In vitro import of the Rieske iron-sulfur protein by trypanosome mitochondria.

Most of the proteins present in the mitochondrion are imported to that location from the cytosol. While this process has been studied extensively in fungal and mammalian systems, little work has been done in other eukaryotic organisms. We are particularly interested in the Trypanosoma brucei system because this organism developmentally regulates mitochondrial function during its life cycle and because one of the imported proteins lacks a conventional targeting sequence. We report here the development of an in vitro import system using crude trypanosome mitochondria and a nuclear encoded, mitochondrial protein. Import of the Rieske iron-sulfur protein subunit of the cytochrome c reductase complex requires a membrane potential, ATP, and a protein component on the mitochondrial surface. The precursor protein is sequentially processed to the mature form in two steps by peptidases that require divalent metal ions for activity. As in other eukaryotic systems, the first processing event occurs inside the inner membrane and is probably catalyzed by a matrix-processing protease. Surprisingly, the second processing activity is located outside the inner membrane. Both processing steps require ATP but are independent of a membrane potential. We suggest that the trypanosome iron-sulfur protein is imported along a “conservative sorting pathway” but that the assembly mechanism of the reductase complex may be unique to trypanosomes.

Most of the proteins associated with the mitochondrion are encoded in the nuclear genome. These proteins are translated on free cytosolic ribosomes and must be transported from the cytosolic pool across one or both of the organelle's membranes (Reid and Schatz, 1982). The import of proteins into the matrix requires the sequential interaction of the precursor with a receptor, a general insertion pore (GIP), 1 a component of the contact sites between the outer and inner mitochondrial mem-branes, and the mitochondrial matrix heat shock proteins (HSPs). Precursors are first recognized at the mitochondrial surface by protein-specific receptors and, in the presence of ATP, are presented to the GIP in an unfolded, import-competent state (Schlossmann et al., 1994;Sollner et al., 1989Sollner et al., , 1990Harkness et al., 1994;Kiebler et al., 1993;Wachter et al., 1994;Pfanner et al., 1987). The precursors are inserted into the outer membrane at the GIP (Pfaller et al., 1988). Insertion into the inner membrane then occurs at contact sites by an undefined process that requires a membrane potential (Schleyer and Neupert, 1985;Pfanner and Neupert, 1985;Martin et al., 1991). Proteins that enter the matrix are associated with HSPs, and ATP is required for their release and for proper folding (Kang et al., 1990;Ostermann et al., 1989;Berthold et al., 1995;Stuart et al., 1994). The targeting and sorting signals for import into the matrix are usually found in the amino-terminal sequences of the precursor protein (for review see Hartl et al. (1989)). In most cases this signal sequence is removed after import by a general matrix-processing protease (MPP) that is soluble in the matrix and that requires divalent manganese, cobalt, or zinc ions for activity (Ou et al., 1989;Bohni et al., 1983;Hawlitschek et al., 1988).
Some proteins such as subunit 9 of the F 0 -ATPase (subunit 9) and the cytochrome c 1 and Rieske iron-sulfur protein (ISP) subunits of the cytochrome c reductase complex are imported into the matrix, processed, and then redirected into or across the inner membrane (Mahlke et al., 1990;Hartl et al., 1986Hartl et al., , 1987. Elements of this "conservative sorting pathway" are apparently descended from the endosymbiotic bacterial ancestor of the mitochondrion (Hartl et al., 1986). For subunit 9, ATP is required for the export from the matrix, and a pH gradient is required for the reinsertion of the protein into the inner membrane (Rojo et al., 1995). Precursor proteins imported along this pathway are processed a second time by intermediate peptidases that appear to have different specificities, requirements, and locations. In yeast, the peptidase that cleaves the ISP subunit of the reductase complex, the mitochondrial intermediate peptidase (MIP), is located on the matrix side of the inner membrane, while the peptidase that cleaves the cytochrome c 1 subunit, the inner membrane protease I, is located in the intermembrane space (Fu et al., 1990;Ramabadran and Beattie, 1992;Isaya et al., 1994;Nunnari et al., 1993). Both peptidases require divalent metal ions for activity, but the metal ion preferences are different (Ramabadran and Beattie, 1992;Nicholson et al., 1989). In addition, the insertion of the iron-sulfur center into the apoprotein is not a prerequisite either for cleavage or for assembly of the ISP into the cytochrome c reductase complex, but heme addition is required for cytochrome c 1 maturation (Graham and Trumpower, 1991;Nicholson et al., 1989).
During its developmental cycle, Trypanosoma brucei, a protozoan parasite of humans and domestic animals, regulates mitochondrial function to coincide with the nutrients available in the mammalian and insect hosts (Vickerman, 1985). We have purified one of the developmentally regulated mitochondrial protein complexes, the cytochrome c reductase, and have cloned two of its subunit proteins, the Rieske ISP and cytochrome c 1 Hajduk, 1992, 1995;Priest et al., 1993). The trypanosome ISP has a conventional cleaved presequence that may function as a mitochondrial targeting signal, but the cytochrome c 1 lacks a presequence (Priest et al., 1993;Priest and Hajduk, 1995). We suggested that the 17-residue aminoterminal ISP presequence directs import into the trypanosome mitochondrion along a conservative sorting pathway and that it is processed in two steps like the fungal ISP presequences. We report here that a crude mitochondrial fraction prepared from the insect (procyclic) form of T. brucei is able to both import and process in vitro translated trypanosome ISP. The import requires a protein component on the exposed surface of the mitochondrion as well as ATP and a membrane potential. Similar import requirements have recently been reported for T. brucei and Leishmania tarentolae mitochondria following extensive density gradient purification (Hauser et al., 1996). Both of the import substrates used by Hauser et al., a heterologous yeast mitochondrial alcohol dehydrogenase and an artificial precursor protein containing the 9-amino acid presequence of the trypanosome dihydrolipoamide dehydrogenase, were cleaved once (as expected for matrix-localized proteins). However, the cleavage activity was not characterized, nor was its submitochondrial location determined. In our system, the trypanosome ISP precursor is processed in two steps by metalloproteases that reside on opposite sides of the mitochondrial inner membrane. The processing from the precursor to the mature form does not require a membrane potential but does require a significant amount of ATP in the matrix. The intermembrane space location of the T. brucei ISP intermediate peptidase is different from that described in other eukaryotic systems, suggesting that the assembly pathway of the cytochrome c reductase complex may have some unusual features in trypanosomes.

MATERIALS AND METHODS
Construction of Clone, Transcription, and Translation-Clone pT-bFe/SNsi contains the entire 894-base pair T. brucei Rieske iron-sulfur protein coding sequence as well as 1.3 kilobase pairs of upstream noncoding sequence and 1 kilobase pair of downstream coding sequence (Priest and Hajduk, 1995). The ATG initiation codon was moved closer to the plasmid-encoded T7 transcription initiation site by the deletion of a 1.2-kilobase pair HinCII fragment from the 5Ј-noncoding sequence. A template for transcription of the ISP coding sequence was generated by cleavage of the subclone (pTbFe/SNsi⌬HinC) at a BspH I site situated 34 base pair downstream of the termination codon. From this template a 1166-nucleotide RNA was transcribed with a 5Ј 7-methyl-GpppG cap (Boehringer Mannheim) using T7 RNA polymerase (Promega, Madison, WI) (Sambrook et al., 1989). Analysis of the transcript by methylmercury-agarose gel electrophoresis demonstrated that essentially all of the RNA was full-length (data not shown).
Radiolabeled ISP was synthesized in vitro using a rabbit reticulocyte translation system with [ 35 S]methionine (DuPont NEN; approximately 1000 Ci/mmol, 10 mCi/ml) as directed by the manufacturer (Promega). At an input RNA concentration of 40 ng/l, each microliter of translation mixture incorporated between 2.9 and 5.7 ϫ 10 4 cpm into trichloroacetic acid-precipitable protein.
Trypanosome Growth and Mitochondria Isolation-T. brucei procyclic cells (TREU 667) were grown in shaking cultures at 26°C using SSMH medium as described elsewhere (Torri and Hajduk, 1988;Priest and Hajduk, 1994a). When the cells reached a density of about 1 ϫ 10 7 cells/ml, they were collected by centrifugation at 4°C (7100 ϫ g, 15 min) and then washed at 4°C with a buffer containing 0.15 M NaCl, 20 mM glucose, and 20 mM NaH 2 PO 4 at pH 7.4. All subsequent procedures were performed at 4°C unless otherwise indicated. The cells were resuspended at a density of 1 ϫ 10 9 cells/ml in buffer containing 0.25 M sucrose, 10 mM MOPS/KOH at pH 7.2, and 2 mM EDTA (SME buffer). The cells were equilibrated with 1000 psi N 2 for 15 min in a nitrogen cavitation bomb (minibomb cell disruption chamber; Kontes, Vineland, NJ). Release from the bomb resulted in the disruption of greater than 95% of the cells. Dilution of the sample with 1 volume of SME buffer was followed by centrifugation for 10 min at 1500 ϫ g to remove intact cells, cell ghosts, and large cellular debris. A crude mitochondrial fraction was then collected by centrifugation at 16,000 ϫ g for 15 min. The pellet was resuspended in the original volume of SME buffer and again centrifuged at 1500 ϫ g for 5 min. The crude mitochondrial fraction was collected from the 1500 ϫ g supernatant by centrifugation at 16,000 ϫ g for 15 min. The final pellet was resuspended at a concentration of 3 ϫ 10 9 cell equivalents/ml (about 2 mg of protein/ml) in storage buffer containing 50% glycerol, 0.25 M sucrose, 10 mM MOPS/KOH at pH 7.2, and 2 mM EDTA. After homogenization in a Dounce homogenizer (5-10 strokes, B pestle), aliquots were frozen at Ϫ70°C. Mitochondrial import activity was stable for several weeks under these conditions. Standard Import Assay Protocol-Crude mitochondria in glycerol storage buffer were warmed to 4°C, centrifuged for 3 s at 14,000 ϫ g (12,000 rpm, Sorvall microcentrifuge, Wilmington, DE) to remove aggregated material, and then aliquoted for individual assays. To remove the glycerol storage buffer, crude mitochondria were diluted with 9 volumes of SME buffer and then collected by centrifugation at 14,000 ϫ g for 15 min. The resulting pellets were resuspended in import buffer, and the assays were begun by adding radiolabeled rabbit reticulocyte lysate translation mixture. Each 25 l of assay mixture (final volume) contained the mitochondria from 1.5 ϫ 10 8 cells (about 100 g of protein), 5 l of rabbit reticulocyte translation mixture (20% v/v), and 20 l of basal import buffer at a final concentration of 0.25 M sucrose, 80 mM KCl, 5 mM MgCl 2 , 5 mM dithiothreitol, 1 mg/ml fatty acid-free bovine serum albumin, and 10 mM MOPS/KOH at pH 7.2 (modified from Pfanner and Neupert, 1985). In most cases additional ATP was provided at a final concentration of 2 mM, and 10 mM creatine phosphate and 0.1 mg/ml creatine phosphokinase were added as an ATP regeneration system. In the indicated experiments, 8 mM potassium ascorbate, 0.2 mM N,N,NЈ,NЈ-tetramethylphenylenediamine, and 5 mM NADH were included to energize the mitochondria and provide reducing equivalents (Pfanner and Neupert, 1986). "Complete" import buffer contained the additional ATP, the ATP regeneration system, and the reagents to energize the mitochondria. Protein import assays were incubated at room temperature for 30 min. Assays were terminated by transfer to 4°C, and the mitochondria-containing fraction was collected by centrifugation at 14,000 ϫ g for 10 min at 4°C. Where possible, samples that were to be directly compared (i.e. Ϯ proteinase K treatment) were taken from the same import assay mixture with the total volume scaled up appropriately.
Post-import Treatment and Proteinase K Digestion-To remove labeled protein that was bound to the membranes but not imported, the mitochondria-containing pellet was resuspended at a concentration of 3 ϫ 10 9 cell equivalents/ml in SME buffer, and proteinase K (Life Technologies, Inc.) was added to 30 g/ml (unless otherwise indicated in the figure legend). Digestion was carried out at room temperature for 15 min. One volume of cold SME buffer containing 4 mM phenylmethylsulfonyl fluoride (PMSF) was added to terminate the digestion, and the mitochondrial fraction was collected by centrifugation at 14,000 ϫ g for 10 min at 4°C. The resulting pellet was solubilized by heating for 3 min at 95°C in SDS sample loading buffer with 2 mM PMSF.
Mitochondria that were not to be digested with proteinase K were resuspended in cold SME buffer at the same concentration described above. Following dilution with 1 volume of cold SME with 4 mM PMSF, the mitochondrial fraction was collected by centrifugation at 4°C and dissolved in SDS loading buffer as described above.
Analysis of Import Reactions-Proteins were resolved on 12% SDSpolyacrylamide gels as described by Laemmli (1972). Each lane was loaded with the crude mitochondrial protein from 1.2 ϫ 10 8 cells (80% of the protein from a 25-l import assay). Gels were stained with Coomassie Blue to ensure that each lane contained an equivalent protein load. Gels were impregnated with Fluoro-Hance (Research Products International, Mount Prospect, IL), dried, and exposed to x-ray film at Ϫ70°C (X-Omat AR, Eastman Kodak Co., Rochester, NY). The relative amounts of protected radiolabeled proteins were quantitated by storage phosphor autoradiography using a Molecular Dynamics model 400E PhosphorImager.
Miscellaneous Procedures-Protein concentration was measured with the dye binding assay of Bradford (1976) using Bio-Rad dye reagent (Richmond, CA). Concentrated stocks of potato apyrase (Grade VIII, 1000 units/ml), fatty acid-free bovine serum albumin (10 mg/ml), and creatine phosphokinase (10 mg/ml) (all from Sigma) were made in 10 mM MOPS/KOH at pH 7.2. Restriction enzymes were purchased from Life Technologies, Inc. (Gaithersburg, MD). All other reagents were purchased from Sigma unless otherwise specified. Valinomycin, carbonylcyanide m-chlorophenylhydrazone (CCCP), oligomycin, antimycin A, and PMSF were all made as 100-fold concentrated stocks in ethanol.

A Crude Mitochondrial Fraction Isolated from T. brucei Procyclic Cells Imports and Processes the Rieske ISP Precursor-
Mitochondria isolation from a kinetoplastid protozoan like T. brucei has been confounded by the fact that each cell has only one large and extensively convoluted mitochondrion (Vickerman, 1965). The nitrogen cavitation lysis protocol described in this work was developed to minimize the disruption of the mitochondrion while efficiently lysing the T. brucei procyclic cells. The protocol uses an isotonic buffer to encourage the formation and maintenance of an intact inner mitochondrial membrane during and after lysis, and it is carried out at 4°C to preserve the import competence of the mitochondria. Since the crude mitochondrial fraction from our differential centrifugation was capable of importing between 4 and 22% of the input labeled mitochondrial precursor proteins, further purification was deemed unnecessary. Tubulin from the disrupted flagella was one of the major contaminants in the crude mitochondrial fraction. The tubulin is apparent on SDS-polyacrylamide gels as two large protein bands in the 50-kDa mass range (see Fig.  1A, lane 2). As shown in lane 3 of Fig. 1A, treatment of the mitochondrial fraction with proteinase K at 30 g/ml removed much of the contamination, yet left many of the other proteins intact. Despite the obvious impurity of our mitochondria, those isolated by our protocol were at least 5-fold more efficient (per mg of protein) at import than mitochondria isolated by the hypotonic lysis and Percoll gradient technique developed earlier in our laboratory (Harris et al., 1990) (data not shown).
The radiolabeled protein profile of in vitro translated T. brucei Rieske ISP is shown in lane 1 of Fig. 1B. The full-length ISP precursor (35-kDa apparent molecular mass, 33.6 kDa predicted) and a truncation product from translation initiation at an internal methionine (29-kDa apparent molecular mass, 27.6 kDa predicted) were represented in the translation products in approximately a 4:3 ratio (for the sequence of the ISP, see Priest and Hajduk (1995)). When these proteins were incubated with the mitochondrial fraction described above, both the full-length ISP (p) and the truncation product were recovered with the mitochondria after one wash with buffer ( Fig. 1B, lane 2). In addition, two new protein products derived from the full-length ISP precursor were associated with the mitochondria: an intermediate form with an apparent molecular mass of 34 kDa (i) and a mature form with an apparent molecular mass of 33 (Ϯ1) kDa (m). The size of the mature form corresponds to the size of the ISP found in the native cytochrome c reductase complex (32 kDa) and to the 31.6-kDa size predicted from the sequence of the ISP minus the 17-residue cleaved presequence Hajduk, 1994a, 1995). Following treatment of the mitochondrial fraction with proteinase K at 30 g/ml (Fig. 1B, lanes 3 and 3Ј), only the two new proteins and a small amount of the precursor remained. The truncated ISP form was completely sensitive to external protease. A total of 7% of the input full-length ISP precursor was recovered in the precursor, intermediate, and mature proteins after protease digestion.
The recovery of labeled ISP following incubation with mitochondria cannot be attributed to nonspecific trapping, incomplete protease digestion, or the innate resistance of ISP to digestion. Nonspecific vesicle trapping can be ruled out, since neither the ISP truncation product (which completely lacks the 17-residue cleaved amino-terminal presequence) nor radiolabeled luciferase (data not shown) were protected from protease digestion under the same import conditions. The recovery was also not a result of incomplete protease digestion. An increase in the proteinase K concentration from 30 to 60 g/ml resulted in only a 15% reduction in the amount of recovered ISP precursor and products (6% overall recovery) (Fig. 1C, lanes 2 and  3). Lowering the proteinase K to concentration to 15 g/ml (Fig.  1C, lane 1) increased the recovery of labeled protein from 7 to 11% without changing the ratio of precursor to intermediate and mature forms. The truncated ISP was not apparent even at this lower protease concentration. Since the two smaller products are only generated upon import (see Fig. 4), this result suggests that some of the mitochondria are unstable after extensive manipulation and become permeable to proteinase K. Thus, the estimate of the level of protein import may, in fact, be low. Solubilization of the mitochondrial membranes with CHAPS detergent rendered the ISP precursor and products FIG. 1. The Rieske iron-sulfur protein is imported into isolated T. brucei mitochondria. A, the mitochondria-containing pellet from a 50-l standard import assay in complete buffer (mitochondria from 3 ϫ 10 8 cells along with 10 l of rabbit reticulocyte translation mixture as detailed under "Materials and Methods") was resuspended in 100 l of ice-cold SME buffer and divided into two samples. One sample was diluted with 1 volume of ice-cold SME buffer and 4 mM PMSF, collected by centrifugation at 4°C, and dissolved in SDS loading buffer (lane 2). The other sample was digested with proteinase K at 30 g/ml as described under "Materials and Methods" (lane 3). Proteins were resolved on a 12% SDS-polyacrylamide gel and stained with Coomassie Blue. Standards with the indicated molecular mass were run in the lane marked M. Lane 1 contained 0.5 l of rabbit reticulocyte translation mixture. Lanes 2 and 3 contained mitochondria from 1.2 ϫ 10 8 T. brucei cells (80% of each sample). B, autoradiograph of the labeled ISP proteins found in the gel shown in A. Lane designations are the same as in A except that lane 3Ј is a longer exposure of the proteinase K-digested mitochondria shown in lane 3. The positions of the ISP precursor (p), the intermediate protein (i), and the mature form (m) are indicated. C, three 25-l standard import assays were performed in complete buffer as described under "Materials and Methods." Mitochondrial pellets were resuspended in SME buffer and digested with proteinase K at the indicated concentrations for 15 min at room temperature. Digestions were terminated with 1 volume of ice-cold SME buffer and 4 mM PMSF, and the mitochondria were collected by centrifugation and analyzed as described above. D, 50-l standard import assay in complete buffer (mitochondria from 3 ϫ 10 8 cells) was divided into two samples, and the crude mitochondria were collected by centrifugation. One pellet was resuspended in 50 l of SME buffer containing 100 mM KCl and 15 g/ml proteinase K (lane 1). The other pellet was resuspended in the same buffer but with the addition of 2% CHAPS detergent (lane 2). Both samples were allowed to digest for 15 min at room temperature. After the addition of 1 volume of cold SME buffer and 4 mM PMSF, total protein was collected by precipitation in the presence of 80% (v/v) cold acetone. Precipitated proteins were collected by centrifugation, dried, dissolved in SDS sample buffer at 95°C, resolved on a 12% SDSpolyacrylamide gel, and visualized by autoradiography. sensitive to protease digestion even at a proteinase K concentration (15 g/ml) lower than that routinely used in the import experiments (Fig. 1D, lane 2). The apparent decrease in the recovery of the mature ISP form in the import assay lacking detergent ( Fig. 1, compare panel D, lane 1, with panel C, lane 1) is probably an artifact of the acetone precipitation technique used to recover the protein in this experiment.
Import and Processing of the Rieske ISP Are Time-dependent- Fig. 2A demonstrates that the accumulation of labeled ISP within a protease-inaccessible compartment of the mitochondrion is time-dependent. Total ISP import (sum of precursor, intermediate, and mature forms) had a t1 ⁄2 of approximately 6.5 min and reached a maximum at about 20 min. Although this t1 ⁄2 is considerably longer than the t1 ⁄2 reported for ISP in yeast (Ͻ2 min) (Fu et al., 1990), it is similar to that found for other T. brucei mitochondrial proteins in this system. 2 The order and kinetics of appearance of the precursor ISP and of the two processed products as determined from the gel in Fig. 2A by storage phosphor image quantitation (Fig. 2B) suggest a linear sequence of conversion from precursor to intermediate to mature forms. The precursor began to accumulate immediately after the import reaction was initiated (Ͻ1 min), while the intermediate and mature forms were not apparent until after 2.5 min and 5 min of incubation with the mitochondria, respectively. The t1 ⁄2 values for the accumulation of the individual ISP forms were 4.5 min for the precursor, 8 min for the intermediate form, and 12 min for the mature form. The t1 ⁄2 of the mature form may be an underestimate, since processing continued after 30 min even though import had ceased. These results are consistent with a pathway of precursor ISP import and subsequent two-step processing within the T. brucei mitochondrion.
Import of ISP Requires a Protein Component Located on the Exterior Surface of the Mitochondrion-In order to determine if an exposed surface component is required for import, the crude mitochondrial fraction was treated with proteinase K prior to incubation with the labeled ISP precursor. The activity of the protease was modulated by performing the pretreatment at either 4°C (mild digestion) or at 25°C (extensive digestion). As controls, crude mitochondria were also preincubated at these two temperatures in the absence of protease. The protein profiles of the 4°C (Fig. 3A, lane 1) and the 25°C (Fig. 3A, lane 2) controls are identical and are similar to the profile observed earlier without any preincubation (Fig. 1A, lane 2). Preincubation at 4°C in the absence of protease also did not affect the ability of the mitochondria to bind ISP (Fig. 3B, lane 1, ϪProt. K) or to import and process ISP precursor to the intermediate and mature forms (Fig. 3B, lane 5, ϩProt. K). However, preincubation at 25°C (Fig. 3B, lane 6, ϩProt. K) did slightly reduce the overall import activity of the mitochondria (decrease of about 20% relative to the 4°C preincubation shown in lane 5), but the processing activities were unaffected. In contrast to preincubation in buffer alone, the amounts of both the ISP precursor and the truncated form of ISP bound by the 4°C and 25°C protease pretreated mitochondria were decreased to 50 and 30% of the respective amounts bound by the control mitochondria (Fig. 3B, lanes 3 and 4 versus lane 1, ϪProt. K). In addition, pretreatment of mitochondria with proteinase K completely abolished import even under mild digestion conditions at 4°C (Fig. 3B, lanes 7 and 8, ϩProt. K). An analysis of the soluble ISP recovered from import buffer supernatant after incubation with the mitochondria demonstrated that the observed absence of import in the protease-pretreated mitochondria did not result from the digestion of the ISP substrate by residual proteinase K. As expected, more ISP was recovered from the supernatants of the protease pretreated mitochondria that did not import than from the supernatants of the importcompetent preincubation control mitochondria (data not shown).
A comparison of the protein profiles in Fig. 3A suggests that the outer mitochondrial membrane components necessary for ISP binding and import may be hypersensitive to protease digestion. Mild digestion at 4°C (lane 3) clearly did not eliminate all of the tubulin contamination in the 50-kDa size range (compare with controls in lanes 1 and 2), nor did it reduce the protein profile to that observed after extensive digestion at 25°C (lane 4); yet ISP was not imported by these mitochondria (Fig. 3B, lane 7, ϩProt. K). These results indicate that ISP binding and import require protein components that are exposed on the external mitochondrial surface. Interestingly, two of the most prominent proteins that appear to be hypersensitive to protease treatment and whose levels appear to be correlated with import activity have a relative molecular mass of approximately 68 kDa. These proteins are in the same size range as the 72-and 70-kDa MOM 72 receptor molecules of Neurospora crassa and yeast, respectively (Steger et al., 1990).
ISP Import Requires a Membrane Potential-Several inhibitors were used to determine if a membrane potential is required for ISP import. CCCP, a protonophore that uncouples oxidative phosphorylation, was used to dissipate the membrane potential by eliminating the pH gradient across the inner 2 J. W. Priest and S. L. Hajduk, unpublished observations.

FIG. 2. Time course of import of ISP into mitochondria.
A 225-l assay in complete buffer (1.35 ϫ 10 9 cell equivalents of mitochondria) was prepared by scaling up the standard assay protocol. After the addition of the rabbit reticulocyte lysate mixture and shift to room temperature, 25-l aliquots were removed at the indicated time points. The import reaction in each aliquot was terminated by the dilution of the aliquot with 75 l of ice-cold SME buffer containing 66 M CCCP and 0.66 M valinomycin (final concentrations of 50 M and 0.5 M, respectively). Samples were centrifuged at 4°C, digested with 30 mg/ml proteinase K, and dissolved in SDS buffer as described under "Materials and Methods." A, the protected radiolabeled proteins were resolved on a 12% SDS-polyacrylamide gel and visualized by autoradiography. The amounts of protein loaded in each lane were as described previously. The positions of the precursor (p), intermediate form (i), and mature form (m) of the ISP protein are indicated. B, the amounts of each labeled protein were determined by storage phosphor autoradiography and are expressed in arbitrary units. mitochondrial membrane. Similarly, the potassium ionophore valinomycin was used in the presence of 80 mM K ϩ to dissipate the membrane potential while leaving the pH gradient intact. A combination of antimycin A and oligomycin was used to block the physiological formation of the membrane potential; antimycin A inhibits the cytochrome c reductase complex, thereby preventing electron transport and H ϩ translocation, and oligomycin inhibits the H ϩ -translocating F 1 F 0 -ATPase, thereby preventing the ATP-driven formation of a membrane potential (Priest and Hajduk, 1992;Williams et al., 1991).
In the absence of proteinase K (Fig. 4, lanes 1-5, ϪProt. K) both the full-length ISP precursor and the truncated translation product were bound to the mitochondria in approximately the same amounts even when the membrane potential had been dissipated (lanes 3-5). In contrast, import of the ISP into mitochondria was completely abolished by all of the inhibitors tested (Fig. 4, lanes 8 -10, ϩProt. K). This effect was not caused by the inhibitor solvent since ethanol by itself (lane 7) had only a moderate effect (22% reduction) compared with untreated mitochondria (lane 6). In a separate experiment the concentration of CCCP necessary to inhibit 50% of the ISP import was determined to be 17 M (data not shown). Given that an intact membrane potential is a general requirement for insertion into the inner mitochondrial membrane during protein import, this finding suggests that the T. brucei ISP crosses both mitochondrial membranes. Fig. 4 also demonstrates that ISP processing does not occur in the absence of mitochondrial protein import. In the mitochondrial samples that were not treated with proteinase K (lanes 1-5), the processed intermediate and mature ISP forms were only visible in the control and ethanol-treated mitochondria (lanes 1 and 2). Mitochondria that lacked a membrane potential and did not import proteins (lanes 3-5) did not process the ISP. This observation shows that the processing activities are located within the mitochondria and are not simply a contaminant of the mitochondria-enriched fraction that was used for the import assay.
ISP Import and Processing Require ATP-To assess the ATP requirements of T. brucei mitochondrial protein import, various concentrations of potato apyrase were used to deplete the rabbit reticulocyte translation mixture of ATP prior to the import assay. The mitochondria were not themselves pretreated with apyrase, but they were exposed to the apyrase (at 20% of the indicated concentration) and to a large extramitochondrial pool of ADP after the import reaction had been initiated. Since the ADP/ATP carrier protein was not inhibited with carboxyatractyloside, the matrix was probably depleted of ATP by carrier protein-mediated exchange with this exterior FIG. 3. Proteinase K pretreatment of mitochondria abolishes import. Crude mitochondria were recovered from glycerol storage buffer as described under "Materials and Methods." Equivalent samples of mitochondria from 3 ϫ 10 8 cells were resuspended in 100 l of SME buffer and pretreated for 15 min at the indicated temperature (4 or 25°C) either in the absence or presence of 15 g/ml proteinase K. Following the addition of 9 volumes of ice-cold SME buffer and 1 mM PMSF, the mitochondrial fraction was collected by centrifugation at 4°C and washed once in the same buffer. Resulting pellets were resuspended in 50 l (final volume) of complete import buffer and assayed for import activity using the standard protocol. The Coomassie Bluestained protein profiles of the pretreated crude mitochondria are shown in A with the pretreatment conditions indicated under each lane. B, autoradiograph of the labeled ISP bound and imported by the pretreated mitochondria shown in A. Postimport mitochondrial pellets were divided into two samples as described in Fig. 1A. To assess the total amount of ISP associated with the pretreated mitochondria (bound plus imported), one sample from each assay was analyzed by SDSpolyacrylamide gel electrophoresis without further protease treatment (lanes 1-4, ϪProt. K). The positions of the precursor (p) and truncated ISP form (t) are indicated. The other sample from each assay was treated with 30 g/ml proteinase K for 15 min at room temperature to remove labeled ISP protein that was bound to the mitochondria but not imported (lanes 5-8, ϩProt. K). The positions of the precursor (p), intermediate (i), and mature (m) forms are indicated. The pretreatment conditions for the mitochondria are given under each lane. The amounts of protein loaded in each lane were as described in Fig. 1A.   FIG. 4. Import of ISP into mitochondria and protein processing require a membrane potential. Standard import assays (50-l final volume containing mitochondria from 3 ϫ 10 8 cells) were carried out in basal import buffer supplemented with 2 mM ATP and an ATP regeneration system as described under "Materials and Methods." The mitochondrial membrane potential was dissipated by preincubation of the crude mitochondria (10 min at 4°C) with 50 M CCCP (lanes 3 and 8), 0.5 M valinomycin (valin., lanes 4 and 9), or 15 M antimycin A and 30 M oligomycin (anti. A ϩ oligo., lanes 5 and 10). All of the inhibitors were added to the import buffer from 100-fold concentrated stock solutions in ethanol (final concentration of ethanol was 2%). To ensure that the solvent did not inhibit import, an assay was performed in the presence of 2% ethanol (lanes 2 and 7). Import in the absence of inhibitor is shown in lanes 1 and 6. Postimport mitochondrial pellets were divided into two samples and processed as described in Fig. 1A. Mitochondria were analyzed either in the absence of protease digestion (ϪProt. K, lanes 1-5) or after digestion with 30 g/ml proteinase K for 15 min at room temperature (ϩProt. K, lanes 6 -10). The amounts of protein loaded in each lane were as described previously. Positions of the various ISP forms are indicated as in Fig. 3. ADP pool. Also, the import assays were conducted in basal import buffer lacking any supplements in order to minimize the amount of ATP generated by the mitochondria. Fig. 5, lane 1 (Ϫoligomycin), shows that the ATP present in the untreated translation mixture (about 0.5 mM) (Jackson and Hunt, 1983) and in the mitochondria was sufficient to support import in basal import buffer. The energy source for the maintenance of a membrane potential under these conditions has not been identified, but it may be a metabolite present in the reticulocyte lysate. Hartl et al. (1986) noted that reticulocyte lysate could supply enough of an oxidizable substrate to support a low level of ISP import into N. crassa mitochondria. Apyrase hydrolysis of the ATP in the translation mixture completely abolished import (lanes 2-4) even at the lowest apyrase concentration tested (2 units/ml). This effect did not result from a nonspecific interference of the apyrase, since heat-inactivated enzyme did not inhibit import (lane 6). The uptake of precursor ISP was partially restored when ATP and an ATP regeneration system were added to the basal import buffer (lane 5), but the level of import did not reach that observed in the untreated control (lane 1). Apyrase treatment also had a significant effect on the processing of the ISP within the mitochondria. The relative amounts of intermediate and mature protein observed in the presence of ATP (lane 5) were far below the levels observed in the untreated control (lane 1). The inability to fully restore import and processing activities in lane 5 may be explained by the fact that the large amount of apyrase present in the translation mixture (25 times more than that required to eliminate import in lane 2) was not inactivated prior to the import assay. Thus, the overall ATP concentration probably did not approach that present in the control assay, and the peak ATP concentration was probably short lived.
Because apyrase depleted the levels of ATP both outside and inside the mitochondria, we were unable to determine the location of the ATP requirement from the experiment described above. To address the location of the ATP requirement, the experiment was repeated in the presence of oligomycin (Fig. 5,  lanes 7-12, ϩoligomycin), an inhibitor of the mitochondrial F 1 F 0 -ATPase that is known to selectively decrease the ATP concentration within the mitochondrial matrix (Stuart et al., 1994). In the presence of oligomycin, untreated mitochondria took up a large amount of precursor ISP, but they were virtually unable to process the precursor to the intermediate or mature forms (Fig. 5, lane 7). Thus, the ATP concentration in the translation mixture was sufficient to support ISP precursor uptake but was not sufficient to maintain ISP processing within the mitochondria even though the movement of ATP into the matrix was not blocked with carboxyatractyloside. Consistent with the results described above, apyrase treatment of the translation mixture completely blocked import into oligomycin-treated mitochondria (lanes 8 -10), and the uptake (but not processing) of the precursor was partially restored by adding ATP to the reaction (lane 11).
These results demonstrate that ATP is required for both import and processing, and they suggest that ATP is required outside the mitochondria for uptake of precursor and inside the mitochondria for precursor processing. The import observed in the presence of oligomycin also demonstrates that matrix ATP is not required to generate a membrane potential in procyclic cell mitochondria. As described earlier, an additional effect of oligomycin is the inhibition of the ATP-driven formation of a membrane potential by the F 1 F 0 -ATPase. In bloodstream T. brucei cells, oligomycin inhibition of the ATPase causes a complete collapse of the membrane potential (Nolan and Voorheis, 1992;Vercesi et al., 1992). The protection of a substantial amount of precursor ISP in the presence of oligomycin (Fig. 5,  lane 7) and the complete inhibition of import observed in the presence of both oligomycin and antimycin A (Fig. 4, lane 5) demonstrate that the membrane potential generated by an active electron transport chain is sufficient to drive normal physiological processes such as import into procyclic mitochondria. Thus, the ATP requirement for import and processing observed in these mitochondria is probably indicative of an interaction between the ISP precursor and an ATP-hydrolyzing heat shock protein.
The Two ISP-processing Peptidases Require Divalent Cations for Activity and Reside on Opposite Sides of the Mitochondrial Inner Membrane-All of the mitochondrial processing endopeptidases thus far identified require a divalent metal ion for activity. Divalent metal ion chelators were used to determine if the T. brucei enzymes have a similar requirement and to determine the location of these activities within the mitochondrion. In the absence of inhibitor (No Inhib.), mitochondria were able to import ISP (Fig. 6, lane 1), but, probably as a result of the elimination of Mg 2ϩ from the import buffer, the relative amounts of the ISP precursor and the processed products were different from that usually observed. No significant changes were noted in the protein profiles after a 15-min chase (lanes 2-4) except when Mn 2ϩ was included in the chase buffer (lane 5). In that instance much of the intermediate form was converted to the mature form, suggesting that at least some of the mature ISP was processed by a two-step mechanism via the intermediate form.
In the presence of a water-soluble divalent metal ion chelator such as bathophenanthroline disulfonic acid or EDTA (ϩBathophen. or ϩEDTA), the formation of mature ISP during import was largely inhibited, while the formation of the intermediate ISP was unaffected (Fig. 6, lanes 6 and 11). No additional mature ISP was formed during a 15-min chase in the presence of either inhibitor (lanes 7 and 12). However, when the chase was accompanied by the addition of a divalent metal ion, varying amounts of mature ISP were observed (lanes 8 -10 and FIG. 5. Import of ISP into mitochondria and protein processing require ATP. Apyrase or heat-inactivated apyrase (95°C for 10 min) were added to aliquots of rabbit reticulocyte translation mixture at the indicated concentrations, and the lysates were incubated for 15 min at 30°C and for 15 min at 25°C (Pfanner and Neupert, 1986). Treated (lanes 2-4), mock-treated (lane 6), and untreated (lane 1) lysates were then used in the standard import assay protocol with crude mitochondria in unsupplemented basal import buffer (ϪOligomycin) as described under "Materials and Methods" (25-l final volume containing 1.5 ϫ 10 8 cell equivalents of mitochondria and 5 l of rabbit reticulocyte lysate). Treated lysate was also added to one import assay in the presence of 2 mM ATP and an ATP regeneration system in an attempt to restore import activity (lane 5). Postimport mitochondria were treated with 30 g/ml proteinase K prior to analysis as described under "Materials and Methods." Each lane was loaded with the mitochondria from 1.2 ϫ 10 8 cells. The positions of the various forms of the ISP proteins are indicated as in the previous figures. The assays were repeated in the presence of 30 M oligomycin (ϩOligomycin, lanes 7-12) to prevent additional ATP synthesis by the mitochondrial F 1 F 0 -ATPase. [13][14][15]. Of the three ions tested, Mn 2ϩ was the most efficient at restoring the mature ISP-forming activity (lanes 10 and 15) and Mg 2ϩ was the least efficient (lanes 8 and 13). Similar results were obtained when bathophenanthroline disulfonic acid and EDTA were used together (data not shown). Since these two metal ion chelators are charged and are not able to penetrate the inner mitochondrial membrane, these results demonstrate that the activity required for mature ISP formation resides outside the inner mitochondrial membrane. The location of this activity and its sensitivity to EDTA inhibition were entirely unexpected. In other eukaryotes, the mature ISP-forming activity (mitochondrial intermediate peptidase) is not sensitive to EDTA in intact mitochondria (Ramabadran and Beattie 1992;Hartl et al., 1986). Since EDTA was included in our isotonic SME buffer, it is possible that this peptidase was inadvertently inhibited to some extent during the mitochondria isolation protocol. This could explain the Mn 2ϩ -dependent stimulation observed in the absence of inhibitor pretreatment (lane 5) as well as the unusual precursor:product ratio observed when Mg 2ϩ was omitted from the import buffer (Fig. 1A, compare lane 1 and lane 3Ј).
The lack of inhibition of the intermediate ISP-forming activity by the water-soluble chelators described above could result from either the absence of a metal ion requirement or from the localization of the activity within an inaccessible compartment of the mitochondrion. To distinguish between these two possibilities, import was carried out in the presence of both EDTA and o-phenanthroline, a lipid-soluble divalent metal ion chelator. These two inhibitors were used together, since 1 mM ophenanthroline by itself was not able to completely inhibit ISP processing (data not shown). A similar result has been reported by other laboratories (Fu et al., 1990;Hartl et al., 1986;Zwizinski et al., 1983). Neither the intermediate ISP form nor the mature ISP form were observed when both inhibitors were included in the import assay (Fig. 6, lane 16, ϩEDTA/o-phen.). When imported precursor ISP was chased for 15 min in the presence of the inhibitors, only a trace of the intermediate form was detected (lane 17). More of the intermediate ISP was apparent when Mg 2ϩ was included in the chase buffer, but the mature form was still absent (lane 18). In contrast, both forms of ISP were apparent when Ca 2ϩ or Mn 2ϩ were included in the chase buffer (lanes 19 and 20). As with the water-soluble chelator results described above, more processing occurred in the presence of Mn 2ϩ than in the presence of Ca 2ϩ . We conclude that the endopeptidase that cleaves the precursor ISP to the intermediate form resides within the mitochondrial inner membrane and that it requires a divalent metal ion (probably Mn 2ϩ ) for activity.
Three additional conclusions can be drawn from the chelation experiment shown in Fig. 6. First, processing is not required for import of the ISP by isolated mitochondria. Import of the ISP precursor continued even when both of the processing activities were inhibited by the combination of EDTA and ophenanthroline (lane 16). Second, the formation of the ironsulfur center is not a requirement for import or for processing. Much of the Fe 2ϩ should have been chelated by the conditions used in lane 16, yet import occurred and processing was restored by the addition of divalent cations other than iron (lanes 18 -20). This result is consistent with the observation that amino acid substitutions in the Fe/S binding site of the yeast ISP do not affect the processing or the assembly of the protein into the cytochrome c reductase complex (Graham and Trumpower, 1991). Third, the processing of the T. brucei ISP does not require a membrane potential. The inclusion of CCCP in the chase buffer at a concentration that completely inhibited import (50 M, see Fig. 4) did not affect the ability of the mitochondria to process the precursor or the intermediate proteins after the divalent cation requirement had been satisfied (lanes 10, 15, and 20). Given that the two processing activities are located on opposite sides of the inner membrane and that the processing events appear to be sequential, this result implies that the movement of the amino terminus (and perhaps the entire protein) from the matrix to the intermembrane space is independent of a membrane potential.

DISCUSSION
The results presented here demonstrate that the T. brucei ISP precursor is imported into isolated T. brucei mitochondria with requirements similar to those described in other eukaryotic systems. Like the N. crassa ISP, the T. brucei protein must interact with one or more protease-sensitive components on the outer mitochondrial surface in order for import to occur (Pfaller et al., 1988;Zwizinski et al., 1984;Sollner et al., 1989;Schlossmann et al., 1994) (Fig. 7). Given that a mild protease digestion of the T. brucei mitochondria reduced the binding of the ISP precursor by 50%, we suggest that one of these protease-sensitive components is a receptor molecule similar to the MOM 19 and MOM 72 receptors described in N. crassa (Fig. 7, Receptor) (Sollner et al., 1989. Although the definitive identification of an import receptor will require further purification of the mitochondria and a careful analysis of the outer membrane components, it is interesting to note that two of the prominent T. brucei proteins in the 68-kDa size range were hypersensitive to protease digestion and that a hypersensitive protein in the FIG. 6. Processing of imported ISP requires two metal iondependent peptidases. Four 125-l import assays in complete import buffer (each containing the mitochondria from 7.5 ϫ 10 8 cells) were prepared as described in the standard assay protocol (see "Materials and Methods") except that the MgCl 2 in the import buffer was replaced with water (No Inhib.), 1 mM bathophenanthroline disulfonic acid (ϩBathophen.), 2.5 mM EDTA (ϩEDTA), or 2.5 mM EDTA and 1 mM o-phenanthroline (ϩEDTA/o-Phen.). Radiolabeled ISP was imported into the mitochondria using the standard import protocol. Each import assay was then divided into five samples, and the crude mitochondria were collected by centrifugation at 4°C. Each mitochondrial pellet was resuspended in 50 l of buffer containing 0.25 M sucrose, 10 mM MOPS/ KOH at pH 7.2, and 50 M CCCP. One sample from each assay was kept at 4°C (no chase) (lanes 1, 6, 11, and 16); one sample from each assay was chased at room temperature for 15 min in the presence of buffer alone (lane 2) or in the presence of buffer and chelator (at the same concentrations used in the import assays) (lanes 7, 12, and 17); the remaining three samples from each assay were chased for 15 min at room temperature in buffer supplemented with either 5 mM MgCl 2 , 5 mM CaCl 2 , or 5 mM MnCl 2 . All of the samples were then digested with 30 g/ml proteinase K as described under "Materials and Methods." The protected radiolabeled proteins were resolved and analyzed as described. Each lane contained the protein from 1. 16-kDa range was also occasionally observed. 2 The interaction of the ISP precursor with this putative receptor appears to be mediated to some extent by the mature part of the protein and is clearly independent of the membrane potential. We demonstrated that a truncated form of the ISP (lacking the 17-amino acid cleaved presequence and an additional 36 amino-terminal residues) bound to both untreated (specific and nonspecific binding) and protease-pretreated (nonspecific binding) mitochondria to the same extent as did the full-length precursor with an intact presequence. The binding of the truncated and full-length proteins was unaffected by all of the membrane potential poisons tested. These observations are consistent with the suggestion that yeast ISP, in association with ATP-hydrolyzing cytosolic chaperones, may bind to a receptor complex that recognizes the mature part of the protein (Haucke et al., 1995). We do have some evidence that extramitochondrial factors are required for trypanosome protein import: proteins translated in reticulocyte lysate were imported, but the same proteins were not imported when translated in wheat germ extract. 2 Apparently the reticulocyte lysate supplies factors necessary for mitochondrial import (probably HSPs) that are absent or present in insufficient quantity in the wheat germ extract (Murakami et al., 1988). We have not yet determined if the specific binding directed by the mature part of the trypanosome ISP places the protein on the native import pathway, nor do we know what role (if any) the cleaved ISP presequence plays in receptor recognition. In the yeast system, the 31-amino acid ISP presequence is sufficient for both binding and import in the absence of any other targeting signal (Japa and Beattie, 1994). As additional mitochondrial proteins with cleaved presequences are isolated, we hope to identify the features necessary for receptor recognition in this new system.
When incubated with energized mitochondria in the presence of ATP, the T. brucei ISP precursor is imported into a protease-protected compartment (Fig. 7). Although the T. brucei homologs of the GIP and of the inner membrane contact sites shown in Fig. 7 have not yet been demonstrated, the observation that a membrane potential (⌬⌿) is required for import does point to the presence of a conventional import apparatus that would presumably be localized at the contact sites (Schleyer and Neupert, 1985;Schwaiger et al., 1987). The import of the ISP through the mitochondrial membrane is probably directed by the 17-amino acid presequence or by a signal located in the amino terminus of the mature protein, since the truncated ISP form described above was not imported. We are currently working with the ISP and with two additional mitochondrial proteins (the Crithidia fasciculata heat shock protein MCP72 and the T. brucei cytochrome c 1 ) in an effort to identify the precise sequences required for mitochondrial protein import (Effron et al., 1993;Priest et al., 1993). The cytochrome c 1 is of special interest, since it lacks a cleaved presequence (Priest et al., 1993). One interesting feature of these three proteins is that they all have two paired, positively charged amino acids within three residues of the amino terminus of the presequence (or the mature terminus in the case of cytochrome c 1 ). This feature is shared with at least three other trypanosomatid mitochondrial proteins, and it may play a role in mitochondrial targeting or import (Else et al., 1994;Peterson et al., 1993;Bringaud et al., 1995).
The import of the trypanosome ISP has two distinct ATP requirements. Under conditions where ATP is completely removed from both outside and inside the mitochondria, import is abolished; under conditions where the matrix ATP level inside the mitochondria is selectively reduced, precursor ISP is protected but not processed. The role of ATP outside the mitochondrion is probably to maintain the precursor in an "importcompetent" conformation in association with the extramitochondrial heat shock proteins (Pfanner et al., 1987Wachter et al., 1994). We do not yet know if ATP is required for FIG. 7. Proposed import pathway for the Rieske ISP in T. brucei mitochondria. ISP protein is translated in the cytosol as a precursor with a mitochondrial targeting presequence (open box and zigzag line) on the amino terminus. The mature part of the protein interacts with a protein component (RECEPTOR) on the outer surface of the outer mitochondrial membrane. In the presence of a membrane potential (⌬⌿) and ATP, the ISP is moved into the mitochondrial matrix, probably through the general insertion pore (GIP) (shaded squares) and the membrane contact sites (shaded circles). In the matrix, the Mn 2ϩ -requiring matrix processing protease (MPP) cleaves the precursor to the intermediate form by removing a short sequence from the amino terminus (zigzag line). The intermediate form is reinserted into the inner membrane so that the amino terminus is exposed to the intermembrane space (IMS). The Mn 2ϩ -requiring mitochondrial intermediate peptidase (MIP) removes the remainder of the presequence (open box) to form the mature protein, and an iron-sulfur center is added to yield the mature holoprotein (Mature Fe/S). The relative timing of holoprotein formation and complex assembly have yet to be examined. trypanosome ISP receptor recognition or for transfer of the precursor to the import apparatus (Fig. 7, ATP?). The absence of processing observed when the matrix was depleted of ATP with oligomycin may reflect the inability of an ATP-dependent mitochondrial HSP-70 to translocate the amino terminus of the precursor into the matrix (Kang et al., 1990). Hwang et al. (1991) demonstrated that depletion of ATP from the yeast mitochondrial matrix led to the accumulation of a proteaseprotected import intermediate within the inner membrane. Cyr et al. (1993) proposed that the cleavage of the presequence of the transfer intermediate depended upon the stability of the interaction of the transfer intermediate with the inner membrane and upon the depth of presequence insertion into the inner membrane. We suggest that the trypanosome ISP precursor is arrested within the membrane in the absence of matrix ATP and that the 17-amino acid presequence does not sufficiently penetrate the inner membrane to expose the MPP cleavage site.
As predicted from the primary sequence, the T. brucei ISP precursor is processed to the mature form by a two-step mechanism similar to that observed in yeast and N. crassa (Priest and Hajduk, 1995;Fu et al., 1990;Hartl et al., 1986). Using divalent metal ion chelators with different membrane permeabilities, we were able to show that the two ISP processing activities are located in different compartments of the mitochondrion. The peptidase that cleaves the precursor ISP to the intermediate form was inhibited by o-phenanthroline and EDTA but not by the water-soluble chelators alone, suggesting a matrix or inner membrane location. Since the activity of this peptidase was restored by manganese, we believe that this peptidase is the equivalent of the matrix-processing peptidase (Fig. 7, MPP) found in the matrix of other eukaryotes (Ou et al., 1989;Bohni et al., 1983;Hawlitschek et al., 1988).
The processing activity that cleaves the intermediate ISP form to the mature form appears to be the equivalent of the mitochondrial intermediate peptidase found in yeast and N. crassa, but it is uniquely located in the trypanosome (Fig. 7,  MIP). The sensitivity of this peptidase to inhibition by water soluble metal ion chelators such as EDTA and bathophenanthroline disulfonic acid indicates a location outside the mitochondrial inner membrane. This is in contrast to the matrix or inner membrane location of the mitochondrial intermediate peptidases that cleave the yeast and N. crassa ISPs (Fu et al., 1990;Ramabadran and Beattie, 1992;Isaya et al., 1994). The metal ion requirements of the trypanosome peptidase (Mn 2ϩ Ͼ Ca 2ϩ Ͼ Mg 2ϩ ) are similar to those reported for the yeast MIP (Ca 2ϩ Ͼ Mn 2ϩ ; Mg 2ϩ ineffective) and distinct from those of the inner membrane protease I that is normally found in the yeast intermembrane space (Mg 2ϩ Ͼ Ca 2ϩ Ͼ Mn 2ϩ ) (Ramabadran and Beattie, 1992;Nicholson et al., 1989). Since the sequence of the trypanosome cleaved presequence is also consistent with the octapeptide motif recognized by the yeast MIP, we suggest that the trypanosome peptidase is a relocated form of the MIP (Priest and Hajduk, 1995;Isaya et al., 1992).
The presence of the two ISP-processing peptidases on opposite sides of the mitochondrial inner membrane is consistent with import along both the conservative import pathway and the stop-transfer pathway (Hartl et al., 1987;Glick et al., 1992aGlick et al., , 1992b. In Fig. 7 we propose that the trypanosome ISP, like the yeast and N. crassa ISPs, is imported along the "conservative" pathway; therefore, the ISP is shown completely transported into the matrix prior to its reinsertion into the inner membrane (Hartl et al., 1986;van Loon and Schatz, 1987). While we cannot specifically rule out a stop-transfer type pathway similar to that described for the yeast cytochromes c 1 and b 2 , the unusually short length of the trypano-some ISP cleaved presequence makes this pathway less likely. The 61-and 80-amino acid cleaved presequences of the yeast cytochromes c 1 and b 2 , respectively, contain 19 and 14 residue regions of hydrophobic and uncharged amino acids that have been implicated in the stop-transfer mechanism (Glick et al., 1992a). These stop-transfer signals are proposed to lie entirely within the Ͼ38-residue (c 1 ) and 48-residue (b 2 ) presequence fragments that remain on the intermediate forms of the cytochromes after MPP cleavage (Glick et al., 1992a). The longest such region in the presequence of the T. brucei ISP contains 12 amino acids, but only about 7 residues of the region would remain after MPP cleavage. This estimate is based on the presence of the octapeptide MIP recognition sequence in the trypanosome presequences (Hendrick et al., 1989;Priest and Hajduk, 1995). Similarly, the presequence of the closely related C. fasciculata ISP would contain a region of only four hydrophobic and uncharged amino acids following MPP cleavage (Priest and Hajduk, 1995). Thus, these proteins probably do not contain conventional stop-transfer signals in their presequences. The presence of a stop-transfer sequence in the mature portion of the ISP is also unlikely, since that would leave the entire presequence exposed to the matrix and inaccessible to the second processing peptidase.
The location of the ISP intermediate peptidase outside the inner membrane of the trypanosome mitochondrion is difficult to reconcile with the expected orientation of the protein within the membrane. The ISP is usually anchored in the exterior face of the inner membrane by a domain that is predicted to form a membrane-spanning, hydrophobic ␣-helix (Graham et al., 1992). The MIP cleavage site (located near the amino terminus of this domain) would ordinarily be exposed to the matrix side of the inner membrane where the MIP could process it. Since the trypanosome MIP is located outside the inner membrane, this domain must be oriented differently for the cleavage site to be exposed at the correct membrane surface (Fig. 7). Thus, the ISP may use an alternative mechanism of insertion into the inner membrane, and the location of the ISP intermediate peptidase may simply be an adaptation required by this insertion pathway. Although the ISP insertion pathway has not been well characterized, we believe it is distinct from the one described for the conservatively sorted N. crassa subunit 9.
Because ISP processing appears to occur sequentially on opposite sides of the inner membrane and in the absence of a pH gradient, we believe that ISP insertion into the inner membrane is also independent of a pH gradient across the membrane. In contrast, Rojo et al. (1995) reported that dissipation of the pH gradient with CCCP prevented the inner membrane insertion of a matrix-localized subunit 9 fusion protein.
Another subunit of the cytochrome c reductase complex, cytochrome c 1 , also appears to utilize an unusual insertion pathway in trypanosomes. This protein lacks a conventional mitochondrial targeting presequence and has a heme binding site sequence that can form only one thioether linkage with the prosthetic group (Priest and Hajduk, 1992;Priest et al., 1993). Preliminary in vitro import results suggest that import of the cytochrome c 1 is largely independent of the mitochondrial membrane potential (Priest and Hajduk, unpublished observations). This raises the possibility that, rather than a conventional stop-transfer or conservative sorting pathway, the trypanosome cytochrome c 1 protein is using a nonconservative pathway to the intermembrane space similar to that described for the N. crassa cytochrome c or cytochrome c heme lyase (Nicholson et al., 1987;Lill et al., 1992). If this is indeed the case, the cytochrome c 1 would enter the inner membrane from the intermembrane space, and the complex assembly mechanism would have to be modified to accept the protein from this location. Since the cytochrome c 1 and the ISP must interact directly in the reductase complex, we think it possible that both proteins are assembled into the complex differently in trypanosomes than in other eukaryotes.
The ability to study the import of the T. brucei mitochondrial proteins in vitro represents a significant advance toward the understanding of the regulation of mitochondrial function. In addition to the studies on the specific import pathways that are involved in the coordinate assembly of a multisubunit complex, it may also be possible to determine the role of mitochondrial protein import in the developmental cycle of T. brucei. The two biochemically distinct life cycle forms of this organism developmentally regulate the levels of cytochromes and other mitochondrial proteins involved in energy metabolism (reviewed in Priest and Hajduk (1994b)). The bloodstream form lacks cytochromes, while the procyclic form has fully functional electron transport. Since at least one of the cytochromes is translated and rapidly turned over in the bloodstream form, we have postulated that regulation occurs at the point of mitochondrial protein import Priest and Hajduk, 1994b). We are currently working to adapt the mitochondria isolation protocol to the bloodstream form so that the import competence of the mitochondria for the cytochromes can be assayed.