A Proposed Common Structure of Substrates Bound to Mitochondrial Processing Peptidase*

,

Numerous mitochondrial proteins are translated on cytoplasmic ribosomes as larger precursors. An N-terminal presequence of the mitochondrial protein precursor functions as a targeting signal for their transport to mitochondria (1)(2)(3). During import of precursors into mitochondria, the presequences are recognized by multiple proteins (4,5), such as molecular chaperones, translocases of the mitochondrial outer and inner membranes, and peptidases from inside mitochondria. Despite the identification of various proteins that interact with the mitochondrial precursors, the mechanism of recognition of the presequence by these components has not yet been understood.
The lack of sequence homology of the presequences, even though they are characterized by the positively charged residues and the formation property of amphiphilic ␣-helices, has inhibited clarification of the recognition mechanism (6).
Mitochondrial processing peptidase (MPP), 1 located in the matrix of the mitochondria, cleaves off most presequences of the imported precursors. MPP consists of two structurally related subunits, ␣-MPP and ␤-MPP. Complex formation with the two subunits is essential for both enzymatic activity (7,8) and substrate binding (9).
Earlier studies indicated that some structural elements of the presequence are required for recognition by MPP. An arginine residue at position Ϫ2, the so-called "proximal arginine," from the cleavage site, which is usually found among most precursor proteins, plays a critical role in cleavage reaction (10 -12). Distal basic amino acid residue(s) around position Ϫ10 are also important for effective cleavage (10 -12). The length between the proximal arginine and the distal basic residues is not so fixed, and 4 -10 amino acids are allowed (13). Our more recent studies have shown a requirement for effective cleavage of flexible linker sequences containing proline and glycine between the two basic residues (13,14), a hydrophobic residue at position ϩ1 (12,13), and serine or threonine residues at position ϩ2 and/or ϩ3 (12,15).
Some functional amino acid residues in MPP were determined using mutational analysis; His-101, Glu-104, and His-105 in rat ␤-MPP, which form a metal binding site, HxxEH, conserved among a pitrilysin metallopeptidase superfamily (16), are the catalytic center of MPP (17,18). Glu-181 is the third metal-binding residue (19). Glu-174 may participate in the catalytic reaction (18). Glu-124, which is in a characteristic acidic amino acid cluster conserved in ␤-MPP, may interact with the N-terminal portion of the presequence in the cleavage reaction (18). On the other hand, Glu-390 and Asp-391 in yeast ␣-MPP interact with the distal arginine residues, which are required especially for cleavage of precursors with a longer presequence (19). Deletion of three residues in the glycine-rich segment characteristic of ␣-MPP resulted in a drastic reduction in affinity to the substrate (20). 2 Findings on functional amino acid residues both in precursor 1 The abbreviations used are: MPP, mitochondrial processing peptidase; CPM, 7-diethylamino-3-(4Ј-maleimidylphenyl)-4-methylcoumarin; DAC, 7-diethyl aminocoumarin-3-carbonic acid; FRET, fluorescence resonance energy transfer; IANBD, N,NЈ-dimethyl-N-(iodoacetyl)-NЈ-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)ethylenediamine; MDH, malate dehydrogenase; ␣and ␤-MPP, ␣and ␤-subunits, respectively, of the mitochondrial processing peptidase. 2 In our previous works, the residues of MPP were numbered from the N terminus of the mature protein reported in the data base. In this studies, we numbered the residues according to the full-length MPP precursors. For instance, His-101, Glu-104, and His-105 in rat ␤-MPP were represented as His-56, Glu-59, and His-60, respectively, in our previous papers. proteins and MPP required for the processing reaction, especially for precursor recognition by MPP, suggest that the two subunits of MPP cooperatively form the substrate binding pocket and that they have several substrate binding sites to cope with different structural elements in the extension peptide. To elucidate the recognition mechanism that makes feasible strict substrate specificity for MPP, it is necessary to determine the structure of the presequence bound to the enzyme.
In the present study, fluorescence resonance energy transfer (FRET) experiments provide the first evidence that the distal arginine and the portion around the cleavage site of the presequence are located at specific sites in the MPP molecule, irrespective of the position of the distal arginine. An induced-fit mechanism of substrate recognition of MPP seems likely.
Preparation of Fluorescence-labeled MPP-A hexahistidine-tagged yeast ␣-MPP and yeast ␣/␤E73Q complex were purified as described (9). Purification and fluorescent labeling of a hexahistidine-tagged yeast ␤E73Q were done as follows. The supernatant from the BL21(DE3) strain carrying pET-␤E73QHis was loaded on a 5-ml Hi-trap chelating column (Amersham Pharmacia Biotech) equilibrated with buffer A (20 mM Hepes-KOH, pH 7.4, containing 500 mM NaCl). The column was washed with 50 ml of buffer A containing 50 mM imidazole. The ␤-MPP was eluted with buffer A containing 200 mM imidazole. To the fractions containing ␤-MPP were added 0.1 mM CPM, and then the sample was left to react for 1 h at 25°C. The reaction was terminated using 1 mM cysteine for 30 min at 25°C. The free dye was removed on a PD-10 desalting column equilibrated with 20 mM Hepes-KOH, pH 7.4, containing 200 mM NaCl. Labeled ␤-MPP was then applied onto 2 ml of Q-Sepharose FF (Amersham Pharmacia Biotech) equilibrated with 20 mM Hepes-KOH, pH 7.4, containing 20 mM NaCl. The column was washed with 30 ml of the same buffer, and then the protein was eluted with 20 mM Hepes-KOH, pH 7.4, containing 100 mM NaCl. The purity of CPM-labeled ␤-MPP was confirmed by SDS-polyacrylamide gel electrophoresis followed by UV transillumination and Coomassie Blue staining. The labeling efficiency was calculated from a molar extinction coefficient of 33,000 (M Ϫ1 cm Ϫ1 ) for CPM. The labeling procedure resulted in the incorporation of 0.11 Ϯ 0.01 mol of CPM/mol of ␤-MPP. The CPM-labeled ␤-MPP was mixed with an equal mol amount of the purified ␣-MPP, and intermolecular FRET measurements were made.
Fluorescence Measurements-Fluorescence was measured at 25°C using a Hitachi F-4500 fluorescence spectrophotometer. Steady-state fluorescence anisotropy measurements were performed with a Hitachi F-4500 fluorescence spectrophotometer equipped with automatic fluorescencce polarization system. In the intramolecular FRET measurement of the double-labeled peptide, excitation of DAC was measured at 390 nm and the emission intensity at 470 nm. MPP (0.5 M) was diluted into 20 mM Hepes-KOH, pH 7.4, containing 30% glycerol, and then various concentrations of the double-labeled peptides were added. The emission spectra were taken after each sample and the blank had been thoroughly mixed and allowed to equilibrate for 1-2 min. In the intermolecular FRET, the excitation of CPM was measured at 390 nm and the emission intensity at 480 nm. To the IANBD-labeled peptides (0.5 M), CPM-labeled MPP was added at the indicated concentrations, and then the emission spectra were taken, and the fluorescence intensity at 480 nm was read.
The dissociation constant, K d , was determined as follows: where Q D stands for the unquenched quantum yield of the donor and Q DA is the quantum yield in the presence of the acceptor. Quantum yield was substituted for emission maximum intensity F. From the energy transfer efficiency results, the distance between donor and acceptor was calculated according to the Förster theory (21), where R is the calculated distance and R 0 is the distance at which 50% energy transfer would occur between the donor-acceptor pair; it is given in angstroms, as shown in Equation 3, where J, the overlap integral, is the degree of spectral overlap of donor emission F D () and acceptor absorbance ⑀ A (), as defined by Equation 4.
2 is assumed to be 2/3. The refractive index of the solvent, n, is used at a value of 1.4. Q D , the quantum yield for the donor, was given as where Q R is the quantum yield for the reference dye, F D and F R are the fluorescence intensities for the donor and reference dye, respectively, and A D and A R are the fluorescence intensity for the donor and reference dye, respectively. Fluorescein was used as the reference dye and was assumed to have a quantum yield of 0.92 in 0.1 N NaOH. Although 2 was taken as 2/3 for the calculation of distances, the maximum and minimum values of 2 were estimated according to the method of Dale et al. (21).
; r D and r A are the limiting anisotropies of the donor and acceptor, respectively. Using these values for the orientation factor, the maximum and minimum distances between probes were calculated and were regarded as the probable error limits of the distance (R-limits).

Distance between the N-terminal End and the C-terminal Portion of the Peptide Substrates Bound to MPP-Sequence
data on the presequence of mitochodrial protein precursors show that position of the distal basic acid is not so fixed among the extension peptides and is located Ϫ7 to Ϫ17 from the cleavage site. To elucidate the structure of the substrate peptide bound to the enzyme, it is vital to determine whether the distal basic amino acids at various positions interact with the same amino acid(s) in the enzyme, because the proximal arginine must be fixed with certain amino acid residues near the active center in ␤-MPP. In attempting this elucidation, we measured the distance between the distal arginine and the residue around the cleavage site, using intramolecular FRET. We synthesized a series of peptides with different lengths of intervening sequence between the proximal and distal arginines, labeling them at the N-terminal end with fluorescent donor DAC and at position ϩ4 cysteine with the acceptor IANBD amide (Table I). These sites were two and five amino acid residues away from the distal and proximal arginines, respectively, to avoid interference by probes in the interaction between the arginines and other recognition elements of the peptide and MPP. In MDH-14A, arginine at position Ϫ3 (position 14 from the N-terminal end) was replaced with alanine in the original sequence of rat MDH, as it is not necessary for the residue at this position to be arginine (10). This peptide has a distal arginine at position Ϫ10 and is used as a standard peptide in the present work. In MDH-⌬AAL, three intervening amino acid residues between proximal and distal arginines were deleted so that the distal arginine was at position Ϫ7. This peptide was found to have the minimum length for effective cleavage by the MPP (13). MDH-AdAR and MDH-AdRA have the distal arginine residue at position Ϫ14 and Ϫ17, respectively. In these peptides, the linker sequence in the presequence of bovine adrenodoxin precursor was introduced between the proximal and distal arginines instead of the original one of the MDH presequence.
The emission maximum of DAC is at 470 nm (Fig. 1). The absorption spectrum of IANBD, which gives the peak at 498 nm, has an excellent overlap with the emission spectrum of DAC (data not shown). The spectral overlap, J, between these spectra is calculated to be 9.31 ϫ 10 Ϫ14 M Ϫ1 cm Ϫ1 nm 4 . The spectral characteristics of all the peptides were essentially the same. For the intermolecular FRET measurements, fixed concentrations (0.5 M) of the DAC or DAC/IANBD-labeled peptides were added to various concentrations (usually 0 -3 M) of the purified ␣/␤E73Q, in which ␤-MPP is an inactive mutant with glutamine substituting for the glutamate residue of the active center (18) (Fig. 2). As demonstrated in our previous study (9), the fluorescence of DAC introduced to the peptide substrate bound to MPP increases with environmental change around the dye (Fig. 1). Titration of the DAC-labeled peptides gave the dissociation constant, K d , of peptides for binding to MPP (Fig. 2). All of the peptides bound to MPP with a high affinity to the same extent.
Introduction of IANBD into the peptides led to a drastic suppression of increase in DAC fluorescence through FRET in all of the peptides studied (Fig. 2). In the titration of DAC-or DAC/IANBD-labeled peptides with MPP, the fitting curves showed a biphasic nature (Fig. 2). Because fluorescence anisotropy change of the peptides was saturated at a stoichiometric amount of the enzyme (data not shown), the biphasic fluorescence change in the titration experiments might be due to increased scattering by increasing the concentration of the enzyme. For calculation of FRET efficiency, E, the fluorescence intensities, F D and F DA for DAC-and DAC/IANBD-labeled peptides, respectively, were taken by extrapolating the second phase curve to the ordinate. The calculated FRET efficiencies of all the peptides showed a range of 80 -85% (Table II). The quantum yield, Q, of the fluorescence donor DAC showed a gradual decrease with the increasing length of the peptides; this also resulted in a decrease in the distance at which 50% FRET occurred between the donor-acceptor pair, R 0 , suggest- ing that the mobility of the N-terminal portion of the peptides bound to MPP increases with length of the presequence. Taken together with the parameters described above, the distances of the donor-acceptor pair, R, were substantially the same among all of the peptides measured and were calculated to be ϳ28 Å (Table II); this indicates that the distance between the Nterminal end of the peptides and cysteine residue at position ϩ4 is fixed, irrespective of the length of the intervening sequence between the proximal and distal arginines. Our findings suggest that the distal basic amino acid interacts with a specific site, probably the acidic residue(s), in the enzyme.

Location of Putative Sites in MPP Interacting with the Nterminal and C-terminal Portions of the Substrate Peptide-
The similarity of the calculated distances between donor-acceptor pairs on the substrates might be because of compensation for changes in distance by changes in orientation of the fluorescent dyes. To eliminate this possibility, we next carried intermolecular FRET measurements between a fluorescence donor in the ␤-MPP and the acceptor in the substrate peptides bound to MPP. This measurement could also estimate the location in the MPP of the N-terminal end of the peptide substrate and the amino acid residue at ϩ4 position from the cleavage site. We first modified a single cysteine residue, Cys-252, in yeast ␤-MPP (␤E73Q) by CPM, as the fluorescent donor. The histidine-tagged ␤E73Q partially purified by nickel-chelating column chromatography was reacted with CPM (Fig. 3,  lane 1). When CPM-labeled ␤E73Q was purified further with Q-Sepharose chromatography (Fig. 3, lane 3), a single band was obtained by both Coomasie Blue staining and UV transillumination. To confirm the specificity of the labeling to the cysteine residue, the labeling with CPM was done with a mutant enzyme, ␤E73Q/C252S, in which the single cysteine residue was substituted to serine. No visible band in UV transillumination was obtained with the mutant enzyme (Fig. 3, lanes 2 and 4), confirming the specific labeling with CPM of Cys-252 in ␤-MPP.
A series of peptides labeled with IANBD amide at the Nterminal end or at position ϩ4 was synthesized (Table III). For labeling with IANBD at the N-terminal end of the peptides, a cysteine residue, instead of leucine, was introduced at the N terminus. The ␣-amino group of the N-terminal end of the peptides was acetylated to avoid the effect of the positive charge of the ␣-amino group in the peptides. In intermolecular FRET measurements, titration of the fixed concentration (0.5 M) of CPM-labeled ␣/␤E73Q was done with various concentrations (typically 0 -1.5 M) of the IANBD-labeled peptides. The emission spectrum of CPM, which has a fluorescence maximum at 475 nm, is similar to that of DAC, and overlaps the excitation spectrum of IANBD amide (J ϭ 9.68 ϫ 10 Ϫ14 M Ϫ1 cm Ϫ1 nm 4 ). The quantum yield of CPM was extremely high (0.90 Ϯ 0.02) relative to that of DAC. The addition of the acceptor-labeled peptides led to a decrease in the fluorescence of CPM and a small increase in that of IANBD, which has a peak at 545 nm (Fig. 4). Because FRET can be detected when donor and acceptor molecules are in close proximity (typically 10 -100 Å) and fluorescence quenching by collision between the donor-acceptor pair was negligible, under the conditions of this measurement (data not shown), the observed decrease in fluorescence indicates binding of the IANBD-labeled peptide to the CPM-labeled MPP molecule. The absorption spectra of IANBD did not change with binding to MPP. From the spectral parameters obtained, the Förster distance, R 0 , was calculated to be 48.6 Ϯ 0.3 Å. The titration curves were shown in Fig. 5 as a plot of 1 Ϫ (F DA /F D ) versus [Pep], where [Pep] is the concentration of the IANBD-labeled peptides. The K d values calculated from the nonlinear least-squares fit of the data increased slightly relative to those of DAC-labeled peptides. In both cases of the peptides labeled at the N-terminal end and at position ϩ4, the FRET efficiency, E, which is calculated as the limitation value of 1 Ϫ (F DA /F D ), showed essentially the same value among the peptides that have distal arginine at different positions (Table  IV). Distances between CPM in ␤-MPP and IANBD at position ϩ4 of the peptides were calculated to be in the range of 44 to 46 Å, whereas those between CPM in MPP and IANBD at the N terminus of the peptides were in the range of 43 to 45 Å, irrespective of the position of the distal arginine residue. The finding showing that the distance between the cysteine residue in ␤-MPP and the N terminus of the peptides is substantially the same among the peptides with various lengths of intervening sequences between the proximal and distal arginines confirmed that the distal basic amino acid residue interacts with the specific site of the enzyme if the distal basic residue is at least within Ϫ7 to Ϫ17 of the cleavage site. DISCUSSION We found that when mitochondrial protein precursors are bound to MPP, distal basic amino acids in its presequences interact with the specific site in the enzyme if the basic residues are present at positions Ϫ7 to Ϫ17. This means that the intervening sequence between the proximal arginine and the

TABLE II
Distance between DAC at the N terminus and IANBD at position ϩ4 in the various peptide substrates bound to MPP The amino acid sequences of the peptides are shown in Table I. K d , dissociation constant; Q, quantum yield; R 0 , distance at which 50% FRET would occur between the donor-acceptor pair; E, FRET efficiency; R, calculated distance between the donor-acceptor pair; R-limits, probable error limits of the distance including the orientation factor. See "Experimental Procedures" for the calculation of these parameters. distal basic amino acid is flexible so that the distal basic residue can fit into a specific binding site in MPP. Thus, the present study is the first to propose the structure of the presequence bound to MPP. We generated an energy-minimized model of rat MPP (22) based on the crystal structure of the core proteins in the bovine bc1 complex (23,24). The two subunits form a ball with a crack leading to the internal cavity. Functional amino acid residues predicted in our previous works (17)(18)(19) are arranged around the cavity. In the simulation model (Fig. 6), the distance between Glu-73 in yeast ␤-MPP (corresponding to Glu-104 in rat ␤-MPP), which is a catalytic center (17,18), and Glu-390/Asp-391 in yeast ␣-MPP (corresponds to Glu-446/Asp-447 in rat ␣-MPP), which we assumed to be residues interacting with the distal arginine in the presequence (19), was calculated to be about 30 Å. The glycine-rich loop in ␣-MPP, which is conserved among different organisms and has been shown to be essential for MPP function, is close to the metal binding active center in ␤-MPP and is located about 30 Å from Glu-390/Asp-391 in ␣-MPP. These values are close to those (about 28 Å) obtained from FRET measurements between the N-terminal end and the amino acid residue at position ϩ4 of the peptides (Table II) amino acid residue at position ϩ4 of the peptide interact with the acidic residue cluster and the glycine-rich region in ␣-MPP, respectively, and that the substrate peptide in the cavity of MPP is needed to form the loop structure, depending on the length of the intervening sequence (Fig. 6).
In the intervening sequence between the proximal arginine and the distal basic amino acid in the presequence, several prolines and/or glycines are usually present, and substitution of these residues for alanine led to a reduction in cleavage efficiency (13). Insertion of ethoxy linkage instead of peptide bonds, as a more flexible linker, between the distal and proximal arginines led to a 2-fold increase in cleavage efficiency compared with that of the control peptide (14). Studies using two-dimensional proton NMR combined with circular dichro-ism on synthetic peptides corresponding to the presequences of the precursor proteins have also demonstrated that peptides to be cleaved by MPP have the potential ability to form a helixlinker-helix structure, in which glycine and/or proline residues serve as an ␣-helix-breaking linker (25)(26)(27). Taking those findings together with the present results, one could surmise that the intervening sequence forms a flexible loop structure and function to aid structural elements, including the distal basic and proximal arginine residues, in binding to multiple subsites in MPP.
During import into mitochondria, multiple proteins, including molecular chaperones, receptors, and processing peptidases, recognize the presequences of mitochondrial protein precursors. The formation of an ␣-helix of the presequence is required apparently for interaction with these components (25)(26)(27). The NMR structure of the cytosolic domain of Tom20, a component of the translocase complex in mitochondrial outer membrane, with a synthetic peptide based on the aldehyde dehydrogenase precursor has recently been resolved (28). The peptide that forms an amphiphilic ␣-helix in a crack of the core structure of Tom20 consists of four helices. The present results indicate that structures required for targeting and processing differ and that a flexible structure is required for the processing, although basic residues in the presequences function as recognition signals for both processes.
Structural convergence between MPP and thermolysin, a Zn 2ϩ -peptidase with a typical metal-binding motif, HExxH, has been discussed recently (29). Superimposition between the N-terminal domain of core 1 protein of the bc1 complex and the portion around the active site of thermolysin showed a similar arrangement of secondary structural elements but with different topological connections and in a reverse main chain orientation. This structural architecture is based on four helices, which contain metal ligands and the catalytic glutamate residue, and the neighboring five strands of a ␤-sheet. A main chain of substrates of thermolysin interacts with the edge of the ␤-sheet through hydrogen bonds. Like thermolysin, the ␤-sheet structure around the active center of MPP might interact with a nonhelical structure around the cleavage site of the precursors through hydrogen bonding, to present the scissile bond of the substrate to the active center in ␤-MPP. Further studies, especially on the crystal structure of MPP, should reveal the precise structure of the presequence bound to MPP and the mechanisms of strict recognition and specific cleavage of precursor proteins by the enzyme.  Table III. The quantum yield of DAC, Q, and the distance at which 50% FRET would occur between the donor-acceptor pair, R 0 , were calculated to be 0.90 Ϯ 0.02 and 48.6 Ϯ 0.3 Å, respectively. K d , dissociation constant; E, FRET efficiency; R, calculated distance between donor-acceptor pair; R-limits, probable error limits of the distance including the orientation factor. See "Experimental Procedures" for the calculation of these parameters. FIG. 6. Predicted conformation of the peptide substrate bound to MPP. Shown is the model of MPP, based on the core proteins of the bovine bc1 complex, as described (24). ␣and ␤-MPP are shown in green and blue, respectively. The bold line indicates the predicted conformation of peptide substrate bound to MPP. N and C indicate the amino and carboxyl terminus of the substrate, respectively. The red arrow indicates the cleavage site of substrate by MPP. Glu-390 in yeast ␣-MPP, which corresponds to Glu-446 in rat ␣-MPP, and Glu-73 and Cys-252 in yeast ␤-MPP, which correspond to Glu-104 and Thr-280 in rat ␤-MPP, respectively, are shown in the space-filled presentation. The glycinerich segment in ␣-MPP is shown in yellow. These residues and segment are described in detail under "Discussion".