All the following steps were carried out on ice (11). The cell pellet with the expressed PIC was suspended in TE (10 mM Tris base, 0.1 mM EDTA, 1 mM DTT, adjusted to pH 7.0 with HCl) and passed twice through a French pressure cell, followed by centrifugation at 12,100 ϫ g for 10 min. The pellet was homogenized in 10 ml of TE and centrifuged at 1,100 ϫ g for 5 min; 8.8 ml of the supernatant was centrifuged at 12,100 ϫ g for 3 min. The pellet was stored at Ϫ20°C. Isolation of PIC from the inclusion bodies was carried out as described (12,13). The pellet was suspended three times in TE buffer containing 2% Triton X-114, followed by a centrifugation at 12,100 ϫ g for 2.5 min. The supernatant was discarded. Finally the pellet was solubilized in 400 l of TE containing 1.5% SLS. After addition of 1% C 12 E 8 the SLS-solublized protein was used for reconstitution directly or for subsequent dialysis. Functional reconstitution was not possible without addition of C 12 E 8 or another nonionic detergent (see "Results" and Table I).
Chemical Modification of Monomeric PIC-Chemical modification of the monomeric PIC was achieved by incubating the SLS-solubilized PIC (before the addition of C 12 E 8 ) with a freshly prepared solution of NEM (final concentration 2 mM) for 15 min in the dark at 4°C. Excess NEM was removed by adding 10 mM DTT, 1% C 12 E 8 was added afterward as described above.
Isolation of the Heterodimer-The Ni-NTA-agarose column was equilibrated in buffer A (300 mM NaCl, 50 mM Na 2 HPO 4 , pH 8.0). After application of the protein the column was washed with buffer B (buffer A containing 10% glycerol (v/v)). Elution was carried out using an imidazole gradient (0 -0.5 M imidazole in buffer B). PIC protein was eluted at an imidazole concentration of 0.2 M. The fractions were collected and concentrated using Centricon 30 tubes (Amicon, Eschborn, Germany). After a 5-fold dilution with TBS buffer plus detergent (50 mM Tris-HCl, 150 mM NaCl, 0.1% C 12 E 8 (v/v), pH 7.4) the protein was applied to an anti-FLAG affinity column equilibrated with TBS buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4) in a recycling procedure, i.e. the affinity chromatography was repeated five times. After washing the column with TBS the protein was eluted with TBS containing the FLAG-peptide. The eluted fractions were concentrated and used for reconstitution.
Polyacrylamide Gel Electrophoresis and Western Blot Analysis-SDS-polyacrylamide gel electrophoresis was carried out according to the method of Schä gger and von Jagow (41) using gels containing 10% acrylamide and 6 M urea. For analysis under native conditions, urea was replaced by 12% glycerol, and SDS by SLS. For immunological detection of the different constructs the SDS gels were blotted to polyvinylidene difluoride membranes in a semi-dry blotter (Pharmacia, Freiburg, Germany) (38). The FLAG-specific antibody or the Ni-NTAconjugate was added. In the case of the FLAG antibody the second alkaline phosphatase-conjugated antibody was added subsequently. The color reaction was initiated by adding freshly prepared 5-bromo-4chloro-3-indolyl phosphate (BCIP/NBT reagent) solution. The reaction was stopped after 10 min by adding 10 mM EDTA.
Reconstitution Procedure-PIC was reconstituted into pre-formed phospholipid vesicles by using the amberlite method (42). This method was modified (13) with regard to the applied phospholipid/ protein and phospholipid/detergent ratio. Maximum exchange rates were obtained with a phospholipid concentration of 16 mg/ml, a phospholipid/protein ratio of 6.25 g/mg, and a detergent/phospholipid ratio of 0.62 mg/mg. This means that 70 l of Triton X-114 (10%, v/v), 112 l of preformed liposomes (10% of egg yolk phospholipids in 50 mM KCl, 20 mM HEPES, KP i 20 mM, pH 6.5), 20 l of protein solution; HEPES (pH 6.5, final concentration 50 mM) and phosphate (final concentration 30 mM) were added up to 700 l. A detergent/ amberlite ratio of 12 mg/g in combination with 15 column passages was used to remove the detergent (43).
Measurement of Transport Activity and Calculation of Rates-The methods were identical to those described for the analysis of the aspartate/glutamate carrier (44,45) and PIC (13,46). Transport activity was determined using forward exchange experiments (44). The assay was started by adding labeled phosphate. The time course of isotope equilibration was fitted to the data points according to a single exponential function {y ϭ A ⅐ (1 Ϫ e Ϫkt ) ϩ B} which delivered the apparent time constant k [min Ϫ1 ]. The specific activity (mol/min ⅐ mg of protein) was calculated from k (min Ϫ1 ), from the final value of the isotope equilibration (dpm), the specific radioactivity (dpm/nmol), the volume of the proteoliposome fraction (ml), and the protein concentration (g/ml) as published previously (44).

Construction of Different
Monomers-In order to be able to monitor the formation of defined dimeric proteins, the individual monomers and the different types of dimers must be experimentally distinguishable. In the solublized state mitochondrial carriers are assumed to be dimers of identical monomers, PIC monomers had thus to be rendered different artificially. For this purpose, we constructed monomers of the wild type PIC with two different tags, namely the His-tag and the FLAGtag. These tagged PIC monomers were separately and heterologously expressed in E. coli inclusion bodies, solubilized, and reconstituted. These two constructs, after reconstitution by themselves into proteoliposomes showed specific activities of phosphate transport comparable to the wild type protein (see below).
Reconstitution of Solublized PIC from Different Association States-We did not succeed in directly reconstituting PIC solubilized from inclusion bodies in SLS as the only detergent, however, after the simple addition of appropriate other detergents, reconstitution of the SLS-solubilized PIC was successful (see "Experimental Procedures and Table II). For proper formation of dimers and for optimum reconstitution, we tried several detergents for replacing SLS as the solublizing agent, some of which are listed in Table I. When considering both protein recovery after the dialysis step, in which protein is lost due to aggregation, as well as the specific transport activity obtained after reconstitution of the dialyzed protein, the polyoxyethylene detergent C 12 E 8 proved to be most favorable. Consequently, this detergent was used in all further experiments. The various constructs and combinations were all active when reconstituted from preparations in which C 12 E 8 was used as detergent (cf. Fig. 3 and Table II).
Detailed studies on the mitochondrial ADP/ATP carrier and uncoupling protein using analytical ultracentrifuge techniques have shown that these mitochondrial carrier proteins exist as dimers after solubilization with Triton X-100 (19,20). Mitochondrial carriers used for these studies were isolated from intact mitochondria. The presence of monomers of these transport proteins has only been demonstrated in SDS gels. It is not known, however, under which conditions these dimers can be formed. The experimental strategy used in the present study could be applied due to the fact that we found that PIC was in a monomeric state when solubilized from E. coli inclusion bodies using SLS. This was established by several kinds of experiments.
By gel electrophoresis using SLS as detergent instead of SDS, we showed that PIC, when solubilized from inclusion bodies, is in monomeric form (about 30 kDa), just like the protein in SDS gels (Fig. 1). Although addition of the polyoxyethylene detergent C 12 E 8 significantly improved the reconstitutibility of PIC, it did not change its behavior in the nondenaturing SLS gel. After dialysis and detergent exchange to C 12 E 8 , the apparent molecular weight in SLS was significantly increased, although not as much as would have been expected for a dimer. Besides the observation of an increased apparent molecular weight in SLS gels, however, the oligomeric (dimeric) state of PIC after dialysis was proven by the fact that stable homo-and heterodimers could be formed under these conditions (see below). It was not possible to reconstitute PIC in functionally active form from the SLS-solublized state directly (experiments not shown). This, however, was achieved by using SLS-solubilized protein after addition of nonionic detergents (see below). The conformational state of PIC in SLS micelles after solubilization from inclusion bodies was investigated using attenuated total reflection Fourier transform infrared spectroscopy. This lead to an estimation of the ␣-helix content which is clearly too low (experiments not shown) when compared with results obtained for the purified ADP/ATP carrier by CD spectroscopy (47). A further experimental indication for PIC being in an unfolded state when solublized in SLS is the drastically changed accessibility of cysteines to the alkylating reagent NEM in comparison to PIC in functionally active form (see below). It is noteworthy, that by cross-linking studies applying intermolecular disulfide bridges it was found, that the ADP/ATP carrier is not in the correctly folded state even when solubilized in Triton X-100 (21).
An interesting reconstitution experiment corroborated our findings of SLS-solubilized PIC being in the monomeric state as obtained by gel electrophoresis. In Fig. 2, the observed absolute transport activity of the reconstituted PIC protein is correlated to the amount of PIC incorporated into a constant amount of phospholipid. A linear dependence at low protein/phospholipid ratios for this kind of experiment has been found for many mitochondrial carriers (5,6) and other secondary transporters, too (48). This is also true for PIC, when isolated from intact mitochondria (Fig. 2). The same result holds for PIC isolated from inclusion bodies when the detergent SLS has been exchanged for the polyoxyethylene detergent C 12 E 8 . Interestingly, PIC solubilized from inclusion bodies by SLS in the presence of added C 12 E 8 , i.e. before removal of the detergent SLS, showed a completely different pattern. At low PIC/phospholipid ratios, the inserted PIC protein was not active in  phosphate transport, only after a significant amount of PIC has been added was a sharp (sigmoidal) rise of the reconstituted carrier activity observed. In order to be sure that the sigmoidal dependence was not caused by an improper insertion of PIC into the proteoliposomes at low protein/lipid ratios, we applied the proteoliposomes containing His-PIC to polyvinylidene difluoride membranes and subjected the samples to Western blotting. The inset in Fig. 2 demonstrates that PIC was inserted properly into the proteoliposomes.
Construction and Reconstitution of Heterodimers-For the experimental strategy used it was necessary to obtain a pure preparation of PIC heterodimers. The first prerequisite, the controlled formation of dimers in solution, was already described above. Another prerequisite is a sufficient stability of the dimers formed. Experiments which proved the stability of the dimeric constructs will be described at the end of "Results." The third prerequisite, finally, is the ability to experimentally select particular heterodimers from the set of different forms of PIC obtained after solublization and dimer formation. When starting with the two differently tagged monomers, after dialysis we obtain a mixture of both types of homodimers ([His-PIC] 2

and [FLAG-PIC] 2 ), the desired heterodimer ([His-PIC/ FLAG-PIC]), and in addition presumably residual amounts of not correctly folded and assembled monomers ([His-PIC] and [FLAG-PIC])
. In order to select the heterodimer from this mixture, we applied two consecutive affinity columns (Fig. 3). In the first step, the starting mixture was applied to a Ni-NTA affinity column. The monomer [FLAG-PIC] and the homodimer [FLAG-PIC] 2 was not bound but was directly eluted, as was proven by Western blotting for the two tags (experiment not shown). The bound species which all carry the His tag ([His-PIC], [His-PIC] 2 , and [His-PIC/FLAG-PIC]) were then eluted by imidazole buffer. After concentration and exchange of buffer this eluate was applied to a FLAG-antibody affinity column, to which the heterodimer [His-PIC/FLAG-PIC] was bound as the only protein species. After elution the protein was reconstituted and analyzed kinetically. As a control, also the other species, i.e. homodimers formed and the eluates from the different columns, were reconstituted and analyzed for phosphate transport (data not shown). In Fig. 4 the kinetics of reconstituted PIC proteins from different steps of purification is shown. FIG. 3. Schematic drawing of the procedure to isolate defined heterodimers. PIC constructs with two different molecular tags, namely His-tag and FLAG-tag, respectively, were expressed in E. coli. After solublization of the inclusion bodies by the detergent SLS, PIC was found to be in monomeric form. The monomers were mixed and the detergent SLS was exchanged for C 12 E 8 which led to a mixture of different monomeric, homo-and heterodimeric forms. This mixture was applied to Ni-NTA-agarose columns and the imidazole eluate contains only constructs with His-tags. After a second FLAG-antibody column, the pure heterodimer is in the FLAG-peptide eluate. Both the specific activity, resembling the functionality of the protein, and the shape of the kinetics, being correlated with the size and integrity of the proteoliposomes (49), was found to be very similar for the different preparations. Table II summarizes the phosphate transport activity of the relevant PIC constructs, whether starting from monomeric or from different dimeric forms.
It should be mentioned that the use of the Ni-NTA column was only possible after exchanging the detergent SLS for nonionic detergents, otherwise binding of His-tagged PIC to the affinity columns was significantly reduced. Although the procedure as described here could successfully be used to isolate pure heterodimers, there was a problem with protein stability. The consecutive application of solublized PIC to two different columns with intermediate dialysis steps led to inactivation and precipitation and therefore to a significant loss of protein, i.e. the final yield of active heterodimer was low. Based on the results described so far, we therefore developed a modification of this method, which makes possible the use of one single column only, at least for particular experiments. For this purpose, the two different monomeric PIC constructs were mixed not in an 1/1 ratio but in a ratio of 1/7 ([His-PIC] and [FLAG-PIC], respectively). On the one hand, by this strategy the content of the desired [His-PIC]/[FLAG-PIC] in the mixture applied to the Ni-NTA column was reduced from 50% in case of a 1/1 mixture to 22% after mixing in a 1/7 ratio. On the other hand, after application of the Ni-NTA column the relative content of the heterodimer in the eluate is more than 93%, i.e. the contribution of the contaminating [His-PIC] 2 homodimer is only about 6% and thus the second FLAG antibody column could be neglected for qualitative experiments in which we were interested in a high specific activity of the heterodimer.
The Heterodimeric PIC in the Phospholipid Membrane-We have used the monomeric and dimeric PIC constructs so far to define the state of aggregation and the reconstitutibility of different forms of the carrier. The major aim of this work, however, to prove whether PIC when inserted into the bilayer membrane is functioning in the dimeric form only, cannot be achieved by this approach, since all kinds of dimers used were similarly active. A discrimination of the actual state of PIC in the membrane requires (i) the construction of a dimer from two different monomers which are characterized by a clearly different state of activity each and (ii) the analysis of the functional properties of this construct after reconstitution into the membrane.
For this purpose, we applied chemical modification by NEM. In contrast to PIC from beef heart, it has been shown for intact PIC from S. cerevisiae that NEM is unable to block its function, since the NEM-sensitive cysteine at position 42, as present in the beef heart PIC, is lacking in the yeast carrier (50). This is of course true for PIC isolated from inclusion bodies, too, when NEM is applied to the reconstituted protein, i.e. when PIC is in the dimeric form and in the native state of conformation (Table  III) (11). The activity of S. cerevisiae PIC, however, was completely blocked by NEM when the alkylating reagent was applied to the SLS-solubilized protein, i.e. before reconstitution. Obviously, this is due to the fact that the protein is in the monomeric state and (partially) unfolded under these conditions (see above) which makes at least one of the three available cysteines of PIC from yeast accessible for modification.
Reconstitution of monomers treated with NEM, as well as reconstitution of homodimers constructed from NEM-treated monomers not surprisingly led to proteoliposomes which were completely inactive in phosphate transport (Table III). When we reconstituted dimers consisting of one active and one inactive, NEM-treated monomer each, the heterodimeric state of which was proven by application of different molecular tags as described above, the membrane-inserted complexes were completely inactive, too. This result is a clear indication for the fact that only the dimer can be the active form of PIC in the membrane and not the monomer.
There are, however, two possible objections against this interpretation. First, it may be argued that NEM-treated PIC proteins are per se able to inactivate other basically active monomers in some kind of trans-effect. A related argument is based on the possibility that NEM might be carried over during the isolation, which then would lead to inactivation of the added active PIC monomers which were not NEM treated. These arguments were ruled out by control experiments in which NEM-inactivated monomers ([His-PIC(NEM)]) and active monomers ([FLAG-PIC]) were mixed and reconstituted directly, i.e. without forming and separating defined heterodimers. The data of Table III prove that  the same assay mixture. The reduced activity of the reconstituted mixture of active and inactive monomers is due to the fact that the expected share of [FLAG-PIC] 2 homodimers, the only active species available in the mixture, is about 25% of all dimers formed. In order to rule out a further argument concerning a possible bias in the combination of modified and unmodified monomers, which would lead to deviations from the calculated amount of active dimer formed, we carried out a set of control experiments where the unmodified and modified monomers, respectively, from both wild type and tagged constructs were mixed in various ratios and reconstituted (Fig. 5). The two independent series of experiments make sure that (i) the monomers combine according to statistical predictions, and (ii) the result perfectly fits to the assumption of the dimer being the active species in the reconstituted proteoliposomes.
A second type of argument which might weaken our interpretation of the results reported above, namely to prove that the dimeric state is in fact essential for function of PIC in the membrane, is based on the fact that we have not clearly shown so far that the dimeric constructs are stable during the course of the experiments. The stability of the constructs, however, at least within the time of our experiments, is an obvious prerequisite for a correct and quantitative interpretation of the reported experiments. Although the ability by itself to isolate heterodimers on the basis of the presence of the two tags in one single protein complex argues for their stability, at least on a qualitative basis, we performed additional experiments to ver-ify this statement.
If the PIC dimers in the membrane are in a dynamic state, i.e. if they dissociate and re-associate in a reasonable time scale, we should observe the formation of active [FLAG-PIC] 2 homodimers after an appropriate time from the originally inserted [His-PIC(NEM)/FLAG-PIC] heterodimers. Fig. 6 shows that this is definitely not the case, even after a prolonged incubation time of 22 h. As a control, the active [His-PIC/ FLAG-PIC] heterodimer was measured, too, in order to exclude that a significant inactivation of the protein occurs during the time course of the experiment. There is an obvious objection against the interpretation of this experiment arguing for the stability of the complexes in the membrane. It must be guaranteed that at least two dimers are present in one liposome, otherwise there would in principle never be the chance to form active dimers from the two active monomers of the heterodimers constructed. The reconstitution conditions of the experiment reported in Fig. 6, however, are carried out in a protein/ phospholipid ratio of milligrams/gram (5.1 g of protein/mg of phospholipid) which is far beyond the critical value found in the titration experiment (Fig. 2), thus providing a sufficient number of active monomers for the formation of active homodimers if in fact the state of aggregation would be a dynamic one. DISCUSSION The definition of the state of aggregation of membrane-embedded carrier proteins was the topic of numerous studies  involving techniques of cross-linking, electron microscopy, electrophoresis, chromatography, rotational diffusion, ultracentrifugation, radiation inactivation, as well as reconstitution titration. As a matter of fact, the state of aggregation is not only a question of correct numbers. It is highly relevant for the functional models of solute carriers whether carrier proteins act as monomers, dimers, or higher aggregates. In view of the fact that the three-dimensional structure of not a single solute carrier protein is available so far, solving this question becomes even more interesting. The most advanced structural analysis of a solute carrier correlating the spatial arrangement of particular amino acid residues with carrier function, the lactose permease of E. coli (51), offers an interesting view on the hypothetical substrate translocation pathway. Nevertheless, even in this case the correct arrangement of transmembrane segments as well as the true state of aggregation is not yet completely clear. The major conclusion reached from reconstituting defined constructs of PIC are that the dimer is functionally active, whereas the monomer is not. Particularly striking was the finding that reconstitution of the monomeric form of PIC resulted in a strongly sigmoidal titration curve, indicating that the first monomers inserted into liposomes are not able to function in phosphate transport. We interpret the sigmoidal shape of the titration curve by the obvious assumption that more than one functionally active monomer must be present in one liposome, in order to be able to form a transport-active complex. It should be noted that the result of the experiment on reconstituting partially inactivated heterodimers does not exclude higher states of aggregation. On the basis of experiments with the mitochondrial ADP/ATP carrier using fluorescent nucleotide analogs, in fact a tetrameric state of the functional unit of this carrier has been suggested (18). However, the observed shape of the titration curves of reconstitution argues against the presence of higher aggregates being essential for transport function. If this would be the case, a sigmoidal shape of the dependence of activity on the amount of added carrier protein would have been expected also for the insertion of dimers into liposomes.
A further conclusion can be drawn from the experiment on reconstitution of heterodimers of both active and inactive subunits. The fact that this construct is inactive not only proves the stability of the dimeric form. It also indicates that mono-mers do not function by themselves within the complex, i.e. that there is some kind of cross-talk between the two subunits. It may be assumed that only one pathway for phosphate exists within the dimer, and that the inactivation of one subunit is sufficient to render the whole complex inactive. An alternative explanation would be the existence of two independent pathways through the two PIC subunits, the function of each of which depends on the integrity of the corresponding pathway in some kind of mutual interaction. It should be noted that previous kinetic studies both on the mitochondrial aspartate/ glutamate and phosphate carrier, respectively, were the basis for defining of the family of mitochondrial carriers with a common kinetic mechanism, namely a simultaneous bisubstrate kinetics (5,6,52,53). This analysis argued for the presence of two substrate pathways in mitochondrial carrier proteins. An experimental basis to decide this question on a molecular level seems to be at hand now by using more sophisticated versions of heterodimers created by methods described in the present publication.
Finally, a further conclusion can be drawn. For a correct interpretation it was necessary to prove that the dimeric constructs were stable in the course of the experiments. Whereas the stability of the dimeric forms in the solublized state, at least in the time range of a few hours, was already proven by the successful application of the strategy for selective isolation, this was not equally simple for the membrane-inserted state. By comparing the activity of defined heterodimers consisting of two active monomers on the one hand, and of both an active and an inactive monomer on the other, we showed that PIC dimers are stable at least within a time range of 22 h. Thus, beside the fact that the dimeric form is essential for function, it has, to our knowledge, been shown for the first time here that there is no dynamic equilibrium between monomers and dimers of mitochondrial carriers in the membrane-inserted state.
FIG. 6. Activity of heterodimers after prolonged incubation time. The transport activity of the different constructs was determined as uptake of labeled phosphate in the homologous phosphate/phosphate exchange mode. The activity of untreated heterodimers (HIS-PIC/FLAG-PIC) (closed symbols, two independent experiments) and of partially NEM-treated heterodimers (HIS-PIC-(NEM)/FLAG-PIC) (open symbols, two independent experiments) was measured.