The Type 4 Prepilin Peptidases Comprise a Novel Family of Aspartic Acid Proteases*

Type 4 prepilins or prepilin-like-proteins are secreted by a wide range of bacterial species and are required for a variety of functions including type 4 pilus formation, toxin and other enzyme secretion, gene transfer, and biofilm formation. A distinctive feature of these proteins is the presence of a specialized leader peptide that is cleaved off by a cognate membrane-bound type 4 prepilin peptidase (TFPP) during the process of secretion. In this report we show that the TFPPs represent a novel family of bilobed aspartate proteases that is unlike any other protease. The active site pairs of aspartic acids of the two TFPPs in Vibrio cholerae are found at positions 125 and 189 of TcpJ and 147 and 212 of VcpD. Corresponding aspartate residues are completely conserved throughout this extensive peptidase family.

Members of the family of leader peptidases known as type 4 prepilin peptidases (TFPP) 1 have been identified in a vast number of species of Gram-negative bacteria and an increasing number of Gram-positive bacteria. The TFPP is responsible for the cleavage and N-methylation (known collectively as processing) of the highly conserved type 4 leader peptide present at the N terminus of the families of secreted proteins known as type 4 prepilins and type 4 prepilin-like proteins (1). Whereas the cleavage of the type 4 leader peptide is a necessary step for secretion of the mature pilin, methylation of the N terminus of the mature pilin is not required for secretion and has no known function (2). Type 4 pilins are polymerized as the structural subunits of type 4 pili (Tfp), which are surface organelles required for diverse activities that contribute to broad attributes such as genetic transfer, virulence, and environmental persistence of a wide array of Gram-negative bacterial pathogens. Prototypical examples include microcolony formation mediated by toxin-coregulated pilus (TCP) of Vibrio cholerae, 2 bacterial surface dispersal mechanisms mediated by bundle forming pilus of enteropathogenic Escherichia coli (3), colonization and natural transformation mediated by PilE of Neisseria sp. (4), and gliding motility mediated by the Tfp of Myxococcus sp. (5).
The related type 4 pilin-like proteins partially comprise the main terminal branch of the general secretory pathway in Gram-negative bacteria, also known as the type 2 secretion pathway (6). The general secretory pathway is the pathway utilized by the majority of proteins that are secreted outside of Gram-negative bacteria. Such proteins include toxins such as cholera toxin of V. cholerae (7) and exotoxin A of Pseudomonas aeruginosa (8), as well as other enzymes such as pullulanase of Klebsiella pneumoniae (9) and hemolysin of Aeromonas hydrophila (10). Type 4 pilin-like proteins are also required for natural transformation of Bacillus subtilus (11) and Streptococcus pneumoniae (12).
The type 4 leader peptide is characterized by several features that are highly conserved, yet differ from those of standard leader peptides. The N-terminal leader that is cleaved from the mature protein is highly charged and immediately precedes a region of approximately 20 residues that are predominantly hydrophobic and are retained within the mature protein. The majority of type 4 prepilins or prepilin-like proteins are designated type 4a, containing leader peptides that are short (ϳ6 residues). Type 4b preproteins contain a longer leader (ϳ25 residues). The peptide bond on the prepilin that is hydrolyzed by the TFPP lies between the charged and hydrophobic domains, immediately C-terminal to an invariant glycine. The new N-terminal amino acid that arises upon processing is a N-methylated phenylalanine for type 4a proteins and is typically a N-methylated methionine for type 4b proteins. The TFPPs themselves are integral cytoplasmic membrane proteins with numerous transmembrane domains. The processing reaction has been postulated to occur on the cytoplasmic side of the membrane because of the membrane orientation of the prepilin, the location of all the major nonmembrane peptidase domains, and the likelihood that the cytoplasmically located Sadenosyl methionine would provide an available source of methyl groups for the methylation step of the reaction (13).
Previous studies on the proteolytic mechanism of the TFPPs have focused on the largest cytoplasmic region of the protein designated cytoplasmic domain 1. These studies were performed on PilD, a prototypic TFPP from P. aeruginosa. Domain 1 of the PilD peptidase contains two pairs of cysteine residues that are conserved among many members of the TFPP family. Alteration of the 4 highly conserved cysteine residues by substitution of alanine or serine residues resulted in up to an 80 -100% reduction of peptidase activity in an in vitro assay. However, the processing defect varied from approximately 0% to 80% for the various mutants in vivo, suggesting that although pilin cleavage was less efficient, it was not abolished (14). Chemical inhibitor studies supported the notion that the cysteine residues contributed to protease activity and PilD was classified as a cysteine protease, specifically as a member of the C20 family of cysteine proteases (15). The retention of partial activity by some of the pilD mutants as well as the subsequent discovery of TFPP's that lack the consensus cysteine residues have brought this mechanism of protease activity into question. For example, the XpsO TFPP of Xanthomonas campestris was characterized and shown to lack all the domain 1 cysteine residues, yet the xpsO gene fully complements a pilD deletion for prepilin processing (16). Some newer members of the rapidly growing family lack the entire domain 1 region. Analysis of a current TFPP alignment also reveals a complete lack of any conserved histidine residue, a necessary component of the cysteine protease catalytic mechanism (15). These observations have led to a revised classification of PilD, and therefore the complete family of TFPPs, to the U12 category of proteases with unknown catalytic mechanism (17).
In the study reported here, an approach similar to the comprehensive mutational strategy employed to identify the signal peptidase I active site (18) was used to determine the active site residues of TcpJ, the TFPP responsible for processing the TcpA type 4 prepilin of V. cholerae. A protein sequence alignment of the TFPP family revealed several highly conserved potential protease active site residues in addition to the two cysteine pairs that exist in this protease family. These include serine, aspartic acid, lysine, and glutamic acid residues, most of which are located in cytoplasmic domains of TFPPs. Mutations in tcpJ corresponding to these positions, as well as a deletion of the entire region encoding cytoplasmic domain 1, were constructed. In vivo activity and in vitro protease assays indicate that only Asp 125 and Asp 189 , which are completely conserved throughout the entire TFPP family, are absolutely required for protease activity in TcpJ. Chemical protease inhibition studies are consistent with this finding. The mutational analysis was extended to a second TFPP of V. cholerae, VcpD, which is required for toxin secretion (19). Taken together, the results lead to the conclusion that members of the TFPP family are bilobed aspartic acid proteases and due to their complete insensitivity to pepstatin should be regarded as a novel family of non-pepsinlike acid proteases. These findings allow the rational development of inhibitors with practical therapeutic value.
E. coli strains were grown in LB at 37°C. V. Cholerae strains were grown in LB, pH 6.5, at 30°C to induce expression of TCP. Antibiotics were added to the media when needed at the following concentrations: 100 g/ml ampicillin, 45 g/ml kanamycin, 34 g/ml chloramphenicol.

Preparation of Membranes Containing TcpJ Prepilin Peptidase or
TcpA Prepilin-The method for membrane preparation was adapted from a previously described cell fractionation protocol (25). A 100-ml overnight culture was placed on ice for 20 min, and the cells were then collected by centrifugation at 5000 ϫ g for 10 min. The cell pellet was resuspended in 1 ml of 200 mM Tris, pH 8.0, to which 1 ml of 50 mM Tris, pH 8.0, 1 M sucrose, 2 ml of H 2 O, 20 l of 0.5 M EDTA, and 20 l of lysozyme (10 mg/ml) were added. The cell suspension was allowed to incubate on ice for an additional 30 min. 100 l of 1 M MgSO 4 was added, and the suspension was centrifuged at 5000 ϫ g for 10 min. The cell pellet is resuspended in 5 ml of 50 mM Tris, pH 8.0, sonicated 3 ϫ 15 s at 50% duty, and centrifuged at 5000 ϫ g for 10 min. The supernatant was then centrifuged at 230,000 ϫ g (60,000 rpm in a TLA 100.3 rotor) for 15 min. The resulting pellet, which contained the total membrane fraction, was resuspended in 200 l of 50 mM Tris, pH 8.0. TcpA Prepilin Quantitation-Membrane preparations were prepared from 42°C grown overnight cultures of E. coli K38, which does not express TcpA, and MK90, which overexpresses TcpA prepilin at elevated temperature. 20 l of a membrane preparation from K38 and from MK90 were subjected to electrophoresis on a 12.5% SDS-PAGE gel alongside a standard of 3 g of trypsin, followed by staining with Coomassie Blue. The stained gel was dried using the Novex gel drying system. The dried gel was scanned by a Molecular Dynamics Personal Densitometer SI, and densitometry analysis was performed using Im-ageQuant version 1.2 software. The intensities (measured in pixels) of the 3-g trypsin band, the TcpA prepilin band (MK90), the corresponding location of the TcpA prepilin band in the TcpA-negative control lane (K38), and an approximately 28-kDa band from both K38 and MK90 were measured while adjusting to a single background value. The mass of the TcpA prepilin present in the MK90 lane was determined by first subtracting the intensity of the corresponding area in the K38 TcpA prepilin-lacking lane. This corrected for the staining intensity of minor protein bands that co-migrate with TcpA prepilin. By comparing the intensity values of the 28-kDa bands from the TcpA-positive and TcpAnegative lanes, a further adjustment was made to the TcpA prepilin value to account for any loading difference between the lanes. The trypsin band intensity was divided by 3 g to determine an intensity/g value, which was applied to the final TcpA prepilin band value to determine the mass of TcpA prepilin/l in the membrane preparation. The TcpA prepilin membrane preparation was diluted with 50 mM Tris, pH 8.0, buffer to a concentration of 0.2 g/l.
In Vitro Cleavage Assay-The in vitro TcpJ cleavage assay was designed, based on the assay developed for PilD of P. aeruginosa (26), except that pre-TcpA was provided in a membrane preparation rather than as purified protein. TcpJ derivatives were provided in membrane preparations of the TcpJ-expressing strain JM109 (pCL9) (wild-type and mutant alleles). The 100-l processing reaction was prepared by combining a 50-l substrate fraction with a 50-l enzyme fraction (an equivalent of 1 unit of wild-type activity) and incubated at 37°C for 1 h. The substrate fraction was prepared by combining 5 l of the TcpA prepilin-containing membrane preparation (1 g of TcpA prepilin), 10 l 0.5% (w/v) cardiolipin, 20 l of 5ϫ assay buffer (125 mM triethanolamine HCl, pH 7.5, 2.5% v/v Triton X-100), and brought up to the total volume with H 2 O. The enzyme fraction was prepared by combining a standardized volume of TcpJ-containing membrane preparation and H 2 O to bring the volume to 50 l. The cleavage reaction was stopped by the addition of 100 l of 2ϫ protein sample buffer. The volume of TcpJ preparation for all TcpJ derivatives was standardized to be equivalent to the volume containing an amount of wild-type TcpJ sufficient to cleave 50% of 1 g of TcpA prepilin in 1 h, which is defined as 1 unit of TcpJ peptidase activity. The formation of mature TcpA from precursor was monitored by Western immunoblot analysis.
Chemical Protease Inhibitors-Chemical inhibitors were tested for their ability to prevent the cleavage of TcpA from the prepilin to mature form in the in vitro cleavage assay. Various concentrations of inhibitors were added to both the enzyme (1 unit of TcpJ) and substrate fraction so that, when combined, the inhibitor concentration would remain constant. The inhibitor was incubated in the enzyme fraction at room temperature for 30 min prior to the combination of the two fractions. The 100-l reaction was allowed to proceed for 1 h at 37°C before 100 l of 2ϫ SDS protein sample buffer was added to stop the reaction. The specific concentrations of each inhibitor are listed in Table I. The 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDAC)/glycinamide inhibition protocol is a two-step chemical reaction based on the procedure developed for the selective modification of carboxyl side groups of aspartic and glutamic acids in proteins (27). Takahashi et al. (28) first suggested the use of this procedure as a method to specifically inhibit acid proteases. Membranes containing the equivalent of 6 units of TcpJ activity are incubated in 80 l of 25 mM potassium biphthalate, pH 4.0, and 100 mM EDAC for 30 min at room temperature. 20 l of 1 M glycinamide was added to the mixture, and incubation at room temperature was continued for an additional 30 min before the reaction was stopped by the addition of 150 l of 200 mM Tris buffer, pH 8.0. The mixture was spot dialyzed against 50 mM Tris buffer, pH 8.0, for 1 h and then centrifuged at 230,000 ϫ g (60,000 rpm in TLA 100.3 rotor) for 15 min. The resulting membrane pellet, which contains the chemically modified TcpJ was resuspended in 60 l of 50 mM Tris, pH 8.0. 20 l of the membrane suspension, which contains 2 units of TcpJ, was used in the in vitro cleavage assay to determine the peptidase activity. Activity of the modified TcpJ was compared with an equivalent amount of TcpJ, which had been subjected to a mock modification procedure that did not include EDAC or glycinamide.
Mutagenesis of tcpJ-The TcpJ S46A, C76A, D189A, and K191A alterations were constructed in pCL10 by the Quikchange TM (Stratagene) method of mutagenesis, which utilizes inverse PCR primed by divergent overlapping primers containing the desired complementary nucleotide changes. The S18A, C48A, C51A, S65A, C73A, S81A, E88A, D125A, D125E, D125N, D125Amb, S172A, D183N, D189E, D189N, D189Amb, S212A/S213A, and ⌬35-81 were constructed in pCL10, and D147A and D212A were constructed in pJM294 using other methods of inverse PCR primed by nonoverlapping divergent primers. Both primers contained an engineered common restriction site that allowed for the restriction of the PCR product by the appropriate enzyme followed by the ligation at the complementary DNA overhangs. In these cases, only one primer carried the desired mutation that was incorporated into the amplified DNA. In some cases the engineering of restriction sites into the primers was accompanied by the incorporation of silent mutations. In other primers, seamless cloning and mutagenesis was performed by incorporating the restriction enzyme site EarI or SapI at the 5Ј end of each primer in the pair. These enzymes cleave just outside their recognition site, which makes possible a primer design that ensures that the recognition site is cleaved off the end of the PCR product, while complementary gene sequence overhangs remain to promote annealing prior to ligation.
Immunoblot Analyses-In vivo TcpA processing was monitored in E. coli and V. cholerae strains grown to mid-log phase except where indicated. 0.5 ml of cell culture was centrifuged and resuspended in 100 l of LB. 100 l of 2ϫ SDS-PAGE sample buffer was added to the sample.
The total sample was boiled for 5 min, and 50 l was subjected to electrophoresis on a 12.5% SDS-PAGE gel. For both in vivo and in vitro processing, the proteins were electroblotted to nitrocellulose and probed with anti-TcpA peptide 6 antiserum. The secondary Ab, an anti-rabbit IgG conjugated to horseradish peroxidase, was used to detect TcpA by chemiluminescence using ECL reagents (Amersham Pharmacia Biotech), followed by autoradiography.
For detection of TcpJ, 20 l of a membrane preparation from a TcpJ-expressing strain was combined with 20 l of 2ϫ SDS-PAGE sample buffer and incubated at 37°C for 15 min. 20 l of the sample was then subjected to electrophoresis on a 12.5% SDS-PAGE gel. Proteins were transferred to nitrocellulose and probed with anti-TcpJ peptide J3-1 antiserum or, in the case of TcpJ (6)His and TcpJ (6)His⌬35-81, with TetraHis Ab (Qiagen). Detection with secondary antibody was carried out as described for TcpA.

A Limited Set of Potential Peptidase Active Site Residues Are
Conserved among TFPPs-In order to characterize the biochemical mechanism of the cleavage reaction catalyzed by TcpJ and other TFPPs, we sought to identify the critical active site amino acids. First, a limited set of candidate active site residues were identified for targeted changes. Since only the amino acids cysteine, serine, aspartic acid, histidine, lysine, or glutamic acid have been identified as critical active site residues in proteases, an alignment of 27 TFPPs was used to quantitate the conservation of these residues (Fig. 1). Since the active site is thought to be located on the cytoplasmic face of the cellular membrane, the analysis was further refined to focus on those residues located within the three major theoretical cytoplasmic domains of TcpJ. Twelve conserved positions were identified, including eight in cytoplasmic domain 1, one in domain 2, and three in domain 3 as indicated by asterisks ( Fig. 2A). Of these residues, only Asp 125 and Asp 189 are present in every single TFPP examined. The positions of all potentially catalytic residues with respect to the membrane topology model of TcpJ are FIG. 1. Percentage of conserved homology of potential protease active site residues from 27 TFPP homologs. All potential protease active site residues (Cys, Ser, Asp, Lys, His, Glu) in TcpJ were analyzed for their percentage of homology in a protein sequence alignment of the TFPP family. Potential protease active site residues that are the major conserved amino acids at a position in the peptidase family alignment but are not conserved in TcpJ are enclosed in parentheses and are designated with respect to the relative position in the TcpJ protein sequence. When a residue in TcpJ and a conserved residue not found in TcpJ exist at the same relative position, then both are indicated by a sectored bar with the TcpJ portion represented by the lower portion of the bar. The remaining portion of the bar represents the conserved residue not in TcpJ. Predicted membrane topology is indicated by C (cytoplasmic), P (periplasmic), or M (transmembrane) below each residue. Gray bars indicate the conserved residues that have been altered during this study. Alignment was done in DNASTAR MegAlign, using the Clustal method of alignment. indicated in Fig. 2B. 16 residues including the 7 serines, 4 cysteines, 3 aspartic acids, 1 lysine, and 1 glutamic acid indicated as shaded residues in Fig. 1 and 2B were selected for mutagenesis. 13 of the 16 potential active site residues targeted for mutagenesis are all of the residues conserved at greater than the arbitrary level of 50% in TcpJ. The three selected residues that did not meet this criterion are Ser 172 , Asp 183 , and Ser 212 . Ser 172 of TcpJ may be functionally equivalent to residues Ser 168 and Ser 171 indicated as (S168) and (S171) in Fig. 1, which, in combination, are approximately 65% conserved. Asp 183 and Glu 183 (Fig. 1, (E183)) in combination are conserved in nearly 80% of the 27 TFPPs. Aspartic acid and glutamic acid can function equivalently as the active site of an acid protease (29). Mutagenesis of Ser 212 was required due to the construction of the S212A/S213A double mutant, which was necessary so the analysis of the phenotype of the S213A would not be ambiguous due to the remaining serine in the adjacent position.
FIG. 2. Primary sequence alignment of the three major cytoplasmic domains of the TFPP family and topology relative to TcpJ. A, partial protein sequence alignment of 27 TFPP homologs. The first region shown corresponds approximately to the large cytoplasmic domain 1. The following two regions correspond precisely to the predicted residues, which constitute cytoplasmic domains 2 and 3. Asterisks designate positions in these regions for which mutations were constructed for this study. The alignment was done in DNASTAR MegAlign, using the Clustal method for alignment. B, membrane topology model of TcpJ. The predicted topology of TcpJ was deduced by comparing the hydrophobicity profile of TcpJ with the ␤-lactamase fusion data of Erwinia carotovora OutO (36) and the alkaline phosphatase fusion data of P. aeruginosa PilD (37). Cytoplasmic domains 1, 2, and 3 are indicated. All potential protease active site residues (Cys, Ser, Asp, Lys, His, Glu) have been designated. Gray shaded residues indicate each position that has been altered during this study. Active site residues are additionally indicated with a star.
Asp 125 and Asp 189 Are the Sole Residues Required for Peptidase Activity-Alteration of the targeted residues was performed by site-directed mutagenesis of the 16 selected positions in tcpJ present on plasmid pCL10, which also carries tcpA encoding the TcpA type 4 prepilin that is the cognate substrate of TcpJ. Mutant and wild-type constructs were introduced into E. coli JM109, and total cell protein extracts from midlog cultures were analyzed for TcpJ protease activity by Western immunoblot detection of precursor and mature forms of TcpA (Fig. 3A). TcpJ proteins with alterations S18A, S46A, C51A, S65A, C73A, C76A, S81L, E88A, S172A, D183N, and K191A exhibited peptidase activity identical to wild-type TcpJ, whereas C48A displayed an intermediate level of peptidase activity. The four mutants D125A, D125N, D189A, and D189N completely lacked all peptidase activity, as would be expected when an active site residue of a protease is converted to a nonfunctional substitute. The D125E and D189E mutants retain the carboxylic acid moiety at the R-group terminus and both retain partial peptidase activity. Processing by S212A/ S213A double mutant is not shown because it was later determined that the mutant TcpJ is unstable.
In order to test the peptidase activity expressed by the tcpJ mutant constructs in V. cholerae, they were introduced into V. cholerae strain CL381, an O395 derivative with an insertional disruption in tcpJ and an in-frame deletion of vcpD, which render it completely devoid of TcpA processing activity. The resultant strains were grown under TCP-inducing conditions (LB, pH 6.5, 30°C) and total cell protein extracts were examined by Western immunoblot analysis of TcpA (Fig. 3B). The CL381 double mutant was utilized for this analysis, because disruption of tcpJ alone does not lead to a complete loss of TcpA processing, as evidenced by the detection of some processed TcpA in the J71K-1 extract. The CL381 extract shows that the residual processing is due to VcpD, a second TFPP present in V. cholerae. Similar to the results seen in JM109, tcpJ mutants S18A, S46A, C51A, S65A, C73A, C76A, S81L, E88A, S172A, D183N, and K191A exhibited peptidase activity identical to wild-type TcpJ while C48A possessed an intermediate peptidase activity. D125A, D125N, D189A, and D189N all completely lacked peptidase activity, whereas the D125E and D189E mutants restored peptidase activity, providing further support for the assignment of Asp 125 and Asp 189 as the active site residues.
To further extend the number of different amino acid substitutions analyzed at the 125 and 189 positions of TcpJ, amber (TAG) codons were individually engineered at each corresponding position on the pCL10 plasmid and the resultant constructs were introduced into Amber-Lys, Amber-Leu, Amber-Gln, Amber-Ser, and Amber-Tyr suppressor strains of E. coli. The strains were grown to mid-log phase, and whole cell protein extracts were examined for cleavage of TcpA by Western immunoblot analysis (Fig. 3C). No amino acid substitution at either position 125 or 189 exhibited any peptidase activity. The protein levels and membrane localization of the altered TcpJ proteins were all comparable to wild-type TcpJ.
TcpJ peptidase activity was also be measured by an in vitro assay modeled after the in vitro peptidase assay devised for PilD of P. aeruginosa (26). TcpA-containing membranes from MK90 and TcpJ-containing membranes from CL318 were combined in the in vitro assay. A series of assays with an increasing volume of TcpJ-containing membranes was used to determine the volume of membrane preparation corresponding to 1 unit of TcpJ activity, which is the amount required to cleave 50% of 1 g of TcpA prepilin into the mature form in 1 h (Fig. 4A).
The peptidase activity of proteins encoded by a subset of the tcpJ mutants was determined in the in vitro cleavage assay. TcpJ proteins with alterations S46A, C51A, S65A, C73A, and C76A were all capable of cleaving TcpA prepilin in the in vitro assay. C48A, the double mutant S212A/S213A, and the Ala, Glu, and Asn changes at positions 125 and 189 abolished all peptidase activity (Fig. 4B). These results are consistent with the in vivo analysis except for the C48A, D125E, and D189E proteins, which exhibited intermediate peptidase activity in vivo had no peptidase activity in vitro. This is not surprising since studies on PilD of P. aeruginosa have shown that the in vitro assay is a more stringent test of peptidase activity than the in vivo conditions (14). The double mutant S212A/S213A has no peptidase activity due to the fact that the enzyme is unstable as shown by the lack of any TcpJ present in TcpJ immunoblot. All other TcpJ derivatives were detected at levels approximately equal to wild-type except for C48A and C51A, which gave faint or undetectable TcpJ signals. This is likely due to the anti-TcpJ antiserum, which is directed against a peptide that encompasses Cys 48 and Cys 51 . It is likely that C48A and C51A mutants alter the epitope in TcpJ sufficiently so that the protein is not detected by the antiserum.
Cytoplasmic Domain 1 Is Not Required for TcpJ Peptidase Activity-In order to definitively determine whether the cysteine residues or any portion of cytoplasmic domain 1 is required for TFPP cleavage activity, an internal in-frame deletion corresponding to positions 35-81 of TcpJ was constructed. Since our TcpJ antibody is directed against a peptide within the region removed by the deletion, the construction was made in the 6-His epitope-tagged version of tcpJ contained on plasmid pCL11, yielding plasmid pCL11⌬35-81. TcpJ (6)His and the deletion derivative expressed from pCL11 and pCL11⌬35-81 were both able to cleave TcpA to equivalent extents in E. coli (Fig. 3A), V. cholerae (Fig. 3B), or in vitro (Fig.  4B). Interestingly, TcpJ (6)His⌬35-81 was found to process TcpA efficiently enough to restore piliation to the tcpJ mutant strain, J71K-1 (Fig. 5).
VcpD Requires a Pair of Aspartic Acid Residues Corresponding to Those Necessary for TcpJ Activity-In order to determine the extent to which the TcpJ requirement for the aspartic acid residue pair can be generalized to other members of the TFPP family, the corresponding changes of D147A and D212A were made in VcpD. The VcpD TFPP is responsible for processing type 4 prepilins such as those involved in mannose-sensitive pilus formation and processes the extracellular protein secretion type 4 prepilin-like proteins required for the secretion of toxin and other enzymes by V. cholerae. VcpD can also cleave TcpA prepilin, although the cognate TFPP for TcpA is TcpJ (19). Wild-type, D147A, and D212A forms of VcpD expressed from plasmid pJM294 derivatives were introduced into E. coli JM109 carrying the tcpA-expressing plasmid pRTH3G7. Strains were grown to midlog phase, whole cell protein extracts were made prepared, and the extracts were examined by SDS-PAGE and Western immunoblot analysis with anti-TcpA antiserum (Fig. 6A). Wild-type VcpD cleaved TcpA prepilin completely under these conditions, whereas the D147A and D212A mutant derivatives were totally defective. Protein levels of VcpD in the four strains were determined by Western immunoblot analysis using the anti-TcpJ antiserum, which crossreacts with VcpD. The D147A and D212A forms of VcpD are present at levels similar to those in wild-type VcpD. An anal-ogous result was seen with the cleavage of the type 4 prepilinlike protein substrate of VcpD, EpsI (Fig. 6B). These experiments demonstrate that alanine substitutions at Asp 147 and Asp 212 , the aspartic acid residues in VcpD that correspond to protease active site Asp 125 and Asp 189 of TcpJ, completely eliminate peptidase activity for multiple substrates.
Protease Inhibitor Studies of TcpJ Activity Support the Essential Role of the Aspartic Acid Residues for Protease Activity-Utilizing the TcpJ in vitro cleavage assay, a number of chemical protease inhibitors were examined for their ability to block the peptidase activity of TcpJ (Table I). Two or more inhibitors specific to each class of protease family were tested for their ability to prevent TcpJ peptidase activity. The majority of the inhibitors had little or no effect on the peptidase activity, as measured by the comparison of cleaved TcpA produced in the in vitro assay in the presence or absence of inhibitor. All inhibitors were tested at the maximum suggested concentration except for the EDAC/glycinamide inhibition protocol, for which it was not required. Of all the common protease inhibitors used, only NEM inhibited cleavage greater than 50%. This agrees with the strong inhibitory effect of NEM on PilD of P. aeruginosa (14). In contrast, a combination of EDAC and glycinamide, which is known to modify aspartic acid residues and is the only method known to inhibit the non-pepsintype acid proteinases such as proteinase A from Aspergillus (28), completely inhibited TcpJ peptidase activity. A 20 -100fold excess of the maximum suggested concentration of pepstatin only exhibited a minor inhibitory effect on peptidase activity, thus ruling out a pepsin-like active site in TcpJ.

DISCUSSION
The action of a family of leader peptidases known as the TFPPs is required for extracellular protein secretion, genetic exchange, and organelle formation in a wide array of bacterial systems. Their involvement in the processing of prepilins to mature forms is a prerequisite to type 4 pilin secretion and assembly of the pilus structure. Other secreted proteins that have type 4 leader peptides, but are not pilin subunits, are referred to as type 4 prepilin-like proteins. Such proteins are components of general secretory pathways involved in protein secretion and pilus biogenesis, the natural transformation pathways, and conjugal plasmid DNA transfer pathways.
Despite the probable ubiquitous presence of TFPPs among bacterial species, identification of the protease active site has been elusive. This may have been due in part to the insensi-tivity of TFPPs to commercially available protease inhibitors. In addition, none of the consensus protease active site motifs present in the vast majority of other protease families are obvious in this peptidase family. For example, the typical catalytic residues of serine proteases are the three amino acids serine, aspartic acid, and histidine, but several families of serine proteases such as the signal peptidase I family utilize only serine and lysine as the catalytic residues (30). The most conserved histidine residue in the TFPP family, His 30 , is only present in 30% of the members (Fig. 1, (H30)), thus ruling out involvement of the S/D/H motif in the catalytic mechanism. The numerous highly conserved serines as well as Lys 191 , conserved in nearly 80% of the TFPP family members, raised the possibility that the proteolytic activity might involve a Ser/Lys catalytic mechanism. A further possibility was that the TFPP could be a cysteine protease. This potential mechanism was suggested by the pilD mutational analysis of the 4 highly conserved cysteines in the peptidase family. Mutations at the 4 cysteines failed to completely eliminate peptidase activity in vivo, but in an in vitro assay these altered proteins were nearly completely deficient of peptidase activity (14). However, since all cysteine proteases require both a cysteine and a histidine as the catalytic dyad (15), the lack of a highly conserved histidine in the TFPP family excludes any known cysteine catalytic mechanism. The possibility of an aspartic acid protease mechanism is strongly suggested for the TFPP family due to the presence of two completely conserved aspartic acid residues corresponding to TcpJ positions 125 and 189 in domains 2 and 3, respectively (Fig. 2B). However, Asp 125 and Asp 189 do not exist in the motif D(T/S)G, which is common to the majority of aspartic acid proteases (31). In addition, the pH optimum, which has been determined to be near neutral for TcpJ (data not shown) and PilD of P. aeruginosa (26), is in contrast to the highly acidic pH optimum of the majority of aspartic acid proteases (32). The final protease family comprises the metalloproteases, for which over half contain the active site motif, HEXXH (32). The TFPP family lacks any homology to the HEXXH metalloprotease motif. In light of this sequence alignment analysis, an aspartic acid protease mechanism was deemed to be the most probable correct assignment for TFPPs. However, all potential protease active site residues conserved at greater than 50% were targeted for mutagenesis in order to determine the active site with certainty.
The results from the mutational analysis of TcpJ at the 16 selected positions clearly demonstrate that only the two aspartic acids corresponding to positions 125 and 189 of TcpJ are essential for protease activity. Substitution of the amino acids alanine, asparagine, leucine, serine, tyrosine, glutamine, and lysine into positions 125 and 189 of TcpJ produced a stable and properly localized enzyme that lacked any protease activity. The cause for this loss of protease activity of the mutants at positions 125 and 189 was determined to be the loss of the carboxylic acid group of the aspartic acid. The carboxylic acid functional groups of the active site aspartic acids directly participate in the general acid-base catalysis that causes the hydrolysis (cleavage) of the peptide bond (32). When the carboxylic acid functional group was restored at positions 125 or 189 by the substitution of glutamic acid, the protease activity was also restored. The additional one carbon in length of the glutamic acid R-group was less tolerated at position 189 than 125, suggesting that the location in space of the R-group carboxylic acid at position 189 is critical to protease activity. It can be inferred from the constraint at position 189 that the carboxylic acid group of Asp 189 directly participates in the initial chemical reduction of the prepilin peptide bond while Asp 125 contributes to the hydrolysis through interaction with a water molecule. FIG. 6. Mutations that alter the predicted protease active site aspartic acid residues in VcpD eliminate peptidase activity. A, Western immunoblot analysis using anti-TcpA antisera to detect cleavage levels of TcpA in JM109 by VcpD mutants. All strains are JM109 and carry the tcpA-expressing plasmid, pRTG7H3. The VcpD negative strain contains no other plasmid. The wild-type and mutant alleles of vcpD are expressed from pJM294 derivatives. A Western immunoblot analysis showing levels of VcpD for each strain is also shown. B, Western immunoblot analysis using Tetra-His Ab to detect cleavage levels of EpsI(6)His in JM109 by VcpD derivatives. All strains are JM109 and carry the EpsI(6)His-expressing plasmid, pJM290 and vcpD derivatives as in A.
Demonstration that the complete loss of protease activity for all amino acid substitutions at positions 125 and 189 was not simply due to a conformational change was also addressed by the asparagine and serine substitutions at these positions. Asparagine and serine have small polar R-groups similar to the R-group of aspartic acid and thus would likely maintain the wild-type conformation of the enzyme.
Only one of the altered forms of TcpJ, the double mutant S212A/S213A form, was unstable and could not be assessed for protease activity. The replacement of both hydrophilic serine residues with hydrophobic alanine residues occurs at a loop region between transmembrane domains and likely prevented the proper integration of the mutant peptidase into the membrane, resulting in its degradation. In lieu of a cleavage activity result of the S212A/S213A mutant, other evidence indicates a protease active site serine does not exist at position 212 or 213. Ser 212 and Ser 213 are both located on the periplasmic side on the inner membrane and could not participate in a protease reaction occurring on the cytoplasmic side on the membrane. Additionally, the only possible catalytic partners are Lys 191 , which is not required for protease activity or Lys 180 , which if considered functionally equivalent to Lys 182 (Fig. 1, (K182)) would only be found in approximately 60% of peptidases.
Conclusive evidence that no amino acid residue between position 35 and 81 is required individually, or in combination, for peptidase activity or for TCP biogenesis was provided by the internal deletion in TcpJ that removed the majority of cytoplasmic domain 1. This eliminates the possibility that two or more of the four conserved cysteines act collectively in some manner as a protease active site cysteine. The deletion also demonstrates that domain 1 is not in any way required for substrate recognition, binding, orientation, or any function that leads to the assembly of the pilin into a pilus structure.
Evidence for conservation of the protease active site among multiple members of the TFPP family was provided by mutations that altered either of the two aspartic acids of VcpD corresponding to the active site residues in TcpJ. These alterations abolished VcpD activity, thus strongly suggesting that all TFPPs share the aspartic acid active site. The result with VcpD was of further importance because its cognate targets are within the type 4a subgroup of type 4 prepilins or prepilin-like proteins, which have a short signal peptide of typically 5-8 amino acids. In contrast, the cognate substrate of TcpJ is TcpA, which is a type 4b prepilin with a 25-amino acid leader peptide.
The chemical protease inhibitor studies served to confirm the mutational analysis that determined that TcpJ is not a cysteine protease, serine protease, metalloprotease, or a typical aspartic acid protease. The protocol using EDAC and glycinamide strongly implicates the dependence on aspartic acids for activity of the enzyme. Inhibition by the EDAC/glycinamide protocol also suggests that the active site aspartic acids are accessible to chemical inhibitors and not shielded by the protein conformation of the cytoplasmic domains. The only protease inhibitor other than EDAC/glycinamide that inhibited at greater than 50% was NEM, which acts by modifying thiol groups of the cysteine R-groups of cysteine proteases. The partial inhibition of TcpJ could occur due to the modification of Cys 48 , which when altered by mutation to an alanine, caused a comparable loss of protease activity. This would suggest that, when Cys 48 is not in its natural form, it can lead to a conformational change in the enzyme that partially prevents the functioning of the protease active site.
The TFPPs differ from the majority of aspartic acid proteases in that the active site aspartic acids are not found in the D(T/S)G motif, the optimum pH for in vitro activity is near neutral as opposed to pH 2-4, and peptidase activity is not inhibited by pepstatin. Other aspartic acid proteases exist that possess these unusual traits individually. Signal peptidase II, an aspartic acid protease that cleaves prolipoproteins, does not possess the D(T/S)G motif at the active site (33). The human aspartic acid protease renin has a pH optimum range of 5.5-7.5 (34). Pseudomonas sp. 101 carboxyl proteinase is a bacterial aspartic acid protease that is insensitive to pepstatin (35). Aspartic acid proteases, which are insensitive to pepstatin, are often referred to as non-pepsin-like acid proteases (28). The majority of aspartic acid proteases like pepsin and human The effect of chemical inhibitors on the in vitro TcpJ cleavage assay The effect of chemical protease inhibitors on the peptidase activity of TcpJ was determined by incubating the inhibitor with an amount of membrane preparation known to contain 1 unit of activity of TcpJ for 30 min at room temperature. The TcpJ/inhibitor mixture is then tested for peptidase activity in the in vitro processing assay (see "Materials and Methods"). Amounts of cleaved TcpA prepilin were quantitated and applied to the formula: % cleavage ϭ (amount of cleaved TcpA in inhibition assay/amount of cleaved TcpA in no inhibitor control assay) ϫ 100. The inhibition by EDAC/glycinamide is a two-step protocol conducted in acidic conditions (see "Materials and Methods").
Chemical(s) Specificity Suggested concentration Concentration Cleavage immunodeficiency virus type 1 protease are bilobed molecules in which the active site cleft is located between the lobes with an active site aspartic acid contributed from each lobe. In this respect, the TFPPs display a remarkable similarity. The active site can be envisioned as a cleft between two lobes corresponding to domains 2 and 3 with one active site aspartic acid residue in each lobe. The comprehensive mutational analysis and chemical inhibitor studies that led to the discovery of the active site of TcpJ are ultimately most significant in that they allow for the prediction of the protease active site of the entire TFPP family. The identity of the active site, as well as the ability to specifically target the active site aspartic acids by a chemical protocol, provide the foundation for the development of specific protease inhibitors of all members of the TFPP family. Such inhibitors would have substantial antimicrobial effects on a wide range of pathogenic bacteria.