The Phosphate Carrier Has an Ability to be Sorted to either the TIM22 Pathway or the TIM23 Pathway for Its Import into Yeast Mitochondria*

Most mitochondrial proteins are synthesized in the cytosol, imported into mitochondria via the TOM40 (translocase of the mitochondrial outer membrane 40) complex, and follow several distinct sorting pathways to reach their destination submitochondrial compartments. Phosphate carrier (PiC) is an inner membrane protein with 6 transmembrane segments (TM1–TM6) and requires, after translocation across the outer membrane, the Tim9-Tim10 complex and the TIM22 complex to be inserted into the inner membrane. Here we analyzed an in vitro import of fusion proteins between various PiC segments and mouse dihydrofolate reductase. The fusion protein without TM1 and TM2 was translocated across the outer membrane but was not inserted into the inner membrane. The fusion proteins without TM1–TM4 were not inserted into the inner membrane but instead translocated across the inner membrane. Functional defects of Tim50 of the TIM23 complex caused either by depletion of the protein or the addition of anti-Tim50 antibodies blocked translocation of the fusion proteins without TM1–TM4 across the inner membrane, suggesting that lack of TM1–TM4 led to switch of its sorting pathway from the TIM22 pathway to the TIM23 pathway. PiC thus appears to have a latent signal for sorting to the TIM23 pathway, which is exposed by reduced interactions with the Tim9-Tim10 complex and maintenance of the import competence.

Newly synthesized organellar and secretory proteins are translocated across or inserted into biological membranes before they become functional. Among several different eukaryotic organelles, mitochondria are unique in being bounded by two biological membranes, the outer and inner mitochondrial membranes, rendering delivery of mitochondrial proteins to their destination suborganellar compartments highly complex. Mitochondria have translocators, membrane protein complexes that recognize targeting/sorting signals of mitochondrial proteins and facilitate their translocation across and/or insertion into the membranes. The outer membrane contains two TOM 1 (translocase of the mitochondrial outer membrane) complexes as translocators, the TOM40 complex and TOB/SAM complex, and the inner membrane contains two TIM (translocase of the mitochondrial inner membrane) complexes as translocators, the TIM23 complex and TIM22 complex (1)(2)(3).
The TOM40 complex mediates translocation across or insertion into the outer membrane for nearly all mitochondrial proteins. Most matrix proteins and some inner membrane proteins are synthesized in the cytosol as precursor proteins with an N-terminal cleavable presequence. After translocation through the TOM40 channel in the outer membrane, the proteins are passed onto the TIM23 complex and move through the TIM23 channel with the aid of MMC (mitochondrial Hsp70-associated motor and chaperone) proteins, i.e. mitochondrial Hsp70 and its partner proteins (4,5). Polytopic inner membrane proteins, including metabolic carrier proteins, are synthesized in the cytosol without a presequence. After crossing the outer membrane via the TOM40 complex, these proteins are bound by the 70-kDa Tim9-Tim10 complex in the intermembrane space (IMS), which likely functions as a chaperone to prevent aggregate formation of the hydrophobic substrate proteins in the aqueous IMS (6,7). Then the Tim9-Tim10 complex delivers the substrate proteins to the 300-kDa TIM22 complex consisting of the core complex (Tim22, Tim18, and Tim54) and the peripheral subunits (Tim9, Tim10, and Tim12), which facilitates insertion of substrate proteins into the inner membrane. The IMS contains many small and soluble proteins, including small Tim proteins that lack presequences. These small IMS proteins reach the IMS through the TOM40 channel with the aid of the inner membrane proteins Tim40/Mia40 (8,9) and Hot13 (10). Although some single-membrane spanning outer membrane proteins are directly inserted into the outer membrane from the TOM40 complex, ␤-barrel outer membrane proteins first reach the IMS side of the outer membrane through the TOM40 complex and are then inserted into the outer membrane with the assistance of small Tim proteins in the IMS and the TOB/ SAM complex in the outer membrane (11).
The TOM40 complex is thus the entry point for mitochondrial protein import pathways, which subsequently diverge from the TOM40 complex into distinct pathways for different sub-mitochondrial destinations. Then a question may arise as to how proteins are sorted among different import pathways after translocation through the TOM40 channel. Although current evidence suggests that at the exit of the TOM40 channel  both Tim50 and the Tim9-Tim10 complex are waiting for the  incoming polypeptide and pass it onto one of the translocators,  i.e. the TIM23, TIM22, and SAM/TOB complexes, it is still  unclear what the sorting signals are and how they are sorted to  different pathways. In the present study, we analyzed the import of fusion proteins containing various deletion mutants of phosphate carrier (PiC), which is a substrate for the TIM22 complex (12,13). Mitochondrial carrier proteins such as PiC and the ATP/ADP carrier are synthesized in the cytosol without an N-terminal cleavable presequence but consist of three structurally related modules (6,7). Each module contains a pair of hydrophobic transmembrane segments (TM1-TM6) connected by a loop that contains a conserved "carrier signature" motif and is exposed to the matrix in the final assembled form. We found here that some of the fusion proteins containing truncated PiC segments were unexpectedly localized to the matrix via the TIM23 pathway instead of the TIM22 pathway. Possible mechanisms for this switch of the sorting pathways will be discussed.  GAL-TIM10, a yeast strain expressing Tim10 under the control of the GAL7 promoter, was constructed as follows. A DNA fragment containing the GAL7 promoter was amplified from plasmid pCgHIS3-GAL7 (15,16) by PCR using primers Tim10-off-F (5Ј-ATAAACAAGCAAAAG-AAGGAACTATCATACTTAGAAAAACAAGCTTGGGTCTTCTGGAGC-3Ј) and Tim10-off-R (5Ј-ATAATTGAGGCTGACCACCACCGAAACCTA-AGAAAGACATTTTTGAGGGAATATTCAACT-3Ј). The amplified DNA fragment was integrated into the chromosome upstream of the TIM10 gene of the wild-type haploid strain W303-1A. Yeast strains were grown in YPD (1% yeast extract, 2% polypeptone, 2% dextrose), lactate (plus 0.1% glucose) medium and lactate (plus 0.1% galactose) medium (17) (with minor modifications). Import Assays-Radiolabeled precursor proteins were synthesized with rabbit reticulocyte lysate by coupled transcription/translation in the presence of [ 35 S]methionine. Mitochondria were isolated from yeast strains D273-10B, W303-1A, and GAL-TIM10 (17) and incubated with radiolabeled precursor proteins in import buffer (600 mM sorbitol, 50 mM Hepes-KOH, pH 7.2, 80 mM KCl, 2.5 mM potassium Pi, 2 mM methionine, 5 mM MgCl 2 , 2 mM ATP, 2 mM NADH, and 1% bovine serum albumin) at 23 or 25°C. Import reactions were stopped by addition of valinomycin to 10 g/ml. The mitochondria were reisolated by centrifugation, and proteins were analyzed by SDS-PAGE and radioimaging. Treatments of mitochondria with proteinase K (PK) and Na 2 CO 3 , and preparation of mitoplasts were performed as described previously (16). Import block with anti-Tim50 antibodies was carried out according to Yamamoto et al. (16).

Truncated PiC Segments
Can Direct DHFR to the Matrix-To reveal the roles of internal segments of PiC in its import into mitochondria, we made DNA constructs for a series of PiC-DHFR fusion proteins between various PiC segments and DHFR (Fig. 1). We synthesized radiolabeled PiC-DHFR fusion proteins as well as PiC as a control with reticulocyte lysate and subjected them to in vitro import experiments with isolated yeast mitochondria. After incubation with mitochondria, PiC became resistant against PK added outside the mitochondria (Fig. 2, PiC, section labeled PiC, lane 2). When mitoplasts were generated from the mitochondria by rupturing the outer membrane and treating with PK, PiC was clipped and a protease-resistant fragment was generated, indicating its insertion into the inner membrane (Fig. 2, section labeled PiC, lane 4). This assembly process of PiC depends strictly on the membrane potential across the inner membrane (⌬⌿), which is essential for the function of the TIM22 complex. Like PiC, PiC[TM1-TM6]-DHFR (the full-length PiC fused to DHFR) was sequestered to the protease-protected compartment after incubation with mitochondria in the presence of ⌬⌿ ( Fig. 2A, section labeled [TM1-TM6], lane 2). Because both the N terminus and the C terminus of PiC are exposed to the IMS (18), we expected that if PiC-DHFR fusion proteins were correctly sorted to the inner membrane like PiC, the DHFR part would be exposed to the IMS. This membrane topology of the DHFR domain can be assessed by testing to determine whether protease added outside mitoplasts can cleave off the DHFR part and generate a protease-resistant PiC fragment in the inner membrane. Indeed, Ͼ70% of the imported fraction of PiC[TM1-TM6]-DHFR was degraded by PK added outside the mitoplasts to generate the protease-resistant 32-kDa PiC part (Fig. 2  PiC[TM3-TM6]-DHFR (the PiC fragment lacking the first module fused to DHFR) and PiC[TM4-TM6]-DHFR (the PiC fragment starting from TM4 fused to DHFR) became proteaseresistant in mitochondria, whereas 75 and 65%, respectively, of the imported fractions were degraded in mitoplasts and did not generate protease-resistant PiC fragments (Fig. 2, A  We next examined possible insertion of the imported PiC-DHFR fusion proteins into the inner membrane by alkaline extraction. Imported PiC and PiC[TM1-TM6]-DHFR were, like the integral membrane proteins Tim50, Tim17, Tim23, and endogenous PiC, recovered in the pellet fraction after Na 2 CO 3 treatment and ultracentrifugation (Fig. 3), suggesting that they were correctly inserted into the inner membrane. S-DHFR in the assembly into the inner membrane may perhaps be due to the lack of the first and/or second modules, which might contain a correct sorting signal to the inner membrane. Because PiC[TM1-TM2]-DHFR, a fusion protein between the first module of PiC and DHFR, could not be imported into mitochondria efficiently (Fig. 2, A and B, section/bars labeled [TM1-TM2]), we cannot test to determine whether the first module actually possesses a sorting signal for the TIM22 pathway. Nevertheless, PiC[TM3-TM4]-DHFR, a fusion protein between the second module of PiC and DHFR, was slightly imported into mitochondria but did not generate a protease-protected PiC part (Fig. 2, A and B, section/bars labeled [TM3-TM4]). This result shows that at least the second module does not contain a correct sorting signal for the TIM22 pathway.
Fusion Proteins with Truncated PiC Can Use the TIM23 Complex for Their Import-By which pathway are the fusion proteins PiC[TM5-TM6]L-DHFR and PiC[TM5-TM6]S-DHFR delivered to the matrix, the TIM23 pathway or TIM22 pathway? To test the possible involvement of the TIM23 complex, we analyzed the in vitro import of PiC-DHFR fusion proteins FIG. 2. Import of PiC-DHFR fusion proteins into isolated mitochondria. A, radiolabeled PiC and PiC-DHFR fusion proteins were incubated with isolated mitochondria for 20 min at 25°C in the presence (ϩ) or absence (Ϫ) of ⌬⌿. After the import, the samples were divided into halves, and one aliquot was subjected to osmotic swelling (SW) to generate mitoplasts. The mitochondria and mitoplasts were treated with 50 g/ml PK for 30 min on ice. Black arrowheads indicate the full-length fusion proteins, and asterisks indicate those proteins translated from an ATG codon other than the first one. into mitochondria isolated from Tim50-depleted (Tim502) cells (16). Tim50 is a subunit of the TIM23 complex and is essential for protein translocation through the TIM23 channel. Mitochondria were isolated from GAL-TIM50 cells in which the TIM50 gene was under the control of the galactose-inducible GAL7 promoter after cultivation in a galactose-free medium at 23°C for 13.5 h.
Because the translocation of presequence-containing precursor proteins through the TIM23 channel requires Tim50, the in vitro import of pSu9-DHFR (a fusion protein between the presequence of subunit 9 of F 0 -ATPase and DHFR) into Tim502 mitochondria was impaired as compared with that into wildtype (WT) mitochondria (Fig. 4A). In contrast, PiC, which uses the TIM22 complex instead of the TIM23 complex for its insertion into the inner membrane, was imported into Tim502 mitochondria as efficiently as into WT mitochondria (Fig. 4B). Although PiC[TM1-TM6]-DHFR was imported into both Tim502 mitochondria and WT mitochondria with similar efficiency (Fig. 4C) (Fig. 4, D-G). The import rate of PiC[TM5-TM6]S-DHFR into Tim502 mitochondria was reduced to 45% of that into WT mitochondria.
The involvement of Tim50, i.e. the TIM23 pathway for the import of the [PiC]-DHFR fusion proteins, can also be assessed by testing the effects of anti-Tim50 antibodies on their import

FIG. 3. Alkaline extraction of imported PiC-DHFR fusion proteins.
A-C, radiolabeled mitochondrial Hsp60 precursor (Hsp60), PiC, and PiC-DHFR fusion proteins were incubated with isolated mitochondria for 20 min at 25°C and treated with 50 g/ml PK for 30 min on ice. The mitochondria were recovered after centrifugation, treated with 0.2 M Na 2 CO 3 (pH 11.2) and then the pellets (ppt) and supernatants (sup) were separated by centrifugation (100,000 ϫ g for 30 min at 4°C). Imported proteins were analyzed by SDS-PAGE and radioimaging (A), and endogenous mitochondrial proteins by SDS-PAGE and immunoblotting with their antibodies (B). The results were quantified, with the total amounts in the pellet and supernatant being set to 100% (C). Asterisks indicate those proteins translated from an ATG codon other than the first one. into mitoplasts. In mitoplasts where the outer membrane was selectively ruptured, antibodies can get access to Tim50 so that an antibody binding to Tim50 can block protein import via the  Tim23 (B), and PiC-DHFR fusion proteins (C-G) were incubated with IgG-treated mitochondria or mitoplasts for 4 min (pSu9-DHFR) or 12 min (Tim23 and PiC-DHFR fusion proteins) min at 25°C. The mitochondria and mitoplasts were treated with PK and recovered by centrifugation, and the proteins were analyzed by SDS-PAGE and radioimaging. The amount of protein imported into mitochondria without IgG was set to 100%. Treatment of mitochondria and mitoplasts with IgG prepared from the preimmune serum did not affect import of pSu9-DHFR (data not shown). Asterisks indicate those proteins translated from an ATG codon other than the first one. TIM23 complex (16). Although pSu9-DHFR was efficiently imported into mitoplasts, the import was significantly inhibited by the presence of anti-Tim50 antibodies (Fig. 5A) as shown previously (16). In contrast, the import of Tim23, a substrate for the TIM22 pathway, into mitoplasts was not blocked by anti-Tim50 antibodies (Fig. 5B). Because PiC strictly requires small Tim proteins for its insertion into the inner membrane, PiC (data not shown) and PiC[TM1-TM6]-DHFR (Fig. 5C) were not virtually imported into mitoplasts where substantial fractions of soluble IMS proteins were lost (19). Import of Fusion Proteins with Fewer PiC TM Segments Are Less Dependent on Tim10 -Because different PiC-DHFR fusion proteins apparently require small Tim proteins differently, we analyzed the role of Tim10 in the import of PiC-DHFR fusion proteins. Mitochondria were isolated from GAL-TIM10 cells in which the TIM10 gene was under the control of the galactoseinducible GAL7 promoter after cultivation in galactose-free medium at 23°C for 13 h. Although pSu9-DHFR was imported into both Tim102 mitochondria and WT mitochondria efficiently (Fig. 6A), the import of PiC and PiC[TM1-TM6]-DHFR into Tim102 mitochondria was significantly reduced to 15% of that into WT mitochondria (Figs. 6, B and C). This finding confirms that the import of PiC and PiC[TM1-TM6]-DHFR requires Tim10 in the IMS and is consistent with the fact that they are

DISCUSSION
In the present study, we revealed that PiC, which is inserted into the inner membrane via the TIM22 pathway, has a latent ability to guide a passenger protein, DHFR, to the matrix via the TIM23 pathway. When fused to DHFR, the full-length PiC behaves like an authentic PiC and uses the Tim9-Tim10 complex in the IMS and the TIM22 complex to get itself inserted into the inner membrane (Fig. 7). Import or at least translocation across the outer membrane of the truncated PiC segments fused to DHFR (PiC[TM3-TM6] (Fig. 7). It is not clear if the TIM23 complex receives these fusion proteins only from the TOM40 complex or from the Tim9-Tim10 complex as well. PiC[TM3-TM6]-DHFR and PiC[TM4-TM6]-DHFR, which passed through the TOM40 channel, remained in the IMS but did not behave like integral membrane proteins. This means that, after translocation across the outer membrane, these fusion proteins cannot be inserted into the inner membrane with the aid of the TIM22 complex or cross the inner membrane through the TIM23 channel (Fig. 7).
What controls the sorting between the TIM23 pathway and the TIM22 pathway for the PiC-DHFR fusion proteins? A trivial possibility is that the DHFR has a cryptic sorting signal to the TIM23 pathway, which is exposed by the unfolding of the DHFR part upon translocation through the TOM40 channel and that this sorting signal competes with the sorting signal for the TIM22 pathway carried by the TM segments in the PiC part. However, this possibility seems unlikely, because PiC[TM5-TM6]L-DHFR and PiC[TM5-TM6]S-DHFR, whose DHFR domains are folded on the basis of the protease digestion FIG. 7. Model of the sorting of PiC-DHFR fusion proteins between the TIM23 and TIM22 pathways. Translocation of PiC-DHFR fusion proteins through the TOM40 complex is facilitated by either the Tim9-Tim10 complex in the IMS or the TIM23 complex in the inner membrane. However, the fusion proteins with fewer TM segments are less dependent on the Tim9-Tim10 complex, and interactions with this complex are essential for sorting to the TIM22 pathway. The fusion proteins that are translocated across the outer membrane independently of the Tim9-Tim10 complex can probably be translocated through the TIM23 complex because of the latent TIM23 sorting signal in the PiC segments. 40, the TOM40 complex; 23, the TIM23 complex; 22, the TIM22 complex; 9 -10, the Tim9-Tim10 complex; OM, the outer membrane; IM, the inner membrane. experiments (data not shown), could be imported into mitoplasts (Fig. 5), whereas folded DHFR alone could not (data not shown).
Presumably, the six hydrophobic TM segments of PiC tend to aggregate and require chaperone functions of the Tim9-Tim10 complex to prevent aggregate formation (18,20). Because PiC and PiC[TM1-TM6]-DHFR have six TM segments, they are efficiently associated with the Tim9-Tim10 complex and passed on to the TIM22 complex. On the other hand, the fusion proteins with a truncated PiC segment have fewer TM segments so that the Tim9-Tim10 complex binds to them only partly. Apparently, even the fractions of these proteins bound to the Tim9-Tim10 complex are not productive transport intermediates and are not correctly transferred to the TIM22 complex. What happens to the fractions of these proteins that passed the outer membrane but are not associated with the Tim9-Tim10 complex? Although as much as 50% of PiC[TM5-TM6]L-DHFR and PiC[TM5-TM6]S-DHFR were transferred to the TIM23 complex, only 20% of PiC[TM3-TM6]-DHFR and PiC[TM4-TM6]-DHFR were received by the TIM23 complex (Fig. 5). This result is likely due to the possibility that PiC[TM3-TM6]-DHFR and PiC[TM4-TM6]-DHFR, which are not associated with the Tim9-Tim10 complex, may form an aggregate or stick nonspecifically to the membrane, thereby becoming transportincompetent in the IMS. Therefore, PiC apparently has a latent sorting signal for the TIM23 pathway that is exposed if the PiC segment is not concealed by the Tim9-Tim10 complex but still remains import-competent. A precise determination of the PiC segments recognized by the Tim9-Tim10 complex and the TIM23 complex will help us to understand further the mechanism of sorting between the TIM22 and TIM23 pathways.