Auto-catalytic Cleavage of Clostridium difficile Toxins A and B Depends on Cysteine Protease Activity*

The action of Clostridium difficile toxins A and B depends on processing and translocation of the catalytic glucosyltransferase domain into the cytosol of target cells where Rho GTPases are modified. Here we studied the processing of the toxins. Dithiothreitol and β-mercaptoethanol induced auto-cleavage of purified native toxin A and toxin B into ∼250/210- and ∼63-kDa fragments. The 63-kDa fragment was identified by mass spectrometric analysis as the N-terminal glucosyltransferase domain. This cleavage was blocked by N-ethylmaleimide or iodoacetamide. Exchange of cysteine 698, histidine 653, or aspartate 587 of toxin B prevented cleavage of full-length recombinant toxin B and of an N-terminal fragment covering residues 1–955 and inhibited cytotoxicity of full-length toxin B. Dithiothreitol synergistically increased the effect of myo-inositol hexakisphosphate, which has been reported to facilitate auto-cleavage of toxin B (Reineke, J., Tenzer, S., Rupnik, M., Koschinski, A., Hasselmayer, O., Schrattenholz, A., Schild, H., and Von Eichel-Streiber, C. (2007) Nature 446, 415–419). N-Ethylmaleimide blocked auto-cleavage induced by the addition of myo-inositol hexakisphosphate, suggesting that cysteine residues are essential for the processing of clostridial glucosylating toxins. Our data indicate that clostridial glucosylating cytotoxins possess an inherent cysteine protease activity related to the cysteine protease of Vibrio cholerae RTX toxin, which is responsible for auto-cleavage of glucosylating toxins.

Responsible for the diseases induced by C. difficile are two exotoxins, toxin A and B, that are produced by the pathogen (3)(4)(5). The toxins have masses of 269 to 308 kDa and are suggested to be structured in a tripartite manner (6,7). The N terminus has glucosyltransferase activity and represents the biologically active part of the protein (8) (Fig. 1, upper panel). The crystal structure of the glucosyltransferase domain of toxin B was solved recently and revealed a folding typical for GT-A family glycosyltransferases (9,10). The middle part of the toxins has been suggested to be involved in translocation of the toxins into the cytosol. At the C terminus the toxin consists of polypeptide repeats that appear to be involved in receptor binding (11,12). Recently, it was shown that this part of toxin A has a ␤-solenoid-like structure that binds to carbohydrate structures (13,14). However, the precise nature of the membrane receptors is not known at present. After binding of the toxins to their specific receptors, the toxin-receptor complex is endocytosed and ends up in the early endosomes. The acidic compartment causes conformational changes of the toxin molecules and allows insertion into vesicle membrane, a process that is accompanied by pore formation (15). Then the toxin is translocated into the cytosol, where Rho GTPases, the eukaryotic targets of the toxins, are glucosylated and inactivated. How this translocation occurs is not well understood. Apparently, only the N-terminal catalytic domain is able to reach the cytosol, whereas the rest of the molecule remains in the vesicle membrane (16,17). Thus, the toxin has to be processed by proteolytic cleavage. Recently it was reported that the toxin is autocatalytically activated by a protease activity harbored to the C-terminal side of the translocation domain (18) (Fig. 1, upper  panel). The authors reported that myo-inositol hexakisphosphate (InsP 6 ) 3 is involved in the processing of the toxin. Because the aspartate protease inhibitor 1,2-epoxy-3-(p-nitrophenoxypropane) blocked the processing of toxin B and labeled a motif, which is also found in aspartate proteases, it has been suggested that an inherent aspartate protease is responsible for the auto-catalytic process. We present evidence that not or not only an aspartate protease domain but rather a cysteine protease domain, which is located downstream of the glucosyltransferase domain (Fig. 1, upper panel), is required for processing of the toxins.
Cloning of Toxin B Fragment CDB 1-955 -Primers CDB-1-BamHI-sense and CDB-955-NotI-antisense (stop) were used accordingly to amplify a fragment of C. difficile toxin B (amino acid 1-955) with flanking BamHI and NotI restriction sites FIGURE 1. Multi-domain structure of clostridial glucosylating cytotoxins and partial sequence alignment with V. cholerae RTX toxin. Upper panel, the multi-domain structure of clostridial glucosylating cytotoxins is given on the basis of C. difficile toxin B. At the N terminus the glucosyltransferase domain is located. The DXD motif, which is involved in manganese coordination, is typical for this domain (9,29). In the middle part of the protein, a short hydrophobic region (residues 956 -1128) is located (24) that may be involved in membrane translocation of the catalytic domain. More than 1000 amino acid residues located distant from the proposed auto-cleavage site (amino acid 543), a DXG motif was recently identified by modification with the aspartate protease inhibitor 1,2-epoxy-3-(p-nitrophenoxypropane) and suggested to be involved in processing of the toxin (18). The C terminus of the clostridial glucosylating cytotoxins consists of polypeptide repeats most likely involved in receptor binding (13,14,30). The cysteine protease domain (residues 544 -767) studied in the present communication, which is similar to V. cholerae RTX toxin (23), is located downstream of the glucosyltransferase domain and may be characterized by the catalytic triad aspartate, histidine, and cysteine (DHC). Our studies indicate that this domain is essential for auto-cleavage of the toxin at amino acid 543 and release of the catalytic domain into the cytosol (16). In the lower part, regions of C. difficile toxin A and B, Clostridium sordellii lethal toxin and C. novyi ␣-toxin are shown, which exhibit similarity with the cysteine protease domain of V. cholerae RTX toxin (23). Identical amino acid residues are in bold type. Residues suggested to be part of the cysteine protease catalytic triad are indicated by arrows. Sequences according to UniProt data base (C. difficile toxin B, accession number, P18177; C. sordellii lethal toxin, accession number Q46342; C. difficile toxin A, accession number P16154; C. novyi alpha-toxin, accession number Q46149; V. cholerae RTX toxin, accession number Q9KS12). from genomic DNA of C. difficile VPI 10463 by PCR with Pfu Turbo polymerase (Fermentas GmbH, St. Leon-Rot, Germany). The generated PCR products were cloned into the pCR-Blun-tII-TOPO vector (Invitrogen) and subcloned into pET-28a(ϩ) (Novagen, EMD Biosciences) as BamHI-NotI fragments. pET-28a(ϩ) T7 promotor activity was used for eukaryotic expression.
Generation of Point Mutations within the Recombinant Holotoxin B and CDB 1-955 -Point mutations within pGEX-ToxB (holotoxin B in pGEX 4T-1 (16) and pET-28a-CDB 1-955 were engineered using the QuikChange mutagenesis method (Stratagene) utilizing the according primers. Following PCR, the methylated DNA was digested by the addition of 1 l of DpnI (10 units/l) for 1 h at 37°C and transformed into chemical competent E. coli TG1 cells. Plasmid DNA was prepared using standard procedures and sequenced with the ABI PRISM TM dye terminator cycle sequencing ready reaction kit and an ABI 310 cycle sequencer (PerkinElmer Life Sciences) to confirm gene sequences.
Expression and Purification of Native and Recombinant Proteins-Native toxins A and B from C. difficile VPI 10463 were purified as described (20). Toxin A was additionally purified by thyroglobulin affinity chromatography (21). Recombinant glutathione S-transferase fusion proteins holotoxin B (ToxB rec wt) and holotoxin B mutants were expressed in E. coli BL21(DE3) as recombinant glutathione S-transferase fusion proteins and purified by affinity chromatography with glutathione-Sepharose 4B according to the manufacturer's instructions and as previously described (16). Glutathione S-transferase was cleaved off by thrombin followed by inactivation of thrombin utilizing benzamidine beads. The proteins were analyzed by SDS-PAGE. Recombinant Rac1 was expressed and purified as described (22).
In Vitro Cleavage Assay-Cleavage assays were performed in Tris-HCl buffer, pH 7.5, at a final volume of 20 l with 3 g of toxin A (for Coomassie staining) and 1 g of toxin B (for Western blotting), respectively. The reactions were initiated by addition of DTT, ␤-mercaptoethanol, or InsP 6 at indicated final concentrations. Following incubation for indicated times (at room temperature or at 37°C as indicated), the reaction was stopped by heating the samples at 95°C in Laemmli buffer, and the probes were separated by SDS-PAGE (3-12% gradient gel).
MALDI-TOF Analysis-3 g of native toxin A isolated from culture supernatant of C. difficile strain VPI 10463 were treated with 2 mM DTT at 37°C overnight. The resulting fragments were excised from Coomassie-stained SDS-PAGE gel. The gel fragment was discolored for 2 h at 50°C in stripping buffer (50 mM NH 4 HCO 3 :acetonitrile (4:6), pH 7.8). Dried gel fragments were treated with 2 l of trypsin (1 g/l) in 18 l of trypsin digestion buffer (50 mM NH 4 HCO 3 , 1% acetonitrile, pH 7.8) for 3 h at 37°C. The obtained peptide solution was mixed with a saturated matrix solution of 4-hydroxy-␣-cyanocinnamic acid according to the dried droplet method for matrix crystallization, resulting in a fine granular matrix layer. MALDI-TOF mass spectrometry was performed on a Bruker Biflex mass spectrometer equipped with a nitrogen laser (, 337 nm). The mass spectra were recorded in the reflector-positive mode in combination with delayed extraction. For calibration an exter-nal two-point calibration with 5 pmol of a fragment of human adrenocorticotropic hormone (amino acids 18 -39, (MϩH) ϩ 2,465.20 Da) and 5 pmol of human angiotensin II ((MϩH) ϩ 1,046.54 Da) was used. The data base program ProFound (Rockefeller University, New York, NY) was used for protein identification and assignment of the measured peptides (peptide mapping). The measured peptides were compared with peptides of the theoretically trypsin-digested protein fragments.
Western Blotting-The proteins separated by SDS-PAGE were transferred to polyvinylidene difluoride membrane with a semi-dry blotter (Bio-Rad). The catalytic domain of toxin B was detected using a monoclonal anti-CDB 1-546 at a dilution of 1:100,000 followed by the anti-mouse IgG horseradish peroxidase secondary antibody (1:5000; Biotrend, Cologne, Germany). The corresponding bands were detected using enhanced chemoluminescence (ECL).
In Vitro Transcription/Translation of Toxin B Fragments-The in vitro transcription/translation was performed with TNT coupled reticulocyte lysate systems from Promega (Madison WI) using radioactive labeled [ 35 S]L-methionine (specific activity, 37 TBq/mmol; Hartmann Analytic GmbH, Braunschweig, Germany) and DNA (pET28a constructs) purified with a Plasmid Midi Kit (Qiagen) according to the manufacturer's instructions. Following 90 min of incubation at 30°C, 10 l of the lysate were separated by SDS-PAGE (12%), and the gels were stained with Coomassie Brilliant Blue and dried. The signals were visualized by autoradiography (phosphorimaging; GE Healthcare, Freiburg, Germany).
Intoxication of HeLa and Swiss 3T3 Cells-The cells were seeded in 12-well dishes for microscopic studies or 10-cm dishes for Rac glucosylation assays and cultivated in Dulbecco's minimum Eagle's medium. For intoxication the cells were washed with phosphate-buffered saline, and the medium was changed to fetal calf serum-free Dulbecco's minimum Eagle's medium. The cells were incubated with 1 g/ml recombinant toxin B (ToxB rec wt) and toxin B point mutants (ToxB rec D587N/H653A/C698A) for 4 -6 h at 37°C and 5% CO 2 . Pictures of Swiss 3T3 cells were taken with an AxioCam HRC camera and Axio Vision software (Carl Zeiss AG, Oberkochen, Germany). For the detection of the glucosylation status of Rac1 after intoxication, HeLa cells from one 10-cm dish were washed with phosphate-buffered saline, scraped off, and resuspended in 100 l of hypotonic lysis buffer (20 mM HEPES, pH 7.9, 10 mM KCl, 1 mM EDTA, 10% glycerol, 0.2% Nonidet P-40). After 2 h of shaking at 4°C, the cell solution was centrifuged for 30 min at 13,000 rpm and 4°C. Protein concentration in the supernatant was determined by Bradford, and 20 g of each cytoplasma extract was subjected to SDS-PAGE and Western blotting with anti-Rac1 antibodies as indicated.
Glucosylation Assay-2 g of recombinant Rac1 were incubated with 5 ng of native toxin B or 50 ng of recombinant holotoxin B and holotoxin B mutants, respectively, in the presence of 10 M UDP-[ 14 C]glucose in a buffer containing 50 mM HEPES, pH 7.5, 100 mM KCl, 2 mM MgCl 2 , 1 mM MnCl 2 , and 100 g/ml bovine serum albumin for 15 min at 30°C. The total volume was 20 l. The labeled proteins were analyzed by SDS-PAGE followed by phosphorimaging.
Inhibitor Assay with N-Ethylmaleimide and Iodoacetamide-For InsP 6 -activated cleavage, N-ethylmaleimide was added simultaneously at the indicated concentrations and incubated at room temperature for 30 min. In the case of DTT-activated cleavage, toxin A (3 g) was preincubated at 37°C for 10 min with 0.5 mM DTT. After the addition of 10 mM N-ethylmaleimide or iodoacetamide, 20 mM DTT were added either simultaneously or after 10 min of incubation with N-ethylmaleimide or iodoacetamide at 37°C, respectively. The samples were incubated overnight at 37°C. All of the samples were analyzed by SDS-PAGE and immunoblotting.

RESULTS
Attempting to increase the stability of C. difficile toxin A, we observed that addition of 2 mM DTT did not increase the stability of the toxin but rather caused cleavage of the protein.
Analysis of the resulting fragments, which were generated in the presence of 2 mM DTT by MALDI-TOF mass spectrometry, revealed that the toxin was split into the N-terminal catalytic domain of ϳ60 kDa and a C-terminal part of ϳ250 kDa ( Fig. 2A and supplemental Fig. S1). Closer analysis of the concentration dependence of the effect of DTT showed that 0.01-0.1 mM DTT was necessary for cleavage of toxin A (Fig. 2B). ␤-Mercaptoethanol also caused cleavage of the toxin, but higher concentrations compared with DTT were necessary (Fig. 2C). The process of cleavage was rather slow with a very slight effect after 30 min, which increased with incubation of the toxin for several hours (Fig. 2D). Next we tested the effects of N-ethylmaleimide and iodoacetamide (Fig. 2E). When N-ethylmaleimide or iodoacetamide was added, DTT-induced cleavage of toxin A was blocked. Further addition of DTT could not restore the inhibitory effect of N-ethylmaleimide or iodoacetamide. These findings are in line with the notion that cysteine residues are essential for cleavage.
A similar effect of DTT was observed for the in vitro processing of toxin B. Here we used for detection of the catalytic fragment an anti-toxin B antibody. As shown in Fig. 2F, DTT caused cleavage of the toxin with release of the catalytic domain. The concentration dependence of DTT and the time course (Fig. 2G) of the cleavage reaction of toxin B were very similar to toxin A. It has been reported that toxin A and toxin B have sequence similarity with a part of the RTX toxin of Vibrio cholerae (23). The region of similarity starts shortly after the catalytic domain of toxin B and covers at least residues 546 -767 (Fig. 1, upper  panel). Recently, this similar part of the RTX toxin was identified as a potential cysteine protease (23). Among the identical amino acids in RTX and toxin B, cysteine 698, histidine 653, and aspartate 587 are conserved in RTX toxins and in all clostridial glucosylating toxins (Fig. 1, lower panel) resembling a putative catalytic triad of cysteine proteases. We changed these residues in toxin B and expressed the mutant proteins in E. coli. During purification, the recombinant wild-type toxin B was partially cleaved in a ϳ210-kDa fragment and the catalytic domain (Fig.  3, A and B). By contrast, the C698A, H653A, and D587N mutants showed much less cleavage of the toxin, and the observed bands are consistent with normal breakdown of the purified toxin (Fig. 3, A and B). We also tested the glucosyltransferase activity of the recombinant proteins (Fig. 3C). All of the recombinant toxin proteins, including wild type and mutants, glucosylated Rac protein in vitro. Moreover, we tested the cytotoxicity of the recombinant toxins in cell culture. Although high concentrations were necessary, the recombinant wild-type toxin caused rounding up of cells like the native toxin B, which was purified from C. difficile (Fig. 3E). By contrast, the recombinant toxin mutants D587N, H653A, and C698A were without any cytotoxic effects. Accordingly, only native toxin B and recombinant wild-type toxin B caused glucosylation of Rac in intact cells, which was studied by means of an anti-Rac antibody (19), which does not recognize glucosylated Rac protein (Fig. 3D).
To gain more insights into the role of the putative cysteine protease domain of toxin B, we generated mutants of fragments of toxin B, covering residues 1-955 (Fig. 1) and expressed the protein by in vitro transcription and translation. We observed that a wild-type fragment of toxin B was always recovered as a cleaved product (with fragments of 63 and 47 kDa, according to a specific cleavage at position 543/544 (17)), whereas the mutant with change in His 653 or Cys 698 were expressed in full length (Fig. 4). Also the change of Asp 587 to alanine largely increased the stability of the toxin fragment, although its stability was less than that of the other mutants (not shown). Therefore, we suggested that in wild-type fragment 1-955 of toxin B, processing within the reticulocyte lysate occurred immediately after production of the protein by auto-cleavage, whereas the mutant fragments were not cleaved, because the inherent protease activity was blocked. To corroborate this hypothesis two controls were made. First, we changed Cys 595 in toxin B, which is not strictly conserved in all clostridial glucosylating toxins and also not present in RTX toxin. Like the wild-type fragment, this mutant was cleaved during in vitro expression. Second, we changed the putative autocatalytic cleavage site and studied the expression of the double mutant L543A/ G544A in fragment 1-955 of toxin B. As shown in Fig. 4, this mutant was fully expressed in vitro.
Recently it was reported that InsP 6 is an essential cytosolic factor (18) that facilitates cleavages of clostridial glucosylating toxins. We tested the effect of InsP 6 and confirmed the increase in proteolytic processing of toxin B by this factor. The efficiency of InsP 6 was much higher with toxin B than with toxin A. Cleavage facilitated by InsP 6 was much more rapid than with DTT. However, DTT and InsP 6 displayed synergistic effects (Fig. 5, A and B). Whereas after short term incubation (30 min) in the presence of 1 M InsP 6 only minor cleavage of toxin B was observed, DTT (2 mM) largely increased cleavage at corresponding InsP 6 concentrations (Fig. 5B). When we studied the processing of toxin B in the presence of InsP 6 and DTT, we observed a slightly different migration behavior of the corresponding cleavage products. We asked whether the cleavage caused by InsP 6 may be independent of the effects induced by DTT. To clarify this, we used N-ethylmaleimide. As shown in Fig. 5B and C, co-treatment of toxin B with N-ethylmaleimide blocked the processing induced by InsP 6 . This finding indicates that InsP 6 -dependent processing of toxin B depends on cysteine residues. HeLa cells were intoxicated with 5 ng/ml native toxin B or 1 g/ml recombinant wild type and mutants of toxin B for 6 h at 37°C. The cells were lysed, and 20 g of the cytoplasmic extract were separated by SDS-PAGE and transferred to immunoblotting with either a Rac1-specific antibody that recognizes only nonglucosylated Rac1 or, as an input control, Rac1 antibody recognizing both glucosylated and nonglucosylated protein (total Rac1). E, cell intoxication by native and recombinant wild-type and mutant toxin B. Swiss 3T3 cells were treated with 5 ng/ml native toxin B (ToxB nat) or 1 g/ml recombinant wild-type (ToxB rec wt) and mutant toxin B proteins (ToxB rec D587N, ToxB rec H653A, and ToxB rec C698A), and cell rounding was monitored by light microscopy. The pictures were taken after 4 h. All of the toxin B mutants displayed no intoxication within this time frame.

DISCUSSION
Recently it was shown that not the holotoxin of clostridial glucosylating cytotoxins but only the catalytic domain is translocated into the cytosol of target cells, indicating a processing of the toxins during toxin uptake (16,17). More recently the autocatalytic cleavage of toxin B has been proposed on the basis of a putative inherent aspartate protease activity, which is increased in the presence of InsP 6 (18). We observed that DTT induced cleavage of toxin A and B. MALDI-TOF mass spectrometric analysis (see supplemental material) revealed that cleavage occurred at the N-terminal region of the toxins where the toxin is suggested to be processed (16,17). Toxin cleavage was inhibited by N-ethylmaleimide and iodoacetamide, indicating the essential role of cysteine residues.
These findings are in line with a recent report showing that clostridial glucosylating toxins share sequence similarity with the auto-catalytic cysteine protease domain of RTX toxin from V. cholerae (23). According to this report, clostridial glucosylating toxins belong to a new family of cysteine proteases. Therefore, we suggest that the active center of the cysteine protease domain is blocked by a disulfide bridge. DTT causes activation of the cysteine protease activity. Accordingly, also ␤-mercaptoethanol facilitated the cleavage, and N-ethylmaleimide or iodoacetamide blocked the processing of the toxins. For RTX toxin from V. cholerae, a catalytic dyad has been suggested (23). Cysteine and histidine residues have been identified in RTX toxin, which correspond with cysteine 698 and histidine 653 in toxin B. Interestingly, it has been shown earlier that these residues are important for the activity of toxin B (24). In addition, we assume that aspartate 587 of toxin B is the third residue of a catalytic triad. We expressed wild-type toxin B and mutants of the putative protease catalytic triad in E. coli and observed cleavage of the wild-type toxin and significantly reduced cleavage of the mutants. Although the mutant toxin B proteins were able to glucosylate Rac in vitro, indicating intact expression of the glucosyltransferase domain, they were not able to intoxicate cells. These observations are in line with the hypothesis that a cysteine protease catalytic triad is involved in processing of the toxin and auto-cleavage is essential for toxin activity.  A, native toxin B (1 g) was incubated with rising concentrations of InsP 6 , as indicated, and/or 2 mM DTT (lower panel) for 30 min at room temperature. The probes were separated by SDS-PAGE, and toxin B was visualized by immunoblot with an antibody recognizing the catalytic domain. The corresponding densitometric analysis of signal intensities from three independent experiments is shown in Fig. 5B. The densitometric data (ϮS.E.) are given as percentages of maximal toxin cleavage observed at a concentration of 100 M InsP 6 . C, native toxin B (1 g) was incubated with 10 mM InsP 6 and at increasing concentrations of N-ethylmaleimide (NEM, as indicated) for 30 min at room temperature. The samples were transferred to immunoblot, and toxin B was detected with a monoclonal antibody directed against the catalytic domain. High molecular mass band corresponds to full-length toxin (f.l.), and the low molecular mass band corresponds to the N-terminal glucosyltransferase domain (N-term.).
How can these results fit into the model of the recently proposed aspartate protease-dependent auto-cleavage (18)? The putative aspartate protease domain was identified by 1,2-epoxy-3-(p-nitrophenoxypropane), which is an aspartate protease inhibitor known to bind covalently to the active site. Thus, the hypothesis of the aspartate protease is mainly based on the presence of a short DXG motif observed in many aspartate proteases. Although Reineke et al. (18) have shown that aspartate 1665 is a residue important for toxin activity, their data do not necessarily suggest that aspartate 1665 is the catalytic residue for processing of toxins downstream of the glucosyltransferase domain. Indeed, this residue is located Ͼ1000 amino acids away from the processing site and to the C-terminal side of the translocation domain (Fig. 1), suggesting that the DXG motif is not translocated but remains within the endosome. By contrast, the cysteine protease domain identified in this work is located adjacent to the processing site to the N-terminal side of the translocation region, indicating that it would be transferred to the cytosol along with the glycosyltransferase domain and thereby accesses intracellular InsP 6 . Consistent with this hypothesis, our recombinant protein, comprising only the portion of toxin B to the N-terminal side of the translocation domain (amino acids 1-955), is processed after in vitro translation to the 63-kDa glycosyltransferase domain dependent upon histidine 653, cysteine 698, and aspartate 587. Moreover, we show that the change of the amino acids leucine 543/glycine 544 at the proposed cleavage site to alanine prevents cleavage of the toxin fragment. These findings show that leucine 543 is the relevant cleavage site in the toxin fragment and strongly indicate that contamination by nonspecific proteases is not responsible for the observed processing.
Further evidence against the aspartic protease model is that the DXG motif is not present in the ␣-toxin of Clostridium novyi, even though ␣-toxin undergoes auto-cleavage by the addition of InsP 6 similar to other clostridial toxins (18). By contrast, ␣-toxin of C. novyi shares identical amino acid residues in the putative cysteine protease domain, indicating a common mechanism for cleavage of all large clostridial toxins. Therefore, we would like to offer the following explanations for these discrepancies. First, identification of the aspartate protease activity may not be correct, and the aspartate protease inhibitor, which is not totally specific, has additional effects on the toxin that disrupt function of the cysteine protease. Alternatively, protease activity of the toxins could depend on two catalytic domains. It is well known that some proteases have more than one catalytic domain in the same translation product. Examples are the angiotensin I-converting enzyme (25) and carboxypeptidase D (26), which are metalloproteases. Other proteases form a catalytic active site by dimerization. Recently a cysteine protease from hepatitis C virus NS2-3 has been described, which has two protease subdomains, and the proteolytically active center is formed by dimerization of the polypeptides (27). Interestingly, this cysteine protease from hepatitis C virus exhibits sequence similarity with toxin B and RTX cysteine proteases (23). Further examples are tandem proteases like polyserases 1-3, which are serine proteases with several distinct protease domains. However, in all these cases, the same family type of proteases is combined (e.g. a cysteine protease domain dimerizes in the case of NS2-3 and serine proteases form polyserases). To our knowledge a combination of different types of proteases domains in one molecule has not been reported.
We confirmed that InsP 6 is a potent activator of toxin B cleavage (18). What is the function of InsP 6 ? InsP 6 is a highly charged molecule and appears to have diverse functions. It might be involved in stabilization of a toxin protein conformation, which is essential for protease activity and/or proper cleavage. Many structural effects of InsP 6 have been reported. Recently InsP 6 was shown to be essential for assembly for human immunodeficiency virus, type 1 Gag molecules (28). We show that the effect of InsP 6 is synergistically enhanced by DTT. Even more important, we found that the effect of InsP 6 is blocked by N-ethylmaleimide treatment, showing that a cysteine residue is absolutely essential for auto-cleavage activity. This is again in line with the hypothesis that the cysteine protease is essential for cleavage.