In Vivo Mitochondrial Import

The positive charges and structural properties of the mitochondrial leader sequence of aldehyde dehydrogenase have been extensively studied in vitro. The results of these studies showed that increasing the helicity of this leader would compensate for reduced import from positive charge substitutions of arginine with glutamine or the insertion of negative charged residues made in the native leader. In this in vivo study, utilizing the green fluorescent protein (GFP) as a passenger protein, import results showed the opposite effect with respect to helicity, but the results from mutations made within the native leader sequence were consistent between the in vitro and in vivo experiments. Leader mutations that reduced the efficiency of import resulted in a cytosolic accumulation of a truncated GFP chimera that was fluorescent but devoid of a mitochondrial leader. The native leader efficiently imported before GFP could achieve a stable, import-incompetent structure, suggesting that import was coupled with translation. As a test for a co-translational mechanism, a chimera of GFP that contained the native leader of aldehyde dehydrogenase attached at the N terminus and a C-terminal endoplasmic reticulum targeting signal attached to the C terminus of GFP was constructed. This chimera was localized exclusively to mitochondria. The import result with the dual signal chimera provides support for a co-translational mitochondrial import pathway.

The positive charges and structural properties of the mitochondrial leader sequence of aldehyde dehydrogenase have been extensively studied in vitro. The results of these studies showed that increasing the helicity of this leader would compensate for reduced import from positive charge substitutions of arginine with glutamine or the insertion of negative charged residues made in the native leader. In this in vivo study, utilizing the green fluorescent protein (GFP) as a passenger protein, import results showed the opposite effect with respect to helicity, but the results from mutations made within the native leader sequence were consistent between the in vitro and in vivo experiments. Leader mutations that reduced the efficiency of import resulted in a cytosolic accumulation of a truncated GFP chimera that was fluorescent but devoid of a mitochondrial leader. The native leader efficiently imported before GFP could achieve a stable, import-incompetent structure, suggesting that import was coupled with translation. As a test for a co-translational mechanism, a chimera of GFP that contained the native leader of aldehyde dehydrogenase attached at the N terminus and a C-terminal endoplasmic reticulum targeting signal attached to the C terminus of GFP was constructed. This chimera was localized exclusively to mitochondria. The import result with the dual signal chimera provides support for a co-translational mitochondrial import pathway.
The majority of mitochondrial proteins are encoded by nuclear DNA and are synthesized on cytosolic ribosomes as precursor proteins. As a consequence of their cellular location during synthesis, mitochondrial proteins must be specifically targeted to the organelle and be imported. The N-terminal portion of these precursor proteins possesses a leader sequence that contains sufficient information to be recognized by the mitochondrial import apparatus, leading to import into mitochondria. Mitochondrial presequences do not share any primary sequence identity, but they do show a statistical bias of positively charged amino acid residues, provided mostly through arginine residues; very few presequences contain negatively charged amino acids (1). The positive charges in the presequences have the potential to interact with the negatively charged surface of mitochondrial membranes (2), the mitochon-drial import receptor (3), and the mitochondrial processing peptidase (4,5). Mitochondrial leader sequences also share an ability to form an amphiphilic, ␣-helix (1,6,7). Numerous two-dimensional NMR studies of peptides corresponding to mitochondrial leaders show that these leaders do form an amphiphilic helix and have provided more structural detail in showing different structural motifs between leader sequences (8 -13). Thus, there is an apparent charge and structural component needed to target precursor proteins to the mitochondria.
Mutations designed to study the role of positive charge, structure, or hydrophobicity of leader sequences have been performed under in vitro conditions (14 -19). In this setting, numerous cytosolic components that assist the import process may be absent or in limited quantity due to the conditions of the assay. Due to the membrane binding property of leaders, in an in vivo setting the leader sequence may additionally need to discriminate among the various organelles (6). Although there is an apparent preference of the leader for the lipid composition found in mitochondria, potential nonspecific interactions with membrane surfaces of other organelles may hinder the process of locating mitochondria, especially in mutants designed to study the import properties of the signal. Additionally, the total number of positive charges of the leader may be more important in vivo although in vitro experiments have shown that numerous positive charges can be removed without apparent interference with import. Thus, out of the diversity among all mitochondrial leader sequences, positive charges, structural motif, and the hydrophobicity of the leader may be balanced to minimize nonspecific membrane binding that may not be detected in vitro (15).

EXPERIMENTAL PROCEDURES
Materials-The pEGFP-N1 vector and anti-GFP 1 monoclonal antibody were purchased from CLONTECH; Dulbecco's modified Eagle's medium, calf serum, and trypsin-EDTA were obtained from Life Technologies, Inc.; the antibiotics and HEPES were purchased from Sigma; paraformaldehyde was from Aldrich; the SuperFect Transfection Reagent was obtained from QIAGEN; the TNT coupled transcription and translation system and the pGEM-7Zf(ϩ) vector were obtained from Promega; [ 35 S] methionine and [␣-35 S]dATP were purchased from Amersham Pharmacia Biotech; the SequiTherm EXCEL DNA Sequencing kit was from Epicentre Technologies Company; the alkaline phosphatase-conjugated goat anti-mouse IgG was from Bio-Rad; nitrocellulose membranes were from Schleicher and Schuell; and the Complete protease inhibitor mixture was from Roche Molecular Biochemicals. Su-perScript II RNaseH Ϫ reverse transcriptase was obtained from Life Technologies; random hexamers were obtained from Promega; and the pET19b vector was from Novagen. * This work was supported in part by National Institutes of Health Grants GM53269 and AA10795. This is paper 15901 from the Purdue University Agricultural Experiment Station. 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 1 The abbreviations used are: GFP; green fluorescent protein; EGFP, enhanced green fluorescent protein; pALDH, precursor aldehyde dehydrogenase; ALDH, mature aldehyde dehydrogenase; pALDH-RGP, linker-deleted leader of precursor aldehyde dehydrogenase; SP1, signal peptide of pALDH with mature amino acid segment; SP2, signal peptide without mature amino acid segment; PCR, polymerase chain reaction; ER, endoplasmic reticulum; PAGE, polyacrylamide gel electrophoresis.
Cells-HeLa cells were provided by Dr. Steven Broyles (Department of Biochemistry, Purdue University). HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 100 units penicillin, 100 g/ml streptomycin, and 0.25 g/ml amphotericin B at 37°C under an atmosphere of 5% CO 2 and 95% air.
Cloning of Rat Microsomal Aldehyde Dehydrogenase 3 cDNA by Reverse Transcription-PCR-The cDNA for rat microsomal ALDH3 was cloned by reverse transcription-PCR based on the published sequence (23). Rat liver total RNA was isolated by Dr. Yie Lane Chen, formerly of our laboratory. Total RNA (about 5 g) was used for the synthesis of the first strand of cDNA by following the protocol from Life Technologies, Inc. Random hexamers were used for the reverse transcription reaction, which was performed at 42°C for 1 h. The first strand of cDNA was then used as the template for PCR. Two specific primers corresponding to the 5Ј (contains a NdeI site) and 3Ј (contains a BamHI site) ends of the rat microsomal ALDH3 cDNA sequence were used to amplify the full length of the rat microsomal ALDH3.
The reverse transcription-PCR products were digested with NdeI and BamHI and then cloned into the pET19b vector. The amplified cDNA was confirmed by sequencing. The constructs, SP1-EGFP, EGFP-ER, and SP1-EGFP-ER ( Fig. 1), were constructed by the splicing by overlap extension PCR method (20 -22) and cloned into the pEGFP-N1 vector. The SP1 R3Q, R3Q/R10Q, and linker-deleted (RGP from position 11 to 13 of the rat pALDH2 leader sequence was deleted) (ϪRGP)-EGFP-ER were also constructed. All constructs were confirmed by sequencing the pALDH leader sequence, the N-and C-terminal EGFP fusion region, and the the ER-targeting signal sequence.
Direct Observation of EGFP Fluorescence in Cultured Cells-HeLa cells were plated on coverslips in six-well plates in 3 ml of growth medium (Dulbecco's modified Eagle's medium plus 10% calf serum) the day before transfection. The cells were 40 -80% confluent on the day of transfection. The cells were transfected with 2.5 g of plasmid DNA at 37°C for 2-3 h using the SuperFect Transfection Reagent from QIA-GEN. Between 10 and 72 h after transfection, the cells were washed twice with phosphate-buffered saline, fixed on the coverslips with 4% paraformaldehyde in PBS at room temperature for 30 min, and washed twice with phosphate-buffered saline. Cells were observed with an Olympus BX60 fluorescence microscope.
Immunoblot Analysis-Culture and transfection of HeLa cells was performed as described above, except that 100-mm dishes and 10 g of plasmids were used for the transfection. The whole cell extracts were separated by SDS-polyacrylamide gel electrophoresis and electrotransferred onto a nitrocellulose membrane. Monoclonal antibody against Aequoria victoria GFP was used as the primary antibody; the alkaline phosphatase-conjugated goat anti-mouse IgG was used as the second antibody.
Isolation of Mitochondria from HeLa Cells Expressing either the Native or the Linker-deleted SP1-EGFP-Culture and transfection of HeLa cells were performed as described above, except that 150-mm dishes and 20 g of plasmid DNA were used for the transfection. 48 h after transfection, the cells were washed twice with ice-cold phosphatebuffered saline, harvested, and then resuspended in ice-cold 0.32 M sucrose and 5 mM HEPES, pH 7.4. The subcellular fractionation was performed as described (24 -27). The cells were homogenized by using a Polytron homogenizer for 30 s at 10,500 rpm. The homogenate was centrifuged with an SS34 rotor at 3,000 rpm for 15 min to pellet the unbroken cells and the nucleus. The supernatant was centrifuged with an SS34 rotor at 10,000 rpm for 20 min to obtain the mitochondria-rich fraction.
Degradation of in Vitro Translated Proteins by the HeLa Cell Cytosol Extracts-HeLa cells were washed twice with ice-cold phosphate-buffered saline, harvested, and then resuspended in ice-cold 0.32 M sucrose and 5 mM HEPES, pH 7.4. The cells were homogenized by using a Polytron homogenizer for 30 s at 10,500 rpm. To remove the unbroken cells, the nucleus, and the mitochondria, the homogenates were centrifuged with an SS34 rotor at 10,000 rpm for 20 min. The cytosolic supernatant was removed, aliquoted, and saved at Ϫ70°C. Radiola-FIG. 1. The N-terminal amino acid sequence of the SP1-and SP2-EGFP constructs. A, the leader sequence of the precursor ALDH is shown in boldface type, while those of the mature portion of ALDH are shown in italics. The processing site is indicated by the triangle above the signal sequence. The hatched blocks below the leader show the two helical segments of the leader between the flexible RGP linker. The molecular weights of pALDH-EGFP, ALDH-EGFP, and EGFP are indicated by the arrows and are rounded to whole numbers. B, EGFP-ER and SP1-EGFP-ER constructs, used to test for co-translational import, are illustrated to show the sequence of the 40 amino acids from the rat microsomal ALDH3 ER signal.
beled SP1-EGFP, SP1(R3Q)-EGFP, and EGFP were produced by the TNT coupled transcription and translation with the pGEM7Zf(ϩ) constructs in the presence of [ 35 S]methionine following the manufacturer's instruction. The in vitro translated proteins were mixed with the HeLa cell cytosol extracts in equal volume at room temperature for fixed periods of time. EGFP was added in the both the SP1-EGFP and SP1-(R3Q)-EGFP incubations to test for protease resistance over the time points used. An aliquot was withdrawn and added to SDS-polyacrylamide gel electrophoresis treatment buffer, which contained protease inhibitors. The mixture was heated to 95°C for 5 min. The samples were analyzed on a 12.5% SDS-polyacrylamide gel, and labeled proteins were visualized by Bio-Rad phosphor imaging.
Subcellular Fractionation of HeLa Cells Expressing either EGFP-ER or SP1-EGFP-ER-Culture and transfection of HeLa cells were performed as described above, except that 150-mm dishes and 20 g of plasmid DNA were used for the transfection. HeLa cells were transfected by plasmids expressing either EGFP-ER or SP1-EGFP-ER. Between 48 and 72 h after transfection, the cells were washed twice with ice-cold phosphate-buffered saline, harvested, and then resuspended in ice-cold 5 mM HEPES, pH 7.4, 0.32 M sucrose. The subcellular fractionation was performed as described (24 -27). The cells were homogenized by using a Polytron homogenizer for 30 s at 10,500 rpm. The homogenate was centrifuged with an SS34 rotor at 3,000 rpm for 15 min; the pellet (P1) was the unbroken cells and the nucleus. The supernatant was centrifuged with an SS34 rotor at 10,000 rpm for 20 min; the pellet (P2) was the mitochondria-rich fraction. The supernatant was centrifuged with an TLA100.3 rotor at 100,000 rpm for 90 min; the pellet (P3) was the microsome-rich fraction, and the final supernatant was the cytosol from HeLa cells. The different fractions (homogenate, P1, P2, P3, and supernatant) were analyzed by a Western blot. Succinatecytochrome c dehydrogenase was used as the mitochondria marker enzyme, and the enzyme assay was performed as described (28). Lactate dehydrogenase was used as the cytosol marker enzyme, and the enzyme assay was performed as described (29). NADPH cytochrome P-450 reductase was used as the ER marker enzyme, and the enzyme assay was performed as described (30,31).

Detection of Fluorescence in HeLa Cells
Transfected with EGFP and the SP1 Constructs of pALDH-EGFP-Cells that were transiently expressing EGFP displayed an observable fluorescence that was distributed throughout the whole cell, including the nucleus (Fig. 2), as shown by others (32). Specific targeting of EGFP to any subcellular organelle was not observed. Cells, which expressed the EGFP protein with an attached pALDH mitochondrial leader sequence, displayed fluorescence localized to mitochondria. Rhodamine 123, a mitochondrial specific stain was used to visually confirm EGFP targeting to the organelle (data not shown). Portions of some cells that expressed pALDH-EGFP appeared to have cytosolic fluorescence. This was due to the fluorescence of mitochondria above and below the focal plane of the cell. To further verify that the pALDH-EGFP protein was localized to the mitochon-FIG. 2. Fluorescent microscopy of the HeLa cells transiently expressing SP1 and SP2 forms of the pALDH-EGFP chimeras. HeLa cells were cultured on coverslips and transfected with 2.5 g of plasmid DNA for the native and mutant forms of the SP1-and SP2-EGFP constructs (first 16 photographs). The expressed EGFP, without a mitochondrial signal, was distributed throughout the cell, including the nucleus. The SP1-EGFP localized to mitochondria, which are seen either as small dots or long cylinders. The absence of fluorescence in the cytosol and nucleus is indicative of efficient import. The intracellular location of the EGFP-ER and SP1-EGFP-ER proteins are shown in the last four photos. The EGFP-ER was localized to the ER that appeared as a lacy network of membranes. The R3Q(R10Q)-EGFP-ER and (ϪRGP)-EGFP-ER constructs do not efficiently target EGFP-ER to the mitochondria. The extent of ER localization is more for R3Q(R10Q)-than (ϪRGP)-EGFP-ER as mitochondrial fluorescence were not readily observed. dria, cells that were transiently expressing pALDH-EGFP were analyzed by a Western blot to determine the molecular weight of the expressed protein. As shown in Fig. 3, a band of 30 kDa was observed, consistent with the imported and processed 32-kDa pALDH-EGFP chimera. The fluorescence microscopy and Western blot results from the transient expression of the EGFP and pALDH-EGFP demonstrated that the leader sequence of pALDH could target EGFP into the mitochondria.
It was shown that only one positive charge in the first helix of the leader was required for import in vitro, since the single point mutations of R3Q and R10Q all imported to native levels, while the R3Q(R10Q) mutant was severely impaired at import (5). As shown in Fig. 2, under in vivo conditions, R3Q and R10Q were capable of targeting EGFP to mitochondria, but cytosolic and nuclear fluorescence were observed, indicating that not all of the precursor proteins were imported. The R3Q and R10Q mutants were not imported at similar levels, since the cytosolic fluorescence of EGFP appeared to be greater with the R3Q mutant. The double mutant R3Q(R10Q) was the most impaired of the three mutants, since fluorescence was mostly observed in the cytosol.
The in vivo results with the Arg to Gln mutations in the first helix were consistent with the in vitro results. However, two unexpected properties emerged in subsequent in vivo experiments. First, there was no apparent necessity for positive charges in the C-helix of the native leader. In vitro, the two arginines in the C-helix of the native leader could be substituted with glutamines as single point mutations with no effect on import, but substituting both positive charges with glutamine resulted in a 2-fold reduction in import. However, in vivo, substitution of both C-helix arginines with glutamine did not effect import. The second unexpected result was that the linker-deleted leader was imported poorly in vivo, whereas it was found to import better than the native leader in vitro. Removing the helix-breaking residues of RGP in the sequence in the native leader allows the leader to form a continuous helix. This property has been shown to consistently recover in vitro import in mutations of the native leader that reduced or eliminated import. In the in vitro experiments, the R3Q(R10Q) mutation in the native leader essentially eliminated import, but the R3Q(R10Q)-RGP leader restored import to approximately 50% of native level. In the in vivo experiments, the pALDH-RGP imported less than native as shown in Fig. 2, and the R3Q(R10Q)-RGP leader was essentially nonfunctional.
The import properties of the pALDH leader were shown to be surprisingly tolerant to the mutagenic introduction of negatively charged amino acids, provided that the charge of first helix of the leader remained positive (33). As shown in Fig. 2, the import of the S7E was less than native level, and the S7E(R11E) mutant did not appear to import, consistent with the in vitro results. Although the first helix of the leader remained positive in the S7E mutant, the loss of import in the S7E(R11E) mutant was probably due to the negative charge being placed adjacent to the first helix. In contrast to S7E(R11E), the S7E(R14E) was imported in vivo. The S7E(R14E) mutant surpassed the ability of the native leader to import in vitro, but there appears to be only a slight reduction in import from what was observed with the S7E mutant in vivo. The S7E(R14E) mutant does not have the additional negative charge adjacent to the first helix to influence the apparent critical positive charge in the first helix. The results of these in vivo experiments reveal that positive charges necessary for import are identical to those necessary for locating the mitochondria and do not require all five positive charges. Subcellular Localization as Determined by Molecular Weight of the EGFP Constructs-After the precursor proteins are imported into mitochondria, many of the leader sequences are proteolytically removed by a matrix space peptidase. The pALDH leader sequence contains a terminal -RLLS-sequence, which serves as the recognition site for the matrix-processing protein (5). As a direct measure of import in transfected HeLa cells, we harvested cells expressing the various constructs of EGFP and analyzed them by a Western blot using monoclonal antibodies against EGFP. For the majority of the precursor proteins used in this in vivo study, a 30-kDa protein would be the expected molecular mass of an imported and processed protein. The R17Q and R14Q(R17Q) mutants have lost the peptidase recognition motif, and these leaders were not proteolytically removed by the peptidase in rat liver mitochondria (5). Additionally, the pALDH-RGP and R3Q(R10Q)-RGP leaders were also nonprocessable leaders, although they contain the recognition site (5,11). These nonprocessable proteins should retain the leader after import and display a molecular mass of 32 kDa.
As shown in Fig. 3A (lane 4), the native pALDH-EGFP leader resolved to the predicted, processed mass of 30 kDa. No observable precursor band of 32 kDa was observed, suggesting that little, if any, of the precursor protein remained in the cytosol. If a processable mutant leader was less capable of importing EGFP, we would expect to observe a 32-kDa band and a 30-kDa band in the Western blot, corresponding to the nonimported precursor and the imported and processed protein, respectively. Mutant leader sequences that reduced import displayed the expected 30-kDa band, indicative of import and processing, but also displayed an unexpected 28-kDa band, Fig. 3. This 28-kDa band corresponds to the molecular mass of EGFP, which apparently had lost the entire N-terminal fusion sequence due to nonspecific proteolysis. For many of the mutant leader sequences, the intensity of the 28-kDa band found in the Western blot paralleled the pattern of cytosolic fluorescence shown in Fig. 2. Good import of the native protein resulted in a single 30-kDa band, and less efficient import resulted in both a 30-and 28-kDa band. Total loss of import resulted in a single 28-kDa band.
An unexpected band was observed for the nonprocessed leaders, pALDH-RGP, R17Q, and R14Q(R17Q). Data are shown only for pALDH-RGP in Fig. 3A (lane 8). These proteins all displayed a 30-kDa band, suggesting that they were processed. This was not observed in our previous in vitro import studies using rat liver mitochondria or processing studies with purified mitochondrial processing peptidase (11,34). A Western blot of isolated mitochondria from HeLa cells expressing either the native or linker-deleted EGFP constructs displayed a 30-kDa band for both proteins (data not shown). Based on the current knowledge of the processing motifs of matrix space processing enzymes, we cannot offer an explanation for the removal of the leaders in some of the mutants used in these experiments. However, independent of how the pALDH-RGP leader could be processed, the experimental evidence does point to a mitochondrial proteolysis as the source of the 30-kDa bands observed in these Western blots.
N-terminal Cytosolic Degradation of pALDH-EGFP-expressed Proteins-The observation of cytosolic fluorescence in the microscopy experiments with the import-impaired constructs indicates that EGFP could properly fold and was resistant to proteolysis. To discern whether or not the mutations made in the leader were susceptible to proteolysis, thereby corrupting the signal, the native leader and the R3Q mutant were exposed to HeLa cell cytosol extracts as described under "Experimental Procedures." As shown in Fig. 4, both pALDH-EGFP and the R3Q-EGFP mutant were rapidly reduced to a slightly smaller molecular weight. After 30 min, all of the leader sequence and the joining amino acids were degraded, but the EGFPs were still present and resistant to protease digestion (data not shown). The susceptibility to N-terminal protease digestion of the native and the R3Q mutant were similar, suggesting that the native leader could be just as likely to be rapidly proteolyzed in vivo as it was in vitro.
SP2 Constructs of pALDH-EGFP and Mutant pALDH-EGFP-The susceptibility of both the native leader sequence and the R3Q mutant to proteolysis with HeLa cell extracts were similar but did not clearly indicate whether proteolysis occurred within the leader or within the amino acids of the mature portion of pALDH. To address this issue, a new chimeric of EGFP (SP2) was made in which six amino acids were used to bridge the leader sequence to EGFP as illustrated in Fig. 1. As shown in Fig. 2, the cellular fluorescence of the native leader, R3Q, and the linker-deleted mutants in the SP2-EGFP constructs were indistinguishable from the SP1 constructs.
In the Western blot using the SP2 constructs, the expected molecular mass of the imported and processed leader is 28 kDa. We would expect to see only a 28-kDa band and little, if any, 30-kDa precursor band if they were imported well. For a protein that was not imported and was retained in the cytosol, the molecular mass would be 30 kDa if the leader were resistant to proteolysis. The Western blot from cells transiently expressing the native leader, R3Q, and the linker-deleted leader in SP2 form all displayed only a single 28-kDa band (Fig. 3). This was the expected result for the native SP2-EGFP protein. Finding no 30-kDa precursor band for the R3Q and linker-deleted SP2-EGFP proteins, which were imported poorly, suggests that the leader sequence was susceptible to cytosolic degradation. The import results with the SP1-and SP2-EGFP constructs indicate that import needed to be rapid enough to protect the leader from cytosolic proteolysis.
Co-translational Versus Post-translational Import-The rapid proteolysis of the leader, in addition to the proteaseresistant structure of EGFP, suggested that in vivo import needed to be on the same time scale as protein synthesis. Using the same in vivo system, we explored the hypothesis that pALDH-EGFP may have been imported in a co-translational manner. A C-terminal ER signal was fused to the C terminus of the SP1 pALDH-EGFP construct in order to test if import was occurring before the protein was free from the ribosome. The C-terminal ER targeting sequence was from rat microsomal ALDH3, which is composed of the C-terminal 35 amino acids (25). These C-terminal 35 amino acids contain both the transmembrane domain and the ER targeting signal. If import were a post-translational event, the EGFP passenger would be found in both mitochondria and ER. However, if it were a co-translational event, then, despite having two attached targeting signals, the EGFP passenger would only be found in the mitochondria. The pALDH-EGFP-ER and EGFP-ER constructs are illustrated in Fig. 1.
In the cells expressing EGFP-ER, the fluorescence was localized in what appeared to be ER, which was observed as a lacy network of membranes surrounding the nucleus. The ER localization of the expressed EGFP-ER proteins was confirmed by subcellular fractionation. In cells expressing the construct with both signals, SP1-EGFP-ER, the EGFP fluorescence was found only in the mitochondria with no detectable ER localization (Fig. 2). The cellular localization of the SP1-EGFP-ER protein in mitochondria was consistent with a co-translational mecha- nism, since no detectable ER or cytosolic localization was observed with the dual signal protein.
To examine whether the C-terminal ER signal was functional in the SP1-EGFP-ER construct, the ER targeting signal was fused to the less efficient mitochondrial signals of SP1(R3Q), SP1(R3Q/R10Q), and SP1(ϪRGP)-EGFP. Data are not shown for SP1(R3Q)-EGFP. When these chimeras were expressed in HeLa cells, the EGFP fluorescence was localized to both mitochondria and ER (Fig. 2). This verified that the ER signal would function when both signals were presented on the same protein.
Subcellular Localization as Determined by Molecular Weight of the Different ER Targeting Constructs-The molecular weight difference between the precursor and mature forms was used to indicate whether or not the expressed chimeras were imported. The molecular mass of the precursor SP1-EGFP-ER is 36 kDa, and that of the imported and processed protein is 34 kDa. A Western blot of cells expressing SP1-EGFP-ER showed only a 34-kDa band consistent with mitochondrial import (Fig.  5). Western blotting of cells expressing the 32-kDa EGFP-ER protein showed a single band of 32 kDa, consistent with ER targeting. This data is in agreement with the fluorescent images that showed that the SP1-EGFP-ER construct was imported exclusively into mitochondria and further supports a co-translation import mechanism.
Subcellular Fractionation of HeLa Cells Expressing either EGFP-ER or SP1-EGFP-ER-To confirm that SP1-EGFP-ER was not localized on the ER membrane, the intracellular distributions of EGFP-ER or SP1-EGFP-ER were analyzed by subcellular fractionation. The separation of ER membranes from mitochondria was confirmed by assaying two typical marker enzymes, NADPH-cytochrome P450 reductase, which has been shown to be an integral ER membrane protein (35), and a mitochondrial marker, succinate-cytochrome c reductase (28). The purity of the fractions was higher than 70%. The whole cell extracts and different organelle fractions from HeLa cells expressing either EGFP-ER or SP1-EGFP-ER were assayed by a Western blot (Fig. 6). From HeLa cells expressing the EGFP-ER, a strong band was found in the ER-rich fraction (P3), and a faint band was observed in the mitochondria-rich fraction (P2). This was most likely due to some fragments of the ER membrane pelleting with the mitochondrial fraction. Western blotting of HeLa cells expressing SP1-EGFP-ER displayed strong bands associated with the P2 fraction and no observable band in the P3 fraction. These results were consistent with the fluorescence microscopy observation that SP1-EGFP-ER was in the mitochondria and not associated with the ER. All of these results are consistent with co-translational mitochondrial import.

DISCUSSION
The basic notion of how pALDH was imported into mitochondria came from mutagenic investigations and the use of model peptides for synthetic membrane binding and two-dimensional NMR structural studies. From these studies, the two helices of the pALDH leader were thought to have different roles in the import process. The N-helix was proposed to be responsible for targeting to the receptor, while the C-helix was more responsible for binding and maintaining a stable N-terminal helix (36,37). Binding studies showed that the C-terminal helix was more protected from amide exchange when bound to synthetic membranes, and mutations involving the N-terminal charges did not reduce the ability of the peptides to bind membranes as it did in the C-helix (5,37). Thus, it was modeled that the C-helix may bind the outer membrane and allow the N-helix to search for a receptor in two-dimensional space. The positive charges in the first helix of the leader are essential for both import and mitochondrial targeting in vivo, and there is no apparent necessity for positive charges in the C-terminal helix of the leader. Thus, the proposed membrane-binding model may not be valid in vivo for pALDH. Surprisingly, and unexplainably, improving the structural properties of the leader through the linker-deleted leader reduced import efficiency and would not compensate for a reduction in positive charge as it consistently did in vitro (5,33).
What became apparent throughout this study was that mutant leaders that resulted in reduced import efficiency were proteolyzed by cytosolic enzymes. Thus, import needed to be sufficiently rapid to avoid degradation of the signal. The remarkable resistance of EGFP to cytosolic proteolysis suggests that it forms a stable tertiary structure. Mitochondrial proteins must remain unfolded in the cytosol in order to be imported. The EGFP passenger protein, however, will fold regardless of whether or not it is successfully targeted into the mitochondria. This is corroborated by a recent study, which has demonstrated efficient autonomous folding of GFP as well as chaperoninmediated folding (38). This suggests that if pALDH-EGFP were to be released into the cytosol for post-translational import, folding would occur and would prevent import. Conversely, if protein synthesis were occurring simultaneously with import, the leader would be protected from proteolytic degradation, and the EGFP passenger protein would not fold. The co-transla- tional import of pALDH is further evident from the import results obtained with a C-terminal ER signal attached to the pALDH-EGFP protein that was not targeted to the ER. In the dual signal construct of pALDH-EGFP-ER, the combination of rapid folding and stability of EGFP, in addition to the alternative targeting signal that did not target the protein to the organelle ER, suggests that post-translational import of pALDH-EGFP is improbable.
Whether or not mitochondrial import can be co-translational is controversial. Evidence supporting post-translational and co-translational routes exist, but no direct in vivo evidence proves which mechanism exists. This discrepancy is potentially due to the fact that both routes of import are possible, but the import studies have largely focused on one pathway. It is generally accepted that mitochondrial import occurs post-translationally (39 -50). In vivo evidence exists for a post-translational mechanism, since carbonyl cyanide m-chlorophenylhydrazonetreated cells accumulated F1-␤-ATPase in the cytosol and were then imported after reestablishing potential (46). However, this same experiment with cytochrome c 1 demonstrated the opposite results and also showed that the protein was proteolized if it accumulated in the cytosol. It is clear that a posttranslation route has been established, and numerous cytosolic factors such as mitochondrial stimulating factors and heat shock proteins assist the precursor protein until import is achieved. However, numerous in vitro and in vivo studies suggest that a co-translational import mechanism might actually be required for some precursor proteins (51)(52)(53)(54)(55)(56)(57)(58)(59)(60). Ribosomal association with mitochondria has been observed in yeast (51)(52)(53), and recently, it has been reported that ribosomes could stably bind to purified rat liver mitochondria (61), which would be consistent with a co-translational pathway. The mechanism by which a leader is recognized early in protein synthesis and potentially routed to either pathway remains to be discovered. It is clear that for a leader to import in vivo by a post-translational mechanism, resistance of the leader sequence to protease degradation is required, and this requirement is not met by all leader sequences or the mature portion of the protein. This study, in combination with studies from other laboratories, provides strong evidence for a co-translational mechanism coexisting with the established post-translational mechanism.