Structural basis of yeast Tim40/Mia40 as an oxidative translocator in the mitochondrial intermembrane space

Abstract

The mitochondrial intermembrane space (IMS) contains many small cysteine-bearing proteins, and their passage across the outer membrane and subsequent folding require recognition and disulfide bond transfer by an oxidative translocator Tim40/Mia40 in the inner membrane facing the IMS. Here we determined the crystal structure of the core domain of yeast Mia40 (Mia40C4) as a fusion protein with maltose-binding protein at a resolution of 3 Å. The overall structure of Mia40C4 is a fruit-dish-like shape with a hydrophobic concave region, which accommodates a linker segment of the fusion protein in a helical conformation, likely mimicking a bound substrate. Replacement of the hydrophobic residues in this region resulted in growth defects and impaired assembly of a substrate protein. The Cys296-Cys298 disulfide bond is close to the hydrophobic concave region or possible substrate-binding site, so that it can mediate disulfide bond transfer to substrate proteins. These results are consistent with the growth phenotypes of Mia40 mutant cells containing Ser replacement of the conserved cysteine residues.

Mitochondria are essential double-membrane organelles in eukaryotic cells and contain ≈1,000 different proteins. Mitochondrial functions rely on correct transport of resident proteins synthesized in the cytosol to mitochondria, which is mediated by >30 mitochondrial proteins (1–3). Most mitochondrial proteins use the TOM40 complex in the outer membrane to cross the outer membrane, and subsequently sorting pathways branch out for different mitochondrial subcompartments. The intermembrane space (IMS) between the outer and inner mitochondrial membranes contains many proteins with a molecular mass <20 kDa and characteristic cysteine motifs. Import of those small cysteine-containing proteins is driven by a redox-regulated translocator Tim40/Mia40 (4–6) and its partner proteins including Erv1 (7, 8) and Hot13 (9, 10).

Yeast Mia40 is anchored to the inner membrane through its N-terminal hydrophobic segment and exposes a large C-terminal domain to the IMS (4, 6). The IMS domain of Mia40 contains six conserved Cys residues (C1-C6) at positions 296, 298, 307, 317, 330, and 340, which form a CPC motif (C1 and C2) and twin CX₉C motifs (C3-X₉-C4 and C5-X₉-C6). These invariant Cys residues are essential for the Mia40 function (4), and form three disulfide bonds (C1-C2, C3-C6, and C4-C5) in the oxidized state (11). After crossing the outer membrane via the TOM40 complex, substrate proteins for the Mia40 pathway bind to Mia40 transiently, suggesting the role of Mia40 in recognition and trapping of the substrates to prevent their retrotranslocation (4, 5). The interactions between the oxidized form of Mia40 and its substrates involve disulfide bonds, and subsequent release of the substrate proteins allows transfer of a disulfide bond from Mia40 to them (7, 8). The resultant reduced form of Mia40 is oxidized by a sulfhydryl oxidase Erv1 in the IMS for the next round of disulfide bond transfer (7, 8). Reduced Erv1 is oxidized by cytochrome c, which is then oxidized by electron transfer to either the cytochrome oxidase complex or to cytochrome c peroxidase (12, 13). The disulfide transfer reaction mediated by Mia40 is promoted by another small protein in the IMS, Hot13 (9). Hot13 specifically interacts with Mia40 and presumably prevents Mia40 from binding to Zn²⁺; thereby, facilitating its efficient oxidation by Erv1 (10).

To gain more insight into the functions of Mia40 in the protein import into the IMS and subsequent redox folding, we decided to determine the X-ray structure of the core domain of Mia40. Combined with the results of serine replacement of the conserved cysteine residues in Mia40, the revealed structure of the Mia40 core domain allowed us to propose roles of a possible substrate binding region and the conserved three disulfide bonds in disulfide bond transfer to substrate proteins in the IMS.

Results and Discussion

Roles of Three Disulfide Pairs in Mia40.

Yeast Mia40 contains six conserved cysteine residues (C1-C6) that can form three disulfide pairs in its oxidized form: the C1-C2 disulfide pair in the CPC motif, and the C3-C6 and C4-C5 disulfide pairs (11). Replacement of one of the three cysteine pairs is lethal (4), suggesting the importance of the three cysteine pairs in Mia40 functions. To gain insight into the roles of each cysteine pair, we systematically replaced one or two of the six cysteine residues with serine residue(s) in Mia40. Because Mia40 is essential for viability of yeast cells, we tested the abilities of the mutant Mia40 to replace WT Mia40 in vivo at 23 and 37 °C. For yeast cells, replacement of the first, second, or third cysteine residue of Mia40 with serine (C1S, C2S, or C3S) was lethal at 23 °C (Fig. S1A). However, the C4S and C5S mutants can complement the depletion of WT Mia40 both at 23 and 37 °C, yet the C6S mutant can complement at 23 °C, but not at 37 °C (Fig. 1A). The double mutations with Ser replacement of C1 or C2 with the other were lethal (C1,2S, C1,3S, C1,4S, C1,5S, C1,6S, C2,4S, C2,5S, and C2,6S) or caused severe growth defects (C2,3S) (Fig. S1A). The double mutants with Ser replacement of C3 or C6 and the other were not viable except for the C3,6S mutant, which showed temperature-sensitive growth defects, in the absence of WT Mia40 (Fig. 1A; Fig. S1A). Therefore, the C1-C2 pair is essential for the Mia40 functions, whereas C3 and C6 are important in such a way that cells can tolerate simultaneous Ser replacement of C3 and C6, but not a single Ser replacement of one of the C3 and C6 pair. C4 and C5 are not essential for the functions of Mia40. Temperature-sensitive growth of C6S and C3,6S was suppressed by the overexpression of Erv1p (Fig. 1B), whereas lethality of C1,2S or C2S was not suppressed by overexpressed Erv1p, suggesting that the defects for C6S and C3,6S arose from the process involving Erv1p. These results are in principle consistent with the recent analyses by Terziyska et al. (14), in which, however, cell growth at elevated temperature or effects of the Erv1 overexpression was not tested.

Fig. 1.

Growth of strains carrying Mia40 Cys mutants. (A) Cells of Δtim40/pRS314-Tim40 (WT), Δtim40/pRS314-Tim40 (C4S), Δtim40/pRS314-Tim40 (C5S), Δtim40/pRS314-Tim40 (C6S), Δtim40/pRS314-Tim40 (C3,6S), and Δtim40/pRS314-Tim40 (C4,5S) were grown to an early stationary phase in SCD (0.67% yeast nitrogen base without amino acids, 0.5% casamino acids, 2% glucose) lacking tryptophan at 23 °C. Cells were diluted in 10-fold increments, and 10 μL of each dilution (starting from 100-fold dilution) was spotted onto SCD lacking tryptophan and incubated at 23 °C for 4 days or at 37 °C for 2 days. (B) Cells of Δtim40/pRS314-Tim40 (WT), Δtim40/pRS314-Tim40 (C6S), and Δtim40/pRS314-Tim40 (C3,6S) harboring the pYO326 vector alone (vector) or pYO326-ERV1 (ERV1) were grown to an early stationary phase in SCD lacking tryptophan and uracil at 23 °C. Cells were diluted in 10-fold increments, and 10 μL of each dilution (starting from 100-fold dilution) was spotted onto SCD lacking tryptophan and uracil and incubated at 23 °C for 3 days or at 37 °C for 2 days.

Overexpression of WT Mia40 from a galactose-inducible GAL7 promoter led to slight defects in cell growth (Fig. S1B). However, when we overexpressed single or double Cys→Ser mutant Mia40 in the presence of WT Mia40, the double Cys→Ser mutants lacking C3 or C6 (C1,3S, C2,3S, C3,4S, C3,5S, C2,6S, C4,6S, and C5,6S) tended to exhibit strong dominant negative effects for yeast cells (C1,6S is the exception) (Fig. S1B). The C1S, C3S, and C1,6S mutants exhibited moderate dominant negative effects (Fig. S1B). Mia40 interacts not only with substrate proteins to introduce disulfide bonds, but also with Erv1 to become oxidized after oxidation of substrates. Indeed, most of these dominant negative effects were relieved by overexpression of Erv1 (Fig. S1B). Therefore, the C3-C6 disulfide bond appears important for the proper interaction with Erv1, and if this interaction is impaired, Mia40 might exhibit dominant negative phenotypes by, for example, generating a dead-end reaction intermediate.

Overall Structure of Mia40C4.

To gain further insight into the structure-function relationship of Mia40, we decided to determine the 3D structure of Mia40. For this purpose, we designed a series of core domains of Mia40 on the basis of limited protease digestion and of the sequence comparison of Mia40 homologs from various organisms. Thus, we found that residues 284–365 of Mia40 (named as Mia40C4), which includes a minimal functional unit of residues 284–352 (14), is most suitable for structural analyses (Fig. 2A). Mia40C4 isolated from Escherichia coli cells took folded structures as judged from CD and NMR spectral characterization (described below). However, despite extensive screenings for crystallization of Mia40C4, crystals suitable for X-ray structural determination were not obtained, perhaps due to tendency of Mia40C4 to oligomerize at high concentrations. To overcome this oligomerization problem, we attempted to increase solubility of Mia40C4 by attaching a large soluble protein domain. Thus, we fused Mia40C4 to the C terminus of maltose binding protein (MBP) with a 16-residue spacer segment at the DNA level (Fig. 2A), and prepared the recombinant fusion protein (MBP-Mia40C4) from E. coli cells. MBP-Mia40C4 was now successfully crystallized as described in Material and Methods.

Fig. 2.

Structure of Mia40C4. (A) Diagrams of Mia40, Mia40C4, and MBP- Mia40C4. (B) The crystal structure of Mia40C4 in ribbon form viewed from different angles. The S atoms of Cys of the three disulfide bonds are presented as yellow spheres. (C) Electrostatic potential mapped onto the surface of Mia40C4 as viewed in B. Red and blue regions represent localization of negative and positive charge, respectively.

The crystal structure of the MBP-Mia40C4 fusion protein was determined at a resolution of 3.0 Å (Fig. S2A and Table S1). The overall structure of residues 3–453 of MBP-Mia40C4 including residues 284–353 of Mia40C4 (residues 384–453 in MBP-Mia40C4 numbering) is well defined, although the N-terminal two residues and the C-terminal 13 residues are disordered. For description of the structure of the Mia40C4 domain of MBP-Mia40C4, we use residue numbering of Mia40 hereafter (residues 284–353; to convert MBP-Mia40C4 numbering, add 100).

Mia40C4 is a fruit-dish-like shape with approximate dimensions of 20 × 27 × 36 Å (Fig. 2B). The overall structure consists of a long N-terminal loop (residues 284–307) followed by two 13-residue α-helices (H1, residues 308–320, and H2, residues 331–343). Helices H1 and H2 form an antiparallel hairpin connected by a 10-residue loop, which are virtually connected by the two disulfide bonds C3-C6 (Cys307-Cys340) and C4-C5 (Cys317-Cys330). Although Mia40C4 was found to have a previously undescribed fold, based on analysis using the DALI server, similar topological arrangements of two helices connected by two disulfide bridges are found in several proteins including yeast Cox17, cytochrome oxidase copper chaperone, which is, like Mia40 itself, an import substrate for Mia40 (Fig. S2B) (5).

Three Disulfide Bonds of Mia40C4.

Mia40C4 has three disulfide bonds, C1-C2, C3-C6, and C4-C5. In the determined crystal structure of Mia40C4, the most N-terminal C1-C2 bond, which forms a CPC motif, is exposed at the tip of the loop (Fig. 2B). The accessible surface areas of S_γ atom for Cys296 and Cys298 are 24.90 and 12.60 Ų, respectively. However, the C3-C6 and C4-C5 disulfide bonds, which involve the twin CX₉C motif, are sandwiched by the two helices, and the accessible surface areas of S_γ atoms for Cys307, Cys340, Cys317, and Cys330 are 0, 11.60, 8.5, and 0 Ų, respectively. Those results are consistent with the previous report that the C1-C2 bond is more redox-sensitive than the two others (11, 14). Because Mia40 with the C3S, C6S, or C3,6S mutation becomes sensitive to protease digestion, the C3-C6 disulfide bond was proposed to have a structural role in Mia40 (14). However, the purified C3S, C5S, C6S, and C3,6S mutants of Mia40C4 (plus 4 residues) exhibited CD spectra similar to that of WT Mia40C4 (Fig. S3), suggesting that the C3-C6 or C4-C5 disulfide bond is not essential for proper overall folding of Mia40C4 (the C1-C2 disulfide bond is not required for correct folding, either; see structural determination and refinement in SI Materials and Methods). Rather, because the Cys→Ser mutations of the C3-C6 disulfide pair (except for the C3,6S mutation) in Mia40 led to dominant negative effects (Fig. S1B) and the Mia40 variant with the C6S or C3,6S mutation was suppressed by overexpression of Erv1 in vivo (Fig. 1B), the C3-C6 disulfide bond may be important for interactions with its partner oxidase Erv1. This interpretation is also consistent with the previous report that purified Mia40 with the C3S, C6S, or C3,6S mutation cannot interact with purified Erv1 (14).

Linker Segment Mimics Substrate Binding.

In the crystal structure, Mia40C4 has a hydrophobic concave surface (10 × 12 Å) formed by the hydrophobic side chains of Phe-311, Phe-315, Phe-318, Val-319, Phe-334, Met-337, and Phe-341 protruding from helices H1 and H2 (Fig. 3 A and B). Interestingly, 11-residue segment (residues 373–383 in MBP-Mia40C4 numbering) of the linker between MBP and Mia40C4 is accommodated, partly forming a helical structure, in the hydrophobic concave region (Fig. 3 A and B). We reasoned that the linker segment may mimic a substrate peptide segment recognized by Mia40C4. The total accessible surface area buried by the linker segment is 428.5 Ų, which can be ascribed to Phe-311, Phe-315, Phe-318, Val-319, Phe-334, Met-337, and Phe-341 with buried accessible surface areas of 23.8, 42.5, 71.9, 13.0, 36.9, 25.0, and 8.0 Ų, respectively. When we examined the proximate relationship for the residues between the linker segment (residues 373–383 in MBP-Mia40C4 numbering) and the Mia40C4 domain (in Mia40 numbering), the hydrophobic residues on the concave region (Phe-311, Phe-315, Phe-318, and Phe-334) are indeed close to Ser-376, Ile-377, Glu-378, Pro-381, and Glu-382 of the linker segment (Fig. 3 C and D).

Fig. 3.

Putative substrate binding region of Mia40C4. (A) The crystal structure of Mia40C4 is drawn as in Fig. 2B with the linker segment (GRGSIEGRPEF; residues 373–383 in MBP-Mia40C4 numbering) in stick form (main chain only). The hydrophobic residues (Phe-311, Phe-315, Phe-318, Val-319, Phe-334, Met-337, and Phe-341 in Mia40C4 numbering) of Mia40C4 interacting with the linker segment are drawn in yellow stick form. (B) Molecular surface of Mia40C4 with the linker segment (in stick form) are drawn with hydrophobic residues (Leu, Ile, Val, Phe, Tyr, Ala, Thr, Trp) in yellow, acidic residues (Asp, Glu) in red, and basic residues (Lys, Arg) in blue, as viewed in A. (C) The Mia40C4 domain (ribbon model with the S atoms of Cys of the three disulfide bonds as yellow spheres) with side chains (as in A) and the linker segment (Left, blue ribbon form; Right, blue stick form) with side chains (stick model in blue with N in dark blue and O in red) viewed from different angles. Residue numbers are shown for the linker segment in blue (in MBP-Mia40C4 numbering) and for the Mia40C4 domain in black (in Mia40 numbering). (D) Proximate relationship for the residues between the linker segment (residues 373–383) and the Mia40C4 domain within a distance of 4 Å. Residues are indicated in blue for the linker segment (in MBP-Mia40C4 numbering) and in black for the Mia40C4 domain (in Mia40 numbering) (Left). The C296-P297-C298 segment and the closest part of the bound linker segment drawn in stick form. The shortest distance is indicated (Right).

Relative geometry of the three disulfide bonds with the hydrophobic substrate binding site shows that the C1-C2 disulfide bond is closer to the linker segment bound to the hydrophobic surface than the C3-C6 and C4-C5 disulfide bonds located on the opposite side of the molecule (Fig. 3C). Indeed, minimal distances between the disulfide bonds and the linker helix is 3.64, 10.76, and 7.85 Å for the C1-C2, C3-C6, and C4-C5 bonds, respectively, and in particular Cys298 (C2) is close to Phe-383 of the linker segment (Fig. 3D). Therefore, the C1-C2 disulfide bond is likely responsible for transferring a disulfide bond to substrate proteins, which is consistent with the results of the Cys→Ser mutants of Mia40 (Fig. S1).

Role of Possible Substrate Binding Site of Mia40.

To further assess the roles of the possible substrate binding region of Mia40, we replaced Phe-311, Phe-315, Phe-318, or Phe-334 of Mia40 with Glu, and analyzed the abilities of those Mia40 mutants to replace WT Mia40 in vivo. The results showed that F315E and F318E mutations were lethal, and that F311E and F334E mutations led to temperature-sensitive growth defect at 37 °C (Fig. 4 A and B). Because Mia40C4 (plus 4 residues) variants with one of the four Phe→Glu mutations purified from E. coli cells showed CD spectra similar to WT Mia40C4 (Fig. S3), Phe-311, Phe-315, Phe-318, or Phe-334 may be directly involved in the function of Mia40. Replacement of Phe-315 or Phe-318 with Ala or Leu did not cause growth defects (Fig. S4A), suggesting that hydrophobic residues in this region are important for the Mia40 functions such as substrate binding. NMR titration analyses showed that the recently identified substrate segment for Mia40 (MSP2; SNLVERSFTD) (15) indeed binds to isolated WT Mia40C4 with higher affinity (K_d = ≈0.7 mM) than to its F318E mutant (Fig. S5). Interestingly, temperature-sensitive growth defects of F311E and F334E mutants were suppressed by overexpression of Erv1 from the multicopy plasmid (Fig. 4B). Out of the four Phe→Glu mutants, only F315E mutant showed a dominant negative growth phenotype when expressed from the GAL7 promoter at 30 °C (Fig. S4B).

Fig. 4.

Effects of Mia40 Phe mutations on cell growth and protein import. (A) Δtim40/pRS316-Tim40 harboring the pRS314 vector alone (vector) or plasmid expressing WT or a series of Phe→Glu mutants of the MIA40 gene from its own promoter were grown on a SCD plate lacking tryptophan and uracil (−Trp −Ura) or a SCD plate lacking tryptophan, but containing 0.1% 5-fluoroorotic acid (−Trp FOA) at 23 °C for 3days. (B) Cells of Δtim40/pRS314-Tim40 (WT), Δtim40/pRS314-Tim40 (F311E), and Δtim40/pRS314-Tim40 (F334E) harboring vector (pYO326) alone or pYO326-ERV1 (ERV1) were grown to an early stationary phase in SCD lacking tryptophan and uracil at 23 °C. Cells were diluted in 10-fold increments, and 10 μL of each dilution (starting from 100-fold dilution) was spotted onto SCD lacking tryptophan and uracil and incubated at 23 °C for 3 days or at 37 °C for 2 days. (C) Mitochondria were isolated from the WT strain harboring pYO326 vector, F311E mutant strains harboring pYO326 vector (F311E), or pYO326-ERV1 (F311E, Erv1↑), which were grown in lactate medium at 23 °C. Radiolabeled Tim9 was incubated with the indicated mitochondria at 32 °C for indicated times in the presence of 50 μM ZnSO₄. The reisolated mitochondria were solubilized with 1% digitonin and analyzed by BN-PAGE followed by radioimaging. An asterisk indicates nonspecific bands. (D) Proteins of indicated mitochondria (4.2 and 50 μg for each mitochondrial prepartion) were analyzed by reducing (Left, with β-mercaptoethanol) or nonreducing (Right, without β-mercaptoethanol) SDS/PAGE followed by immunoblotting with anti-Mia40 (Upper) and anti-Erv1 (Lower) antibodies. The oxidized and reduced forms of Mia40, the Erv1 homodimer, and the Mia40-Erv1 complex are indicated.

We then isolated mitochondria with the F311E or F334E mutant of Mia40 and tested their import ability of small Tim proteins in vitro. F311E mutant mitochondria showed normal levels of the translocator components (Fig. S6A) and the normal TIM22 complex of ≈300 kDa (Fig. S6B), whereas F334E mutant mitochondria showed reduced levels of Erv1, small Tim proteins (Tim9, Tim10, and Tim12), and Tim22 (Fig. S6A). The mitochondrial Hsp60 precursor (a TIM23 pathway substrate), ADP-ATP carrier, and dicarboxylate carrier (both are TIM22 pathway substrates) were imported into isolated WT and F311E mitochondria with similar efficiency while their import into F334E mitochondria was retarded at 37 °C (Fig. S6C). We next incubated radiolabeled Tim9, a genuine substrate for the Mia40 pathway, with WT, F311E, and F334E mitochondria at 25 °C or 32 °C, and subjected the mitochondria to solubilization with digitonin and blue-native (BN)-PAGE analyses (Fig. 4C; Fig. S6E). On incubation with WT mitochondria, Tim9 formed a mixed-disulfide intermediate with Mia40 (140 kDa), and then, the mature Tim9-Tim10 complex (70 kDa) and the TIM22 complex (300 kDa) were formed (Fig. 4C; Fig. S6E) (4, 5). Tim9 was imported into WT, F311E, and F334E mitochondria efficiently at 25 °C (Fig. S6D), yet formation of the Tim9-Mia40 intermediate as well as the final assembly of the Tim9-Tim10 and TIM22 complexes were retarded in F334E mitochondria (Fig. S6E), likely due to the reduced level of Erv1 and the mature TIM22 complex (Fig. S6 A and B). However, although F311E mitochondria contain normal levels of Erv1 and other translocator components, formation of the final TIM22 complex was retarded at 25 °C (Fig. S6E) and 32 °C (Fig. 4C). Interestingly, formation of the Tim9-Mia40 intermediate was even enhanced at 25 °C (Fig. S6E), which was also observed for mia40-4 mutant mitochondria previously (5). Overexpression of Erv1 (≈10-fold) partly relieved the assembly defects of Tim9 (F311E/Erv1↑; Fig. 4C). Although overexpression of Erv1 did not change the ratio of the oxidized and reduced forms of Mia40, it significantly increased the amount of the Mia40-Erv1 complex with an intermolecular disulfide bond (Fig. 4D). Together with the observed suppression of the growth defects of the F311E mutant by Erv1 overexpression (Fig. 4B), we suppose that Phe-311 in the hydrophobic region of Mia40 is not essential for mere substrate binding, but is also important for the later step of disulfide bond transfer perhaps involving the substrate-Mia40-Erv1 ternary complex (16).

Conclusion

Here, we determined the crystal structure of the core domain of Mia40, Mia40C4, as a fusion protein with MBP at a resolution of 3 Å. The overall structure of Mia40C4 is a fruit-dish-like shape with a hydrophobic concave region accommodating a linker segment of the fusion protein in a helical conformation, which probably mimics the bound substrate. The C1-C2 disulfide bond is close to the “pseudo” substrate of the linker segment in the possible substrate-binding region, so that it can mediate the formation of a mixed disulfide intermediate, in which C1 or C2 is linked to a cysteine residue of substrate proteins. During preparation of the manuscript, Banci et al. (17) reported the NMR structure of the core domain of human Mia40, which is similar to the crystal structure of yeast Mia40C4, although no direct structural information on the interaction with substrates were assessed. To reveal the precise structure-based mechanism of the disulfide bond transfer to the substrate, followed by oxidation of C1 and C2 by Erv1, structural determination of the Mia40-Erv1 complex is the essential subject of future studies.

Materials and Methods

Plasmids and Strains.

Constructions of Mia40 mutant plasmids and yeast strains and growth conditions for yeast cells are described in SI Materials and Methods and Table S2.

MBP-Mia40C4 Fusion Protein.

Preparation of the MBP-Mia40C4 fusion protein is described in SI Materials and Methods.

Crystallization and X-Ray Diffraction.

Block shaped crystals were grown by the sitting vapor diffusion method at 20 °C from 100 mM Na-acetate, pH 4.6, with 30% (wt/vol) PEG 4000 and 200 mM NH₄-acetate. The crystals belonged to an orthombic space group (P2₁2₁2₁ with unit cell dimensions of a = 41.26 Å, b = 101.40 Å, and c = 109.36 Å) and contained one molecule of MBP-Mia40C4 in an asymmetric unit with a solvent content of 44.9%. After cryoprotection with 10% (vol/vol) glycerol, the crystals were mounted in a cryoloop and flash cooled with liquid nitrogen stream at 100 K. Diffraction data were collected using a Rigaku FR-E SuperBright (Cu Kα, 45 kV, 45 mA) and a R-AXIS VII area detector with a detector distance of 200 mm, and processed with DENZO and SCALEPACK (18). The procedures for structural determination and refinement (Figs. S2 and Fig. S7) are described in SI Materials and Methods.

In Vitro Import.

Radiolabeled precursor proteins were synthesized in a cell-free translation system with reticulocyte lysate in the presence of [³⁵S]-methionine. Mitochondria were isolated from yeast strains (WT, F311E, and F334E) cultivated in lactate medium at 23 °C. In vitro import into isolated mitochondria was performed as described previously (4).