Clostridium perfringens ε-Toxin Forms a Heptameric Pore within the Detergent-insoluble Microdomains of Madin-Darby Canine Kidney Cells and Rat Synaptosomes*

Clostridium perfringens ε-toxin, which is responsible for enterotoxaemia in ungulates, forms a heptamer in rat synaptosomal and Madin-Darby canine kidney (MDCK) cell membranes, leading to membrane permealization. Thus, the toxin may target the detergent-resistant membrane domains (DRMs) of these membranes, in analogy to aerolysin, a heptameric pore-forming toxin that associates with DRMs. To test this idea, we examined the distribution of radiolabeled ε-toxin in DRM and detergent-soluble membrane fractions of MDCK cells and rat synaptosomal membranes. When MDCK cells and synaptosomal membranes were incubated with the toxin and then fractionated by cold Triton X-100 extraction and flotation on sucrose gradients, the heptameric toxin was detected almost exclusively in DRMs. The results of a toxin overlay assay revealed that the toxin preferentially bound to and heptamerized in the isolated DRMs. Furthermore, cholesterol depletion by methyl-β-cyclodextrin abrogated their association and lowered the cytotoxicity of the toxin toward MDCK cells. When ε-protoxin, an inactive precursor able to bind to but unable to heptamerize in the membrane, was incubated with MDCK cell membranes, it was detected mainly in their DRMs. These results suggest that the toxin is concentrated and induced to heptamerize on binding to a putative receptor located preferentially in DRMs, with all steps from initial binding through pore formation completed within the same DRMs.

Most mammalian cells have at least two types of discrete lipid microdomains, lipid rafts and caveolae. These microdomains, also known as detergent-resistant membranes (DRMs) 1 or detergent-insoluble glycosphingolipid-enriched membranes, are lateral assemblies of sphingolipids and cholesterol that form a separate liquid-ordered phase in the liquid-disordered matrix of the lipid bilayer (1,2). They favor partitioning of distinct classes of proteins, such as glycosylphosphatidylinositol (GPI)-anchored proteins, transmembrane proteins, and diacylated signaling proteins (1,2). The proposed biological relevance of these observations is that the microdomains function as platforms that compartmentalize signals (1)(2)(3). Increasing evidence strongly supports the involvement of DRMs in membrane sorting and ligand-mediated signaling (for reviews, see Refs. 4 and 5).
DRMs are utilized by some microbial pathogens as a portal for their entry or that of toxins, thereby avoiding the classic endosome-lysosome pathway mediated by clathrin-coated vesicles (6 -8). They are also targeted by certain pore-forming toxins, e.g. Aeromonas hydrophila aerolysin (9,10), Clostridium perfringens perfringolysin O (11), and Bacillus thuringiensis Cry1A (12). C. perfringens ⑀-toxin (ET) possesses biochemical properties similar to those of aerolysin, which led us to speculate that ET also attacks DRMs and also to expect that it may be potentially useful, as aerolysin is, for understanding the molecular organization and functional entity of DRMs (4).
ET is a potent toxin produced by C. perfringens types B and D that causes rapidly fatal enterotoxaemia in livestock (13). This toxin accumulates mainly in the brain and kidneys when intravenously injected into rats (14), causing injury to neuronal cells (15)(16)(17) and cerebral blood vessels (18). One characteristic feature of the toxin is its extraordinarily high potency; 20 ng of ET kills a mouse within 1 h. 2 We have previously shown that proteolytic activation of ⑀-protoxin (ProET) is due to the removal of a C-terminal peptide (19,20) and that activated ET forms a heptamer in rat synaptosomal membranes (20). Similar ET oligomerization has been reported for Madin-Darby canine kidney (MDCK) cell membranes, where the potassium ion permeability is increased by the toxin (21). Recently, the ability of heptamerized ET to form a ion-permeable channel was demonstrated by an in vitro study involving artificial lipid bilayers (22). Furthermore, ET is rich in ␤-strand (23), suggesting that it functions as a ␤-barrel pore-forming toxin. Considering these characteristic features of ET, it may be possible that ET utilizes DRMs as a concentration device in a similar way to aerolysin and thereby rapidly exhibits its high toxicity.
Functional and biochemical studies on the association of trafficking, signaling, and pathogen entry with DRMs have been based mainly on two operational strategies: depletion of cell cholesterol with methyl-␤-cyclodextrin (MbCD), a cholesterol-extracting agent, and isolation of membrane fractions that are resistant to cold detergent extraction (24). By means of these strategies, we have assessed the possible association between ET and DRMs that were prepared from rat synaptosomal membranes and MDCK cells, the only ET-sensitive cell line so far available. The results presented here show that ET heptamerizes exclusively in DRMs like aerolysin, which constitutes convincing evidence of the proposed action of DRMs as a device for the concentration and heptamerization of poreforming toxins (9). Furthermore, we have shown by means of a ProET binding experiment that it binds mainly to DRMs. This suggests that ET binds predominantly to DRMs, and that all the steps from binding to pore formation are completed within the same DRMs, unlike aerolysin in which the precursor is activated in non-DRMs and then moves into DRMs. Finally, to address the question of whether or not the cytotoxicity of ET is reversed by inhibition of ET heptamerization, we examined the effect of cholesterol depletion on ET toxicity toward MDCK cells. The finding that MbCD had an inhibitory effect on ET cytotoxicity provides an insight not only into the molecular mechanism underlying the cytotoxicity of ET but also into the cases of other pore-forming toxins.

EXPERIMENTAL PROCEDURES
Materials-MbCD, lovastatin, mevalonate, and a protease inhibitor mixture were obtained from Sigma. Mouse monoclonal antibodies against caveolin-1 and flotillin-1 were purchased from BD Transduction Laboratories (Lexington, KY). A rabbit polyclonal antibody against the ␣-1 subunit of Na ϩ ,K ϩ -ATPase was purchased from Upstate Biotechnology (Lake Placid, NY). Horseradish peroxidase-labeled goat antimouse IgG and anti-rabbit IgG were purchased from BioRad.
Construction of Recombinant ProET and ET-A plasmid, pEP1, that encodes ProET (20) was digested with EcoRI and XhoI to abolish the HindIII site in the non-coding region, blunt-ended, and then religated. Two synthetic oligonucleotides, 5Ј-TATGGGCCATCATCATCATCATC-ACAGCAGCGGCATCGAAGGTCGTATGAGACGTGCGTCTGTTAA-3Ј (sense; the His 6 tag, factor Xa, and protein kinase A recognition sequences are underlined, doubly underlined, and italicized, respectively) and 5Ј-AGCTTTAACAGACGCACGTCTCATACGACCTTCGATGCCG-CTGCTGTGATGATGATGATGATGGCCCA-3Ј (antisense), were annealed and cloned into the above plasmid, which had been digested with NdeI and HindIII. The resultant plasmid, which encodes His 6 -⌬N-proET (ProET lacking an N-terminal propeptide of 13 amino acid residues but having a His 6 -tag, a factor Xa cleavage site, and a protein kinase A-specific phosphorylation site at the N terminus), was named pEL18. A NdeI-AvrII insert DNA fragment from pEL18 was cloned into pEP4, which was described previously (20). The resultant plasmid, named pEL20, enabled the expression of a His 6 -⌬N-ProET derivative (His 6 -⌬N-ProET with an additional factor Xa cleavage site at the C terminus). The identity of each of the His-tagged toxins was verified by nucleotide sequencing.
Preparation of Recombinant ProET and ET-Transformants of Escherichia coli BL21(DE3)pLysS (Novagen) carrying plasmids pEL18 and pEL20 were cultured, and the recombinant proteins were expressed as described previously (20). The toxins were purified by affinity chromatography on a Ni 2ϩ -chelating column (Amersham Biosciences) according to the manufacturer's instructions. Fractions containing His 6 -⌬N-ProET or the His 6 -⌬N-ProET derivative, as determined by SDS-PAGE, were pooled. Their N-terminal His 6 -tags and the C-terminal pro-regions of the His 6 -⌬N-ProET derivative were cleaved off at factor Xa-sensitive sites, as described previously (20), to generate recombinant ProET and ET, respectively. The purified toxins were dialyzed against Tris-buffered saline (TBS; 150 mM NaCl, 20 mM Tris-HCl, pH 7.4) and stored at Ϫ80°C. Because the cytotoxic activities toward MDCK cells of recombinant ProET and ET were comparable with those of ProET from C. perfringens cultures and the trypsin-activated ET, respectively (data not shown), they were used as ProET and ET throughout this study.
Radiolabeling of ProET and ET was performed as described previously except that [␥- 35  Preparation of MDCK Cell and Rat Synaptosomal Membranes-MDCK cells were grown to confluency in MEM (Eagle's minimum essential medium containing Earle's salts, 100 units/ml penicillin, and 100 g/ml streptomycin) supplemented with 10% heat-inactivated fetal bovine serum under 5% CO 2 at 37°C. All incubations of MDCK cells were carried out in this medium, unless otherwise stated. MDCK cell plasma membranes were prepared as described by Shortt et al. (25). Synaptosomal membranes were prepared from rat brains as described previously (20). The membranes were suspended in a small volume of TBS containing the protease inhibitor mixture.
MbCD Treatment and Triton X-100 Extraction of ET-bound Membranes-[ 35 S]-ET binding and heptamerization in the MDCK cell or rat synaptosomal membranes were carried out as described previously (20). After incubating the membranes with [ 35 S]-ET, the incubation mixture was centrifuged at 17,000 ϫ g for 15 min. The membrane pellet was washed three times with TBS at 4°C and then resuspended in TBS with or without 10 mM MbCD. After incubation at 37°C for 30 min, the incubation mixture was centrifuged at 17,000 ϫ g for 15 min. The precipitate was further extracted with TBS containing 1% TX-100 at 4°C for 30 min and then centrifuged at 17,000 ϫ g for 15 min. The supernatant and precipitate were analyzed by SDS-PAGE, followed by autoradiography with an imaging plate (Fuji Photo Film, Kanagawa, Japan).
Flotation-Centrifugation on a Sucrose Gradient-Separation of DRMs from detergent-soluble membranes (non-DRMs, non-detergentresistant membranes) and a cytosoluble fraction was carried out by flotation-centrifugation on a sucrose gradient (26), as follows. MDCK cells were incubated in fresh medium containing 100 ng/ml [ 35 S]-ET for 90 min. The cells were rinsed with Dulbecco's phosphate-buffered saline and then lysed by exposure to 1% TX-100 for 30 min at 4°C in TBS containing the protease inhibitor mixture. The lysate was scraped from the dishes with a cell scraper and then homogenized by passage through a 22-gauge needle. Synaptosomal membranes were also incubated with 2 g/ml [ 35 S]-ET in TBS containing 1 mg/ml BSA at 37°C for 90 min and then lysed with 1% TX-100 in TBS containing the protease inhibitor mixture. Portions of the lysates were brought to 40% sucrose in a final volume of 0.5 ml and overlaid with 3.5 ml of 30% sucrose in TBS, followed by 1 ml of 5% sucrose in TBS. The gradients were centrifuged for 18 h at 50,000 rpm (ϳ250,000 ϫ g) at 4°C in a Hitachi RPS65T rotor. Fractions (ϳ0.5 ml each) were collected from the bottom of the centrifuge tube, and aliquots were subjected to SDS-PAGE and immunoblot analysis.
Immunoblot Analysis of DRM Marker Proteins-Immunoblot analysis was carried out as described previously (27). In brief, aliquots of membrane samples (10 g of protein each) or 100-l aliquots of flotation sucrose gradient fractions were precipitated with trichloroacetic acid and then dissolved by heating in 2ϫ SDS-sample buffer at 99°C for 5 min. Samples were electrophoresed on a SDS-PAGE gel, followed by transfer to a nitrocellulose membrane. The blots were blocked with TTBS containing 5% skim milk and incubated first with the primary antibody in TTBS containing 1% skim milk, then with an appropriate horseradish peroxidase-conjugated secondary antibody, and finally with an enhanced chemiluminescence reagent (PerkinElmer Life Sciences).
ET Overlay Assaying-MDCK cell and synaptosomal membranes were fractionated by flotation-centrifugation in the same manner as described above except for omission of the preincubation with ET. The ET overlay assay for the detection of membrane-bound toxins was performed as follows. MDCK cell or synaptosomal membranes (1 g of protein each) or 10-l aliquots of flotation sucrose gradient fractions were diluted with TBS to 200 l and then slot-blotted onto a nitrocellulose membrane. The nitrocellulose membrane was incubated in TTBS supplemented with 5% skim milk at 25°C for 90 min, followed by a 90-min incubation in TTBS containing 1% skim milk and 1 g/ml [ 35 S]-ET or [ 35 S]-ProET. The membrane was then washed five times with TTBS for 10 min each. ET bound to the cell membranes was visualized by autoradiography. When the effect of MbCD on ET binding was examined, the membranes were preincubated in TBS with and without 10 mM MbCD at 37°C for 30 min, washed three times with TBS, and then slot-blotted.
The Effect of MbCD Treatment on ET Heptamerization-MDCK cell and synaptosomal membranes were incubated in TBS with and without 10 mM MbCD at 37°C for 30 min. The membranes were washed three times with TBS and then incubated with 2 g/ml [ 35 S]-ET at 37°C for various times. After incubation, the reaction mixture was dissolved by heating in SDS-sample buffer and then subjected to SDS-PAGE followed by autoradiography. The amounts of heptameric ET were determined with Image Gauge software (Fuji Photo Film) and are presented as percentages of the total amount of ET added to the incubation mixture.
Effect of Cholesterol Depletion on ET Cytotoxicity toward MDCK Cells-Cholesterol depletion from MDCK cells was carried out as described by Keller and Simons (28). Briefly, MDCK cells (2.0 ϫ 10 4 ) were cultured in the medium containing 4 M lovastatin and 0.25 mM mevalonate in a 96-well microculture plate for 48 h to minimize the de novo synthesis of cholesterol. Then the residual cholesterol was extracted by treatment with 10 mM MbCD in MEM for 80 min at 37°C. After washing two times with MEM, the MDCK cells were incubated with various concentrations of ET in MEM for 4 h and then subjected to the cytotoxicity assay.
Assay for ProET Binding to MDCK Cell Membranes-MDCK cell membranes were incubated with 2 g/ml [ 35 S]-ProET in TBS containing 0.1% BSA at 37°C for various times in a centrifuge tube, which had been pretreated overnight with 1% BSA in TBS to avoid nonspecific adsorption of the toxin to the tube wall. The reaction mixture was centrifuged at 17,000 ϫ g for 5 min and washed three times with TBS. The membrane precipitate was treated with cold 1% TX-100 and separated into non-DRM and detergent-insoluble DRM fractions by centrifugation at 17,000 ϫ g for 15 min as described above.

RESULTS
Association of ET Heptamers with DRMs-To assess the possible association of ET heptamerization with the DRMs of MDCK cell and rat synaptosomal membranes, we first examined the distribution of ET heptamers in the detergent-soluble and detergent-insoluble fractions after TX-100 extraction at low temperature. The SDS-resistant ET heptamer formed in MDCK cell membranes was found almost exclusively in the precipitate (detergent-insoluble fraction) after treatment with 1% TX-100 at 4°C followed by centrifugation, there being very little in the supernatant (detergent-soluble fraction) (Fig. 1A,  lanes 2 and 3). Its association with the detergent-insoluble fraction was affected only marginally by MbCD treatment alone (Fig. 1A, lane 4). However, the ET heptamer was found mainly in the detergent-soluble fraction when the precipitate after MbCD treatment was extracted with cold 1% TX-100 (Fig.  1A, lanes 5 and 6), suggesting that the heptameric ET is associated with DRMs. Similar results were obtained when the heptameric ET in synaptosomal membranes was analyzed (Fig.  1B). These results suggest that most ET heptamers are associated with DRMs in both MDCK cell and synaptosomal membranes. Although monomeric ET showed an apparently similar TX-100 extraction profile, its band intensity was too low to evaluate accurately. To analyze more precisely the distribution of heptameric ET in membranes, we fractionated ET-bound membranes by flotation-centrifugation. As shown in Fig. 2, the heptameric ET was detected almost exclusively in the DRMs, i.e. not in the non-DRMs, of ET-bound MDCK cells, as revealed by the distribution of a DRM marker, caveolin-1. Likewise, when fractions of ET-bound synaptosomal membranes were analyzed (Fig. 2B), the heptameric ET was detected almost exclusively in the DRM fraction containing a DRM marker, flotillin-1, i.e. not in that containing a non-DRM marker, Na ϩ ,K ϩ -ATPase ␣-1 subunit. These results indicated that ET heptamers are located predominantly within their DRMs.
The Effect of Cholesterol Depletion on ET Heptamerization-The association of heptameric ET with DRMs may result from the heptamerization of monomeric ET in the DRMs or from the translocation of ET heptamerized outside the DRMs. If ET heptamerization occurs in DRMs, it would be abrogated by MbCD treatment, which depletes cholesterol and disrupts the integrity of DRMs. To assess this possibility, we examined the effect of MbCD treatment on the rate of ET heptamerization. As shown in Fig. 3, treatment with MbCD decreased the rate of heptamerization; the level of heptameric ET in MDCK cell and

FIG. 2. Flotation gradient fractionation of ET-bound MDCK cells and synaptosomal membranes. MDCK cells (A) and synaptosomal membranes (B) were incubated in the presence of [ 35 S
]-ET (100 ng/ml medium and 2 g/ml TBS, respectively) at 37°C for 90 min and then extracted with TBS containing 1% TX-100 at 4°C for 30 min. Aliquots of the extracts were brought to 40% sucrose and then loaded at the bottom of a centrifuge tube. After flotation sucrose gradient ultracentrifugation, 11 (A) or 10 (B) different fractions (ϳ0.5 ml each) were collected from the bottom of the tube. The gradient fractions were precipitated with trichloroacetic acid and then dissolved by heating in 2ϫ SDS-sample buffer. Samples were subjected to SDS-PAGE, followed by immunoblotting (a) or autoradiography (b). Lanes 1-11, fractions from the top to bottom of the gradient. synaptosomal membranes was reduced to 1 and 12% of those in untreated control membranes, respectively. This suggests that ET heptamerization occurs at least mainly in DRMs, depending on their integrity. However, it still does not exclude the possibility that heptameric ET is translocated from non-DRMs, because MbCD may also affect cholesterol surrounding and outside of DRMs, altering the diffusion of ET therein, as has been suggested for the MbCD effect of proaerolysin on non-DRMs (9,29).
ET Binding to and Heptamerization in Isolated DRMs-We examined the ability of ET to heptamerize in DRMs and non-DRMs fractionated by flotation-centrifugation (Fig. 4). The heptameric ET was detected exclusively in the DRMs, i.e. not in the solubilized non-DRMs, of MDCK cells, as revealed by the distribution of a DRM marker, caveolin-1 (Fig. 4A). Likewise, when fractions of synaptosomal membranes were analyzed (Fig. 4B), heptameric ET was detected only in the fraction containing a DRM marker, flotillin-1, i.e. not in that containing a non-DRM marker, Na ϩ ,K ϩ -ATPase ␣-1 subunit. The distribution of [ 35 S]-ET detected in the toxin overlay assay with the same fractionated samples coincided with that of heptameric ET, suggesting that ET binds preferentially to the DRMs. Consistent with this, ProET, which is able to bind but unable to heptamerize, was also detected in DRMs but not in solubilized non-DRM fractions (Fig. 4, A and B). Therefore, it can be assumed that ET binds to a putative ET receptor predominantly located in DRMs. This was reinforced by the results of a toxin overlay assay with MbCD-treated membranes (Fig. 5). MbCD treatment decreased the binding of ET to MDCK cell and synaptosomal membranes to 22 and 9.6%, respectively, of the levels in untreated control membranes, as was calculated from the band intensity of each slot-blot.
Preferential Binding of ProET to DRMs of MDCK Cell Membranes-The levels of bound ET detected in the toxin overlay assay may reflect ET heptamerization rather than ET binding, because monomeric ET might have heptamerized during the incubation. Furthermore, the TX-100 used for the fractionation of membrane domains may affect toxin binding in non-DRMs. To confirm that ProET binds predominantly to DRMs but not to non-DRMs, MDCK cell membranes were incubated with ProET in the absence of TX-100 for various times and then fractionated by cold TX-100 extraction. As shown in Fig. 6, ϳ2-fold more ProET was detected in DRMs than in non-DRMs even at 5 min after initiation of the incubation; this proportion increased slightly during 80 min of incubation. This result indicates that a ProET receptor, which must be the same as the ET one (see "Discussion"), is predominantly located in DRMs. . After incubation, each reaction mixture was dissolved by heating in SDS-sample buffer and then subjected to SDS-PAGE followed by autoradiography. The insets show autoradiographs of heptameric ET. The amounts of heptameric ET were determined with Image Gauge software (Fuji Photo Film) and are presented as percentages of the total amount of ET added to the incubation mixture.

FIG. 4. ET binding to and ET heptamerization in DRMs of MDCK cells and synaptosomes. MDCK cells (A) and synaptosomal membranes (B)
were extracted with 1% Triton X-100 at 4°C for 30 min. Aliquots of the extracts were brought to 40% sucrose and then loaded at the bottom of a centrifuge tube. After flotation sucrose gradient ultracentrifugation, 11 fractions (ϳ0.5 ml each) were collected. a, 100-l aliquots of the gradient fractions were precipitated with trichloroacetic acid and then dissolved by heating in 2ϫ SDS-sample buffer. Samples were subjected to SDS-PAGE, followed by immunoblotting. b, 10 l of each fraction was diluted with 200 l of TBS and then slot-blotted onto a nitrocellulose membrane. The blotted membrane was subjected to overlay assay with [ 35 S]-ET or [ 35 S]-ProET. After washing, the membrane was subjected to autoradiography. c, [ 35 S]-ET (20 ng) was added to 10 l of the gradient fractions, followed by incubation at 37°C for 90 min. After incubation, the reaction mixture was heated in the SDSsample buffer and then subjected to SDS-PAGE, followed by autoradiography. Lane C, MDCK cell membranes or synaptosomal membranes without fractionation Lanes 1-11, fractions from the top to bottom of the gradient.
Thus, ET is very likely to bind predominantly to DRMs and to heptamerize therein.
Effect of Cholesterol Depletion on ET Cytotoxicity-To examine the effect of cholesterol depletion on ET cytotoxicity, MDCK cells were grown in the presence of lovastatin and mevalonate to minimize the de novo synthesis of cholesterol, followed by an 80-min incubation in the presence of MbCD. The cell viability revealed by the MTS assay was compared between MbCDtreated and untreated cells. The MbCD treatment decreased the ET cytotoxicity toward MDCK cells; the cell viability with 100 ng/ml ET of MDCK cells treated with MbCD was 5-fold lower than that of untreated cells (Fig. 7). DISCUSSION This study has revealed that the ET heptamer exists almost exclusively in DRMs of MDCK cells and rat synaptosomal membranes by showing the distribution of [ 35 S]-labeled ET heptamers in DRM fractions floating on sucrose density gradients. ET heptamer association with DRMs was further confirmed by its dislocation from the detergent-insoluble to the detergent-soluble fraction upon cholesterol depletion, as has been widely performed for the demonstration of DRM compo-nents. It was also shown that heptamerization is inhibited by cholesterol depletion. Furthermore, ET was shown by the ET overlay assay to bind to DRMs but not to solubilized non-DRMs, both of which were isolated by cold detergent extraction and flotation-centrifugation. To overcome the disadvantage of the ET overlay assay so that it might reflect ET heptamerization rather than ET binding and that TX-100 might affect the ET binding to non-DRMs, ProET (which is able to bind but unable to heptamerize) was used to monitor only toxin binding. The results showed that ProET binds predominantly to DRMs even in detergent-free conditions. ProET and ET can be assumed to bind to the same receptor, because ProET inhibits the binding of ET to the synaptosomal fraction (14), the heptamerization of ET therein (20), and also its mouse lethality (30). Therefore, a putative ET receptor appears to be located mainly in DRMs. The finding that a minor fraction of ProET was detected in non-DRMs does not rule out the possibility that some ET monomers diffuse into DRMs after binding to the outside of the DRMs. Overall, however, it seems likely that most ET molecules complete all the steps from initial binding to heptamerization within the same DRMs.
The hypothesis that DRMs act as a specialized concentration device for pore-forming toxins has been proposed based on the finding that aerolysin binds specifically to GPI-anchored proteins (31) and that proaerolysin is highly abundant in DRMs (32). It has been further reinforced by the finding that aerolysin heptamers formed in DRMs move to a non-DRM fraction upon treatment with saponin, a cholesterol-binding agent (9). The cytotoxicity and heptamerization of aerolysin in T lymphocytes, however, are insensitive to MbCD treatment, thus arguing against the dependence of aerolysin heptamerization on DRMs (33). A plausible explanation for these apparently contradictory observations is that cholesterol depletion affects heptamerization in most cell types but not in T lymphocytes expressing unusually high levels of aerolysin receptors (10). Another likely explanation is that MbCD only affects cholesterol surrounding and outside of DRMs (9,29). Although the effects of MbCD on the distribution of GPI-anchored proteins are controversial (9), the finding that ET preferentially binds to DRMs and hep- tamerizes therein in a cholesterol-dependent manner lends some support to these explanations and adds weight to the proposed function of DRMs as a platform that promotes the multimerization of pore-forming toxins.
ET and aerolysin are similar in that their heptamers are almost exclusively found in DRMs. However, they are different in that only a minor fraction of ProET binds to the outside of DRMs in contrast to proaerolysin, which binds equally to the inside and outside of DRMs (9). Proaerolysin is activated by furin recruited outside of DRMs (34), and hence binding of proaerolysin to non-DRMs is necessary for it to be activated. On the contrary, ProET must be activated before reaching its target, e.g. by digestive proteases in the intestine or C. perfringens -protease at the site of multiplication of the organism. Furthermore, a putative receptor seems to be confined to specific sites, i.e. the DRMs of only a few cell types, in contrast to the aerolysin receptor, which is ubiquitously distributed in GPI-anchored proteins. Thus, ET seems to use the simplest strategy: it completes all steps from binding to pore-formation within the same region.
ET and aerolysin also differ in the response of their heptamerization to MbCD treatment. The heptamerization of ET but not of aerolysin is sensitive to MbCD, although the latter is sensitive to saponin, a cholesterol-sequestering agent (9). This is not surprising, because abrogation of the receptor function by cholesterol depletion is not always a sure sign that DRMs are associated with signaling (24,35), and MbCD and another cholesterol-sequestering agent, filipin, exhibit different effects on antigen receptor-mediated signaling in B lymphocytes (36). If MbCD affects cholesterol surrounding and outside of DRMs more preferentially than cholesterol inside of DRMs (9,29), the difference in the MbCD effect between ET and aerolysin may suggest that heptamerization of the two toxins occurs in different environments within DRMs, e.g. ET heptamerizes at the MbCD-sensitive boundary between DRMs and non-DRMs. Whatever the mechanism, these different effects suggest the need for caution in the interpretation of the results of experiments that utilize these drugs.
There is no evidence indicating the intracellular location of aerolysin and ET (21,37). However, the cytotoxicity of these toxins may involve not only membrane disruption due to pore formation but also other cellular events, especially with low concentrations, e.g. apoptosis (10,38) and vacuolation of the endoplasmic reticulum (10, 32) by aerolysin, and mitotic disturbance by ET (39). Thus, the questions remain as to whether or not these events are linked to the heptamerization of poreforming toxins in DRMs, and if the answer is yes, how they are linked to each other. The inhibition of ET cytotoxicity and ET heptamerization by MbCD suggests that heptamerization is at least mainly responsible for the cytotoxicity of ET.
The present study has provided useful information pertaining not only to the molecular basis of ET cytotoxicity but also to biochemical approaches for identification of an ET receptor, which is a prerequisite for addressing the above questions. It should also be noted that ET is highly toxic, whereas ProET is non-toxic unless proteolytically converted to ET. It preferentially binds to the DRMs of a narrow range of susceptible cells such as renal epithelial and neuronal cells in vivo and MDCK cells in vitro. Thus, ProET may constitute a potential tool for studies on the topology, distribution, and dynamic properties of the DRMs in these cells by taking advantage of its specific association with their DRMs.