Membrane-embedded C-terminal Segment of Rat Mitochondrial TOM40 Constitutes Protein-conducting Pore with Enriched β-Structure*

TOM40 is the central component of the preprotein translocase of the mitochondrial outer membrane (TOM complex). We purified recombinant rat TOM40 (rTOM40), which was refolded in Brij35 after solubilization from inclusion bodies by guanidine HCl. rTOM40 (i) consisted of a 63% β-sheet structure and (ii) bound a matrix-targeted preprotein with high affinity and partially translocated it into the rTOM40 pore. This partial translocation was inhibited by stabilization of the mature domain of the precursor. (iii) rTOM40 bound preprotein initially through ionic interactions, followed by salt-resistant non-ionic interactions, and (iv) exhibited presequence-sensitive, cation-specific channel activity in reconstituted liposomes. Based on the domain structure of rTOM40 deduced by protease treatment, we purified the elastase-resistant and membrane-embedded C-terminal segment (rTOM40(ΔN165)) as a recombinant protein with 62% β-structure that exhibited properties comparable with those of full-size rTOM40. We concluded that the membrane-embedded C-terminal half of rTOM40 constitutes the preprotein recognition domain with an enriched β-structure, which forms the preprotein conducting pore containing a salt-sensitive cis-binding site and a salt-resistant trans-binding site.

Mitochondrial precursor proteins synthesized in the cytosol are delivered to the preprotein translocase of the outer membrane (TOM 1 complex) where the precursors destined to the inner compartments are translocated across the membrane, and those destined to the outer membrane are sorted into the lipid bilayer of the membrane (1)(2)(3)(4)(5)(6). In yeast, the ϳ400-kDa TOM holocomplex is composed of the import receptors Tom70, Tom20, and Tom22, the import channel Tom40 with a predicted ␤-barrel structure, Tom 5, which regulates precursor transfer from the receptor to the channel, and Tom6 and Tom7, which regulate channel assembly (4,7). Tom22, Tom5, Tom6, and Tom7 are tightly associated with Tom40 and form the ϳ350-kDa TOM core complex (7)(8)(9). The TOM holocomplex and TOM core complex exhibit two to three ring structures with an ϳ20-Å diameter and voltage-dependent, cation-specific channel activity (3, 8 -10). The oligomeric form of Neurospora crassa Tom40 (ϳ350 kDa), purified after dissociating the TOM holocomplex with dodecylmaltoside, is mainly composed of a onering structure with a 20 -30-Å diameter (10) and exhibits channel activity comparable with that of the TOM core and holocomplexes. Upon reconstitution into liposomes, they actively import preproteins destined for the outer membrane, intermembrane space, inner membrane, and matrix (3,11). Purified recombinant yeast Tom40 reconstituted into liposomes forms a cation-selective and voltage-dependent high conductance channel with multiple conductance states, which specifically bind mitochondria-targeting sequences added to the cis-side of the membrane (12). Matrix-targeted precursors or synthetic presequence peptides added to the cis-side strongly reduced the channel open probability and increased the frequency of channel gating. In a recent report, site-specific crosslinking revealed that the Tom40 channel binds to unfolded segments of non-native proteins and prevents their aggregation. Furthermore, it has the capacity to sequester ϳ90 residues of unfolded or loosely folded preproteins (13).
Despite these advances, mechanism of preprotein recognition and sorting by the Tom40 machinery remains unclear, probably due, partly, to the difficulty in isolating Tom40 in the correctly folded and soluble form. In the present study, we expressed rat mitochondrial TOM40 (rTOM40) (14) in Escherichia coli, and we purified it from the inclusion bodies by solubilizing it in 6 M guanidine hydrochloride (GdnHCl) with subsequent refolding in the non-ionic detergent Brij35. The purified rTOM40 consisted of a ϳ63% ␤-structure and, when incorporated into liposomes, exhibited presequence-sensitive, cation-specific channel activity. A pull-down assay and surface plasmon resonance (SPR) revealed that purified rTOM40 directly bound loosely folded matrix-targeted preprotein, pSU9-DHFR, with high affinity (K D range of 10 Ϫ10 ). Salt sensitivity of the binding indicated that rTOM40 recognized preproteins by two distinct sequential interactions: initial ionic interactions followed by salt-resistant hydrophobic interactions. These sequential interactions drove partial translocation of preproteins and sequestration of the mitochondrial processing peptidase (MPP)-processing site of preproteins within the rTOM40 pore. We then narrowed the location of the ␤-structure-enriched channel domain to the membrane-embedded C-terminal half of rTOM40 (residues 166 -361). Purified recombinant rTOM40(⌬N165) was in the oligomeric form of ϳ170-kDa, as assessed by gel filtration, and exhibited both structural and preprotein-binding characteristics almost identical to those of rTOM40.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-For construction of the E. coli expression plasmid of rat TOM40, the 1086-bp cDNA encoding N-terminal His 6tagged TOM40 was cloned into the NdeI-BamHI sites of pET28a (Novagen) to obtain pET28-NHis40 (14). The expression vector of rTOM40(⌬N165) was constructed as follows. A DNA fragment encoding His 6 -TOM40(⌬N165) was prepared by PCR using pET28a-NHis40 as the template, and the following oligonucleotides as the primer: sense strand, 5Ј-GGGAATTCCATATGCACCAGCTGAGCCCAG GC-3Ј, and antisense strand, 5Ј-GCGGGATCCTCAGCCGATGGTGAG GCC AAA-3Ј. The NdeI and BamHI sites are underlined. The fragment was inserted into the NdeI-BamHI sites of pET28a to create pET28a-⌬N165. The expression vector for the N-terminal His 6 -tagged TOM40-(1-165) was constructed as follows. A DNA coding for N-terminal His 6tagged TOM40-(1-165) was amplified by PCR using pET28-NHis40 as the template and the following oligonucleotides as the primers: sense strand, 5Ј-GGGAATTCCATATGGGGAACGTGTTGGCTGCT-3Ј, and antisense strand, 5Ј-GCGGGATCCTCAGATGACCTGTGCATTGAG-3Ј. The NdeI and BamHI sites are underlined. The amplified fragment was inserted between the NdeI-BamHI sites of pET28a to create pET28a- .
Preparation of Elastase Fragments of rTOM40 for the N-terminal Sequencing-The reaction mixture (50 l) containing 3 g of rTOM40 and elastase (2 g/ml) was incubated at 0°C for 30 min. The protein fragments were recovered by trichloroacetic acid precipitation and separated by SDS-PAGE. The fragments were transferred to polyvinylidene difluoride membrane followed by Ponceau S-staining. The stained bands were cut out and subjected to protein sequencing.
Preparation of Proteoliposomes-Preparation of proteoliposomes con-taining rTOM40 was essentially according to the method described by Jackson and Litman (15). Briefly, 1.2 mg of L-␣-phosphatidylcholine and 0.8 mg of L-␣-phosphatidylethanolamine in chloroform were mixed in a test tube, and the solvent was evaporated by flushing with N 2 . One milliliter of 50 mM Tris-HCl buffer (pH 7.5) containing 0.5% Brij35 and 20 g of rTOM40 was added to the tube, vortexed, and incubated on ice for 4 h with intermittent mixing. After incubation, the reaction mixture was diluted in 50 ml of 50 mM Tris-HCl buffer (pH 7.5) containing 50 mM sodium acetate. The reaction mixture was then dialyzed against 5 liters of 10 mM Tris-HCl buffer (pH 7.5) containing 0.15 M NaCl and 20 g of Bio-Beads SM-2 (Bio-Rad). The dialysate was concentrated using a membrane filter. Approximately 80% of the obtained proteoliposomes assumed unilamellar vesicles with ϳ0.1 m diameter. Preparation of Mitochondria from HeLa Cells Expressing the Epitope-tagged rTOM40 -HeLa cells were transfected with N-terminal His 6 -tagged rTOM40 or C-terminal HA-tagged rTOM40 using FuGENE 6 (Roche Applied Science) and cultured for 24 h. The cells were collected and homogenized in the homogenization buffer (10 mM Hepes-KOH buffer (pH 7.4), containing 0.22 M mannitol, 0.07 M sucrose, and 1 mM PMSF) by passing them through a 27-gauge needle 20 times using a syringe. The homogenate was centrifuged at 600 ϫ g for 5 min, and the supernatant was then centrifuged at 6000 ϫ g for 10 min to obtain the mitochondrial fraction.
Measurement of Channel Activity-Electrical measurements were performed with nystatin-perforated patch recordings applied on lipid bilayer vesicles containing rTOM40. SCC1-19 (10 M) dissolved in the perfusate was applied using the Y-tube method, which allowed the external solution to be exchanged within 20 ms (16). The resistance between the patch pipette and the reference electrode in the external solution was 10 -12 megohms. Ionic currents were measured with a patch clamp amplifier (EPC-7, List-Medical, Germany) and low pass filtered at 1 kHz (E-3210A, NF Electronic Instruments, Japan). All experiments were performed at room temperature. The composition of the pipette solution was 100 mM KCl, 50 mM potassium methane sulfonate, and 10 mM Hepes (pH 7.2). The external solution contained 150 mM KCl, 10 mM Hepes (pH 7.2), and 1 mM CaCl 2 . In some experiments, the reversal potentials were obtained as the membrane potential at which the current responses to ramp voltage steps from Ϫ100 mV to ϩ100 mV with and without SCC1-19 intersected with each other. The presence of the TOM40 pore was verified by measuring release of [ 14 C]sucrose from the proteoliposomes containing rTOM40 or rTOM40(⌬N165). The assay was carried out essentially as described by Zalman et al. (17) using [ 14 C]sucrose and [ 3 H]dextran (mean M r , 70,000) as the permeable and impermeable substrates, respectively, except that rTOM40, rTOM40(⌬N165), rTOM40-(1-165), or lactate dehydrogenase, in lieu of mitochondrial outer membrane, was reconstituted into asolectin liposomes. The reaction mixture was passed through a Sepharose CL-4B column (0.9 ϫ 9.5 cm) equilibrated with 10 mM Hepes-KOH buffer (pH 7.4) containing 100 mM NaCl, 0.1 mM MgCl 2 , and 3 mM NaN 3 . 0.3-ml fractions were collected and assayed for radioactivity by using a liquid scintillation counter. Membrane vesicles were eluted in fractions 6 -7 (Fig. 7B).
Isolation of Proteoliposomes by Centrifugal Floatation-After reconstitution of proteoliposomes, the reaction mixture was adjusted to 1.6 M sucrose, placed under the layers of 1.25 M and 0.25 M sucrose, and centrifuged in a Hitachi RP-S120AT3 rotor at 100,000 rpm for 90 min. Proteoliposomes floated to the 0.25 and 1.25 M sucrose layers were recovered and analyzed by SDS-PAGE.
Binding Assay of Preproteins to rTOM40, rTOM40(⌬N165), or rTOM40-(1-165) by Co-precipitation-The mixture containing 1 g of pAd and the indicated amounts of rTOM40, rTOM40(⌬N165), or rTOM40-(1-165) in 200 l of the binding buffer (20 mM Hepes-KOH buffer (pH 7.4) containing 0.1% Brij35, and 50 mM NaCl) was incubated at 30°C for 30 min. The reaction mixture was centrifuged at 45,000 rpm for 5 min. The obtained supernatant was mixed with 30 l (50% slurry) of TALON metal affinity resin (Clontech) and incubated at 4°C for 1 h. The beads were washed with the binding buffer, and the bound proteins were analyzed by SDS-PAGE and subsequent immunoblotting using anti-adrenodoxin IgG.
Protection of the MPP-processing Site of pAd by rTOM40 or rTOM40(⌬N165)-Binding of pAd to rTOM40, rTOM40(⌬N165), or rTOM40-(1-165) and isolation of the complex by TALON beads were performed as described above. The isolated beads were suspended in 25 l of the binding buffer, and the suspension was then incubated with 3 g of yeast recombinant MPP in the presence of 2 mM MnCl2 at 30°C for 30 min. The reaction was terminated with the sample loading buffer and analyzed by SDS-PAGE and subsequent immunoblotting using anti-adrenodoxin IgG.
CD Spectrum Measurement-CD spectra of rTOM40, rTOM40(⌬N165), and rTOM40-(1-165) were measured in 10 mM Tris-HCl buffer (pH 7.4) containing 0.5% Brij35 at 25°C using a JASCO J-720 spectropolarimeter and a cuvette with a 1-mm light-path. Each spectrum represents an average of five scans from 195 to 250 nm at 0.1-nm intervals. The base line was established by subtracting the spectrum of the buffer alone. Analysis of the secondary structure was performed using the method of Reed and Reed (18).
Surface Plasmon Resonance Measurements-The SPR measurements were performed at 25°C with a Biacore 3000 (Biacore AB). Purified rTOM40, rTOM40(⌬N165), or rTOM40-(1-165) was immobilized onto the sensor chip CM5 by amine-coupling according to the manufacturer's protocol. Briefly, the coupling was performed in 10 mM sodium acetate buffer (pH 6.5) at a protein concentration of 10 g/ml. The level of immobilization typically corresponded to 2000 resonance units, which corresponded to ϳ2 ng of protein/mm 2 (Fig. 6, A-D). In Fig.  6E, rTOM40 was immobilized to the chip at 22,000 resonance units (corresponding to ϳ22 ng of protein/mm 2 ). Binding analyses were performed in 20 mM Hepes-KOH buffer (pH 7.4) containing 150 mM NaCl and 0.05% Brij35 (running buffer) at a flow rate of 20 l/min. The sensor chip surface was regenerated by 50 mM HCl. Binding curves were analyzed using BIA-Evaluation software (version 3.2). The kinetic data fitting was performed using a Langmuir 1:1 binding model.

RESULTS
Purification of Rat TOM40 -N-terminal His 6 -tagged rTOM40 expressed in E. coli as inclusion bodies was solubilized by 6 M GdnHCl, applied to a Ni-NTA affinity column, and subjected to a refolding reaction by exchanging GdnHCl slowly with Brij35. rTOM40 was then eluted by imidazole and subjected to Mono-S column chromatography. Purified rTOM40 ( Fig. 1A) was eluted through a Superose 6 column with a peak at ϳ250 kDa, although with a rather broad elution profile (see Fig. 5B).
Secondary Structure of rTOM40 -A CD spectrum of rTOM40 in 0.5% Brij35 had a minimum value at 213 nm, crossover of the base line at 201 nm, and zero ellipticity at a wavelength 235 nm (Fig. 1B). The secondary structure of rTOM40 estimated from the CD spectrum using the program of Reed and Reed (18) comprised 62.9% ␤-sheet, 10.0% ␣-helix, 5.9% turn, and 21.1% random structures in the protein. The content of the ␤-sheet structure of rTOM40 was comparable with that of recombinant Saccharomyces cerevisiae Tom40 (Ͼ60%) purified from inclusion bodies after solubilization with 8 M urea and either reconstituted into liposomes or solubilized in Mega9, although the CD spectra were distinct (12). The secondary structure of rTOM40 was significantly different from that of the oligomeric form of N. crassa Tom40, which was purified after dissociation of the purified TOM complex with octyl glucoside (less ␤-sheet, ϳ31%; more ␣-helix, 30%) (10). The reason for this structural difference between rat and N. crassa Tom40 is not known.
Channel Properties of rTOM40 -The purified TOM holocomplex, TOM core complex, oligomeric N. crassa Tom40 isolated from the purified TOM core complex, and recombinant Tom40 of S. cerevisiae form cation-selective high conductance channels when incorporated into lipid membranes and the presequence peptide block the channel in a voltage-dependent manner (3,8,10,12,19). We therefore examined whether purified rTOM40 was correctly folded to exhibit channel activity. rTOM40 was incorporated into unilamellar ϳ0.1-m-diameter liposomes, and electrical measurements were performed with nystatinperforated patch recordings. A functional presequence peptide SCC1-19 (10 M) (20) induced an immediate outward shift of current at a holding potential of Ϫ100 mV in less than 1 s (Fig.  2). During SCC1-19 application, there was no desensitization of the current but current noise increased. The current returned to base line within 10 s upon washing out the presequence peptide. Less hyperpolarization of the membrane reduced SCC1-19-induced current amplitude and a positive membrane potential reversed the direction of SCC1-19-induced current inward. The reversal potential of SCC1-19-induced current was 8.3 Ϯ 3.2 mV (mean Ϯ S.E., n ϭ 5). This result suggested that the lipid bilayer with rTOM40 contains cation-permeable ion channels that are rapidly and reversibly blocked by the presequence peptide.
Binding of Matrix-targeted Preproteins to Purified rTOM40 -Confirming that recombinant rTOM40 had refolded to constitute a dominant ␤-sheet structure and to exhibit channel activity, we examined its interaction with matrix-targeted preproteins. Recombinant preadrenodoxin (pAd) (21) was incubated with rTOM40 (N-terminal His 6 -tagged), and rTOM40 was recovered with nickel resin. Immunoblot analysis with anti-adrenodoxin antibody revealed that pAd was recovered to the nickel resin depending on the amount of rTOM40 added to the reaction mixture (Fig. 3A). As a control, the mature form of adrenodoxin (mAd) did not bind to rTOM40 (data not shown). The interaction between rTOM40 and pAd was sensitive to NaCl, and the interaction was almost completely abolished by 500 mM NaCl (Fig. 3B). The pAd⅐rTOM40 complex, once formed, was stable in high salt (Fig. 3B, Af in the lower panel). These results suggested that the preprotein initially binds to rTOM40 mainly through ionic interactions, which is followed by the other interactions involving hydrophobic interactions. The so-called "cis-binding sites" or "cis-sites" of mitochondria or mitochondrial outer membranes, which are located in the protease-sensitive surface receptors Tom20 and Tom22, are sensitive to salt concentrations as low as 100 mM (22)(23)(24)(25)(26). Therefore, the "cis"-site of TOM40 involved in the initial precursor recognition binds preproteins through stronger ionic interactions than that for the cis-binding site of the mitochondrial outer membrane. Stan et al. (27) demonstrated that the isolated N. crassa TOM holocomplex and the proteinase K-treated core not only bind pSU9-DHFR but protect the precursor from cleavage by MPP, indicating partial translocation of the precursor protein into the TOM complex and that the MPP cleavage site is protected by the TOM complex against MPP. Because this MPP protection is a suitable criterion to assess Tom40 function, we examined MPP protection with rTOM40. Recombinant pAd was incubated with rTOM40 and then the pAd⅐rTOM40 complex was isolated using nickel resin, which was then subjected to MPP digestion. As shown in Fig. 3C, the pAd recovered as the complex with rTOM40 was protected against MPP, whereas unbound pAd was efficiently processed. This protection occurred as a function of the amount of rTOM40 (Fig. 3, D  and E), and the reaction was essentially saturated by 4 -8 g of rTOM40 per 1 g of pAd (roughly calculated, ϳ2-4 mol of rTOM40/mol of pAd, assuming the molecular size of rTOM40 and pAd to be 38 and 20 kDa, respectively). The rTOM40-dependent MPP protection was also observed for pSU9-DHFR (data not shown). Therefore, recombinant rTOM40 had properties similar to those of the isolated N. crassa TOM holocomplex or the proteinase K-treated core, which is composed solely of the oligomeric form of Tom40 (27). The TOM core complex was unable to partially translocate the preprotein unless phospholipids from the mitochondrial outer membrane were supplied externally (27). Because phospholipid P i was not detected in our rTOM40 preparation (data not shown), rTOM40 seemed to have folded correctly in Brij35 during the FIG. 3. Recognition of preprotein by purified rTOM40. A, binding of pAd to rTOM40 as measured by pull-down assay. pAd (1 g) was incubated with the indicated amounts of rTOM40 (N-terminal His 6 -tagged), and then the reaction mixtures were subjected to pull-down reaction by Ni-NTA beads. The recovered pAd⅐rTOM40 complex was subjected to SDS-PAGE followed by immunoblotting using anti-adrenodoxin antibodies and subsequent image analysis by a LAS1000 plus (Fuji Film Co.). The band intensities were calculated by setting the total pAd signal to 100% (shown in the right panel). B, salt sensitivity of the pAd⅐rTOM40 complex. The binding assay was performed as in A using 5 g of rTOM40 in the presence of the indicated concentrations of NaCl. In the lower panel, the pAd⅐rTOM40 complex formed after 30 min of incubation was incubated with 500 mM NaCl (Af) and then subjected to the pull-down assay as in A. In a separate experiment, pAd and rTOM40 were incubated in the presence of 500 mM NaCl (Bf) followed by the pull-down assay. The quantified results are shown in the right panel. C, sequestration of the MPP-processing site of pAd within the rTOM40 molecule. pAd (1.4 g) and rTOM40 (3.8 g) were incubated in 50 l at 30°C for 30 min. The reaction mixtures were then incubated with Ni-NTA beads to separate into pAd (unbound) and the pAd⅐rTOM40 complex (bound). Both fractions were then incubated with (ϩ) or without (Ϫ) MPP at 30°C for 30 min, and the reaction mixtures were analyzed by SDS-PAGE and subsequent immunoblotting with anti-adrenodoxin antibodies. D, dosedependent sequestration of the MPPprocessing site of pAd by rTOM40. pAd (1 g) and the indicated amounts of rTOM40 were incubated at 30°C for 30 min. The reaction mixtures were then incubated with MPP at 30°C for 30 min, which were subjected to SDS-PAGE and subsequent image analysis. The protection efficiency (%) was calculated as the ratio of pAd to pAd plus mAd, and shown in E.
purification process to acquire the activity of partial translocation of preproteins even in the absence of phospholipids.
Domain Structure of rTOM40 and Its Membrane Topology-We then probed the domain structure of purified rTOM40 using protease digestion. Elastase (2 g/ml) treatment of rTOM40 at 0°C for 30 min produced at least three distinct fragments (Fig.  4A). Fragments 1 and 2 were formed transiently, and fragment 3 was formed stably. These fragments were detected even after 50 g/ml elastase digestion (see Fig. 4B). N-terminal amino acid sequencing revealed that fragments 1-3 had lost residues 1-7, 1-65, and 1-165, respectively. The antibodies against rTOM40 (␣-TOM40) recognized only fragment 1, whereas two antibodies raised against the synthetic peptides corresponding to the regions near the C terminus (residues 323-345: peptide FIG. 4. Domain structure and membrane topology of rTOM40. A, elastase (Ela) digestion of rTOM40. rTOM40 (3 g) was digested with 2 g/ml elastase in 50 l at 0°C for 30 min. The reaction mixture was trichloroacetic acid-precipitated and analyzed by SDS-PAGE and subsequent Coomassie Brilliant Blue staining. B, recognition of the elastaseproduced fragments of rTOM40 by various antibodies. rTOM40 (3 g) was digested in 50 l with the indicated concentrations of elastase at 0°C for 30 min. The reaction mixtures were trichloroacetic acid-precipitated, and the precipitates were solubilized in the loading buffer, divided into 4 aliquots, and analyzed by SDS-PAGE and subsequent immunoblotting using the indicated antibodies. C, elastase susceptibility of endogenous rTOM40 in the mitochondrial outer membrane. Mitochondria (50 g/100 l) were digested with the indicated concentrations of elastase at 0°C for 30 min under isotonic (Ϫ) or hypotonic (ϩ) conditions. The reaction mixtures were trichloroacetic acid-precipitated and subjected to SDS-PAGE and subsequent immunoblotting using the indicated antibodies. Asterisk, nonspecific band. D, membrane topology of the N-and C-terminal ends of rTOM40 in the outer membrane as probed by elastase and proteinase K (Pro.K). Mitochondria (10 g/100 l) harboring either N-terminal His 6tagged rTOM40 (His 6 -TOM40) or C-terminal HA-tagged rTOM40 (TOM40-HA) were treated with or without 100 g/ml elastase or 100 g/ml proteinase K at 0°C for 30 min under the indicated conditions. The trichloroacetic acid-precipitates were resolved by SDS-PAGE and analyzed by immunoblotting using the antibodies against His 6 or HA. TX-100, Triton X-100. E, topology of the C-terminal end of rTOM40 in the mitochondrial outer membrane as probed by carboxypeptidase Y (CPY). Mitochondria (50 g/100 l) were treated with 5 g/ml carboxypeptidase Y at 30°C for 30 min under the indicated conditions. The reaction mixtures were analyzed by SDS-PAGE and subsequent immunoblotting using the indicated antibodies. F, schematic representation of the topology of rTOM40 in the mitochondrial outer membrane. The ␤-structure-enriched C-terminal half is shown as the cylindrical structure. The sites Ala 7 -Ser 8 and Ala 65 -Ala 66 are accessible to elastase from outside mitochondria and from the intermembrane space, respectively, whereas Ile 165 -His 166 , which is accessible to elastase in the purified rTOM40, is resistant in the outer membrane to elastase treatment from either side of the membrane. The segment 1-165 should span the membrane at least once, although the detailed membrane disposition of this segment is not known. OM, outer membrane; IMS, intermembrane space.

FIG. 5. Structural characteristics of purified rTOM40(⌬N165) and its properties of preprotein recognition.
A, SDS-PAGE profile of purified rTOM40(⌬N165). B, elution profiles of rTOM40 and rTOM40(⌬N165) through Superose 6 column equilibrated with 20 mM Tris-HCl (pH 7.5) containing 0.5% Brij35 and 150 mM NaCl. C, CD spectrum of rTOM40(⌬N165). D, binding of pAd by rTOM40(⌬N165) as assessed by pull-down assay. pAd (1 g) was incubated with the indicated amounts of rTOM40(⌬N165) (N-terminal His 6 -tagged), and then the reaction mixtures were subjected to pull-down reaction. Other conditions were as described in the legend to Fig. 3. E, salt sensitivity of the pAd⅐rTOM40(⌬N165) complex. The binding assay was performed as in D using 5 g of rTOM40(⌬N165) in the presence of the indicated concentrations of NaCl. F, sequestration of the MPP-processing site of pAd within rTOM40(⌬N165) molecule. pAd and rTOM40(⌬N165) were incubated at 30°C for 30 min. The reaction mixtures were then incubated with Ni-NTA beads. The pAd⅐rTOM40(⌬N165) complex was incubated with (ϩ) or without (Ϫ) MPP at 30°C for 30 min, and the reaction mixtures were analyzed by SDS-PAGE and subsequent immunoblotting with anti-adrenodoxin antibodies. In a separate experiment (⌬N165 ϭ 0 g), pAd was incubated with MPP at 30°C for 30 min, and the reaction mixture was analyzed by SDS-PAGE and subsequent immunoblotting. The band intensities were quantified, and the processing efficiency (mAd/(pAd ϩ mAd)) was calculated by setting the efficiency in the absence of rTOM40(⌬N165) to 100% (shown in the right panel). G, rTOM40-(1-165) binds pAd but does not sequester the MPP-processing site within the molecule. The indicated amounts of N-terminal His 6 -tagged rTOM40-(1-165) were incubated in 200 l with 1 g of pAd, and the reaction mixtures were subjected to the MPP protection assay as described in F. 40 -1, and 189 -207: peptide 40 -2) recognized all three fragments (Fig. 4B). On the other hand, anti-His 6 antibodies only recognized the full-size rTOM40 (Fig. 4B). Thus, rTOM40 an-tibodies recognized the epitopes located within the N-terminal 65 residues of rTOM40. These results indicated that the Cterminal half of rTOM40 (residues 166 -361; 21.4 kDa) folded   FIG. 6. Interaction of the presequence peptide or pSU9-DHFR with rTOM40 as assessed by SPR. A, SCC1-19; B, SCC1-19M; or C, pSU9-DHFR in running buffer containing 0.05% Brij35 was injected to rTOM40-immobilized sensor chip at 25°C. D, pSU9-DHFR pretreated with or without methotrexate (Mtx) at 0°C for 30 min was injected into the sensor chip. Other conditions were described under "Experimental Procedures." E, interaction of rTOM40 and pSU9-DHFR was performed in running buffer containing the indicated concentrations of NaCl.
to form a stable domain structure.
We then addressed the topology of rTOM40 in the mitochondrial outer membrane using antibodies against rTOM40 (␣-TOM40) and against a synthetic peptide corresponding to residues 189 -207 (␣-peptide 40-2). Elastase treatment of rat liver mitochondria under isotonic conditions produced a fragment, which was recognized both by ␣-TOM40 and ␣-peptide 40-2 (Fig. 4C, left and middle panels). Under hypotonic conditions, elastase produced a fragment, which was detected only by ␣-peptide 40-2 (Fig, 4C, middle panel). From the size and reactivity with the antibodies, the bands produced under isotonic conditions and hypotonic conditions were considered to correspond to fragment 1 and fragment 2, respectively. These results indicated that the N-terminal site (Ala 7 -Ser 8 ) of rTOM40 is exposed to the outer surface of the mitochondria, whereas the site Ala 65 -Ala 66 is localized in the intermembrane space. The Ile 165 -His 166 site, which is accessible to elastase in purified rTOM40 to produce fragment 3, was masked by the membrane or by the components of the TOM complex. The behavior of mitochondrial markers, mHsp70 (matrix protein), TIM23 (inner membrane protein extruding the N-terminal segment out of the inner membrane), and TOM20 (outer membrane protein extruding the bulk C-terminal portion to the cytosol), indicated that the protease digestion reactions were well controlled (Fig. 4C, right panel). Of note, fragments 1 and 2 were resistant to sodium carbonate (pH 11.5) extraction, indicating that they were firmly embedded in the membrane (data not shown). Topology of the N-terminal segment was further confirmed by using mitochondria isolated from HeLa cells expressing N-terminal His 6 -tagged rTOM40. As shown in Fig. 4D, the His 6 epitope tag was removed by elastase treatment under isotonic conditions, indicating that the N-terminal segment of rTOM40 is exposed to the cytosol. We then probed the orientation of the C-terminal segment by using mitochondria harboring C-terminal hemagglutinin (HA)-tagged rTOM40. When the mitochondria were treated with proteinase K under isotonic conditions, the HA tag was unaffected, whereas it was completely removed from rTOM40 when the outer membrane was ruptured by hypotonic treatment, thus indicating that the C-terminal segment is exposed to the intermembrane space (Fig. 4D, right panel). This was further confirmed by carboxypeptidase Y treatment. rTOM40 in the isolated mitochondria was resistant to carboxypeptidase Y treatment under isotonic conditions, whereas it was completely digested under hypotonic conditions or in the presence of Triton X-100 (Fig. 4E). As a control, rTOM22 that is inserted into the outer membrane in the N out -C in orientation exhibited the same susceptibility to the carboxypeptidase Y treatment (Fig. 4E). Taken together, rTOM40 is embedded in the outer membrane exposing its N-terminal segment to the cytosol and the Cterminal segment to the intermembrane space, whereas at least the site Ala 65 -Ala 66 is exposed to the intermembrane space (Fig. 4F). Whether segment 1-165 is embedded in the membrane by a single or multispanning configuration remains to be determined. The predicted overall topology is distinct from that of N. crassa Tom40; the N-and C-terminal ends are exposed to the intermembrane space (28,29). The N-terminal segment of S. cerevisiae Tom40 is exposed to the cytosol (12), but the topology of the C-terminal segment is not known. rTOM40(⌬N165) Has a Secondary Structure and Preproteinbinding Properties Comparable with rTOM40 -Based on the above findings, we purified a recombinant protein rTOM40(⌬N165) (21.4 kDa) in which the N-terminal 165-residue segment of rTOM40 was deleted, essentially according to the procedure adopted for rTOM40 (Fig. 5A). On a Superose 6 gel filtration column, it was eluted at an apparent molecular size of ϳ170 kDa with a sharp elution peak compared with rTOM40 (Fig. 5B). The secondary structure calculated from the CD spectrum in 0.5% Brij35 (Fig. 5C) revealed 62.0% ␤-sheet, 1.8% ␣-helix, 1.8% turn, and 34.5% random structures. rTOM40(⌬N165) had preprotein-binding properties comparable with those of rTOM40 as follows: (i) dose-dependent pAd binding (Fig. 5D); (ii) salt-sensitive initial binding of pAd, followed by salt-resistant binding (Fig. 5E); and (iii) sequestration of the MPP processing site within the rTOM40(⌬N165) pore (Fig. 5F). It should be noted that the purified recombinant form of segment 1-165 of rTOM40 (rTOM40-(1-165)) bound pSU9-DHFR with a K D of 1.4 ϫ 10 Ϫ10 M as assessed by SPR measurements (data not shown), but it failed to protect the precursor from attack by MPP (Fig. 5G), suggesting that the protection against MPP was because of specific interactions with rTOM40(⌬N165). Taken together, rTOM40(⌬N165) exhibited preprotein recognition properties as the import pore comparable with those of rTOM40. In support of these findings, rTOM40 and rTOM40(⌬N165) reconstituted into proteoliposomes exhibited permease activity for the vesicle-entrapped sucrose (see below).
Binding Kinetics of Preproteins to rTOM40 and rTOM40-(⌬N165) as Analyzed by SPR-We next measured the kinetics of interaction of rTOM40 or rTOM40(⌬N165) with either the synthetic presequence SCC1-19 or recombinant pSU9-DHFR by using SPR. rTOM40 or rTOM40(⌬N165) was immobilized to the sensor tips, and various concentrations of presequence or pSU9-DHFR were injected. The binding curves obtained (Fig. 6 for rTOM40; data not shown for rTOM40(⌬N165)) were analyzed by using BIA Evaluation software. Calculated association (k a ) and dissociation (k d ) rate constants and K D (k d /k a ) values are summarized in Table I. The affinity of rTOM40 for presequence peptide SCC1-19 was 3.0 ϫ 10 Ϫ6 M (Table I). No binding was observed with nonfunctional control peptides SCC1-19M (20) (Fig. 6B) and Synb2 (30) (data not shown). In contrast, however, rTOM40 exhibited ϳ10 4 -fold higher affinity (1.2 ϫ 10 Ϫ10 M) for pSU9-DHFR compared with the synthetic presequence, suggesting that the mature segment of the precursor was responsible for the high affinity binding (Fig. 6C and Table I). When the conformation of the DHFR segment was stabilized with methotrexate, binding of pSU9-DHFR was strongly inhibited (Fig. 6D). Taken together, these results suggested that the affinity of rTOM40 for the presequence per se was rather low, and the affinity was greatly increased by the presence of the unfolded mature region of the preprotein.
We then examined the salt sensitivity of the interaction between rTOM40 and pSU9-DHFR. The interaction was saltsensitive, and the binding was almost completely inhibited by 0.5 M NaCl (Fig. 6E), confirming the results of the pull-down assays (see Fig. 3). These results indicated that rTOM40 initially binds preproteins mainly through ionic interactions, which is followed by some other interactions including hydrophobic forces; the unfolded mature segment of preprotein seemed to contribute to the latter interactions (13). rTOM40(⌬N165) exhibited similar but slightly lower affinity for SCC1-19 and pSU9-DHFR compared with full size rTOM40 (Table I and data not shown). The N-terminal 165 segment might contribute to stabilize the correct conformation of the pore-forming segment. rTOM40 and rTOM40(⌬N165) Exhibit Sucrose Passage Activity When Reconstituted into Liposomes-Because our attempts to measure the channel activity for rTOM40(⌬N165) by using electrical methods were unsuccessful for technical reasons, we tried to measure the pore activity biochemically. Because the TOM complex, when reconstituted into proteoliposomes, mediates passage of small molecules (11, 31), we examined whether rTOM40(⌬N165) has sucrose passage activity, using the method adopted for measuring the activity of mitochondrial porin (17,32). This assay measures the retention of large [ 3 H]dextran (mean M r 70,000) versus small [ 14 C]sucrose that had been trapped into proteoliposomes containing rTOM40 proteins by sieving through a Sepharose 4B column. These experiments revealed that both rTOM40 and rTOM40(⌬N165) mediated passage of sucrose to a significant extent ( Fig. 7 and Table II). As the controls, heat-denatured rTOM40(⌬N165) and cytoplasmic enzyme lactate dehydrogenase were inactive in this assay. We thus concluded that the C-terminal half-segment of rTOM40, like full-size rTOM40, mediates passage of small molecules across the membranes. These properties seem to correspond to the ability of rTOM40 and rTOM40(⌬N165) for sequestration of the MPP-processing site of preprotein within the molecule (see Fig. 3 and Fig. 5).

DISCUSSION
Virtually all the nuclear coded mitochondrial proteins are translocated and sorted into mitochondrial subcompartments via the TOM complex; preproteins transported to the inner compartments are translocated through the TOM channel irrespective of whether they are destined to soluble compartments or to the inner membrane. On the other hand, the outer membrane proteins are sorted by the TOM complex from the proteins destined for the inner compartments and anchored to the lipid bilayer of the outer membrane. As an initial step for understanding the mechanism of this diverse preprotein recognition by the TOM channel, we purified active recombinant rTOM40, and we analyzed the recognition properties using matrix-targeted preproteins.
Purified rTOM40 bound preproteins with high affinity and sequestered the MPP-processing site within the molecule. Furthermore, when reconstituted into liposomes, it exhibited presequence-sensitive cation-selective channel activity. Therefore, recombinant rTOM40 was correctly refolded to attain the functional conformation as the preprotein translocation pore. The CD spectrum of rTOM40 did not exhibit the light-scattering effects caused by aggregated species and revealed a greater than 60% ␤-sheet structure. This value coincided well with that for recombinant S. cerevisiae Tom40 (12), although the CD spectra differed considerably. In contrast, the ␤-sheet structure content of N. crassa Tom40 predicted by CD spectra or IR spectra was markedly lower with a maximum of 31% (10). The ␣-helical structure of rTOM40 (10%) was half that of N. crassa Tom40. The reason for the difference in the secondary structure between N. crassa and mammals is not known.
Most importantly, this study demonstrates that the purified membrane embedded C-terminal, half-formed ϳ170-kDa homo-oligomeric complex with a greater than 60% ␤-sheet structure and exhibited preprotein-binding properties comparable with those of rTOM40, suggesting that the C-terminal segment constitutes the preprotein conducting pore. Alignment of Tom40 proteins from several organisms revealed that the sequence conservation is higher in the C-terminal pore-forming segment compared with the N-terminal segment (14). Although attempts to measure the presequence-responsive channel activity of rTOM40(⌬N165) electrically were unsuccessful, we could demonstrate that it mediated passage of sucrose across the membrane. This preparation will help analyze the struc-ture of the pore and preprotein recognition mechanisms. rTOM40(⌬N165) was almost functionally identical with rTOM40 with respect to the preprotein recognition, and both exhibited enriched ␤-sheet structures, thus the ␤-barrel structure is responsible for the pore function as is the case for porin (28,33). The ␤-structure content of rTOM40(⌬N165) was lower (62%) than that of rTOM40. Because the random coil structure was increased in rTOM40(⌬N165), proper refolding might be disturbed to some extent. The N-terminal 1-165 segment might be required for correct formation or stabilization of the pore structure, and this might be reflected in the decreased affinity of rTOM40(⌬N165) for pSU9-DHFR.
Here we demonstrated that purified recombinant rTOM40 and rTOM40(⌬N165) exhibited virtually identical properties with the TOM core complex (24,27). They initially bind the preprotein through predominantly electrostatic interactions and partially translocate the preprotein to the salt-resistant trans-site that is inaccessible to MPP, probably within the translocation channel. Stabilization of the DHFR moiety by methotrexate inhibited binding of pSU9-DHFR to rTOM40 or rTOM40(⌬N165), suggesting that the partial translocation is accompanied by unfolding of the mature segment, and the activity is restricted to the C-terminal half of rTOM40. These results also indicate that purified rTOM40 as well as rTOM40(⌬N165) contain the salt-sensitive cis-binding site. The salt-sensitive binding to the cis-site provided by the surface receptors Tom20 and Tom22 is much weaker than that in rTOM40 or rTOM40(⌬N165); cis-site binding of the preprotein was almost completely inhibited by 100 mM KCl (22)(23)(24). Consistent with this, the K D values of preproteins for the cytoplasmic domain of import receptors Tom70 or Tom20 as measured by SPR were 10 Ϫ7 -10 Ϫ8 M (34). This affinity difference might facilitate vectorial preprotein transfer from the surface import receptors to the cis-binding site of Tom40.
Analysis by SPR revealed that rTOM40 bound pSU9-DHFR with high affinity (in the 10 Ϫ10 M range), and stabilization of the DHFR moiety greatly decreased the affinity. Most interestingly, rTOM40 bound a presequence peptide but with 10 4 -fold lower affinity at 3.0 ϫ 10 Ϫ6 M. These results indicate that the mature portion of the preprotein contributes significantly to the high affinity binding. It should be noted that the rTOM40⅐pSU9-DHFR complex or rTOM40(⌬N165)⅐pSU9-DHFR complex, once formed, was resistant to salt treatment, indicating a mode of interaction different from the initial interactions in the latter binding stage or in the trans-site binding in the purified molecules. The precise nature of the interaction of the preprotein with the trans-binding site remains to be determined. rTOM40 and rTOM40(⌬N165) thus possess virtually all the preprotein-binding properties characteristic of the TOM holocomplex. What might be the function of the N-terminal 165-residue segment? rTOM40-(1-165) was expressed in E. coli as a soluble form. CD spectra of the purified recombinant rTOM40-(1-165) revealed that it has 49% ␣-helix, 6% ␤-sheet, and 45% random structures. The segment consisting of residues 1-65 should span the membrane at least once, although the exact states of membrane disposition of the segment including this and up to 165 residues remains unknown. Recombinant rTOM40-(1-165) bound preprotein with an affinity on the order of 10 Ϫ10 M mainly through hydrophobic interactions; the complex was stable in the presence of 500 mM NaCl. 2 Considering that purified rTOM40 initially binds preproteins by ionic interactions, these results suggest that the 1-165 segment functions in the later stages of preprotein translocation. In N. crassa Tom40, segment 41-60 (corresponds to residues 80 -98 of rTOM40) is essential for proper assembly/stability of Tom40 in the TOM complex (35). In a recent report, residues 51-60 (correspond to residues 90 -98 of rTOM40) and the C-terminal 3 residues (residues 321-323 which correspond to 353-355 of rTOM40) are required for assembly beyond the 250-kDa assembly intermediate of the TOM complex (36). Thus, the N-terminal 1-165 segment might also be involved in the assembly with the TOM components such as Tom22 and small Tom proteins or function as the interface of releasing outer membrane proteins from the import pore into the lipid bilayer. Another possibility is that the N-terminal 1-165 segment is required for coupling the TOM complex with the translocation of inner membrane complex during preprotein transit from the outer membrane to the inner membrane.