Amino acid replacements in the Serratia marcescens haemolysin ShlA define sites involved in activation and secretion

The haemolysin of Serratia marcescens (ShlA) is translocated through the cytoplasmic membrane by the signal peptide-dependent export apparatus. Translocation across the outer membrane (secretion) is mediated by the ShIB protein. Only the secreted form of ShlA is haemolytic. ShIB also converts in vitro inactive ShlA (ShlA*), synthesized in the absence of ShIB, into the haemolytic form (a process termed activation). To define regions in ShlA involved in both processes, ShlA derivatives were isolated and tested for secretion and activation. Analysis of C-terminally truncated proteins (ShlA) assigned the secretion signal to the amino-terminal 238 residues of ShlA. Trypsin cleavage of a secreted ShlA' derivative yielded a 15kDa N-terminal fragment, by which a haemolytically inactive ShlA* protein could be activated in vitro. It is suggested that the haemolysin activation site is located in this N-terminal fragment. Replacement of asparagine-69 and asparagine-109 by isoleucine yielded inactive haemolysin derivatives. Both asparagine residues are part of two short sequence motifs, reading Ala-Asn-Pro-Asn, which are critical to both activation and secretion. These point mutants as well as N-terminal deletion derivatives which were not activated by ShIB were activated by adding a non-haemolytic N-terminal fragment synthesized in an ShIB⁺ strain (complementation). Apparently the activated N-terminal fragment substituted for the missing activation of the ShlA derivatives and directed them into the erythrocyte membrane, where they formed pores. It is concluded that activation is only required for initiation of pore formation, and that in vivo activation and secretion are tightly coupled processes. Complementation may also indicate that haemolysin oligomers form the pores.     

Specific phosphatidylethanolamine dependence of Serratia marcescens cytotoxin activity

The cytolytic and haemolytic activity of Serratia marcescens is determined by the ShlA protein, which is secreted across the outer membrane with the aid of the ShlB protein. In the absence of ShlB, inactive ShlA* remains in the periplasm of Escherichia coli transformed with an shlA-encoding plasmid, which indicates that ShlB converts ShlA* to active ShlA. ShlA* in a periplasmic extract and partially purified ShlA* were activated in vitro by partially purified ShlB. When both proteins were highly purified, ShlA* was only activated by ShlB when phosphatidylethanolamine (PE) or phosphatidylserine was added to the assay, while phosphatidylglycerol contributed little to ShlA* activation. Lyso-PE, cardiolipin, phosphatidylcholine, phosphatidic acid, lipopolysaccharide and various detergents could not substitute for PE. Although radioactively labelled PE was so tightly associated with ShlA that it remained bound to ShlA after heating and SDS–PAGE, it was not covalently linked to ShlA as PE could be removed by thin-layer chromatography with organic solvents. The number of PE molecules associated per molecule of ShlA was 3.9 ± 2.2. Active ShlA was inactivated by treatment with phospholipase A₂, which indicated that PE is also required for ShlA activity. ShlA-255 (containing the 255 N-terminal amino acids of ShlA) reversibly complemented ShlA* to active ShlA and was inactivated by phospholipase A₂, which demonstrated that PE binds to the N-terminal portion of ShlA; this region has previously been found to be involved in ShlA secretion and activation. Electrospray mass spectroscopy of ShlA-255 determined a molar mass that corresponded to that of unmodified ShlA-255. An E. coli mutant that synthesized only minute amounts of PE did not secrete ShlA but contained residual cell-bound haemolytic activity. Since PE binds strongly to ShlA* in the absence of ShlB without converting ShlA* to haemolytic ShlA, ShlB presumably imposes a conformation on ShlA that brings PE into a position to mediate interaction of the hydrophilic haemolysin with the lipid bilayer of the eukaryotic membrane.     

Ionic-strength-dependent changes in the structure of the major protein of the human erythrocyte membrane.

The effect of ionic strength on the proteolysis by trypsin of the major membrane-penetrating protein (polypeptide 3) in the erythrocyte membrane was studied. Both the intracellular and extracellular regions of the protein are susceptible to trypsin proteolysis under hypo-osmotic conditions, whereas under iso-osmotic conditions the extracellular region of the protein is resistant to trypsin, and the intracellular region yields only two cleavage products with trypsin. Studies of the fragments obtained from polypeptide 3 by trypsin digestion under iso-osmotic conditions of 'ghosts' radioiodinated with lactoperoxidase confirmed our earlier conclusions that the polypeptide chain of polypeptide 3 traverses the membrane twice. Ionic-strength-dependent changes were also observed in the incorporation of iodine by lactoperoxidase into the individual extracellular tyrosine sites of the protein. These results show that polypeptide 3 undergoes ionic-strength-dependent changes in structure.     

Investigation of the haemolytic activity of Proteus mirabilis strains

Young broth cultures of all P. mirabilis strains tested exhibited haemolytic activity. This activity seemed to be strongly cell-associated as only a very small fraction of this activity was found in the cell-free supernatant. The haemolysin was only produced by actively growing cells. Inhibition studies with trypsin and chloramphenicol suggested that the haemolysin is of protein nature. Lecithin and serum of several species had an inhibitory effect on the haemolysin. Besides erythrocytes of various species also VERO cells were affected by the haemolysin. A correlation was found between the haemolytic activity of a strain and its virulence in an experimental mouse model.     

Utilization of membranous lipid substrates by membranous enzymes. Hydrolysis of sphingomyelin in erythrocyte ''ghosts'' and liposomes by the membranous sphingomyelinase of chicken erythrocyte ''ghosts''.

Incubation at 37 degrees C of haemolysed chicken erythrocytes ('chicken erythrocyte ghosts') resulted in hydrolysis of the membrane sphingomyelin, suggesting an activation of a latent sphingomyelinase during the haemolysis procedure. When this incubation was continued for several hours, the entire sphingomyelin of the erythrocyte 'ghosts' was hydrolysed and membranes were obtained that were devoid of sphingomyelin, but had an active sphingomyelinase. Mixing the latter membranes with human erythrocyte 'ghosts' or positively charged liposomes led to hydrolysis of the sphingomyelin in these two membranes. This suggested that, after haemolysis, the activated sphingomyelinase in the membrane of the chicken erythrocyte 'ghosts' could hydrolyse sphingomyelin in its own membrane ('intramembrane utilization') or adjacent membranes ('intermembrane utilization').     

The repeat domain of Escherichia coli haemolysin (HlyA) is responsible for its Ca2+-dependent binding to erythrocytes

Summary The haemolysin protein (HlyA) of Escherichia coli contains 11 tandemly repeated sequences consisting of 9 amino acids each between amino acids 739 and 849 of HlyA. We removed, by oligonucleotide-directed mutagenesis, different single repeats and combinations of several repeats. The resulting mutant proteins were perfectly stable in E. coli and were secreted with the same efficiency as the wild-type HlyA. HlyA proteins which had lost a single repeat only were still haemolytically active (in the presence of HlyC) but required elevated levels of Ca²⁺ for activity, as compared to the wild-type haemolysin. Removal of three or more repeats led to the complete loss of the haemolytic activity even in the presence of high Ca²⁺ concentrations. The mutant haemolysins were unable to compete with the wild-type haemolysin for binding to erythrocytes at low Ca²⁺ concentrations but could still generate ion-permeable channels in artificial lipid bilayer membranes formed of plant asolectin, even in the complete absence of Ca²⁺. These data indicate that the repeat domain of haemolysin is responsible for Ca²⁺-dependent binding of haemolysin to the erythrocyte membrane. A model for the possible functional role of Ca²⁺ in haemolysis is presented.     

Subcellular location and unique secretion of the hemolysin of Serratia marcescens

It is shown that Serratia marcescens exports a hemolysin to the cell surface and secretes it to the extracellular space. Escherichia coli containing the cloned hemolysin genes shlA and shlB exported and secreted the S. marcescens hemolysin. A nonhemolytic secretion-incompetent precursor of the hemolysin, designated ShlA*, was synthesized in a shlB deletion mutant and accumulated in the periplasmic space of E. coli. Immunogold-labeled ultrathin sections revealed ShlA* bound to the outer face of the cytoplasmic membrane and to the inner face of the outer membrane. A number of mutants carrying 3' deletions in the shlA gene secreted truncated polypeptides, the smallest of which contained only 261 of the 1578 amino acids of the mature ShlA hemolysin, showing that the information for export to the cell surface of E. coli and secretion into the culture medium is located in the NH2-terminal segment of the hemolysin. We propose a secretion pathway in which ShlA and ShlB are exported across the cytoplasmic membrane via a signal sequence-dependent mechanism. ShlB is integrated into the outer membrane. ShlA is translocated across the outer membrane with the help of ShlB. During the latter export process or at the cell surface, ShlA acquires the hemolytically active conformation and is released to the extracellular space. The hemolysin secretion pathway appears to be different from any other secretion system hitherto reported and involves only a single specific export protein.     

Cultural conditions forAeromonas hydrophila affect the production of haemolysins with differing host specificities

Five strains ofAeromonas hydrophila were studied for production of haemolysin specific for erythrocytes of various animal species using three cultural methods. All the strains produced haemolysin for all the erythrocyte species when the organisms were cultured on blood agar.Using cellophane overlay method, all the strains produced haemolysin for fish erythrocytes and variable activity to mammalian erythrocytes. Only one strain produced haemolytic activity for various though not all of the erythrocyte species when grown in brain heart infusion broth.Data suggest thatA. hydrophila produces multiple haemolysins with specificities for erythrocytes of different animals. This was confirmed for trout and horse erythrocyte targeted haemolysins, by using iso-electric focussing separation and by measuring the effect of addition of ammonium sulphate to the growth medium.     

The haemolysin-secreting ShlB protein of the outer membrane of Serratia marcescens : determination of surface-exposed residues and formation of ion-permeable pores by ShlB mutants in artificial lipid bilayer membranes

The ShlB protein in the outer membrane of Serratia marcescens is the only protein known to be involved in secretion of the ShlA protein across the outer membrane. At the same time, ShlB converts ShlA into a haemolytic and a cytolytic toxin. Surface-exposed residues of ShlB were determined by reaction of an M2 monoclonal antibody with the M2 epitope DYKDDDDK inserted at 25 sites along the entire ShlB polypeptide. The antibody bound to the M2 epitope at 17 sites in intact cells, which indicated surface exposure of the epitope, and to 23 sites in isolated outer membranes. Two insertion mutants contained no ShlB(M2) protein in the outer membrane. The ShlB derivatives activated and/or secreted ShlA. To gain insights into the secretion mechanism, we studied whether highly purified ShlB and ShlB deletion derivatives formed pores in artificial lipid bilayer membranes. Wild-type ShlB formed channels with very low single channel conductance that rarely assumed an open channel configuration. In contrast, open channels with a considerably higher single channel conductance were observed with the deletion mutants ShlB(Delta65-186), ShlB(Delta87-153), and ShlB(Delta126-200). ShlB(Delta126-200) frequently formed permanently open channels, whereas the conductance caused by ShlB(Delta65-186) and ShlB(Delta87-153) did not assume a stationary value, but fluctuated rapidly between open and closed configurations. The results demonstrate the orientation of large portions of ShlB in the outer membrane and suggest that ShlB may function as a specialized pore through which ShlA is secreted.     

Superlytic hemolysin mutants of Serratia marcescens.

Hemolysis by Serratia marcescens is caused by two proteins, ShlA and ShlB. ShlA is the hemolysin proper, and ShlB transports ShlA through the outer membrane, whereby ShlA is converted into a hemolysin. Superhemolytic ShlA derivatives that displayed 7- to 20-fold higher activities than wild-type ShlA were isolated. ShlA80 carried the single amino acid replacement of G to D at position 326 (G326D), ShlA87 carried S386N, and ShlA80III carried G326D and N236D. Superhemolysis was attributed to the greater stability of the mutant ShlA derivatives because they aggregated less than the wild-type hemolysin, which lost activity within 3 min at 20 degrees C. In contrast to the highly hemolytic wild-type ShlA at 0 degrees C, the hyperlytic hemolysins were nonhemolytic at 0 degrees C, suggesting that the hyperlytic derivatives differed from wild-type ShlA in adsorption to and insertion into the erythrocyte membrane. However, the size of the pores formed at 20 degrees C by superhemolytic hemolysins could not be distinguished from that of wild-type ShlA. In addition to the N-terminal sequence up to residue 238, previously identified to be important for activation and secretion, sites 326 and 386 contribute to hemolysin activity since they are contained in regions that participate in hemolysin inactivation through aggregation.     

Interaction of the 140/130/110 kDa rhoptry protein complex of Plasmodium falciparum with the erythrocyte membrane and liposomes

During Plasmodium falciparum merozoite invasion into human and mouse erythrocytes, a 110-kDa rhoptry protein is secreted from the organelle into the erythrocyte membrane. In the present study our interest was to examine the interaction of rhoptry proteins of P. falciparum with the erythrocyte membrane. It was observed that the complex of rhoptry proteins of 140/130/110 kDa bind directly to a trypsin sensitive site on intact mouse erythrocytes, and not human, saimiri, or other erythrocytes. However, when erythrocytes were disrupted by hypotonic lysis, rhoptry proteins of 140/130/110 kDa were found to bind to membranes and inside-out vesicles prepared from human, mouse, saimiri, rhesus, rat, and rabbit erythrocytes. A binding site on the cytoplasmic face of the erythrocyte membrane suggests that the rhoptry proteins may be translocated across the lipid bilayer during merozoite invasion. Furthermore, pretreatment of human erythrocytes with a specific peptide derived from MSA-1, the major P. falciparum merozoite surface antigen of MW 190,000-200,000, induced binding of the 140/130/110-kDa complex. The rhoptry proteins bound equally to normal human erythrocytes and erythrocytes treated with neuraminidase, trypsin, and chymotrypsin indicating the binding site was independent of glycophorin and other major surface proteins. The rhoptry protein complex also bound specifically to liposomes prepared from different types of phospholipids. Liposomes containing PE effectively block binding of the rhoptry proteins to mouse cells, suggesting that there are two binding sites on the mouse membrane for the 140/130/110-kDa complex, one protein and a second, possibly lipid in nature. The results of this study suggest that the 140/130/110 kDa protein complex may interact directly with sites in the lipid bilayer of the erythrocyte membrane.     

Serratia marcescens forms a new type of cytolysin

Most Serratia marcescens strains produce a new type of cytolysin (hemolysin) which is also found in other Serratia species. The hemolytic polypeptide ShlA (M(r) 162 101) is secreted across the outer membrane through the help of the ShlB protein which also involves conversion of an inactive precursor in an hemolytically active form. Both proteins are synthesized with signal sequences which are released during export across the cytoplasmic membrane. Mutants expressing inactive ShlB derivatives are impaired in activation and secretion suggesting a tight coupling between both processes. The region of ShlA for activation and secretion is confined to the N-terminal 16% of the polypeptide which contains the sequence NPNG which is also found in the Proteus hemolysin, the Bordetella pertussis filamentous hemagglutinin and two highly expressed outer membrane proteins of Haemophilus influenzae. Substitution of the first asparagine (N) residue by isoleucine converts the Serratia hemolysin into an inactive secretion incompetent form. It is concluded that this region is recognized by ShlB for activation and secretion of ShlA. The Serratia hemolysin forms defined pores in erythrocyte membranes.     

Organic-acid transport in resealed haemoglobin-containing human erythrocyte ''ghosts''.

The transport of organic acids across the membrane of resealed haemoglobin-containing erythrocyte 'ghosts' prepared by a dialysis technique has been studied. The present work forms part of studies directed towards the use of erythrocyte cellular carriers in enzyme-replacement therapy of inherited metabolic diseases. Oxalic acid, glycollic acid and glyoxylic acid were taken as representative of aliphatic acids of low molecular mass and benzoic and cinnamic acids as representative of unsubstituted aromatic acids. These selected acids are important in the diseases with which the present work is concerned. Comparison of influx and efflux transport characteristics showed that erythrocyte 'ghosts' retain transport properties closely similar to those of normal erythrocytes. Rapid transport was observed with all organic acids studied and there was a linear relationship between initial amount of influx and external concentration of aliphatic acid. Saturation of the transport system was not observed up to 1 mM external concentration, and the presence of plasma in the external medium had no effect on transport characteristics. Transport in intact erythrocytes and prepared erythrocyte 'ghosts' from patients with hyperoxaluria was also studied.     

A pore-forming haemolysin from the hookworm, Ancylostoma caninum

Hookworms feed on blood, but the mechanism by which they lyse ingested erythrocytes is unknown. Here we show that Ancylostoma caninum, the common dog hookworm, expresses a detergent soluble, haemolytic factor. Activity was identified in both adult and larval stages, was heat-stable and unaffected by the addition of protease inhibitors, metal ions, chelators and reducing agents. Trypsin ablated lysis indicating that the haemolysin is a protein. A closely migrating doublet of hookworm proteins with apparent molecular weights of 60-65 kDa bound to the erythrocyte membrane after lysis of cells using both unlabeled and biotinylated detergent-solubilised hookworm extracts. In addition, separation of detergent-soluble parasite extracts using strong cation-exchange chromatography, resulted in purification of 60-65 kDa proteins with trypsin-sensitive haemolytic activity. Erythrocytes lysed with particulate, buffer-insoluble worm extracts were observed using scanning electron microscopy and appeared as red cell ghosts with approximately 100 nm diameter pores formed in the cell membranes. Red blood cell ghosts remained visible indicating that lysis was likely caused by pore formation and followed by osmotic disruption of the cell.     

The relationship to knobs of the 92,000 D protein specific for knobby strains of Plasmodium falciparum

A 92,000 D protein was identified associated with the membrane of host erythrocytes infected with the FCB1 Plasmodium falciparum strain from Colombia. The same protein was identified in the knob-forming Gambian (and the Malayan Camp) strain, but was not present in all the corresponding knobless strains. In the FCB1 strain as well as in the FCR3 strain the protein is synthesized during the ring-stage period. The cleavage products of the 92,000 D protein were investigated by peptide mapping following limited proteolytic digestion with Staphylococcus aureus V8 protease. The 92,000 D protein cleavage products from both the Colombian and the Gambian strains were identical. Moreover, both the proteins were sensitive to trypsin and chymotrypsin and also to treatment with neuraminidase. Enzymatic removal of the protein from the erythrocyte membrane by trypsin or chymotrypsin did not affect parasite maturation. The merozoites thus produced were fully invasive and the morphology of the knobs was unaltered. When the erythrocyte membrane was treated with trypsin before the time of synthesis of the 92,000 D protein, it was not possible to identify the protein in membranes of later stages of infected erythrocytes, indicating that the protein cannot be inserted into the membrane cytoskeleton compartment. Knobs, however, were formed more or less normally, suggesting that it is not the accumulation of this protein which products the knobs.     

Effect of Vibrio parahaemolyticus haemolysin on human erythrocytes

Haemolysin Kanagawa, a toxin from Vibrio parahaemolyticus, is known to trigger haemolysis. Flux studies indicated that haemolysin forms a cation channel. In the present study, channel properties were elucidated by patch clamp and functional significance of ion fluxes by fluorescence-activated cell sorting (FACS) analysis. Treatment of human erythrocytes with 1 U ml-1 haemolysin within minutes induces a non-selective cation permeability. Moreover, haemolysin activates clotrimazole-sensitive K+ channels, pointing to stimulation of Ca2+-sensitive Gardos channels. Haemolysin (1 U ml-1) leads within 5 min to slight cell shrinkage, which is reversed in Ca2+-free saline. Erythrocytes treated with haemolysin (0.1 U ml-1) do not undergo significant haemolysis within the first 60 min. Replacement of extracellular Na+ with NMDG+ leads to slight cell shrinkage, which is potentiated by 0.1 U ml-1 haemolysin. According to annexin binding, treatment of erythrocytes with 0.1 U ml-1 haemolysin leads within 30 min to breakdown of phosphatidylserine asymmetry of the cell membrane, a typical feature of erythrocyte apoptosis. The annexin binding is significantly blunted at increased extracellular K+ concentrations and by K+ channel blocker clotrimazole. In conclusion, haemolysin Kanagawa induces cation permeability and activates endogenous Gardos K+ channels. Consequences include breakdown of phosphatidylserine asymmetry, which depends at least partially on cellular loss of K+.     

Protein and lipid components of the pigeon erythrocyte membrane.

The plasma membrane of the nucleated pigeon erythrocyte was isolated by a method that is simple, reproducible and minimally disruptive, the final preparation consisting of whole cell 'ghosts', recovered at over 40% yield. Alternative methods, which yield membrane fragments, were also tested and some of their possible disadvantages demonstrated. Analysis of the protein components of the isolated membranes by gel elctrophoresis in the presence of sodium dodecyl sulphate revealed that their composition is very similar to that of the proteins of human erythrocyte membranes. However, two major proteins are unique to the nucleated cell membrane; these have apparent mol.wts. of 97000 and 57000. Also, the bands designated 4.2 (74500 mol.wt.) and 6 (35000 mol wt.) by Steck [(1974) J. Cell Biol. 62, 1-19] for the human cell membrane are absent from pigon cell membrane. Glycosylated membrane proteins could not be detected in gels stained with the periodate-Schiff-base procedure. Analysis of membrane phospholipids revealed the same components known to be present in mammalian erythrocytes, though in different proportions. These findings are discussed in the light of known physiological and biochemical differences between avian and mature mammalian erythrocytes.     

Activation and mechanism of Clostridium septicum alpha toxin

Clostridium septicum produces a single lethal factor, alpha toxin (AT), which is a cytolytic protein with a molecular mass of approximately 48kDa. The 48kDa toxin was found to be an inactive protoxin (AT^pro) which could be activated via a carboxy-terminal cleavage with trypsin. The cleavage site was located approximately 4kDa from the carboxy-terminus. Proteolytically activated AT^pro had a specific activity of approximately 1.5 × 10⁶ haemolytic units mg^-1. The trypsin-activated toxin (AT^act) was haemolytic, stimulated a prelytic release of potassium ions from erythrocytes which was followed by haemoglobin release, induced channel formation in planar membranes and aggregated into a complex of M_r >210000 on erythrocyte membranes. AT^pro did not exhibit these properties. AT^act formed pores with a diameter of at least 1.3-1.6 nm. We suggest that pore formation on target cell membranes is responsible for the cytolytic activity of alpha toxin.     

Water exchange through erythrocyte membranes: Nuclear magnetic resonance studies on the effects of inhibitors and of chemical modification of human membranes

The changes in water diffusion across human erythrocyte membranes following exposure to various inhibitors and proteolytic enzymes have been studied on isolated erythrocytes suspended in isotonic buffered solutions. An important issue was to investigate whether the sulfhydryl reacting reagents that have been applied in osmotic experiments showed similar effects on diffusional permeability. It was found that mercurials, including mersalyl, were the only sulfhydryl reacting reagents that were efficient inhibitors. Under optimal conditions a similar degree of inhibition (around 45%) was found with all mercury-containing sulfhydryl reagents. Other reagents, including the sulfhydryl reagent DTNB, phloretin, or H2DIDS, the specific inhibitor of the anion transport system in erythrocyte membrane, did not appear to inhibit significantly the diffusional permeability. No changes in water diffusion were noticed after exposure to erythrocytes to trypsin and chymotrypsin. A new kind of experiments was that in which the effects of exposure of erythrocytes to two or more agents were studied. It was found that none of the chemical manipulations of membranes that did not affect water diffusion hampered the inhibitory action of mercurials. These findings show that the SH groups involved in water diffusion across erythrocyte membrane do not react with any of the other SH reagents aside from mercurials and that the molecular mechanism of water transport is not affected by chymotryptic cleavage of band 3 protein into the 60 and 35 kD fragments. The NMR method appears as a useful tool for studying changes in water diffusion in erythrocyte membranes following various chemical manipulations of the membranes with the aim of locating the water channel.     

Distinct steps in the penetration of adenylate cyclase toxin of Bordetella pertussis into sheep erythrocytes. Translocation of the toxin across the membrane.

Adenylate cyclase (AC) toxin from Bordetella pertussis penetrates eukaryotic cells and upon activation by calmodulin generates unregulated levels of intracellular cAMP. The process of toxin penetration into sheep erythrocytes was resolved into three consecutive steps including insertion, translocation, and intracellular cleavage. Insertion of the toxin into the cell membrane occurred over a wide temperature range (4-36 degrees C). In contrast, translocation of the toxin, i.e. transfer of the NH2-terminal catalytically active fragment across the membrane, occurred only above 20 degrees C and was highly temperature-dependent. While a single exposure of the toxin to Ca2+ was sufficient for its insertion into the plasma membrane, toxin translocation required exogenous Ca2+ at mM concentrations. Translocation was not affected by pretreatment of cells with trypsin, N-ethylmaleimide, and sodium carbonate at alkaline pH. The NH2-terminal fragment of the toxin was cleaved in the cell releasing the 45-kDa active AC into the cytosol. The cleavage was blocked by treatment of cells with N-ethylmaleimide. It is hypothesized that the COOH-terminal portion of the toxin creates in the membrane a channel through which the NH2-terminal fragment is translocated.