Identification of Potential Active-site Residues in the Escherichia coli Leader Peptidase*

Leader peptidase of Escherichia coli cleaves the leader sequence from the amino terminus of membrane and secreted proteins after these proteins insert across the membrane. Despite considerable research, the mechanism of catalysis of leader peptidase remains unknown. This peptidase cannot be classified using protease inhibitors to the serine, cysteine, aspartic acid, or metallo- classes of proteases (Zwizinski, C., Date, T., and Wickner, W. (1981) J. Biol. Chem. 256, 3593-3597). Using site-directed mutagenesis, we have attempted to place leader peptidase in one of these groups. We found that leader peptidase, lacking all of the cysteine residues, can cleave the leader peptide from procoat, the precursor to bacteriophage M13 coat protein. Replacement of each histidine residue with an alanyl residue was without effect on catalysis. Among all the serine and aspartic acid residues, serine 90 and serine 185 as well as aspartic acid 99, 153, 273, and 276 are necessary to cleave procoat in a detergent extract. However, only serine 90 and aspartic acid 153 were required for processing using a highly sensitive in vivo assay. In addition to the residues directly af-fecting catalysis, aspartic acid 99 plays a role in maintaining the structure of leader peptidase. Replacement of this residue with alanine results in a very unstable leader peptidase protein. This study thus defines two critical residues, serine 90 and aspartic acid 153, that may be directly involved in catalysis and provides evidence that leader

ported to the outer and inner membrane. All other known preproteins are processed by leader peptidase, an inner membrane protein of 37,000 Da (Zwizinski and Wickner, 1980;Wolfe et al., 1982). Its orientation across the membrane is shown in Fig. 1. Apolar domains H1 and H2 are transmembrane, the short polar P1 region is in the cytoplasm, and the third apolar domain H3 and the large carboxyl-terminal polar P2 region protrude into the periplasm (Wolfe et al., 1983a;Moore and Miura, 1987). The regions of bacterial leader peptidase needed for catalysis have recently been explored using a genetic deletion approach (Bilgin et al., 1990). H1 and P1 are not directly involved in catalysis, but H2 and the region immediately following it are important for the activity, suggesting that these latter regions may be close to the active site of leader peptidase.
The substrate specificity of leader peptidase has been investigated extensively (Watts et al., 1983;Koshland et at., 1982;Fikes et al., 1990;Kuhn and Wickner, 1985;Shen et al., 1991). In detergent extracts, it cleaves preproteins, ranging from bacterial exported proteins through yeast preacid phosphatase, honeybee prepromellitin, and human prehormones. Although there is no sequence similarity in the leader peptides of these preproteins, there is a common pattern of small amino acids at -1 and -3 with respect to the cleavage site and a helix-breaking residue in the region -4 to -6. Mutations that block processing indeed have been isolated to these leader peptide positions in @-lactamase (Koshland et al., 1982), maltose-binding protein (Fikes et al., 1990), and procoat (Kuhn and Wickner, 1985). Furthermore, we have recently found that the M13 precursor protein, termed procoat, can be effciently cleaved only if it has the helix breaker proline at -6, a leucine, valine, threonine, glycine, or a serine at -3, and a glycine, serine, proline, or an alanine at -1 (Shen et al., 1991). With few exceptions, almost any residue can be tolerated for leader peptidase processing at +1, -2, -4, and -5 of the M13 procoat.
While major progress has been made in understanding the substrate specificity, the mechanism of catalysis of leader peptidase remains elusive. Protease inhibitors such as phenylmethylsulfonyl fluoride, tosylamido-2-phenylethylchloromethylketone, EDTA, o-phenanthroline, N-ethylmaleimide, dinitrophenol, carboxyphenanthroline, or 2,6-pyridinedicarboxylic acid are ineffective at inhibiting leader peptidase (Zwizinski et al., 1981). Using site-directed mutagenesis, we now show that leader peptidase does not require a histidine or a cysteine for catalysis. Of all the serine and aspartic acid residues, serine 90 and 185 as well as aspartic acid 99, 153, 273, and 276 are required for processing of the M13 procoat protein in a Triton X-100 extract. In contrast, only the serine 90 and aspartic acid 153 mutants are inactive using a more sensitive in vivo assay. These results, along with sequence similarity with other prokaryotic leader peptidases and the Saccharomyces cerevisiae mitochondrial leader peptidase, sug- coli. E , amino acid sequence of leader peptidase. The amino acids which were changed by site-directed mutagenesis are depicted by the following symbols: I (cysteine), n (histidine), 0 (serine), and 0 (aspartic acid). All residues were replaced with alanine, except for cysteines, which were replaced with serine residues. gest that serine 90 and aspartic acid 153 may participate directly in catalysis and that leader peptidase is likely to belong to a novel class of serine proteases.

Bacterial Strains and Plasmids"MC1061
(AlacX74, ara-D139, A(ara, leu)7697, gam, galK, hsr, hsm, strA) and JM103 (Alacpro) thi, strA, supE, endA, sbcB, hsdR, traD36, proAB, lacIqZ M15) were from our collection. The PING plasmid (Johnston et al., 1985), which has the arabinose operon regulatory elements and the arabinose promoter, was from Dr. Gary Wilcox (Ingene, Inc). Insertion of the leader peptidase gene into the PING plasmid was described in Dalbey and Wickner (1985). DNA Methods-Oligonucleotide-directed mutagenesis was performed using the M13 mp8 vector containing the lepB gene as a template, as described in Dalbey and Wickner (1987). The techniques described in Maniatis et al. (1982) were used for the DNA manipulations. Restriction enzymes were purchased from BRL and New England Biolabs. For DNA transformations, the method of Cohen et nl. (1973) was used.
In Vitro Assay-Leader peptidase activity was measured by the posttranslational conversion of procoat to coat protein and leader peptide. In this assay, procoat was synthesized in vitro according to the procedure of Zalkin et al. (1974), except for minor modifications the E. coli Leader Peptidase 13155 (Yamane et al., 1987). Cultures (2 ml) of MC1061 expressing different leader peptidase mutants were grown to an optical density at 600 nm of 0.2, induced with arabinose for 2 h, followed by centrifugation for 1 min to concentrate the cells. After resuspending in 0.3 ml of lysis buffer containing 20% sucrose, 10 mM Tris-HC1, (pH 8.0), 10 mM EDTA, 1% Triton X-100, lysozyme (1 mg/ml), deoxyribonuclease (5 pg/ml), ribonuclease (1 pg/ml), and phenylmethylsulfonyl fluoride (5 mM) and incubating for 30 min at room temperature, cell extracts were used directly or diluted (1:10, etc.) and incubated with 35Slabeled procoat at 37 "C for 30 min. The processing of procoat was analyzed on a 23% polyacrylamide gel (Boeke et al., 1980). I n Vivo Assay-Leader peptidase activity of the mutants was measured by examining the processing of outer membrane protein A precursor (pro-OmpA) in IT41, a temperature-sensitive leader peptidase strain (Inada et al., 1989). Briefly, IT41 was grown at 32 "C in M9 medium containing 0.5% fructose and 50 pg/ml each amino acid (except methionine). After reaching the early-log phase, cultures were shifted to 42 "C for 1 h to inactivate the temperature-sensitive leader peptidase. Arabinose (0.2%) was added to the medium to induce synthesis of leader peptidase. Cells were labeled with [3sS]methionine for 15 s, and unlabeled methionine was added to a final concentration of 500 pg/ml. At indicated times, samples were removed and quenched in 20% trichloroacetic acid. Samples were immunoprecipitated with antibody directed against outer membrane protein A and analyzed on a 12% SDS'-polyacrylamide gel (acrylamide:bis, 30:0.8) with a 5% stacking gel using a discontinuous buffer system. The gels were then fixed and subjected to fluorography (Ito et al., 1980).

RESULTS
We have used site-directed mutagenesis in an attempt to place leader peptidase in either the cysteine, serine, aspartic acid, or metallopeptidase class of proteases. All oligonucleotide-directed mutations were made as described by Zoller and Smith (1983) with some modifications (Dalbey and Wickner, 1987). The mutants were designated XNY (X, amino acid in the wild-type protein; N , the position of the amino acid; Y, the new substituted amino acid). The sequences of the mutagenic primers are shown in Table I, and the sites of the mutations are indicated in Fig. 1B. Each of the mutants was evaluated first for its catalytic activity using an in vitro assay (Table 11).
As a first step, we replaced all of the cysteine residues with serines. This was achieved in two successive steps. In the first, the cysteine at position 21 was replaced with a serine. In the second step, the cysteine residues at positions 170 and 176 were replaced with serines using as a template a leader peptidase gene encoding the C21S substitution. Leader peptidase lacking all three cysteines was then assayed for enzymatic activity with an in vitro assay (Fig. 2). In this assay, radiochemically labeled [35S]pro~oat was used as a substrate. Leader peptidase cleaves procoat to the coat protein (Fig. 2,  bottom right panel). Briefly, the in vitro assay is based on a detergent extract of MC1061 cells transformed with the appropriate plasmid. The plasmid pRD8 encodes for wild-type leader peptidase. As can be seen in Fig. 2, a 200-fold diluted extract of MC1061 pRD8 catalyzes comparable cleavage of procoat, as does undiluted extract of MC1061 without plasmid (compare no plasmid and pRD8 panels). The background activity is due to the leader peptidase coded by the chromosomal lepB gene. Cell extract containing leader peptidase C21S, C170S, and C176S, which lacks all cysteines, has full enzymatic activity. This indicates that leader peptidase is not a cysteine protease.
In a second step, we evaluated whether leader peptidase can be grouped into the classical metallo-or serine class of proteases. Since a histidine residue is essential in these groups we replaced each of the histidine residues with an alanine. Cell extracts bearing H124A, H235A, or H323A leader pepti-' The abbreviation used is: SDS, sodium dodecyl sulfate. The synthetic oligonucleotides were prepared on an Applied Biosystem model 380B instrument by the Ohio State University Biochemical Instrument Center to change potential residues involved in catalysis. The codons, which were chanEed to make the mutations. are underlined.

CCGGCTACCTAC~AACGTGGAACCG-
dase have normal enzyme activity ( Fig. 2 and Table 11). This shows that leader peptidase is not a standard serine protease. Nor is this result consistent with leader peptidase being a typical zinc metalloprotease since in this class of proteases histidine residues have been shown to chelate the metal ion (Dunn, 1989) and, hence, are likely to be indispensable for catalysis.
To determine whether aspartic acid and serine residues are required for catalysis, we replaced each of them with alanines. In this study, to simplify the number of oligonucleotides needed, two or three of these residues were changed at once using one oligonucleotide (Table I). Table I1 shows that three of the triple mutants, leader peptidase D153A, D158A, S161A, leader peptidase D191A, S185A, S190A, and leader peptidase D273A, D276A, S278A had 1/100 to 1/200 the activity of wild-type protein. In addition, two of the six double mutant proteins were inactive (Table 11). This includes leader peptidase S88A, S90A and leader peptidase D280A, S281A. Therefore, we also made a number of single mutations to identify which of the amino acids of the double or triple mutants were important for catalysis. As can be seen in Table 11, leader peptidase S90A (see also Fig. 2), S185A, D99A, D153A, D273A, or D276A have very low activity (background enzymatic activity) while all the other single serine and aspartic acid mutants had normal substrate processing (100-ZOO-fold higher). It is interesting to note that while the double mutant leader peptidase D280A, S281A was inactive, each of the c 2 1 s D23A D43A, S44A, D46A S65A S78A S88A, S90A S90A D99A D112A H124A D129A D138A, D142A D153A, D158A, S161A D153A D158A D161A C170S, C176S S171A, S172A S185A, S190A, D191A S185A S190A D191A S197A S206A S224A, D232A H235A D245A D273A, D276A, S278A D273A D276A S278A D280A, S281A D280A S281A S302A, D304A S317A H323A mutant proteins containing the single mutation was active (Table 11). We believe this is due to a conformational change that arises by the replacement of two amino acids.
As a second assay, we tested the activity of selected inactive leader peptidase mutants using a sensitive in uiuo assay (Bilgin et al., 1990). It is based on the fact that overproduction of leader peptidase accelerates the processing of the precursor to outer membrane protein A (pro-OmpA) at the nonpermissive temperature in IT41 (Inada et al., 1989). This strain bears a temperature-sensitive leader peptidase encoded by the chromosomal lepB gene (Fig. 3). Various IT41 strains containing either no plasmid, pRD8 (expressing wild-type leader peptidase), or a plasmid encoding the mutant proteins (S90A, D153A or S185A) were pulse-labeled for 15 s with [35S] methionine and chased with an excess of unlabeled methionine. At the indicated times, aliquots were removed and analyzed by immunoprecipitation to OmpA, SDS-polyacrylamide gel electrophoresis, and fluorography. While processing is very rapid with a tIl2 < 10 s in IT41 pRD8, it is slow in cells without plasmid ( t1/2 z 90 s). However, extracts prepared from cells expressing either leader peptidase S90A or D153A have very low activity comparable with that of IT41 without plasmid. Slightly higher activity is observed in cells expressing leader peptidase S185A (Fig. 3), and normal rapid processing is seen with leader peptidase D273A and D276A (data not shown). Since these latter residues are not required for in vivo processing, this indicates that serine 185 as well as aspartic ++++ D153A, D158A, S161A D191A, S185, S190A D273A, D276A, S278A Enzymatic activity is defined relative to the activity in a MC1061 extract (without plasmid), which is arbitrarily defined as -. Each + corresponds to roughly 50-fold higher activity than MC1061. Activity of the leader peptidase mutants in the cell extract was determined by comparing which dilutions (l:l, l:lO, 150, 1:200, and 1:400) were comparable with the undiluted MC1061 extract. For example, 1:200diluted extracts from pRD8 cells have processing comparable with that for undiluted extracts without plasmid.
acids 273 and 276 do not play a catalytic role. However, the critical residues, serine 90 and aspartic acid 153, may participate directly in catalysis. In contrast to the serine 90 and aspartic acid 153 mutant proteins, leader peptidase D99A was not detected by immunoblot analysis (data not shown). Pulse-chase experiments revealed that leader peptidase D99A is unstable (Fig. 4), with a half-life of roughly 2 min. This is compared to a half-life of > 60 min for wild-type leader peptidase (data not shown).

DISCUSSION
Proteases can be typically classified into four different groups: serine and cysteine proteases that form covalent enzyme complexes and aspartic acid and metalloproteases that do not form covalent intermediates (Neurath, 1989). To date all serine, cysteine, and metallopeptidase always have a histidine residue that is required for catalysis. The histidine residue acts as the proton donor and acceptor in serine and cysteine proteases. The metal ion, which is a zinc ion in all known metalloproteases, functions to polarize the carbonyl bond of the substrate and is coordinated by two histidine Cultures expressing the wild-type or mutant leader peptidase proteins were tested for leader peptidase activity by measuring the posttranslational processing of the M13 procoat a t various dilutions of the cell extract. For procoat synthesis, a transcription/ translation system was used as described under "Experimental Procedures." Briefly, cell extracts were prepared as described in the "Experimental Procedures." Undiluted or diluted (l:lO, 150, 1:200, or 1:400) extracts were incubated for 30 min at 37 "C with in uitro synthesized ["S]procoat. [RsS]Procoat was separated from the processed coat protein by using a 23% SDS-polyacrylamide gel. In the bottom right panel, in uitro synthesized procoat (10 pl) was incubated for 30 min at 37 "C with 2 pl of 1 mg/ml leader peptidase. residues and usually a glutamic acid (Dunn, 1989). Aspartic acid proteases almost always have a consensus sequence comprised of the motif Asp-Thr-Gly (Davies, 1990).
In this report, we have used site-directed mutagenesis to place leader peptidase into one of these four classes of proteases. Except for cysteines (which were replaced with serines), each of the histidine, aspartic acid, and serine residues was replaced with alanine. This analysis has revealed that cysteine residues are not needed for catalysis, showing that leader peptidase is not a cysteine protease. In addition, since histidine residues are not required for cleavage, it rules out leader peptidase as a classical serine protease. It is also highly unlikely that leader peptidase is a classic aspartic acid protease since it does not contain the motif Asp-Thr-Gly, which is found for all aspartic acid protease (Davies, 1990). The other group of proteases, metalloproteases, has a zinc atom that chelates two histidine residues (Dunn, 1989). Three lines of evidence indicate that leader peptidase is not a standard metallopeptidase. First, it does not contain a critical histidine, a result not consistent with a histidine being required to coordinate the metal ion. Second, the enzyme is not inhibited by metal chelating agents (Zwizinski et al., 1981).
Third, atomic absorption analysis shows that leader peptidase does not contain a zinc atom.* Although leader peptidase is not a standard serine protease or an aspartic acid protease, it does require several serine and aspartic acid residues for optimal activity. With an in vitro system, we found that serine 90 and serine 185, as well as aspartic acid 99,153,273, and 276, were essential for cleaving the M13 procoat protein to coat protein. In addition, using an in uiuo assay, we found that catalytic activity is completely lost when serine 90 and aspartic acid 153 are replaced with * G. Renkes and R. Dalbey, unpublished data. alanyl residues. In contrast, mutants S185A, D273A, and D276A have activity in this in vivo assay. This indicates that these latter serine and aspartic acid residues are not catalytic. Aspartic acid 99, which is critical for processing the substrate procoat (Table 11), plays a role in maintaining the native structure of leader peptidase. Pulse-chase studies (Fig.  4) established that the replacement of aspartic acid 99 with alanine results in a very unstable leader peptidase molecule. Therefore, it is unlike!y that this residue is directly involved in catalysis, but rather plays an important role in stabilizing the protein structure.
It is striking that the essential serine 90 of the E. coli leader peptidase is conserved in all known prokaryotic and eukaryotic signal peptidases." In prokaryotes, there are now four leader peptidases in which the amino acid sequence is known: E. coli (Wolfe et al., 1983a), Salmonella typhimurium (van Dijl et ul., 1990), Bacillus subtilis,: ' and Pseudomonas fluorescem (Black et al., 1992). In eukaryotes, there are four: S. cereuisiue mitochondrial peptidase (Behrens et al., 1991), the Sec 11 subunit from the endoplasmic reticulum signal peptidase (Bohni et al., 1988), and subunits Spc 18 and Spc 21 from the canine endoplasmic reticulum signal peptidase (Greenburg et al., 1989). In each of these enzymes, the serine is preceded by a long stretch of apolar residues that almost certainly anchor the protein to the membrane. Thus, the serine is spatially close to the membrane surface where it can attack the target sequences of preproteins as they emerge from the cytoplasmic side of the bilayer.
Our current working model is that leader peptidase belongs to a new class of serine proteases that does not require a histidine residue as a proton donor and acceptor. Although this is a reasonable hypothesis based on the site-directed mutagenesis and the sequence similarity results, we cannot rule out that the mutation of the serine may cause a conformational change which destroys the active site made up of other amino acids. It is also possible that other amino acids are involved in catalysis, other than serine and aspartic acid residues. However, recent studies strongly suggest that the serine is ~a t a l y t i c .~ We have been able to generate an active leader peptidase enzyme by replacing serine 90 with a cysteine. In contrast to the wild-type protein, this thiol leader peptidase is inactivated by N-ethylmaleimide, a cysteinespecific reagent. Current effort in this laboratory is centered around isolating a catalytically active leader peptidase comprising the periplasmic domain. The long term objective is to solve the structure by x-ray crystallography to determine which residues are at the active site.