The C-terminal domain of perfringolysin O is an essential cholesterol-binding unit targeting to cholesterol-rich microdomains Yukiko Shimada 1 , Mikako Maruya 2 , Shintaro Iwashita 3 and Yoshiko Ohno-Iwashita 1 1 Biomembrane Research Group, Tokyo Metropolitan Institute of Gerontology; 2 Department of Cell Biology, Tokyo Metropolitan Institute of Medical Science; 3 Mitsubishi Kagaku Institute of Life Sciences (MITILS), Machida, Tokyo, Japan There is much evidence to indicate that cholesterol forms lateral membrane microdomains (rafts), and to suggest their important role in cellular signaling. However, no probe has been produced to analyze cholesterol behavior, especially cholesterol movement in rafts, in real time. To obtain a potent tool for analyzing cholesterol dynamics in rafts, we prepared and characterized several truncated fragments of h-toxin (perfringolysin O), a cholesterol-binding cytolysin, whose chemically modified form has been recently shown to bind selectively to rafts. BIAcore and structural analyses demonstrate that the C-terminal domain (domain 4) of the toxin is the smallest functional unit that has the same cho- lesterol-binding activity as the full-size toxin with structural stability. Cell membrane-bound recombinant domain 4 was detected in the floating low-density fractions and was found to be cofractionated with the raft-associated protein Lck, indicating that recombinant domain 4 also binds selectively to cholesterol-rich rafts. Furthermore, an enhanced green fluorescent protein- 1 domain 4 fusion protein stains mem- brane surfaces in a cholesterol-dependent manner in living cells. Therefore, domain 4 of h-toxin is an essential choles- terol-binding unit targeting to cholesterol in membrane rafts, providing a very useful tool for further studies on lipid rafts on cell surfaces and inside cells. Keywords: raft; microdomain; cholesterol; BIAcore; perfringolysin O. In recent years, accumulating evidence has indicated that cholesterol forms lateral membrane microdomains (lipid rafts) in which sphingolipids are also enriched [1,2]. This microdomain is a scaffold where specific proteins assemble and plays a pivotal role in signal transduction and many other cellular functions [3]. Lipid rafts have been isolated by sucrose density gradient centrifugation after treatment of cells with Triton X-100 (TX-100) by taking advantage of their insolubility in detergent at 4 °C [2]. Changes in cholesterol content, either by inhibition of its biosynthesis or by its removal from the plasma membrane, affect the localization of proteins associated with rafts, and thus affect raft function [4,5], suggesting (an) essential role(s) of cholesterol in the structural maintenance and function of rafts. Although interest in cholesterol functions and demand for their analyses have been increasing rapidly, there are almost no probes that have been used to detect and monitor cholesterol in rafts. Filipin is a reagent currently used for the cytochemical staining of cholesterol in fixed cells [6]. However, filipin permeabilizes the cell membrane and binds to cell cholesterol indiscriminately [6,7]. We have examined the cytolytic mechanism of perfringo- lysin O (h-toxin) secreted by Clostridium perfringens,which binds to membrane cholesterol and causes cell disruption. Cholesterol-mediated binding to a membrane is a trigger for forming toxin oligomers, leading to the formation of large pores. This pore formation directly causes cell membrane damage resulting in cell disruption. We prepared several h-toxin derivatives that retain specific binding activity to cholesterol but lack cytolytic activity. Ch [8] is a protease- nicked derivative and loses the capacity to oligomerize below 20 °C. MCh [9] and BCh [10,11] are methylated and biotinylated derivatives of Ch, respectively, and both have the same binding specificity and affinity for membrane cholesterol as intact h-toxin, but cause no damage to membranes at 37 °Corbelow. The h-toxin derivatives bind to liposomes with high cholesterol content but not to liposomes that contain less than 20 mol% of cholesterol [12], which strongly suggests their selective binding to cholesterol-enriched membrane domains. Recently we demonstrated that BCh selectively binds to cholesterol in cholesterol-rich microdomains of intact cells, domains that fulfill the criteria of lipid rafts [7]. The BCh bound to various types of cells was found to be TX-100 insoluble at 4 °C[7].WhenBCh-bound platelets were treated with TX-100 and fractionated on a sucrose-density gradient, BCh was predominantly localized in the floating low-density fractions (FLDF) where cholesterol, sphingomyelin and Src family kinases are enriched [7]. Depletion of one-third of the cholesterol from cells with b-cyclodextrin, which was accompanied by more than a 70% reduction in cholesterol from FLDF, almost Correspondence to Y. Shimada or Y. Ohno-Iwashita, Biomembrane Research Group, Tokyo Metropolitan Institute of Gerontology, 35–2 Sakae-cho, Itabashi-ku, Tokyo 173–0015, Japan. Fax: +81 3 35794776, Tel: +81 3 39643241 ext. 3063 or 3068, E-mail: yshimada@tmig.or.jp or iwashita@tmig.or.jp Abbreviations: Br-DSPC, brominated distearoylphosphatidylcholine; DOPC, dioleoylphosphatidylcholine; DSPC, distearoylphospha- tidylcholine; EGFP, enhanced green fluorescent protein; FLDF, floating low-density fractions; IPTG, isopropyl thio-b- D -galactoside; 2OHpbCD, 2-hydroxypropyl-b-cyclodextrin; PE, phosphatidyl- ethanolamine; TX-100, Triton X-100. (Received 5 August 2002, revised 10 October 2002, accepted 30 October 2002) Eur. J. Biochem. 269, 6195–6203 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03338.x completely abolished BCh binding to lipid rafts. This indicates that the binding of BCh to lipid rafts depends on their cholesterol content. BCh, coupled with fluorescent- avidin or colloidal gold-avidin, has been used as a probe to analyze the distribution of membrane cholesterol by fluorescence microscopy and electron microscopy [10,11,13]. Rafts are abundant at the plasma membrane surface, and are also found in intracellular compartments in the endo- cytic pathway [14]. In a further study on lipid rafts, analysis of the dynamic movement of intracellular rafts, for instance raft assembly and raft trafficking, is necessary as well as that on the membrane surface. However, as staining with BCh requires fluorescent avidin, it is not suitable for real-time imaging of the dynamic movement of lipid rafts in living cells. Especially, such movement inside the cell is hard to trace by the indirect fluorescence method. To establish a system for real-time imaging of rafts, we have attempted to isolate the cholesterol-binding domain of the toxin. Based on the 3D crystal structure [15], h-toxin comprises four b-sheet-rich domains, and only domain 4, located at the C-terminus, is structurally autonomous [15]. There is evidence to suggest that a cholesterol-binding site is located within domain 4. For example, a C-terminal fragment obtained by trypsin digestion (T2), including predominantly domain 4, binds to cholesterol and to cholesterol-containing membranes [16]. Furthermore, experiments with many toxins mutated in the tryptophan-rich motif at the C- terminus have revealed a significant reduction in the membrane-binding activity [17]. However, the cholesterol- binding site of h-toxin has not been clearly defined as yet. We have characterized cholesterol binding activity in relation to toxin stability and identified the smallest region necessary for its activity. In this paper, we show that domain 4 of h-toxin is an essential cholesterol-binding unit targeting to cholesterol in lipid rafts. Furthermore, we demonstrate that enhanced green fluorescent protein (EGFP)-tagged domain 4 may be a promising tool for analyzing raft dynamics in living cells. EXPERIMENTAL PROCEDURES Materials Anti-h-toxin antibody was raised in rabbits as described previously [16]. A rabbit antibody specific to the C-terminus of h-toxin was produced using the peptide antigen CGTTLYPGSSITYN (amino acids C449-N472 of the mature form of h-toxin). Cholesterol and isopropyl thio-b- D -galactoside (IPTG) were purchased from Sigma (St Louis, MO, USA). Hydroxylapatite was from Seikagaku (Tokyo, Japan). Peroxidase-conjugated anti-rabbit IgG was pur- chased from Medical & Biological Laboratories (Nagoya, Japan). BIAcore sensor chip SA was from BIAcore (Uppsala, Sweden). Alexa Fluor TM 546 was from Molecular Probes, Inc. (Eugene, OR, USA). Plasmid construction Plasmid pNSP10 containing the perfringolysin O gene (pfoA) [17] was used to construct pfoA derivatives encoding the C-terminal region of the toxin. The DNA fragment containing the T2¢ (V337-N472)-encoding region was pre- pared by digesting pNSP10 with SpeIandXhoIand inserting it between the NheIandXhoIsitesonthe3¢ side of the sequence encoding His-tag and a thrombin-cleavage site in the expression vector pET-28b. The DNA fragment containing the D4 (K363-N472)- or DN-D4 (S371-N472)- encoding region was amplified from pNSP10 by a poly- merase chain reaction and ligated into pET-28b digested with NheIandXhoI. The NheI restriction sites of the polymerase chain reaction-amplified products were created as noncomplementary ends of the amplification primers. For DC-D4 (K363-T470), a pNSP10-derived plasmid enco- ding D471 (K1-T470) was used as a PCR template. The poly- merase chain reaction primers used were 5¢-CTCAGGC TAGCAAGGGAAAAATAAACTTAGATC-3¢ (for D4 and DC-D4), 5¢-TCAGAGCTAGCAGTGGAGCCTATG TTGCACAG-3¢ (for DN-D4) and 5¢-TGGTGGTGG TGCTCGAGTGC-3¢. For the construction of a plasmid encoding a His-tag-EGFP-D4 fusion protein, a DNA frag- ment containing the EGFP-encoding region was amplified from pEGFP-N3 (Clontech) by the polymerase chain reaction with forward primer A (5¢-CGTTCTAGAGT GAGCAAGGGCGAGGAGCTG-3¢) and reverse primer B(5¢-ATCTACGTCGGCTAGCCTTGTACAGCTCGT CCATGCCGAG-3¢). The fragment was then ligated into the NheI site of the plasmid encoding His-tag-D4. The DNA sequences in the resulting plasmids were confirmed by the dideoxynucleotide chain-termination method [18]. Plasmids were introduced into E. coli strain BL21 (DE3) [19] (Novagen, Madison, WI, USA) by transformation of competent cells. Protein production and purification E. coli strain BL21(DE3) was used for the overexpression of His-tag-T2¢, His-tag-D4, His-tag-DN-D4 and His-tag-DC- D4 fusion proteins. After induction with IPTG, E. coli cells were harvested by centrifugation and lysed in native lysis buffer (50 m M phosphate buffer, pH 8.0, 300 m M NaCl, 10 m M imidazole) by ultrasonication. The overexpressed proteins were partially purified from the cytoplasmic fraction of E. coli by Ni + -NTA agarose column chroma- tography. His-tagged toxin fragments bound to Ni + -NTA agarose were eluted with 250 m M imidazole. For further purification, the fractions containing His-tagged toxin fragments were loaded onto a hydroxylapatite column equilibrated with 20 m M phosphate buffer, pH 8.0, and the flow through fraction was collected. Active fragments were recovered in the flow through fraction, while inactive ones were adsorbed to hydroxylapatite. The pooled fraction was incubated with thrombin at an enzyme to substrate ratio of 1 : 100 for 5 h at room temperature to cleave the His-tag and the cleavage reaction was stopped by the addition of 1m M phenylmethanesulfonyl fluoride. After protease treat- ment, the pooled fraction was applied to a butyl-agarose column equilibrated with 20 m M Tris/HCl, pH 7.5, con- taining 0.8 M (NH 4 ) 2 SO 4 . The toxin fragments were eluted with 0.2 M (NH 4 ) 2 SO 4 and dialyzed against Hepes-buffered saline, pH 7.0, at 4 °C. The purity of the toxin fragments was checked by SDS/PAGE [20]. The sequence GSHMAS remains attached to the N termini of purified fragments after thrombin cleavage. Toxin derivatives MCh and BCh, and the T2 fragment were prepared as described previously [9,10,16]. 6196 Y. Shimada et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Binding to cholesterol on TLC plates The cholesterol-binding activity of each toxin fragment was examined on TLC plates as described previously [16]. Toxin fragments bound to the TLC plates were detected with anti- whole h-toxin antibody. Preparation of lipid vesicles Phospholipids [dioleoylphosphatidylcholine (DOPC) or brominated distearoylphosphatidylcholine (Br-DSPC)] alone or 1 : 1 (mol/mol) mixtures of cholesterol and phospholipids were evaporated to make lipid films. Hepes-buffered saline was added to the lipid films and mixed vigorously. The lipid dispersion was sonicated with a Branson Sonifier and centrifuged at 9000 g for 5 min to remove undispersed lipids. For kinetic analysis by the BIAcore system, 0.1 mol% of biotin-phosphatidylethanol- amine (PE) was added to the lipid mixture before evapor- ation. The lipid dispersion containing biotin-PE was frozen and thawed 10 times, and the resultant multilamellar vesicles were extruded through 100 nm polycarbonate membranes in a Liposofast apparatus (Avestin Inc., Ottawa, Canada). Binding of toxin fragments to liposomes and MOLT-4 cells The toxin fragments were incubated individually with liposomes in Hepes-buffered saline containing 1 mgÆmL )1 bovine serum albumin for 30 min at room temperature. The mixtures were centrifuged at 350 000 g for 30 min at 4 °C. Both the pellets and the supernatants were analyzed by Western blotting using anti-C-terminal peptide antibody. For measurement of binding to cultured cells, MOLT-4 (10 5 cells) [21] were washed twice with phosphate-buffered saline (NaCl/P i ), and then incubated with or without 5 m M 2-hydroxypropyl-b-cyclodextrin (2OHpbCD) in serum-free RPMI 1640 for 15 min at 37 °C. The cells were then washed twice with NaCl/P i , and then incubated with toxin frag- ments (0.3 lg) in NaCl/P i containing 1 mgÆmL )1 bovine serum albumin for 30 min at 37 °C. Toxin fragments bound to cells were obtained in the pellet by centrifugation. Measurement of Trp fluorescence A 0.3 nmol sample of each purified toxin fragment was mixed with either phospholipid-liposomes [toxin fragment/ phospholipids 1 : 30 (mol/mol)] or cholesterol/phospho- lipid-liposomes [toxin fragment/cholesterol 1 : 30 (mol/mol)] in 1.5 mL of Hepes-buffered saline. After incubation for 10 min at room temperature, emission spectra were recor- ded in the range of 300–400 nm at an excitation wavelength of 295 nm with a Shimadzu spectrofluorophotometer RF-5000. Circular dichroism spectra Circular dichroism spectra were recorded on a JASCO J-720 spectropolarimeter with 5 mm pathlength cells. Toxin fragments in NaCl/P i were scanned from 250 to 200 nm. Molecular ellipticity ([h]) was calculated based on the mean residue weight of each fragment. Kinetic analysis of toxin fragments in cholesterol binding The cholesterol-binding kinetics of toxin fragments were determined by surface plasmon resonance [22] using BIA- core 1000 TM . All experiments were carried out at 25 °Cin degassed Hepes-buffered saline. DOPC/cholesterol lipo- somes containing biotin-PE were injected and immobilized on a sensor chip SA that has dextran matrix-attached streptavidin. After immobilization, the final signal increase was 1000 response units (RU). MCh,T2¢ and D4 were dialyzed against the same buffer and applied to the liposome-immobilized sensor chip. Analyses were per- formed at a flow rate of 20 lLÆmin )1 . Another sensor chip SA bearing immobilized DOPC/biotin-PE liposomes was used as a control. Susceptibility of toxin fragments to a protease Liposome-bound toxin fragments were obtained by ultra- centrifugation after incubation with liposomes as described above. Subtilisin BPN¢ was mixed with toxin fragment or liposome-bound toxin fragment preparations in 50 m M phosphate buffer, pH 7.0. The mixture was incubated for 30 min at 27 °C and the cleavage reaction was stopped by the addition of phenylmethanesulfonyl fluoride at a final concentration of 1 m M . TX-100 treatment and sucrose density gradient fractionation In order to isolate TX-100-insoluble membranes, MOLT-4 cells were extracted on ice for 20 min with 1% TX-100 in TNE buffer (25 m M Tris/HCl, pH 7.5, 150 m M NaCl, 5m M EDTA) containing 2 m M phenylmethanesulfonyl fluoride, 1 m M leupeptin, 25 lgÆmL )1 aprotinin and 20 lgÆmL )1 soybean trypsin inhibitor. Then the TX-100- soluble and -insoluble fractions were separated by centri- fugation at 15 000 g for 15 min and analyzed by Western blotting. For sucrose density gradient fractionation, TX-100-treated cells were homogenized with a Potter– Elvehjem homogenizer and mixed with an equal volume of 80% sucrose, overlaid with 2.4 mL of 35% sucrose and 1.3 mL of 5% sucrose in TNE buffer. After centrifugation at 250 000 g for 18 h at 4 °C in a SW55 rotor, 11 fractions of 0.4 mL each were collected from the top and the pellet was suspended in 0.4 mL of TNE buffer. Fluorescence microscopy For fluorescence microscopic observation, EGFP-D4 was overexpressed in E. coli and purified with Ni + -NTA agarose as described for the purification of the T2¢ and D4 fragments. The fractions eluted from the Ni + -NTA agarose column were applied to a butyl-agarose column equilibrated with 20 m M Tris/HCl, pH 7.5, containing 0.8 M (NH 4 ) 2 SO 4 . EGFP-D4 was eluted with 20 m M Tris/ HCl, pH 7.5. Cells were incubated with EGFP-D4 in serum-free RPMI-1640 for 5 min at 37 °C. After washing with RPMI 1640, fluorescence images of living cells were observed using an Olympus fluorescent microscope. No significant difference in cell viability was found before and after EGFP-D4 addition by checking with trypan blue Ó FEBS 2002 A probe for raft cholesterol in living cells (Eur. J. Biochem. 269) 6197 exclusion. More than 95% of the cells were viable after being labeled with EGFP-D4, washed and incubated for one hour at room temperature. Others Tricine-SDS/PAGE was performed by the method of Schagger [23]. N-terminal sequences of toxin fragments were analyzed with a precise cLC protein sequencer (Applied Biosystems) according to the manufacturer’s recommendations. RESULTS Isolation of cholesterol-binding fragments of h-toxin As the C-terminal portion of h-toxin might retain choles- terol binding activity, several N-terminal truncated frag- ments were constructed and expressed in E. coli.Toxin fragments T2¢,D4,DN-D4, and DC-D4 [Fig. 1(A)] were purified from the cytoplasmic fraction of E. coli by Ni + -NTA agarose, hydroxylapatite and butyl agarose column chromatographies. We also prepared the T2 fragment [16] and a toxin derivative, MCh [9], from recombinant h-toxin protein by biochemical modification as described before. Three fragments (T2, T2¢ and D4) were obtained as single proteins [Fig. 1(B)]. On the other hand, DN-D4, which is truncated by eight-amino acids from the N terminus of domain 4, and DC-D4, which has only two amino acids deleted from the C terminus of domain 4, were not stable during the purification process. Therefore small amounts of the DN-D4 and DC-D4 fragments were recovered. N-terminal sequence analysis of the T2¢ and D4 fragments revealed that they have the expected N-terminal sequence. Binding specificity and affinity of toxin fragments for cholesterol We first examined the binding specificity of the toxin fragments to cholesterol by immunostaining with anti-h- toxin antibody on lipid-developed TLC plates. The T2¢ and D4 fragments, as well as T2 and MCh, bound only to free cholesterol among lipids [Fig. 2(A)], indicating specific recognition of free cholesterol by these toxin fragments. To investigate the binding specificity to cholesterol as a mem- brane component further, liposomes containing cholesterol were prepared. After incubation with DOPC/cholesterol liposomes, the T2¢ and D4 fragments were detected in the pellet fraction [Fig. 2(B)], which shows that the fragments have binding activity similar to those of intact h-toxin, MCh and T2. Negligible amounts of toxin fragments were bound to liposomes prepared without cholesterol, indicating the specific binding for cholesterol in membranes. These results show that cholesterol-binding activity resides in domain 4, and the binding specificity is the same as that of h-toxin. They also indicate that the amino acid sequence of whole domain 4 is required for folding into a stable structure for cholesterol binding. As expected, neither T2¢ nor D4 showed hemolytic activity (data not shown) despite of their ability to bind to cholesterol-containing membranes. We next examined the cholesterol-binding kinetics of these toxin fragments by surface plasmon resonance using a sensor chip on which cholesterol-containing liposomes were immobilized (Table 1). Association and dissociation rate constants for T2¢ and D4 binding to cholesterol-containing liposomes were almost the same as corresponding constants for MCh binding (Table 1), indicating that the deletion of domains 1–3 from the toxin did not influence the binding kinetics. As a result, the dissociation constants also exhibit similar values. These experiments with liposomal mem- branes show that domain 4 retains the same binding specificity and binding affinity for membrane cholesterol as h-toxin. Protease susceptibility of membrane-bound toxin fragments To investigate the state of the toxin fragments during membrane binding, we analyzed the susceptibilities of the T2¢ and D4 fragments to protease in the presence and absence of membranes (Fig. 3). In the absence of liposomal Fig. 1. Isolation of toxin fragments. (A) Schematic drawings of h-toxin and its derivatives. Recombinant toxin fragments T2¢,D4,DN-D4 and DC-D4 were produced in E. coli with an N-terminal His-tag for purification. After thrombin digestion, an extra of six amino acids remained at the N terminus of each fragment (dotted rectangles). T2 is a tryptic fragment of h-toxin and the N-terminal sequence was deter- mined previously [8]. Amino acid numbers shown in each toxin derivative correspond to the positions in h-toxin. The black rectangles represent the tryptophan-rich motif in domain 4. The arrowhead indicates the position of a protease-nicked site located between T144 and H145 in MCh and BCh. (B) SDS/PAGE of purified toxin frag- ments and derivatives. Toxin fragments (T2¢ and D4) were expressed in E. coli and purified from the cytoplasmic fraction by a series of column chromatographies as described in Experimental procedures. MCh and T2 were obtained from recombinant h-toxin by biochemical modifi- cations as described before. Lane M shows molecular size marker proteins. 6198 Y. Shimada et al. (Eur. J. Biochem. 269) Ó FEBS 2002 membranes, both fragments were digested by subtilisin BPN¢ into undetectable pieces (Fig. 3, lanes 2 and 6). When the D4 fragment bound to DOPC/cholesterol liposomes was treated with subtilisin BPN¢, no changes in fragment size were observed on Tricine-SDS/PAGE (Fig. 3, lane 8). In the case of T2¢, after binding to DOPC/cholesterol liposomes, subtilisin BPN¢ digestion produced a proteolytic fragment with a molecular size similar to that of the D4 fragment (Fig. 3, lane 4). The resultant proteolytic frag- ments were recovered from the gel (lanes 4 and 8) and their N-terminal sequences were analyzed by a protein sequencer. The N-terminal amino acid sequence of the liposome- bound D4 fragment after digestion was found to be GSHMASKGKI, which corresponds to the N-terminal sequence of the intact D4 fragment, indicating that no cleavage occurred. On the other hand, the N-terminal amino acid sequence of the digested product of the liposome-bound T2¢ fragment was determined to be STE- YSKGKIN, indicating that 27 amino acid residues were cleaved from the N terminus of T2¢. The cleaved position is shown in the 3D structure of the T2¢ fragment (Fig. 3), demonstrating that the entire domain 4 region is protected from protease digestion. This finding is consistent with the Fig. 2. Binding of h-toxin fragments to cholesterol. (A) Specific binding of h-toxin fragments to cholesterol on TLC plates. Lipid mixtures containing 2 lg each of standard neutral lipids were applied to TLC plates and the plates were developed. The plates were then incubated with toxin fragments or derivatives and bound proteins were detected by immunostaining with anti-(whole h-toxin) Ig. Lipids were detected with 3% cupric acetate/8% phosphoric acid by heating at 140 °C (ÔlipidsÕ lane). PC, phosphatidylcholine; SM, sphingomyelin. (B) Toxin binding to liposomal membranes. h-Toxin, MCh and toxin fragments were incubated with DOPC liposomes or DOPC/cholesterol liposomes for 20 min at room temperature. After centrifugation, the total frac- tion (T), and the resulting supernatant (S) and pellet (P) fractions were separated and analyzed by SDS/PAGE followed by immunoblotting with an antibody against h-toxin C-terminal peptide. Lane M shows molecular size marker proteins. Table 1. Kinetic analysis of toxin fragment binding to cholesterol by surface plasmon resonance. Kinetic analysis of toxin fragment binding to immobilized cholesterol-containing liposomes was performed as described in ÔExperimental proceduresÕ. The binding kinetics were analyzed by the software BIAEVALUATION 2.1. Each value is given as mean ± SE, n ¼ 6. Toxin fragment k on ( M )1 Æs )1 ) k off (s )1 ) K D ( M ) D4 (1.1 ± 0.29) · 10 5 (6.0 ± 0.47) · 10 )3 (5.2 ± 0.14) · 10 )8 T2¢ (1.7 ± 0.47) · 10 5 (1.4 ± 0.17) · 10 )2 (8.8 ± 0.30) · 10 )8 MCh (5.1 ± 2.5) · 10 4 (6.5 ± 0.11) · 10 )3 (1.5 ± 0.81) · 10 )7 Fig. 3. Susceptibility of T2¢ and D4 to protease. T2¢ and D4 fragments were digested with subtilisin BPN¢ in the presence or absence of cho- lesterol-containing liposomes. After protease treatment, the resultant fragments were separated by Tricine-SDS/PAGE and analyzed by Western blotting with an antibody against h-toxin C-terminal peptide. In the lower panel the 3D structures of T2¢ and D4 are shown in black against a gray background of the whole h-toxin structure. The arrow indicates the position of cleavage by the protease in the presence of cholesterol-containing liposomes. The N-terminal sequences of T2¢ and D4 are also shown in the lower panel. Ó FEBS 2002 A probe for raft cholesterol in living cells (Eur. J. Biochem. 269) 6199 observation that the resultant fragment was nearly the same size as the D4 fragment on Tricine-SDS/PAGE. As liposomes not containing cholesterol do not protect the fragments against protease digestion (data not shown), cholesterol-dependent membrane binding is required for protection. Tryptophan fluorescence and circular dichroism spectra of the fragments To examine the conformation of the toxin fragments that bind to liposomes, intrinsic tryptophan fluorescence was measured (Fig. 4). Among seven tryptophan residues in h-toxin, domain 4 contains six, while only one (Trp137) is located in the N-terminal region (domain 1). To eliminate the spectral contribution of Trp137, we used W137F, a mutant h-toxin in which Trp137 in the N-terminal region is replaced by Phe [17]. Some tryptophan residues in domain 4, especially those within the 11-amino acid consensus sequence (tryptophan-rich motif), insert into the membrane lipid layer when intact h-toxin binds to the membrane [24]. Upon binding to cholesterol-containing membranes, tryp- tophan fluorescence in all toxin fragments was markedly enhanced (Fig. 4, dotted and dashed line), as is observed in intact h-toxin [25]. Next we examined whether the enhanced tryptophan fluorescence of the toxin fragments is quenched by liposomes containing Br-DSPC [24,25], in which bro- mines were placed on the 9,10-carbon atoms of the acyl chains. It is known that bromine atoms quench intrinsic tryptophan fluorescence in their vicinity [26,27]. When T2, T2¢ and D4 bound to Br-DSPC/cholesterol liposomes, the tryptophan fluorescence enhanced by binding to cholesterol was remarkably quenched by Br-DSPC (Fig. 4, long dashed line), in a similar manner to those of h-toxin and W137F. Compared to W137F and MCh,themaximalemission wavelengths of the T2, T2¢ and D4 fragments are distinctly longer (336–338 vs. 345–346 nm) in the absence of lipo- somes (Fig. 4, solid line). These results show that the environment of the tryptophan residues in the T2, T2¢ and D4 fragments is more exposed to solvent than in the case of the full-size toxins. However, after binding to cholesterol- containing liposomes, T2, T2¢ and D4 showed a blue shift in the maximal emission wavelength resembling that (336 nm) of intact h-toxin, MCh and W137F (Fig. 4, dotted and dashed line). These results suggest that the toxin fragments may change the conformation of their cholesterol-binding site to that of the full-size toxin, when they bind to membrane cholesterol. We also analyzed their secondary structures by circular dichroism measurement. The D4 fragment is enriched in b-sheets (data not shown), which is consistent with the structure predicted from X-ray crystallography, supporting the idea that biosynthesized domain 4 automatically folds into the secondary structure of the native toxin. On the other hand, the spectra of the T2 and T2¢ fragments exhibit more disordered structures than that of D4. These data imply that their extra N-terminal sequences other than domain 4 might be a disordered structure. Selective binding of D4 to lipid rafts The binding characteristics of the D4 fragment to intact cell membranes was examined. The D4 fragment was detected in the cell fraction after incubation with MOLT- 4 cells (Fig. 5A, )2OHpbCD, pellet fraction). Treatment with 5 m M 2OHpbCD for 15 min at 37 °C, which depletes cholesterol by 30%, caused significant reduction in the number of D4 fragments bound to MOLT-4 cells (Fig. 5A, 5 m M 2OHpbCD), demonstrating cholesterol- dependent binding. We have previously shown that BCh binds selectively to lipid rafts in intact platelets [7]. To analyze the selectivity of binding, we first examined the detergent-insolubility of the D4 fragment bound to MOLT-4 cells. After extraction with 1% TX-100 on ice, the membrane-bound D4 fragment was recovered in the Triton-insoluble membrane fraction (Fig. 5A, TX). Next, we examined the distribution of the cell membrane- bound D4 fragment in sucrose gradient fractionation following solubilization with TX-100. The D4 fragment was recovered predominantly in the FLDF where lipid rafts are known to be located (Fig. 5B). Cholesterol and a tyrosine kinase, Lck [2,28], were recovered in the same fractions (Fig. 5B), confirming that these fractions are enriched in lipid rafts. In parallel experiments, BCh- bound cells were treated with TX-100 and fractionated on a sucrose gradient, which showed that the distribution pattern of D4 is the same as that of BCh (data not shown). These results show that the D4 fragment binds selectively to lipid rafts, indicating that the binding characteristics of BCh targeting to lipid rafts can be ascribed to domain 4. Visualization of the probe To understand the biological function of lipid rafts, it is necessary to examine the distribution of cell membrane Fig. 4. Fluorescence emission spectra of toxin fragments and derivatives. The intrinsic tryptophan fluorescence of toxin fragments and deriva- tives was examined (solid lines) as described in Experimental proce- dures. These proteins were incubated with three types of liposomes, DOPC liposomes (dotted lines), DOPC/cholesterol liposomes (dotted and dashed lines), and Br-DSPC/cholesterol liposomes (long dashed lines). After 10 min at room temperature, the fluorescence emission spectra were measured. Maximal emission wavelengths in the presence of DOPC/cholesterol liposomes and in the absence of liposomes are displayed on the graphs in nanometers. 6200 Y. Shimada et al. (Eur. J. Biochem. 269) Ó FEBS 2002 cholesterol in real time. Thus we attempted to visualize membrane cholesterol using domain 4 directly labeled with a fluorescent dye in live cells. As domain 4 has no membrane-damaging activity by itself, it is a good candidate for the construction of cholesterol-specific probes. To prepare the fluorescent toxin fragment, the T2¢ and D4 fragments were labeled with Alexa 546. Alexa- labeled toxin fragments stained cell surfaces and 2OHpbCD treatment abolished this staining, indicating cholesterol-specific binding (data not shown). We tried another approach in which EGFP was fused to the N terminus of D4. The EGFP-D4 fusion protein was overproduced in E. coli and purified as described in ÔExperimental proceduresÕ. Following incubation with EGFP-D4 for 5 min at room temperature, clear fluores- cent labeling was observed on the surface of live MOLT-4 cells (Fig. 6A). EGFP-D4 stains cells in a cholesterol- dependent manner as no staining was observed in 2OHpbCD-treated cells (Fig. 6B). Together with the finding that the D4 fragment binds selectively to lipid rafts, EGFP-D4 allows us to visualize membrane choles- terol in lipid rafts of live cells. DISCUSSION To clarify the physiological significance of lipid rafts, many experimental tools have been used. Bacterial toxins that target components in rafts are often used as raft markers [29]. Cholera toxin is used to detect ganglioside GM1, which is enriched in lipid rafts [30]. Sphingomyelin, a major component of rafts, is a target for lysenin, which is secreted by earthworms [31]. However, no tool has been reported for the detection of cholesterol in lipid rafts in living cells. In this paper, to obtain a probe for targeting raft cholesterol, we isolated and characterized the minimal toxin fragment of h-toxin that binds to cholesterol with the high specificity of native h-toxin without cytolytic activity. The D4 fragment, which corresponds to domain 4, retains a stable structure and has cholesterol-binding activity with high affinity. As to the structural features of the purified D4 fragment, the circular dichroism measurements revealed that the isolated domain 4 exhibits a secondary structure enriched in b-sheets, in good agreement with that predicted from crystal structure [15]. However, the maximal emission wavelength of the intrinsic tryptophan fluorescence is red- shifted in the D4 fragment (Fig. 4), which indicates that the environment of the tryptophan residues in the D4 fragment is more exposed to solvent than in the native toxin. As no tryptophan is present at the interface of domain 4 with domains 1–3, the above results imply a structural change accompanied by the removal of domains 1–3. Therefore, domain 4 is an autonomous domain, but some co-operation with domains 1–3 is necessary for folding into the native tertiary structure. However, it should be noted that this structural difference does not affect the binding affinity and specificity of D4 to membrane cholesterol. In addition, upon binding to cholesterol, the polarity of the local environment around the tryptophan residues in the D4 fragment becomes the same as in the intact toxin (Fig. 4), suggesting that binding to membranes restores the conformation of the D4 fragment to that in the membrane-bound form of the full- size toxin. This is in striking contrast with some C-terminal Fig. 5. D4 fragment is enriched in FLDF after binding to membrane cholesterol. (A) MOLT-4 cells (10 6 cellsÆmL )1 ) treated with (5 m M 2OHpbCD) or without () 2OHpbCD) 2OHpbCD were incubated with D4 fragment for 15 min at 37 °C. After centrifugation, the total fraction (T) and the resultant supernatant (S) and pellet (P) fractions were analyzed by SDS/PAGE followed by Western blotting with an anti C-terminal peptide antibody. A portion of the resultant pellet obtained from the 2OHpbCD untreated sample was extracted with 1% TX-100 for 20 min on ice and the soluble (TX, S) and insoluble (TX, P) fractions were separated by centrifugation at 15 000 g.(B)D4frag- ment-bound MOLT-4 cells were treated with 1% TX-100, homogen- ized and subjected to a sucrose density gradient centrifugation. Eleven fractions (starting at the top) and the pellet (P) were collected. The distributions of D4, Lck and cholesterol were measured. Fig. 6. Staining of cell surface cholesterol in living cells with EGFP-D4. (A) MOLT-4 cells were incubated with EGFP-D4 for 5 min at room temperature. Labeled cells were observed under a fluorescent micro- scope. (B) MOLT-4 cells were treated with 5 m M 2OHpbCD to deplete cholesterol and then stained with EGFP-D4. Left panel, phase con- trast; right panel, fluorescence. Ó FEBS 2002 A probe for raft cholesterol in living cells (Eur. J. Biochem. 269) 6201 truncated mutants, in which a similar red shift of tryptophan fluorescence is accompanied by the complete loss of cholesterol-binding activity [32]. It is noteworthy that even a small truncation of domain 4 at either the N or C terminus causes instability of the products. Especially, the deletion of only two amino acids from the C terminus might cause product instability, resulting in little production. Therefore, we conclude that the D4 fragment, which corresponds to domain 4, is the smallest cholesterol-binding unit that is structurally stable. Recently recombinant domain 4 (r-d4) of streptolysin O was isolated by a combination of TX-100- extraction, denaturation with urea and ethanol precipitation [33]. Such drastic procedures might affect the conformation of native structure. Furthermore, bovine serum albumin was required for the renaturation of r-d4, resulting in failure in analysis of the toxin structure. On the other hand, our procedures for the purification of recombinant domain 4 are under native conditions without the use of any detergents or denaturants, which makes it possible to analyze its structure precisely. At present, the amino acid residues involved in cholesterol binding remain unidentified. Recombinant domain 4 should provide a good tool for analyzing cholesterol-binding sites, for instance, by NMR on the toxin–cholesterol interaction. The cytolytic activity of h-toxin depends on its oligo- merization, which leads to pore formation in the cell membrane. We have previously shown that T2 lacks oligomerization activity due to the deletion of the N-terminal portion, resulting in no cytolytic activity [16]. T2¢ and D4, shorter fragments than T2, also lack cytolytic activity. In this report we demonstrate that the isolated domain 4 has the ability to recognize and bind to membrane cholesterol, the first step in h-toxin action. The isolated domain 4 has a binding affinity for cholesterol similar to that of the full-size toxin as revealed by surface plasmon resonance measurement (Table 1). We previously showed that binding of the toxin to membrane cholesterol triggers its conformational change around tryptophan residues in domain 4 [24,25]. Such conformational change occurs without oligomerization, as a similar change was observed for isolated T2 fragment (Fig. 4). Recently a model was proposed that toxin oligomerization triggers the insertion of a portion of domain 3 as b-hairpins, which contribute to the formation of a transmembrane pore [34,35]. Although domain 3 might play a role for pore formation, our results clearly demonstrated that the insertion of domain 3 is not required for maintaining the high affinity of the toxin for the membrane, as revealed by BIAcore analysis (Table 1). The D4 fragment has selective binding affinity for cholesterol in lipid rafts (Fig. 5B) as demonstrated with BCh [7], a protease-nicked and biotinylated full-size h-toxin. Therefore the selectivity of binding to lipid rafts is ascribed to domain 4. We also show that EGFP-D4 is a potent probe for the detection of cell surface cholesterol in live cells. As most membrane-bound EGFP-D4 was recovered in FLDF (data not shown), the staining should display the distribution of cholesterol in membrane rafts. The cell- staining profile with EGFP-D4 is quite similar to that with Alexa-labeled D4, suggesting the fusion with EGFP does not influence the cholesterol-binding activity of D4. Previ- ously we reported that BCh is a specific probe for the detection of cholesterol by fluorescent microscopy [10,11]. Cholesterol on the outer surface of the plasma membrane can be stained with BCh coupled with FITC-conjugated avidin. On the other hand EGFP-D4 can be used to visualize cell surface cholesterol in both living and fixed cells in only one step. Therefore staining with EGFP-D4 has the advantage that the distribution and movement of choles- terol in living cells can be monitored without artifacts linked to the cross-linkage of FITC-conjugated avidin. Further- more, it is possible to use EGFP-D4 to detect intracellular distribution of cholesterol. We are now constructing an expression vector for intracellular EGFP-D4 to analyze the movement of cholesterol-containing microdomains within live cells. Thus, the recombinant domain 4 could become a key tool for investigating the distribution and movement of not only the membrane surface, but also intracellular cholesterol in lipid rafts. ACKNOWLEDGMENTS We thank Dr Ichiro Yahara for providing facilities of kinetic analysis by the BIAcore systems. We thank Drs Yoshitaka Nagai and Koichi Suzuki for support and encouragement. We are grateful to Drs Inomata and Hayashi for technical advice and to Dr M.M. Dooley- Ohto for reading the manuscript. This work was supported by grants from the Japan Science and Technology Corporation, the Japan Science Society (to Y.S.) and from a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science, ONO Medical Research Foundation and Life Science Foundation of Japan (to Y.O I.). Y.S. is a Domestic Research Fellow of the Japan Science and Technology Corporation. REFERENCES 1. Simons, K. & Ikonen, E. (1997) Functional rafts in cell mem- branes. Nature 387, 569–572. 2. Brown, D.A. & London, E. (2000) Structure and function of sphingolipid- and cholesterol-rich membrane rafts. J. Biol. Chem. 275, 17221–17224. 3. Simons, K. & Toomre, D. (2000) Lipid rafts and signal trans- duction. Mol. Cell Biol. 1, 31–39. 4. Hooper, N.M. (1999) Detergent-insoluble glycosphingolipid/ cholesterol-rich membrane domains, lipid rafts and caveolae. Mol. Membr. Biol. 16, 145–156. 5. Ostermeyer, A.G., Beckrich, B.T., Ivarson, K.A., Grove, K.E. & Brown, D.A. (1999) Glycosphingolipids are not essential for for- mation of detergent-resistant membrane rafts in melanoma cells. methyl-beta-cyclodextrin does not affect cell surface transport of a GPI-anchored protein. J. Biol. Chem. 274, 34459–34466. 6. Neufeld, E.B., Cooney, A.M., Pitha, J., Dawidowicz, E.A., Dwyer, N.K., Pentchev, P.G. & Blanchette-Mackie, E.J. (1996) Intracellular trafficking of cholesterol monitored with a cyclo- dextrin. J. Biol. Chem. 271, 21604–21613. 7. Waheed, A.A., Shimada, Y., Heijnen, H.F.G., Nakamura, M., Inomata, M., Hayashi, M., Iwashita, S., Slot, J.W. & Ohno- Iwashita, Y. (2001) Selective binding of perfringolysin O derivative to cholesterol-rich membrane microdomains (rafts). Proc. Natl Acad. Sci. USA 98, 4926–4931. 8. Ohno-Iwashita, Y., Iwamoto, M., Mitsui, K., Kawasaki, H. & Ando, S. (1986) Cold-labile hemolysin produced by limited pro- teolysis of theta-toxin from Clostridium perfringens. Biochemistry 25, 6048–6053. 9.Ohno-Iwashita,Y.,Iwamoto,M.,Ando,S.,Mitsui,K.& Iwashita, S. (1990) A modified theta-toxin produced by limited proteolysis and methylation: a probe for the functional study of membrane cholesterol. Biochim. Biophys. Acta 1023, 441–448. 10. Iwamoto, M., Morita, I., Fukuda, M., Murota, S., Ando, S. & Ohno-Iwashita, Y. (1997) A biotinylated perfringolysin O 6202 Y. Shimada et al. (Eur. J. Biochem. 269) Ó FEBS 2002 derivative: a new probe for detection of cell surface cholesterol. Biochim. Biophys. Acta 1327, 222–230. 11. Fujimoto, T., Hayashi, M., Iwamoto, M. & Ohno-Iwashita, Y. (1997) Crosslinked plasmalemmal cholesterol is sequestered to caveolae: analysis with a new cytochemical probe. J. Histochem. Cytochem. 45, 1197–1205. 12. Ohno-Iwashita, Y., Iwamoto, M., Ando, S. & Iwashita, S. (1992) Effect of lipidic factors on membrane cholesterol topology – mode of binding of theta-toxin to cholesterol in liposomes. Biochim. Biophys. Acta 1109, 81–90. 13. Mobius, W., Ohno-Iwashita, Y., van Donselaar, E.G., Oorschot, V.M.J., Shimada, Y., Fujimoto, T., Heijnen, H.F.G., Geuze, H.J. & Slot, J.W. (2002) Immunoelectron microscopic localization of cholesterol using biotinylated and non-cytolytic perfringolysin O. J. Histochem. Cytochem. 50, 43–55. 14. Simons, K. & Ikonen, E. (1998) Protein and lipid sorting from the trans-Golgi network to the plasma membrane in polarized cells. Semin. Cell Dev. Biol. 5, 503–509. 15. Rossjohn, J., Feil, S.C., McKinstry, W.J., Tweten, R.K. & Parker, M.W. (1997) Structure of a cholesterol-binding, thiol-activated cytolysin and a model of its membrane form. Cell 89, 685–692. 16. Iwamoto,M.,Ohno-Iwashita,Y.&Ando,S.(1990)Effectof isolated C-terminal fragment of theta-toxin (perfringolysin O) on toxinassemblyandmembranelysis.Eur. J. Biochem. 194, 25–31. 17. Sekino-Suzuki, N., Nakamura, M., Mitsui, K. & Ohno-Iwashita, Y. (1996) Contribution of individual tryptophan residues to the structure and activity of theta-toxin (perfringolysin O), a choles- terol-binding cytolysin. Eur. J. Biochem. 241, 941–947. 18. Sanger, F., Nicklen, S. & Coulson, A.R. (1977) DNA sequencing with chain-terminating inhibitors. Proc. Natl Acad. Sci. USA 74, 5463–5467. 19. Studier, F.W., Rosenberg, A.H., Dunn, J.J. & Dubendorff, J.W. (1990) Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60–89. 20. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 21. Minowada, J., Onuma, T. & Moore, G.E. (1972) Rosette-forming human lymphoid cell lines. I. Establishment and evidence for origin of thymus-derived lymphocytes. J. Natl. Cancer Inst. 49, 891–895. 22. Jin, L., Shieh, J J., Grabbe, E., Adimoolam, S., Durbin, D. & Jonas, A. (1999) Surface plasmon resonance biosensor studies of human wild-type and mutant lecithin cholesterol acyltransferase interactions with lipoproteins. Biochemistry 38, 15659–15665. 23. Schagger, H. & von Jagow, G. (1987) Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa. Anal. Biochem. 166, 368– 379. 24. Nakamura, M., Sekino-Suzuki, N., Mitsui, K. & Ohno-Iwashita, Y. (1998) Contribution of tryptophan residues to the structural changes in perfringolysin O during interaction with liposomal membranes. J. Biochem. 123, 1145–1155. 25. Nakamura, M., Sekino, N., Iwamoto, M. & Ohno-Iwashita, Y. (1995) Interaction of theta-toxin (perfringolysin O), a cholesterol- binding cytolysin, with liposomal membranes: change in the aro- matic side chains upon binding and insertion. Biochemistry 34, 6513–6520. 26. Bolen, E.J. & Holloway, P.W. (1990) Quenching of tryptophan fluorescence by brominated phospholipid. Biochemistry 29, 9638– 9643. 27. Gonzalez-Manas, J.M., Lakey, J.H. & Pattus, F. (1992) Bromi- nated phospholipids as a tool for monitoring the membrane insertion of colicin A. Biochemistry 31, 7294–7300. 28. Kasahara, K., Watanabe, Y., Yamamoto, T. & Sanai, Y. (1997) Association of Src family tyrosine kinase Lyn with ganglioside GD3 in rat brain. Possible regulation of Lyn by glycosphingolipid in caveolae-like domains. J. Biol. Chem. 272, 29947–29953. 29. Fivaz, M., Abrami, L. & van der Goot, F.G. (1999) Landing on lipid rafts. Trends Cell Biol. 9, 212–213. 30. Parton, R.G. (1994) Ultrastructural localization of gangliosides; GM1 is concentrated in caveolae. J. Histochem. Cytochem. 42, 155–166. 31. Yamaji, A., Sekizawa, Y., Emoto, K., Sakuraba, H., Inoue, K., Kobayashi, H. & Umeda, M. (1998) Lysenin, a novel sphingo- myelin-specific binding protein. J. Biol. Chem. 273, 5300–5306. 32. Shimada, Y., Nakamura, M., Naito, Y., Nomura, K. & Ohno- Iwashita, Y. (1999) C-terminal amino acid residues are required for the folding and cholesterol binding property of perfringolysin O, a pore-forming cytolysin. J. Biol. Chem. 274, 18536–18542. 33. Weis, S. & Palmer, M. (2001) Streptolysin O: the C-terminal, tryptophan-rich domain carries functional sites for both mem- brane binding and self-interaction but not for stable oligomer- ization. Biochim. Biophys. Acta 1510, 292–299. 34. Shepard, L.A., Shatursky, O., Johnson, A.E. & Tweten, R.K. (2000) The mechanism of pore assembly for a cholesterol-depen- dent cytolysin: formation of a large prepore complex precedes the insertion of the transmembrane beta-hairpins. Biochemistry 39, 10284–10293. 35. Heuck, A.P., Hotze, E.M., Tweten, R.K. & Johnson, A.E. (2000) Mechanism of membrane insertion of a multimeric beta-barrel protein: perfringolysin O creates a pore using ordered and coupled conformational changes. Mol. Cell 6, 1233–1242. Ó FEBS 2002 A probe for raft cholesterol in living cells (Eur. J. Biochem. 269) 6203 . change accompanied by the removal of domains 1–3. Therefore, domain 4 is an autonomous domain, but some co-operation with domains 1–3 is necessary for folding. should provide a good tool for analyzing cholesterol-binding sites, for instance, by NMR on the toxin–cholesterol interaction. The cytolytic activity of