Studies on a Novel Serine Protease of a ΔhapAΔprtV Vibrio cholerae O1 Strain and Its Role in Hemorrhagic Response in the Rabbit Ileal Loop Model

Background Two well-characterized proteases secreted by Vibrio cholerae O1 strains are hemagglutinin protease (HAP) and V. cholerae protease (PrtV). The hapA and prtV knock out mutant, V. cholerae O1 strain CHA6.8ΔprtV, still retains residual protease activity. We initiated this study to characterize the protease present in CHA6.8ΔprtV strain and study its role in pathogenesis in rabbit ileal loop model (RIL). Methodology/Principal Findings We partially purified the residual protease secreted by strain CHA6.8ΔprtV from culture supernatant by anion-exchange chromatography. The major protein band in native PAGE was identified by MS peptide mapping and sequence analysis showed homology with a 59-kDa trypsin-like serine protease encoded by VC1649. The protease activity was partially inhibited by 25 mM PMSF and 10 mM EDTA and completely inhibited by EDTA and PMSF together. RIL assay with culture supernatants of strains C6709 (FA ratio 1.1+/−0.3 n = 3), CHA6.8 (FA ratio 1.08+/−0.2 n = 3), CHA6.8ΔprtV (FA ratio 1.02+/−0.2 n = 3) and partially purified serine protease from CHA6.8ΔprtV (FA ratio 1.2+/−0.3 n = 3) induced fluid accumulation and histopathological studies on rabbit ileum showed destruction of the villus structure with hemorrhage in all layers of the mucosa. RIL assay with culture supernatant of CHA6.8ΔprtVΔVC1649 strain (FA ratio 0.11+/−0.005 n = 3) and with protease incubated with PMSF and EDTA (FA ratio 0.3+/−0.05 n = 3) induced a significantly reduced FA ratio with almost complete normal villus structure. Conclusion Our results show the presence of a novel 59-kDa serine protease in a ΔhapAΔprtV V. cholerae O1 strain and its role in hemorrhagic response in RIL model.


Introduction
Proteases are enzymes that catalyze the hydrolysis of peptide bonds in proteins or peptides. They are either exopeptidases, whose actions are restricted to the N-or C-termini of proteins, or endopeptidases which cleave internal peptide bonds. Microbial peptides are predominantly secreted enzymes and can be classified based on the essential catalytic residue at their active site. They include serine proteases, cysteine proteases, aspartate proteases and metalloproteases. Proteases produced by pathogenic microorganisms play an important role in virulence [1]. Tissue barriers to pathogen invasion, such as extracellular matrices, epidermal keratinocyte layers and blood vessel walls, may be targeted by bacterial proteases. Proteolysis of host tissue components such as extracellular matrix proteins, including collagen, laminin, fibronectin and elastin, can induce necrotic tissue damage [2,3]. V. cholerae O1, the causative agent of epidemic cholera, secretes a 32-kDa zinc-containing hemagglutinin protease that may play a role in the pathogenesis of cholera. V. cholerae secretes hemagglutinin/protease (HAP), which is encoded by the hap gene [4,5]. HAP can perturb the paracellular barrier function in epithelial cells by degrading occludin in tight junctions [6,7]. HAP nicks the cholera toxin [8] and digests proteins, such as mucin, fibronectin, lactoferrin, and secretory immunoglobulin A, that may participate in host defense against cholera [9]. HAP can also hydrolyze mucin to enhance the detachment of V. cholerae from cultured epithelial cells [10]. A CTXQand hap-defective vaccine strain, 638, was not reactogenic in human volunteers and induced lower levels of IL-8 than its parent wild-type strain in HT29 cells [11]. The protease activity in V. cholerae vaccine strains reduced the transcellular epithelial resistance of polarized T84 intestinal epithelial cells [6]. These results suggested a role of HAP in reactogenicity, including inflammatory diarrhea. In our earlier studies, we have reported that HAP may play a role in the pathogenesis of a ctx-negative V. cholerae non-O1, non-O139 strain by inducing a hemorrhagic fluid response in the RIL assay [12]. Histopathological examination of 20 mg of purified protease-treated rabbit ileum showed the presence of erythrocytes and neutrophils in the upper part of the villus lamina propria [12].
Although HAP is a very active virulence factor, an isogenic strain of V. cholerae mutated in the hap gene was no less virulent in infant rabbits than the parental strain [13]. Fullner et al reported that a hap mutant is more lethal in a pulmonary murine model and caused more severe histopathological damage than its wild-type parent in the lungs of survivors, although no difference was seen in the induction of inflammation [14]. Studies by Zhou et al suggest that an IL-8 stimulator other than HAP may be responsible for inflammation contributing to the reactogenicity of attenuated V. cholerae vaccine strains [15]. An earlier study by Hase et al., [5] showed that a hapA-deleted mutant of V. cholerae O1 had reduced extracellular proteolytic activity compared with the parental strain in a skim milk assay, indicating that the mutant still produces some extracellular proteolytic activity. In addition, residual proteolytic activity expressed by the hapA-deleted mutant is distinct from HAP, as demonstrated by failure of anti-HAP serum to inhibit the activity of this secondary protease on milk agar. The mutant strain also failed to agglutinate chicken erythrocytes [5]. Young and Broadbent [16] described several extracellular proteases in V. cholerae that could explain the residual proteolytic activity of the hap-negative V. cholerae mutant. Besides HAP, the other major wellcharacterized protease in V. cholerae is a 97-kDa Vibrio cholerae protease, PrtV. PrtV plays a role in virulence in a C. elegans infection model [17].
In the present study, a hap and prtV double knock out mutant of V. cholerae strain CHA6.8DprtV still had residual protease activity. This protease was partially purified from strain CHA6.8DprtV and MS peptide mapping and sequence analysis of the protein revealed homology with a 59-kDa trypsin-like serine protease encoded by VC1649. To our knowledge, this is the first report of a serine protease in V. cholerae O1 and demonstration of its role in hemorrhagic response in the RIL model.

Bacterial strains, plasmids and primers used in this study
The bacterial strains used in this study and their relevant properties are summarized in Table 1. All the strains were maintained at 280uC in 30% glycerol in tryptic soy broth (TSB, Difco laboratories). For protease purification, a DhapADprtV mutant of V. cholerae O1 CHA6.8DprtV was used. Antibiotics were used at the following concentrations unless otherwise indicated: ampicillin (Am), 100 mg/ml; streptomycin (Sm), 100 mg/ml; carbenicillin (Cb), 100 mg/ml; kanamycin (Km), 50 mg/ml for E. coli and 40 mg/ml for V. cholerae.
Construction of a hapA knock -out mutant in Vibrio cholerae O1 strain C6709 The bacterial mutant was constructed by replacing the hapA gene with its deletion-insertion allele DhapA::kan using published methods [18]. One such mutant showing a Sm r Km r phenotype was selected for further study and designated CHA6.8 ( Table 1). The in-frame hapA gene deletion in the strain C6709 was confirmed by PCR with hapA internal primers ( Table 1).
Construction of a DhapADprtV V. cholerae O1 strain, CHA6.8DprtV The bacterial mutant was constructed by double crossover method using the construct as mentioned previously [17]. The knock out mutant (CHA6.8DprtV) was confirmed with prtV internal primers ( Table 1).
A VC1649 in-frame deletion mutant was constructed using published methods [18,19]. Several colonies were purified from the plates, tested for Cb sensitivity and then analyzed for the deletion and confirmed with internal primers for hapA, prtV, VC1649 and ctx (Table 1).

Azocasein assay
Casein was chosen as the substrate to assay proteolytic activity. The substrate-enzyme mixture was incubated at 37uC and the reaction was terminated with 10% trichloroacetic acid after 1 hour. The precipitated protein was removed by centrifugation (12,0006 g for 4 mins), and the supernatant was transferred to a clean tube containing 525 mM NaOH. Absorbance was measured at 440 nm using a Smartspec spectrophotometer (Bio-Rad, Hercules, CA). Substrate with buffer and substrate with inhibitors were used as negative controls.

Skim milk assay
Single colonies of V.cholerae O1 strains C6709, CHA6.8, CHA6.8DprtV and CHA6.8DprtVDVC1649 were streaked onto nutrient agar (NA) plates containing 1.5% skim milk and incubated at 37uC overnight. Protease activity was detected by clearing of the opaque milk proteins incorporated into the NA.

Inhibition of protease activity
The azocasein assay was done with 30 mg of ammonium sulphate precipitated proteins from culture supernatants of C6709, CHA6.8, CHA6.8DprtV and CHA6.8DprtVDVC1649. The protease activity of crude proteins was also tested for inhibition with 25 mM PMSF, 10 mM EDTA and 10 mM 1,10-phenanthroline. Twenty-five mM Tris-HCl and 25 mM Tris-HCl in the presence of 25 mM PMSF, 10 mM EDTA and 10 mM 1,10-phenanthroline were used as negative controls. The NB (non-binding) fraction (5 mg) of the CHA6.8DprtV strain with protease inhibitors (10 mM EDTA, 10 mM EGTA, 25 mM PMSF, 10 mM EDTA with 25 mM PMSF, 1 mg/ml aprotinin, 28 mM E-64, 1 mg/ml leupeptin and 10 mM 1,10-phenanthroline) were incubated for 30 mins at 37uC and assayed by azocasein assay. Twenty-five mM Tris-HCl was used as a negative control. The protease activity with EDTA was measured both in the presence and absence of 10 mM CaCl 2 . The mean with standard deviation of three experiments was considered for data analysis.
Partial purification of a novel protease from strain CHA6.8DprtV The DhapADprtV V. cholerae O1 strain CHA6.8DprtV was grown in 3 l of tryptic soy broth (TSB) for 18 h under agitation in an orbital shaker (OSI503; Firstek Scientific). Cells were harvested by centrifugation at 8,0006 g for 20 min at 4uC in a SS34 rotor (Sorvall, Newtown, Connecticut). The protein in the cell-free

Native PAGE
The proteins were separated by electrophoresis on a 10% native polyacrylamide gel according to the procedures described by Davis et al [20] in the absence of SDS and 2-mercaptoethanol. Protein samples were mixed with sample buffer containing 10% glycerol, 0.05% bromophenol blue and Tris-HCl pH-6.8, resolved in the gel and bands were visualized by staining with Coomassie brilliant blue.

Protein identification by MS peptide mapping and sequencing analysis
The major band observed in the native PAGE of the nonbinding pooled fraction from the DE-52 column was excised from the Coomassie blue stained gel and analysed on a Bruker Autoflex III MALDI TOF/TOF instrument at Alphalyse, Odense, Denmark. The peptide mixture was analyzed in positive reflector mode for accurate peptide mass determination and 5-10 of the peptides were selected for analysis by MS/MS fragmentation for partial peptide sequencing. The MS and MS/MS spectra were combined and used for a database search in an in-house protein database using the Mascot software. The peptides used for the identification are highlighted in the sequence. Peptides confirmed by MS/MS sequencing are shown in bold.

DNA sequencing
The VC1649 gene (ORF) was amplified by using primers 59CCGTTCATACTCTGCAATAG39and59CACGGTTGGCC-TCGAGTAAACAAAATTGGC39. The resultant PCR product was analyzed by 0.7% agarose gel electrophoresis. The sequencing was done on an ABI 3130 DNA analyzer (Applied Biosystems, Foster City, CA) and the sequences were aligned, analyzed using Clustal X and NCBI/BLAST programs. The nucleotide sequence data reported in this paper will appear in the DDBJ/EMBL/ GenBank nucleotide sequence database with the accession number AB572560.

Rabbit ileal loop assay
The rabbit ileal loop (RIL) assay was performed in young New Zealand White rabbits (2 kg) essentially by the method described by De and Chatterjee [21]. Culture supernatants (one ml) and washed bacterial cells (10 9 cfu/ml) of C6709, CHA6.8, CHA6.8DprtV, CHA6.8DprtVDVC1649 grown in tryptic soy broth were inoculated in rabbit ileum. Tryptic soy broth was used as negative control in the above assay. The partially purified protease at a concentration of 50 mg (non-binding pooled fraction eluted from DE-52 column) from CHA6.8DprtV strain, similar concentration of protease inhibited with 25 mM PMSF and EDTA and 25 mM Tris-HCl with PMSF and EDTA (negative control) in a volume of 1 ml were also assayed in RIL assay. The animals were sacrificed after 18 hrs and the enterotoxic response was determined by measuring the fluid accumulation (FA) ratio, which is the ratio of the volume of fluid accumulated in the intestinal loop to the length of the loop. A ratio of greater than 1.0 indicated a strong positive response, while a negative response was defined as FA ratio of less than 0.2.

Histopathological studies
Tissue samples (2 cm in length) from RIL assays were collected and placed in 10% neutral-buffered formalin for histopathological analysis. Tissues were embedded in paraffin and processed following the standard protocol. Sections (3 to 4 mm thick) prepared with a Leica rotary microtome were stained with hematoxylin and eosin and examined by light microscopy. Photographs were taken under different magnifications with a Leica DMLB microscope (Solms, Germany), equipped with a digital imaging system.

Construction of DhapA and DhapADprtV V. cholerae O1 mutant strains
The in-frame hapA deletion in the strain C6709 was confirmed by PCR using internal primers ( Table 1). The CHA6.8DprtV mutant was constructed by the double crossover method as described previously (Vaitkevicius et al 2006). The knock out mutant CHA6.8DprtV was confirmed using prtV specific internal primers (Table 1).
Protease activity in C6709, CHA6.8 and CHA6.8DprtV strains The proteins from culture supernatants of the above strains were precipitated with ammonium sulphate and after dialysis, 30 mg of crude proteins were tested for protease activity by azocasein assay. As shown in Fig. 1A, V. cholerae strain C6709 had maximum protease activity compared to the other strains. The protease activity in strain C6709 was inhibited in the presence of EDTA and 1, 10-phenanthroline. 1, 10-phenanthroline, a specific metalloprotease inhibitor, inhibited 90.7% of the protease activity, confirming that the major protease activity in this strain is due to metalloproteases (Fig. 1A). EDTA, a non-specific metalloprotease inhibitor, repressed 61% of the protease activity (Fig. 1A). PMSF, a serine protease inhibitor, did not change the protease activity, suggesting that there is no serine protease activity in C6709 (Fig. 1A). V. cholerae O1 strain CHA6.8, in which the hapA gene has been knocked out, had a 78.7% decrease in protease activity when compared to strain C6709. The protease activity of crude proteins from CHA6.8 in the presence of 1, 10-phenanthroline decreased by 21.1%, indicating the presence of other metalloprotease in the absence of HAP in strain CHA6.8. Both EDTA and PMSF significantly decreased protease activity in strain CHA6.8 by 58.9% and 63.9%, respectively. The major protease present in strain CHA6.8 is a serine protease, although metalloproteases are also present. Besides HAP, the other well-characterized metalloprotease in V. cholerae is the 97-kDa V. cholerae protease (PrtV). The Dhap and DprtV V. cholerae O1 strain CHA6.8DprtV had decreased protease activity by 29.6% compared to strain CHA6.8, indicating that PrtV also contributes to the protease activity of strain CHA6.8 (Fig. 1A). Both EDTA and PMSF significantly decrease protease activity, by 48.7% and 47%, respectively, suggesting that the major protease activity in the CHA6.8DprtV strain is due to a serine protease. The 48.7% decrease in CHA6.8DprtV due to EDTA is not due to a metalloprotease, as 1, 10-phenanthroline, a specific metalloprotease, only reduced protease activity by 8.4% (Fig. 1A).
All strains included in this study were also tested for protease activity in a skim milk assay. As shown in Fig. 1B, the zone of proteolysis created by C6709 on skim milk agar was clear, indicating complete degradation of the milk proteins. With the other two strains, CHA6.8 and CHA6.8DprtV, the zone of proteolysis was hazy, indicating that not all of the milk proteins were degraded. These results suggest that the substrate specificity of proteases in CHA6.8, and CHA6.8DprtV may be different from that of C6709, which secretes HAP.
Partial purification of protease from the strain CHA6.8DprtV The ammonium sulphate precipitated proteins from culture supernatants of CHA6.8DprtV were loaded onto an anionexchange chromatography column (DE-52). The proteins in the non-binding fraction of the column ( Fig. 2A) were pooled and concentrated. The bound proteins were eluted with 0.1 M (Fig. 2B) and 0.3 M (Fig. 2C) NaCl, dialyzed against 25 mM Tris-HCl buffer and concentrated. When protease activity in the NB, 0.1 M and 0.3 M NaCl eluted fractions were tested by azocasein assay, the major protease activity was present in the NB fraction (Fig. 2D). The NB-pooled fraction was concentrated and run on a native PAGE (Fig. 2E). The major protein band was excised and analyzed by MS/MS sequencing (Fig. 2E). The sequences highlighted showed homology with a 59-kDa serine protease encoded by the gene VC1649 (Fig. 2F). The sequence GDSGGP (underlined) flanks the serine residue in trypsin-like serine proteases (Fig. 2F).

The presence of a calcium-dependent serine protease
To determine the nature of the partially purified protease from CHA6.8DprtV eluted in the non-binding fraction of a DE52 anion- exchange column, we performed protease inhibition assays with several inhibitors (Fig. 2G). The protease was partially inhibited in the presence of EDTA (60.3%), EGTA (59.2%) and PMSF (60.3%). The partially purified protease was completely inhibited when PMSF and EDTA are used together (Fig. 2G). There was significantly less inhibition of protease activity in the presence of 1,10-phenanthroline (9%), aprotinin (10.5%), leupeptin (8.7%) and E64 (1.8%). Although EDTA inhibited protease activity by 60.3%, EDTA in the presence of CaCl 2 inhibited activity by only 3.6% inhibition (Fig. 2G). The serine protease secreted by CHA6.8DprtV is a calcium-dependent serine protease.

Construction of CHA6.8DprtVDVC1649 deletion mutant
The CHA6.8DprtVDVC1649 deletion mutant was constructed as described in the text and confirmed by internal primers for hapA, prtV and VC1649 as shown in Fig. 3A. PCR with internal primers for the ctx gene confirmed the presence of ctx in both the C6709 and the CHA6.8DprtVDVC1649 strains. PCR with internal primers for hapA, prtV and VC1649 in strain CHA6.8DprtVDVC1649 showed the absence of any band by agarose gel electrophoresis (Fig. 3A) and also confirmed the absence of the deleted gene sequences. On the other hand, strain C6709 showed the presence of PCR products with the internal primers confirming the presence of hapA, prtV and VC1649 genes (Fig. 3A). A native PAGE profile of crude proteins from CHA6.8DprtV and CHA6.8DprtVDVC1649 showed the absence of a 59-kDa band in the CHA6.8DprtVDVC1649 strain (Fig. 3B).
The VC1649 gene sequence showed complete homology with the published sequence of VC1649 from V. cholerae O1 biovar El Tor strain N16961 (NCBI).
We initially started the study to characterize the proteases present in the CHA6.8DprtV strain. Partial purification of proteases from CHA6.8DprtV strain showed the presence of a 59-kDa trypsin-like serine protease encoded by the VC1649 gene. The major protease present in the CHA6.8DprtV strain is a serine protease, but protease activity in CHA6.8DprtVDVC1649, in which the 59-kDa serine protease is not secreted, increased by 11.4% (Fig. 1A). These results indicated that besides HAP, PrtV and a 59-kDa serine protease, there are still other proteases secreted by the CHA6.8DprtVDVC1649 strain. PMSF and EDTA inhibited protease activity in CHA6.8DprtVDVC1649 by 64.5% and 46.2%, respectively, whereas 1,10-phenanthroline only inhibited protease activity by 6.2% suggesting the presence of another serine protease (Fig. 1A). Skim milk assay with CHA6.8DprtVDVC1649 still showed residual protease activity (Fig. 1B).

The serine protease induces a hemorrhagic fluid response in RIL
To study the role of the 59-kDa serine protease in virulence, 50 mg of the partially purified protease was injected into the rabbit ileum, which induced significant hemorrhagic fluid accumulation (FA ratio 1.2+/20.2, n = 3, Fig. 4A). When a similar concentration of the protease was incubated in the presence of PMSF and EDTA and injected into the rabbit ileum, there was a significant decrease in fluid accumulation (FA ratio 0.3+/20.05, n = 3, Fig. 4A). Histopathological analysis of the rabbit ileum revealed that the protease caused extensive damage to all the layers of the mucosa. There was damage to the villus structure, which was completely destroyed. We observed gross damage of the villus surface structure with hemorrhage in all layers of the mucosa (Fig. 5A). On the other hand, analysis of the ileal tissues treated with the protease in presence of both PMSF and EDTA, revealed normal microvillus structure with no gross alteration in villus structure, although the villus lamina propria was slightly dilated and RBCs had accumulated in a few places in the basal area (Fig. 5B). PMSF and EDTA completely inhibited protease activity (Fig. 4A), but still we observed some residual effect in the rabbit ileal loop. This effect could be due to some other domain in the protease, which may not be its proteolytic domain, and could be responsible for causing damage to the ileal tissue. Tissues treated with 25 mM Tris-HCl and PMSF + EDTA did not cause fluid accumulation in RIL (FA ratio 0.12+/20.002 n = 3, Fig. 4A) and histopathology of the ileal tissue showed normal microvillus structure (Fig. 5C).
One ml of culture supernatant of C6709, CHA6.8 and CHA6.8DprtV strain induced significant fluid accumulation (Fig. 4B)   in the sub-mucosal layer (Fig. 5D). Ileal tissues treated with culture supernatant from CHA6.8 strain showed widely dilated villi with rupture at places with gross hemorrhage and inflammatory cells in mucosa and sub-mucosa (Fig. 5E). Ileal tissues treated with culture supernatant of CHA6.8DprtV strain also showed dilated villi with gross hemorrhage in all layers of the mucosa (Fig. 5F). The same section at a higher magnification 40X showed ruptured villi with hemorrhage and inflammatory cells in mucosa and sub-mucosa (Fig. 5G). The ileal tissues treated with culture supernatant of CHA6.8DprtVDVC1649 strain protease showed villous architecture almost normal with minimum hemorrhage in mucosa and sub-mucosa (Fig. 5H). TSB treated ileal tissue showed normal gut mucosa (Fig. 5I).

Discussion
Bacterial proteases are an important virulence factor in a variety of organisms, causing massive tissue damage which may aid the bacteria in host cell entry [22]. The major protease secreted by V. cholerae is HAP, a member of a large family of metalloproteases. HAP is produced by both non-pathogenic and pathogenic species, including the elastase of P. aeroginosa [23]. It acts on potentially relevant substrates like mucin, fibronectin, lactoferrin and the A subunit of CT [8,10]. In our earlier studies, we demonstrated that  (Fig. 4A, NB). Gross damage of the villus surface structure was observed with hemorrhage in all layers of the mucosa. Magnification, 20X. B) Almost normal villous architecture observed in ileal tissues treated with 50 mg of partially purified protease inhibited with 25 mM PMSF and 10 mM EDTA (Fig. 4A, NB+PMSF+EDTA). This photomicrograph shows no gross alteration in villus structure but villus lamina propria are slightly dilated and RBC have accumulated at a few places in the basal area. Magnification, 20X. C) Ileal tissues treated with 25 mM Tris-HCl buffer with PMSF and EDTA (Fig. 4A, control)  HAP may play an important role in the pathogenesis of ctxnegative V. cholerae non-O1, non-O139 strains [12]. We have also shown that the processed 35-kDa form of HAP induces a dosedependent hemorrhagic response in the RIL assay, a decrease in the intestinal short circuit current (Isc) in an Ussing chamber, and a cell rounding effect on HeLa cells. Fullner et al., reported that the deletion of hap in V. cholerae did not affect the production of IL-6, or macrophage inflammatory protein 2 in a murine pulmonary model, and the hap mutant was more virulent than its wild-type parental strain, although the mechanism was not clear [14]. When V. cholerae hapA mutants were tested in the C. elegans killing assay, hapA deleted strains were not attenuated compared to wild-type V. cholerae O1 [17]. The culture supernatant from a hapA mutant contained proteins bands encoded by the ORFs VCA0812, VCA0813, and VCA0223 [24], as determined by mass spectrometry. The protein products are a leucine aminopeptidase-related protein, leucine aminopeptidase (Lap) [25], and the PrtV protease [26], respectively. When DprtV, Dlap, and DlapX mutants were tested in the C. elegans assay, the DprtV mutant was completely attenuated compared to the wild-type strain. The PrtV protein is a factor required for the V. cholerae lethal infection of C. elegans [17]. Earlier studies with hapA mutant strains showed that deletion of this gene still produces some extracellular proteolytic activity [5] Measurements of proteolytic activity against azocasein indicated that 10-20% of total activity in culture supernatants was abolished by the DprtV mutation [17]. In the same study, deletion of hapA reduced the total protease activity to 10% of the wild-type level. HAP, being the major protease in V. cholerae O1, could mask the other secretory proteases. As shown earlier, strain C6709 does not exhibit serine protease activity, but in absence of hapA the serine protease is secreted by strain CHA6.8. Our results suggest that the expression of proteases in V. cholerae may follow a cascade of events. HAP controls secretion of the 59-kDa serine protease, which in turn may control secretion of other proteases. Further experiments are being done to confirm these results.
The role of proteases other than HAP can best be studied in hapA mutant strains. The objective of our study was to identify the protease present in the DhapA, DprtV V. cholerae O1 strain CHA6.8DprtV and study the role of this protease in pathogenesis. The protease was partially purified and its activity was observed in the non-binding fraction of an anion exchange column. The major band of this partially purified protease, as visualized by native PAGE, was analyzed by mass peptide sequencing and found to be homologous to a trypsin-like serine protease encoded by the VC1649 gene. The serine protease also had the GDSGGP sequence normally associated with trypsin-like serine protease [27]. Interestingly, when the nature of this partially purified protease was studied using inhibitors, it was observed that EDTA, EGTA and PMSF could partially inhibit its protease activity. Protease activity was completely inhibited in the presence of PMSF and EDTA together. The specific metalloprotease inhibitors, like 1, 10-phenanthroline, could not inhibit the protease activity, nor could the other inhibitors like E-64, aprotinin, leupeptin and bestatin. EDTA with CaCl 2 failed to inhibit protease activity. Our results showed that the protease present in the hapA,prtV-deleted strain is a calcium-dependent serine protease. In an earlier study by Young and Broadbent [16], 100 strains of V. cholerae El Tor from different parts of the world were screened for protease production by a rapid assay with gelatin agar plates. Based on protease production, the strains were classified as high, medium and low protease producers. Protease I activity (as shown by PMSF inhibition) was detected only in low protease producers, whereas protease II activity (marked stimulation by EDTA) was associated with the high protease producers. Protease III activity (EDTA inhibition) was difficult to detect in the presence of large amounts of protease II, but it was seen in some low protease producers. It is possible that the absence of protease I in the high protease producers is simply due to masking of this activity by the much larger amounts of protease II. Filtrates from the low protease producing strain 1621 contained predominantly type I protease activity, which is sensitive to serine protease inhibitors such as PMSF and the lima bean trypsin inhibitor. Activation of CT by limited proteolysis of the A subunit is also sensitive to serine protease inhibitors [28,29], and it seems likely, therefore, that this involves the type I protease. These results were, however, carried out with strain 569B, a low protease producer in which type I protease is readily detectable. This report clearly suggests that in high protease producers, in which EDTA can inhibit activity, the protease could be hemagglutinin protease; in strains in which HAP is not secreted, the serine protease could be the major protease. The 59-kDa serine protease could be the major protease in classical strains like 569 B. Molecular genetic analysis of classical biotype V. cholerae strains that caused cholera outbreaks in 1942 in Russia showed that these strains contain the gene hapA, demonstrated by PCR, but produce no soluble HAP [30]. It would be interesting to study the presence and role of the 59-kDa serine protease in such classical V. cholerae O1 strains.
The genus Vibrio consists of many pathogenic species that include V. cholerae, V. parahemolyticus, V. vulnificus, V. mimicus and V. fluialis [31]. In addition to toxins and hemolysin produced by vibrios, protease is also recognized as one of the pathogenic factors in some Vibrio species [3]. The proteases in vibrios are divided into two main groups, the zinc metalloproteases and the serine proteases. V. cholerae and V. vulnificus secrete proteases belonging to the thermolysin family of metalloproteases which have a zinc ion and are immunologically cross-reactive with each other [3]. There are no studies on the role of serine protease in V. cholerae. Existence of the thermolysin family of zinc metalloproteases has not been recognized in V. parahemolyticus, although the production of other kinds of proteases including serine proteases have been reported [32]. A 50-kDa serine protease designated as VPPI (Vibrio parahemolyticus protease I) was purified from the culture supernatant of a clinical strain of Vibrio parahemolyticus [32]. VPPI activity was inhibited by EDTA, EGTA and serine protease inhibitors, but not when EDTA was incubated in the presence of CaCl 2 suggesting that it is a calcium-dependent serine protease [32]. The N-terminal amino acid sequence of VPPI was quite similar to that of the Vibrio metschnikovii protease. It was also demonstrated that VPPI or its related proteases is widely distributed in not only V. parahemolyticus but also V. alginolyticus. V. parahaemolyticus protease possesses various toxic activities, including the collagenolytic, cytotoxic, hemolytic and edema-forming activity [33]. V. vulnificus also secretes a 59-kDa serine protease, which is the free form, while the 69-kDa protein may be a complex form associated non-covalently with small peptide(s) [34]. The proteolytic activity of the final preparation was almost completely abolished by treatment with 5 mM PMSF, a well-known inhibitor of serine proteases. By contrast, tetraethylenepentamine, a specific inhibitor of metalloproteases including VVP (Vibrio vulnificus protease), showed no inhibitory effect on the proteolytic activity. Vibrio vulnificus strain NCIMB 2137, in which vvp has been deleted, secretes a serine protease, VVA0302. The serine protease, VVA0302, is an orthologue of an extracellular protease produced by V. parahaemolyticus. V. vulnificus serine protease may be a virulence factor in vibriosis, which is characterized by external and internal hemorrhages affecting the major organs [35], or human wound infection with necrotic tissue damage [36].
Our results show the presence of several proteases in V. cholerae, such as HAP, PrtV, 59-kDa serine protease and other novel proteases. The serine protease from a DhapADprtV V. cholerae O1 strain induced hemorrhagic response in rabbit ileal loop. The strains used for rabbit ileal loop experiments were grown in tryptic soy broth under conditions which are not optimal for CT production (Results not shown). Although Young and Broadbent [16] had earlier reported that strain 569B secretes a protease that is inhibited by PMSF, our study may be the first to demonstrate the presence of a novel 59-kDa serine protease in V. cholerae and its role in hemorrhagic response in RIL model. Studies have shown wide variation in extracellular protease production among different strains of V. cholerae [16]. Among V. cholerae El Tor strains, there was a 100-fold variation in protease production and the two classical strains tested differed in protease production by a factor of 80 [16]. Further studies are in progress to characterize the expression levels of this 59-kDa serine protease in V. cholerae strains of both El Tor and classical biotypes.