Characterization of membrane-associated Clostridium perfringens enterotoxin following pronase treatment

Cryo-EM structures of Clostridium perfringens enterotoxin bound to its human receptor, claudin-4

Clostridium perfringens enterotoxin (CpE) causes prevalent and deadly gastrointestinal disorders. CpE binds to receptors called claudins on the apical surfaces of small intestinal epithelium. Claudins normally regulate paracellular transport but are hijacked from doing so by CpE and are instead led to form claudin/CpE complexes. Claudin/CpE complexes are the building blocks of oligomeric β-barrel pores that penetrate the plasma membrane and induce gut cytotoxicity. Here, we present the structures of CpE in complex with its native claudin receptor in humans, claudin-4, using cryogenic electron microscopy. The structures reveal the architecture of the claudin/CpE complex, the residues used in binding, the orientation of CpE relative to the membrane, and CpE-induced changes to claudin-4. Further, structures and modeling allude to the biophysical procession from claudin/CpE complexes to cytotoxic β-barrel pores during pathogenesis. In full, this work proposes a model of claudin/CpE assembly and provides strategies to obstruct its formation to treat CpE diseases.

Effects of Claudin-1 on the Action of Clostridium perfringens Enterotoxin in Caco-2 Cells

Clostridium perfringens enterotoxin (CPE) contributes to diarrhea and an often-lethal enterotoxemia. CPE action starts when it binds to claudin receptors, forming a small complex (90 kDa). Six small complexes then oligomerize to create prepores, followed by insertion of beta-hairpins from CPE to form beta-barrel pores named CH-1 or CH-2. Of the ~27 members of the human claudin protein family, only some bind CPE. However, both receptor claudins and the nonreceptor claudin-1 (CLDN-1) are associated with the small and CH-1/CH-2 CPE complexes. Therefore, this study evaluated whether claudin-1 affects CPE action by generating a CLDN-1 null mutant in Caco-2 cells using CRISPR-Cas9. Compared to wild-type Caco-2 cells, paracellular permeability of the CLDN-1 mutant was significantly enhanced, suggesting that claudin-1 may reduce CPE absorption during enterotoxemia. The CLDN-1 mutant was also markedly more sensitive than wild-type Caco-2 cells to apically-applied CPE. The mechanism behind this increased sensitivity involved higher CPE binding by the CLDN-1 mutant vs. wild-type Caco-2 cells, which led to more CH-1/CH-2 complex formation. However, the CH-1/CH-2 complexes formed by the CLDN-1 mutant were less stable or trypsin resistant than those of wild-type cells. These results indicate that, although a nonreceptor, CLDN-1 positively and negatively influences CPE action.

Identification and characterization of Clostridium perfringens beta toxin variants with differing trypsin sensitivity and in vitro cytotoxicity activity

By producing toxins, Clostridium perfringens causes devastating diseases of both humans and animals. C. perfringens beta toxin (CPB) is the major virulence determinant for type C infections and is also implicated in type B infections, but little is known about the CPB structure-function relationship. Amino acid sequence comparisons of the CPBs made by 8 randomly selected isolates identified two natural variant toxins with four conserved amino acid changes, including a switch of E to K at position 168 (E168K) that introduces a potential trypsin cleavage site into the CPB protein of strain JGS1076. To investigate whether this potential trypsin cleavage site affects sensitivity to trypsin, a primary host defense against this toxin, the two CPB variants were assayed for their trypsin sensitivity. The results demonstrated a significant difference in trypsin sensitivity, which was linked to the E168K switch by using site-directed recombinant CPB (rCPB) mutants. The natural CPB variants also displayed significant differences in their cytotoxicity to human endothelial cells. This cytotoxicity difference was mainly attributable to increased host cell binding rather than the ability to oligomerize or form functional pores. Using rCPB site-directed mutants, differences in cytotoxicity and host cell binding were linked to an A300V amino acid substitution in the strain JGS1076 CPB variant that possessed more cytotoxic activity. Mapping of sequence variations on a CPB structure modeled using related toxins suggests that the E168K substitution is surface localized and so can interact with trypsin and that the A300V substitution is located in a putative binding domain of the CPB toxin.

Human claudin-8 and -14 are receptors capable of conveying the cytotoxic effects of Clostridium perfringens enterotoxin

Clostridium perfringens enterotoxin (CPE) contributes to several important human gastrointestinal (GI) diseases. This toxin and its derivatives are also being explored for translational applications, i.e., cancer therapy or drug delivery. Some, but not all, members of the 24-member claudin (Cldn) family of mammalian tight junction proteins can serve as CPE receptors. Among the human Cldns (hCldns), hCldn-3 and -4 are known to convey CPE sensitivity when expressed by fibroblast transfectants. However, other Cldns are also reportedly expressed in the intestines, where they might contribute to natural CPE-mediated GI disease, and in other organs, where they might react with CPE-based therapeutics. Therefore, the current study assessed whether two additional hCldns beside hCldn-3 and -4 are also functional CPE receptors. Using Cldn-expressing transfectants, hCldn-8 and -14 were shown to convey CPE-mediated cytotoxicity at pathophysiologically relevant concentrations of this toxin, although ~2-to-10-fold less efficiently than hCldn-4. Site-directed mutagenesis then demonstrated that the N¹⁴⁶ residue in hCldn-14 and the S¹⁵¹ residue in hCldn-8 are largely responsible for modulating the weaker CPE binding properties of hCldn-8 and -14 versus hCldn-4, which broadens understanding of Cldn:CPE binding interactions. Since Cldn-8 and -14 are reportedly expressed in mammalian intestines, the current results support the possibility that these two hCldns contribute to natural CPE-mediated gastrointestinal disease and could be CPE-based therapeutic targets for cancers overexpressing those claudins. However, these results also suggest caution during therapeutic use of CPE, which might trigger toxic side effects in normal human tissues producing hCldn-8 or -14, as well as in those producing hCldn-3 or -4.

IMPORTANCE Clostridium perfringens enterotoxin (CPE) is responsible for the gastrointestinal symptoms of the second-most-common bacterial food-borne illness and is also being explored for use as a cancer therapeutic or for increasing drug delivery. Until now, the only known human CPE receptors were claudin-3 and -4. This work shows that human claudin-8 and -14 can also bind CPE and convey cytotoxicity, although slightly less efficiently than claudin-3 and -4. The claudin-8 and -14 residues responsible for this weaker CPE binding were identified, shedding new light on CPE:claudin interactions.

Clostridium perfringens enterotoxin (CPE) is responsible for the gastrointestinal symptoms of the second-most-common bacterial food-borne illness and is also being explored for use as a cancer therapeutic or for increasing drug delivery. Until now, the only known human CPE receptors were claudin-3 and -4. This work shows that human claudin-8 and -14 can also bind CPE and convey cytotoxicity, although slightly less efficiently than claudin-3 and -4. The claudin-8 and -14 residues responsible for this weaker CPE binding were identified, shedding new light on CPE:claudin interactions.

Cysteine-scanning mutagenesis supports the importance of Clostridium perfringens enterotoxin amino acids 80 to 106 for membrane insertion and pore formation

Clostridium perfringens enterotoxin (CPE) causes the gastrointestinal symptoms of the second most common bacterial food-borne illness. Previous studies suggested that a region named TM1, which has amphipathic characteristics and spans from amino acids 81 to 106 of the native CPE protein, forms a β-hairpin involved in β-barrel pore formation. To further explore the potential role of TM1 in pore formation, the single Cys naturally present in CPE at residue 186 was first altered to alanine by mutagenesis; the resultant rCPE variant, named C186A, was shown to retain cytotoxic properties. Cys-scanning mutagenesis was then performed in which individual Cys mutations were introduced into each TM1 residue of the C186A variant. When those Cys variants were characterized, three variants were identified that exhibit reduced cytotoxicity despite possessing binding and oligomerization abilities similar to those of the C186A variant from which they were derived. Pronase challenge experiments suggested that the reduced cytotoxicity of those two Cys variants, i.e., the F91C and F95C variants, which model to the tip of the β-hairpin, was attributable to a lessened ability of these variants to insert into membranes after oligomerization. In contrast, another Cys variant, i.e., the G103C variant, with impaired cytotoxicity apparently inserted into membranes after oligomerization but could not form a pore with a fully functional channel. Collectively, these results support the TM1 region forming a β-hairpin as an important step in CPE insertion and pore formation. Furthermore, this work identifies the first amino acid residues specifically involved in those two steps in CPE action.

Cell type-dependent internalization of the Escherichia coli STb enterotoxin

Previous studies have suggested that internalization of the Escherichia coli STb enterotoxin in human and rat intestinal epithelial cells is involved in STb pathogenesis, but toxin uptake in porcine jejunum epithelium, the in vivo target tissue, still remains elusive. Using flow cytometry, we studied the internalization of fluorescein isothiocyanate-labelled STb in porcine intestinal epithelial IPEC-J2 and murine fibroblast NIH-3T3 cell lines. In contrast to the selective pronase resistance of STb in NIH-3T3 cells at 37 °C, but not at 4 °C, indicative of toxin internalization, most of the toxin was pronase-sensitive at both temperatures in IPEC-J2 cells, indicating reduced uptake, but significant cell surface binding. Actin reorganization is required for STb internalization by NIH-3T3 cells, confirming STb endocytosis in these cells. The toxin receptor, sulfatide, could not explain these internalization differences because both cell lines possessed surface sulfatide and internalized antisulfatide antibodies over time at 37 °C. Inhibition of lipid rafts endocytosis, known to contain sulfatide, with methyl-β-cyclodextrin or genistein, did not influence toxin uptake by either cell line. STb internalization is therefore differentially regulated depending on the cell type, possibly by factors other than sulfatide. Although a small STb fraction could be internalized by porcine intestinal epithelial cells, our findings suggest the ability of STb to induce, from the cell surface, intracellular signalling leading to fluid secretion in porcine intestinal epithelium.

On the interaction of Clostridium perfringens enterotoxin with claudins

Clostridium perfringens causes one of the most common foodborne illnesses, which is largely mediated by the Clostridium perfringens enterotoxin (CPE). The toxin consists of two functional domains. The N-terminal region mediates the cytotoxic effect through pore formation in the plasma membrane of the mammalian host cell. The C-terminal region (cCPE) binds to the second extracellular loop of a subset of claudins. Claudin-3 and claudin-4 have been shown to be receptors for CPE with very high affinity. The toxin binds with weak affinity to claudin-1 and -2 but contribution of these weak binding claudins to CPE-mediated disease is questionable. cCPE is not cytotoxic, however, it is a potent modulator of tight junctions. This review describes recent progress in the molecular characterization of the cCPE-claudin interaction using mutagenesis, in vitro binding assays and permeation studies. The results promote the development of recombinant cCPE-proteins and CPE-based peptidomimetics to modulate tight junctions for improved drug delivery or to treat tumors overexpressing claudins.

Clostridial toxins

Clostridia produce the highest number of toxins of any type of bacteria and are involved in severe diseases in humans and other animals. Most of the clostridial toxins are pore-forming toxins responsible for gangrenes and gastrointestinal diseases. Among them, perfringolysin has been extensively studied and it is the paradigm of the cholesterol-dependent cytolysins, whereas Clostridium perfringens epsilon-toxin and Clostridium septicum alpha-toxin, which are related to aerolysin, are the prototypes of clostridial toxins that form small pores. Other toxins active on the cell surface possess an enzymatic activity, such as phospholipase C and collagenase, and are involved in the degradation of specific cell-membrane or extracellular-matrix components. Three groups of clostridial toxins have the ability to enter cells: large clostridial glucosylating toxins, binary toxins and neurotoxins. The binary and large clostridial glucosylating toxins alter the actin cytoskeleton by enzymatically modifying the actin monomers and the regulatory proteins from the Rho family, respectively. Clostridial neurotoxins proteolyse key components of neuroexocytosis. Botulinum neurotoxins inhibit neurotransmission at neuromuscular junctions, whereas tetanus toxin targets the inhibitory interneurons of the CNS. The high potency of clostridial toxins results from their specific targets, which have an essential cellular function, and from the type of modification that they induce. In addition, clostridial toxins are useful pharmacological and biological tools.

Identification of a prepore large-complex stage in the mechanism of action of Clostridium perfringens enterotoxin

Clostridium perfringens enterotoxin (CPE) is the etiological agent of the third most common food-borne illness in the United States. The enteropathogenic effects of CPE result from formation of large CPE-containing complexes in eukaryotic cell membranes. Formation of these approximately 155- and approximately 200-kDa complexes coincides with plasma membrane permeability changes in eukaryotic cells, causing a Ca2+ influx that drives cell death pathways. CPE contains a stretch of amino acids (residues 81 to 106) that alternates markedly in side chain polarity (a pattern shared by the transmembrane domains of the beta-barrel pore-forming toxin family). The goal of this study, therefore, was to investigate whether this CPE region is involved in pore formation. Complete deletion of the CPE region from 81 to 106 produced a CPE variant that was noncytotoxic for Caco-2 cells and was unable to form CPE pores. However, this variant maintained the ability to form the approximately 155-kDa large complex. This large complex appears to be a prepore present on the plasma membrane surface since it showed greater susceptibility to proteases, increased complex instability, and a higher degree of dissociation from membranes compared to the large complex formed by recombinant CPE. When a D48A mutation was engineered into this prepore-forming CPE variant, the resultant variant was unable to form any prepore approximately 155-kDa large complex. Collectively these findings reveal a new step in CPE action, whereby receptor binding is followed by formation of a prepore large complex, which then inserts into membranes to form a pore.

The Plasmodium circumsporozoite protein is proteolytically processed during cell invasion

The circumsporozoite protein (CSP) is the major surface protein of Plasmodium sporozoites, the infective stage of malaria. Although CSP has been extensively studied as a malaria vaccine candidate, little is known about its structure. Here, we show that CSP is proteolytically cleaved by a papain family cysteine protease of parasite origin. Our data suggest that the highly conserved region I, found just before the repeat region, contains the cleavage site. Cleavage occurs on the sporozoite surface when parasites contact target cells. Inhibitors of CSP processing inhibit cell invasion in vitro, and treatment of mice with E-64, a highly specific cysteine protease inhibitor, completely inhibits sporozoite infectivity in vivo.

The enteric toxins of Clostridium perfringens

The Gram-positive pathogen Clostridium perfringens is a major cause of human and veterinary enteric disease largely because this bacterium can produce several toxins when present inside the gastrointestinal tract. The enteric toxins of C. perfringens share two common features: (1) they are all single polypeptides of modest (approximately 25-35 kDa) size, although lacking in sequence homology, and (2) they generally act by forming pores or channels in plasma membranes of host cells. These enteric toxins include C. perfringens enterotoxin (CPE), which is responsible for the symptoms of a common human food poisoning and acts by forming pores after interacting with intestinal tight junction proteins. Two other C. perfringens enteric toxins, epsilon-toxin (a bioterrorism select agent) and beta-toxin, cause veterinary enterotoxemias when absorbed from the intestines; beta- and epsilon-toxins then apparently act by forming oligomeric pores in intestinal or extra-intestinal target tissues. The action of a newly discovered C. perfringens enteric toxin, beta2 toxin, has not yet been defined but precedent suggests it might also be a pore-former. Experience with other clostridial toxins certainly warrants continued research on these C. perfringens enteric toxins to develop their potential as therapeutic agents and tools for cellular biology.

Bacterial-associated diarrhea in the dog: a critical appraisal

The clinical documentation of enteropathogenic bacteria causing diarrhea in dogs is clouded by the presence of many of these organisms existing as normal constituents of the indigenous intestinal flora. The diagnosis of a putative bacterial enteropathogen(s) in dogs should be made based on a combination of parameters, including signalment and predisposing factors, clinical signs, serologic assays for toxins, fecal culture, and PCR. Relying on results of fecal culture alone is problematic, because C perfringens, C difficile, Campylobacter spp, and pathogenic and non-pathogenic E coli are commonly isolated from apparently healthy dogs [10,13,33]. Nevertheless, culture may be useful in procuring isolates for the application of molecular techniques, such as PCR, for detection of specific toxin genes or molecular typing of isolated strains to establish clonality in suspected outbreaks. The oversimplistic attempt to characterize bacterially associated diarrhea by anatomic localization of clinical signs should be discouraged, because most of the previously mentioned bacteria have been associated with small and large intestinal diarrhea. Accurate diagnosis of infections may require diagnostic laboratories to incorporate PCR-based assays using genus- and species-specific primers to facilitate detection of toxin genes and differentiation of species that appear phenotypically and biochemically similar. There has been tremendous interest in the application of microarray technology for the simultaneous detection of thousands of genes or target DNA sequences on one glass slide. This powerful tool could be used for detection of specific pathogenic bacterial strains in fecal specimens obtained from dogs in the future.

The role of bacterial and non-bacterial toxins in the induction of changes in membrane transport: implications for diarrhea

Bacterial toxins induce changes in membrane transport which underlie the loss of electrolyte homeostasis associated with diarrhea. Bacterial- and their secreted toxin-types which have been linked with diarrhea include: (a) Vibrio cholerae (cholera toxin, E1 Tor hemolysin and accessory cholera enterotoxin); (b) Escherichia coli (heat stable enterotoxin, heat-labile enterotoxin and colicins); (c) Shigella dysenteriae (shiga-toxin); (d) Clostridium perfringens (C. perfringens enterotoxin, alpha-toxin, beta-toxin and theta-toxin); (e) Clostridium difficile (toxins A and B); (f) Staphylococcus aureus (alpha-haemolysin); (g) Bacillus cereus (cytotoxin K and haemolysin BL); and (h) Aeromonas hydrophila (aerolysin, heat labile cytotoxins and heat stable cytotoxins). The mechanisms of toxin-induced diarrhea include: (a) direct effects on ion transport in intestinal epithelial cells, i.e. direct toxin interaction with intrinsic ion channels in the membrane and (b) indirect interaction with ion transport in intestinal epithelial cells mediated by toxin binding to a membrane receptor. These effects consequently cause the release of second messengers, e.g. the release of adenosine 3',5'-cyclic monophosphate/guanosine 3',5'-monophosphate, IP(3), Ca2+ and/or changes in second messengers that are the result of toxin-formed Ca2+ and K+ permeable channels, which increase Ca2+ flux and augment changes in Ca2+ homeostasis and cause depolarisation of the membrane potential. Consequently, many voltage-dependent ion transport systems, e.g. voltage-dependent Ca2+ influx, are affected. The toxin-formed ion channels may act as a pathway for loss of fluid and electrolytes. Although most of the diarrhea-causing toxins have been reported to act via cation and anion channel formation, the properties of these channels have not been well studied, and the available biophysical properties that are needed for the characterization of these channels are inadequate.

The complex interactions between Clostridium perfringens enterotoxin and epithelial tight junctions

Clostridium perfringens enterotoxin (CPE) is responsible for the diarrheal symptoms of C. perfringens type A food poisoning and antibiotic-associated diarrhea. The CPE protein consists of a single 35 kDa polypeptide with a C-terminal receptor-binding region and an N-terminal toxicity domain. Under appropriate conditions, CPE can interact with structural components of the epithelial tight junctions, including certain claudins and occludin. Those interactions can affect tight junction structure and function, thereby altering paracellular permeability and (possibly) contributing to CPE-induced diarrhea. However, the tight junction effects of CPE require cellular damage as a prerequisite. CPE induces cellular damage via its cytotoxic activity, which results from plasma membrane permeability alterations caused by formation of a approximately 155 kDa CPE-containing complex that may correspond to a pore. Thus, CPE appears to be a bifunctional toxin that first induces plasma membrane permeability alterations; using the resultant cell damage, CPE then gains access to tight junction proteins and affects tight junction structure and function.

Comparative biochemical and immunocytochemical studies reveal differences in the effects of Clostridium perfringens enterotoxin on polarized CaCo-2 cells versus Vero cells

Since most in vitro studies exploring the action of Clostridium perfringens enterotoxin (CPE) utilize either Vero or CaCo-2 cells, the current study directly compared the CPE responsiveness of those two cell lines. When CPE-treated in suspension, both CaCo-2 and Vero cells formed SDS-resistant, CPE-containing complexes of approximately 135, approximately 155, and approximately 200 kDa. However, confluent Transwell cultures of either cell line CPE-treated for 20 min formed only the approximately 155-kDa complex. Since those Transwell cultures also exhibited significant (86)Rb release, approximately 155-kDa complex formation is sufficient for CPE-induced cytotoxicity. Several differences in CPE responsiveness between the two cell lines were also detected. (i) CaCo-2 cells were more sensitive when CPE-treated on their basal surface, whereas Vero cells were more sensitive when CPE-treated on their apical surface; those sensitivity differences correlated with CPE binding the apical versus basolateral surfaces of these two cell lines. (ii) CPE-treated Vero cells released (86)Rb into both Transwell chambers, whereas CaCo-2 cells released (86)Rb only into the CPE-containing Transwell chamber. (iii) Vero cells express the tight junction (TJ) protein occludin but (unlike CaCo-2 cells) cannot form TJs. The ability of TJs to affect CPE responsiveness is supported by the similar effects of CPE on Transwell cultures of CaCo-2 cells and Madin-Darby canine kidney cells, another polarized cell forming TJs. Confluent CaCo-2 Transwell cultures CPE-treated for >1 h formed the approximately 200-kDa CPE complex (which also contains occludin), exhibited morphologic damage, and had occludin removed from their TJs. Collectively, these results identify CPE as a bifunctional toxin that, in confluent polarized cells, first exerts a cytotoxic effect mediated by the approximately 155-kDa complex. Resultant damage then provides CPE access to TJs, leading to approximately 200-kDa complex formation, internalization of some TJ proteins, and TJ damage that may increase paracellular permeability and thereby contribute to the diarrhea of CPE-induced gastrointestinal disease.

CaCo-2 cells treated with Clostridium perfringens enterotoxin form multiple large complex species, one of which contains the tight junction protein occludin

The previous model for the action of Clostridium perfringens enterotoxin (CPE) proposed that (i) CPE binds to host cell receptor(s), forming a small ( approximately 90 kDa) complex, (ii) the small complex interacts with other eucaryotic protein(s), forming a large ( approximately 160 kDa) complex, and (iii) the large complex triggers massive permeability changes, thereby inducing enterocyte death. In the current study, Western immunoblot analysis demonstrated that CPE bound to CaCo-2 human intestinal cells at 37 degrees C forms multiple large complex species, with apparent sizes of approximately 200, approximately 155, and approximately 135 kDa. These immunoblot experiments also revealed that occludin, an approximately 65-kDa tight junction protein, is present in the approximately 200-kDa large complex but absent from the other large complex species. Immunoprecipitation studies confirmed that occludin physically associates with CPE in large complex material and also indicated that occludin is absent from small complex. These results strongly suggest that occludin becomes associated with CPE during formation of the approximately 200-kDa large complex. A postbinding association between CPE and occludin is consistent with the failure of rat fibroblast transfectants expressing occludin to bind CPE in the current study. Those occludin transfectants were also insensitive to CPE, strongly suggesting that occludin expression is not sufficient to confer CPE sensitivity. However, the occludin-containing, approximately 200-kDa large complex may contribute to CPE-induced cytotoxicity, because nontoxic CPE point mutants did not form any large complex species. By showing that large complex material is comprised of several species (one containing occludin), the current studies indicate that CPE action is more complicated than previously appreciated and also provide additional evidence for CPE interactions with tight junction proteins, which could be important for CPE-induced pathophysiology.

Identification of a Clostridium perfringens enterotoxin region required for large complex formation and cytotoxicity by random mutagenesis

Clostridium perfringens enterotoxin (CPE), a single polypeptide of 319 amino acids, has a unique multistep mechanism of action. In the first step, CPE binds to claudin proteins and/or a 50-kDa eukaryotic membrane protein receptor, forming a small ( approximately 90-kDa) complex. This small complex apparently then associates with a 70-kDa eukaryotic membrane protein, resulting in formation of a large complex that induces the onset of membrane permeability alterations. To better define the boundaries of CPE functional regions and to identify specific amino acid residues involved in various steps of CPE action, in this study we subjected the cloned cpe gene to random mutagenesis in XL-1 Red strains of Escherichia coli. Seven CPE random mutants with reduced cytotoxicity for Vero cells were phenotypically characterized for the ability to complete each step in CPE action. Five of these seven recombinant CPE (rCPE) random mutants (G49D, S59L, R116S, R137G, and S167P) exhibited binding characteristics similar to those of rCPE or native CPE, while the Y310C and W226Stop mutants showed reduced binding and no binding, respectively, to brush border membranes. Interestingly, two completely nontoxic mutants (G49D and S59L) were able to bind and form small complex but they did not mediate any detectable large complex formation. Another strongly attenuated mutant, R116S, formed reduced amounts of an anomalously migrating large complex. Collectively, these results provide further support for large complex formation being an essential step in CPE action and also identify the CPE region ranging from residues approximately 45 to 116 as important for large complex formation. Finally, we also report that limited removal of extreme N-terminal CPE sequences, which may occur in vivo during disease, stimulates cytotoxic activity by enhancing large complex formation.