interaction of the eukaryotic pore-forming...

doi:10.1016/j.jmb.2004.12.058 J. Mol. Biol. (2005) 347, 27–39

Interaction of the Eukaryotic Pore-forming CytolysinEquinatoxin II with Model Membranes: 19F NMR Studies

Gregor Anderluh1, Andrej Razpotnik1, Zdravko Podlesek1, Peter Macek1

Frances Separovic2 and Raymond S. Norton3*

1Department of BiologyBiotechnical Faculty, Universityof Ljubljana, Vecna pot 1111000 Ljubljana, Slovenia

2School of ChemistryUniversity of MelbourneVIC 3010, Australia

3Walter and Eliza Hall Instituteof Medical Research, 1G RoyalParade, Parkville 3050Australia

0022-2836/$ - see front matter q 2005 E

Abbreviations used: DHPC, dihephosphatidylcholine; DMPC, dimyrphosphatidylcholine; DPC, dodecylEqTII, equinatoxin II; F-Trp-EqTII, efive Trp replaced with 5-F-Trp; IPTGD-galactopyranoside; PC, phosphatiphosphocholine; SM, sphingomyeliunilamellar vesicles.

E-mail address of the [email protected]

Sea anemones produce a family of 18–20 kDa proteins, the actinoporins,which lyse cells by forming pores in cell membranes. Sphingomyelin playsan important role in their lytic activity, with membranes lacking this lipidbeing largely refractory to these toxins. As a means of characterisingmembrane binding by the actinoporin equinatoxin II (EqTII), we have used19F NMR to probe the environment of Trp residues in the presence ofmicelles and bicelles. Trp was chosen as previous data from mutationalstudies and truncated analogues had identified the N-terminal helix ofEqTII and the surface aromatic cluster including tryptophan residues 112and 116 as being important for membrane interactions. The five tryptophanresidues were replaced with 5-fluorotryptophan and assigned by site-directed mutagenesis. The 19F resonance of W112 was most affected in thepresence of phospholipid micelles or bicelles, followed by W116, withfurther change induced by the addition of sphingomyelin. Althoughbinding to phosphatidylcholine is not sufficient to enable pore formation inbilayer membranes, this interaction had a greater effect on the tryptophanresidues in our studies than the subsequent interaction with sphingo-myelin. Furthermore, sphingomyelin had a direct effect on EqTII in bothmodel membranes, so its role in EqTII pore formation involves more thansimply an indirect effect mediated via bulk lipid properties. The lack ofchange in chemical shift for W149 even in the presence of sphingomyelinindicates that, at least in the model membranes studied here, interactionwith sphingomyelin was not sufficient to trigger dissociation of theN-terminal helix from the b-sandwich, which forms the bulk of the protein.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: cytolysin; actinoporin; pore formation; membrane; NMR
*Corresponding author
Introduction

The actinoporins are a family of sea anemonetoxins that function by forming pores in cellmembranes.1–3 These highly basic proteins, ofmass 18–20 kDa, display permeabilising activity inmodel lipid and cell membranes that is markedlyenhanced by the presence of sphingomyelin (SM).

lsevier Ltd. All rights reserve

xanoyl-istoyl-

phosphocholine-2H38;quinatoxin II with all, isopropyl-1-thio-b-

dylcholine; POC,n; SUV, small

ing author:

The actinoporins differ from the various classes ofbacterial pore-forming toxins4–7 in several respects:they are more potent, the pore they form does nothave a stable structure and has not yet beenvisualized directly, and they are smaller andextremely stable towards proteolytic degradation.Indeed, the potency and properties of these cytoly-sins have prompted their evaluation as the toxiccomponent of chimeric proteins targeted at tumourcells8,9 and human parasites.10 The nature of theirinteraction with lipids in bilayer membranes andthe specific role of sphingomyelin in pore formationare not understood at the molecular level.

In common with many pore-forming toxins, theactinoporins are highly water-soluble, stableproteins, and yet their only known activity is theformation of oligomeric pores in membranes,consisting of three or, more likely, fourmonomers.11–15 Moreover, tetramers of the

d.

28 Membrane Binding of Equinatoxin II

actinoporin sticholysin II have been observed insolution16 and when crystallized on lipid mono-layers.17 The resulting pores have a radius of about1 nm13,15,18,19 and are permeable to small moleculesand solutes, with the resulting osmotic imbalancepromoting cell lysis.

Sphingomyelin plays a key role in the lyticactivity of the actinoporins. The haemolytic activityof a cytolysin from Stoichactis (now Stichodactyla)helianthus was inhibited by pre-incubation with SM,and treatment of erythrocyte membranes withsphingomyelinase rendered them resistant tolysis.20,21 The haemolytic activity of the actinoporinequinatoxin II (EqTII) was also inhibited by pre-incubation with SM.22 More recent studies withsticholysins I14 and II15 on model membranesconfirmed that SM enhances lytic activity andsuggested that cholesterol may have a minorrole.15 The interaction of EqTII with largeunilamellar vesicles containing phosphatidyl-choline (PC) was reversible and did not involvemajor conformational changes,23 but the presence ofSM enabled irreversible insertion and pore for-mation, which were associated with major confor-mational changes. Some EqTII-induced leakage wasobserved from large unilamellar vesicles containingonly PC and cholesterol, again suggesting acontribution of this sterol lipid to actinoporincytolytic activity, albeit a much less significant onethan that of SM. Fluorescence studies of EqTIIbinding to lipid vesicles showed that associationwas markedly enhanced by the presence of SM.24

Equinatoxin II, the subject of this work, is a 179residue, 19.8 kDa cytolysin isolated from theMediterranean anemone Actinia equina L.25 It isessentially identical with tenebrosin-C, a cytolysinisolated from the Australian red waratah anemoneActinia tenebrosa,26 and the first of this class of toxinsto be sequenced in full;27 tenebrosin-C has a variantS177T and EqTII has a variant P81D.28 Its structure,solved in solution29,30 and by X-ray crystallo-graphy,31 consists of two short helices packedagainst opposite faces of a b-sandwich structureformed by two five-stranded b-sheets.

A number of studies have been carried out in aneffort to elucidate the molecular mechanism ofactinoporin pore formation. CD and FTIR studiesboth detected small increases in b-sheet anda-helical content at the expense of random structurein the presence of unilamellar vesicles.18,32,33 Bysite-directed mutagenesis, it was shown that at leasttwo regions of EqTII became embedded in lipidmembranes, the N-terminal region (residues 10–28)34,35 and the surface aromatic cluster includingtryptophan residues 112 and 116.34,36,37 The currentmodel of pore formation proposes that EqTII bindsto the membrane initially via the aromatic-richregion, then the N-terminal region is transferred tothe lipid membrane and, finally, across the bilayerto form a final functional pore. The structuraldetails of the final oligomeric assembly are notknown.

The perturbation by EqTII of model lipid

membranes composed of PC and SM has beeninvestigated using solid-state NMR.38,39 NMR andelectron microscopy showed the formation of anisotropic lipid phase in SM-containing bilayers withEqTII.39 The toxin enhanced slow motions in themembrane lipids and destabilized membranescontaining SM, suggesting a preferential interactionbetween the toxin and SM.

Our goal is to elucidate the structural basis for theinteraction of EqTII with lipid membranes, and therole of SM in that interaction. Here, we describe 19FNMR studies of EqTII with all five Trp residuesreplaced with 5-fluoro-Trp. Previous studies ofother membrane-binding proteins have shownthat 19F NMR is a sensitive and informative probeof protein interactions with lipid membranes,40,41

and this proves to be the case for EqTII.

Results

Expression, isolation and characterization ofF-Trp-EqTII

A Trp auxotroph strain was constructed that grewonly when Trp was present in the medium, thelowest Trp concentration required for normalgrowth being 0.4 mM (Figure 1(a)). As 5-F-Trpproved to be inhibitory for growth (Figure 1(a)),bacteria were first grown in complex LB mediumuntil they reached the middle of the exponentialphase, then transferred to M9 medium containing1% Casamino acids that lack any Trp. After onehour shaking, 5-F-Trp was added to the mediumand after another half-hour expression of EqTII wasinduced by addition of 0.4 mM (final) IPTG.

EqTII with all five Trp replaced by 5-F-Trp (F-Trp-EqTII) was isolated as described.42 The isolatedprotein was O95% pure according to SDS-PAGE(Figure 1(b)) and RP-HPLC. The RP-HLPC reten-tion times were virtually the same for EqTII andF-Trp-EqTII (data not shown). Importantly, the 19F-labelled protein was also of the expected size asdetermined by mass spectrometry (EqTII expected19,833 Da (Thr179 variant), obtained 19,831.6G2;F-Trp-EqTII, expected 19,923 Da; obtained19,922.5 Da), indicating that 5-F-Trp was incorpor-ated quantitatively at all five positions in thesequence. The same purification procedures wereused for the four Trp/Phe mutants used in makingresonance assignments.

Absorbance spectra of F-Trp-EqTII were quali-tatively and quantitatively different from those ofunlabelled EqTII (Figure 1(c)). There was a shoulderpresent at around 300 nm in F-Trp-EqTII that wasnot visible in spectra of EqTII. The A280 ratiobetween F-Trp-EqTII and unlabelled EqTII is0.96G0.01 (nZ2; GSD). It appears that F-Trp-EqTII has an increased Trp fluorescence quantumyield, the fluorescence of F-Trp-EqTII being morethan twice that of EqTII. Fluorescence was blue-shifted in both proteins relative to equimolarconcentrations of tryptophan in solution, indicating

Figure 1. Expression and characterisation of F-Trp-EqTII.(a) Growth-rate curves. Bacteria, transformed with pAG2.1(the plasmid used for expression of wild-type EqTII), weregrown in M9 medium with 100 mg/ml of ampicillin and15 mg/ml of tetracycline. Filled squares, E. coli BL21(DE3);filled circles, E. coli BL21(DE3) induced with final 0.4 mMIPTG; open squares, E. coli BL21(DE3) trpTTn10 in thepresence of 0.4 mM Trp; open circles, E. coli BL21(DE3)trpTTn10 in the presence of 0.4 mM Trp induced with(final) 0.4 mM IPTG; filled up triangles, E. coli BL21(DE3)trpTTn10 in the presence of 0.4 mM 5-F-Trp; open uptriangles, E. coli BL21(DE3) trpTTn10 in the presence of0.4 mM 5-F-Trp induced with (final) 0.4 mM IPTG; filleddown triangles, E. coliBL21(DE3) trpTTn10. (b) SDS-PAGEof F-Trp-EqTII and Trp/Phe mutants: 1 mg of protein wasapplied per lane and separated on SDS-15% PAGE gels.Gels were stained with Coomassie brilliant blue. M,molecular mass standard. Lane 1, native EqTII; 2, F-Trp-EqTII; 3, F-Trp-EqTII-W45F; 4, F-Trp-EqTII-W112F; 5, F-Trp-EqTII-W116F; 6, F-Trp-EqTII-W149F. (c) Absorption andfluorescence spectra of unlabelled and labelled EqTII.EqTII, continuous line; F-Trp-EqTII, broken line. Proteinconcentration was 0.1 mg/ml (11.5 mM) in water forabsorption spectra and 1 mM in 10 mM Tris–HCl (pH 8.0)for fluorescence spectra. Protein concentrations weremeasured by BCA assay. The spectra were measured atroom temperature. Fluorescence spectra of equimolaramino acid residues (5 mM) are shown as thinner lines(Trp, continuous line; 5-F-Trp, broken line).

Membrane Binding of Equinatoxin II 29

folding and burial of some of the Trp residues. Theshift was from 356 nm for free Trp to 339 nm forEqTII, which is in accord with published data,24,36,37

and slightly less for F-Trp-EqTII, from 357 nm forfree F-Trp to 343 nm for F-Trp-EqTII.

The activity of F-Trp-EqTII was compared withthat of EqTII in two different assays. As shown inFigure 2, both the haemolytic activity and calceinrelease from small unilamellar vesicles (SUV) ofF-Trp-EqTII were similar to that of unlabelledprotein.

These data indicate that replacement of the fiveTrp residues in EqTII with 5-F-Trp did not affect thestructure or activity of EqTII. This conclusion isfurther supported by analysis of one and two-dimensional 1H NMR spectra of F-Trp-EqTII and itsfour Phe-containing mutants recorded at 600 MHz.These spectra (not shown) differed slightly in the9.5–11 ppm region where the indole NH resonancesoccur, but this was a direct consequence of thefluorine substitution on chemical shifts rather thanan indicator of structural differences. Other regions

Figure 2. Activity of F-Trp-EqTII. (a) Haemolysis ofbovine red blood cells. The proteins were serially twofolddiluted in 100 ml of erythrocyte buffer (130 mM NaCl,20 mM Tris–HCl, pH 7.4). A portion (100 ml) of suspen-sion of washed bovine red blood cells (A630Z0.5) wasthen added and hemolysis was monitored at 630 nm for20 minutes. Open squares, EqTII; filled squares, F-Trp-EqTII (nZ4GSD). (b) Calcein release from small uni-lamellar liposomes. Calcein flourescence was excited at495 nm and emission monitored at 520 nm at 25 8C. SUVwere in 140 mM NaCl, 20 mM Tris–HCl (8.5), 1 mMEDTA. Toxin was then added to reach the desired lipid totoxin molar ratio (L/T) and stirred. Continuous line,EqTII; broken line, F-Trp-EqTII.


of the spectra that are sensitive monitors of nativestructure, such as 5–6 ppm and 0–0.7 ppm, weresimilar for EqTII and F-Trp-EqTII. Spectra of F-Trp-EqTII, F-Trp-EqTII-W112F and F-Trp-EqTII-W116Fwere very similar to one another. Differences wereevident in F-Trp-EqTII-W45F and F-Trp-EqTII-W149F, suggesting that some local structuralchanges accompanied these Trp to Phe mutations;this is discussed below in relation to 19F NMRspectral changes in the mutants. All of these F-Trp-EqTII mutants displayed activity similar to that ofF-Trp-EqTII itself. The key result, however, is thatthe structure and activities of EqTII and F-Trp-EqTIIare essentially identical.

NMR spectrum in aqueous solution

The 19F NMR spectrum of F-Trp-EqTII in aqueoussolution shows five well-resolved peaks, asshown in Figure 3(a). The peaks were assigned toindividual Trp residues using a series of four singleTrp/Phe mutants. The resolution of the five peaks

Figure 3. 19F NMR spectra of F-Trp-EqTII and Trp/Phe mutants in aqueous solution. Spectra were recordedat 470 MHz and 21 8C. (a) Wild-type F-Trp-EqTII, pH 5.1,2K scans; (b) F-Trp-EqTII-W45F, pH 6.9, 2K scans(a second spectrum recorded on this sample after furtherdialysis and pH adjustment to 5.6 was identical);(c) F-Trp-EqTII-W112F, pH 5.0, 2K scans; (d) F-Trp-EqTII-W116F, pH 5.1, 2K scans; (e) F-Trp-EqTII-W149F,pH 5.2, 6K scans

in the spectrum of wild-type F-Trp-EqTII wassufficient to enable clear assignment of the missingpeak in each Trp/Phe mutant, even though therewere slight changes in the chemical shifts ofremaining Trp resonances in some mutants. Inparticular, F-Trp-EqTII-W45F showed a downfieldshift of the W149 peak and a smaller upfield shift ofW117 (Figure 3(b)), suggesting some structuralperturbation associated with this Trp/Phe substi-tution. In F-Trp-EqTII-W116F, the W149 peak movesdownfield and W112 and W117 were slightlyaffected (Figure 3(d)), while in F-Trp-EqTII-W149Fthe W116 peak moves downfield (Figure 3(e)).Substitution of W112 did not perturb any of theother Trp resonances, consistent with its highdegree of solvent exposure as part of a surfaceloop. Despite these small secondary shifts in someTrp/Phe mutants, the assignment of the missingpeak in each case was unequivocal. Because of this,assignment of the fifth resonance, from Trp117, wasmade by elimination, without production of theW117F mutant.

The 5-F-Trp resonances in F-Trp-EqTII span achemical shift range of about 3.3 ppm. The chemicalshift of the central resonance, from W112, which isalso the sharpest, corresponds to the chemical shiftof acid-denatured F-Trp-EqTII. This is consistentwith the observation that W112 is located on ahighly solvent-accessible eight residue loop con-necting strands b5 and b6, which includes fouraromatic residues, Y108, Y110, W112 and Y113.30,31

However, there is no simple correlation betweenchemical shift and degree of solvent exposure, asthe other two exposed tryptophan residues, 116 and149, are well separated from the W112 resonance(Figure 3(a)). The closest resonance to W112 is fromW45, which is buried.30,31 Both van der Waals andelectrostatic interactions can have a significant effecton 19F chemical shifts in proteins,40,43 and it seemsthat both are important in F-Trp-EqTII.

NMR relaxation in aqueous solution

The 19F NMR relaxation times of F-Trp-EqTII aresummarised in Table 1. The T2 values wereremarkably uniform at both resonance frequenciesand, therefore, did not correlate with degree ofsolvent exposure. By contrast, the T1 values showeda significant spread, with the most solvent-exposedresidue, W112, having the longest T1 at bothfrequencies. Of the three solvent-exposed trypto-phan residues, W112 and W116 had T1 valuesclearly longer than the others, but the third, W149,was the same as the buried residues at 376 MHz andonly slightly longer at 470 MHz. Thus, there is arough, but by no means perfect, correlation betweenT1 and solvent exposure.

Interaction with micelles

Significant changes in 19F chemical shifts ofF-Trp-EqTII were observed in the presence of DPCmicelles, as illustrated in Figure 4. The W112

Table 1. 19F NMR relaxation data for Trp residues of EqTII

Trp 376 MHz 470 MHz

T1 (s) T2 (ms) T1 (s) T2 (ms)

W45 0.28 19 0.33 16W112 0.74 23 0.99 20W116 0.43 22 0.64 16W117 0.32 24 0.45 19W149 0.31 20 0.52 18

Values measured at 376 MHz are the mean of three sets of data, all recorded at 25 8C with a recycle time of 1.5 seconds; rangeswere G0.02–0.03 seconds for T1 and G2 ms for T2. Values at 470 MHz were measured at 21 8C with a recycle time of 3.5 seconds.A second set of data at 470 MHz was recorded with a more dilute sample (ca one-sixth the concentration of the first sample); the T2

values were similar to those shown in this table, as were the T1 values for W45, W117 and W149. The T1 values for W112 and W116 werelonger (1.37 seconds and 0.76 seconds, respectively).


resonance shifted upfield by 0.9 ppm, W116 shifteddownfield by 0.4 ppm and W117 shifted downfieldby 0.2 ppm (Figure 4(b)). As a result of its upfieldshift, the W112 resonance was close to that of W149,

Figure 4. 19F NMR spectra of F-Trp-EqTII in DPCmicelles with increasing amounts of SM. Spectra wererecorded at 470 MHz and 21 8C. (a) F-Trp-EqTII, pH 5.1,2K scans; (b) F-Trp-EqTII in DPC micelles. 0.3 mM F-Trp-EqTII, 123 mM DPC in 50 mM phosphate buffer at pH 6.1(note that a spectrum of a sample of DPC micelles withhalf this concentration of F-Trp-EqTII and pH 6.3 gaveidentical chemical shifts); (c) 1.3 mM SM, ratio of SM toDPC 0.01; (d) 16 mM SM; ratio 0.13; (e) 27 mM SM; ratio0.22. All spectra in DPC micelles, with or without SM,were run with 32K scans.

but was distinguishable by its sharper linewidth.These assignments were confirmed by parallelexperiments on the W149F mutant (spectra notshown).

When a small amount of SM was added theoverall appearance of the spectrum was similar, butthe intensity of the W112 peak decreased and therewas a small broad peak around K47.9 ppm(indicated in Figure 4(c)). In fact, a weak peak atthis position was visible even in the absence of SM(Figure 4(b)), and was seen also in correspondingspectra at 376 MHz of both F-Trp-EqTII and F-Trp-EqTII-W149F (not shown). Importantly, the samepattern of shifts was observed when half theconcentration of F-Trp-EqTII was used, indicatingthat the EqTII binding capacity of the DPC micelleswas not saturated at the concentration used inFigure 4. As the ratio of SM to DPC increased, thesharp W112 peak disappeared and the broaddownfield peak grew (Figure 4(d) and (e)). TheW116 resonance continued to move downfield andbecame broader, while W117 was largely unaffectedby SM. When the ratio of SM to DPC reached 1:1, allpeaks in the spectrum (not shown) became quitebroad but the distribution of peak intensities wassimilar to that in Figure 4(e) except that the W112linewidth had become more similar to the others.

A similar pattern of changes was seen at 376 MHzupon interaction of F-Trp-EqTII and F-Trp-EqTII-W149F, although the peak at K47.9 ppm wassharper than at 470 MHz. One possible explanationfor this is that there is some conformationalaveraging occurring on the millisecond timescalethat produces more broadening at the higherobservation frequency. In addition, chemical shiftanisotropy contributions to relaxation41,44 will bemore pronounced at the higher field. For thepurpose of this study we have not attempted todistinguish between these possible sources ofbroadening at 470 MHz. The spectra of F-Trp-EqTII in DPC, with and without SM, showed nolong-term time-dependence, being reproducibleover a period of weeks to months.

The effect of SM on the spectrum of F-Trp-EqTIIin these experiments is likely to be mediatedentirely by SM bound to micelles. SM is veryinsoluble in aqueous solution, and when EqTII was


mixed with a suspension of SM in phosphate bufferit gave no detectable 1H NMR signal at 600 MHz.Furthermore, previous experiments have shownthat EqTII bound to SM in aqueous solutions is notavailable to cause haemolysis.22 Thus, essentially allof the SM added to DPC micelles is likely to beassociated with the micelles. Any free SM will beinsoluble and any F-Trp-EqTII associated with itwould not be observable in our NMR experiments,as confirmed by 1H NMR (data not shown). It is alsoreasonable to assume that most of the toxin isassociated with micelles under our experimentalconditions (lipid to toxin molar ratio 500 at thehighest SM concentration). At even lower lipid totoxin ratios, the majority of the toxin was associatedwith the membranes in liposomes composed of 1:1PC/SM, with approximately 80% of EqTII36 andsticholysins I and II33 bound to small and largeunilamellar vesicles, respectively.

To determine whether the order of addition hadany influence, we mixed SM with DPC beforeadding F-Trp-EqTII. At an SM to DPC ratio similarto that shown in Figure 4(d) the W112 peak had notmoved as far downfield as when SM was added toF-Trp-EqTII already mixed with DPC. This suggeststhat the distribution of F-Trp-EqTII and SM acrossthe micelle population is affected slightly by theorder of addition, although the pattern of chemicalshift changes is the same. The toxin may induce achange in lipid micelle structure that enables easierpartition of SM into the micelle or easier access ofSM to F-Trp-EqTII, or F-Trp-EqTII may preferen-tially attract SM to micelles in which it is embedded.

An impurity peak at K42.3 ppm was evident inmost, but not all, spectra in DPC, with or withoutSM (Figure 4) and in bicelles with SM (see below).Its sharper linewidth suggests that it arose froman impurity in the lipids, rather than from nativeF-Trp-EqTII. This was confirmed by the appearanceof this peak in spectra of DPC micelles and bicellesin the absence of SM or F-Trp-EqTII. Its T1 at470 MHz was similar to that of W45 in F-Trp-EqTIIin DPC with 16 mM SM (sample as in Figure 4(d)),i.e. about 0.4 second.

Figure 5. 19F NMR spectra of F-Trp-EqTII and F-Trp-EqTII-W149F in DMPC/DHPC bicelles with or withoutSM. Spectra were recorded at 376 MHz and 30 8C exceptfor the micelle spectrum in (a), which was run at 25 8C.(a) F-Trp-EqTII in DPC micelles; (b) F-Trp-EqTII inDMPC/DHPC bicelles. 0.3 mM F-Trp-EqTII, 123 mM PC(23 mM, DMPC, 92 mM DHPC) in 50 mM phosphatebuffer, pH 6.8; (c) F-Trp-EqTII in DMPC/DHPC bicelles,plus 15 mM SM (ratio of SM to PC 0.13); (d) F-Trp-EqTII-W149F in DMPC/DHPC bicelles; (e) F-Trp-EqTII-W149Fin DMPC/DHPC bicelles, plus 15 mM SM (ratio of SM toPC 0.13). With time, the W112 peak narrowed andappeared to increase in intensity slightly, but did notchange its chemical shift. All spectra in bicelles were runwith 30–50K scans.

Interaction with bicelles

Binary bilayered mixed micelles, known asbicelles, represent a model membrane systemmore similar to natural biomembranes than DPCmicelles. They are considered to be discoidalbilayers typically composed of long-chaindimyristoylphosphatidylcholine (DMPC) localizedin the planar section, and short-chain dihexanoyl-phosphatidylcholine (DHPC) stabilizing the torusof the discs.45–49 Because of the diamagneticsusceptibility of their phospholipid components,bicelles can spontaneously align in the magneticfield with the bilayer normal perpendicular to thedirection of the field, but this alignment depends onthe temperature, concentration, and long-chain toshort-chain lipid ratio (q). In this study, we haveused bicelles with a small q ratio (0.25), which are

expected to be cylindrical in shape, with a heightaround 45 A and a radius of about 25 A.47 Althoughthey are larger than micelles they undergo suffi-ciently rapid tumbling in solution to allow studiesof bicelle-associated peptides and proteins bysolution-state NMR.48,49

The changes in 19F chemical shift in thepresence of DMPC/DHPC bicelles were quali-tatively similar to those in the presence of DPCmicelles, but there were some quantitativedifferences. Comparison of 376 MHz spectra ofF-Trp-EqTII in DPC micelles (Figure 5(a)) andDMPC/DHPC bicelles (Figure 5(b)) showed thatthe chemical shift changes for W112 and W116followed the same pattern, but were larger. Thus,W112 moved slightly further upfield in bicelles such

Figure 6. [2-13C]Trp-labelled EqTII. This two-dimen-sional 13C{1H}-HSQC spectrum of [2-13C]Trp-labelledEqTII in water at pH 4.9 was recorded on a BrukerDRX-600 at 25 8C (total spectral acquisition time fourhours). Assignments are based on Zhang et al.29 Becauseonly the 1H resonances were assigned for W116 and W149and their 1H chemical shifts are very similar, the assign-ments of their 1H–13C cross-peaks (*) are tentative.


that it overlapped with W149, and W116 movedfurther downfield. Even W117, which moved onlyslightly in micelles, shifted further in bicelles. Itappears that the interactions of F-Trp-EqTII withmicelles and bicelles are very similar, but theassociation with lipids is somewhat tighter inbicelles. Interestingly, the small peak atK47.9 ppm, which was present in spectra ofF-Trp-EqTII and F-Trp-EqTII-W149F in DPCmicelles and became stronger as SM was added,was not observed in spectra of F-Trp-EqTII or F-Trp-EqTII-W149F in bicelles in the absence of SM(Figure 5(b) and (d)).

Addition of SM to bicelles caused changesqualitatively similar to those seen in micelles, butagain there were quantitative differences. W112 didnot move as far downfield in the presence of similaramounts of SM, and W116, which had alreadymoved further downfield in bicelles than micelles,did not move much further. Addition of more SMshifted the W112 peak further downfield (notshown). The same trends were evident for F-Trp-EqTII-W149F (Figure 5(d) and (e)) but W112 movedfurther in F-Trp-EqTII-W149F (Figure 5(e)) than F-Trp-EqTII (Figure 5(c)) even though the amounts ofPC lipid and the ratios of SM were the same. It ispossible that the Trp149/Phe mutation altered theinteraction with SM, but the amount of protein wasmuch lower in the case of F-Trp-EqTII-W149F(hence the much lower signal-to-noise ratio in thespectra), and perhaps the effect of SM depended notonly on the SM to PC ratio but also the ratio of SM toEqTII (unlike the effect of PC, which was clearlysaturating even at the highest F-Trp-EqTII concen-trations added). An effect of SM that was dependenton the SM to F-Trp-EqTII ratio would be moreconsistent with a specific interaction with theprotein than with a bulk effect on the lipid. Inorder to test this, we repeated the experiment usinga concentration of F-Trp-EqTII even lower than thatof F-Trp-EqTII-W149F in Figure 5. With much lesstoxin, the same ratio of SM to PC caused a largershift in W112, very similar to that in Figure 5(e)(Figure S2 in Supplementary Material). By contrast,in DPC micelles the 376 MHz spectra of F-Trp-EqTIIand F-Trp-EqTII-W149F had similar chemical shiftsin the presence of SM, although the W112 resonancewas considerably sharper for the more dilutemutant.

It appears that EqTII associated more tightly withbicelles than micelles and, perhaps because of this,SM was less effective at causing subsequentchanges in the EqTII–lipid interaction, even thoughthe changes were qualitatively similar. This isconsistent with previous reports of actinoporinpreference for binding to and permeabilisation ofSM-containing large unilamellar vesicles, whichhave a lower curvature than SUV.14,50 It is alsopossible that SM may partition differently intobicelles because of their less curved surface.

NMR spectra of the micelles and bicelles did notreveal any change in the size of the phospholipidstructures upon addition of EqTII or SM, with both

the line-width and chemical shift of the phospho-lipid 31P resonance being unaffected.51 Some broad-ening was seen in the 31P spectra of 1:1 DPC to SMmicelles, but even here the lineshape was consistentwith small, isotropically tumbling “micelles”.

19F NMR relaxation in lipid micelles

19F NMR spin-lattice relaxation times weremeasured at 376 MHz for F-Trp-EqTII in DPCmicelles in the presence of SM at a ratio of 0.13,similar to that shown in Figure 4(d) but with asharper W112 peak because of the lower fieldstrength. The T1 values were: W45 0.41 second,W112 0.74 second, W116 0.74 second, W1170.53 second, W149 0.62 second. The least solvent-exposed Trp in the native structure, W45, was stillthe most rapidly relaxing, but W112 was now muchmore similar to the other three than for the aqueousprotein (Table 1), where it was almost twice as longas the others. These data suggest that W112 is nolonger as highly solvent-exposed, presumablyreflecting its interaction with micellar lipids.

Comparison with 13C NMR

The sensitivity of the 5-F-Trp resonances of F-Trp-EqTII to its interaction with model membranesprompted a comparison with 13C NMR studies of13C-labelled Trp in the same protein. The aromaticresonances of Trp are very sensitive to theirenvironment within native proteins,52 and partialassignments for the Trp aromatic 1H and 13Cresonances are available.29 [2-13C]Trp was incorpor-ated quantitatively using the same expressionsystem as for 5-F-Trp, and the labelled EqTII wasfully active (Supplementary Material). A 2D


13C{1H}-HSQC spectrum (Figure 6) showed that thefive resonances were well dispersed in both the 13Cand 1H dimensions. The sharpest resonance in the1H dimension is that of W112, reflecting its highdegree of solvent exposure. When this sample wasmixed with DPC micelles in the absence of SM,good-quality spectra could still be obtained butonly with significantly longer spectral acquisitiontimes. However, few changes in chemical shift wereobserved, the only significant shift being that of the13C resonance of W116, from 129.7 ppm to130.1 ppm (Table S3, Supplementary Material).When SM was added to a molar ratio of 0.14, thespectral quality decreased to the point where onlythe two sharpest resonances (W112 and W149) wereobservable in the HSQC at 25 8C; W112 showed asignificant change in chemical shift with thisaddition of SM, moving from 127.2 ppm to127.6 ppm. At 35 8C all five sites were observableand there was significant broadening of someresonances, but the spectral quality was poor evenafter 24 hours spectral acquisition on a cryoprobe-equipped spectrometer. When the SM ratio wasincreased to 0.27, only the W112 resonance wasvisible even after 40 hours of spectral acquisition.Thus, whatever spectral changes may haveoccurred upon lipid binding, they wereconsiderably more difficult to observe than thecorresponding 19F NMR spectral changes.

Discussion

By creating an E. coli strain auxotrophic fortryptophan we have been able to produce ananalogue of EqTII with all five Trp residuesreplaced by 5-F-Trp. This enabled 19F NMR studiesof the binding of EqTII to lipid membranes, whichoffer the advantages of high NMR detectionsensitivity (83% that of 1H), good peak resolutionand no natural background. Trp residues werelabelled because previous studies of EqTII inter-actions with membranes by means of fluorescenceand mutagenesis had suggested that two regions ofEqTII were particularly important, namely theN-terminal helical region and a surface aromaticcluster involving Y108, W112, Y113, W116, Y133,Y137, and Y138.34,35,37 In addition, Trp residues playan important role in orienting peptides and proteinsin the interfacial region of lipid membranes.53 The 5position of the Trp ring was chosen as 5-F-Trpresonances exhibit greater dispersion than 4 or 6-F-Trp (7 ppm versus 6 ppm or 3 ppm, respectively)40

and show less broadening at higher magnetic fieldsdue to chemical shift anisotropy.41 A significantadvantage of 19F NMR, which becomes importantin interpreting our data, is its ability to observe allfive Trp residues, rather than a subset that, forexample, gives rise to fluorescence spectra.24,37 Thelabelled protein, F-Trp-EqTII, was fully active inhaemolytic and calcein release assays and had 1HNMR spectra similar to those of native EqTII,although some differences were evident in its UV

absorbance and fluorescence spectra. The 19Fresonances of F-Trp-EqTII were well resolved, andwere readily assigned by a series of single-siteTrp/Phe mutations.

Compared with 19F labelling, 13C labelling offersthe potential advantages that 13C is likely to be morereadily incorporated without the risk of cell toxicityand to cause less perturbation of the local structure.Moreover, if resonance assignments have beenmade in the course of other NMR studies of thetarget protein, there should be no need to resort tomutational studies for resonance assignments. Onthe other hand, 13C-labelled amino acid residues areconsiderably more expensive than their 19F-labelledcounterparts. In the current case, 19F labellingproved to be far more informative, with largerspectral perturbations observed upon interactionwith lipids and better-quality spectra obtained inthe presence of model membranes. The additionaleffort required to assign the resonances via single-site mutants was minimal and, in any case, themutants proved to be valuable in their own right forprobing lipid interactions.

In the presence of DPC micelles, W112 showedthe largest shift (upfield), while W116 moveddownfield and W117 moved slightly downfield.The downfield shift of W116 suggests that it hasmoved to a less polar environment upon membranebinding40,43 whereas the upfield shift of W112 mayreflect its proximity to the zwitterionic headgroupsof the lipids. To test this possibility, we addedphosphocholine (POC) to F-Trp-EqTII. By analogywith sticholysin II,17 POC should bind in thevicinity of W112. At a molar ratio of 1:1, POCcaused a slight (0.04 ppm) upfield shift of the W112resonance while not affecting other resonances, butat a ratio of 10:1 aggregation occurred and spectracould not be obtained. These preliminaryobservations suggest that interaction with thephosphatidylcholine head group may contributeto the upfield shift of W112 in micelles and bicelles,but further work is required to establish the exactmagnitude of this contribution.

The addition of SM shifted W116 even furtherdownfield, but caused a shift reversal of W112,which now moved downfield. This shows that SMeither interacts directly with W112 in themembrane-bound state of EqTII or modifies theenvironment of W112 by changing the interaction ofEqTII with the membrane or the conformation ofmembrane-bound EqTII. This is the first demon-stration that SM directly affects EqTII, as opposed tosimply stabilising the membrane-bound form of thetoxin. Recently, Barlic et al.54 concluded that,although specific interactions between actinoporinsand SM could not be excluded,13,14 it was thecoexistence of lipid phases that seemed to modulatethe interaction of the toxin with the membrane. Ourresults show that SM does have a direct effect onEqTII in two different model membranes, so its rolein EqTII pore formation involves more than what-ever effects it has on bulk lipid bilayer properties.Evidence of an interaction between Trp residues


and SM has been reported recently for the unrelatedhaemolytic toxin lysenin.55

The fact that significant shifts in the resonancesfrom W112 and W116 are observed upon binding ofF-Trp-EqTII to DPC micelles, and even more so,DMPC/DHPC bicelles, shows that EqTII makessignificant interactions with PC even in the absenceof SM. Although binding to PC is not sufficient toenable pore formation,13,14,23,54 our results showthat the interaction has as much effect on thetryptophan residues of EqTII as the subsequentinteraction with SM. They also show that EqTII doesnot undergo a major conformational changeupon binding to membranes, and thus rule outthe possibility that the pore could be formed byb-barrel structures of the type seen for bacterialcytolysins.56

The lack of effect on the W149 19F resonance inmicelles and bicelles, with or without SM, proves tobe very valuable in defining how EqTII binds tomembranes and the extent of any conformational

changes accompanying that binding. F-Trp-EqTIIcannot be lying on the surface of the bicelles ormicelles in such a way that the hydrophobic surfacethat is normally protected from solvent by theN-terminal helix (i.e. strands b1 (residues 32–40)and b3 (residues 69–75), and to a lesser extent b10(residues 170–177)) is associated with the lipidsurface, because this would bring W149 into closeproximity to the lipid and affect its chemical shift(Figure 7(a)). In the model of the tetrameric poreproposed by Mancheno et al.17 on the basis of EMimages of two-dimensional crystals of sticholysin II,the a2 helix (residues 129–134 in EqTII) is intimatelyassociated with the membrane (Figure 7 ofMancheno et al.). It is possible to orient EqTII suchthat this helix, as well as W112 and W116, are closeto the lipid (Figure 7(a)), and this may be a goodmodel of the micelle-bound and bicelle-boundEqTII in our experiments. In this orientation W112and W116 are intimately associated with themembrane, W117 is the next closest, and W45 and

Figure 7. (a) Possible models ofthe orientation of EqTII at themembrane surface. The magni-tudes of the 5-F-Trp chemical shiftchanges upon membrane bindingare consistent with the left-handand middle, but not the right-hand,orientations. The middle orien-tation provides the best fit of thethree to the observed shifts. In eachcase the closest-to-average struc-ture of EqTII in solution29 is shown,with the Trp side-chains colouredas follows: W112 red, W116magenta, W117 gold, W149 green,W45 blue. (b) Close-up view show-ing the proximity of W149 (green)to F16 (violet). Given the sensitivityof 19F chemical shifts to theirenvironment, the lack of change inthe W149 chemical shift uponbinding to micelles or bicellesindicates that the N-terminal helixhas not dissociated from the b-sheet, even in the presence of SM.


W149 are far enough away not to be affected. The N-terminal helix is still associated with the protein asin native EqTII and would have to undergo a largerearrangement in order to intercalate into the lipidbilayer. However, the lack of effect of lipid bindingon W149 suggests that such a change has not yetoccurred in micelles or bicelles. W149 is in closeproximity to the side-chain of Phe16 in the nativestructure (Figure 7(b); the distance between W149H-5 (Hz3) and F16 Cg is 5.4 A in the closest-to-average solution structure30) and a change in thisinteraction would be expected to change its 19Fchemical shift.

The observation that the N-terminal dissociationdoes not appear to occur even in the presence of SMimplies that the interaction with SM per se is notsufficient to trigger this conformational change, atleast in the model membranes investigated here. Itis possible that a membrane bilayer is required forN-terminal dissociation to occur. Alternatively,some additional interaction with the membrane,and/or with other molecules of EqTII that willparticipate in formation of the tetrameric pore, maybe required. In the model of the tetrameric poreproposed by Mancheno et al.,17 the hydrophobicsurface normally protected from solvent by theN-terminal helix appears to be exposed, whichwould be energetically unfavourable. Possibly thehydrophobic surfaces exposed upon lipid bindingcause these protein surfaces to aggregate andpromote pore formation.

In conclusion, we have shown that EqTII under-goes significant interactions with model mem-branes in the absence of SM, and that the additionof SM causes further changes in the membrane-bound protein, particularly in the vicinity of W112.The significant 19F chemical shift changes observedfor some Trp residues, coupled with the absence ofsuch changes for others, place significant restric-tions on putative models of membrane-boundEqTII. In particular, the absence of change forW149 indicates that the interaction with SM is notsufficient to trigger dissociation of the N-terminalhelix from the b-sandwich, at least in the modelmembranes studied here. Our results emphasizethe value of 19F NMR in studies of the interactionsof proteins with membranes, with good-qualityspectra attainable in relatively short acquisitiontimes and chemical shifts that are exquisitelysensitive to the local environment. Moreover, thistechnique is applicable in solid-state NMR experi-ments on lipid bilayers,57 and further studies willnow proceed in that direction with the aim ofcharacterising the subsequent steps towards for-mation of a tetrameric pore in bilayer membranes.

Materials and Methods

Creation of Trp auxotroph strain

Standard molecular biology protocols58 were used tocreate a Trp auxotroph strain (E. coli BL21(DE3)

trpTTn10). It was constructed by transduction of E. coliBL21(DE3) with P1 phage prepared on E. coli strain Y1090(trpTTn10), which harbours transposon Tn10, bearingtetracycline resistance, inserted within the trp operon. Theresulting strain, selected as tetracycline-resistant trans-ductant, was not able to grow in minimal media, incontrast to the original E. coli BL21(DE3) strain(Figure 1(a)). It grew only when Trp was present in themedium, the lowest Trp concentration required fornormal growth being 0.4 mM. Bacteria were growninitially in minimal medium with the addition of0.4 mM 5-fluoro-DL-tryptophan (5-F-Trp; Sigma), but, as5-F-Trp proved to be inhibitory for growth (Figure 1(a)),we developed a protocol to bypass this limitation.Bacteria were grown overnight in LB medium sup-plemented with 100 mg/ml of ampicillin and 15 mg/mlof tetracycline at 37 8C with shaking. In the morning,500 ml of LB with added M9 salts and antibiotics wasinoculated with 20 ml of the overnight culture and cellswere grown under the same conditions until they reachedmid-exponential phase (A600 ca 1.5), then pelleted bycentrifugation for five minutes at 4400g at room tempera-ture. After washing with M9 salts they were transferred toM9 medium with 2% (w/v) glucose, 1 mg/ml of biotin,1 mg/ml of thiamine, and 1% Casamino acids that lackTrp. After one hour shaking, 5-F-Trp was added to themedium to a final concentration of 0.4 mM and, afteranother 30 minutes, expression of EqTII was induced byaddition of 0.4 mM (final) IPTG. Cells were incubated foran additional four hours, centrifuged at 4400g and 4 8Cand then frozen.

Isolation and characterization of F-Trp-labelled EqTIIand mutants

F-Trp-EqTII and the mutants W45F, W112F, W116F andW149F were expressed in the Trp auxotroph E. coliBL21(DE3) strain as described above. The plasmids forthe mutants were prepared as described.36 Recombinantproteins were isolated from the bacterial cytoplasm asdescribed.42 Native EqtII was isolated from Actinia equinaas described.25 The absorption coefficient at 280 nm forwild-type EqtII was taken as 2.21.36 The concentration ofF-Trp labelled proteins was determined by BCA ProteinAssay (Pierce), using bovine serum albumin (Merck) forthe calibration curve.

Electrospray ionization MS analysis was performedusing a high resolution magnetic-sector AutospecQ massspectrometer (Micromass). The protein samples wereintroduced into an electrospray nebulizer at a flow rate of10 ml/minute with a syringe pump. The spectra wereobtained by scanning from m/z 2500 to 200 at 10 sec-onds/scan. Calibration was performed by sodium iodidecluster ions.

Haemolysis of bovine red blood cells was measuredwith a microplate reader on microtiter plates at roomtemperature (Dynex, Germany). The proteins wereserially twofold diluted in 100 ml of erythrocyte buffer(130 mM NaCl, 20 mM Tris–HCl, pH 7.4); 100 ml of asuspension of washed bovine red blood cells (A630 0.5)was then added and haemolysis monitored at 630 nm for20 minutes.

Calcein release was measured from SUV. SUV loadedwith calcein were prepared by sonication, with excesscalcein removed on a small Sephadex G50 column.Calcein flourescence was excited at 495 nm and emissionmonitored at 520 nm at 25 8C. SUV were in 140 mM NaCl,20 mM Tris–HCl (pH 8.5), 1 mM EDTA. Toxin was thenadded to reach the desired lipid to toxin ratio (L/T) and


stirred. The permeabilisation (in %) was compared to themaximal, obtained at the end of the assay by the additionof (final) 2 mM Triton X-100.

Absorbance spectra were measured on a Shimadzu UV-2101PC spectrophotometer. Fluorescence spectra weremeasured with a Jasco FP-750 spectrofluorimeter at 25 8C.The excitation wavelength was set to 295 nm andemission was scanned from 300–400 nm. Excitation andemission slits were set to 5 nm.

NMR spectroscopy

Samples were prepared for NMR by dissolving solidprotein in aqueous solution then adjusting the pH at roomtemperature; F-Trp-EqTII and its Trp/Phe mutants werehighly water-soluble. Micelle and bicelle solutions wereprepared by dissolving the appropriate lipids in water.SM is essentially insoluble in aqueous solution, butdissolved in the micelle and bicelle solutions with gentleagitation at room temperature over a period of 0.5–1 hour.Protein concentrations were typically in the range 0.8–2 mM for aqueous solutions and 0.1–0.4 mM in micellesand bicelles. DMPC, DHPC, DPC and brain SM werepurchased from Avanti Polar Lipids, Inc. (Alabaster, AL).POC chloride calcium salt was purchased from Sigma-Aldrich (St. Louis, MO). All lipids were used withoutfurther purification.

19F NMR spectra were recorded at 470 MHz on aBruker Avance500 spectrometer equipped with a dedi-cated 1H probe de-tuned to 19F, or at 376 MHz on a VarianInova 400 (Palo Alto, CA) equipped with a triple-resonance gradient probe. Probe temperatures were21 8C and 25 8C at 470 MHz and 376 MHz, respectively,for spectra of F-Trp-EqTII in aqueous solution andmicelles, and 30 8C for spectra in bicelles at 376 MHz.Recycle times (acquisition plus delay) were 1.5 secondsand 1.7 seconds, respectively. One and two-dimensional1H NMR spectra were recorded at 25 8C on a Bruker DRX-600 equipped with a triple-resonance gradient probe. 31Pspectra were obtained on a Varian Inova 400 operating at162 MHz with proton decoupling; micelles were run at25 8C and bicelles at 30 8C.

Spin-lattice relaxation times (T1) were measured usingthe inversion recovery (p-t-p/2) pulse sequence, with tvalues in the ranges 1 ms to 1.5 seconds at 376 MHz and10 ms to 2.0 seconds at 470 MHz. Spin-spin relaxationtime (T2) measurements were made using a spin-echoexperiment (p/2-t-p) with t values in the ranges 0.1–30 ms at 376 MHz and 2–100 ms at 470 MHz. Recycletimes (acquisition plus delay) were 1.5 seconds and 3.5seconds at 376 MHz and 470 MHz, respectively.

Spectra were processed with XWINNMR (version 3.5,Bruker) or VMR (version 6.1C, Varian). Unless statedotherwise, spectra were processed using an exponentialwindow function with a line-broadening of 15 Hz.Structural Figures were created using MOLMOL.59

Acknowledgements

This work was supported, in part, by grants fromthe Australian Research Council (to F.S.) and theSlovenian Ministry for Education, Science and Sport(to G.A., Z.P. and P.M.). G.A. is recipient of aWellcome Trust International Research Develop-ment Award. We are grateful to David Bourne for

access to the Avance500, Aphrodite Anastasiadis-Pool for assistance with some spectra, Chunxiao C.Wang for assistance with the Figures and BogdanKralj for MS analysis.

Supplementary Data

Supplementary data associated with thisarticle can be found, in the online version, atdoi:10.1016/j.jmb.2004.12.058

References

1. Turk, T. (1991). Cytolytic toxins from sea anemones.J. Toxicol. Toxin Rev. 10, 223–262.

2. Macek, P. (1992). Polypeptide cytolytic toxins from seaanemones (Actiniaria). FEMS Microbiol. Immunol. 105,121–130.

3. Anderluh, G. & Macek, P. (2002). Cytolytic peptideand protein toxins from sea anemones (Anthozoa:Actiniaria). Toxicon, 40, 111–124.

4. Prevost, G., Mourey, L., Colin, D. A. & Menestrina, G.(2001). Staphylococcal pore-forming toxins. Curr. Top.Microbiol. Immunol. 257, 53–83.

5. Fivaz, M., Abrami, L., Tsitrin, Y. & van der Goot, F. G.(2001). Aerolysin from Aeromonas hydrophila andrelated toxins. Curr. Top. Microbiol. Immunol. 257,35–52.

6. Lakey, J. H. & Slatin, S. L. (2001). Pore-formingcolicins and their relatives. Curr. Top. Microbiol.Immunol. 257, 131–161.

7. Billington, S. J., Jost, B. H. & Songer, J. G. (2000). Thiol-activated cytolysins: structure, function and role inpathogenesis. FEMS Microbiol. Letters, 182, 197–205.

8. Avila, A. D., Mateo de Acosta, C. & Lage, A. (1989).A carcinoembryonic antigen-directed immunotoxinbuilt by linking a monoclonal antibody to a hemolytictoxin. Int. J. Cancer, 43, 926–929.

9. Pederzolli, C., Belmonte, G., Dalla Serra, M., Macek, P.& Menestrina, G. (1995). Biochemical and cytotoxicproperties of conjugates of transferrin with equina-toxin II, a cytolysin from a sea anemone. Bioconjug.Chem. 6, 166–173.

10. Tejuca, M., Anderluh, G., Macek, P., Marcet, R., Torres,D., Sarracent, J. et al. (1999). Antiparasite activity ofsea-anemone cytolysins on Giardia duodenalis andspecific targeting with anti-Giardia antibodies. Int.J. Parasitol. 29, 489–498.

11. Michaels, D. W. (1979). Membrane damage by a toxinfrom the sea anemone Stoichactis helianthus I. For-mation of transmembrane channels in lipid bilayers.Biochim. Biophys. Acta, 555, 67–78.

12. Varanda, W. & Finkelstein, A. (1980). Ion and none-lectrolyte permeability properties of channels formedin planar lipid bilayer membranes by the cytolytictoxin from the sea anemone, Stoichactis helianthus.J. Membr. Biol. 55, 203–211.

13. Belmonte, G., Pederzolli, C., Macek, P. & Menestrina,G. (1993). Pore formation by the sea anemonecytolysin equinatoxin II in red blood cells andmodel lipid membranes. J. Membr. Biol. 131, 11–22.

14. Tejuca, M., Dalla Serra, M., Ferreras, M., Lanio, M. E.& Menestrina, G. (1996). Mechanism of membranepermeabilization by sticholysin I, a cytolysin isolatedfrom venom of the sea anemone Stichodactylahelianthus. Biochemistry, 35, 14947–14957.

http://dx.doi.org/doi:10.1016/j.jmb.2004.12.058


15. de los Rios, V., Mancheno, J. M., Lanio, M. E.,Onaderra, M. & Gavilanes, J. G. (1998). Mechanismof the leakage induced on lipid model membranes bythe hemolytic protein sticholysin II from the seaanemone Stichodactyla helianthus. Eur. J. Biochem. 252,284–289.

16. de los Rios, V., Mancheno, J. M., Martinez del Pozo,A., Alfonso, C., Rivas, G., Onaderra, M. & Gavilanes,J. G. (1999). Sticholysin II, a cytolysin from the seaanemone Stichodactyla helianthus, is a monomer-tetramer associating protein. FEBS Letters, 455, 27–30.

17. Mancheno, J. M., Martin-Benito, J., Martinez-Ripoll,M., Gavilanes, J. G. & Hermoso, J. A. (2003). Crystaland electron microscopy structures of sticholysin IIactinoporin reveal insights into the mechanism ofmembrane pore formation. Structure (Camb.), 11,1319–1328.

18. Anderluh, G., Barlic, A., Potrich, C., Macek, P. &Menestrina, G. (2000). Lysine 77 is a key residue inaggregation of equinatoxin II, a pore-forming toxinfrom sea anemone Actinia equina. J. Membr. Biol. 173,47–55.

19. Tejuca, M., Dalla Serra, M., Potrich, C., Alvarez, C. &Menestrina, G. (2001). Sizing the radius of the poreformed in erythrocytes and lipid vesicles by the toxinsticholysin I from the sea anemone Stichodactylahelianthus. J. Membr. Biol. 183, 125–135.

20. Bernheimer, A. W. & Avigad, L. (1976). Properties of atoxin from the sea anemone Stoichactis helianthus,including specific binding to sphingomyelin. Proc.Natl Acad. Sci. USA, 73, 467–471.

21. Linder, R., Bernheimer, A. W. & Kim, K. S. (1977).Interaction between sphingomyelin and a cytolysinfrom the sea anemone Stoichactis helianthus. Biochim.Biophys. Acta, 467, 290–300.

22. Galettis, P. & Norton, R. S. (1990). Biochemical andpharmacological studies of the mechanism of actionof tenebrosin-C, a cardiac stimulatory and haemolyticprotein from the sea anemone Actinia tenebrosa.Toxicon, 28, 695–706.

23. Caaveiro, J. M. M., Echabe, I., Gutierrez-Aguirre, I.,Nieva, J. L., Arrondo, J. L. & Gonzalez-Manas, J. M.(2001). Differential interaction of equinatoxin II withmodel membranes in response to lipid composition.Biophys. J. 80, 1343–1353.

24. Macek, P., Zecchini, M., Pederzolli, C., Dalla Serra, M.& Menestrina, G. (1995). Intrinsic tryptophan fluor-escence of equinatoxin II, a pore-forming polypeptidefrom the sea anemone Actinia equina L., monitors itsinteraction with lipid membranes. Eur. J. Biochem. 234,329–335.

25. Macek, P. & Lebez, D. (1988). Isolation and character-ization of three lethal and hemolytic toxins from thesea anemone Actinia equina L. Toxicon, 26, 441–451.

26. Norton, R. S., Bobek, G., Ivanov, J. O., Thomson, M.,Fiala-Beer, E., Moritz, R. L. & Simpson, R. J. (1990).Purification and characterisation of proteins withcardiac stimulatory and haemolytic activity from theanemone Actinia tenebrosa. Toxicon, 28, 29–41.

27. Simpson, R. J., Reid, G. E., Moritz, R. L., Morton, C. &Norton, R. S. (1990). Complete amino acid sequence oftenebrosin-C, a cardiac stimulatory and haemolyticprotein from the sea anemone Actinia tenebrosa. Eur.J. Biochem. 190, 319–328.

28. Belmonte, G., Menestrina, G., Pederzolli, C., Krizaj, I.,Gubensek, F., Turk, T. & Macek, P. (1994). Primary andsecondary structure of a pore-forming toxin from thesea anemone Actinia equina L., and its association withlipid vesicles. Biochim. Biophys. Acta, 1192, 197–204.

29. Zhang, W., Hinds, M. G., Anderluh, G., Hansen, P. E.& Norton, R. S. (2000). Sequence-specific resonanceassignments of the potent cytolysin equinatoxin II.J. Biomol. NMR, 18, 281–282.

30. Hinds, M. G., Zhang, W., Anderluh, G., Hansen, P. E.& Norton, R. S. (2002). Solution structure of theeukaryotic pore-forming cytolysin equinatoxin II:implications for pore formation. J. Mol. Biol. 315,1219–1229.

31. Athanasiadis, A., Anderluh, G., Macek, P. & Turk, D.(2001). Crystal structure of the soluble form ofequinatoxin II, a pore-forming toxin from the seaanemone Actinia equina. Structure (Camb.), 9, 341–346.

32. Poklar, N., Fritz, J., Macek, P., Vesnaver, G. &Chalikian, T. V. (1999). Interaction of the pore-formingprotein equinatoxin II with model lipid membranes: acalorimetric and spectroscopic study. Biochemistry, 38,14999–15008.

33. Menestrina, G., Cabiaux, V. & Tejuca, M. (1999).Secondary structure of sea anemone cytolysins insoluble and membrane bound form by infraredspectroscopy. Biochem. Biophys. Res. Commun. 254,174–180.

34. Anderluh, G., Barlic, A., Podlesek, Z., Macek, P.,Pungercar, J., Gubensek, F. et al. (1999). Cysteine-scanning mutagenesis of an eukaryotic pore-formingtoxin from sea anemone. Topology in lipid mem-branes. Eur. J. Biochem. 263, 128–136.

35. Malovrh, P., Viero, G., Serra, M. D., Podlesek, Z.,Lakey, J. H., Macek, P. et al. (2003). A novel mechanismof pore formation: membrane penetration by theN-terminal amphipathic region of equinatoxin. J. Biol.Chem. 278, 22678–22685.

36. Malovrh, P., Barlic, A., Podlesek, Z., Macek, P.,Menestrina, G. & Anderluh, G. (2000). Structure-function studies of tryptophan mutants of equina-toxin II, a sea anemone pore-forming protein. Biochem.J. 346, 223–232.

37. Hong, Q., Gutierrez-Aguirre, I., Barlic, A., Malovrh,P., Kristan, K., Podlesek, Z. et al. (2002). Two-stepmembrane binding by equinatoxin II, a pore-formingtoxin from the sea anemone, involves an exposedaromatic cluster and a flexible helix. J. Biol. Chem. 277,41916–41924.

38. Anderluh, G., Dalla Serra, M., Viero, G., Guella, G.,Macek, P. & Menestrina, G. (2003). Pore formation byequinatoxin II, a eukaryotic protein toxin, occurs byinduction of nonlamellar lipid structures. J. Biol.Chem. 278, 45216–45223.

39. Bonev, B. B., Lam, Y. H., Anderluh, G., Watts, A.,Norton, R. S. & Separovic, F. (2003). Effects of theeukaryotic pore-forming cytolysin equinatoxin II onlipid membranes and the role of sphingomyelin.Biophys. J. 84, 2382–2392.

40. Danielson, M. A. & Falke, J. J. (1996). Use of 19F NMRto probe protein structure and conformationalchanges. Annu. Rev. Biophys. Biomol. Struct. 25,163–195.

41. Ho, C., Pratt, E. A. & Rule, G. S. (1989). Membrane-bound D-lactate dehydrogenase of Escherichia coli: amodel for protein interactions in membranes. Biochim.Biophys. Acta, 988, 173–184.

42. Anderluh, G., Pungercar, J., Strukelj, B., Macek, P. &Gubensek, F. (1996). Cloning, sequencing, andexpression of equinatoxin II. Biochem. Biophys. Res.Commun. 220, 437–442.

43. de Dios, A. C., Pearson, J. G. & Oldfield, E. (1993).


Secondary and tertiary structural effects on proteinNMR chemical shifts: an ab initio approach. Science,260, 1491–1496.

44. Hull, W. E. & Sykes, B. D. (1975). Fluorotyrosinealkaline phosphatase: internal mobility of individualtyrosines and the role of chemical shift anisotropy as a19F spin relaxation mechanism in proteins. J. Mol. Biol.98, 121–153.

45. Sanders, C. R. & Landis, G. C. (1995). Reconstitutionof membrane proteins into lipid-rich bilayered mixedmicelles for NMR studies. Biochemistry, 34, 4030–4040.

46. Glover, K. J., Whiles, J. A., Wu, G., Yu, N-J. , Deems, R.,Struppe, J. O. et al. (2001). Structural evaluation ofphospholipid bicelles for solution-state studies ofmembrane-associated biomolecules. Biophys. J. 81,2163–2171.

47. Luchette, P. A., Vetman, T. N., Prosser, R. S., Hancock,R. E. W., Nieh, M-P. , Glinka, C. J. et al. (2001).Morphology of fast-tumbling bicelles: a small angleneutron scattering and NMR study. Biochim. Biophys.Acta, 1513, 83–94.

48. Glover, K. J., Whiles, J. A., Wood, M. J., Melacini, G.,Komives, E. A. & Vold, R. R. (2001). Conformationaldimorphism and transmembrane orientation of prionprotein residues 110–136 in bicelles. Biochemistry, 40,13137–13142.

49. Marcotte, I., Separovic, F., Gagne, S. M. & Auger, M.(2004). A multidimensional 1H NMR study of theeffect of model membranes on the conformation ofenkephalin. Biophys. J. 86, 1587–1600.

50. Macek, P., Belmonte, G., Pederzolli, C. & Menestrina,G. (1994). Mechanism of action of equinatoxin II, acytolysin from the sea anemone Actinia equina L.belonging to the family of actinoporins. Toxicology,1994, 205–227.

51. Seelig, J. (1978). 31P nuclear magnetic resonance andthe head group structure of phospholipids in mem-branes. Biochim. Biophys. Acta, 515, 105–140.

52. Norton, R. S. & Allerhand, A. (1977). Participation oftryptophan-62 in the self-association of hen egg-whitelysozyme. Application of natural-abundance carbon-13 nuclear magnetic resonance spectroscopy. J. Biol.Chem. 252, 1795–1798.

53. Killian, J. A. & von Heijne, G. (2000). How proteinsadapt to a membrane–water interface. Trends Biochem.Sci. 25, 429–434.

54. Barlic, A., Gutierrez-Aguirre, I., Caaveiro, J. M. M.,Cruz, A., Ruiz-Arguello, M. B., Perez-Gil, J. &Gonzalez-Manas, J. M. (2004). Lipid phase coexistencefavours membrane insertion of equinatoxin-II, a pore-forming toxin from Actinia equina. J. Biol. Chem. 279,34209–34216.

55. Kiyokawa, E., Makino, A., Ishii, K., Otsuka, N.,Yamaji-Hasegawa, A. & Kobayashi, T. (2004). Recog-nition of sphingomyelin by lysenin and lysenin-related proteins. Biochemistry, 43, 9766–9773.

56. Shatursky, O., Heuck, A. P., Shepard, L. A., Rossjohn,J., Parker, M. W., Johnson, A. E. & Tweten, R. K. (1999).The mechanism of membrane insertion for acholesterol-dependent cytolysin: a novel paradigmfor pore-forming toxins. Cell, 99, 293–299.

57. Salgado, J., Grage, S. L., Kondejewski, L. H., Hodges,R. S., McElhaney, R. N. & Ulrich, A. S. (2001).Membrane-bound structure and alignment of theantimicrobial b-sheet peptide gramicidin S derivedfrom angular and distance constraints by solid state19F-NMR. J. Biomol. NMR, 21, 191–208.

58. Miller, J. H. (1972). Experiments in Molecular Genetics,Cold Spring Harbor Laboratory Press, Cold SpringHarbour, NY.

59. Koradi, R., Billeter, M. & Wuthrich, K. (1996).MOLMOL: a program for display and analysis ofmacromolecular structures. J. Mol. Graph. 14, 51–55.

Edited by G. von Heijne

(Received 8 November 2004; received in revised form 23 December 2004; accepted 28 December 2004)

interaction of the eukaryotic pore-forming...

Documents