Glycine-rich Region of Mitochondrial Processing Peptidase α-Subunit Is Essential for Binding and Cleavage of the Precursor Proteins*

Mitochondrial processing peptidase, a metalloendopeptidase consisting of α- and β-subunits, specifically recognizes a large variety of mitochondrial precursor proteins and cleaves off amino-terminal extension peptides. The α-subunit has a characteristic glycine-rich segment in the middle portion. To elucidate the role of the region in processing functions of the enzyme, deletion or site-directed mutations were introduced, and effects on kinetic parameters and substrate binding of the enzyme were analyzed. Deletion of three residues of the region, Phe289 to Ala291, led to a dramatic reduction in processing activity to practically zero. Mutation of Phe289, Lys296, and Met298 to alanine resulted in a decrease in the activity, but these mutations had no apparent effect on interactions between the two subunits, indicating that reduction in processing activity is not due to structural disruption at the interface interacting with the β-subunit. Although the mutant enzymes, Phe289Ala, Lys296Ala, and Met298Ala, had an approximate 10-fold less affinity for substrate peptides than did that of the wild type, the deletion mutant, Δ289–291, showed an extremely low affinity. Thus, shortening of the glycine-rich stretch led to a dramatic reduction of interaction between the enzyme and substrate peptides and cleavage reaction, whereas mutation of each amino acid in this region seemed to affect primarily the cleavage reaction.

Most mitochondrial proteins encoded in the nucleus are synthesized as precursor proteins with extension peptides at the amino termini and are targeted to the mitochondria. After import of the precursors into the mitochondria, the extension peptides are proteolytically cleaved off by a matrix-located metallopeptidase, mitochondrial processing peptidase (MPP) 1 (1)(2)(3)(4). Purification of MPP from various species revealed that this peptidase is a heterodimer consisting of larger (␣-MPP) and smaller (␤-MPP) subunits (5)(6)(7)(8)(9)(10). Sequence analysis (11)(12)(13)(14)(15)(16) revealed that subunits of MPP have significant sequence homology with a family of endopeptidases, the pitrilysin family, that includes Escherichia coli pitrilysin (also called protease III), the insulin-degrading enzymes from mammals and insects, and the N-arginine dibasic convertase from the rat (17,18). All members, except the ␣-MPPs, have a metal-binding motif, His-X-X-Glu-His. Site-directed mutagenesis showed that this site is responsible for the catalytic activity of MPP (19,20). The ␣and ␤-subunits of the MPP are also homologous to core 2 and core 1 proteins, respectively, of the mitochondrial ubiquinol-cytochrome c oxidoreductase (bc 1 complex), a component of the respiratory chain (21). Plant MPP was purified as subunits of the bc 1 complex located in an inner membrane (9. 10). In Neurospora crassa, a core 1 protein of the bc 1 complex is identical with ␤-MPP in the matrix (22).
For correct recognition of the cleavage site, MPP requires information located within the extension peptides of the precursor proteins. Contrary to the strict substrate specificity of the enzyme, amino acid sequences of the extension peptides have a broad range of lengths with little similarity (23). In the extension peptide, several basic amino acids intervene into a long sequence of neutral amino acid residues, and negatively charged residues are rarely present (24,25). Studies using radiolabeled precursors and synthetic peptides as substrates demonstrated that an arginine residue at Ϫ2 position and basic amino acid residues distal to the cleavage site are essential for specific cleavage by MPP (26 -30). Aromatic and bulky hydrophobic amino acids at position ϩ1 and flexible linker regions containing proline and/or glycine between the two basic residues are required (28,30). Amino acids at position ϩ2 and ϩ3 are also important for the specific recognition and cleavage reaction by MPP (31).
Although ␤-MPP has a catalytic center containing a metal responsible for the reaction, the subunit alone has no apparent activity, and cooperative actions of the two subunits are essential for MPP activity (32)(33)(34). ␣-MPP seems to participate in functions other than catalytic ones. Cross-linking analyses suggested that a substrate-binding domain might reside in ␣-MPP (35,36). Indeed, mutation experiments revealed that acidic amino acids in the carboxyl-terminal portion of ␣-MPP are involved in substrate recognition and are mainly responsible for binding of distal basic amino acids of longer extension peptides (37). Our recent results further showed that both subunits of MPP are required for formation of a substrate binding pocket, and different parts of the extension peptide of precursor protein interact with individual subunit at the interface (38).
Sequence data demonstrate that there is a glycine-rich segment in the middle portion of ␣-MPP and that it is characteristic of this subunit and is absent in ␤-MPP and other members of the pitrilysin family (11)(12)(13)(14)(15)(16)(17)(18). Core 2 protein of the fungal or animal mitochondrial bc 1 complex that corresponds to ␣-MPP has a short stretch of glycine-rich segment. Crystal structures of the bc 1 complex from bovine and chicken heart mitochondria have been reported (39 -41); hence, one can expect that MPP is similar in structure to the complex. The core 1 and core 2 proteins, which are structurally similar, consist of two domains of roughly equal size and almost identical folding topology. The two components representing these proteins were described as coming together in the form of a ball with a crack leading to internal cavity. Based on available data, the middle portion of core 2, which corresponds to the glycine-rich region of ␣-MPP conserved among many species, is located at the face toward the core 1 in the cavity. These findings suggest the possible involvement of segment characteristics of ␣-MPP in dimer formation, substrate recognition, or catalytic reaction of MPP.
It thus seemed important to determine the role of the glycine-rich segment of ␣-MPP in the processing reaction. Several amino acid residues in the segment were mutated, and effects on kinetic parameters and substrate binding of the mutated enzymes were analyzed. Deletion of some amino acids in the segment led to an almost complete loss of the activity, without affecting formation of a stable heterodimeric complex. From these results, together with a simulation model calculated from the reported crystallographic data on the core proteins (38 -40), the possible role of the glycine-rich segment in the processing of precursor proteins by MPP is discussed.

Construction of Expression Vectors and Site-directed Mutation of
␣-MPP--A hexahistidine tag was introduced into the carboxyl termini of yeast ␣-MPP and rat ␤-MPP (42), using polymerase chain reaction (PCR). The resultant cDNAs were finally inserted into a multiple cloning site of pTrc99A (Amersham Pharmacia Biotech), leading to pTrcy␣-His and pTrcr␤-His. To introduce mutations in ␣-MPP, site-directed mutagenesis was done using PCR, and the mutations were confirmed by DNA sequencing.
Expression and Purification of Recombinant MPP-E. coli BL21 (DE3) cells transformed with each plasmid were cultured in LB medium containing 0.1 mM isopropyl ␤-D-thiogalactopyranoside for 24 h at 25°C. After harvesting the cells by centrifugation for 10 min at 1,000 ϫ g, the cells were suspended in 10 mM Tris-HCl buffer, pH 7.5, containing 10 mM MgSO 4 , 50 mM KCl, and 10% glycerol, then lysed by sonication, followed by solubilization with the addition of Tween 20 to the final concentration of 0.1%. The soluble extract was recovered by centrifugation for 20 min at 15,000 ϫ g. The extract was loaded on a nickel-chelating Sepharose column (Amersham Pharmacia Biotech; His-trap) equilibrated with 20 mM HEPES-KOH buffer, pH 7.5, containing 200 mM NaCl, 50 mM imidazole, 30% glycerol, 0.1% Tween 20. After washing the column with the buffer, histidine-tagged MPP subunits were eluted with 500 mM imidazole. The purity was confirmed by SDS-polyacrylamide gel electrophoresis.
Determination of MPP Activity-The purified wild-type or mutant ␣-MPP was mixed with purified ␤-MPP, and then activity of the in vitro reconstituted MPP was measured using real time fluorometric assay, as described by Ogishima et al. (28). A fluorogenic peptide, aminobenzoyl-LARPVGAALRRSFSTY(NO 2 )AQNN, synthesized based on the extension peptide of mouse malate dehydrogenase (MDH), was used as the substrate. Excitation was at 315 nm, and emitted light was measured at 420 nm. Fluorometry was done using a Hitachi F-2000 fluorescence spectrophotometer at 25°C.
Kinetic parameters were determined using as substrate MDH1-21, a synthetic peptide corresponding to the first 21 amino acid residues of MDH precursor (27). The reconstituted MPP was incubated with the synthetic peptide in 440 l of 20 mM HEPES-KOH buffer, pH 7.4, containing 0.1% Tween 20 at 25°C. The reaction was stopped by adding 0.05% trifluoroacetic acid or 0.4% trifluoroacetic acid and 200 mM NaCl. The processing products were analyzed by reversed phase high pressure liquid chromatography, as described (27). K m and k cat values for MPPs were determined from Lineweaver-Burk plots.
Association between ␣and ␤-MPPs-Yeast ␤-MPP cDNA was inserted into the tetracycline-resistant region of pACYC184 (Nippon Gene). pTrcy␣-His and pACYCy␤ were co-expressed in BL21 strain. The cell extract was applied onto a nickel-chelating Sepharose column, and the adsorbed proteins were eluted with 0.5 M imidazole. The coeluted proteins were analyzed by SDS-polyacrylamide gel electrophoresis, as described by Kitada et al. (42).
Binding of Fluorescence-labeled Peptide to MPP-The MDH5-25 peptide was labeled at the amino-terminal ␣-amino group with 7-diethylaminocoumarin (DAC), N-DAC-MDH5-25, as described by Kojima et al. (38). The coumarin-labeled peptides of indicated concentrations added to the purified MPP were preincubated with 5 mM EDTA on ice for 30 min. Change in fluorescence was monitored at 470 nm (excitation at 390 nm) at 25°C using a Hitachi F-2000 fluorescence spectrophotometer. K d values were calculated as described (38).
Molecular Modeling-Modeling was performed using the Insight II/ Homology Software Package (Molecular Simulations, Inc., San Diego, CA). To build a model of MPP, core proteins cytochrome bc 1 complex (1BCC, 1BGY, and 1QCR) was used as reference protein. The structurally conserved regions (SCRs) were assigned on condition that the value of the root mean square deviations of segment backbone atoms were under 0.5 Å. The intervening regions were defined as variable regions (VRs). Coordinates of side chains in the SCRs were automatically established according to reference proteins. The side chains that differed from the template were mutated to the corresponding amino acids in such a way that conformation of the common area was retained whenever possible. In the VRs, coordinates of the side chains were also automatically established according to reference proteins and were generated using the Loop-search command in homology. To avoid unacceptable van der Waals contacts, all side chains were optimized using the Auto-rotamer command. To refine the model structure, the energy minimization calculations were done using the Discover program and the consistent valence force field with a distance-dependent dielectric constant. We carried out energy minimizations in a two-step procedure. (i) All VRs were minimized using the conjugated gradient method while the backbone atoms were tethered (50 kcal mol-1A-2) and all SCRs were fixed. (ii) All VRs and the side chains of the SCRs were minimized using the conjugated gradient method while the backbone atoms of all SCRs were tethered (1000 kcal/mol⅐Å 2 ). The two steps were carried out for 2000 iterations.

RESULTS
Mutation in the Glycine-rich Region in ␣-MPP-To investigate the role of the characteristic glycine-rich region in ␣-subunit, which is a highly conserved segment among ␣-subunits of various MPPs but not other members of the related proteins, including ␤-MPP and two core proteins (Fig. 1A, cf. Ref. 43), conservative residues in the region with a large side chain, Phe 289 , Lys 296 , Met 298 , Tyr 299 , Arg 301 and Leu 302 , were individually mutated to alanines or to similar amino acids, using PCR. Three, five, and eight amino acid residues at the center of this region, from Phe 289 to Ala 291 , Ser 288 to Gly 292 , and Gly 287 to Pro 294 , respectively, were deleted (Fig. 1B). Each mutant subunit was purified, and the role of these residues in substrate recognition and/or catalytic reaction was examined.
Cleavage of Fluorescence-labeled Peptide by Wild-type and Mutant MPPs-After purification of each subunit so as to be homogeneous with the nickel-chelating column, two subunits were mixed to reconstitute the whole enzyme, as described under "Materials and Methods". The processing activity of mutant MPPs was measured using as substrate a fluorogenic peptide (Fig. 2). No activity was detected in the enzyme deleted even three amino acid residues from Phe 289 to Ala 291 , ⌬289 -291. Phe289Ala, Phe289Leu, Lys296Ala, and Met298Ala also showed little activity, less than 1% of the wild-type MPP. On the other hand, other mutant proteins examined, including Met298Leu, Tyr299Ala, Tyr299Ser, Arg301Ala, Arg301Lys, and Leu302Ala, retained substantial activity, 40 -80% of the wild-type MPP. Differential effects of mutation on Met 298 to alanine and leucine indicate that the important feature of the methionine side chain is its hydrophobic character. Thus, Phe 289 , Lys 296 , and Met 298 and residues 287-294 seem to be primarily involved in MPP function, including subunit assembly, substrate binding, and/or catalysis.
Effect on Mutation of ␣-MPP on Association with ␤-MPP-Since both subunits of MPP are required for formation of a substrate binding pocket and different parts of the extension peptide of precursor protein interact with individual subunits at the interface (38), quantitative analysis of heterodimer formation between mutant ␣and native ␤-subunits must be sensitive monitoring to detect structural change near substrate binding and/or catalytic sites in the mutant enzymes. To determine whether loss of processing activity of the mutant enzymes, as shown in Fig. 2, would result in failure in association of the subunits, we examined the association of the histidinetagged wild-type and mutated ␣-subunits to non-tagged ␤-subunit on the nickel-chelating column (Fig. 3A). The histidinetagged ␣-subunit was co-expressed with the non-tagged ␤-subunit in E. coli; the bacterial extract was applied onto a nickel-chelating column, and the adsorbed proteins were eluted with 0.5 M imidazole. Nearly the same amount of non-tagged ␤-MPP was eluted together with the mutant ␣-MPP as well as with the wild-type one, indicating that mutation of the glycinerich region of ␣-MPP has no effect on interactions between the two subunits.
The same conclusion was derived from observations on quantitative analysis of the interaction between the mutant ␣-subunits and the wild-type ␤-subunit. We reported that addition of the inactive ␤-MPP mutated at either the putative active site or at essential glutamate residues efficiently decreased the wild-type MPP activity by replacing the wild-type ␤-MPP (19,42). Therefore, we used this system to examine interactions between the mutant ␣-MPP and the wild-type ␤-MPP (Fig. 3B). When increasing amounts of mutant ␣-MPPs were added to the wild-type MPP, the mutant subunits clearly had decreased MPP activity, and a 50% reduction was observed when the equimolar amounts of the wild-type and mutant ␣-subunits were used. Thus, the mutant ␣-subunits can bind to the ␤-subunit with a similar affinity seen with the wild-type one.
These results indicate that loss of processing activity observed in the mutants, even in the mutant-deleted 8-amino acid residues in the glycine-rich region, is not due to structural disruption of ␣-MPP at the interface interacting with ␤-MPP.
Effect of Mutation on Binding of Extension Peptide-To de- termine effects of the mutation on direct interactions between the enzyme and substrate, we synthesized the fluorescencelabeled peptide, in which an environment-sensitive coumarin derivative was covalently introduced into the peptide (MDH1-25) at the amino terminus, N-DAC-MDH5-25, and then we analyzed the substrate binding ability of the enzymes. The coumarin-labeled peptide gave a fluorescence emission spectrum with the maximum at 482 nm, and addition of the MPP, which had been inactivated in the presence of EDTA, led to a large increase in fluorescence intensity and to a blue shift in emission spectrum (emission maximum at 470 nm) (data not shown). As reported earlier (38,42), titration of the enzyme with the fluorescence-labeled peptide produced a simple saturation curve with an emission intensity at 470 nm. The dissociation constants, K d , were calculated from the titration curves (Table I). The K d value of the wild-type MPP was determined to be 0.06 M, a value essentially the same as noted earlier, 0.048 M (38). The mutant enzymes, Lys296Ala and Met298Ala, had over a 10-fold less affinity than that of the wild type and a lesser effect on affinity to the N-DAC-MDH5-25 peptide was obtained with the Phe 289 mutants. On the other hand, deletion mutants, ⌬289 -291, ⌬288 -292, and ⌬287-294, showed an extremely low affinity to the peptide, i.e. the K d values were over 200-fold larger than that of the wild type.
Effect of Mutations on Kinetic Parameters-To examine effects of mutations on the overall reaction, including substrate recognition and catalytic activity, we determined kinetic parameters of the mutant enzymes, using the synthetic peptide substrate, MDH1-21 (Table II). As expected from the results for Fig. 2, the k cat values were dramatically decreased in the mutants, Phe289Ala, Phe289Leu, Lys296Ala, Met298Ala, and ⌬289 -291, although the extent of the effect varied among the mutants. Mutation at Phe 289 and Lys 296 led to reduction of the activity to less than 1% of the wild type. Although these mutants exhibited slight increases in the K m value, there was a marked decrease in the catalytic efficiency, k cat /K m , to less than 1% of that of the wild-type enzyme. The three deletion mutants exhibited no detectable activity. Thus, taken together with data in Table I, shortening the glycine-rich stretch led to a dramatic reduction of interactions between the enzyme and substrate peptides and cleavage reaction, whereas mutation of each amino acid in this region seemed to affect primarily the cleavage reaction. DISCUSSION Our present observations indicate that the glycine-rich segment in the middle portion of ␣-MPP, which is characteristic of this subunit, is important for MPP functions. We found that lack of a part of the segment led to a dramatic reduction of interaction between the enzyme and substrate peptide and of the catalytic activity, the k cat value. Even in the deletion mutant, the subunit protein retained correct interaction with the wild-type ␤-MPP, demonstrating that the mutation caused little structural distortion of ␣-MPP at the interface interacting with ␤-MPP.
The ␣and ␤-subunits of the MPP are homologous to the core proteins of mitochondrial ubiquinol-cytochrome oxidoreductase (21). Crystal structures of the bc 1 complex from mammalian and avian mitochondria have been reported (39 -41), and one can expect that MPP is similar in structure to the core proteins in the complex. Based on reported crystallographic data (39,40) on bovine core proteins, we generated an energy-minimized model of MPP (Fig. 4). In dimer form, the two subunits (␣-and ␤-MPP) form a crack leading to the internal cavity, the wall of which is lined with mostly hydrophilic amino acid residues. The glycine-rich region of ␣-MPP is predicted to form a loop structure that protrudes at the surface of internal cavity formed by two subunits and reaches close to the zinc-binding active center of ␤-MPP. We found that deletion of a part of the glycine-rich segment seemed to primarily affect binding of the extension peptides of the precursor proteins to the enzyme. The result suggests that the glycine-rich loop could open the crack in the molecule and present the extension peptide to the region near the active site. Shortening of the glycine-rich loop may lead to narrowing the crack.
An extension peptide of the "Rieske" iron-sulfur protein,   subunit 9, was reported to be caged and mainly interacts with some amino acids in the carboxyl-terminal region of the core 2 protein, which corresponds to ␣-MPP. It is surmised from the findings that subunit 9 has been cleaved directly from the Rieske protein by these core proteins and remained in the crack, although the isolated bovine heart core proteins lack processing activity (40). The core proteins in plant mitochondrial bc 1 complexes, however, have processing activity and do not contain subunit 9 in the molecule (9). It has been reported that core proteins of bovine heart bc 1 complex exhibit peptidase activity after removal or weakening of the binding of the subunit 9 peptide with the core proteins in the complex by mild detergent treatment (44). It is worthwhile to note that ␣-MPPs from mammalian and yeast mitochondria and the core 2 protein from plant bc 1 complex have a long glycine-rich segment in the middle portion of the subunit, whereas the corresponding segments of the core 2 proteins in the bc 1 complexes from mammalian, avian, and yeast are all short (Fig. 1A). Thus, evidence of processing activity seems to correlate with the lack of subunit 9 and the presence of a long glycine-rich segment. Alternative possible role of the glycine-rich segment might be promotion of rapid release of the extension peptide and mature protein from the enzyme after the processing reaction. Lack of a long loop would impair the process and result in a dramatic reduction in rates of reaction. We also found in the segment that Phe 289 , Lys 296 , and Met 298 are responsible for catalytic activity rather than for interaction with the precursor proteins, because the k cat value is markedly decreased by mutation of these residues to alanine, whereas K d and K m values are not markedly affected. We reported that alteration of amino acid at position ϩ2 in the peptide substrates had a greater effect on the V max value than on the K m value (31). Other residues near the cleavage site, i.e. positions Ϫ2, ϩ1, and ϩ3, are also essential for the processing reaction, and the catalytic activity was influenced by mutation of these amino acids, albeit to a lesser extent (27,28,31). Although it is difficult to state the precise orientation of each residue in the loop because the structure presented in Fig. 4 is a simulation model calculated from the data based on the core proteins, the well conserved amino acids in the glycine-rich segment, like Phe 289 , Lys 296 and Met 298 , seem to interact directly with these essential amino acids near the cleavage site of the precursor proteins. Since Phe 289 could not be replaced by leucine, the aromatic nature would be required for its function, suggesting that the residue could interact with the proximal arginine residue at position Ϫ2, which is one of the most important resides both for recognition of the substrate peptide and cleavage reaction (27), through cation-interaction to fix the arginine residue at around reaction center. Lys 296 could also interact with amino acids at position ϩ2 and/or ϩ3, where hydroxyl residues are highly effective for cleavage reaction (28,29), probably through hydrogen bonds, whereas Met 298 might interact with aromatic or hydrophobic residue at position ϩ1 through its flexible hydrophobic stalk. However, the precise structural information, such as crystallographic data on the MPP, is required for further discussion.