An Electron-Spin-Resonance Spin-Label Study of the Interaction of Purified Mojave Toxin with Synaptosomal Membranes from Rat Brain

Eur. J. Biochem. / 3 / , 559-565 (1983) (c FEBS 1983

An Electron-Spin-Resonance Spin-Label Study of the Interaction of Purified Mojave Toxin with Synaptosomal Membranes from Rat Brain

Joseph HARRIS, Timothy J. POWER, Allan L. BIEBER, and Anthony WATTS

Department of Biochemistry, University of Oxford

(Received October 12/December 13, 1982)-EJB 6095

The structural properties of isolated purified rat brain synaptosomal membranes, both in the presence and absence of purified active toxin of the Mojave snake Crofalus scutulatus scutulatus, were studied by spin-label electron spin resonance techniques. The spectra from eight different positional isomers of nitroxide-labelled stearic acids, a rigid steroid androstanol, and a spin-labelled phosphatidylcholine intercalated into the synaptosomal membranes, were obtained as a function of temperature from 4 - 40 "C.

The flexibility gradient (from spin-label order parameters) and polarity profile (from isotropic splitting factors) across the synaptosomal membranes, was characteristic for lipid bilayers. The nitroxide spin-labelled steroid, androstanol, intercalated into the synaptosomal membrane, revealed the abrupt onset of rapid cooperative rotation about the long axis of the molecule at 12°C showing that the lipid molecules are rotating rapidly around their long axes at physiological temperatures.

The presence of the Mojave toxin affected the synaptosomal membrane in a complex manner, depending upon the temperature and the position of the nitroxide label on the alkyl chain of the stearic acid probe. Mojave toxin exerted little effect on the flexibility gradient of the synaptosomal membrane at 20 "C, a temperature at which the acyl chain labels detected a structural change in the membranes. At temperatures lower than 20"C, the Mojave toxin produced a change in the flexibility gradient of the synaptosomal membrane which indicated an increased disordering in the upper region of the membrane and a concomitant increased ordering of the acyl chains in the deeper regions of the membrane. At temperatures higher than 20 'C, the order profile of the synaptosomal membrane was shifted by the presence of the Mojave toxin in a manner which indicated that the outer parts of the membrane were more rigid and the inner regions more fluid, than in controls. A cross-over point for the perturbation occurred at C8-9, which is about 12- 14A into the membrane. This is the approximate depth of the hydrophobic pocket shown in pancreatic phospholipase A, [Drenthet al. (1976) Nature (Lorzd.) 261,373 - 3771, a protein likely to be homologous to the basic subunit of the toxin. At all temperatures, rotational lipid motion was inhibited by the toxin as indicated by the steroid probe.

The electron spin-resonance spin-label results are interpreted in terms of the partial penetration of the basic subunit of the intact toxin into the membrane, disordering the ordered chains at low temperature and ordering the disordered chains at physiological temperatures. The purified individual toxin subunits did not perturb the membrane lipids at physiological temperatures implying that both subunits must be associated for activity of the toxin which is confirmed by toxicily studies.

Some snake venoms contain neurotoxins with prcsynaptic action, e.g. P-bungarotoxin from Bunguruis multicirlutus [I], and these, in general have received less attention than the post synaptically-acting ones, i.e. the x-toxins. Other presynaptically-active snake neurotoxins have now been isolated which are equally or even more toxic than p-bungarotoxin, notably, taipoxin from the venom of the Australian taipan 0.xyurem.s scutellatus scutellatus [2], notexin from the Australian tiger snake Notechis scutatus srutatus [3], crotoxin from the South American tropical rattlesnake Crotalus durrissuc terriJicus [4] and Mojave toxin from the North American rattlesnake Crotulus scutulutus scutulatus [5], the latter being identified as

Abhreviutions. ESR, electron spin resonance; I(m, P I ) , ( 1 1 + 2 ) ( 4 , 4 - dimethyloxazolidine-N-oxy1)stearic acid; Il(r??, n), a-(n + 2)(4,4'-dime~h- yloxazolidine-N-oxyl) stearoyl-cc-palmitoyl-P-phosphatidylcholine; 111, 4',4'-diniethylspiro(5a-androstan-l-ol) ;

Enzyme. Pancreatic phospholipase A, (EC 3.1.1.4).

one of the most toxic species of rattlesnakes found in the continental United States [6]. All of these presynaptically- active toxins have been found to have an intrinsic phospholipase A, activity and, where data are available, a sequence homology exists with pancreatic phospholipase A, although toxicity is seemingly not directly related to this enzymic activity. Much, if not all of the work on snake venom toxins, has utilized the electromotor system of elasmobranch fish belong- ing to the family Torpendinidae, or with other purely choliner- gic systems with synapses analogous to those present in skeletal muscle of vertebrates. Though much information is available about neurotoxin effects on neurotransmission, especially that involving acetylcholine, structural aspects of neurotoxin interaction with membrane organelles and with the membrane lipid component, remain unclear.

Only two ESR spin-label studies of snake toxin interaction with organelles using lipid labels have been reported, namely that of a detoxified derivative of taipoxin and T-sacs of the

560

Torpedo [7], and that of neurotoxin 11 of Nuja naja oxina with an acetylcholine receptor protein preparation of Torpedu marmorata [S]. Other previous spectroscopic studies on the interaction of toxins with membranes have concentrated on the toxin itself, either by using an intrinsic chromophore and fluorescence methods [9] or by spin-labelling the toxin through a maleimide-linked label [lo]. In view of the limited number of studies that have been made with purified toxins of North American rattlesnake, and because of the critical need to have a detailed understanding of the structural changes in mammalian brain synaptosomes induced by neurotoxins, we have applied ESR spin-label techniques to study the effects of purified Mojave neurotoxin on the lipid component in purified synapto- soma1 membranes.

The Mojave toxin used in this study is a 24300-Da, two- subunit protein possessing phospholipase A, activity in its lethal basic subunit. Pharmacologically, Mojave toxin is a phospholytic neurotoxins capable of presynaptic blockade of acetylcholine release and producing myotoxic effects [l I], properties common to a number of phospholytic snake toxins.

In the present study, the physical state of the lipids in the membrane, their rate and amplitude of motion, was measured from the ESR spectra of eight different positional nitroxide labelled isomers of stearic acid. In addition to these probes, a rigid steroid probe, the nitroxide-labelled androstanol, as well as nitroxide-labelled phosphatidylcholines, were intercalated into the synaptosomal membrane. The structural changes in synaptic membranes were studied as a function of temperature between 4°C and 40°C. Both the temperature dependence and the effect of purified Mojave toxin and its individual subunits were studied in an attempt to assess the structural relevance to the independent action of the neurotoxin on the nerve terminal.

MATERIALS AND METHODS

Mojave Toxin

Mojave toxin was isolated from the crude lyophilized venom of the Mojave rattlesnake purchased from the Miami Serpentarium and purified by the method of Cate and Bieber [5]. The purified toxin, designated as M D - ~ v , was fractionated into a basic and an acidic subunits with molecular weights of 14700 and 9600 and isoelectric points of 9.6 and 3.8 respectively.

Synaptosoinal Membranes

Adult male Wistar rats were killed by gassing in CO,, the brains rapidly excised and the cerebella removed. All sub- sequent operations were conducted at 4°C. Six animals were used for each experiment.

Subcellular fractionation was performed essentially as dscribed by Jones and Matus [12]. The crude membrane fraction was purified by lysing in hypotonic 5mM Tris/Cl, pH 8.1 for at least 30 min, followed by subfractionation on a discontinuous gradient of 10%- 28 %- 34 sucrose. Ultra- centrifugation at 60000 x g for IlOmin, yielded three distinct fractions, a fluocculent white upper band of myelin, a brown pellet of mitochondria and a fluffy yellowish-grey middle band containing the purified synaptosomes. The latter were carefully removed, apportioned into 1 ml micro-tubes and stored at - 20 'C until required. Aliquots were taken for protein and phospholipid analyses. Synaptosomes isolated and purified in this manner were, in our hands, found to be free of post- synaptic adhesions as determined by electron microscopy.

0 x N - Q

CH3(CHZ)m 'C' ' ' (CHZ)" COOH

I ( rn ,n)

m x w ' L

Q

Fig. 1. Formuluenfl~id.~]Jiti-lrrbelsusedin thestudy. I(n7,n) = (17 + 2)(4,4-, methyl-oxazolidine-N-oxy1)stearic acid, isomers used were 1(13,2), I ( 12,. l(lO,S), 1(9,6), I(7,8), l(S,lO), 1(3,12), 1(1,14). II(m,n) = B-(n + 2)(4',* dimethyl-oxazolidine-N-oxyl)stearoyl-~-palm~toyl-~-phosphatidylcholin isomer used was 11(13,2). I11 = 4,J'-dime~hyl-spiro-(S-~-androstan-l-c

Protein and Phospholipid Analyses

Synaptosomal membrane protein concentrations were dc termined by the Coomassie Blue method of Bradford [13] usin, purified serum albumin as a standard.

Phospholipid content of synaptosomal membranes werc determined by acid digestion of the membranes according t c the procedure of Barlett [I41 and the analysis of phosphate b) the method of Eibl and Lands [15].

Spin Label Probes

The doxy1 stearic acid spin-label isomers [I(m,n), Fig. 11 with the nitroxide group situated at C-5, C-12 or C-16 position of the hydrocarbon chain and the androstanol spin-label (111, Fig. 1) were purchased from Syva Inc. (Palo Alto, CA, USA). Other stearic acid isomers, with the deoxyl group positioned at C-4, C-7, C-8, C-30 and C-14 of the alkyl chain respectively, were synthesized by methods similar to those described by Hubbell and McConnell [16]. The phosphatidylcholine spin-label [11(13,2), Fig. I], with the nitroxide group situated at the C-4 of the sn-2 alkyl chain was synthesized by acylation of I-palmitoyl lysophosphatidylcholine (Fluka, Buchs, Switzerland) according to the general method of Boss et al. [17]. Spin-labels I and 111 were used from stock solutions of 1 mg/ml in CHCI,, and I1 from an ethanolic solution at 4 mg/ml all stored at - 20-'C.

Spin-Labelling of' Synaptosomal Membranes

A 3-ml aliquot of isolated synaptosomal membrane in sucrose from the density gradient was washed with a lO-fo\d quantity of buffer and recovered by centrifugation (20000 x g, 15 min). The pellet was resuspended in 100 111 of fresh buffer or, if treated with toxin, with 5Op1 of buffer and 50pl of toxin solution (0.2 mgiml), then shaken and incubated at 4 -C for 10min with a dry film of spin-label 1 or 111. The spin-label (1 or 111) film was obtained by placing an aliquot of stock solution in a small glass tube and the organic solvent removed, first i n a stream of oxygen-free N,, then by vacuum dessication for at least 3h. Labelling levels of 1 or 2mol y!;, to membrane phospholipid were utilized. The labelled membrane suspension was transferred to a 100 p1 glass capillary and concentrated by low-speed centrifugation (bench centrifuge, 1500 rev./min for

56 1

3 min). About half of the aqueous phase was removed before ESR experiments. Phospholipid spin-labels were added from a small volume of ethanolic stock solution to 5ml of washed membrane suspensions [18]. In all cases, the ethanol concentration was less than 0.2% (v/v) and the total spin label added was about 1 % with respect to membrane lipid although label incorporation was less efficient than with I or 111. The membrane suspensions were washed and centrifuged (1 5000 x g; 15 min) at least four times to remove unincorporated spin

label which is in the form of bilayer vesicles. The spin-labelled membrane pellet was resuspended in buffer or toxin, and prepared for ESR measurements as detailed earlier for fatty acid or steroid spin-labelled membranes.

Electron Spin Resonance

Instrumentation. ESR spectra were obtained using a Varian Century Line El09 9 GHz spectrometer equipped with a heater and nitrogen gas-flow system. The sample capillaries were inserted into a standard 4-mm quartz ESR tube containing light silicon oil for thermal stability. Sample temperature was measured with a thermocouple placed in the silicon oil just above the ESR cavity. ESR spectra were obtained at a modulation amplitude of 0.125 mT and microwave power of 30 mW.

E S R Spectra of Fatty Acid ( I ) and Phospholipid (111) Spin-Labels

The average amplitude (0) of membrane lipid acyl chain motion is given by the order parameter, S, by the relationship S = 1/2(3 a - 1 ) . The experimentally determined ESR spin- label order parameter is given by [19]:

where A,,, A,, and A,, are the single crystal, principle hyperfine splittings for the doxy1 group. All and A; are the experimentally determined, partially averaged hyperfine splittings as shown in Fig. 2. Computer simulations by Gaffney [20] have shown that, due to line-width limitations in accurately measuring A l , the correction to the experimentally determined parameter A ; must be made, such that:

AL = A;+ 1.4[1- (All - A;)/A,, - l / 2 (Axx + Ayy)l (2) where all values of A are in units of 10-4T. The normalization term of ah, conipensates order parameter determinations for the sensitivity of nitroxide hyperfine splittings to environmen- tal polarity, a relative measure of which is given by the isotropic splitting factor a,, where uo = 1/3 ( A l l + 2 A J and ub = 1/3

Order parameters can only be determined using Eqn (2) when the spin-label motion in fast (zR < 3 ns) on the ESR time- scale. Slow motion results in line-broadening and in this situation the maximum hyperfine splitting A,,,, which is equivalent to Al l for fast motion, can be used be to give an indication of acyl chain motion about the long niolecular axis : i.e. wobbling. Also slow-motion broadening results in a temperature dependent isotropic splitting factor, a,, and only when motion becomes fast does a, become temperature independent.

E S R Spectra of’dndrostanol. For steroid spin labels, which are rigid and planar, i.e. S = 1, long axis rotation can be measured from the separation between the two outer spectral extrema (Amax; in Fig.6) as a consequence of motional

+ A,, + A ~ , ) (see ~ 9 1 ) .

Fig. 2. ESR spectra of sterir acidspin-labels in rut synaptosomal membmizes. Skaric acid spin-labels, I(m,n), were intercalated into the lipid phase of synaptosomal membranes at a concentration of 1 mol”/, of the endogenous lipid and the spectra shown above recorded at 37°C in the absence of toxin. The method of measuring the maximum Al l and minimum, A:, hyperfine splittings is shown (see [19,20])

averaging of the A,, and A,, hyperfine splittings in the plane perpendicular to the long axis of the probe. For slow correlation times zll, i.e. greater than 1 ns, the following empirical equation, derived from simulations [21] can be used, with less sensitivity for fast motion,

711 = (1 - Arnax/Azz)-’ ( 3 ) where a and b are constants of values 0.259611s and 1.396 respectively. For correlation times less than 1 ns, dynamic information can be obtained in terms of line widths and line hights [21] using the effective linewidth parameters B’ and C‘ which are defined as:

(4)

(5)

B’ = AH0/2[(Z0/I- I)”’ - (I,/Z+

C‘ = AH0/2[(I , / I - I)”’ + (Zo/Z+ 1)”’ -21

where AH, in the peak to peak derivative of the central line width and I,, I - I , , are the line hights of the central, high- field and low-field line hights respectively as illustrated in Fig. 6. Correlation times are then obtained from calibration curves [21].

RESULTS

Stearic Acid Spin-Labels, I ( m, n)

Typical ESR spectra obtained with positional isomers of the spin-labelled stearic acids, I(m, n), intercalated into synapto- soma1 membranes are illustrated in Fig.2. The spectra are characteristic for such spin-labels undergoing fast, anisotropic motion in lipid bilayers. Hyperfine splittings, Al l and A;, used to calculate order parameters and isotropic splitting factors were measured as shown.

Amplitude of’h’otion. The temperature dependences of the outer hyperfine splitting, A,,,, of three of the stearic acid spin labels are given in Fig. 3. The A,,, parameter gives a relative measure of the acyl chain segmental motion both for slower and faster motion of the chains. As scen in Fig.3, A,,, decreases as the temperature increases, indicative of increased mobility at higher temperatures. For spin-labels positioned deeper in the membrane, that is for larger values of n, larger segmental motion occurs than in regions closer to the polar- apolar interface of the membrane, as shown by the decrease in

562

0.11, 'A.

2 L 6 8 10 12 1L 16 Carbon number, n ( c - c )

Fig. 4. Order parameier prqfilr Jor the lipid stearic acid spin-labels it7 rat .synuptosornes. The order parameter, S, caiculated from Eqn (1) for the fatty acids labels I(nz,n), as a function of position of the spin-label on the acyl chain at 20°C and 37 'C in the presence @--A) and absence (0 4) of Mojave toxin

J 10 20 30 LO

Temperature ( O C 1 Fig. 3. Tcmpernture dependence of ESR parameters j i v spin-labelled ,fatty acids reporting on the lipidphrue of rat synaptosomal niembranes. The outer hyperfine splitting, A, , , , measured as shown in Fig. 2, for the stearic acid spin-labels, I(m,n), in synaptosomal membranes in the presence of Mojave toxin ( A ~ --A) and in the absencc of toxin (control) (04). Top pair, middle pair and bottom pair of spectra are from 1(13,2), I(9,6) and 1(5,10), respectively

A,,, with increased n. The addition of Mojave toxin to the synaptosomal membranes, at toxin levels of 1 mM per 1.1 - 1.3 mg of phospholipid significantly changes the spin label mobility, depending on the temperature and the spin label. In Fig. 3 it can be seen that for the stearic acid with nitroxide spin label at C-4 on the acyl chain, 1(13,2), the addition of toxin caused a decrease in A,,, (increased mobility) over that of the control at low temperatures (lower than about 17'C), but caused in increase in A,,, at higher temperatures. For the other two spin label isomers, the effect was reversed. For n = 6 and 10, the addition of toxin increased A,,, values at lower temperatures, but decreased them at the higher temperatures. For all of the labels studied there was a cross-over of the treated with the control curves centered around 20"C, a region of temperature at which the addition of toxin appeared to have no effect on label mobility. The measurements of A,,, for all the spin-labels I(m, n ) reveal inflections, both with and without added toxin, at temperatures between 18°C and 24°C as shown for II = 2, 6 and 10 in Fig. 3. This inflection temperature is the same temperature at which the inner hyperfine splitting (from which Al is measured) can be resolved, and is close to the temperature at which the perturbations caused by the toxin on the A,,, values are negligible. Note the cross-over points for the variation in A,,, with temperature, both with and without toxin in Fig. 3. Order parameters can only be measured above the temperature at which A ; is resolved. The maximum hyperfine splitting for fast spin-label motion is now All and order parameters were calculated using Eqn (1) for temperatures greater than 20 "C for all isomers of I(m, H) used. The values of S at 20°C and 37°C are given in Fig. 4 with and without added Mojave toxin. In common with the indications of Fig.3, the presence of toxin does not change the order parameters at 20 'C. However at the physiological temperature o f 37"C, the presence of toxin increases the order parameter above the C-8 positional spin-label isomer, but decreases the order parameter towards the centre of the membrane, that is, for (n + 2) > 8.

Polarity Profile

As in the case of the order parameter, there is a positional dependence of the nitroxide isotropic hyperfine splitting factor,

Fig. 5. Relarive polarity prqfile ,fbr rut brain synuptosonial membrane wiih and without toxin. The polarity parameter a0 was measured as described in Materialsand Methods and plotted as a function of the label position, n, on the acyl chain of the label I(n?,n) in the presence of Mojave toxin (A ---A) and in the absence (0 +) of Mojave toxin

a,, which gives a profile of polarity through the membrane [19]. Values of a, were calculated from direct measurement of A l l and A; from ESR spectra of I(m,n) and were idependent of temperature at temperatures greater than 20 "C. The isotropic hyperfine constants given in Fig. 5 are the averaged values from 20 - 40°C and constitute a relative polarity profile across of the isolated purified synaptosomal membranes. Fig. 5 indicates a sigmoidal form with a region of relatively high polarity in the upper part of the membrane leaflet near the polar-apolar interface, followed by a sharp decrease in polarity leading to a plateau region of lower polarity in the center of the membrane. The presence of Mojave toxin tends to increase the polarity in the upper part of the bilayer followed by an even greater decrease in polarity, as reflected in the steeper profile.

Steroid Spin Label ( I I I )

Information regarding spin-label rotation in the membrane can be obtained from the hyperfine splitting, A,,,, line-widths and line-hights of the androstanol label (111) measured as shown in Fig. 6 in terms of correlation times (Eqns 3, 4, 5 and

563

rA-"l - 3.2 : [ v

1 mT

A I ! - Ho

Fig. 6 . ESR speclru n f the steroid .spin-label in rut synuptosomes. The androstanol (111) was intercalated into the lipid phase of synaptosomal membrane at a concentration of 1 mol label to the endogenous lipid as described in Materials and Methods. The spectrum shown here was recorded at 17.5"C and shows the line-width parameter AH,, and line- heights, I, ] , I, and I- used to calculate correlation times, q, in thc fast motional regime using Eqns (4) and ( 5 ) and, in the slow motional regime from A,,, to give the results of Fig. 7 (see [22])

t

' .e l I0 ' io ' io ' L '0 ' Temperature PCI

Fig. 7. Temperature dependence of the outer hyperfine splitting androtational corrrlution time for the sterol spin-lube1 in rut synuptosomes. The line- splittings, A,,, (open symbols) and line-hights were measured as shown in Fig. 3 and the rotational correlation times, T ~ , (closed symbols), calculated using Eqns ( 3 ) , (4), and (5) over the temperature range shown, in the presence of Mojave toxin (&-A) and in the absence of toxin (control, 0-1

[21]). In Fig. 7 the results are given for the temperature dependence of rotational correlation time in the synaptic membranes. A rapid transition from a long (zll > 7 ns) to a short (tll 1 ns) correlation time for both control and Mojave toxin treated synaptosomes is observed at around 12 OC (Fig. 7). The correlation times for the toxin-treated membranes are slightly, but consistently higher over the measured temperature range than the correlation time for the control membranes.

Phospholipid Spin-Lube1 II(m, n)

Experiments carried out on synaptic membranes with the phosphatidylcholine spin-label with the nitroxide situated at C-4 of the alkyl chain [II(13,2), Fig. I], gave results (not shown) essentially the same as those obtained with the nitroxide labelled at C-4 of the stearic acid probe, [I(13,2), Fig. I] as given in Fig.3. The toxin decreased A,,, at lower temperatures,

Temperature ( O C I Fig. 8. Muximum hypwfine splitting, for the phosphutidyl~holine .xpiti-lrbel intercalated into rut synuprosomul memhrrinrs. Values of A,,, were measured as shown in Fig. 2 for the phospholipid label, ll(13.2), in the absence of Majava toxin (o-), and in the presence of the acid subunit (A-A) and basic subunit (V-- V ) of the toxin as a function of temperature

(< 20°C) but increased it at higher temperatures (> ZO'C). Further the phase transition monitored by C-4 phospholipid label was the same as that obtained with the C-4 stearic acid spin label. These experiments suggest that the toxin showed no specific interaction associated with the fatty acid spin-label but had a general effect monitored also by the phospholipid label.

To investigate the possible effect of either of the two toxin subunits on synaptic membranes, the phosphatidylcholine label was incorporated into synaptosomes and both subunits added separately in the same molar propositions used for whole toxins studies. The results of Fig. 8 indicate that neither subunit in isolation can perturb the acyl chains of the synaptic membranes at physiological temperatures. However, the inflection in the A,,, values at around 20°C still reflects a structural transition, unperturbed by the toxin subunits.

DISCUSSION

The experimental results of the present spin-label studies on isolated and purified synaptosomal membranes show that the stearic acid spin-labels record both a polarity profile of gradually decreasing polarity (Fig. 5 ) and a fluidity gradient of gradually increasing fluidity (Fig. 4) as the spin-label group proceeds down the alkyl chain away from the carboxyl (head) group of the stearic acid molecule. This indicates that the amphiphilic stearic acid probes are located at specific positions in the synaptosomal membrane. Such polarity profiles and flexibility gradients are characteristic of fatty acid labels in model membrane bilayers [19] as well as in chromaffin granule membranes [23,24]. It appears that a substantial proportion of the lipid in the synaptosomal membrane (at least that sensed by spin labels) is organized in a lipid bilayer structure.

The data of Fig. 3 and 8 show that the fatty acid, I(m,n) and phosphatidylcholine II(13,2) labels detect a structural reorganization of the lipids a t M 2 0 T , whereas the androstranol label, 111, reveals the onset of lipid rotation at z 12°C. Similar results have been previously reported from studies using I( 12,3) and I(2,24) [25]. I n light of these low temperatures, it seems unlikely that the transitions are of any direct physiological significance in the functioning of the synaptosomal membrane. These results do however imply that a physiological temperatures the lipid molecules in the membrane are undergoing rapid segmental and rotational motion. The transition to rapid rotation occurs over a very narrow temperature range, showing that the lipid molecules are capable of interacting together in a highly cooperative manner. This could be significant for trigger

564

mechanisms involving the lipid phase of the membrane, as with receptor-initiated action [26]. The temperature discontinuity in segmental motion at z 20°C, reflected in the A,,, measurements for acyl chain labels is more difficult to interpret. Although the transition is one of motion about the long molecular lipid axis, that is wobbling, it could reflect a reorganization of membrane components rather than an ordered-disordered phase transition characteristic of many saturated lipids in model membrane bilayers [I 91. The diverse lipid composition and lack of evidence for laterally separated lipids in the synaptosomal membranes, precludes the possi- bility of a cooperative phase transition of the lipid acyl chains from acyl chains in an all trans conformation to disordered, liquid crystalline chains.

Mojave toxin is seen to affect the structure of the synapto- soma1 membrane in a complex way, depending upon the temperature and the positional depth in the membrane of the nitroxide label on the stearic acid I(rn,n). Mojave toxin increased the amplitude of acyl chain motion at temperatures below 20 "C for n < 6, but decreased the amplitude of motion for n > 6 as seen from the results of Fig. 3. The converse is observed for temperatures above 20 "C. The toxin decreased wobbling motion for n < 6 but increased it for n > 6. An analogous phenornenom arises with the effect of cholesterol on membrane bilayers, but uniformly throughout the whole bilayer. The sterol fluidizes ordered phase lipid bilayers, but condenses or rigidifies bilayers in the fluid phase [27]. In the upper half of the synaptic membrane for n < 6, toxin fluidizes the acyl chains below 20°C but rigidifies them above 20°C relative to the non-toxin-treated controls. This observation implies a physical penetration of the native toxin into the synaptic membrane such as to perturb directly the lipid acyl chain motion.

An indication of the depth of toxin penetration into the synaptosomal membrane can be deduced from the bilayer profile in Fig. 4. At 37"C, the order parameter is unchanged at C8-9, suggesting that the toxin penetrates to this acyl chain segment where no effect is monitored upon interaction with the toxin. The same cross-over point is obtained from an A,,, profile across the bilayer at temperatures both above and below 20°C (not shown). Below 20"C, the disorder induced by the toxin penetration in the upper part of the bilayer is com- pensated for by a decreased amplitude of segmental motion in the lower part of the bilayer chains towards the membrane centre. The converse is true above the structural transition temperature; the chains ordered by the toxin in their upper part nearest the polar-apolar interface become more disordered in their lower part.

The polarity profile of Fig. 5 confirms that on interaction with toxin, the nitroxides sense a more hydrophilic environ- mcnt than in the absence of toxin for temperatures above 20 "C. This may be due to an easier penetration of water into this now relatively more ordered part of the membrane (see Fig. 4). Alternatively, the toxin molecule which clearly has an outer surface less hydrophobic than hydrocarbon-acyl chains, may itself impose a more hydrophilic environment on the nitroxide, thereby increasing the isotropic splitting factor, q,. The relative polarity parameter, a,, is quite independent of the amplitude of segmental chain motion, at least for fast spin-label motion. Thus, although the amplitude of acyl chain motion in the centre of the bilayer (n = 10- 14) is considerably decreased on toxin penetration into the bilayer, the polarity remains essentially unchanged.

The acyl chain spin-labels, I(rn,n) and II(13,2) give information only on the amplitude of segmental motion. The order

parameters are independent of the rate of chain motion for fast motional anisotropy averaging ( T ~ < 1 ns). Information on the rate of lipid rotational motion is given by the sterol spin-label, 111. The results of Fig. 7, constructed using independent spectral features (lines-positions, line-widths and line-hights), give some insight into the perturbation induced by the presence of Mojave toxin on synaptosomal membranes. Rotational motion is reduced in the presence of toxin both above and below the cooperative transition temperature at 10- 12°C. Such a cooperative transition is common to other natural membranes, for example, bovine chromaffin granule membranes [23].

The change in flexibility gradient by Mojave toxin on the synaptosomal membrane of rat brain in the present study is qualitatively similar to the effect of a p-bromophenylacylated derivative of taipoxin [(PBP),-taipoxin], from the venom of the Australian taipan Oxyuranus scutellatus, on a synaptosomal membrane (T-sacs) of the electromotor system of Torpedo as reported by Dowdall et al. [7]. In the presence of (PBP),- taipoxin the spin-label flexibility profile in T-sacs was altered similarly to Fig.4, with the upper bilayer regions becoming more rigid and the inner regions becoming more fluid with a cross-over point or penetration depth occurring at C8-9 into the membrane. N o structural transition was observed in the T-sac membrane and thus the Mojave toxin interaction with synaptosomal membranes is much more complex than the effects reported for (PBP),-taipoxin on T-sac membranes. The Mojave toxin is a 24300-Da, two-subunit protein possessing phospholipase activity in its basic subunit of 14700 Da. Taipoxin is a larger molecule of 49000 Da and three subunits. Phospholipase activity is present in its 18000-Da basic subunit which is associated noncovalently with the other two subunits of 13500Da. Despite the variation in structure of these and other phospholytic presynaptic neurotoxins, a constant structural homology has been observed in the amino acid sequences of phospholipase A, from a variety of sources. The remarkable constant structural feature makes it likely that the overall rectangular box-like structure with a hydrophobic pocket of about 1 .O nm for pancreatic phospholipase A, deduced from X-ray crystallography [28], may also prevail in the various phospholipase presynaptic neurotoxins [7]. The differences in the present work and that on the Torpedo membrane are thus attributed to membrane rather than to the toxin. The taipoxin interaction of Torpedo synaptosomes therefore closely re- sembles the higher temperature (above 20 "C) results for the rat brain synaptosomes in the present study.

The present study gives some indication of the necessity of protein-protein interactions for toxin binding to the membrane. The presence of the individual acidic or basic subunit in the same molar proportions used for the native toxin studies are not detected in the membrane by the spin-labels as shown in Fig. 8. A synergistic interaction between the two subunits appears to be necessary for structural penetration into the membrane although the way in which the subunits interact is not clear from this study. Functionally, each subunit individually shows very low toxic action on mice, with no toxicity up to 10 mg/kg body weight for the acidic and a median lethal dose, LD,,, of 0.58 mg/kg body weight for the basic subunit, compared to an LD,, of 0.056 mg/kg body weight (mice) [5] for the native toxin. An analogous synergism has been reported for the non-toxic acidic subunit and phospholytic basic subunit of crotoxin [22].

A possible model summarizing the changes in the lipid component of membranes on toxin binding is illustrated schematically in Fig. 9. The whole toxin is directed by elec-

565

Dissociated subunits

T <2OoC T >20°C

Fig. 9. Diagrammalir representation of the mode of Mojave toxin perturbation on rat .synaptosomal membranes deducedjrom the spin-label studies. The dissociated subunits, when added individually to the synaptic membranes in separate experiments, do not perturb the membrane lipids as indicated in the upper scheme. On binding to the membrane through electrostatic associations of the acidic subunit, the intact toxin perturbs the membrane lipids both above and below = 20"C, the temperature of the spin-label detected structural transition in acyl chain motion about the long molecular axis, i.e. wobbling. The depth of penetration to about the 8th or 9th methylene segment is an approximate distance of 1.0- 1.4nm, the depth of the hydrophobic pocket of pancreatic phospholipase A, [28], an homologous polypeptide to the basic subunit of the Mojave toxin

trostatic orientation to the membrane surface by the acid subunit, binds and penetrates neural membranes to a depth of some 1.0- 1.4nm, which is approximately the depth of the n = 6 spin-label moiety and close to the depth of the phospholipase A2 hydrophobic pocket [28]. This causes the hy- drocarbons of neighbouring fluid phospholipids at physiological temperatures to become more ordered in the outer regions of the membrane but more disordered in the inner regions of the membrane. At the present time there is no evidence for the penetration of phospholipase A, into membranes, and its mode of action could involve extraction of lipids before hydrolysis rather than membrane penetration. Also the ESR results here suggest that toxin penetration as detected by the spin-labels, ocurrs over a long time compared to the time needed to average the changes in the hyperfine splittings, that is, with an on-off frequency for the toxin of slower than 10 MHz.

This work was supported in part by the Burroughs-Wellcome Foundation (to A. W.) and in part by the United States National Institute of Health (Grant 5R01 G M 24566-03 to A.L.B). J. Harris acknowledges, with appreciation, a Wellcome Research Travel Grant which made this investigation possible.

J. Harris and A. L. Bieber, Department of Chemistry, Arizona State University, Tempe, Arizona, USA 85281

REFERENCES

1. Chang, C. C. & Lee, C. Y (1963) Arch Int. Pharmacodyn. 144, 241 -

2. Fohlman,J., Eaker,D., Karlsson, E. & Thesleff,S. (1976) Eur. J.

3. Karlsson. E. (1973) Experientia (Basel) 29, 1319-27. 4. Habermann, E. & Breithaupt, H. (1978) Toxicon, 16, 19-30. 5. Cate, R. L. & Bieber,A. L. (1978) Arch. Biochem. Biophys. 189, 397-

408. 6. Glenn, J. L. & Straight, R. C. (1982) in Rattlesnake Venoms: Their

Action and Treatment (Tu, A. T., ed.) pp. 3-119, Marcel Dekker, New York.

7. Dowdall, M. J., Fohlman, J. P. & Watts, A. (1979) Adv. Cytopharma- col. 3, 63 - 76.

8. Tsetlin, V. I., Karlsson, E., Arseniev, A. S., Utkin,Y. N., Surin, A. M., Pashkov, V. S., Pluzhnikov, W. A,, Ivanov, V. T., Bystrov, V. F. & Urchinnikov,Y.A. (1979) FEBS Lett. 106, 47-52.

9. Dufourcq, J. & Faucon, J-F. (1978) Biochim. Biophys. Acta, 436,811 - 824.

10. Rousselet, A., Cartaud, J., Devaux, P. F. & Changeux, J. P. (1982) The EMBO Journal, 1, 439- 445.

11. Gapalakrishmakone, P., Hawgood, B. J., Holbrooke, S. E., Marsh, N. A,, Santana de Sa, S. & Tu,A. T. (1980) Br. J. Pharmacol. 69, 421 -431.

12. Jones, D. H. & Matus,A. J. (1974) Biochim. Biophys. Acta, 356,276- 287.

13. Bradford, M. N. (1976) Anal. Biochem. 72, 248-254. 14. Bartlett,G. R. (1959) J. Bid. Chem, 234, 466-468. 15. Eibl,H. & Lands, W. E. (1969) Anal Biochem, 30, 51-57. 16. Hubbell, W. L. & McConnel1,H.M. (1971) J . Am. Chrm. Soc. 93,

17. Boss, W. F., Kelly, C. J. & Landsberger, F. R. (1975) Anal. Biochem.

18. Watts,A., Volotovski, I. D. & Marsh, D. (1979) Biochemistry, 18,

19. Marsh, D. & Watts, A. (1981) in Liposome.sJrom Physical Structure f o Therapeutic Application (Knight, G. G., ed.) pp. 139- 188, Else- vier, Amsterdam.

20. Gaffney, B. J. (1976) in Spin-labels (Berliner, L. J., ed.) pp. 567-571, Academic Press, New York.

21. Polnaszek, C. F., Marsh, D. & Smith, I. C. P. (1981) J. Magnetic Res.

22. Bon, C . (1982) Toxicon, 20, 105- 109. 23. Marsh, D., Radda, G. K. & Ritchie, G. A. (1976) Eur. J . Biochem. 71,

24. Fretten, P., Morris,S. J., Watts,A. & Marsh, D. (1980) Biochim.

25. Viret, J. & Leterrier, F. (1976) Biochim. Biophys. Acra, 436, 81 1-824. 26. Rimon,G., Hansk, E., Braun, S. & Levitzki,A. (1978) Nature (Lond.)

27. Marsh, D. & Smith, I. C. P. (1973) Biochim. Biophys. Acta, 298, 133-

28. Drenth, J., Enzing, C. M., Kalk, K. H. & Vessies, J. C. A. (1976) Nature

257.

Biochrm. 68, 457 - 469.

314- 326.

64, 289-292.

5006-5013.

43, 54-64.

53-61.

Biophys. A m , 598, 247 - 259.

276, 394- 396.

144.

(Lond.) 264, 373 - 377.

T. J. Power and A. Watts, Department of Biochemistry, University of Oxford, South Parks Road, Oxford, Great Britain, OX1 3QU

Documents

An Electron-Spin-Resonance Spin-Label Study of the Interaction of Purified Mojave Toxin with Synaptosomal Membranes from Rat Brain