Báo cáo khoa học: Effects of sphingomyelin, cholesterol and zinc ions on the binding, insertion and aggregation of the amyloid Ab1)40 peptide in solid-supported lipid bilayers ppt

When the SM bilayer included 35% cholesterol, an increase of 2.5-fold occurred in the amount of peptide bound, with a similar increase in the extent of aggregation, the latter resulting

binding, insertion and aggregation of the amyloid Ab1)40 peptide in solid-supported lipid bilayers

Savitha Devanathan1, Zdzislaw Salamon1, Go¨ran Lindblom1, Gerhard Gro¨bner2and Gordon Tollin1

1 Department of Biochemistry and Molecular Biophysics, University of Arizona, Tucson, AZ, USA

2 Department of Biophysical Chemistry, Umea˚ University, Sweden

The 39–42 amino acid residue amyloid b peptide (Ab)

is a seminal etiologic factor in Alzheimer’s disease

(AD), a member of the large family of

neurodegenera-tive disorders with a common pathology in the form

of aberrant protein folding [1–4] The unifying theme

for all of these amyloidogenic diseases is the

pathologi-cal conversion of specific proteins into toxic assem-blies In the case of AD, its key substance, Ab peptide,

is released as a soluble monomer, but seems to require

a minimal level of aggregation to exert its neurotoxic action [3–9] In particular, soluble oligomeric and pro-tofibrillar Ab structures are the primary toxic agents

Keywords

Alzheimer’s disease; amyloid toxicity;

microdomains; plasmon-waveguide

resonance spectroscopy; rafts

Correspondence

G Tollin, Department of Biochemistry and

Molecular Biophysics, University of Arizona,

Tucson, AZ 85721, USA

Fax: +1 520 621 9288

Tel: +1 520 621 3447

E-mail: gtollin@u.arizona.edu

(Received 8 December 2005, revised 25

January 2006, accepted 2 February 2006)

doi:10.1111/j.1742-4658.2006.05162.x

We utilized plasmon-waveguide resonance (PWR) spectroscopy to follow the effects of sphingomyelin, cholesterol and zinc ions on the binding and aggregation of the amyloid b peptide1)40 in lipid bilayers With a dioleoyl-phosphatidylcholine (DOPC) bilayer, peptide binding was observed, but no aggregation occurred over a period of 15 h In contrast, similar binding was found with a brain sphingomyelin (SM) bilayer, but in this case an exponential aggregation process was observed during the same time inter-val When the SM bilayer included 35% cholesterol, an increase of

2.5-fold occurred in the amount of peptide bound, with a similar increase

in the extent of aggregation, the latter resulting in decreases in the bilayer packing density and displacement of lipid Peptide association with a

bilay-er formed from equimolar amounts of DOPC, SM and cholestbilay-erol was fol-lowed using a high-resolution PWR sensor that alfol-lowed microdomains to

be observed Biphasic binding to both domains occurred, but predomin-antly to the SM-rich domain, initially to the surface and at higher peptide concentrations within the interior of the bilayer Again, aggregation was observed and occurred within both microdomains, resulting in lipid dis-placement We attribute the aggregation in the DOPC-enriched domain to

be a consequence of lipid mixing within these microdomains, resulting in the presence of small amounts of SM and cholesterol in the DOPC micro-domain When 1 mm zinc was present, an increase of approximately three-fold in the amount of peptide association was observed, as well as large changes in mass and bilayer structure as a consequence of peptide aggrega-tion, occurring without loss of bilayer integrity A structural interpretation

of peptide interaction with the bilayer is presented based on the results of simulation analysis of the PWR spectra

Abbreviations

Ab, amyloid b1)40peptide; AD, Alzheimer’s disease; AFM, atomic force microscopy; DOPC, dioleoylphosphatidylcholine; POPC,

palmitoyloleoylphosphatidylcholine; PWR, plasmon-waveguide resonance; SM, brain sphingomyelin; TFA, trifluoroacetic acid; TFE,

trifluoroethanol.

currently associated with the neuropathological events

occurring in patients with AD [3,7–9] Nevertheless,

globular and nonfibrillar Ab peptides are continuously

released during normal metabolism in healthy people,

with no problems observed, and therefore fundamental

questions behind the toxic mechanism in AD are

unsolved [10–12] Recently, the discovery of various

soluble amyloid oligomers having a common structure,

independent of their location, has brought new insight

into possible mechanisms of toxicity [10] The

inhibi-tion of their toxicity by a common oligomer-specific

antibody, connected to cell parts that are accessible by

extra- and intracellular regions, has pointed strongly

to cell membranes as a potential prime target [9,10,13–

15] This is not surprising because Ab has inherited a

transmembrane domain from its precursor protein (a

highly conserved integral membrane protein with a

single transmembrane domain), providing it with an

amphipathic nature, which makes it an ideal target for

toxic events associated with neuronal membranes

[5,13,16–22] The effects of Ab on membranes and

lipid systems, and their possible roles in neurotoxicity,

include changes in membrane fluidity, leading to

membrane depolarization and disorder [23],

mem-brane-mediated aggregation of Ab triggering neuronal

apoptotic cell death [24], lipid peroxidation via H2O2

produced by Cu2+ reduction by Ab [22,25], and even

the formation of calcium-permeable membrane ion

channels [26]

Owing to the amphipatic nature of Ab, a second

process plays a potential key role in AD, namely the

enhancing effects of specific neuronal membranes on

Ab peptide conversion into toxic-acting oligomers

Various lipid membranes have been shown to induce

an electrostatically driven surface accumulation,

fol-lowed by dramatically increased misfolding of Ab, at

rates much higher than in a membrane-free

environ-ment [13–23,25–28] Membrane components, such as

anionic lipids, gangliosides or cholesterol, were shown

to be involved in various stages of Ab aggregation,

and raft-like neuronal membranes seem to play a

signi-ficant role in the regulation of Ab-production and its

cytotoxic products [20–23,29,30] Interestingly, brain

lipid composition in patients with AD is significantly

altered, suggesting a link between lipid composition

and increased susceptibility to neuronal cell death

[16,31,32] Because of its amphipathic nature and the

fact that patients with AD have altered neuronal lipid

compositions [16,17], in the present study we

investi-gated the role of raft-mimicking model neuronal

membranes on the behavior of Ab peptide Using

plas-mon-waveguide resonance (PWR) spectroscopy [33,34]

we elucidated features of the peptide–membrane

inter-action, which might be important for raft membrane-dependent aggregation and neurotoxic action, in par-ticular the presence of sphingomyelin, cholesterol and zinc ions

Results and Discussion

In order to characterize the interaction of Ab with lipid membranes, we used PWR spectroscopy to study the association with bilayers composed of single lipids, of binary lipid mixtures, or of a ternary mixture composed

of dioleoylphosphatidylcholine (DOPC)⁄ sphingomyelin (SM)⁄ cholesterol (1 : 1 : 1 mole ratio), the latter in the presence and absence of added zinc ions This methodo-logy has previously been used in our laboratory to characterize the composition of, the formation of, and insertion into, microdomains in bilayer membranes formed from binary mixtures of DOPC⁄ SM and palm-itoyloleoylphosphatidylcholine (POPC)⁄ SM, both in

1 : 1 mole ratios [35,36]

Interaction of Ab with single lipid bilayers Figure 1A,B shows that for a DOPC bilayer, interac-tion with increasing Ab concentrainterac-tions produced small shifts to higher-incident angles in p-polarized and s-polarized spectra (11 and 5 mdeg shifts, respectively,

at a peptide concentration of 5 lm in the aqueous cell compartment; s-polarized data not shown) The spec-tral shifts can be ascribed to an overall mass increase

in the membrane as a result of peptide association with the bilayer These shifts followed a single hyperbolic curve with an apparent dissociation constant (KD) of 0.16 ± 0.02 lm (Fig 1B; Table 1) No further spectral changes were observed with time (over a time period

of
15 h), indicating that no peptide aggregation occurred during this interval

The addition of Ab to a lipid bilayer containing only

SM again resulted in increasing spectral shifts to higher incident angles at peptide concentrations up to 5 lm (Fig 2A,B) for both p- and s-polarized spectra (17 ver-sus 14 mdeg, respectively; s-polarized data not shown) This is similar to what was observed for the DOPC bilayer, and occurred with a similar binding affinity (Fig 2B; Table 1) However, in the SM bilayer, after peptide binding and upon further equilibration with time (up to
15 h), a slow progressive increase in spec-tral position to higher-incident angles was observed These changes, which we attribute to peptide aggrega-tion, are plotted in Fig 2C; they occurred exponentially with a half-time of
4.6 h (Table 2) We have obtained similar results to these with bilayers composed of DOPC and SM in a 1 : 1 mole ratio (data not shown)

Interaction of Ab with binary lipid bilayers

As shown by the data in Fig 3, and by the KDvalues

in Table 1, Ab was bound fivefold more tightly to an

SM bilayer containing cholesterol (1 : 0.35 mole ratio)

and resulted in an approximately twofold larger

mag-nitude in the spectral shifts obtained at 5 lm peptide

concentration (40 mdeg for p-polarization versus 27

mdeg for s-polarization; s-polarized data not shown)

Again, time-dependent aggregation of the peptide

occurred, with a half-time of
4.2 h (Fig 3C;

Table 2) However, in this case the spectral changes

involved shifts towards lower-incident angles (for

p-polarization, )23 mdeg and for s-polarization, )9 mdeg; s-polarized data not shown), contrary to that observed with the SM bilayer in the absence of choles-terol A spectral shift to smaller angles is caused by a decrease in refractive index As this parameter is pro-portional to mass density, we attribute this shift to a net loss in mass resulting from the removal of lipid molecules from the bilayer and transfer to the Gibbs border, which occurs upon peptide insertion into the bilayer and aggregation leading to lipid displacement (see below for further discussion) This contrasts with the process of peptide interaction and aggregation with the SM bilayer in the absence of cholesterol, where a net mass increase was observed as a result of peptide accumulation and possibly also lipid recruitment from the Gibbs border

Interaction of Ab with ternary lipid bilayers For a ternary mixture composed of DOPC⁄ SM ⁄ choles-terol (1 : 1 : 1 mole ratio), and using a PWR resonator design with a higher resolution that enabled the obser-vation of membrane microdomains [35], two resonances were obtained corresponding to less-ordered thinner domains at lower-incident angles (DOPC enriched), and more ordered and more densely packed thicker microdomains (SM enriched) occurring at higher-inci-dent angles (Fig 4A; cf ref 35 for evidence supporting this assertion) In this case, the Ab-binding process was biphasic (the initial phase is shown in the inset to Fig 4B), with an initial positive shift followed by a neg-ative shift as the peptide concentration was increased, and occurred preferentially into the SM-rich microdo-main The latter is shown by the larger shift observed for the resonance corresponding to this domain than for the DOPC-enriched domain (compare Fig 4B,C)

Table 1 Amyloid b1)40 peptide (Ab) binding affinities (p-polariza-tion) DOPC, dioleoylphosphatidylcholine; SM, sphingomyelin.

Resonance shifts (mdeg) a

DOPC ⁄ SM ⁄ cholesterol

SM-rich domain 0.004 ± 0.001 (K D1 ) 3 ± 1

0.110 ± 0.01 (KD2) )15 ± 2 DOPC ⁄ SM ⁄ cholesterol (+ 1 m M Zn)

SM-rich domain 0.003 ± 0.001 (KD1) 20 ± 2

0.022 ± 0.001 (KD2) )32 ± 1

a Extrapolated to infinite peptide concentration.

63.5 63.6 63.7 63.8

0.0

0.2

0.4

0.6

2 1

Incident angle (deg)

A

0

5

10

15

B

A β (µ M )

Fig 1 Binding of amyloid b1)40peptide (Ab) to a

dioleoylphosphat-idylcholine (DOPC) bilayer (A) Plasmon-waveguide resonance

(PWR) spectra for p-polarization are shown for a solid-supported

DOPC bilayer before (curve 1) and after (curve 2; squares) the

addi-tion of Ab (5 l M bulk concentration in the aqueous cell

compart-ment) The buffer used in the sample compartment was 10 m M

Tris (pH 7.4) Experiments were performed at 25 ± 0.1
C (B) Plot

of p-polarized PWR resonance minimum spectral shifts as a

func-tion of the concentrafunc-tion of Ab added to the PWR sample

compart-ment The binding data were fit by a single hyperbola (solid line),

and the binding affinity values and the magnitude of the spectral

shift are given in Table 1.

It is worth noting that accumulation of peptide at the

microdomain surface, rather than insertion into its

interior, is also possible We attribute the initial positive

shift to a mass increase resulting from peptide binding

to the bilayer surface, and the subsequent negative shift

to peptide insertion into the bilayer and lipid

displace-ment The higher spectral resolution allowed us to

observe both of these processes in this experiment

Con-sistent with the larger spectral shifts for the SM-rich

microdomain, and thus a higher peptide concentration, the binding affinity of the peptide for this domain was approximately fourfold larger than for the DOPC-enriched domain (Table 1) Note that a shift to lower-incident angles occurred for both microdomains, again indicating insertion of peptide within the bilayer, pro-ducing a less densely packed bilayer as a result of expulsion of lipid molecules That this occurred in both microdomains is probably a consequence of the fact that during microdomain formation a small amount of

SM and cholesterol is mixed into the DOPC portion of the bilayer, and a small amount of DOPC and choles-terol is mixed into the SM-enriched microdomain [35] Peptide aggregation also occurred in this system, and within the first 3 h the spectral changes occurred pre-dominantly in the SM-enriched region (Fig 4A, curve 3), as evidenced by the smaller spectral shift that occurred in the DOPC-enriched domain Figure 4D shows the time course of resonance minimum shifts upon peptide aggregation At 15 h, the magnitude of the spectral changes was approximately twice as large for the SM-rich microdomain In the DOPC-rich microdomain, the half-time was 4.5 h, whereas it was slightly shorter in the SM-rich domain (Table 2)

0 10 20

A β (µ M )

B

63.6 64.0 64.4

0.0

0.2

0.4

0.6

0.8

3 2

1

Incident angle (deg)

A

15 30

45

C

Time (h)

Fig 2 Binding and aggregation of Ab in a sphingomyelin (SM) bilayer (A) p-Polarized plasmon-waveguide resonance (PWR) spec-tra are shown for an SM bilayer before (curve 1) and after (curve 2; circles) the addi-tion of Ab (5 l M bulk concentration in the aqueous cell compartment) Curve 3 (trian-gles) shows the spectrum after equilibration for 15 h Other conditions were as in Fig 1 (B) Plot of PWR spectral shifts induced by

Ab binding to the bilayer with an increasing concentration of added peptide The data were fit by a single hyperbola, with the binding constant and total spectral shift given in Table 1 (C) Plot of the time course

of spectral changes for p-polarization associ-ated with peptide aggregation A single exponential fit to the data is shown (solid line) with a half-time and total spectral shift

as presented in Table 2.

Table 2 Aggregation kinetics (p-polarization) DOPC,

dioleoylphos-phatidylcholine; SM, sphingomyelin.

Bilayer

Half-times (h)

Resonance shifts (mdeg)a

DOPC ⁄ SM ⁄ cholesterol

DOPC ⁄ SM ⁄ cholesterol

(+ 1 m M Zn)

t1¼ 0.53 ± 0.06 )15 ± 3

t 2 ¼ 2.96 ± 0.12 194 ± 5

t 3 ¼ 17.9 ± 0.51 58 ± 3

a

Extrapolated to infinite time.

Effect of Zn2+on Ab interaction with a ternary

lipid bilayer

The binding of metal ions, such as Zn2+, to Ab

pep-tides has been shown to facilitate peptide penetration

of the hydrocarbon core of the membrane and

subse-quent aggregation [22,37] In order to test the effect of

zinc addition in the present experiments on peptide

binding and bilayer structural changes caused by

aggregation, we used a DOPC⁄ SM ⁄ cholesterol (1 : 1 : 1

mole ratio) bilayer in contact with a buffer containing

1 mm Zn2+, present on both sides of the bilayer The

spectral changes produced as a result of peptide

bind-ing and aggregation in the presence of zinc are shown

in Fig 5A,B Control experiments showed that no

PWR spectral changes occurred as a result of Zn2+

interaction with the lipid bilayer in the absence of

pep-tide Figure 5A shows the spectral changes for

p-polar-ization at various early time-points up to 200 min after

the addition of 5 lm peptide The binding and

aggre-gation process was observed to follow biphasic

kinet-ics, with an initial shift to lower-incident angles that

occurred in
30 min, followed at later time-points by

a shift to higher angles accompanied by a decrease in

spectral amplitude for the lower-incident angle reson-ance (owing to the DOPC-rich domain), and an increase in amplitude and a large shift to higher angles for the resonance associated with the SM-rich domain (Fig 5A,B) These results suggest that a major change

in bilayer structure occurred as a result of peptide insertion into the bilayer and aggregation The spectral changes are consistent with a high degree of mass accumulation occurring predominantly within the SM-rich microdomain However, it is important to note that the bilayer remained intact (i.e no large holes were formed that exposed resonator surface to the aqueous phase) This is evidenced by the fact that

no resonances were observed corresponding to the bare prism surface in direct contact with the aqueous medium

The binding isotherms for peptide association with the bilayer are shown in Fig 5C,D, with the binding constants given in Table 1 Again, the initial binding process occurred with high affinity (inset to Fig 5C), similar to that observed in the absence of zinc How-ever, in this case the second phase of binding had a fivefold higher affinity than in the absence of zinc (Fig 5C; Table 1) The peptide interacted with the

0.4 0.6

0.8

A

1

Incident angle (deg)

B

0 10 20 30 40

Aβ (µM)

-20 0 20

40

C

Time (h)

Fig 3 Binding and aggregation of Ab in a

SM ⁄ cholesterol (1 : 0.35) bilayer (A)

p-Polar-ized PWR spectra are shown for a bilayer

before (curve 1) and after (curve 2; squares)

Ab was added to the sample cell (5 l M bulk

concentration in the aqueous compartment).

Curve 3 (triangles) shows the spectrum

after equilibration for 15 h Other conditions

were as in Fig 1 (B) Plot of the spectral

shifts associated with binding of the peptide

to the bilayer as a function of added Ab

con-centration The binding constant was

obtained by a hyperbolic fit to the data (solid

line) with a KDvalue and total spectral shift

as given in Table 1 (C) Plot of the time

course for p-polarization spectral changes

associated with peptide aggregation The

data were fit with a single exponential (solid

line), and the half-time and total spectral

shift are given in Table 2.

DOPC-rich domain with a single KD value 20-fold

weaker than with the SM-rich domain, and with a

smaller spectral shift (Fig 5D)

The kinetics of the spectral changes are shown in

Fig 6, and the half-times obtained by curve fitting are

given in Table 2 These reveal at least three kinetic

phases: two at early time-points with half-time values

of 0.53 ± 0.06 h and 2.96 ± 0.12 h; and a third,

slower, phase corresponding to a half-time of

17.9 ± 0.5 h Thus, the mechanism by which

aggrega-tion occurs is quite complex in this system, and further

studies will be necessary to obtain mechanistic insights

These results are consistent with the reported effect of zinc on the formation of a helical peptide structure that facilitates insertion and pore formation as a result of aggregation [22,37,38] Furthermore, the atomic force microscopy (AFM) studies of Lin et al [39], on solid-supported bilayers, have shown that the interaction of

Ab results in the formation of conducting ion channels that have rectangular or hexagonal shapes with four or six subunits, are 80–120 A˚ in diameter and protrude

10 A˚ from the bilayer surface The large PWR spec-tral changes shown in Fig 5 are consistent with struc-tures of this magnitude inserted into the bilayer

Fig 4 Binding and aggregation of Ab in a DOPC ⁄ SM ⁄ cholesterol (1 : 1 : 1) bilayer Spectra were obtained using a high-resolution sensor Other experimental conditions are as given in the legend to Fig 1 (A) p-Polarized PWR spectra are shown for the membrane before (curve 1) and after (curve 2; squares) addition of the peptide (5 l M bulk concentration in the aqueous cell compartment) Based on previous results [35], the resonance at smaller-incident angles is ascribed to a DOPC-enriched microdomain, and the resonance at larger incident angles to

an SM-enriched microdomain The initial addition of peptide resulted in binding to both microdomains, but to a larger extent in the SM-rich domain Spectra are also shown after the sample had equilibrated for 3 h (curve 3; circles) and 15 h (curve 4; triangles) These changes are ascribed to peptide aggregation (B,C) Plots of initial peptide binding, resulting in shifts to larger angles (inset to B) and binding at higher con-centrations resulting in shifts to smaller angles Binding to the SM-rich domain is shown in (B) and to the DOPC-rich domain in (C) Data were fit with three hyperbolic curves (solid lines), yielding the affinity constants and spectral shifts in Table 1 (D) Plots of the time course of spectral changes for the DOPC-rich microdomain (open triangles) and the SM-rich microdomain (closed triangles) associated with peptide aggregation Solid lines correspond to single exponential fits, with half-times and total shifts given in Table 2.

Spectral simulation and structural modeling

The purpose of PWR spectral simulation is to quantify

the changes in the optical properties of the membrane

that occur during the processes of lipid bilayer–peptide

binding and subsequent time-dependent peptide

aggre-gation This can provide insights into the changes in

mass density and structure of the system caused by

these events The simulation procedure is described in

detail in our previous publication [35] Briefly, it is

based on Maxwell’s equations that provide an

analyt-ical relationship between experimental spectral

parame-ters and the optical properties of the bilayer, the latter

defined by refractive index, n, extinction coefficient, k,

and thickness, t These three parameters can provide a unique fit to the experimental spectra (In these experi-ments, the k-value is caused by light-scattering effects and will be ignored.) In the present analysis, based on the structure of the peptide and the literature data, we have assumed a working model that allows the peptide

to either partially penetrate the lipid membrane with its hydrophobic tail (i.e residues 29–40), or to stay on the surface of the bilayer membrane without significant penetration This generates two separate layers, com-posed of peptide and of the lipid bilayer Using this two-layer model we have been able to quantify how much of the peptide molecule protrudes beyond the bilayer–water interface, and how much the lipid

Fig 5 Binding and aggregation of Ab in a DOPC ⁄ SM ⁄ cholesterol (1 : 1 : 1) bilayer in the presence of 1 m M zinc ions Spectra were obtained using a high-resolution sensor Other experimental conditions were as in Fig 1 (A) and (B) p-Polarized PWR spectra obtained for the bilayer

in the presence of 1 m M Zn ions, as a function of time, after the addition of Ab (5 l M bulk concentration in the sample cell) Initial binding occurred mainly in the SM-rich microdomain (data not shown) With time, additional resonance shifts occurred, mainly in the SM-rich micro-domain, along with large changes in amplitude The resonance for the DOPC-rich domain diminished in intensity, whereas that for the SM-rich domain increased in intensity and large shifts to longer angles occurred Note that the t ¼ 208 min resonance is repeated in both panels for reference purposes (C,D) Plots of initial peptide binding, resulting in shifts to larger angles (inset to C) and binding at higher concentra-tions resulting in shifts to smaller angles Binding to the SM-rich domain is shown in (C) and to the DOPC-rich domain in (D) Data were fit with three hyperbolic curves (solid lines) to yield affinity constants and spectral shifts as given in Table 1 Initial insertion into the SM-rich domain was followed by incorporation into both microdomains.

membrane properties have been changed by the

pep-tide–bilayer interaction This is accomplished by

evalu-ation of the optical parameters for the lipid layer

before and after binding of peptide and subsequent

aggregation, and of the layer composed of peptide (or

a portion of the peptide) that extends beyond the lipid

surface Both p- and s-polarized data were used in such

simulations Figure 7 shows an example of the

simula-ted spectra superimposed on the experimental spectra,

obtained with s-polarized red light (k¼ 632.8 nm)

The optical parameters resulting from such simulations

for bilayers composed of DOPC, of SM and of

SM⁄ cholesterol are summarized in Tables 3–5 for the

binding and aggregation processes The error limits

shown in these tables correspond to the standard

errors obtained from the fitting procedure It must be

noted that we have not been able to satisfactorily

simulate the spectra obtained with the DOPC⁄

SM⁄ cholesterol bilayers, with and without zinc, using

the high-resolution resonator This requires additional

information about the effects of varying amounts of

cholesterol on DOPC and SM bilayers, which we do

not have at present Thus, further experiments are

nee-ded in order to accomplish this Therefore, for the time

being we have limited our discussion to qualitative

aspects of Ab binding and aggregation in this system,

as presented above

There are several important conclusions that can be

obtained from the values in Tables 3–5 First, the

bilayer membranes consisting of DOPC, SM, or

SM⁄ cholesterol have quite different optical parameters

before peptide has been bound (Table 3) This has been observed previously for DOPC and SM bilayers [35] and discussed in the context of the membrane structure and properties The present work has now extended this to an SM bilayer containing 0.35 mole percent cholesterol Thus, the refractive indices clearly indicate that the bilayer consisting of DOPC is much less densely packed with lipid molecules than that of either SM or SM⁄ cholesterol (note that refractive index is proportional to mass density, i.e molecules per unit surface area) The lipid-packing density influ-ences the ordering of hydrocarbon tails, resulting in a higher degree of order in the latter two membranes, and this difference is reflected in an increased thickness

of both membranes In addition, it is clear that the presence of cholesterol significantly increases both the packing density of lipid molecules (as shown by the increase in refractive index) as well as the thickness

Fig 6 Plot of the time course of spectral changes occurring in the

SM-rich domain The solid line is a fit to the data with three

expon-entials; half-times and spectral shifts are given in Table 2.

0.75 0.85 0.95

Incident angle, deg

1

2

3

Fig 7 Examples of simulated spectra (solid lines) superimposed

on experimental spectra (symbols), obtained with a SM ⁄ cholesterol membrane using s-polarized red light Spectra are shown for the membrane before (curve 1) and after (curve 2) addition of peptide Curve 3 was obtained after aggregation.

Table 3 Optical parameters of lipid bilayers prior to Ab binding DOPC, dioleoylphosphatidylcholine; np, p-polarized refractive index;

ns, s-polarized refractive index; SM, sphingomyelin; tav, average thickness.

Bilayer n p (± 0.003) n s (± 0.002) t av (nm) (± 0.1)

of the SM bilayer membrane Therefore, peptide

bind-ing is occurrbind-ing to three different membrane structures

in these studies

The data presented in Table 4 show that the binding

interaction of the peptide with the three membranes

creates different effects within the lipid layer and in

the peptide layer Although the peptide layers in both

the DOPC and SM membranes are similar in

thick-ness, indicating that the peptide protrusions from the

bilayer surface are similar, they differ in the values of

the refractive indices, indicating a higher peptide

sur-face mass density in the case of the SM membrane as

compared with the DOPC membrane It is also

important to note that, in both of these cases, the

val-ues of the refractive indices are the same for both

polarizations This strongly supports the idea that the

internal structures of the peptide layers are quite

iso-tropic, without any significant degree of long-range

molecular order Based on this, one can conclude that

these layers are formed from anisotropic molecular

structures which are arranged in the peptide layer in a

random way, thereby creating an average isotropic

dis-tribution of conformations It is worth noting that

such anisotropies can result from either molecular

con-formation or molecular orientation, or both; these

can-not be distinguished by the present methods

In none of the three membrane systems did the

ini-tial binding of peptide cause any significant changes in

the lipid bilayer parameters (Table 4) This suggests

that the peptide was not anchored deeply within these

membranes, indicating that the bulk of the peptide

mass remained largely on the surface of the membrane

However, the peptide layer was appreciably different

for the SM⁄ cholesterol membrane, having a higher mass density, a significant anisotropy (np> ns), and a larger thickness Thus, more peptide was bound when cholesterol was included in the bilayer, and the struc-ture of the peptide layer was different In order to explain the anisotropic nature of the peptide layer, it seems reasonable to assume that some of bound pep-tides had an extended conformation (b-sheet like) and bound with their long axis perpendicular to the plane

of the lipid membrane Furthermore, the 4.7 nm thick-ness of the peptide layer indicates that the hydropho-bic tail of the peptide must be buried within the lipid bilayer This type of arrangement of the bound peptide

is in agreement with previous studies showing that the addition of cholesterol to a DOPC⁄ SM bilayer conver-ted a b-sheet Ab form into a-helix structures, a change that was necessary to allow incorporation of the C-ter-minal tail into the membrane [40] Such incorporation

is probably driven by hydrophobic interactions caused

by the high nonpolar amino acid content in the C ter-minus of Ab A hydrophobic peptide segment would

be expected to have an orientation and anisotropic refractive index values similar to both the fatty acyl tails of the lipid and the cholesterol molecule, and thus would not produce much change in the lipid portion

of the bilayer However, the peptide layer would be appreciably altered by such a conformational change

of the peptide

It is generally accepted that carefully prepared fibril-free Ab in an aqueous environment exists mainly as monomers, dimers, trimers, and tetramers, in a rapid equilibrium, within which the predominant secondary structural element is random coil with smaller amounts

Table 4 Optical parameters of lipid and peptide layers after amyloid b1)40peptide (Ab) binding chol, cholesterol; DOPC, dioleoylphosphat-idylcholine; np, p-polarized refractive index; ns, s-polarized refractive index; SM, sphingomyelin; tav, average thickness.

Bilayer

np(± 0.003) ns(± 0.002) tav(nm) (± 0.2) np(± 0.005) ns(± 0.005) tav(nm) (± 0.2)

Table 5 Optical parameters of lipid and peptide layers after amyloid b 1 )40peptide (Ab) aggregation chol, cholesterol; DOPC,

dioleoylphos-phatidylcholine; np, p-polarized refractive index; ns, s-polarized refractive index; SM, sphingomyelin; tav, average thickness.

Bilayer

n p (± 0.003) n s (± 0.002) t av (nm) (± 0.2) n p (± 0.005) n s (± 0.005) t av (nm) (± 0.2)

of b-sheet, and even smaller amounts of a-helix [41].

We therefore assume that the peptide layer on the

DOPC and SM membrane surface consists of such a

mixture of oligomers and secondary conformations

This accounts for the isotropic nature of this layer in

these two systems It is interesting, however, to note

that both the binding affinity and the amount of

accu-mulated peptide (i.e mass density) were significantly

higher with the SM bilayer than with the DOPC

bilayer This is presumably a result of differences in

both the type of lipid molecule (i.e sphingolipid versus

glycerolipid), as well as the topology and morphology

of the surfaces of each of the bilayers This requires

further study

The differences between the membranes increased

further with time after the initial binding process As

noted above, and as shown in Table 5, after 15 h there

were no significant changes in either the bilayer or the

peptide layer with the DOPC membrane (i.e no

pep-tide aggregation occurred) In contrast, changes

occurred in both of these layers with the SM

mem-brane Thus, the surface mass density of the peptide

layer decreased, as reflected by the smaller value of ns,

and a significant refractive index anisotropy was

induced (np> ns) A corresponding increase in the

mass density in the lipid bilayer occurred, without a

significant change in anisotropy These results clearly

indicate that some peptide mass was inserted into the

lipid membrane, and that the mass distribution within

the peptide layer was altered In order for the latter to

occur, a significant conformational rearrangement

must take place, involving either creating more

aniso-tropic conformations or reorganizing the existing

anisotropic conformations so as to create long-range

molecular order, or both A plausible interpretation of

these effects is that some of the surface-bound peptide

(possibly monomeric forms with a b-sheet structure)

created higher oligomeric structures that inserted their

hydrophobic tails into the interior of the membrane

We suggest that this process was favored, in the case

of the SM membrane, because the surface mass density

(i.e concentration) of peptide was much higher than

with the DOPC membrane

As can be seen from the data presented in Table 5,

the aggregation of the peptide in the lipid membrane

consisting of SM⁄ cholesterol resulted in large decreases

in the refractive index parameters for the lipid layer

This indicates a significant expulsion of lipid mass

(presumably SM molecules) from the bilayer Thus,

the aggregates formed in the presence of cholesterol

must have a structure that occupies a larger fraction of

the bilayer volume than the unaggregated peptide, in

contrast to those formed in the absence of cholesterol

This suggests that cholesterol may have been incorpor-ated into these aggregincorpor-ated peptide structures, which requires further study It is also worth noting that the insertion and aggregation of Ab with membranes con-taining at least 30% cholesterol has been shown to form channel-like structures [42] This is consistent with our observation of lipid removal from the bilayer upon peptide insertion and aggregation in the

SM⁄ cholesterol system

In summary, the data presented here provide strong support for a mechanism in which peptide aggregation requires an SM-rich environment and occurs largely

on the bilayer surface in the absence of cholesterol The presence of cholesterol facilitates peptide insertion into the bilayer and promotes the aggregation process This leads to the formation of a less densely packed bilayer The presence of Zn2+ also enhances insertion and aggregation, and promotes large bilayer structural changes by peptide aggregates resulting in a more por-ous membrane The large effects of SM and cholesterol

on peptide insertion and aggregation are consistent with the reported occurrence of the amyloid precursor protein and Ab in neuronal membrane rafts [20– 23,29,30]

Experimental procedures

Materials

Solid-supported lipid bilayers (DOPC, SM, cholesterol and their mixtures; Avanti Polar Lipids, Birmingham, AL, USA) were made using solutions of either the single lipid components or mixtures containing various molar ratios

of lipids (10 mgÆmL)1 total lipid concentration) in buta-nol⁄ squalene (10 : 0.1, v ⁄ v) The buffer solution in contact with the bilayer in the sample cell for all the experiments was 10 mm Tris (pH 7.4) at 25
C, either in the absence of additional salt ions or in the presence of 1 mm Zn2+ions The 40-residue amyloid peptide (Ab1)40) was synthesized by standard solid-phase Fmoc chemistry, subsequently purified

by HPLC and found to be over 90% pure, and character-ized by MALDI-TOF mass spectroscopy It was stored in

a lyophilized form by a combined trifluoroacetic acid (TFA)⁄ trifluoroethanol (TFE) protocol to keep it in a monomeric form prior to further use [19,28]

PWR spectroscopy

The principles of PWR spectroscopy have been thoroughly described in previous publications [34–36,43,44] Here, we will briefly review those aspects that are especially relevant

to the present work Figure 8 shows a schematic diagram

of a PWR spectrometer Resonances can be excited with