Processing of colicin V-1, a secretable marker protein of a bacterial ATP binding cassette export system, requires membrane integrity, energy, and cytosolic factors.

Extracellular secretion of the peptide antibiotic colicin V (ColV) in Escherichia coli is mediated by a dedicated exporter system consisting of host TolC protein and the products of two specific genes, cvaA and cvaB, the latter being a member of the ATP binding cassette (ABC) superfamily. An amino-terminal export signal of ColV is specific for the CvaA-CvaB-TolC exporter and is processed concomitant with secretion. In this study, we attempt to characterize this processing with a secretable marker protein, ColV-1, using a newly developed in vitro assay. Processing is found to be dependent on both CvaA-CvaB transporters and the TolC protein and to require membrane integrity. An additional cytoplasmic soluble factor(s) is also necessary for the processing. Although the sequence of the cleavage site suggests it could be a substrate, ColV-1 cannot be processed in vitro by the purified leader peptidase I. Moreover, ColV-1 processing is inhibited by antipain and N-ethylmaleimide. Furthermore, the processing requires energy in the form of nucleotide hydrolysis. These results indicate that the processing of ColV-1 is specific and more complex than expected, requiring the CvaA-CvaB-TolC transporter intact in the membrane, energy, and cytosolic factors for rapid cleavage.

The localization of proteins to different cellular and extracellular compartments has been studied intensively because of its fundamental importance to all living cells. The best characterized process is the classic, signal sequence-dependent export pathway (1,2). In Gram-negative bacteria, this pathway requires the products of several sec genes and precursor proteins bearing an amino-terminal signal sequence, which usually has a positively charged amino terminus and a hydrophobic core, followed by a cleavage site for signal peptidases (3).
There is, however, a group of proteins and peptides that are exported by signal sequence-independent pathways. Such substrates include yeast a factor (4), Escherichia coli ␣-hemolysin (5,6), Bordetella cyclolysin (7), Erwinia protease (8), and an E. coli peptide antibiotic, microcin B17 (9). In each case, export is mediated by a dedicated export apparatus. This group is sometimes referred to as the multidrug resistance-like family because one of its best known members is the P-glycoprotein or multidrug resistance protein, which is responsible for the multidrug resistant phenotype of tumor cells when overexpressed (10,11). It is also referred to as the ABC 1 transporter superfamily, which describes the ATP binding domain common to all members (12)(13)(14)(15)(16)(17).
The bacterial ABC export system is exemplified by the secretion of ␣-hemolysin from E. coli (18,19). The targeting signal for its secretion resides in the carboxyl-terminal 50 amino acids and is not cleaved (20 -23). The transport process requires three specific envelope proteins: the ABC exporter (HlyB), accessory protein (HlyD), and outer membrane TolC protein (5, 6, 20 -25). The export of peptide antibiotic colicin V (ColV) from E. coli also utilizes an ABC export system similar to ␣-hemolysin (26,27). The structure gene for colicin antibiotic is encoded in large, low copy number virulence plasmids found in E. coli and other members of Enterobateriaceae (28). ColV kills sensitive cells by disrupting their membrane potential (29). Four linked genes are identified in the ColV operons (26,27): the cvaC structural gene, the cvi immunity gene, and the two exporter genes cvaA and cvaB. Products of cvaA and cvaB, together with the TolC protein, function as a dedicated export system in producing cells to transport ColV through two membranes (26,27). Although biochemical data have shown that ColV does not accumulate in the periplasm (30), the data from genetic assays indicate that a small amount of ColV appears to be present transiently in the periplasm during secretion (31).
ColV is synthesized as a 103-amino acid primary translation product (pre-ColV) which does not contain a typical Sec-dependent amino-terminal signal peptide, even though its aminoterminal 15 residues are processed concomitant with extracellular secretion (32,33). Homology analysis reveals that the sequence of this amino-terminal signal has significant similarity to the leader peptides of nonlantibiotics and some lantibiotics found in Gram-positive bacteria (32)(33)(34). The resulting 88 residues, comprising the mature colicin V peptide, has aberrant mobility on SDS-polyacrylamide gel electrophoresis (predicted M r , 8700; observed M r , 5800; Ref 32). It has been suggested that active ColV probably contains a disulfide bond between the only cysteines at positions 91 and 102 of the polypeptide (33). ColV-1 is a Tn5 ColV mutant form with the 21 carboxylterminal amino acids of colicin V replaced by 8 heterologous residues (32). It can be secreted efficiently into extracellular media from cells with a plasmid containing cvaA and cvaB, indicating that the carboxyl-terminal 21 amino acids of ColV are not required for secretion (32), in contrast to that of hemolysin. It loses ColV toxicity but exhibits normal mobility on SDS-polyacrylamide gel electrophoresis (predicted M r , 7200; observed M r , 7300), providing a convenient tool for certain studies.
In this study, we have developed an in vitro assay using ColV-1 as a marker protein to characterize the processing of ColV. The results show that the processing of ColV-1 in vitro requires the intact CvaA, CvaB, and TolC transporters in the membranes. Additionally, nucleotide hydrolysis and soluble factor(s) are required for the processing. Moreover, the specific inhibition in processing by antipain and NEM suggests that the peptidase responsible for ColV-1 processing might be a cysteine protease.
Manipulation of DNA-Isolation of plasmid DNA, transformation of bacteria, restriction enzyme digestion, and other routine DNA manipulations were performed using standard procedures (36). For cloning the cvaC-1 gene behind the T7 RNA polymerase promoter, two primers, primer 1, 5Ј-GGGAATTCTTTTACGGCGGTCGCCTTGACCGCCGAA-AGGTTCCGTTCAGGACG-3Ј (underline indicates a -independent transcriptional terminator); and primer 2, 5Ј-CTCCTGCAGTATCCGA-TAAGATAAAAAGGAGAT-3Ј, were used to generate a 350-base pair polymerase chain reaction fragment from template plasmid pHK11-1. This fragment and vector pT7-6 were each digested with EcoRI and PstI, and then appropriate fragments were ligated to form pXZ5 (Fig.  1B). pXZ10 is another T7 cvaC-1 construct that contains double stop codons. Primer 3, 5Ј-GGGAATCTTTTACGGCGGTCGCCTTGACCGC-CGAAAGGTTCTTATCAGGACG-3Ј (underline indicates double stop codons) and primer 2 were used to generate a polymerase chain reaction fragment from template plasmid pXZ5. This fragment and pT7-6 were digested with EcoRI and PstI and then ligated to form pXZ10.
In Vivo Labeling and Pulse Chase-Cells were grown at 37°C in TB medium supplemented with 0.5% glucose and appropriate antibiotics. The overnight culture was diluted 1:50 in the same medium. When the cells had grown to A 600 of 0.2-0.3, 0.1 mM 2,2Ј-dipyridyl (for inducing the ColV transporters, which are iron-regulated) was added for 1 h, and then 2 ml of cells was centrifuged for 30 s. The cell pellet was washed and resuspended with 2 ml of M9 medium containing 19 amino acids and appropriate antibiotics. After cell growth for 40 min at 37°C with shaking, 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside was added for 20 min, and then rifampicin was added to 200 g/ml. The culture was kept at 42°C for 10 min and then shifted to 37°C for an additional 20 min. The cells were labeled by adding 50 Ci/ml [ 35 S]methionine for 5 min. After centrifugation, the cell pellet was resuspended in the sample buffer, boiled, and subjected to gel electrophoresis. To detect the extracellular ColV-1, the cells were labeled for 5 min and chased with cold methionine (0.1%) for 30 min. Following centrifugation, the supernatant was precipitated with ice-cold trichloroacetic acid (TCA) to a final concentration of 10%, and after centrifugation the pellet was washed with ice-cold acetone, dried, dissolved in sample buffer, and analyzed.
In the pulse chase experiment, BL21(DE3)/pXZ5/pHK22-6 cells were pulse-labeled for 10 s and chased by the addition of stop solution (final concentration, 0.1% methionine and 100 g/ml tetracycline) to stop new synthesis of overproduced product. 150 l of cell culture samples was collected and precipitated by ice-cold TCA.
In Vitro Transcription Translation-Synthesis of ColV-1 precursors by a cell-free transcription-translation system from plasmid pXZ5 (or pXZ10) was modified from the described in vitro translation system (35) using S30 from strain D10 or D10/pHK11-1. In addition to the reagents described, the mixtures (in 1 ml) contained 5 mM phosphoenolpyruvate-HCl, 0.88 mM each of UTP-Tris, CTP-Tris, and GTP-Tris, 0.76 mM ATP-Tris, 25 g pXZ5 plasmid DNA, 200 Ci of [ 35 S]methionine, 0.174 mg/ml tRNA, and 40 units of T7 RNA polymerase. Reactions were carried out at 37°C for 10 min, and the mixtures were centrifuged in a Beckman Instruments TL100.2 rotor at 300,000 ϫ g for 20 min to remove ribosomes and any endogenous membrane vesicles. The supernatant was used as the source of ColV-1 precursor.
In Vitro Processing Assay-25 l of supernatant from in vitro transcription-translation of ColV-1 was mixed with 0.1 unit of A 280 membrane vesicles (prepared as described previously; Ref. 35), along with 2 l of energy-regenerating mixture (35) and an appropriate amount of S100 to a final volume of 100 l. S100 was prepared from S30, which was centrifuged at 100,000 rpm for 20 min in a Beckman Instruments TLA100.2 rotor. The reaction mixtures were incubated at 37°C for 15 min. Membrane vesicles were isolated by centrifugation or precipitated by 10% TCA where indicated, then dissolved in the sample buffer and submitted to SDS-polyacrylamide gel electrophoresis. Autoradiographs were quantitated by PDI image analyzing system (Protein Databases Inc., New York, NY). For the calculation of processing ratio, correction was made based on the presence of three residues of methionine in the precursor form and two residues in the processed form.
Removal of Nucleotides-0.3 ml of the 300,000 ϫ g supernatant of the transcription-translation mixture was applied to a P4 polyacrylamide (Bio-Rad) column (0.5 ϫ 40.0 cm), which had been previously equilibrated with 50 mM Tris-HCl, pH 7.6, and 1 mM dithiothreitol. Fractions (0.25 ml) were collected, and those containing ColV-1 were monitored by determinations of a hot trichloroacetic acid-precipitable count (National Diagnostics), verified by gel electrophoresis, and pooled.
Gel Electrophoresis and Immunological Assays-Regular SDS gel electrophoresis with 15% acrylamide was carried out according to the method of Laemmli (37). 15% Tricine SDS gel electrophoresis (38) at 4°C was used for separating precursor and processed forms of ColV-1. For immunoprecipitation, samples were diluted to 1 ml with IPB buffer (20 mM Tris-HCl, pH 8.0, 0.15 M NaCl, 5 mM EDTA, 0.5% Triton X-100, and 0.05% SDS) and treated with preimmune serum (10 l) on ice for 30 min. Then 100 l of 10% Sepharose-protein A suspension in IPB buffer was added for 10 min with occasional shakings. After spinning at 5,000 rpm for 2 min, the supernatant was treated with anti-ColV peptide antiserum (10 l) for 30 min, and 100 l of 10% Sepharose-protein A suspension was added for 10 min. The resin pellet after centrifugation was washed with 1 ml of IBP buffer twice, and then proteins were eluted with 1.5 ϫ sample buffer and boiling.
Chemicals and Reagents-Purified GroEL, GroES, DnaK, DnaJ, and GrpE were from StressGen Biotechnologies Corp. SecA, SecB, and proOmpA were laboratory stocks purified from T7-SecA-and T7-SecBoverproducing strains and BA13/pAM103, respectively. Nucleotides and protease inhibitors were obtained from Sigma. GMP-PNP 3 was obtained from Boehringer Mannheim. Other chemicals were of reagent grade obtained from commercial sources.

ColV-1 Can Be Efficiently Processed and Secreted into the
Extracellular Medium-In this study, we chose ColV-1 (Fig.  1A) as a model marker protein. The whole cvaC-1 gene was subcloned behind the T7 RNA polymerase promoter in plasmid pXZ5 (Fig. 1B). The ColV-1 precursor was uniquely labeled in the strain BL21(DE3)/pXZ5 ( Fig. 2A, lane 1). In the presence of CvaA and CvaB transporters, ColV-1 was processed ( Fig. 2A, lane 2) and secreted into the extracellular medium ( Fig. 2A,  lanes 3 and 4). The precursor and the processed form of ColV-1 could each be immunoprecipitated by ColV peptide antibody (data not shown). Their migration difference provides a convenient criterion for the identification of the processing of ColV-1. The processing was rapid and efficient (Fig. 2B). More than 50% of the T7 promoter-expressed ColV-1 precursor was processed within 1 min (Fig. 2A, lane 4).
Processing of ColV-1 Requires Soluble Factor(s) and Dedicated Transporters -As a first step to elucidate biochemically the mechanism of ColV secretion, we developed an in vitro processing assay using ColV-1 as a marker protein, transcribed and translated from plasmid pXZ5. Inverted inner membrane vesicles were isolated from MC4100/pHK11-1, which could actively secrete ColV-1 in vivo. In the preliminary experiments, this assay was carried out both co-and post-translationally; i.e. the membranes were added either during or after protein synthesis. There was not much difference between these two processes. A very weak processed band of ColV-1, which had the same migration pattern of mature ColV-1, was detected, unless the amount of S100 from D10/pHK11-1 was increased (Fig. 3A, lanes 2-5). On the other hand, there was no processed form without addition of membranes (Fig. 3A, lane 7); therefore, the processing was not due to nonspecific proteases in the S100. However, since S100 from the D10 strain without CvaA and CvaB transporter genes also had the same enhancement effect (Fig. 3A, lane 6), the soluble factor was not specifically derived from the ColV operon.
Surprisingly, only about 20% of total processed ColV-1 was found to co-sediment with membrane vesicles, and 80% remained in the supernatant after centrifugation (data not shown). Those in the supernatant might be unable to complete the translocation in the membranes and were released after cleavage. This result was consistent with the in vivo data that some processed ColV-1 was found in cytosol (data not shown). Since a majority of the processed form was in the supernatant, a processing assay hereafter was then carried out in total reaction mixtures precipitated by TCA without isolating membranes, unless specified.
In the transcription-translation mixture, a protein band was observed above the precursor of ColV-1, which was also processed in the assay (Fig. 3A, lanes 4-6). As shown in Fig. 3B, the precursor of ColV-1 (PColV-1) and its processed form (ColV-1) and the band above the precursor of ColV-1 (PColV-1Ј) together with its proc- A, the processing assay was carried out as described under "Experimental Procedures." All reaction mixtures (100 l) were incubated at 37°C for 15 min. Membrane vesicles were reisolated by centrifugation, dissolved in sample buffer, and submitted to SDS-polyacrylamide gel electrophoresis. Lane 1, ColV-1 precursor; lane 2, reaction without S100 added. In lanes 3-5, S100 from D10/pHK11-1 (cvaA ϩ cvaB ϩ tolC ϩ ) was added; in lane 6, S100 from strain D10 (wild type, cvaA Ϫ cvaB Ϫ ) was added. In lane 7, S100 from D10/pHK11-1 was added without membrane vesicles. Lane 8, markers for precursor and mature forms of ColV-1. B, cell-free T7 transcription-translation mixture of ColV-1 and in vitro assay mixtures with membrane vesicles from MC4100 or MC4100/pHK11-1 were pretreated with preimmune serum (lanes 1, 3, and 5) and then precipitated by ColV peptide antibodies (lanes 2, 4, and 6). C, membrane integrity is required for processing. Lane 1, TCA precipitate of total in vitro assay mixture with membrane vesicles from MC4100/pHK11-1. Lane 2, membrane vesicle was pretreated with 2.5% OG then was diluted 5-fold in the assay. Lane 3, same as lane 2, with addition of 4 g of leader peptidase (SPase I). D, titration of OG and Triton X-100. Membrane vesicles from MC4100/pHK11-1 were pretreated with OG and Triton X-100 in the concentration as indicated and then were diluted 12.5 times in the assay. Lane 1, ColV-1 precursor. essed form (ColV-1Ј) in MC4100/pHK11 membrane vesicles (lane 6) could be immunoprecipitated by ColV peptide antibodies. Therefore, this band is probably a read-through product of ColV-1, because a stem loop structure (-independent transcription factor) lies right behind the translational stop codon of the cvaC-1 gene during the construction of the plasmid. (Indeed, a construct containing double stop codons did not yield the upper band, and the processing was similar (data not shown, but see Fig. 9).) No processed forms were observed with MC4100 membranes without the transporter (Fig. 3B, lane 4), confirming that they were all ColV derivatives and the processing was specific for the dedicated transporters.
Membrane Integrity but not Leader Peptidase I Is Required for Processing-The cleavage site for ColV is preceded by the sequence Ser-Gly-Gly, making it a potential substrate for leader peptidase I, as noted by Fath et al. (32). However, in the in vitro assay, there was no processing of ColV-1 with solubilized membranes from MC4100/pHK11-1, even in the presence of 4 g of purified leader peptidase I (see Fig. 3C). Under the same conditions, both purified and in vitro translated proOmpA were completely processed by 0.8 g of purified leader peptidase I or 50% processed by the leader peptidase I released from octylglucoside (OG)-or Triton X-100-solubilized membranes (data not shown). When the membranes were treated with OG or Triton X-100 concentrations higher than 0.5 or 0.1%, respectively, the processing was completely abolished (Fig. 3D). These data indicated that membrane integrity, i.e. the transporter in the intact membrane, is required for the processing and that there might be another peptidase that processes ColV-1.
Processing of ColV-1 Is Inhibited by Antipain and N-Ethylmaleimide-To characterize the processing of ColV-1, different types of protease inhibitors were tested in the in vitro assay. Based on the nature of their active sites and mechanism of action, four mechanistic classes of protease have been recognized thus far: metalloprotease, serine protease, aspartic protease, and cysteine protease (39). Each type of protease activity can be specifically inhibited by its own inhibitors. Thus, five representative protease inhibitors were tested for their effects. Pepstatin A, phosphoramidon, and phenylmethylsulfonyl fluoride had no inhibition (Fig. 4A, lanes 3-6, 9, and 10). In contrast, antipain at 5 mM showed significant inhibition (Fig. 4A, lane 8), and NEM completely inhibited the processing of ColV-1, even at 1 mM (Fig. 4A, lane 11). When the membrane fraction and soluble factors were treated with NEM separately, the processing activity was inactivated (Fig. 4B), indicating that the membrane and soluble factors contributing to ColV-1 processing are NEM-sensitive.
The Soluble Factor(s) Is Identified as Being Protein in Nature-To characterize further the nature of the soluble factor(s) that enhanced the processing, its sensitivity to protease and to RNase was tested. As shown in Fig. 5A, proteinase K (100 g/ml) completely inactivated the soluble factor, but not in the presence of an inhibitor, phenylmethylsulfonyl fluoride, whereas treatment with the micrococcal nuclease (500 g/ml) had no effect. These results suggest that the soluble factor(s), which enhances the processing, is of a protein nature, and no functional RNA, if any, is accessible to RNase.
As the soluble factor(s) is not specific from ColV operons, it might be a general factor. We tested several candidates that are known to be involved in protein targeting: SecA, SecB, trigger factor, GroEL, GroES, DnaK, DnaJ, and GrpE. None of these purified proteins alone or in different combinations could enhance processing of ColV-1 (Fig. 5, B and C). To determine whether the purified factors need additional factors to function, small amounts of S100 were mixed with these purified factors. No processing enhancement was observed (Fig. 5, B, lanes 6,   11, and 13, and C, lanes 4, 6, 8, and 10). These results indicate that factors other than those tested have the enhancing activity of processing.
Kinetics of ColV-1 Processing-To determine the rate of processing of ColV-1 in vitro, kinetics of processing was performed in the assays with MC4100/pHK11-1 membrane vesicles. As shown in Fig. 7, processing at 37°C was linear up to 8 min, and half of the ColV-1 was processed within 4 min, indicating that processing occurred rapidly in vitro, in a manner similar to that observed in intact cells (Fig. 2B). Processing was much less efficient but appreciable at 0°C than at 37°C. This result is consistent with the finding that processing is still continued at 0°C in vivo (data not shown).

Processing of ColV-1 Requires Energy in the Form of NTP(s), and GTP Appears to Be the Most Efficient Nucleotide for This
Process-It is widely accepted that ATPase activity of ABC transporters contributes energy to all active ABC protein-dependent transport. Since CvaB is a member of the ABC transporters, it was of interest to determine the energy requirement of the ColV-1 processing with our in vitro assay. Small molecules (e.g. ATP and salts), required for the transcription-translation mixtures of ColV-1, were removed by gel filtration after in vitro synthesis. Fractions containing ColV-1 were pooled and used for the assay.

FIG. 4. Processing of ColV-1 is inhibited by NEM and Antipain.
A, membrane vesicles from MC4100/pHK11-1 and D10 S100 (125 g) were used for the in vitro assay as described under "Experimental Procedures," except that final reaction volume was reduced to 25 l, and the reaction mixtures were stopped by TCA. Reaction mixtures were incubated with the inhibitors as indicated on ice for 15 min before adding ColV-1 precursor to start reaction. Lane 1, precursor of ColV-1. B, both membrane vesicles and soluble factors are sensitive to NEM. S100 and membranes were treated with 1 mM NEM at 0°C for 20 min; 10 mM dithiothreitol was added after incubation (lanes 3 and 5) or during incubation (lanes 4 and 6) to prevent the NEM action; and then whole mixture each was dialyzed for 3 h at 4°C before assay. Lane 1, precursor alone.
As shown in Fig. 8A, processing was completely abolished by such nucleotide depletion (lane 1). This activity, however, could be restored when different nucleotides were added to test for the processing of ColV-1 (Fig. 8A, lanes 2-9). As expected, ATP was active in promoting the processing and more active than CTP and UTP. As the concentration was increased from 0.5 to 1.0 mM, the efficiency of ATP was increased significantly. Surprisingly, GTP seemed to be the most efficient among all NTPs. Furthermore, there was little processing with the nonhydrolyzable GTP analogue GMP-PNP (Fig. 8A, lane 10), indicating that the hydrolysis of GTP was required. Addition of 1 mM ATP together with 1 mM GTP was only slightly more active than with 1 mM GTP alone (Fig. 8A, lane 11). Further titration of GTP and ATP on ColV-1 processing showed that 3 mM GTP was sufficient to maximize the observed processing, whereas ATP reached its maximum at 4 mM and was still consistently less efficient than GTP at that level (Fig. 8B). These results show that GTP is as efficient as ATP if not the most efficient nucleotide required for the processing.
CvaA-CvaB Membrane Vesicles Alone Have Little Processing Activity-It has been suggested that the CvaA-CvaB transporter could have "leader peptidase" activity (32). Moreover, CvaB has been proposed to possess protease activity (40,41). It is thus of interest to determine whether CvaA-CvaB membrane vesicles alone have some processing activity without addition of soluble factor(s) and energy. A new construct of ColV-1, pXZ10, that contained a double stop codon and generated a single ColV-1 product (Fig. 9, lane 1) was used. With energy and S100, the product of new ColV-1 construct was processed in 15 min at 37°C as efficiently as that of the old construct (Fig. 9, lane 8).
In contrast, ColV-1 precursors were not processed after incubation with CvaA Ϫ B Ϫ or CvaB Ϫ membrane vesicles, or even with CvaA-CvaB membrane vesicles without energy and factors at 37°C for 18 h (Fig. 9, lanes 2-7). Both precursor and processed forms of ColV-1 were quite stable even after overnight incubation at 37°C. When the same film was exposed longer, a faint processed form-like band is visible in Fig. 9, lane 3; however, this band is also visible, although weaker in the precursor sample (data not shown). Even if this is a real processed band, this activity is more than 6,000 times lower than that in the optimal condition. When the precursors was incubated with CvaA and CvaB membrane vesicles and energy without additional soluble factors, some processing activity (about 2%) was detected (Fig. 9, lanes 10 -12). Since the in vitro synthesized ColV-1 precursor preparation contained some endogenous soluble protein (about 5% of the optimal amount), it may contribute to this residual activity. Hence, taking all of these data together, it is indicated that both energy and soluble factors are essential for efficient ColV-1 processing, and membranes containing the CvaA and CvaB transporter alone is not sufficient for the processing; presumably these findings can be extended to the processing of native ColV processing. DISCUSSION Amino-terminal processing appears to be the first step during the secretion of ColV. In this study, we characterized the processing of colicin V using a marker protein, ColV-1. An in FIG. 5. Preliminary characterization of the soluble factors. A, the soluble factor is protein in nature, based on its sensitivity to proteinase K (PK). The assays were as described in Fig. 4. Lane 1, precursor alone. Treatment of S100 with proteinase K was at 0°C for 10 min. 2 mM phenylmethylsulfonyl fluoride (PMSF) was added after (lane 3) or before (lane 4) proteinase K (100 g/ml) treatment. In lane 6, S100 was treated with micrococcal nuclease (500 g/ml) and Ca 2ϩ at 25°C for 30 min, and 5 mM EGTA was added after incubation. B, Purified GroEL, GroES, DnaK, DnaJ, and GrpE are not the enhancing factors. The purified proteins were tested in the assay as indicated. Lane 1, precursor of ColV-1. Lane 2, 125 g of D10 S100 was added. C, purified SecA, SecB, and trigger factor (T.F.) do not enhance the processing. Purified proteins were tested in the assay as indicated. Lane 1, precursor of ColV-1. Lane 2, 125 g of D10 S100 was added.
Under the condition found in ColV-1, wild type ColV was also similarly processed, but it was very unstable. 2 Mature wild type ColV has been shown to lack any post-translational modification beyond the amino-terminal 15-residue removal (32,33). Here, the processed form of ColV-1 in the in vitro assay has also been radioactively sequenced by double labeling with [ 35 S]Met and [ 3 H]Gly; 3 H activity was specifically detected in 3rd and 11th cycles and 35 S activity was detected in the 8th cycle, 2 corresponding to the predicted processing site. Therefore, the requirements for processing of ColV-1 probably reflect those for wild type ColV.
The ColV leader sequence shares a common predicted secondary structure with the leader sequences of many other peptide antibiotics, especially the lactococcins (32-34, 42, 43). While this work was in progress, two reports (40,41) proposed that bacteriocin ABC transporters have protease functions that process the double glycine type leader peptide concomitant with export. It was reported that a fragment containing the 150 amino-terminal amino acids of LagD (the lactococcin G ABC transporter) was able to cleave its cognate prebacteriocin and that expression of the amino-terminal domain of PedD (the pediocin ABC transporter) in cells was sufficient for prepediocin processing, suggesting that LagD and PedD possess the processing activity. However, the processing activity of the LagD fusion protein was low: it required overnight incubation at 37°C (40), and the expression of one PedD domain in cells does not rule out the involvement of other cellular factors (41). In our system, the processing is efficient with all required components (almost completely processed within a few minutes at 37°C). It therefore seems closer to the physiological enzymatic reaction that occurred in cells. Such processing occurs in the inner membrane with an intact CvaA-CvaB-TolC complex. Our findings that processing of ColV-1 can be inhibited by cysteine protease inhibitors are consistent with the findings (40) that replacement of cysteine with an alanine at residue 13 of LagD completely abolishes the activity of the enzyme. Since both ColV-1 and CvaA do not contain any cysteine residues (27,32), it is possible that CvaB could be the target of NEM and antipain and that CvaB could be the peptidase. Since the CvaA-CvaB membrane alone has little processing activity even with overnight incubation at 37°C, additional soluble factors, which may function as cofactors, are necessary for ColV-1 processing. Although we have not established the protease responsible for the processing activity, it is quite clear that the process is quite specific and dedicated, requiring soluble factors, energy, and membrane integrity with intact CvaA, CvaB, and TolC.
The findings in vitro that no ColV-1 processing was observed with CvaA Ϫ , CvaB Ϫ , or TolC Ϫ membrane vesicles are also consistent with in vivo data (31). A tolC Ϫ mutant abolishes the majority of extracellular ColV activity (27), however, processing in the tolC Ϫ strain appears to be similar to the wild type, albeit slightly slower (31). In our in vitro assay, there is no detectable processing for ColV-1 without CvaA, CvaB, or TolC. This difference may be due to the sensitivity of the assays. The activity of isolated membrane vesicles from the tolC Ϫ mutant strain may also be significantly affected by either the instabil-2 X. Zhong and P. C. Tai, unpublished data.  9. CvaA-CvaB membrane vesicles alone have no processing activity. ColV-1 precursor was generated by using pXZ10 for in vitro synthesis. The in vitro processing assay was the same as described in Fig. 8. Lane 1, precursor marker. In lanes 2, 3, and 8 -12, MC4100/ pHK11 membrane vesicles were added. In lanes 4 and 5, MC4100 membrane vesicles were added. In lanes 6 and 7, MC4100/pHK11-4 membrane vesicles were added. Lanes 10 -12 were from a separate gel. Energy (ATP-regenerating system) and S100 were added as indicated. The reaction mixtures were incubated at 37°C for the times indicated. ity or reduced the expression of CvaA-CvaB transporters in a tolC Ϫ background. 3 Furthermore, our in vitro results also indicate that processing of ColV-1 requires an intact membrane for the transporters to function, suggesting that the processing may require a specific conformation of the transporter, precursor, or peptidase for it to be active in the membranes. Since the mutants we tested are all null mutants, we cannot yet specify the exact role of each component, due to the stability effect of the whole complex in the assay. Missense mutants will be necessary to address this question.
One of the interesting findings in this study, however, is that processing of ColV-1 also requires soluble protein factor(s). It probably serves as a targeting factor, a cofactor, or chaperone for ColV-1 to interact with membrane transporters for processing and secretion. This interaction is specific and critical, since without the soluble factors, little ColV-1 is processed, although some precursors can nonspecifically bind to membrane vesicles in the assay. Many heat shock proteins function as molecular chaperones, which can maintain presecretory proteins in a translocation-competent conformation in the cytosol in the major secretion Sec system. However, we have shown that SecA, SecB, trigger factor, GroEL, GroES, DnaK, DnaJ, and GrpE are not involved in enhancing the processing of ColV-1. Preliminary fractionation of the soluble factors revealed that multiple factors are involved. It is possible that soluble factors may serve as cofactors for the peptidase to function. In this regard, it is interesting that nucleotide hydrolysis is required for the processing, and that GTP seems to be the most efficient for processing. It has been demonstrated that ATP is hydrolyzed during transport in reconstituted systems directed by the eukaryotic multidrug resistance transporter and the bacterial permeases HisP and MalK (44 -48). In a reconstituted histidine transport proteoliposome system, it was reported that GTP could replace ATP reasonably well (70%; Ref. 45), indicating that this type of ATPase has less substrate specificity. However, it does not readily explain why GTP in the ColV-1 system is a more efficient substrate. Therefore, it raises the possibility that a GTP-binding protein might be involved in the processing of ColV-1. Ffh/4.5S ribonucleoproteins contain a GTP binding domain and have greatly stimulated GTPase activity when combined with the receptor protein FtsY, and the complex has been proposed to be involved in the targeting (49,50). The possible role of this complex in ColV-1 processing is under investigation. Computer analysis of the CvaA primary sequence reveals that it also contains a nucleotide binding motif, suggesting that CvaA may be involved in this GTP requirement. Moreover, CvaB is an ABC transporter; therefore, ATP should theoretically be involved. Indeed, in our hands, both GTP and ATP nonhydrolyzable analogs have significant inhibition on processing, 2 suggesting that both GTP and ATP are involved. On the other hand, processing is only one step in the export processes. So far, complete secretion into membrane vesicles, as defined by resistance to protease digestion, has not been demonstrated in this in vitro system. The requirements for this complete secretion process, including involvement of nucleotides, must await further studies.