Expression in Escherichia coli, Functional Characterization, and Tissue Distribution of Isoforms A and B of the Phosphate Carrier from Bovine Mitochondria*

The two isoforms of the mammalian mitochondrial phosphate carrier (PiC), A and B, differing in the sequence near the N terminus, arise from alternative splicing of a primary transcript of the PiC gene (Dolce, V., Iacobazzi, V., Palmieri, F., and Walker, J. E. (1994) J. Biol. Chem. 269, 10451–10460). To date, the PiC isoforms A and B have not been studied at the protein level. To explore the tissue-distribution and the potential functional differences between the two isoforms, polyclonal site-directed antibodies specific for PiC-A and PiC-B were raised, and the two bovine isoforms were obtained by expression in Escherichia coliand reconstituted into phospholipid vesicles. Western blot analysis demonstrated that isoform A is present in high amounts in heart, skeletal muscle, and diaphragm mitochondria, whereas isoform B is present in the mitochondria of all tissues examined. Heart and liver bovine mitochondria contained 69 and 0 pmol of PiC-A/mg of protein, and 10 and 8 pmol of PiC-B/mg of protein, respectively. In the reconstituted system the pure recombinant isoforms A and B both catalyzed the two known modes of transport (Pi/Pi antiport and Pi/H+ symport) and exhibited similar properties of substrate specificity and inhibitor sensitivity. However, they strongly differed in their kinetic parameters. The transport affinities of isoform B for phosphate and arsenate were found to be 3-fold lower than those of isoform A. Furthermore, the maximum transport rate of isoform B is about 3-fold higher than that of isoform A. These results support the hypothesis that the sequence divergence between PiC-A and PiC-B may have functional significance in determining the affinity and the translocation rate of the substrate through the PiC molecule.

The transport of inorganic phosphate across the inner mitochondrial membrane into the matrix compartment is essential for the oxidative phosphorylation of ADP to ATP. This transport is catalyzed by the phosphate carrier (PiC) 1 which has been purified from different sources and reconstituted into liposomes in an active form (1)(2)(3)(4). In the reconstituted system the native bovine PiC protein catalyzes the P i /H ϩ symport as well as the P i /P i exchange, which functions by a sequential mechanism (5). The primary structure of the mature PiC is made up of three tandemly related domains about 100 amino acids in length (6). These repetitive elements are related to those found in the other well characterized members of the mitochondrial carrier family (see Refs. 7-10 for reviews) (11,12). They are also found in a number of other proteins of known sequence but of unknown function, which therefore belong to the same protein superfamily (7)(8)(9)(10). By examination of the transmembrane topography of the PiC in the inner mitochondrial membrane, it has been proposed that both the N-and C-terminal regions of the PiC protrude toward the cytosol and that the polypeptide chain spans the membrane six times (13). Only one gene for the PiC has been detected in man and cow (14), of which the human one has been localized to chromosome 12q23 (15). In both man and cow two closely related exons named IIIA and IIIB appear to be alternatively spliced (14). The alternative splicing mechanism affects amino acids 4 -45 of the mature PiC. More recently, by Northern blot analysis, isoform A has been found to be limited to heart and skeletal muscle, whereas isoform B is expressed in all tissues examined (16).
The biochemical characterization of potential functional differences between isoforms A and B of the mitochondrial PiC has so far been precluded by the lack of a purification procedure for separating the two isoforms. Similarly, tissue distribution studies of PiC-A and PiC-B at the protein level have been prevented by the lack of specific antibodies. We therefore raised antibodies against PiC isoforms A and B and produced two recombinant proteins that have either the PiC-A or the PiC-B sequence. In this study we demonstrate that the two PiC isoforms are tissue-specific, and that the two recombinant isoforms exhibit different transport affinities (K m ) and specific activities (V max ) after purification and reconstitution into liposomes. To our knowledge, this is the first time that a functional comparison of mitochondrial carrier isoforms has been performed.

Construction of Expression Plasmids Coding for PiC Isoforms A and B-
The coding regions for the mature bovine PiC isoforms A and B were amplified from 10 ng of bovine heart cDNA (isoform A) and bovine liver cDNA (isoform B), respectively, by 30 cycles of polymerase chain reactions. The forward and reverse oligonucleotide primers employed in these reactions corresponded to nucleotides 444 -453 linked to 1770 -1781 (forward primer A), 444 -453 linked to 2051-2062 (forward primer B), and 5955-5978 (reverse primers A and B) of the bovine PiC gene sequence (14). The forward and reverse primers carried NdeI and XhoI restriction sites as linkers, respectively. The absence of the stop codon in the reverse primer sequences led to the expression of the two PiC * This work was supported by the CNR Target Project "Biotechnology" and by the Ministero dell'Università e della Ricerca Scientifica e Tecnologica (MURST). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ isoforms with an extra leucine and glutamic acid, encoded by the XhoI restriction site sequence, plus a tail of six histidines at their C terminus. The reaction products were cloned in the expression vector pET21b, and their sequences were verified by the modified dideoxy chain termination method (17).
Bacterial Expression and Purification of the Bovine PiC Isoforms-The expression of PiC-A and PiC-B as inclusion bodies in the bacterial cytosol was accomplished in Escherichia coli BL21(DE3), as described first for the bovine oxoglutarate carrier (18) and then for several other mitochondrial carriers (11, 12, 19 -21). Control cultures containing the empty pET21b vector were processed in parallel. Purified inclusion bodies (18), suspended in TE buffer (10 mM Tris-HCl and 0.1 mM EDTA, pH 8.0), were solubilized in 1.67% (w/v) N-dodecanoylsarcosine (Sarkosyl) for 5 min at 0°C. The solution was diluted 20 times with SSP buffer consisting of 0.1% Sarkosyl, 0.5 M NaCl, and 20 mM P i , pH 8.0, and centrifuged at 12,000 ϫ g for 10 min at 4°C. The supernatant was chromatographed on Ni ϩ -nitrilotriacetic acid-agarose affinity column (Quiagen). Unspecifically bound proteins were washed with the SSP buffer supplemented with increasing concentrations of histidine. Pure PiC isoforms A and B were recovered when histidine reached 10 mM. The purified proteins were desalted by a Sephadex G-25 column (PD-10 Pharmacia) and stored at Ϫ70°C. All chromatographic steps were performed at 4°C.
Reconstitution of the Recombinant PiC Isoforms into Liposomes-Purified PiC-A and PiC-B isoforms were reconstituted by cyclic removal of the detergent with a hydrophobic column (22,23). The composition of the initial mixture used for reconstitution was 200 l of purified PiC isoform (about 400 ng of protein), 100 l of 10% Triton X-114, 90 l of 10% phospholipids in the form of sonicated liposomes (1), 30 mM P i (except where otherwise indicated), 20 mM Pipes (pH 6.5), 0.63 mg of cardiolipin (Sigma), and water to a final volume of 700 l. After vortexing, this mixture was recycled 13 times through an Amberlite column (Fluka) (3.2 ϫ 0.5 cm) preequilibrated with a buffer containing 20 mM Pipes (pH 6.5) and the substrate at the same concentration as in the starting mixture. All operations were performed at 4°C, except the passages through Amberlite, which were carried out at room temperature.
Transport Measurements-The reconstituted P i /P i antiport activity was determined by measuring the uptake (forward exchange) or the efflux (backward exchange) of [ 33 P]phosphate in exchange for unlabeled substrate (22). For backward exchange measurements, as well as for measuring the efflux of phosphate as a P i /H ϩ symport, the proteoliposomes containing internal P i were prelabeled, immediately after reconstitution, by carrier-mediated exchange equilibration, i.e. by adding carrier-free [ 33 P]phosphate of high specific radioactivity for 20 min. The external substrate was removed from proteoliposomes on a Sephadex G-75 column (22) in the presence of a reversible inhibitor, 5 M pchloromercuribenzenesulfonate (23), in order to avoid the efflux of internal substrate by the P i /H ϩ symport. Transport was started by adding 5 mM dithioerythritol and labeled P i at the indicated concentrations (forward exchange), 5 mM dithioerythritol and 30 mM cold P i (backward exchange), or 5 mM dithioerythritol alone (P i /H ϩ symport). In all cases the carrier-mediated transport was terminated by addition of 25 mM pyridoxal 5Ј-phosphate (23). In control samples the inhibitor was added at time 0 according to the inhibitor stop method (22). The assay temperature was 25°C. All transport measurements were carried out at the same internal and external pH value of 6.5. Finally, the external substrate was removed, and the radioactivity in the liposomes was measured (22). In forward exchange kinetic measurements, the initial transport rate was calculated in millimoles/min/g of protein from the time course of isotope equilibration, as has been published previously (22). In the case of backward exchange and P i /H ϩ symport, the decrease in radioactivity inside the liposomes was fitted to the equation ␣ ϭ 100(1 Ϫ e Ϫkt ) (where ␣ is the percentage of isotopic equilibration) (24). The rates were expressed as apparent velocities, i.e. the product of k and the substrate concentration inside the liposomes, and they are directly proportional to the actual transport rate (22,24).
Western Blot Analysis-The antisera anti-PiC-A and anti-PiC-B were raised against the amino acid sequences 1-10 of the mature bovine PiC-A and 7-16 of the mature bovine PiC-B, respectively. The synthesis of peptides, the coupling to ovalbumin of the first peptide through Tyr-6 or Tyr-10, and of the second one through Cys-7, as well as the generation of the antibodies were carried out as described previously (25). Samples, solubilized in SDS sample buffer and boiled for 5 min, were separated by SDS-PAGE (as described in Capobianco et al. (26)) and electrotransferred to nitrocellulose membranes. The membranes were incubated with the antiserum anti-PiC-A or the antiserum anti-PiC-B (each diluted 1:40,000 in PBS-TM (PBS containing 0.5% (w/v) non-fat dried milk and 0.05% (v/v) Tween-20)) for 2 h, washed three times in PBS-TM for 10 min each, and then incubated for 2 h with the secondary antibody, horseradish peroxidase conjugated anti-rabbit IgG (Pierce) diluted 1:2000 in PBS-TM. The membranes were washed and developed using the ECL system (Amersham Pharmacia Biotech). The films were scanned with an LKB 2202 Ultroscan laser densitometer. To determine the amount of PiC-A and PiC-B in mitochondria, standard calibration curves were constructed using 5-30 ng pure recombinant PiC-A or PiC-B as the standard. Nitrocellulose membranes containing the standards and the mitochondrial samples were simultaneously immunodecorated as described above. Once it was checked that mitochondrial sample loading was within the linear range of the calibration curves, the densitometric signal intensity was used to measure the amount of PiC-A and PiC-B (27,28).
Other Methods-After SDS-PAGE, proteins were either stained with Coomassie Blue dye or transferred to polyvinylidene difluoride membranes, stained with Coomassie Blue dye, and their N-terminal sequences determined with a pulsed liquid protein sequencer (Applied Biosystems 477A). The pure recombinant PiC isoforms were estimated from Coomassie Blue-stained SDS-PAGE gels with an LKB 2202 Ultroscan laser densitometer, using carbonic anhydrase as protein standard. To assay the protein incorporated into liposomes, the vesicles were passed through a Sephadex G-75 column, centrifuged at 300,000 ϫ g for 30 min and delipidated with organic solvents as described in Capobianco et al. (26). Then, the SDS-solubilized protein was determined by comparison with carbonic anhydrase in SDS gels and/or by quantitative immunodecoration (as described above) with the antiserum anti-PiC-A and the antiserum anti-PiC-B.

RESULTS
The sequence divergence between isoforms A and B of PiC is confined to the N-terminal region of the PiC protein and is precisely localized between amino acids 4 -45 of the mature protein corresponding to exons III A and III B of the PiC gene (Fig. 1).
The isolation of the cDNA clones encoding PiC isoforms A and B (see "Experimental Procedures") created the possibility to produce, in bacteria, homogeneous A and B isoforms of PiC, which could be used to explore the functional differences between the two isoforms. To this end, PiC-A and PiC-B isoforms were expressed in E. coli, purified, and refolded into the active form using a pET expression system and the Sarkosyl refolding procedure described under "Experimental Procedures." Expression of the Bovine PiC Isoforms-After induction with isopropyl-␤-D-thiogalactopyranoside, PiC-A and PiC-B accumulated in the bacterial cytosol as inclusion bodies (Fig. 2, A and  B, lane 1). The presence of the histidine tail at the C-terminal end of the expressed PiC isoforms allowed their purification by a Ni ϩ -agarose affinity column (Fig. 2, A and B, lane 5). About 2 mg of each isoform were obtained per liter of bacterial culture. The identity of PiC-A and PiC-B was confirmed by the determination of their N-terminal sequences, and by their reaction with specific antisera, as shown in Fig. 3.
Tissue-specific Distribution of PiC Isoforms A and B-To prepare antisera specific for PiC-A and PiC-B, we first raised antibodies against three synthetic peptides corresponding to residues 1-10 and 8 -17 of PiC-A, and residues 7-16 of PiC-B. Only the antibodies generated by using the latter two peptides were found, at the dilution used in this work, to be highly specific for PiC-A and PiC-B, respectively (Fig. 3, A and B,  lanes A and B). We then investigated the presence of the two PiC isoforms in the mitochondria of several bovine tissues by Western blotting. Fig. 3A shows that the antiserum specific for PiC-A immunodecorated a band of 34 kDa in the lysates of heart, diaphragm, and skeletal muscle, but none in the lysates of liver, lung, brain, and kidney. On the contrary, the antiserum specific for PiC-B (Fig. 3B) reacted with a band of 34 kDa in the mitochondrial lysates of all the tissues investigated. These results indicate that the PiC-A isoform is present only in muscles, whereas the PiC-B isoform is present in all the mitochondria tested. It should be noted that the autoradiography films were exposed for 15 s in the case of the immunoblotting with anti-PiC-A antiserum (Fig. 3A) and 60 min in the case of the immunoblotting with the anti-PiC-B antiserum (Fig. 3B). Therefore, the expression levels of PiC-A in mitochondria of heart, skeletal muscle, and diaphragm are much higher than those of PiC-B in the mitochondria of all the tissues. Furthermore, no immunoreaction was observed with the anti-PiC-A antiserum and the lysates of liver, lung, brain, and kidney when the exposure time was prolonged to 60 min.
To quantify PiC-A and PiC-B levels in mitochondria, various amounts of mitochondrial samples were loaded onto the gel and immunoblotted simultaneously with the appropriate range of recombinant PiC-A and PiC-B standards (see "Experimental Procedures"). In four determinations, the abundance of PiC isoforms was calculated to be 69 Ϯ 13 pmol/mg of protein of PiC-A in heart mitochondria, and 10 Ϯ 2 and 8 Ϯ 1 pmol/mg of protein of PiC-B in heart and liver mitochondria, respectively.
Functional Characterization of Recombinant PiC Isoforms A and B-For kinetic analysis of the two PiC isoforms the reconstitution procedure has been optimized by adjusting the parameters that influence the efficiency of carrier incorporation into the liposomes. With both isoforms optimal transport activity was obtained with 0.6 g/ml and 12.8 mg/ml protein and phospholipid concentration, respectively, with a Triton X-114/ phospholipid ratio of 1.1 and with 13 passages through the same Amberlite column (see "Experimental Procedures"). Furthermore, since cardiolipin has been shown to be essential for the reconstitution of purified PiC (1, 29, 30), we tested the effect of this phospholipid on reconstitution of the two recombinant PiC isoforms. Fig. 4 shows that the activity of both isoforms was increased 5-fold by optimal concentrations of cardiolipin and that isoform B was more active than isoform A at Km and V max Values-Kinetic constants were determined for pure recombinant PiC-A and PiC-B over a wide range of P i concentrations from a standard double-reciprocal set of experiments (see Table I). The transport affinities (K m ) for P i on the external membrane surface of the reconstituted PiC-A and PiC-B isoforms, measured by the forward exchange method, were determined to be 2.21 Ϯ 0.15 and 0.78 Ϯ 0.02 mM for PiC-A and PiC-B, respectively. The specific activities of recombinant PiC-A and PiC-B (V max ) were 125 Ϯ 18 and 349 Ϯ 22 mmol/min/g of protein for PiC-A and PiC-B, respectively. These activities were calculated by taking into account the amount of PiC isoforms recovered in the proteoliposomes after reconstitution. Under our experimental conditions the inserted protein varied between 14 and 20% of the protein added to the reconstitution mixture. Furthermore, no difference in the efficiency of reconstitution (i.e. the share of successfully incorporated protein) was observed between PiC-A and PiC-B.
The analysis of the inward-facing substrate binding site was carried out by using the backward-exchange method for the methodological reasons that have been described previously (22). From these experiments an internal K m of 9.7 Ϯ 0.6 mM for PiC-A and 6.3 Ϯ 0.5 mM for PiC-B was calculated. The intraliposomal K m was also determined by measuring the efflux of [ 33 P]phosphate as P i /H ϩ symport (without external phosphate). Under these conditions the calculated internal K m values were 8.1 Ϯ 0.7 and 7.0 Ϯ 0.5 mM for recombinant PiC-A and PiC-B, respectively, which agree well with the internal substrate affinities for PiC-A and PiC-B observed in the case of the P i /P i exchange.
Substrate Specificity and Inhibitor Sensitivity-The substrate specificity of PiC-A and PiC-B was investigated by measuring the uptake of [ 33 P]phosphate into liposomes preloaded with a variety of substrates (Table II). With both isoforms, high uptake was observed only when P i or arsenate was present inside the liposomes. With all the other compounds tested there was a very low uptake similar to that found in the presence of internal NaCl. These results clearly indicate that both isoforms exhibit a very narrow specificity, as found previously for the PiC in mitochondria (31)(32)(33).
The effects of inhibitors on the [ 33 P]phosphate/phosphate exchange reaction catalyzed by the PiC isoforms A and B were also examined. Both isoforms were inhibited by the sulfhydryl reagents N-ethylmaleimide, mersalyl, p-chloromercuribenzoate, p-chloromercuribenzenesulfonate, and HgCl 2 , and by the lysine reagents pyridoxal 5Ј-phosphate and phenylisothiocyanate (data not shown). These findings agree with the results obtained with PiC in mitochondria (34 -37) and with the purified and reconstituted carrier from mammals (1-3, 5, 23). In addition, the reconstituted P i /P i exchange activity catalyzed by both PiC-A and PiC-B was inhibited by the external addition of the substrates P i and arsenate, whereas it was virtually unaffected by sulfate, malate, ATP, thiosulfate, oxoglutarate, citrate, and glutamate (not shown).
Inhibition by Arsenate-As demonstrated above, both recombinant PiC-A and PiC-B are able to transport arsenate. Dixon (38) plots indicated that the inhibition by arsenate of P i /P i exchange catalyzed by both isoforms was competitive with respect to P i . The K i values estimated from these experiments were 2.04 Ϯ 0.12 and 0.84 Ϯ 0.03 mM in the case of PiC-A and PiC-B, respectively (see Table I).

DISCUSSION
In this work we demonstrate the presence of two PiC isoforms at the protein level in mitochondria of bovine tissues by using antibodies specific for each of the two isoforms. Isoform PiC-A is present in muscles, whereas PiC-B is ubiquitous. This tissue specificity of PiC-A and PiC-B corresponds exactly to the organ distribution of the mRNA of the two isoforms published in Dolce et al. (16). The differences between these results and those reported in Dolce et al. (14) may be explained by a Lineweaver-Burk plots were obtained from forward exchange measurements of exchange velocity, v, or from backward exchange measurements of rate constant, k, under variation of external or internal P i as described under "Experimental Procedures." Substrate concentrations were as follows: P i /P i antiport measured by the forward procedure, 30 mM internal P i , and 0.33-9.0 mM external P i ; P i /P i antiport measured by the backward procedure, 0.3-40 mM internal P i and 30 mM external P i ; P i /H ϩ symport, 0.3-40 mM internal P i in the absence of external P i . For the determination of K i of arsenate, Dixon plots were obtained from forward exchange measurements of exchange velocity, v, in the presence of 30 mM internal P i , and three external concentrations of [ 33   problem of polymerase chain reaction product carryover in the earlier study. Furthermore, we determined the absolute amounts of the immunoreactive PiC isoforms in bovine heart and liver mitochondria by quantitative Western blotting. The PiC content values we found are much lower than those previously determined with other methods (2, 39 -41). Previous attempts to express the mitochondrial PiC from mammals in E. coli have not been successful, probably due to the different expression systems employed (42). To characterize the functional properties of the two bovine PiC isoforms, both proteins were expressed in E. coli and purified by Ni ϩ -affinity chromatography in an active state. The results presented in this report demonstrate that the two recombinant bovine isoforms of PiC, when reconstituted into liposomes, differed markedly in their kinetic characteristics. On the other hand, they exhibited the unique properties of the PiC (as known from intact mitochondria) concerning substrate specificity, inhibitor sensitivity, and cardiolipin requirement. The transport affinity (K m ) for P i of the recombinant PiC-B on the external proteoliposomal side is about 3-fold lower than the corresponding kinetic constant of isoform A. On the internal side, however, the K m 's for P i of the two isoforms are similar and are about one order of magnitude higher than the external K m of PiC-B. The very different K m values of PiC-A and PiC-B on the two membrane sides indicate that both recombinant PiC isoforms are oriented unidirectionally in the proteoliposomal membrane, as previously reported for PiC purified from bovine heart (23). The observed difference in the external transport affinities of PiC-A and PiC-B for P i is substantiated by the K i values of PiC-A and PiC-B for arsenate, the only other substrate of the mitochondrial phosphate carrier (31)(32)(33), which are virtually identical to the corresponding K m values for P i . For an accurate comparison of the V max values of recombinant PiC-A and PiC-B, we determined the amount of protein recovered in the proteoliposomes after reconstitution. Our measurements of inserted protein (14 -20% of the protein added to the reconstitution mixture) are similar to those determined previously for the recombinant yeast PiC (20,24,43). The specific activities reported in this study, which were calculated by using the amount of inserted protein, demonstrate that PiC-B is about 3-fold more active than PiC-A. Our V max values for recombinant PiC-A and PiC-B are higher than those previously determined for the purified bovine heart PiC (23,44), most likely because a reconstitution efficiency of 100% had been assumed in these earlier studies.
The two PiC isoforms A and B differ in just 11 amino acids of the 34 (33) nearest the N-terminal end, of which Gln-5 from PiC-A is missing in isoform B (see Fig. 1). These differences are mainly localized in the first putative membrane-spanning ␣-helical structure of the protein (13). Although it is clear that the variation in transport properties is caused by one or several amino acid differences in this ␣-helical structure, additional investigations are needed to elucidate the role of the first transmembrane ␣-helix in the translocation of P i through the mitochondrial membrane.
The tissue specificity of the two isoforms of the PiC resembles that of the isoforms of other mitochondrial proteins involved in oxidative phosphorylation. Among these proteins are the adenine nucleotide carrier (45)(46)(47)(48), the uncoupling protein (49,50), the human ␥-subunit of the ATP synthase complex (51), the mammalian proteolipid subunit of ATP synthase (52), and the three mammalian subunits (VIa, VIIa, and VIII) of cytochrome oxidase (53,54). The common feature of all these proteins is the presence of at least one heart type isoform, which is abundantly expressed only in muscles, and one liver type isoform, which is ubiquitously expressed in all tissues. Apart from the characterization of the yeast adenine nucleotide carrier 1 (55), the functional differences of the above mentioned isoforms of proteins have not yet been elucidated. The observed differences in the kinetic properties of the two PiC isoforms can account for the differential reliance of the tissues where they are present on oxidative phosphorylation. We suggest that the ubiquitous PiC isoform B matches the basic energy requirement of all tissues and isoform A becomes operative to accommodate the higher energy demands associated with contraction of striated muscle fibers. Thus, during muscle contraction the capacity of isoform B, which has a higher affinity for P i , is overwhelmed, and isoform A with its lower substrate affinity will then be brought into operation by the increased concentrations of cytosolic P i . Therefore, the PiC isoforms A and B, due to their different K m and V max , can be used to modulate the rate of ATP production by oxidative phosphorylation for tissuespecific energetic needs.