Cys 32 and His 105 Are the Critical Residues for the Calcium-dependent Cysteine Proteolytic Activity of CvaB, an ATP-binding Cassette Transporter*

CvaB, member of the ATP-binding cassette transporter superfamily, is the central membrane transporter of the colicin V secretion system in Escherichia coli . Cys 32 and His 105 in the N-terminal domain of CvaB were identified as critical residues for both colicin V secretion and cysteine proteolytic activity. By inhibit-ing degradation with N -ethylmaleimide and a mixture of protease inhibitors, a stable wild-type N-terminal domain (which showed cysteine protease activity when activated) was purified. Such protease activity was Ca 2 (cid:1) - and concentration-dependent and could be inhibited by antipain, N -ethylmaleimide, EDTA, and EGTA. At low concentrations, the Ca 2 (cid:1) analogs Tb 3 (cid:1) and La 3 (cid:1) (but not Fe 3 (cid:1) ) significantly enhanced proteolytic activity, suggesting that the size of the cations is important for activity. Together with comparisons of the sequences of members of the cysteine protease family, these results indicate that Cys 32 and His 105 are the critical residues in the CvaB N-terminal domain for the calcium-dependent cysteine protease activity and secretion of colicin V. ATP-binding cassette (ABC) 1 transporters bind ATP and utilize the energy released to transport molecules across the cell membrane (1, 2). These transporters constitute one of the largest protein families found in living organisms. Members of this family transport a wide variety of compounds, including sugars, ions, antibiotics, and peptides (3, 4). Some members of

ATP-binding cassette (ABC) 1 transporters bind ATP and utilize the energy released to transport molecules across the cell membrane (1,2). These transporters constitute one of the largest protein families found in living organisms. Members of this family transport a wide variety of compounds, including sugars, ions, antibiotics, and peptides (3,4). Some members of this family, such as the multidrug resistance protein, P-glycoprotein (which is involved in pumping out anticancer drugs), and the cystic fibrosis transmembrane conductance regulator (which, when defective, leads to this deadly inherited disease), are clinically relevant (5,6). Most ABC transporters consist of two hydrophobic transmembrane domains, which contain 6 -12 membrane-spanning ␣-helices and provide the specificity for the substrates that cross the membrane. They also contain two cytosolic nucleotide-binding domains, which bind and hydro-lyze ATP to provide the energy for translocating substrates (1). A highly conserved region among all ABC transporters is found within the nucleotide-binding domain (ϳ30% identity), which contains Walker A and B motifs separated by ϳ90 -120 amino acids and a signature C motif located upstream of the Walker B site (7,8).
Well studied examples of ABC transporters in prokaryotes are the histidine permease complex of Salmonella typhimurium HisQMP 2 (9) (its ATP-binding subunit HisP has been crystallized (10)), the maltose-binding protein MalEFGK 2 (11) (the crystal structure of its MalK subunit has been solved (12)), the ␣-hemolysin transporter (13), and colicin V (14). Hemolysin B and its associated components transport the toxin hemolysin A across both membranes of Escherichia coli (15). The colicin V system consists of a toxin gene (cvaC) that encodes a 103-amino acid precursor protein, the first 15 residues of which are cleaved behind two glycine residues of the leader peptide during secretion (16,19). A second gene (cvaB) that encodes the central ABC membrane transporter protein is predicted to have an N-terminal cytoplasmic domain that is conserved among bacteriocin ABC transporters; it has been proposed to possess the protease activity that cleaves the leader peptide (14, 18 -20). The N-terminal domains of the similar Lactococcus bacteriocin transporters have been proposed to be essential for substrate recognition and processing (19).
Previous studies have shown that processing ColV-1 in membrane vesicles is dependent on the CvaA-CvaB transporter and TolC proteins and that this activity is sensitive to inhibition by the serine/cysteine protease inhibitor antipain and the cysteine protease inhibitor N-ethylmaleimide (NEM) (21,22). Because neither CvaA nor ColV-1 contains a cysteine residue (14,23), CvaB is probably the target of inhibition of NEM and antipain. Sequence similarities indicate that the CvaB N-terminal domain (BntD) belongs to the cysteine protease calpain family. The catalytic residues in this family are believed to be Cys and His (24 -26). In this study, we identified the residues of BntD that are critical for colicin V secretion, and we show that wild-type BntD (but not certain Cys or His BntD mutants) possesses calcium-dependent cysteine peptidase activity.

EXPERIMENTAL PROCEDURES
Media and Growth Conditions-Both liquid and solid (with 1.5% agar) TB media (10 g/liter Tryptone and 8 g/liter NaCl) were used as growth media for transformation, activity assay, and protein overproduction. Ampicillin and chloramphenicol were used at final concentrations of 100 and 30 g/ml, respectively.
Bacterial Strains, Plasmids, and Reagents-The strains and plasmids used in this study are listed in Table I pTrcHis2B with a His 6 tag at the C terminus was obtained from Invitrogen. E. coli strains DH5␣ and BL21 were used as the bacterial hosts. Restriction enzymes and T4 DNA ligase were obtained from Roche Applied Science or New England Biolabs, Inc. (Beverly, MA) and were used essentially as recommended by the manufacturers. Complete protease inhibitor mixture tablets were obtained from Roche Applied Science. All other chemicals were reagent-grade and were purchased from Sigma unless otherwise noted.
Site-directed Mutagenesis-All mutageneses were carried out by PCR using oligonucleotides containing the desired mutation. The standard PCR contained 20 ng of linearized plasmid template, 400 ng of each oligonucleotide primer, 200 M each deoxynucleotide triphosphate, and 5 units of ExTaq DNA polymerase (PanVera/TaKaRa, Madison, WI) in 1ϫ buffer provided by the manufacturer in a total volume of 50 l. The PCR mixture was amplified by 35 cycles of 20 s at 94°C, 20 s at 50°C, and 1 min at 70°C, followed by one cycle of 10 min at 72°C. Following the reaction, PCR amplification products were purified using QIAquick PCR purification kits (QIAGEN Inc., Valencia, CA), resolved by electrophoresis on a 1% agarose gel in 40 mM Tris acetate and 1 mM EDTA in the presence of ethidium bromide, and isolated from the gel using a QIAquick gel extraction kit (QIAGEN Inc.).
To generate the desired site-directed mutagenesis constructs, a three-round asymmetric PCR strategy was used (30). For example, to generate the H105D mutation, asymmetric PCR using mutagenic primer kw71 containing 29 nucleotides of CvaB sequence and BglIIlinearized pHK11 template was performed to synthesize a singlestranded DNA in the first round. This single-stranded DNA, extended from the mutation through the BglII site at nucleotide 3874 of the ColV operon, was used as a template for a second round of asymmetric PCR primed with primer kw61 to amplify a complementary strand to the first round PCR product. In the third round of PCR, 20 ng of BglIIlinearized pHK11 template was combined with 2 l of the second round PCR product and primer kw60 to generate a symmetric doublestranded DNA product. The amplification products were digested with SacI and KpnI, and the 513-nucleotide fragment was used for subcloning into the same sites in vector pTrcHis2B, generating a new plasmid, pKW21. To incorporate the mutation into full-length CvaB, the SacI-PstI fragment containing 471 nucleotides from the new plasmid pKW21 was used to replace the corresponding fragment in wild-type CvaB. To introduce the FLAG epitope into wild-type CvaB, forward primer kw45 and backward primer kw106 containing eight amino acids of the FLAG epitope and the BglII restriction site were used to perform a standard PCR.
All mutations were confirmed by sequencing using an ABI Prism Big Dye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems Inc., Foster City, CA) and appropriate primers. Sequencing reaction products were purified on Centrisep columns (Princeton Separa-tions, Adelphia, NJ). Sequencing was performed using an ABI 377 or 3100 Sequencer (Applied Biosystems Inc.) in the Georgia State University Biology Core Facilities.
Gel Electrophoresis, Western Blotting, and Peptide Sequencing-Standard SDS-acrylamide gel electrophoresis was performed according to the method of Laemmli (31). For Western blotting, electrophoresed proteins were transferred to polyvinylidene difluoride membranes (Applied Biosystems Inc.) and treated by established procedures for Western blotting (32). Alkaline phosphatase-conjugated goat anti-rabbit IgG (Bio-Rad) was used as the secondary antibody. For N-terminal peptide sequencing, spots from gels transferred by CAPS transfer buffer (2.213 g of CAPS in 1 liter of deionized water, pH 11) were sequenced in the Georgia State University Peptide Sequencing Facility using an Applied Biosystems Procise Sequencer.
Overproduction and Purification of BntD for Eliciting Antibodies-BL21 cells containing the pKW21 plasmid were grown overnight in ampicillin-containing TB medium at 37°C and diluted to A 600 nm ϳ 0.1. At A 600 nm ϳ 0.5, the pKW21-containing BL21 cells were induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) to overproduce H105D BntD, and the cells were harvested 4 h later by centrifugation at 11,300 ϫ g for 8 min at 4°C. The purification procedure was based on the manufacturer's specifications (Invitrogen) with minor modifications as follows. For inclusion body purification, the cell pellets were resuspended in 50 mM Tris-HCl, pH 7.6, 20 mM NH 4 OAc, 40 mM KOAc, and 5 mM Mg(OAc) 2 and then passed twice through a French press (17,000 p.s.i.), followed by centrifugation at 15,000 ϫ g for 15 min at 4°C. The pellet containing overproduced 22-kDa BntD in inclusion bodies was washed sequentially with 1 and 2 M urea and then extracted with 6 M urea containing 50 mM Tris-HCl, pH 7.6. The supernatant was loaded onto a 2-ml nickel-chelate column (Invitrogen) and mixed with the resin by rocking for 30 min. The breakthrough fraction was collected and reapplied to the column, with rocking for 15 min. The resin was settled in the column, which was then washed with 40 ml of wash buffer (20 mM phosphate buffer, pH 7.6, 6 M urea, and 500 mM NaCl) by resuspending the resin, rocking for 2 min, and then draining the resin by gravity. The column was washed once with wash buffer plus 0.1 M imidazole and eluted with 2 ml of elution buffer (50 mM Tris-HCl, pH 8.0, 6 M urea, and 0.3 M imidazole); the resin was rocked for 15 min. The eluent was collected by gravity and used for generating antibodies.
Pieces containing 22-kDa BntD were excised from gels developed by 15% SDS-PAGE and ground into suspension. The purified protein (1.5 mg) was used for the primary injection with Freund's adjuvant, followed by boosters; serum preparations were prepared by general standard procedures (32), and anti-BntD antibodies were further purified using acetone powder (32).
Purification of Soluble BntD for Enzyme Assay-Cells harboring wild-type plasmid pKW24, H105D-expressing plasmid pKW21, or C32S-expressing plasmid pKW26 were grown and induced by IPTG as described above. The induction time was shortened to 1.5 h to reduce formation of inclusion bodies, and no urea was used in the preparations. The cell pellets harvested were resuspended in binding buffer (20 mM sodium phosphate and 0.5 M NaCl, pH 7.8) plus complete protease inhibitors and 1 mM NEM. The cells were lysed and centrifuged as described above. The supernatant containing induced soluble proteins was applied to the 2-ml nickel-chelate column and then washed twice with binding buffer, followed by binding buffer at pH 6.0. The proteins were eluted sequentially with 1 ml of 50, 200, 350, and 500 mM imidazole elution buffer. The samples eluted with 500 mM imidazole elution buffer containing BntD were dialyzed against 10 mM Tris-HCl, pH 8.0, and kept at Ϫ70°C in small aliquots for the protease cleavage assay. Purified wild-type BntD, C32S BntD, and H105D BntD were verified by peptide sequencing and immunodetection.
In Vitro Protease Cleavage Assay-The standard assays were as follow. Purified BntD (normally 4.32 g) was preincubated with 100 l of cleavage reaction buffer (50 mM Tris-HCl, pH 7.6, 5 mM CaCl 2 , 10 mM 2-mercaptoethanol, and 5 mM dithiothreitol) at 37°C for 10 min (33)(34)(35). The substrate L-arginine p-nitroanilide (LAPNA; final concentration of 1 mM) was added to cleavage reaction buffer and incubated at 37°C for 6 h. The assays were performed in 96-well microtiter plates (path length of 0.7 cm), and the cleavage product (p-nitroaniline) was monitored at 405 nm using a Beckman Biomek 2000 plate reader spectrophotometer. Background readings of hydrolysis of the substrate without BntD were subtracted. For kinetic analysis, initial rates of enzymatic activity were used. The V max and K m values were calculated from nonlinear regression, and the inhibition constant (K i ) was determined from IC 50 by fitting the data using GraphPAD Prism software.

Identification of Critical Residues in Cysteine Peptidase
BntD-By comparing the nucleotide sequence of BntD with those of other protease families, it seems likely that BntD belongs to the cysteine protease calpain family (Fig. 1), which contains cysteine and histidine as active sites and presumably asparagine as part of triad residues. A homology search of the cysteine residues in the N-terminal domain involved in proteolysis and colicin V secretion showed that Cys 32 may correspond to the active-site nucleophile Cys 105 of calpain because it is conserved in the motif GDC(W/G) (Fig. 1). To investigate the role of Cys 32 , it was replaced with Ser or Ala. As shown in Fig.  2, neither the serine nor alanine mutant had the colicin V secretion activity of the wild type. The other cysteine residues in the domain (Cys 36 , Cys 41 , and Cys 97 ) were similarly tested; each had normal colicin V secretion activity ( Fig. 2A), suggesting that Cys 32 is indeed the critical cysteine residue for colicin V secretion.
Moreover, amino acid sequence comparisons showed that His 100 and His 105 are conserved within the cysteine protease family (Fig. 1). Replacement of His 100 with Asp, Ala, Lys, or Ser did not impair normal colicin V secretion activity, but a similar replacement of His 105 abolished it ( Fig. 2A), indicating that His 105 is required for catalytic activity.
To verify that Cys 32 and His 105 mutants are not structurally unstable, we determined CvaB proteins by immunoblotting. The FLAG epitope was introduced into CvaB, which had no effect on the colicin V activity. The results show that CvaB, wild-type CvaA, C32S CvaA, and H105D CvaA were all stable (Fig. 2C).
H105D BntD Can Be Stably Induced-We constructed Cterminally His-tagged BntD under the control of the Trc promoter. Cultures containing plasmids encoding wild-type BntD or the H105D mutant were induced. Only the H105D mutant overproduced the product, which was purified by affinity chromatography (Fig. 3); the wild-type domain was not overproduced (data not shown). After the cells were lysed, the inclusion bodies containing H105D mutant proteins were harvested by centrifugation and solubilized by 6 M urea (lane 4). The supernatant was then applied to a His tag affinity column, and the nonspecific binding proteins were removed by washings. The 22-kDa H105D BntD protein was eluted with 0.3 M imidazole and used to elicit antibodies.
To detect wild-type BntD and the H105D mutant, samples were collected before and after IPTG induction. As expected, the total cell extracts of the H105D mutant collected after IPTG induction showed a strong immunological reaction, but only weakly before induction, indicating that the H105D mutant was indeed induced (data not shown). On the other hand, wild-type extracts showed little 22-kDa protein even after induction, confirming that it is difficult to detect stable wild-type BntD under the conditions used (data not shown).
Stabilization by Treatment with NEM and Purification of Wild-type BntD-We reasoned that because NEM is a cysteine protease inhibitor, it might stabilize wild-type BntD. Adding NEM (1 mM) to the buffer before cell lysis did indeed confer stability. The total cell extracts and inclusion bodies of wildtype BntD contained a 22-kDa protein, which could be detected immunologically.
Accordingly, we proceeded to purify wild-type BntD in the presence of NEM. Cells were induced at low temperature for 1.5 h, yielding a soluble supernatant that contained half of the induced proteins, from which we purified wild-type BntD. Its identity (Fig. 3, lane 9) was verified by N-terminal peptide sequencing, confirming the first eight amino acids as -MDPMT-NRN. These preparations were used to test protease activities.
BntD Proteolytic Activity Is Time-and Concentrationdependent-Because the sequence homology suggested that BntD is a cysteine protease, a typical substrate (LAPNA) for metallo-and cysteine aminoprotease activities was tested and found to be adequate for enzyme assay of BntD. The cleavage yielded p-nitroanilide, which could be monitored at A 405 nm .
To characterize proteolytic activity, four concentrations of BntD were tested for various times. At 2.16 g/100 l, BntD showed delayed activity, but reached maximal activity at 4.32 g/100 l; at 1.08 g/100 l, BntD showed low enzymatic activity; and at 0.86 g/100 l, BntD showed virtually no activity (Fig. 4). Low enzymatic activities were observed before a 6-h incubation with 2.16 g of proteins and increased between 12 and 24 h, suggesting that cooperative aggregation or oligomerization of BntD is required for maximal activity. The higher concentration (4 g) of BntD was used for all subsequent assays.
Substrate Specificity-N ␣ -Benzoyl-DL-arginine p-nitroanilide, a typical serine protease substrate, was compared with LAPNA, a typical substrate for metallo-and cysteine aminoproteases. As expected, N ␣ -benzoyl-DL-arginine p-nitroanilide was inactive (Fig. 5A), indicating that BntD belongs to the cysteine protease family. The effect of pH on BntD proteolytic activity was also tested, and the results show that BntD proteolysis occurred over a wide pH range, but was optimal at pH 7.5 (Fig. 5B).
To further characterize BntD, various concentrations of LAPNA were incubated with BntD in cleavage reaction buffer at 37°C for 6 h. The activities gradually increased up to 2 mM, reaching saturation at ϳ3 mM (Fig. 5C). The rate of hydrolysis of LAPNA was dependent on substrate concentrations, with a K m of 1.15 mM and a V max of 91 pmol/min/g BntD (Fig. 5C).
BntD Proteolytic Activity Is Calcium-dependent-As indicated in Fig. 1, the sequence homology suggested that BntD is a cysteine protease calpain-like protein. Calpains, which include the ubiquitously expressedand m-calpains, are a family of Ca 2ϩ -dependent intracellular cysteine proteases from mammals, birds, and insects. The product of the sol gene in Drosophila is a calpain (36). To test whether BntD indeed requires Ca 2ϩ for activity, different concentrations of CaCl 2 were tested. p-Nitroanilide cleavage gradually increased up to 5 mM Ca 2ϩ (Fig. 6A), reaching saturation at 7.5 mM. The data demonstrate that the proteolytic activity of BntD is calcium concentration-dependent. CaCl 2 at 5 mM was used in subsequent experiments.
It has been reported that the enzymatic activity of a cysteine protease of Porphyromonas gingivalis, a black-pigmented Gram-negative anaerobe, is enhanced severalfold by divalent cations such as Ca 2ϩ , Mg 2ϩ , and Mn 2ϩ (37). We tested their effect on BntD proteolytic activity. 5 mM CaCl 2 , MnCl 2 , and CuCl 2 each conferred moderate activity (Fig. 6B). On the other hand, KCl at 50 mM or MgCl 2 at 5 mM was less effective, and MgCl 2 at 10 mM did not compete with the CaCl 2 -induced proteolytic activity of BntD (Fig. 6C). ZnSO 4 at 5 mM did not enhance any activity. Due to their similar ionic radii, Tb 3ϩ and La 3ϩ have been widely used to compete for calcium binding (38). We were surprised to find that TbCl 3 at 150 M and LaCl 3 at 100 M were as active as CaCl 2 at 5 mM and that LaCl 3 at 500 M was 2-fold more active, but higher concentrations enhanced less activity presumably due to precipitation. At the same concentrations, FeCl 3 was inactive. Interestingly, Havarstein et al. (19) reported that 250 mM Na 2 SO 4 increases proteolytic cleavage by LagD by 4-fold. We found that it did not enhance the proteolytic activity of BntD as effectively as Ca 2ϩ .
To determine whether the proteolytic activity of BntD requires free calcium, the divalent ion chelator EDTA and the heavy metal chelator EGTA were incubated in cleavage reaction buffer with BntD (Fig. 6C). In the presence of 5 mM CaCl 2 ,  6. BntD proteolytic activity is calcium-dependent. A, BntD proteolytic activity is calcium-dependent. Increasing concentrations of CaCl 2 were added to cleavage reaction buffer and incubated for 6 h. The cleavage amount increased quickly up to 5 mM and then reached a maximum at 7.5 mM. B, cations are required for the BntD proteolytic reaction. TbCl 3 and LaCl 3 significantly increased BntD activity. MgCl 2 and EDTA at 5 mM reduced BntD activity by almost 90%, whereas EGTA at 20 mM inhibited BntD activity by 75%. These data indicate that BntD proteolytic activity requires free Ca 2ϩ .
BntD Proteolytic Activity Requires Reducing Agents and Effects of Inhibitors-Our preparations contained 1 mM NEM initially in the cell extracts to stabilize the protein to allow purification of BntD, and our typical enzyme assays contained 2-mercaptoethanol and DDT. We evaluated the effects of these compounds on BntD activity. Very low activity was observed in the absence of 2-mercaptoethanol or dithiothreitol (Fig. 7, bar  1). Dithiothreitol at 5 mM or 2-mercaptoethanol at 5-10 mM (data not shown) conferred similar specific activity (32 pmol/ min/g), indicating that 2-mercaptoethanol at 5 mM is sufficient to sustain the reducing environment required for the proteolytic activity of BntD, presumably due to the sulfhydryl group of the critical residue Cys 32 .
Inhibition of BntD activity by NEM was then characterized in the presence of 2-mercaptoethanol (Fig. 7). NEM at 5 mM inhibited proteolytic activity by 90% (lane 3). This observation is consistent with earlier reports that inhibition by NEM at low concentrations is reversible (39,40) and at high concentrations is irreversible, presumably because of the inactivation by alkylation (39,41).
Four classes of protease have been well defined: cysteine protease, serine protease, metalloprotease, and aspartic protease (42). To assign BntD more precisely to the cysteine peptidase class, its sensitivity to other groups of inhibitors was tested (Fig. 7). BntD was sensitive to the cysteine protease inhibitor antipain (in addition to NEM) in a dose-dependent manner. No inhibition was observed at 1 mM antipain (data not shown), and 2.5 and 5 mM antipain inhibited proteolytic activity by 47.5 and 73%, respectively (Fig. 7). Pepstatin A, an aspartic protease inhibitor, and phenylmethylsulfonyl fluoride, a serine protease inhibitor, exerted no inhibitory effect under the conditions used, indicating that BntD does not possess significant aspartic or serine protease activity.
Lack of Proteolytic Activity of Mutants C32S and H105D-The 22-kDa soluble BntD proteins from CvaB mutants C32S and H105D, which lack colicin V secretion activity (Fig. 2), were purified and assayed. As expected, they lack proteolytic activity completely (Fig. 8), thus supporting the notion that Cys 32 and His 105 are critical residues for proteolytic activity as well as colicin V secretion.

DISCUSSION
In this study, Cys 32 and His 105 were identified as critical catalytic residues involved in BntD proteolytic activity. Purification of wild-type BntD was difficult previously because of its proteolytic activity, which is presumably capable of self-cleaving. 2 The stability and purification of the wild-type domain were accomplished in the presence of the cysteine protease inhibitor NEM in cell extracts (ϳ7.2 mg/ml proteins), which presumably interacts with the sulfhydryl group of the cysteine residue. Inhibition by 1 mM NEM in the cell extracts during BntD purification was apparently reversible, as has been reported previously (39,40). In vitro cleavage showed that the proteolytic activity of BntD was indeed inhibited by NEM; addition of dithiothreitol restored inhibition at low concentrations of NEM, indicating that NEM inhibition is reversible at low concentrations, whereas at high concentrations of NEM (Ͼ2 mM), inhibition is not reversible. 2 Although NEM is traditionally regarded as a cysteine protease inhibitor, the inhibition of BntD by NEM in this work does not demonstrate that inhibition by NEM is mediated through covalent modification of Cys 32 because alkylation at a site other than a cysteine residue has been shown (41). The nature of the reversible and nonreversible inhibition by NEM at different concentrations is under investigation.
It has been reported that the enzymatic activity of cysteine proteases is enhanced severalfold by divalent cations such as Ca 2ϩ (37,43) and Mg 2ϩ and Mn 2ϩ (37). Our study shows that Ca 2ϩ stimulated the BntD reaction in a time-dependent (Fig. 4) and concentration-dependent (Fig. 6A) manner. Low concentrations of calcium showed a moderate effect, but high concentrations exerted progressively accelerated BntD activity. Pal et al. (44) reported that the active triad of calpain in the absence of Ca 2ϩ is not assembled into a catalytically active conformation, as evidenced by crystallization in the absence of calcium. Our data support this finding because little proteolytic activity was observed in the absence of Ca 2ϩ . Other divalent cations (Mn 2ϩ and Cu 2ϩ ) exhibited a moderate effect. Tb 3ϩ and La 3ϩ , calcium analogs because of their similar ionic radii and coordination 2 K.-H. Wu and P. Tai, unpublished data.
KCl showed moderate LAPNA cleavage; MnCl 2 and CuCl 2 showed half-activity of the CaCl 2 addition. No activity was detected in the absence of CaCl 2 (negative control), ZnSO 4 , and FeCl 3 addition, suggesting that Ca 2ϩ is required for BntD proteolytic activity. C, Ca 2ϩ is required for BntD proteolytic activity. 10 mM MgCl 2 did not compete with 5 mM CaCl 2 -induced cysteine proteolytic activity. BntD-specific activity was reduced by addition of 5 mM EDTA and 20 mM EGTA by chelating Ca 2ϩ divalent cations. Error bars represent the S.D. of an experiment performed in triplicate.  8. Lack of proteolytic activity of mutants C32S and  H105D. Wild-type BntD proteolytic activity (6 h) was used as a 100% activity control, and wild-type 0-h incubation served as the background activity. Purified C32S and H105D mutant proteins were assayed, but no cleavage was observed. Error bars represent the S.D. of an experiment performed in triplicate.
properties, have long been used to probe presumed calcium binding (38). Interestingly, low concentrations of Tb 3ϩ and La 3ϩ significantly enhanced BntD proteolytic activity, whereas FeCl 3 did not, which is consistent with the physiological role of CvaB in secretion of colicin V under conditions of iron depletion (45). Based on the above results, the size of the cation appears to be important, but trivalent charges ion can be ruled out. It has been reported that Zn 2ϩ inhibits caspase-6 (46) and caspase-3 (47) activities due to the inherent tendency of Zn 2ϩ to react with thiols (46). The present study shows that ZnSO 4 did not enhance BntD proteolytic activity, indicating that this effect was probably due to the interaction of Zn 2ϩ with the thiol group. EDTA and EGTA, chelators of metal ions, abolished the stimulating effect of calcium, indicating that free calcium is required for the proteolytic activity of BntD. A typical calpain protein has approximately four or five EF-hands, which include the calcium-binding domain (48), but a computer search did not reveal an EF-hand motif in BntD. On the other hand, the EF-hand structure is also absent in several other calpain proteins, including human calpain-7; and despite the absence of the EF-hand or Ca 2ϩ -binding motifs, the recombinant human calpain protein still shows catalytic activity in the presence of calcium (49). Moreover, non-EF-hand Ca 2ϩ -binding sites have been revealed in the cysteine protease region of calpain, and Ca 2ϩ binding aligns the active site of calpain (50). Taken together, these results indicate that BntD is a member of the calpain-like cysteine peptidase family and is a Ca 2ϩ concentration dependent cysteine peptidase.
The catalytic residues of cysteine protease are believed to be Cys and His (24,25). It has been suggested that a histidine residue located in the vicinity of the active-site cysteine residue is required to form a thiolate-imidazolium ion pair (51). The amide oxygen of the Asn side chain is believed to form a hydrogen bond with the nitrogen atom of His (24,25), creating a Cys-His-Asn triad that is often considered analogous to the Ser-His-Asp arrangement of serine peptidases. The conserved region around Cys is QGGDC(G/W)A. Cys 32 of CvaB is conserved among the cysteine proteases. We have demonstrated here that it is also involved in secretion activity. The conserved regions around His and Asn of cysteine proteases are -HF and -NPWGW, respectively (Fig. 1). His 105 of CvaB is in this region, but Asn is absent. Mutants in which His 105 was replaced by other amino acids exhibited no protease activity. The presence of the imidazole ring is essential for proteolysis presumably because no other amino acid side chain can substitute for the tautomeric property of imidazole. It has been suggested that the basic nature of the imidazole ring, which can accept a proton released from the nucleophile (Cys residue), and the aromatic property of a protonated imidazole ring contribute to proteolytic activity (52). The C32S and H105D BntD mutants showed no proteolytic activity in vitro (Fig. 8), which further supports the notion that Cys 32 and His 105 are the critical residues for proteolysis. Asn was believed to be involved in hydrogen bonding with His to stabilize the thiolate-imidazolium ion pair. However, the active Asn residue was not identified in BntD. 2 As predicted by three-dimensional modeling, Trp 101 , Asp 102 , and Val 108 of BntD may interact with the imidazole ring of His 105 . The possible involvement of these residues is under investigation. Gln 26 of BntD is also conserved among cysteine proteases. It has been reported that Gln 19 in papain forms an "oxyanion hole" with Cys 25 , which in turn stabilizes the Cys and His ion pair (57). Replacements of Gln 26 with Lys significantly affected ColV secretion activity. 2 The role of Gln 26 is under investigation.
Alignment in different data base systems shows that BntD has high identities to the peptidase C39 family, 3 which includes lantibiotic and non-lantibiotic bacteriocins. They are synthesized as precursor peptides containing N-terminal leader peptides, which are cleaved off during maturation. These bacteriocins have leader peptides of the so-called doubleglycine type: the processing site with two conserved glycine residues in positions Ϫ1 and Ϫ2. Peptide bacteriocins, including colicin V (18,23), are exported across the cytoplasmic membrane by the ABC transporter, in which the proteolytic domain resides in its N-terminal part. Havarstein et al. (19) showed that the N-terminal domain of LagD, a transporter of lactococcin G, possesses proteolytic activity against lactococcin G derivatives containing the double-glycine peptides. Our study indicates that BntD indeed possesses protease activity, which has been shown to be a novel calcium-dependent cysteine protease. BntD in low amounts reacted slowly with the substrate LAPNA, requiring 6 h to show significant cleavage. Even at high protein concentrations, the V max was still relatively low; a cooperative effect of aggregation or oligomerization of BntD is probably required at low protein concentrations. This low enzymatic activity may be a characteristic of energy-dependent proteases. It is generally believed that the nucleotide-binding domain of the ABC transporter binds and hydrolyzes NTPs, which provides energy for ABC transporters to secrete (53). It is not certain whether the proteolytic activity of BntD can be enhanced by its C-terminal domain ATPase. Zhong et al. (21) showed that processing ColV-1 requires the CvaA-CvaB transporter, the TolC protein, membrane integrity, and energy. In addition, the processing of ColV-1 is rapid (21). The slow enzymatic activity of BntD suggests that the cooperative effects with other transporter components and energy might be required for higher proteolytic activity in vitro for cleavage and secretion of colicin V (21).