Signal Peptides Having Standard and Nonstandard Cleavage Sites Can Be Processed by Imp1p of the Mitochondrial Inner Membrane Protease*

We have performed a site-directed mutagenesis study showing that residues comprising the type I signal peptidase signature in the two catalytic subunits of the yeast inner membrane protease, Imp1p and Imp2p, are functionally important, consistent with the idea that these subunits contain a serine/lysine catalytic dyad. Previous studies have shown that Imp1p cleaves signal peptides having asparagine at the −1 position, which deviates from the typical signal peptide possessing a small uncharged amino acid at this position. To determine whether asparagine is responsible for the nonoverlapping substrate specificities exhibited by the inner membrane protease subunits, we have substituted asparagine with 19 amino acids in the Imp1p substrate i-cytochrome (cyt)b 2. The resulting signal peptides containing alanine, serine, cysteine, leucine, and methionine can be cleaved efficiently by Imp1p. The remaining mutant signal peptides are cleaved inefficiently or not at all. Surprisingly, none of the amino acid changes results in the recognition of i-cytb 2 by Imp2p, whose natural substrate,i-cyt c 1, has alanine at the −1 position. The data demonstrate that (i) although the −1 residue is important in substrates recognized by Imp1p, signal peptides having standard and nonstandard cleavage sites can be processed by Imp1p, and (ii) a −1 asparagine does not govern the substrate specificity of the inner membrane protease subunits.

The type I signal peptidase family consists of enzymes located in the plasma membranes of eubacterial cells, the endoplasmic reticulum (ER) 1 membrane, and the inner membrane of mitochondria. These enzymes function similarly to cleave signal peptides from the amino termini of precursor proteins after the delivery of these precursors to their appropriate cellular compartments (reviewed in Ref. 1). Based on the crystal structure of leader peptidase from Escherichia coli (2), the eubacterial signal peptidases exhibit five characteristic residues that make up a type I signature sequence consisting of a serine/lysine catalytic dyad and three structurally important amino acids, an arginine and two aspartic acids, that are posi-tioned close to the catalytic site. The ER homolog of the eubacterial signal peptidases, Sec11p, has a similar signature; however, Sec11p contains histidine in place of lysine. Because this histidine is important for catalysis (3), Sec11p may utilize a serine/histidine dyad or a catalytic triad of serine/histidine/ aspartic acid. The type I signal peptidase found in the inner membrane of mitochondria consists of two subunits, both of which are catalytic (4 -6). Termed inner membrane protease (IMP), sequence comparisons suggest that the IMP subunits, Imp1p and Imp2p, may contain serine/lysine dyads, like their eubacterial counterparts (1).
Imp1p from the yeast Saccharomyces cerevisiae cleaves the signal peptides from the precursors to cytochrome (cyt) b 2 , a nuclear encoded protein, and cyt oxidase subunit II, a protein encoded within the mitochondrion. The signal peptides of these precursors possess asparagine at the Ϫ1 position from the cleavage site (4). The presence of a Ϫ1 asparagine in the Imp1p substrates violates the "Ϫ1, Ϫ3 rule" proposed in previous studies (7)(8)(9), which is based on the fact that signal peptides in the eubacterial and ER systems contain small uncharged amino acids at the Ϫ1 position and, to a lesser extent, the Ϫ3 position. Indeed, site-directed mutagenesis studies have confirmed the critical role of the Ϫ1 amino acid for the correct cleavage of signal peptides in the ER and in eubacterial cells (10 -12). From the crystallographic analysis of leader peptidase, the Ϫ1 and Ϫ3 amino acids probably help to position the signal peptide relative to the active site through their interactions with distinct binding pockets on the enzyme's surface (2).
In agreement with the Ϫ1, Ϫ3 rule, Imp2p cleaves a more conventional signal peptide such as that of the cyt c 1 precursor, which has alanine at the Ϫ1 position (5). A reasonable hypothesis to explain why Imp1p cleaves a different signal peptide from that of Imp2p is that Imp1p is able to recognize a Ϫ1 asparagine (5). An extension of this idea is that asparagine may fit into the appropriate binding pocket of Imp1p and be excluded from the corresponding site in Imp2p. To address this hypothesis, we have asked the following question: do mutations affecting the Ϫ1 asparagine of the cyt b 2 precursor inhibit its cleavage by Imp1p and, conversely, allow for its cleavage by Imp2p? We have also asked whether the type I signature sequences of Imp1p and Imp2p are important for their enzymatic activities.
Biochemical Procedures-Pulse-chase methods have been published (13,15). For immunoprecipitation analysis, the procedure published previously (15) was used, except for the following modifications. Anti-HA antibodies and protein G (instead of protein A)-Sepharose were used. After the wash of the protein G-Sepharose pellet, the material was resuspended in 0.1 ml of 1% SDS, 50 mM Tris (pH 7.5) and boiled for 4 min. PBS containing 1% Triton X-100 (1.0 ml) and a protease inhibitor mixture (15) was added. The mixture was vortexed briefly in an Eppendorf tube and placed on ice for 10 min. The tube was subjected to centrifugation (14,000 ϫ g) for 10 s, and the supernatant was transferred to a new Eppendorf tube. A second aliquot of anti-HA antibody (2 g/3 A 600 cell equivalents) was added, and the immunoprecipitation procedure described previously (15) was followed.
Epitope Tagging-Protein tagging using the HA epitope (16) was performed as follows. The CYB2 gene encoding the precursor to cyt b 2 was amplified by the polymerase chain reaction using upstream primer TCCCCCGGGATGCTAAAATACAAACCTTTAC containing a SmaI site and downstream primer GCGAATTCTTACGCGTAGTCTGGGACGTC-GTATGGGTATGCATCCTCAAATTCTGTTAAAGTAGGTCCC containing a EcoRI site and encoding the HA epitope. The product was inserted between the SmaI and EcoRI sites of pHF454 (2 m TRP1) (3), which placed the CYB2 coding sequence immediately downstream of the ADH1 promoter (17).
The CYT1 gene encoding the precursor to cyt c 1 was amplified using upstream primer CGGGATCCATGTTTTCAAATCTATCTAAACG containing a BamHI site and downstream primer GCGAATTCTTACGCG-TAGTCTGGGACGTCGTATGGGTACTTTCTTGGTTTTGGTGGATTG-AAAACG containing a EcoRI site and encoding the HA epitope. The product was inserted between the BamHI and EcoRI sites of pHF454 (2 m TRP1) immediately downstream of the ADH1 promoter.
Site-directed Mutagenesis-All site-directed mutagenesis reactions were performed as described previously (3). The wild-type IMP1 gene and the mutations constructed in IMP1 were introduced into pRS426 (2 m URA3) (18). The DNA fragment that was used included the native IMP1 promoter carried by a 400-nucleotide stretch upstream of the Imp1p coding sequence. The wild-type IMP2 gene and the mutations constructed in IMP2 were introduced into pHF455 (2 m URA3), which placed the IMP2 coding sequence immediately downstream of the ADH1 promoter. To mutagenize CYB2, the wild-type CYB2 gene and the mutations constructed in CYB2 (tagged with a sequence encoding the HA epitope as described above) were introduced into pHF454 (2 m TRP1), which placed the CYB2 gene downstream of the ADH1 promoter. To mutagenize CYT1, the wild-type CYT1 gene and the mutations constructed in CYT1 (tagged with a sequence encoding the HA epitope as described above) were introduced into pHF454 (2 m TRP1), placing the CYT1 gene immediately downstream of the ADH1 promoter. The sequences of the oligonucleotides used to construct these mutations are available upon request.

Amino Acids Comprising the Type I Signal Peptidase
Signature Are Important for Imp1p and Imp2p Function-Five key amino acids have been found to serve critical roles in catalysis and structural maintenance of the type I signal peptidases in eubacteria, including leader peptidase from E. coli (2, 19 -21) and SipS from Bacillus subtilis (22). This signature sequence includes the catalytic dyad residues, serine and lysine, and three amino acids, an arginine and two aspartic acids, that are structurally important (Fig. 1). Because Imp1p and Imp2p of the mitochondrial signal peptidase from the yeast S. cerevisiae contain this signature, we asked whether these five amino acids were important for enzyme function using a site-directed mutagenesis approach. Mutations altering these five residues were constructed in the IMP1 gene and inserted into pRS426 (2 m URA3) (see "Experimental Procedures"). The constructs were then introduced into yeast strain XCY101 (imp1::HIS3). The genotypes of strains used in this study are listed under "Experimental Procedures." To determine whether these mutations inhibited enzyme activity, a pulse-chase assay was used to monitor for cleavage of the Imp1p substrate i-cyt b 2 . Pre-cyt b 2 (p-cyt b 2 ) contains a bipartite signal sequence (23). The amino-terminal half is cleaved by the mitochondrial processing protease (␣/␤) within the mitochondrial matrix (24 -27). This cleavage event liberates an intermediate form, i-cyt b 2 , that contains the second half of the bipartite signal, which is cleaved by Imp1p. Cells of strain XCY101 bearing the imp1 mutations and bearing pXC2 (2 m TRP1) that contained the wild-type CYB2 gene (encoding p-cyt b 2 that had been tagged at the carboxyl terminus with the HA epitope) (see "Experimental Procedures") were grown to log phase in a glucose-containing medium and treated with a 10-min pulse using [ 35 S]methionine and [ 35 S]cysteine, followed by a 60-min chase with excess unlabeled methionine and cysteine. Proteins were precipitated from cell extracts using anti-HA antibodies and subjected to SDS-PAGE and fluorography.
After the 60-min chase, mature cyt b 2 HA (m-cyt b 2 HA) was present in cells carrying the wild-type IMP1 gene, whereas the intermediate species, but no mature species, was found in XCY101 cells (imp1::HIS3) lacking Imp1p (Fig. 2). The p-cyt b 2 HA protein containing the bipartite signal was not detected under these conditions. In cells carrying the S40A, S40T, K84R, K84H, D131Y, D138N, and D138E mutations, i-cyt b 2 HA was seen exclusively, indicating a strong inhibition of cleavage by these mutated forms of Imp1p. In agreement with this result, yeast cells bearing these imp1 mutations were not viable on agar plates containing glycerol, a nonfermentable carbon source. Imp1p is required for cellular respiration, probably because at least one of the Imp1p substrates is nonfunctional when its signal peptide is still attached. Therefore, growth of yeast cells in the presence of a nonfermentable carbon source is an indicator of Imp1p function (4,5). The remaining imp1 mutations (D131N, D131E, and R85A) inhibited the cleavage of i-cyt b 2 HA less well (Fig. 1). Furthermore, yeast cells bearing these mutations were able to grow on agar plates containing glycerol as the sole carbon source. Taken together, all five amino acids comprising the type I signature of Imp1p were important for function, although Asp 131 could be changed to asparagine and glutamic acid without appreciable loss of activity, and Arg 85 could be changed to alanine, resulting in only a partial enzymatic defect. We next asked whether the type I signature amino acids in Imp2p were important for its function. A series of imp2 mutations was constructed by site-directed mutagenesis. The IMP2 gene and the imp2 mutations were introduced into pHF455 (2 m URA3) (see "Experimental Procedures"). This plasmid series was then introduced into strain JNY34 (imp2⌬), which contained pXC1 (2 m TRP1) that bore the CYT1 gene (encoding p-cyt c 1 that had been tagged at the carboxyl terminus with the HA epitope) (see "Experimental Procedures"). Cells were grown in a medium containing glucose and then subjected to a 15-min pulse and a 30-min chase.
When the wild-type IMP2 gene was present in cells, a large amount of m-cyt c 1 HA was exhibited (Fig. 3). These cells also displayed both p-cyt c 1 HA, which contains a bipartite signal peptide (28), and i-cyt c 1 HA, which contains the second half of the bipartite signal. The presence of these two immature species in cells containing wild-type Imp2p may be due to the fact that the CYT1 gene was overexpressed (see "Experimental Procedures" for a description of the plasmid used). A novel species of slightly smaller molecular mass than m-cyt c 1 HA was also present in cells containing wild-type Imp2p (Fig. 3, see lane IMP2(wt)). The identity of this novel form is not known; however, it may represent a proteolytic fragment arising from overexpression of the apoprotein. In JNY34 (imp2⌬) cells lacking Imp2p, m-cyt c 1 HA was absent (Fig. 3). Likewise, little or no production of m-cyt c 1 HA was detected in cells carrying the S41A, S41T, K91R, K91H, D124N, D124Y, D131N, and D131E mutations. One of these mutations, the S41A mutation, was also constructed in a previous study, and results similar to those reported here were obtained (5). The R92A and D124E mutations only partially inhibited the production of m-cyt c 1 HA (Fig. 3).
None of these imp2 mutations inhibited the growth of JNY34 cells on agar plates containing glycerol, consistent with the fact that Imp2p activity is nonessential for cellular respiration (5). However, a physical interaction between Imp1p and Imp2p is important for the stability of Imp1p and thus for the growth of yeast cells in the presence of a nonfermentable carbon source (5). The fact that none of the imp2 mutations inhibited the growth of JNY34 cells in the presence of glycerol suggests that the imp2 mutations did not prevent the binding of Imp2p to Imp1p. Indeed, we have shown that Imp2p containing the above-mentioned mutations altering Ser 41 or Lys 91 was stable in yeast cells for at least 30 min, whereas mutations altering Arg 92 , Asp 124 , and Asp 131 led to the degradation of Imp2p in vivo (data not shown).
In summary, the data show that, as with Imp1p, the five amino acids comprising the type I signature of Imp2p are important for its enzymatic activity. Considering that mutations altering Ser 41 and Lys 91 of Imp2p inhibit enzyme activity completely but do not affect Imp2p stability in vivo, Imp2p and, by extension, Imp1p probably contain a serine/lysine catalytic dyad.
Imp1p Cleaves Signal Peptides Containing Standard and Nonstandard Cleavage Sites-Imp2p cleaves the signal peptide of i-cyt c 1 that contains the Ϫ1 residue, alanine, whereas Imp1p cleaves the signal peptide of i-cyt b 2 that has a nonstandard Ϫ1 residue, asparagine. Because current models suggest that asparagine may provide a determinant for this substrate specificity, we prepared a series of constructs that introduced 19 amino acids into the Ϫ1 position of i-cyt b 2 HA to identify amino acid substitutions that inhibit cleavage by Imp1p and promote cleavage by Imp2p. These changes were constructed by sitedirected mutagenesis (see "Experimental Procedures"). To determine whether Imp1p was able to cleave signal peptides containing these amino acid substitutions, a series of plasmids was introduced into cells of strain JNY34 (imp2⌬). Because Imp1p is unstable in the absence of Imp2p, the cells also carried plasmid pXC3 that contained the imp2 (S41A) mutation. This mutation renders Imp2p enzymatically inactive (see Fig. 3) but does not affect Imp2p stability and thus Imp1p stability (5). Cells were grown to log phase and subjected to a 10-min pulse followed by a 60-min chase, and proteins were precipitated from cell extracts using anti-HA antibodies.
As shown in Fig. 4A, the placement of five different amino acids into the signal peptide of i-cyt b 2 HA led to its efficient cleavage by Imp1p. Serine, cysteine, methionine, alanine, and leucine were tolerated almost as well as the naturally occurring asparagine residue. Only a small amount of cleavage of i-cyt b 2 HA by Imp1p was detected when glutamine and threonine were present and no cleavage was detected when any other amino acid was present at the Ϫ1 position (Fig. 4A). The inhibition of cleavage by 14 different amino acid substitutions at the Ϫ1 position argues against the idea that cleavage can occur at a nearby asparagine when serine, cysteine, methionine, alanine, and leucine are present.
We next introduced this series of mutations into cells of strain XCY101 (imp1::HIS3), and then we performed a pulsechase analysis to determine whether Imp2p could cleave these mutant forms of i-cyt b 2 HA. As shown in Fig. 4B, Imp2p was unable to cleave i-cyt b 2 HA containing any amino acid at the Ϫ1 position.
We reasoned that the introduction of a Ϫ1 methionine into i-cyt b 2 HA may result in a new translational start site, which could be mistaken for cleavage at the Ϫ1 site by Imp1p (Fig.  4A). We therefore repeated the pulse-chase analysis, only this time, proteins were immunoprecipitated from cell extracts of strain JNY34 (imp2⌬)/pXC3 (imp2 (S41A)) after both the pulse and the chase. The p-cyt b 2 HA protein was present after the pulse, and m-cyt b 2 HA was present after a 30-min chase (data not shown). This precursor-product relationship established that a cleavage event had occurred, supporting the idea that Imp1p can cleave i-cyt b 2 HA after the Ϫ1 methionine. Further support for this conclusion comes from the fact that a protein having the size of m-cyt b 2 HA was absent from strain XCY101 (Fig. 4B).
The data presented in Fig. 4B show that, surprisingly, Imp2p could not cleave i-cyt b 2 HA that had alanine at the Ϫ1 position. Because the natural substrate for Imp2p, i-cyt c 1 , contains a Ϫ1 alanine, we chose to examine the Ϫ3 residue of i-cyt b 2 HA. FIG. 2. Processing of the cyt b 2 precursor by mutant forms of Imp1p. Processing of the cyt b 2 precursor tagged with the HA epitope was examined in XCY101 cells (imp1::HIS3)/pXC2 (CYB2-HA TRP1) bearing a URA3-marked plasmid containing the wild-type (wt) IMP1 gene or one of a number of imp1 mutations that are indicated at the top of the figure (see "Experimental Procedures"). Cells were subjected to a 10-min pulse and a 60-min chase, and proteins were precipitated using anti-HA antibodies (see "Experimental Procedures"). Intermediate (i) and mature (m) cyt b 2 were resolved on a 7% SDS-PAGE gel. Each lane represents two A 600 equivalents of log-phase yeast cells.
Both of the Imp1p substrates, i-cyt b 2 and p-cyt oxidase subunit II, contained isoleucine at the Ϫ3 position, whereas the Imp2p substrate had a Ϫ3 alanine. It thus seemed plausible that in order to promote cleavage of i-cyt b 2 by Imp2p, it may be necessary to change the Ϫ1 and Ϫ3 amino acids to alanine. To this end, we constructed a series of mutations in which different combinations of asparagine and alanine were placed at the Ϫ1 position, and different combinations of isoleucine and alanine were placed at the Ϫ3 position of i-cyt b 2 HA. The mutations were introduced into plasmid pHF454 (2 m TRP1) (see "Experimental Procedures"), and the constructs were transformed into strains XCY101 (imp1::HIS3) and JNY34 (imp2⌬)/ pXC3 (imp2 (S41A)). Cells were then examined by pulse-chase analysis. As shown in Fig. 5, none of the mutational combinations led to a form of i-cyt b 2 HA that could be cleaved by Imp2p, including the (Ϫ1A Ϫ3I) form that was described earlier (Fig.  4B, see lane A). Of particular interest is the (Ϫ1A Ϫ3A) double mutation. Whereas the natural substrate for Imp2p contained a Ϫ1 and Ϫ3 alanine, Imp2p was unable to cleave the double alanine form of i-cyt b 2 HA (Fig. 5). In contrast, Imp1p was able to cleave some forms of i-cyt b 2 HA, including the wild-type species (Ϫ1N Ϫ3I) and two mutant species, (Ϫ1N Ϫ3A; Fig. 5) and (Ϫ1A Ϫ3I; Fig. 4A, see lane A). These two mutations demonstrated that Imp1p can cleave a signal peptide containing alanine at the Ϫ1 or Ϫ3 position. However, introducing alanine at both positions of i-cyt b 2 HA inhibited cleavage by Imp1p (Fig. 5), demonstrating the importance of the Ϫ1 and Ϫ3 amino acids in determining cleavage by Imp1p.

DISCUSSION
In this study, we have asked whether amino acids comprising the type I signal peptidase signatures of the IMP subunits are functionally important. Studies in eubacterial systems have shown that the type I signature consists of a serine, lysine, arginine, and two aspartic acid residues that are important for the function of leader peptidase from E. coli and SipS from B. subtilis (21,22). Imp1p and Imp2p from the yeast mitochondrion contain similar signatures (Fig. 1); however, functional studies have been lacking, except for an analysis of Ser 41 , which has been shown to be essential to Imp2p (5). Here, we have demonstrated that conservative changes of the signature serine and lysine residues of Imp1p (Fig. 2) and Imp2p (Fig. 3) abolish their enzymatic activity. Moreover, Imp2p containing conservative changes of the signature serine and lysine residues is stable in vivo (data not shown), consistent with studies in eubacterial systems showing that the corresponding residues make up a catalytic serine/lysine dyad (2,21,22). In the eubacterial studies, the remaining amino acids, an arginine and two aspartic acid residues, are important structural amino acids, and from the data presented here, it is plausible that the corresponding amino acids in Imp1p and Imp2p serve a similar role. Considering these results, Imp1p and Imp2p contain similar type I signatures, although they cleave different signal peptides inside the mitochondrion.
Another goal of this study is to understand why Imp1p and Imp2p exhibit nonoverlapping substrate specificities. To this end, we have tested the hypothesis that the presence of asparagine at the Ϫ1 position directs signal peptides to Imp1p and away from Imp2p. We have substituted the Ϫ1 asparagine of i-cyt b 2 with 19 amino acids and then examined the mutant signal peptides for cleavage by Imp1p (Fig. 4A) and Imp2p (Fig.  4B). The data reveal that five different amino acids can be introduced into the Ϫ1 position of i-cyt b 2 without appreciable loss of cleavage efficiency by Imp1p. The amino acids tolerated are serine, cysteine, methionine, alanine, and leucine. Including the naturally occurring amino acid, asparagine, this group includes polar (asparagine, serine, and cysteine) and nonpolar (methionine, alanine, and leucine) amino acids and amino acids with varying side-chain lengths. Both of the sulfur-containing amino acids (cysteine and methionine) are permitted. Surprisingly, the alanine substitution is recognized by Imp1p, despite the fact that the Imp2p substrate, i-cyt c 1 , contains a Ϫ1 alanine. Indeed, alanine, serine, cysteine, and leucine have been seen at the Ϫ1 positions of signal peptides targeted to the ER membrane (29 -31). From the fact that Imp1p can cleave these signal peptides and cleave the highly unusual signal peptides containing asparagine and methionine at the Ϫ1 position, we conclude that Imp1p is capable of processing signal peptides exhibiting both standard and nonstandard cleavage sites.
A caveat to this conclusion is that without amino-terminal sequencing of the cleavage product, we cannot be absolutely certain that cleavage did not occur at a nearby asparagine residue in the linear amino acid sequence. However, 14 amino acid substitutions at the Ϫ1 position of i-cyt b 2 produce signal peptides that are recognized poorly or not at all by Imp1p (Fig.  4A). Signal peptides containing glutamine and threonine are cleaved to a small extent, and the remaining amino acid substitutions are not recognized, including the substitutions of the aromatic and charged amino acids and substitutions of proline, glycine, isoleucine, and valine. The absence of efficient cleavage of 14 mutant signal pepides thus argues strongly against the idea that cleavage can occur at a nearby asparagine.
Although the Imp2p substrate i-cyt c 1 contains alanine at the Ϫ1 position, Imp2p cannot cleave i-cyt b 2 containing a Ϫ1 alanine (Fig. 4B). In fact, Imp2p cannot cleave i-cyt b 2 regardless of the amino acid present at the Ϫ1 position. This argues against the hypothesis that asparagine is responsible for directing signal peptides toward Imp1p and away from Imp2p. During the course of this study, we noted that Imp1p substrates i-cyt b 2 and p-cyt oxidase subunit II contain a Ϫ3 isoleucine, whereas the Imp2p substrate, i-cyt c 1 , contains a Ϫ3 alanine. Because the Ϫ3 position is thought to be important for signal peptide recognition in the ER and eubacterial systems (2, 29 -31), we constructed another series of mutations, changing the Ϫ1 and Ϫ3 amino acids of i-cyt b 2 . However, not even i-cyt b 2 containing alanine at both the Ϫ1 and Ϫ3 positions could be cleaved by Imp2p (Fig. 5).
Thus, the goal of identifying factors that govern the substrate specificities exhibited by the IMP subunits has been achieved only partially in this study. We have shown that amino acids at the Ϫ1 and Ϫ3 positions are important for proper cleavage of the signal peptide of i-cyt b 2 . In contrast to previous models, however, a Ϫ1 asparagine is not required to direct signal peptides to Imp1p. Because the attachment of heme c to the i-cyt c 1 apoprotein is essential for cleavage of this substrate by Imp2p (28), and the heme binding domain of i-cyt b 2 is needed for its efficient cleavage by Imp1p (32,33), current studies are focused on determining whether the mature portions of these proteins are important for their recognition by specific IMP subunits.