‘rHE JOURNAL OF BIOLOGICAL CHEMISTRY (c:! 1991 by The American Society for Biochemistry ’ and Moleculal ’ Biology, Inc

Vol. 266, No. 19, Issue of July 5, pp. 12449-12454,1991 Prin&d in U. S. A.

Isolation of a Tryptic Fragment from CZostridium perfringens &Toxin That Contains Sites for Membrane Binding and Self-aggregation*

(Received for publication, January 23, 1991)

Rodney K. Tweten$§, Richard W. Harris$, and Peter J. Sims$1111 From the $Department of Microbiology and Immunology, University of Oklahoma Health Sciences Center, Oklahoma City, Oklahoma 73190 and the lcardiouascular Biology Research Program, Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma 73104

Trypsin cleaves Clostridium perfringens 8-toxin (perfringolysin 0 or PFO) at a single site between residues 303 and 304 (Ohno-Iwashita, Y., Iwamoto, M., Mitsui, K., Kawasaki, H., and Ando, S. (1986) Biochemistry 25, 6048-6053; Tweten, R. K. (1988b) Infect. Immun. 56, 3228-3234) and yields an amino- terminal fragment of 30,208 Da (Tl) and a carboxyl- terminal fragment of 22,268 Da (T2). Both peptides were purified by reverse phase chromatography of trypsin-nicked PFO. Neither peptide retained hemo- lytic activity. Peptide T1 had no apparent effect on the hemolytic activity of PFO, whereas T2 was found to inhibit the hemolytic activity of PFO and was analyzed further. The order of binding of T2 and PFO to mem- branes did not alter the inhibitory effect of T2 on PFO- induced hemolysis, indicating that competitive binding by T2 for PFO membrane binding sites was not the basis for the observed inhibition. Further analysis showed that T2 could inhibit membrane-dependent flu- orescence energy transfer (FET) between PFO mole- cules labeled with fluorescein (fluorescent donor) or tetramethylrhodamine (fluorescent acceptor). This provided evidence that T2 could complex with PFO. T2 was also found to be incapable of self-aggregation (as opposed to PFO), since preincubation of T2 with either erythrocytes or erythrocyte ghost membranes did not affect the T2-dependent inhibition of hemolysis o r FET. These data indicate that T2 inhibits PFO- dependent hemolysis by forming a complex with PFO, which inhibits aggregation and that the membrane binding site and a single aggregation site remain intact on T2.

Clostridium perfringens 6’-toxin (perfringolysin 0 or PFO)’ and related toxins from species of the Streptococcus, Liste- ria,and Bacillus share many mechanistic and structural fea- tures. The cytolytic mechanism of these toxins is thought to involve binding of the toxin to the cell surface via cholesterol, lateral diffusion, and polymerization into supramolecular

*This work was supported by a grant (to R. K. T.) from the Presbyterian Health Foundation (Oklahoma City, OK) and Grant HL36061 from the National Institutes of Health (to P. J. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ii To whom correspondence should be addressed. Tel.: 405-271- 2133.

11 Established Investigator for the American Heart Association. ’ The abbreviations used are: PFO, perfringolysin 0; SDS-PAGE,

sodium dodecyl sulfate-polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; FET, fluorescence energy transfer.

complexes (Bhakdi et al., 1985; Hugo et al., 1986; Mitsui et al., 1979a; 1979b). The primary sequences of five of these toxins are known, and all exhibit significant amino acid sequence homology (Geoffroy et al., 1990; Kehoe et al., 1987; Tweten, 1988b; Walker et al., 1987). Although a considerable amount of information now exists on the primary structure of these toxins, little information is available concerning the location of the functional domains, the conformational change(s) that presumably occurs after membrane binding, or the the domains of PFO which are embedded in the membrane after binding.

PFO has been shown to be susceptible to trypsin cleavage at a single site (Ohno-Iwashita et al., 1986), which yields two peptides when denatured. Trypsin cleaves PFO on the amino- terminal side of residue 304 (Tweten, 1988b) resulting in a 275-residue (30,208 Da) amino-terminal peptide and an 196- residue (22,268 Da) carboxyl-terminal peptide termed T1 and T2, respectively (Ohno-Iwashita et al., 1986). The trypsin- nicked PFO was shown to be functionally similar to native PFO (Ohno-Iwashita et al., 1986), suggesting that the nick site was located in a domain that was not directly required for activity. Single trypsin cleavage sites are frequently found in bacterial toxins and are often found to cleave these toxins into functional domains as in the case for diphtheria toxin and Pseudomonas exotoxin A (reviewed in Stephen and Pie- trowski (1986)) . We have taken advantage of the presence of this trypsin site and have purified the two peptides generated by trypsin cleavage. This report provides evidence that the carboxyl-terminal peptide T2 inhibits the hemolytic mecha- nism of PFO by inhibiting PFO aggregation. Sites that facil- itate membrane binding and the formation of intermolecular contacts with PFO were found to be present on T2.

EXPERIMENTAL PROCEDURES

Media and Reagents-All bacterial media were obtained from Difco, and all chemicals and chromatography gels were obtained through Sigma unless noted otherwise. Trifluoroacetic acid and Op- tima grade acetonitrile were obtained from Fisher Scientific (Dallas, TX). Benzamidine Sepharose was obtained from Pharmacia LKB Biotechnology Inc. Fluorescein 5-isothiocyanate (isomer I) and tet- ramethylrhodamine-5 (and -6) isothiocyanate were obtained from Molecular Probes (Eugene, OR). Carrier-free ‘“1 was obtained from ICN Biomedicals Inc. (Costa Mesa, CA).

Purification of PFO and Peptides TI and T2-PFO was purified by a modification of the procedure of Tweten (1988a). Briefly, a 2- liter brain heart infusion culture of Escherichia coli JM109 carrying plasmid pRTlB (Tweten, 1988a) was grown at 37 “C in the presence of 1 mM isopropyl 8-D-thiogalactopyranoside for 18 h. The cells were harvested by centrifugation and the cell pellet washed with TE buffer (IO mM Tris-HCI, 1 mM EDTA, pH 8.0). PFO was released from the periplasmic space by osmotic shock, then concentrated and pressure dialyzed against 1 liter of TE buffer. The periplasmic extract was applied to a column (1.5 X 10 cm) packed with an anion exchange matrix (Accell QMA, Millipore, Bedford, MA). The hound proteins

12449

12450 Membrane Binding and Aggregation of a &Toxin Peptide were eluted with a linear gradient (200 ml) of 0-0.2 M NaCl in T E buffer. The fractions that contained the hemolytic activity were pooled, concentrated, and applied to a high resolution gel filtration column (1.6 X 50 cm packed with Superose 12 prep grade, Pharmacia LKB Biotechnology Inc.). The fractions that contained the hemolytic activity were pooled, concentrated, and applied to a high resolution anion exchange column (MA7Q, Bio-Rad, Richman, CA). The column was developed with a linear gradient (40 ml) from 0 to 0.2 M NaCl in T E buffer. PFO eluted at approximately 0.08 M NaC1. The recombi- nant-derived PFO was homogeneous as determined by SDS-PAGE (Laemmli, 1970).

Peptides T1 and T2 were generated by the overnight digestion of 2 mg of pure PFO with 80 pg of N-tosyl-L-phenylalanine chlorometh- ylketone-treated trypsin. The total reaction mixture was approxi- mately 0.5 ml and contained 10 mM Tris-HC1, pH 8.0, and 1 mM CaC12. After digestion the trypsin was removed by passing the tryp- sin:PFO mixture over a 0.5-ml column of benzamidine Sepharose. The nicked PFO was washed through the column in 10 mM Tris-HC1, pH 8.0, containing 0.3 M NaCl. Hemolytically active fractions were collected, pooled, and concentrated to approximately 0.5 ml. Separa- tion of peptides T1 and T2 was achieved by passing the trypsinized PFO over a Synchropak RP-P C18 reverse phase high pressure chromatography column (4.6 X 250 mm) (SynChrom, Inc., Lafayette, IN) equilibrated in 0.1% trifluoroacetic acid in distilled water. The peptides were eluted with a linear gradient from 35-50% acetonitrile. The peaks containing T1 and T2 were collected (usually 2 ml) and were simultaneously dialyzed and concentrated to approximately 0.5 ml by vacuum dialysis against 10 mM Tris-HCI, pH 7.5, with 5 mM dithiothreitol in a ProDiCon apparatus (Bio-Molecular Dynamics, Beaverton, OR) equipped with 10,000-Da cutoff membranes. Any precipitate in the samples was removed by centrifugation of the samples for 10 min a t 15,000 X g. The samples were stored on ice in a refrigerator until used.

Radioactive Labeling of PFO and Membrane Binding Analysis- The procedure for labeling PFO with ' 9 was essentially carried out as previously described (Ohno-Iwashita et al., 1986), except that chloramine T was used as the oxidizing agent instead of lactoperoxi- dase. Competitive binding studies between PFO and T2 were carried out as described by Ohno-Iwashita et al. (1990), except that recom- binant PFO was used as the labeled species.

Gel Electrophoresis-Proteins and peptides were analyzed by so- dium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS- PAGE) according to the procedure of Laemmli (1970). A 10% gel was used for separations. After electrophoresis the gel was stained with Coomassie R-250 and destained with a mixture of acetic acidmethano1:water (Laemmli, 1970).

Hemolysis Kinetics-Hemolysis kinetics were followed by changes in the right angle scatter of erythrocytes (Harris et al., 1991). The basis of the analysis relied on the differences in the right angle scatter of intact and lysed erythrocytes and has been shown to correlate with hemoglobin release from PFO-treated erythrocytes (Harris et al., 1991). Washed human erythrocytes were prepared by repeated dilu- tion and centrifugation (5000 X g) of the erythrocytes in PBS (10 mM sodium phosphate, pH 7.0, 145 mM NaC1, 5 mM KC1, 100 pg/ml gelatin (Difco), and 1 mM glucose) until the supernatant was sub- stantially free of hemoglobin. The erythrocytes were resuspended in PBS to approximately 1-5 X lo9 cells/ml and used within 24 h of preparation. Erythrocyte numbers were determined in a Coulter ZBI particle counter (Coulter Electronics, Hialeah, FL). Dynamic changes in the right angle scatter of erythrocytes undergoing lysis was meas- ured using a SLM 8000 spectrofluorimeter (SLM Instruments, Ur- bana, IL) equipped for individual sample cell stirring and temperature control as previously described (Harris et al., 1991). Wavelength selection was provided by double grating excitation monochromaters and single grating emission monochromaters. A 1-cm path length quartz cuvette containing a final reaction volume of 2 ml was used in all experiments. The excitation and emission monochromators were set to 590 nm with a slit width of 2 nm.

Two basic experimental protocols were used to assess the effect of peptide T2 on the lytic activity of PFO. The first set of experiments involved simultaneously injecting active PFO (500 ng) with and without various amounts of T2 into a stirred solution of washed human erythrocytes (5 X 107/ml) in PBS equilibrated at 30 "C. Hemolysis of the erythrocytes was followed by the decrease in the right angle scatter. Data acquisition was initiated 25 s prior to the injection of the PFO and was set to collect data at 1-s intervals.

Further characterization of the inhibitory effect of T2 on PFO- induced hemolysis was accomplished using temperature-shift experi-

ments. In these experiments T2 and PFO (or vice versa) were con- secutively bound to erythrocytes at 0-2 "C and then shifted to 37 "C to initiate hemolysis. This approach facilitated the separation of the binding component from the aggregation component of the lytic mechanism. PFO was decreased to 25 ng for these experiments to slow down the hemolytic rate. The concentration of T2 was set at 225 ng, which resulted in a 21 M excess of T2 over PFO. Either PFO or T2 was first prebound to erythrocytes (2 X lo7 in 40 pl) a t 0-2 "C, which allowed binding of either protein without initiating aggregation or hemolysis (Harris et al., 1991; Hugo et al., 1986). The unbound protein was removed by a 25-fold dilution of the mixture in ice-cold PBS and centrifugation (4 "C) at 10,000 X g for 10 s. The supernatant was discarded, the erythrocyte pellet was resuspended in 40 p1 of ice- cold PBS and the second protein (either PFO or T2) was added. The second protein was prebound for 2 min a t 0-2 "C and the unbound protein removed by another cycle of dilution and centrifugation. The erythrocytes were resuspended in 40 pl of ice-cold PBS and added to 1.96 ml of prewarmed PBS in a thermostated cuvette in the spectro- fluorimeter. This initiated the onset of aggregation and subsequently hemolysis. Data acquisition was initiated 10 s after the erythrocytes were added to the cuvette. The degree of inhibition of hemolysis by T2 was determined by the following equation.

T T

% Inhibition = -

The TS2 is the time to achieve 50% hemolysis for the uninhibited control, and Tir2 is the time required to achieve 50% hemolysis when PFO-induced hemolysis is inhibited by T2.

Erythrocyte Membrane Preparation-Erythrocyte ghosts substan- tially free of hemoglobin were prepared as previously described (Har- ris et al., 1991). Briefly, 10-20 ml of packed human erythrocytes were hemolyzed in 1 liter of ice-cold 5 mM phosphate buffer, 1 mM EDTA, pH 7.5. Removal of the cytoplasmic constituents was achieved by repeated rounds of centrifugation a t 15,000 X g and dilution with the same buffer. Membrane preparations were partially resealed by in- cubation in 10 mM phosphate buffer with 5 mM MgClz for 40 min at 37 "C. The erythrocyte ghosts were enumerated by resistive particle counting in a Coulter ZBI particle counter (Coulter Electronics, Hialeah, FL).

Aggregation of T2 with PFO-It has been shown previously that the aggregation of PFO can be monitored by fluorescence energy transfer (FET) (Harris et al., 1991). The aggregation of PFO is membrane-dependent. Fluorescent derivatives of PFO were generated as previously described (Harris et al., 1991). Fluorescein-labeled PFO was used as the fluorescent donor (PFOD) and tetramethylrhodamine- labeled PFO was used as the fluorescent acceptor (PFOA). The molar ratio of fluorophore to PFO was approximately 1:l for both donor and acceptor-labeled PFO. All fluorescence measurements were per- formed on a SLM 8000 spectrofluorimeter (SLM Instruments, Ur- bana, IL) equipped for individual sample cell stirring and temperature control. Wavelength selection was provided by double grating exci- tation monochromaters and single grating emission monochromaters. A 1-cm path length quartz cuvette containing a final reaction volume of 2 ml was used for all experiments. The excitation monochromator was set at 470 nm with a 16-nm slit width for the excitation of fluorescein 5-isothiocyanate. The excitation maximum for fluorescein is 490 nm; however, 470 nm was chosen to minimize the excitation of the tetramethylrhodamine acceptor. The emission monochromator was set at a 16-nm slit width and set to scan from 500 to 600 nm with a 1-s integration time. Spectra were recorded as total fluorescence emission.

For all FET experiments the PF0D:PFOA molar ratio was main- tained a t 1:4 with a total PFO concentration of 4.8 X lo-' M (250 ng/ ml) for all FET experiments. Erythrocyte ghost membranes (5 X lo7/ ml) were added to the PFOD + PFOA mixture and incubated for 2 min a t 30 "C to allow PFO aggregation to reach completion. An emission scan was then obtained from 500 to 600 nm. Scatter due to the membranes was subtracted from all experimental data. The total assay volume was maintained at 2 ml for all experiments, and the assay buffer was PBS with 1 mM dithiothreitol.

RESULTS

Purification of Peptide T2-Reverse phase separation of T2 from undigested PFO and T1 is shown in Fig. 1. T2 was eluted as the first major peak and was well separated from the other species. Fractions containing T1 and T2 were concentrated

Membrane Binding and Aggregation of a &Toxin Peptide 12451

0.20

h

E 0.15 0 00 e? 0 0.10

e 0 c m

0 0.05 P v)

-l

1 0 0

I

80 ' z zi 1. 60

-l -. - 4 0 13

8 20 -

o.oo 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 l I I I I I I I I I I I I 0

o 1 0 20 30 4 0 50 60

Time (min)

FIG. 1. Reverse phase purification of peptides T1 and T2. Shown is a typical chromatogram obtained when trypsinized PFO is separated by reverse phase chromatography. The peaks that contained T1, T2, and undigested PFO are designated. The dashed line indicates the percent change in the concentration of acetonitrile during the chromatographic run. Peptides T2 and TI eluted a t approximately 40 and 41.5% acetonitrile, respectively.

1 2 3 4

97.4 v

68 v

43 b

29 b

18.4 b

FIG. 2. SDS-PAGE analysis of purified peptide T2. PFO, T I , and T2 were separated by SDS-PAGE. Lane 1, molecular mass markers which are labeled to the left of the gel; lune 2, partially trypsin-nicked PFO showing intact PFO at 52 kDa, TI a t 30 kDa, and T2 at 22 kDa; lune3, purified TI; lane 4, purified T2. The smaller peptide (-19 kDa) located in the T2 preparation is the result of another trypsin nick in T2, and its appearance is variable depending on the length of the incubation with trypsin and trypsin concentra- tion.

and dialyzed to remove the trifluoroacetic acid and acetoni- trile. The extinction coefficient of T2 is significantly higher than T1, probably because of the fact that 6 of the 7 trypto- phan residues found in PFO are present within T2 (Tweten, 198813). During dialysis a precipitate formed which was some of the T2 peptide. After centrifugation of the T2 solution a t 15,000 X g the remaining soluble material was found to inhibit the hemolytic activity of PFO (see below). The amino-termi- nal peptide T1 also precipitated to some extent during di- alysis. After removal of the precipitate by centrifugation, the remaining soluble T1 did not appear to affect PFO-induced hemolysis. T2 was found to be free of T1 and intact PFO by SDS-PAGE analysis (Fig. 2). Neither peptide exhibited any hemolytic activity.

Inhibition of PFO-induced Hemolysis by T2-Two possible mechanisms by which T2 could inhibit the activity of PFO were examined. One possibility was that T2 was competing with PFO for binding sites on the erythrocyte surface. This possibility did not appear to be the explanation, since concen- trations of T2 that inhibited hemolysis did not inhibit lZ5I- labeled PFO binding to erythrocyte membranes (data not shown). Although it may be possible to compete PFO mem- brane binding with high levels of T2, in the experiments presented here competition for binding sites could not account for the observed inhibition of PFO by T2. The alternative

mechanism of inhibition was that T2 complexed with PFO and formed a lytically inactive aggregate on the membrane.

The inhibitory concentrations of T2 were determined by mixing various molar ratios of T2 and PFO and then injecting these mixtures into a solution of erythrocytes. It was found that inhibition by T2 was dose-dependent (Fig. 3). A 4.7, 9.4, or 18.8 M excess of T2 over PFO (9.6 X lo-" mol) resulted in approximately 73,80, and 87% reductions, respectively, in the hemolytic activity of PFO based on the time required for 50% hemolysis. Temperature-shift experiments were performed to determined if the order of addition of T2 and PFO to eryth- rocytes altered the inhibitory effect of T2. Initially, T2 was bound to the erythrocytes before PFO to determine whether T2 could bind independently of PFO and then subsequently inhibit PFO-induced hemolysis (Fig. 4). In the second exper- iment the order of addition was reversed (Fig. 4). Little difference was observed in the inhibition of PFO by T2 in either experiment. In both cases the hemolytic activity of PFO was reduced approximately 67% in the presence of T2. These data suggested that T2 could bind to the membranes of erythrocytes independently of PFO and that T2 distributed in the membrane could inhibit the hemolytic activity a t a step

L 1.20 E.;; 2 g 1.00 V J Z I

m ? - 0.80

5 2 0.60

x L 0.40 .-

r P

.F rn = 0.20

0.00 0 50 100 150 200 250 300 350 400

Time ( 5 )

FIG. 3. Inhibition of PFO-dependent hemolysis of erythro- cytes by peptide T2. Various ratios of T2 to PFO were evaluated by right angle scatter analysis for T2-dependent inhibition of PFO induced hemolysis of erythrocytes. The PFO and T2 were mixed prior to injection into a stirred solution of erythrocytes 25 s after the initiation of the scan. Hemolysis was followed by the changes in the right angle scatter of the erythrocytes. A , 500 ng (9.6 X 10"' moles) of PFO; R, PFO with a 4.7 M excess of T2; C, PFO with a 9.4 M excess of T2; D, PFO with an 18.8 M excess of T2.

0.0 0 200 400 600 800 1000

Time (s)

FIG. 4. Temperature-shift analysis of the T2 inhibition of PFO-dependent hemolysis of erythrocytes. The dependence of the T2 inhibition of PFO on the order of binding of PFO and T2 to erythrocytes was examined by temperature-shift analysis. Each ex- periment was done with its own uninhibited control (A, PFO alone). A. The T2-inhibited hemolytic curves are designated by the letter R. The solid lines indicate the matched data for the experiment in which T 2 was added before PFO, and the dashed line indicates the matched data in which PFO was added before T2.

12452 Membrane Binding and Aggregation of a &Toxin Peptide

after PFO binding to the membrane. Little change in the inhibitory activity of T2 was observed

whether it was added simultaneously with PFO to erythro- cytes or prebound for up to 10 min to the erythrocytes at 23 "C (data not shown). This suggested that T2 did not lose its ability to interact with PFO under these conditions, as would have been expected if T2 had undergone irreversible self-aggregation when bound to the membrane. If T2 had been capable of aggregation with itself, then the monomer to poly- mer conversion should have dramatically decreased its ability to inhibit PFO.

Formation of Aggregation Complexes of T2 and PFO-The hemolytic analysis indicated that T2 inhibited the hemolytic activity of PFO after the PFO was bound to the erythrocyte membrane, which suggested an effect on the conversion of PFO monomers to polymers. This possibility was examined using an aggregation assay based on FET between fluoro- phore-labeled PFO molecules. We have previously shown that aggregation of PFO on the erythrocyte membrane can be monitored by FET (Harris et al., 1.991) when the appropriate fluorescent probes are present on the PFO molecules. FET occurs only during membrane-dependent PFO aggregation which brings the donor !nd acceptor fluorophores within the Forster distance of 40 A for these fluorophores (Fairclough and Cantor, 1978) (shown schematically in Fig. 7A). FET between the donor (PFOD) and acceptor (PFOA) pairs can be eliminated by the addition of unlabeled PFO, which effectively separates the donor-acceptor pairs in the aggregation complex by competing with PFOA for aggregation sites on PFOD (shown schematically in Fig. 7B). Since the Forster di2tance for the fluorescein-tetramethylrhodamine pair is 40 A, the intercalation of 1-2 unlabeled PFO molecules between the PFOD and PFOA abolishes FET (Harris et al., 1991). If T2 inhibited hemolysis by complexing with PFO, as was sug- gested by the hemolytic data, then T2 should be capable of decreasing FET between the donor and acceptor molecules by one or both mechanisms shown in Fig. 7, C or D.

When T2 was prebound to erythrocyte membranes at 23 "C prior to mixing the membranes with a solution of PFOD and PFOA, energy transfer between the donor and acceptor was inhibited by T2 in a dose-dependent manner (Fig. 5). No significant differences were observed if T2 was added simul- taneously with PFO (data not shown), which was expected since the degree of inhibition of PFO-induced hemolysis was not significantly affected by prebinding T2 before PFO to the erythrocytes. If peptide T1, which had no effect on the he- molytic activity of PFO, was added at a 40 M excess over PFO, no effect on FET was observed (data not shown).

FET was also used to examine the question of whether T2 could self-aggregate, since the hemolytic analysis indicated that T2 did not self-aggregate. If true, then preincubation of T 2 with the membranes for an extended period of time should not change its inhibitory activity by the conversion of T2 monomers to polymers. However, if T2 was slowly self-aggre- gating then the conversion of monomeric T2 to polymeric T2 should significantly decrease its inhibitory effect on aggrega- tion-dependent FET between donor and acceptor PFO mole- cules. It was necessary to initially test the hypothesis by using unlabeled PFO to inhibit FET between donor and acceptor- labeled PFO. The basis of the control experiments is shown schematically in Fig. 7, A and B, which shows that if unlabeled PFO is included with the donor and acceptor PFO, then it can effectively separate both and abolish FET by increasing the mean distance separating the fluorophores. Preincubating the PFO with the membranes before mixing the membranes with the PFOD and PFOA should eliminate the inhibitory

- o . o o ~ ' ' " ' " ' ' ' " " " " ' '

500 520 540 560 580 600

Wavelength (nm)

FIG. 5 . Inhibition of fluorescence energy transfer between fluorophore-labeled PFO molecules by T2. The ability of T2 to interfere with aggregation-dependent FET was examined using fluorophore-labeled PFO. Donor (PFOD) and acceptor (PFOA) were added to a cuvette containing PBS at 30 "C. Erythrocyte membranes were preincubated for 2 min a t 23 "C with or without various amounts of T2 and then added to the donor-acceptor solution. The mixture was incubated for an additional 2 min to facilitate aggregation of the donor and acceptor. An emission scan from 500 to 600 nm was then performed. The positive FET control ( A ) is the maximum FET obtained in the presence of acceptor, and the negative FET control ( E ) shows the minimum FET obtained when the acceptor is replaced with unlabeled PFO. The scans are of the following mixtures: A, PFOD and PFOA (positive FET control); B, PFOD, PFOA and T2 at a 4.7 M excess over total PFO; C, PFOD, PFOA, and T2 at a 9.4 M excess over total PFO; and D, PFOD, PFOA, and T2 at an 18.8 M excess over total PFO; E, PFOD and 400 ng of unlabeled PFO (negative FET control).

effect of the unlabeled PFO on FET by the conversion of monomeric PFO to polymers. When unlabeled PFO was added simultaneously with PFOD and PFOA to the membranes the unlabeled PFO completely inhibited FET between the donor and acceptor (Fig. 6A, line C). As expected, when unlabeled PFO was allowed to aggregate on the membranes for as little as 2 min prior to mixing the membranes with PFOD and PFOA there was no effect on FET (Fig. 6A, lines D and E ) . This showed that membrane aggregation of the unlabeled PFO effectively eliminated its ability to compete with donor and acceptor pairing by the conversion of PFO monomers to polymers. In contrast, when T2 was preincubated with the membranes for 10 min before mixing with PFOD and PFOA no significant change in the degree of inhibition of FET was observed (Fig. 6B, line C). These results were almost identical to those results obtained when T2 was added simultaneously to the membranes with PFOD and PFOA (Fig. 6B, line D ) . These data suggest that, in contrast to PFO, T2 did not lose its ability to inhibit FET because of self-aggregation on the membrane.

DISCUSSION

The membrane binding site and at least one aggregation site appear to be present on the carboxyl-terminal fragment of PFO, termed T2, which represents approximately 40% of the PFO molecule. The data shown here and by others (Hugo et al., 1986) indicates that these toxins bind to the membrane and then diffuse laterally to form aggregates. Apparently, T2 can also bind to the membrane, diffuse laterally, and form a nonlytic or weakly lytic complex with PFO although it is not hemolytic by itself. This was supported by the observation that the inhibition of hemolysis by T2 was independent of the order of addition of T2 and PFO to erythrocytes at low temperature. In addition, preincubation of T2 with erythro- cytes for an extended period of time at temperatures that should have facilitated aggregation did not significantly alter its inhibitory activity. This indicated that T2 did not lose its

Membrane Binding and Aggregation of a &Toxin Peptide 12453

5 0 0 5 2 0 5 4 0 5 6 0 5 8 0 6 0 0 Wavelength (nrn)

1 .oo

0.80

0.60

0.40

0.20

-_ o . o o " ~ " " " ~ " ' ~ ' " ~

5 0 0 5 2 0 5 4 0 5 6 0 5 8 0 600 Wavelength (nm)

FIG. 6. Inhibition of FET after extended preincubation of T2 or PFO on erythrocyte ghost membranes. The ability of T2 to self-aggregate on membranes was examined by preincubating T2 on erythrocyte membranes for an extended length of time prior to adding the membranes to a mixture of donor and acceptor. A similar experiment was performed using unlabeled PFO to show that prein- cubation of the PFO effectively eliminated its ability to inhibit FET due to monomer to polymer conversion. Panel A , A, PFOI, (100 ng) and unlabeled PFO (400 ng) (negative FET control); B, same as in A except that the PFOA replaced the unlabeled PFO (positive FET control); C, same as in B except that a 1.6 M excess of PFO (over total PFOD plus PFOA) was added simultaneously to the membranes with PFOD and PFO,: D, same as C except the unlabeled PFO was preincubated on the membranes for 2 min a t 23 "C prior to mixing with the PFOL, and PFOA; E, same as D except the unlabeled PFO was preincubated with the membranes for 10 min prior to mixing with PFO" and PFOA. Panel B, A, same as line A in panel A; B, same as line B in panel A; C, T2 added simultaneously with PFOD and PFOA; D, T2 preincubated with the membranes for 10 min before mixing with PFOD and PFOA.

inhibitory activity by self-aggregation on the membrane, which supported the mechanism shown in Fig. 7C.

Complex formation between PFO and T2 and the capacity of T2 to self-aggregate was examined by the use of FET. This approach can be used to directly examine the interaction of molecules on the membrane without significantly perturbing the system (Harris et al., 1991). Since FET is proportional to l /r6 (r = distance separating donor and acceptor fluorophores) it is exquisitely sensitive to any increase in r that may result from the separation of the donor-acceptor pairs. The separa- tion of the donor-acceptor pair would necessarily result from the specific interaction of a protein with the aggregation sites on PFOD or PFOA that would prevent the juxtapositioning of the donor and acceptor pairs. Therefore, this molecule would have to be another PFO molecule, a PFO-derived peptide such as T2, or perhaps a closely related toxin similar to PFO, that could specifically interact with the aggregation sites present on PFOD and PFOA. The Forster distance (distance for 50% transfer efficiency) for theo fluorescein-rhodamine pair has been determined to be 40 A (Fairclough and Cantor, 1978), which would probably approximate the molecular diameter of

FIG. 7. Proposed models for the interaction of T2 with PFO. The schematic in panel A represents the hypothesized aggregation of PFO molecules on the erythrocyte membrane. The small stippled circle present on the PFO molecule represents covalently attached fluorescein (PFO,,), and the black circle represents covalently attached tetramethylrhodamine (PFOA). Under normal conditions in a FET assay, the PFO aggregates, which juxtaposes the fluorescein and tetramethylrhodamine probes. This has been shown (Harris et al., 1991) to result in FET from the donor (fluorescein) to the acceptor (tetramethylrhodamine). If, as in panel B, unlabeled PFO is added to the mixture, it effectively separates the donor and acceptor PFO so that FET is abolished. Shown in C is one possible mode by which T2 could inhibit FET and the lytic activity of the PFO. If T2 contains a single site for aggregation, inhibition would result when a sufficient number of PFO molecules formed dimers with T2. This would pre- sumably reduce the extent of PFO aggregation to such a degree as to inhibit hemolysis. If more than one aggregation site was retained on T2, as shown in D, then a copolymer of PFO and T2 might form, which would diminish FET and which would presumably be lytically inactive.

a PFO molecule. T2 was found to inhibit FET between PFOD and PFOA suggesting that T2 formed a specific complex with PFO that prevented the formation of donor-acceptor pairs. Also, T2 did not appear to be capable of self-aggregation on erythrocyte membranes, which suggested the presence of a single complementary aggregation site (shown schematically in Fig. 7C) rather than two complementary aggregation sites (shown schematically in Fig. 70 ) . Qualitatively this indicates that T2 inhibits hemolysis by interfering with aggregate for- mation, although the extent to which aggregate size must be reduced to result in a nonfunctional membrane lesion remains unresolved.

Any quantitative correlations between the degree of inhi- bition of FET by T2 and the extent to which T2 inhibits hemolysis must be approached with caution. The results showed that a 4.7 M excess of T2 over PFO resulted in approximately a 70% reduction in the hemolytic activity of PFO, which was accompanied by a 20% decrease in the FET. Since two different phenomena are being measured, a 1:1 correlation cannot be assumed. FET depends on the spatial arrangement and dipole orientation of the donor and acceptor fluorophores that are covalently bound to PFO, whereas he- molysis assays measure the rate of cell lysis, which is a multifaceted process.

It appeared that T2 did not demonstrate 100% activity in the formation of T2-PFO complexes and/or the affinity of a T2-PFO interaction was less than that of a PFO-PFO inter- action. This was based on the observation that a 5 M excess of T2 did not totally abolish FET between PFOo and PFOA,

12454 Membrane Binding and Aggregation of a 9-Toxin Peptide

whereas a 1.6 M excess of unlabeled PFO completely abolished the FET. Currently, the relative affinity of the T2-PFO interaction versus a PFO-PFO interaction cannot be deter- mined; however, the fact that T2 may not be 100% active was not unexpected. Reverse phase purification subjects the pro- tein or peptide to low pH ( ~ 2 . 0 ) and an organic solvent (acetonitrile), so it is possible that some of the T2 did not refold properly after removal of the acid and acetonitrile by dialysis.

Several bacterial toxins can be cleaved by various proteases into functional subunits, but this is the first example (that we know of) of a hemolysin being separated into functional subunits by proteolytic cleavage. Evidence was presented that the membrane binding site and at least one aggregation site are present on T2; however, the functional aspects of the amino-terminal domain encompassed by T1 remain unknown. T1 may contain the other domain required for aggregation and/or may be involved in membrane intercalation; however, the primary sequence of PFO does not exhibit any obvious transmembrane regions or extended hydrophobic domains (Tweten, 1988b).

REFERENCES Bhakdi, S., Tranum, J. J., and Sziegoleit, A. (1985) Infect. Immun.

47,52-60

Fairclough, R. H., and Cantor, C. R. (1978) Methods Enzymol. 48,

Geoffroy, C., Mengaud, J., Alouf, J. E., and Cossart, P. (1990) J.

Harris, R. W., Sims, P. J., and Tweten, R. K. (1991) J. Biol. Chem.

Hugo, F., Reichwein, J., Arvand, M., Kramer, S., and Bhakdi, S.

Kehoe, M. A., Miller, L., Walker, J . A,, and Boulnois, G. J. (1987)

Laemmli, U. K. (1970) Nature 227, 680-685 Mitsui, K., Sekiya, T., Nozawa, Y., and Hase, J. (1979a) Biochim.

Biophys. Acta 5 5 4 , 68-75 Mitsui, K., Sekiya, T., Okamura, S., Nozawa, Y., and Hase, J. (1979b)

Biochim. Biophys. Acta 558, 307-313 Ohno-Iwashita, Y., Iwamoto, M., Mitsui, K., Kawasaki, H., and Ando,

S. (1986) Biochemistry 25, 6048-6053 Ohno-Iwashita, Y., Iwamoto, M., Ando, S., Mitsui, K., and Iwashita,

S. (1990) Biochim. Biophys. Acta 1023,441-448 Stephen, J., and Pietrowski, R. A. (1986) in Bacterial Toxins (Cole,

J . A,, Knowles, C. J., and Schlessinger, D., eds) pp. 25-39, American Society for Microbiology, Washington, D. C.

347-379

Bacterid. 172, 7301-7305

266,6936-6941

(1986) Infect. Zmmun. 54, 641-645

Infect. Immun. 55,3228-3232

Tweten, R. K. (1988a) Infect. Zmmun. 56, 3228-3234 Tweten, R. K. (198813) Infect. Immun. 5 6 , 3235-3240 Walker, J . A., Allen, R. L., Falmagne, P., Johnson, M. K., and

Boulnois, G . J . (1987) Infect. Zmmun. 5 5 , 1184-1189