1. Trang chủ
  2. » Luận Văn - Báo Cáo

Báo cáo khoa học: Secondary structure conversions of Alzheimer’s Ab(1–40) peptide induced by membrane-mimicking detergents pdf

12 268 0
Tài liệu đã được kiểm tra trùng lặp

Đang tải... (xem toàn văn)

Tài liệu hạn chế xem trước, để xem đầy đủ mời bạn chọn Tải xuống

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 12
Dung lượng 465,14 KB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

By further addition of SDS or lithium dodecyl sulfate reaching concentrations close to the critical micellar concentration, CD, NMR and FTIR spectra show that the peptide rearranges to f

Trang 1

peptide induced by membrane-mimicking detergents

Anna Wahlstro¨m1,*, Loı¨c Hugonin1,*, Alex Pera´lvarez-Marı´n1,*, Ju¨ri Jarvet2 and Astrid Gra¨slund1

1 Department of Biochemistry and Biophysics, The Arrhenius Laboratories for Natural Sciences, Stockholm University, Sweden

2 The National Institute of Chemical Physics and Biophysics, Tallinn, Estonia

Introduction

The amyloid b peptide (Ab) is the major component of

the amyloid plaques, which are found in the brains of

Alzheimer’s disease patients The Ab-peptide is a

39–42 residue peptide cleaved by processing of the

amyloid-b precursor protein [1,2] The Ab(1–40)

peptide has a hydrophilic N-terminal domain and a

more hydrophobic C-terminal domain, and contains a

central hydrophobic cluster (residues 17–21) suggested

to play an important role in peptide aggregation Solu-ble oligomeric peptide aggregates are reported to medi-ate toxic effects on neurons and synapses [1,3] and have attracted growing interest because of their proba-ble link to the pathology of the disease The formation

of aggregates occurs in parallel with a conformational change of the peptide structure to b-sheet

In vitro, the Ab monomer is in a dominating random coil secondary structure in solution at room temperature and physiological pH [4–7] However, in

Keywords

amyloid b peptide; CD; NMR; oligomer; SDS

Correspondence

A Gra¨slund, Department of Biochemistry

and Biophysics, The Arrhenius Laboratories

for Natural Sciences, Stockholm University,

SE-10691 Stockholm, Sweden

Fax: +46 8 155597

Tel: +46 8 162450

E-mail: astrid@dbb.su.se

*These authors contributed equally to this

work

(Received 29 April 2008, revised 8 August

2008, accepted 13 August 2008)

doi:10.1111/j.1742-4658.2008.06643.x

The amyloid b peptide (Ab) with 39–42 residues is the major component of amyloid plaques found in brains of Alzheimer’s disease patients, and solu-ble oligomeric peptide aggregates mediate toxic effects on neurons The Ab aggregation involves a conformational change of the peptide structure to b-sheet In the present study, we report on the effect of detergents on the structure transitions of Ab, to mimic the effects that biomembranes may have In vitro, monomeric Ab(1–40) in a dilute aqueous solution is weakly structured By gradually adding small amounts of sodium dodecyl sulfate (SDS) or lithium dodecyl sulfate to a dilute aqueous solution, Ab(1–40) is converted to b-sheet, as observed by CD at 3C and 20 C The transition

is mainly a two-state process, as revealed by approximately isodichroic points in the titrations Ab(1–40) loses almost all NMR signals at dodecyl sulfate concentrations giving rise to the optimal b-sheet content (approxi-mate detergent⁄ peptide ratio = 20) Under these conditions, thioflavin T fluorescence measurements indicate a maximum of aggregated amyloid-like structures The loss of NMR signals suggests that these are also involved

in intermediate chemical exchange Transverse relaxation optimized spec-troscopy NMR spectra indicate that the C-terminal residues are more dynamic than the others By further addition of SDS or lithium dodecyl sulfate reaching concentrations close to the critical micellar concentration,

CD, NMR and FTIR spectra show that the peptide rearranges to form a micelle-bound structure with a-helical segments, similar to the secondary structures formed when a high concentration of detergent is added directly

to the peptide solution

Abbreviations

Ab-peptide, amyloid b peptide; D ⁄ P, detergent to peptide ratio; HSQC, heteronuclear single quantum coherence; LiDS, lithium dodecyl sulfate; ppII, polyproline II; SDS-d25,deuterated SDS; ThT, thioflavin T; TROSY, transverse relaxation optimized spectroscopy.

Trang 2

membrane-mimicking environments, such as SDS

micelles, the Ab-peptide displays an a-helical structure,

with two a-helical segments comprising residues 15–24

and 29–35, separated by a flexible hinge [8], and less

structured N- and C-termini In the presence of

phos-pholipid vesicles, a-helical structures as well as b-sheet

structures have been reported [9] Rangachari et al

[10] have described interfacial aggregation of Ab(1–40)

at a polar⁄ nonpolar interface, with a concomitant

increase in b-structure content, brought about by SDS

micelles In line with this finding, it was recently shown

that the Ab(1–40) and Ab(1–42) peptides form

b-sheet-rich aggregates at SDS concentrations significantly

below the critical micellar concentration [11] These

aggregates give rise to thioflavin T (ThT) fluorescence

and are neurotoxic

In the present study, we report on further properties

of soluble oligomeric b-sheet-rich Ab(1–40) aggregates,

formed at submicellar SDS or lithium dodecyl sulfate

(LiDS) concentrations at detergent⁄ peptide ratios of

approximately 20 The results obtained by CD, NMR,

FTIR and ThT fluorescence are compared and

inter-preted in terms of mixed micelle-like aggregates with

amyloid properties at intermediate detergent

concen-trations, where the peptides show dynamic properties,

particularly in the C-termini

Results

Structural transitions of the Ab-peptide induced by

increasing concentrations of membrane-mimicking

detergents (SDS or LiDS) were studied by CD, NMR

and FTIR spectroscopy at temperatures in the range

3–25C LiDS was used at low temperature

measure-ments because it has a higher solubility at lower

tem-peratures than SDS; however, the critical micellar

concentration is approximately the same for the two

detergents [12,13]

CD spectroscopy

Detergent titration experiments were performed on a

sample with 75 lm Ab(1–40) peptide in 10 mm sodium

phosphate buffer at 3C and 20 C and pH 7.2 The

structural starting point for Ab(1–40) varies to some

extent as a function of temperature At 3C, the

secondary structure includes contributions from a

poly-proline II (ppII) helix, whereas, at 20C, the

second-ary structure is almost exclusively random coil, as

described previously [5]

Figure 1A shows the titration of the Ab(1–40)

pep-tide with microvolumes of LiDS at 3C over a

deter-gent concentration interval in the range 0.05–20 mm,

corresponding to detergent⁄ peptide (D ⁄ P) ratios of 0.7–267, respectively The CD data report on a first structural conversion from a mixture of ppII helix and random coil (weak positive shoulder at approximately

220 nm and negative minimum at 198 nm) occurring

at low LiDS concentrations (up to 0.7 mm, D⁄ P = 9)

to a signal appearing at 1.6 mm LiDS (D⁄ P = 21) with a maximum at 195 nm and a minimum at

218 nm, indicative of a dominating b-sheet structure

It should be noted that, up to this titration point, the spectra show a relatively well defined isodichroic point, implying a two-state transition between the initial structure and the b-sheet structure After increasing the LiDS concentration further (3.0 mm LiDS,

D⁄ P = 40), a new state is observed, mostly consisting

of a-helix structure The conversion to a-helix struc-ture reached its final state at 20 mm LiDS with a char-acteristic maximum and two minima at 193 and

208⁄ 222 nm, respectively

Figure 1B shows the SDS titration experiment at

20C In the absence of SDS, Ab(1–40) gives a CD spectrum with a minimum at 198 nm, indicating a pre-dominantly random coil secondary structure As the detergent concentration was increased, the CD signal disappeared in the wavelength region around 198 nm (SDS concentration of approximately 4 mm,

D⁄ P = 53) Further increase of the SDS concentration (up to 5 mm, D⁄ P = 67) yielded a b-sheet spectrum with a positive maximum at 194 nm and a negative minimum at 218 nm Also at this temperature, there was a relatively well-defined isodichroic point in the titration; however, this was not as clear as in the 3C titration At high SDS concentrations (above 10 mm SDS, D⁄ P = 133), the secondary structure was mainly a-helix, with a characteristic maximum at 192 nm and two minima at 208 and 221 nm

The mean residual molar ellipticity at 195 nm as a function of detergent concentration at 3C and 20 C

is shown in Fig 1C The disappearance of an initial weakly structured state and conversion to b-sheet and then to a-helix are evident The CD intensities at this wavelength allowed us to compare the detergent secondary structure induction at 3C and 20 C (Fig 1C) Only one transition was visible with a mid-point at 1 mm LiDS at 3C At 20 C and with SDS, three transitions could be distinguished The first had a midpoint at 0.7 mm, followed by two more transitions, with midpoints at 2.1 and 4.6 mm SDS

Figure 1D shows the corresponding curves for the mean residual ellipticity at 208 nm as function of detergent concentration At 3C with LiDS, the data show two sigmoidal transitions The first sigmoidal transition (positive) occurred in the range 0–1.6 mm

Trang 3

LiDS with a midpoint at 0.7 mm, corresponding to the

transition from initial structure to the structure

domi-nated by b-sheet The second sigmoidal transition

(neg-ative) had a midpoint at 2.6 mm We interpret this as

corresponding to the transition from b-sheet to a-helix

In the SDS titration at 20C, the intensity at 208 nm

again indicated three sigmoidal transitions Two

posi-tive transitions had midpoints at 0.9 and 2 mm,

respec-tively, and the third (negative) had a midpoint at

6 mm It should be noted that three transitions are

visible at 20C at both wavelengths studied, and that the SDS concentration midpoints are in approximate agreement: the first transition at approximately 0.8 mm SDS (D⁄ P = 11), the second one at approximately

2 mm SDS (D⁄ P = 27) and the third one at approximately 5 mm SDS (D⁄ P = 67) The third transition probably involves the formation of the partly a-helical state, whereas the two first may involve two similar but distinguishable states with b-sheet structures

Fig 1 Circular dichroism spectra of 75 l M Ab(1–40) peptide in 10 m M phosphate buffer at pH 7.2 in the presence of different concentra-tions of detergent (A) At 3 C in LiDS: open square, buffer; open circle, 0.05 m M ; open triangle, 0.1 m M ; filled square, 0.3 m M ; open diamond, 0.5 m M ; filled circle, 0.7 m M ; filled hexagon, 1.0 m M ; open hexagon, 1.3 m M ; open star, 1.6 m M ; cross, 2.0 m M ; filled star, 3.0 m M ; open pentagon, 20 m M (B) At 20 C in SDS: open square, buffer; open circle, 0.1 m M ; filled star, 0.8 m M ; open triangle, 2.0 m M ; open pentagon, 3.8 m M ; filled square, 4.2 m M ; open diamond, 4.3 m M ; filled circle, 5.0 m M ; filled triangle, 6.2 m M ; open hexagon, 7.0 m M ; open star, 12.2 m M (C) Plot of the mean residual molar ellipticity at 195 nm for the experiment in LiDS at 3 C (filled square) and for the experiment in SDS at 20 C (open circle) (D) Plot of the mean residual molar ellipticity at 208 nm for the experiment in LiDS at 3 C (filled square) and for the experiment in SDS at 20 C (open circle).

Trang 4

NMR spectroscopy

Heteronuclear single quantum coherence (HSQC) and

transverse relaxation optimized spectroscopy (TROSY)

NMR spectroscopy were used to follow the structural

transitions of the Ab(1–40) peptide (75 lm) induced by

increasing concentrations of the membrane-mimicking

detergent LiDS The 1H-15N HSQC spectrum of

uniformly 15N-labeled Ab(1–40) in 10 mm phosphate

buffer (pH 7.2, 3C) at the beginning of a titration is

shown in Fig 2 (left) The corresponding spectrum of

the peptide in 128 mm LiDS at the end of a titration is

also shown in Fig 2 (right, green spectrum) There are

significant chemical shift differences in comparison to

the initial state Figure 2 (right) also includes the

HSQC spectrum of the peptide after direct addition of

150 mm LiDS at 3C (red spectrum) The two spectra

shown in Fig 2 (right) were found to overlap very well

with one another However, the intensities (when

corrected for different peptide concentrations) were

significantly smaller in the spectrum after titration

Assignments of the amide groups of Ab(1–40) in

buffer (Table S1) were made by comparison with the

previous assignment [14] Assignment of Ab(1–40) in

150 mm LiDS at 3C (Table S1) was performed by

starting the NMR experiment at 25C where assign-ments are known [8] and decreasing the temperature

by 5C at a time following the gradual changes of the HSQC spectra The similarity of chemical shift patterns at 3C and 25 C suggests that the previously determined a-helical regions involving residues 15–24 and 29–35 are the same at the two temperatures after direct addition of a high concentration of detergent [8] Between the two well defined states shown in Fig 2 (i.e at an intermediate detergent concentration), a new state of the peptide characterized by complete NMR signal loss was observed This state occurred at a criti-cal concentration of LiDS of 1–2 mm, corresponding

to D⁄ P = 13–27

There was no obvious change in chemical shifts, nor linewidth, of the amide HSQC crosspeaks by the grad-ual titration with detergent below the concentration inducing signal loss To study how the signal was influ-enced by an increasing concentration of detergent, the volume of each crosspeak was integrated In a titration series with small titration steps (0.05, 0.1, 0.2, 0.3, 0.5, 0.7, 1, 2, 10 and 20 mm), most of the signals were unchanged or slowly decayed up to a LiDS concentra-tion of 0.5 mm However, beyond 0.5 mm LiDS, the signal from every residue abruptly decreased (Fig 3)

Fig 2 HSQC NMR spectra and assignment of amide crosspeaks for the Ab(1–40) peptide, and the effect of added lithium dodecyl sulfate Left:1H-15N HSQC spectrum of 75 l M uniformly 15N-labeled Ab(1–40) in 10 m M phosphate buffer The two peaks (V39 and V40) found in the TROSY experiment with 75 l M15N-Ab(1–40) in the presence of 2 m M LiDS are indicated with arrows Right: overlay of HSQC spectra;

75 l M15N-labeled Ab(1–40) in 128 m M of LiDS (i.e the end point in the titration series 0, 0.5, 1, 4, 8, 16, 32, 64 and 128 m M LiDS) (green spectrum) and 300 l M15N-labeled Ab(1–40) in 150 m M of LiDS, added in one addition (red spectrum) The peak intensities are corrected in relation to the different peptide concentrations All measurements were performed in 10 m M phosphate buffer at pH 7.2 and 3 C.

Trang 5

At 1 mm LiDS, corresponding to D⁄ P = 13, almost

all the HSQC crosspeaks had disappeared and, at

2 mm, all were lost At LiDS concentrations of 10 and

20 mm, the crosspeaks reappeared, directly with

chemi-cal shifts closely corresponding to those observed after

direct addition of 150 mm LiDS (Fig 2, right, red

spectrum)

The crosspeaks from the amide groups in the amino

acid residues in the N- and C-terminal ends returned

with the strongest signals upon titration with detergent

(Fig 3) This is probably due to an increased mobility

in the N- and C-terminal end segments (i.e residues

up to G9 and beyond G37) The chemical shifts observed at detergent concentrations of 10 and 20 mm were retained in the presence of the higher LiDS con-centrations of 64 and 128 mm, which all coincide with the chemical shifts found at 150 mm LiDS (Fig 2) The disappearance of all NMR peaks at detergent concentrations of 1–2 mm may have more than one explanation An obvious reason for signal loss is that

Fig 3 The crosspeak signal intensity of assigned residues of15N-labeled Ab(1–40) in the1H-15N HSQC spectra as a function of LiDS con-centration (0, 0.05, 0.1, 0.2, 0.3, 0.5, 0.7, 1, 2, 10, 20 and 128 m M ) at pH 7.2 and 3 C The volumes of the HSQC crosspeaks of 75 l M of

15 N-Ab(1–40) were integrated The amino acids are divided into different figures according to the earlier findings indicating that residues 15–24 and 29–35 have a-helical structure, whereas the regions in the ends and in between are unstructured [8] The x-axis (LiDS concentra-tion) is shown as a logarithmic scale.

Trang 6

large, probably heterogeneous, aggregates of peptide

and detergent are formed Exchange on an

intermedi-ate time scale between aggregintermedi-ates of different sizes

may also contribute To study this state further,

TRO-SY experiments and translational diffusion NMR

experiments were performed TROSY makes it

possi-ble to study larger proteins or complexes because it

reduces transverse relaxation rates [15] In the TROSY

spectrum in the presence of 2 mm LiDS (i.e at the

same conditions where all HSQC signals disappeared),

two C-terminal residues were observable (V39 and

V40) (Fig 2, left) This observation suggests a higher

mobility of the charged carboxy terminus of Ab(1–40)

in the aggregated state

NMR translational diffusion experiments were

per-formed to investigate whether NMR-visible complexes

of different sizes could be observed during the

deter-gent titration The results (Fig 4) revealed a diffusion

coefficient of 1.29· 10)10m2Æs)1 for Ab(1–40) in D2O

buffer (pH 7.4, 25C), indicating that the observable

peptide is monomeric [4] The diffusion coefficient did

not change significantly from this value in the presence

of 0.1, 0.5 and 1 mm deuterated SDS (SDS-d25) The

NMR signal disappeared abruptly at 2 mm SDS-d25,

and a diffusion coefficient could not be determined for

this condition At 5 mm, the resonances had

reap-peared in the ‘new’ positions and the associated

diffusion coefficient had decreased This implies

forma-tion of an assembly of peptides, probably also in

complex with detergent molecules (Fig 4) At the same

time, some fibrils could be seen in the sample solution

The diffusion coefficient for Ab(1–40) in the presence

of 10 mm SDS-d25was 0.85· 10)10m2Æs)1, which can

be compared with the diffusion coefficient 0.48· 10)10

m2Æs)1 for 100 lm Ab(1–40) in directly added 150 mm SDS, which comprises a state when one peptide is bound to one micelle [8]

FTIR spectroscopy FTIR spectroscopy was used to obtain further infor-mation about the striking changes in the secondary structure of Ab(1–40) observed at concentrations in the range of 0–4 mm SDS or LiDS The amide I¢ region in the IR spectrum is indicative for the second-ary structure of the peptide It has been shown that Ab(1–40) has a strong secondary structure concentra-tion dependence [16] To increase the signal-to-noise ratio and to eliminate contributions of the baseline drift, the concentration of peptide was as low as possi-ble (100 lm), and only slightly higher than the CD and NMR concentrations The negative second derivative

of the spectra in the amide I¢ band is shown in Fig 5 Assignment of different secondary structures was performed according to Byler and Susi [17] The spectra indicate that, at 20C, the peptide had a mixture of random coil and b-sheet secondary struc-ture in the absence of SDS and with SDS at a low

D⁄ P ratio of 1 (0.1 mm SDS) At 1.4 mm, the random coil contribution disappeared, the b-sheet contribution decreased and a-helix structure became evident At a

Fig 4 The translational diffusion coefficient (Dt) of 75 l M Ab(1–40)

versus increasing SDS-d 25 concentration (0, 0.1, 0.5, 1, 5 and

10 m M ) The experiment was performed in 10 m M phosphate

buffer at 25 C and pH 7.4 , diffusion coefficient for 100 l M

Ab(1–40) in 150 m M SDS at 25 C [8] The gray box indicates the

conditions for which a diffusion coefficient could not be determined

due to signal loss.

Fig 5 Secondary structure induction by SDS of 100 l M Ab(1–40)

in 10 m M phosphate buffer at pH 7.2 and 20 C The negative second derivative of the peptide in the presence of different SDS concentrations is shown: thick black line, 0 m M ; thin black line, 0.1 m M ; gray line, 1.4 m M ; light gray line, 10 m M The spectra were normalized for trifluoroacetic acid intensity (as indicated by an aster-isk) The wavenumber intervals corresponding to the specific secondary structures are also indicated.

Trang 7

high SDS concentration (10 mm), the peptide showed

a predominantly a-helix structure, with a shoulder in

the b-sheet region

ThT interactions

The titration of 75 lm Ab(1–40) with increasing

amounts of SDS at 20C was monitored by ThT

(15 lm) fluorescence, a classical probe for aggregated

amyloid material [18] Figure 6 shows a titration curve

yielding maximum ThT fluorescence (approximately

8· initial intensity) at 2.2 mm SDS Further addition

of SDS decreased the fluorescence intensity to

approxi-mately 3· initial intensity (at 45 mm SDS) The lack

of complete reversal of ThT fluorescence indicates the

presence of some remaining amyloid-like material also

at higher SDS concentrations, although most of the

aggregation appears to have been reversed As a

control, an SDS titration of ThT in the absence of

Ab(1–40) was also performed (Fig 6) This experiment

showed that ThT fluorescence is generally low

com-pared to the results in the presence of the peptide

However, also in the absence of peptide, ThT

fluores-cence under SDS titration follows a sigmoid curve,

with a midpoint at approximately 2 mm This

behav-iour of ThT is in general agreement with previous

results obtained for ThT interacting with anionic

micelles [19] To further characterize the state of the

Ab(1–40) during the SDS titration, five representative

SDS titration points were chosen (0, 1.1, 2.2, 4.6 and 25.5 mm SDS) for investigation by native-PAGE (Fig S1) A preliminary qualitative assessment of the gels could be performed with respect to the presence of low and high molecular weight species in the different samples [20] Whereas the 0 and 25.5 mm SDS samples had a relatively high population of low molecular weight species (presumably peptide monomers), the samples prepared with intermediate SDS concentra-tions showed mainly high molecular weight (aggre-gated) peptide species

Discussion

By combining CD, NMR and FTIR experiments, we have shown that the aggregation process of Ab(1–40) induced by LiDS or SDS detergent gives rise to a variety of secondary structure states, each of which is relatively stable under its given conditions It is demonstrated that the extreme variability of the secondary structure of the peptide is dependent on its environment

The CD results reveal that, in a dilute aqueous solu-tion, Ab(1–40) has a dominating random coil second-ary structure with a low contribution of ppII helix and b-sheet at low temperature Titrations with SDS or LiDS show that a secondary structure conversion of Ab(1–40) can be described essentially as a two-state process, involving conversion of the initial weak struc-ture to b-sheet-rich strucstruc-tures Continued addition of SDS or LiDS, reaching concentrations close to the critical micellar concentration, induces rearrangement

of the peptide structure to a structure with a-helix con-tributions

The NMR results at 3C show that the Ab(1–40) peptide retains its random coil⁄ ppII structure free in solution in the presence of low detergent concentra-tions in the range 0.05–0.5 mm At a detergent concen-tration of 1–2 mm, on the other hand, the NMR signal

is essentially lost and the results suggest peptide aggre-gation and possibly intermediate chemical exchange Preliminary light absorption observations (data not shown) suggest considerable light scattering under these conditions, in agreement with the formation of large particles

A high molecular weight state induced by submi-cellar concentrations of detergent was also recently observed by Tew et al [11] using CD and NMR They observed that the 1D 1H-NMR spectrum disappeared

at a certain SDS concentration but showed up again at SDS concentrations above the critical micellar concen-tration In the present study, we aimed to analyze the aggregated state further after assignment of the amide

Fig 6 SDS titration monitored by ThT fluorescence SDS titrations

in the absence of peptide (open squares) and in the presence of

75 l M Ab(1–40) (open and filled circles) showing the fluorescence

changes of 15 l M ThT in 10 m M phosphate buffer at pH 7.3 and

20 C The SDS concentrations are: 0, 0.09, 0.4, 0.6, 1.1, 1.7, 2.2,

2.8, 4.6, 6.5, 8.4, 10.3, 14.1, 17.9, 25.5 and 45 m M Full circles

indi-cate the concentrations analyzed by native-PAGE (see Supporting

information, Fig S1).

Trang 8

HSQC crosspeaks We were able to follow the

intensity changes for the individual amino acids at

increasing LiDS concentrations We conclude that the

signals only partly recover towards high detergent

concentrations At and above the critical micellar

concentration, the recovered spectrum displays amide

chemical shifts very similar to those seen in a sample

after direct addition of a high concentration of

deter-gent The amide groups of N- and C-terminal residues

return to visibility with the strongest signal intensities,

implying higher mobility at the peptide terminus This

observation is also strengthened by the TROSY

mea-surement showing weak amide crosspeaks from V39

and V40 at conditions where no peaks are visible in

the HSQC experiment Under the same conditions

with high detergent concentrations, the CD spectra

provide evidence of significant a-helix formation

The a-helical state of Ab(1–40) in SDS micelles has

been characterized previously by NMR It was found

to consist of two a-helical segments, involving residues

15–24 and 29–35, respectively, of which the C-terminal

helix is inserted into the micelle [8] It is interesting to

compare this structure with two earlier proposals of

full length Ab structures (1) In complex with an

affi-body protein, Ab(1–40) forms a hairpin between

resi-dues 17–36, where resiresi-dues 24–29 apparently form the

loop connecting the two b strands [21] (2) A model

based on solid state and solution state NMR for a

fibril formed by Ab(1–42) showed a parallel⁄ in-register

b-sheet arrangement between residues 18–26 and 31–42

[22] Obviously, there are two segments of Ab

[approx-imately 16–25 and 30–36⁄ 42 in Ab(1–40) and

Ab(1–42), respectively] that are prone to form stable

hydrogen bonds We hypothesize that these segments

therefore easily form secondary structures; with

affibodies or in fibrils, they may form a b-sheet and,

with dodecyl sulfate, they may form a-helices

The NMR diffusion measurements revealed that, up

to a detergent concentration of 1 mm, Ab(1–40) is

monomeric and then a state follows that cannot be

characterized by diffusion NMR, but probably

involves large aggregates Continued titration with

SDS, reaching a concentration of 5–10 mm, induces

micelle-like formations, which appear to have a more

rapid translational diffusion than the better defined

state at an SDS concentration of 150 mm, when one

peptide is associated with one micelle of normal SDS

micellar size [8]

FTIR spectroscopy indicates the presence of some

b-sheet structure in addition to random coil when

Ab(1–40) is dissolved in dilute aqueous buffer The

NMR and CD measurements report mainly random

coil under similar conditions This is despite the careful

procedures performed when preparing the peptide solutions as described in the Experimental procedures

A possible explanation for this discrepancy is the exis-tence of small amounts of very large aggregates, or seeds, in the sample, which remain in the sample prep-aration and are not detectable by NMR or CD At 1.4 mm SDS (D⁄ P = 14), the IR band indicating a b-sheet is transformed into two shoulders, which might represent the seeds and the b-sheet containing com-plexes of Ab(1–40) and detergent molecules, respec-tively These observations emphasize the problems encountered in spectroscopic studies with respect to an aggregating peptide displaying heterogeneous proper-ties Different techniques visualize different compo-nents of the sample, even when great care has been taken to ensure that the same (or very similar) state of the sample is investigated in all experiments

The ThT and electrophoresis experiments provide further evidence of an aggregated and amyloid-like state of Ab(1–40) at SDS concentrations of approxi-mately 2 mm Obviously, these properties are not fully reversed when the titration continues towards higher concentrations of SDS, above the critical micellar concentration

The b-sheet containing aggregates of Ab(1–40) and detergent formed at a detergent concentration of 1–2 mm (corresponding to detergent⁄ peptide ratios of 13–27) may have different hypothetical arrangements The sample is not homogeneous in this state, as is evident from the ThT and native-PAGE experiments (Fig 6; see also Fig S1) The potential occurrence of chemical exchange between aggregates characterized by different structures and sizes, with intermediate kinet-ics, contributes to making NMR characterization diffi-cult The kinetic exchange effects may in fact be the major reason for the loss of NMR signal intensity towards high SDS concentrations in the titration experi-ments, where only one fraction of the sample is NMR visible (i.e the fraction where detergent micelles solubi-lize individual peptides and induce partial a-helical sec-ondary structure) The major fraction of the peptide molecules remains NMR invisible, suggesting that the aggregates are only partly dissolved after the titration The situation may be compared to that of a partly-folded molten globule structure of a protein like a-lact-albumin [23] In that case, the collapse of a core region

of partly-folded protein structure gave rise to extreme NMR line broadening due to chemical exchange, whereas completely unfolded protein structures allowed NMR observations However, the experiments performed in the present study do not allow us to defi-nitely decide whether there are one or more reasons for the NMR line broadening during the SDS

Trang 9

titration By contrast to the a-lactalbumin molten

globule study, where the overall molecular weight is

constant, there is both increased molecular weight in

the aggregates and heterogeneity in the present study

Both effects may contribute to the loss of NMR

signals

In one hypothetic structural scenario, the aggregates

are constituted by micelle-like oligomers of Ab(1–40)

peptide surrounded by detergent molecules b-sheet

structures would be induced by peptides interacting

with other peptides In this model, the complete loss of

structure and higher mobility of the C-terminus of the

peptide may be due to its positioning in a hydrophobic

interior of the structure (analogous to the hydrophobic

interior of a lipid bilayer) In another structural

sce-nario, the aggregate is formed by assembled Ab(1–40)

peptides, where each peptide is embedded by detergent

molecules Presumably, this model is less likely because

it seems improbable that a b-structure should be

induced in a peptide surrounded by detergent

mole-cules

The detailed molecular properties of the aggregated

complex may represent the state of the peptide when it

aggregates at a crowded cell membrane surface In

turn, this situation may reflect the state of a peptide

that is closely related to the oligomeric toxic species

thought to be involved in the pathology of Alzheimer’s

disease [3] It is interesting to note that similar

aggre-gation⁄ solubilization effects of anionic detergents have

also been described for other molecules, including

chlorin p6, a natural porphyrin compound [24], other

membrane interacting short peptides, such as

dynor-phin neuropeptides [25] and antimicrobial peptides

[26], or the intrinsically disordered protein a-synuclein

implicated in misfolding and fibril formation in

Parkinson’s disease [27]

Experimental procedures

Materials

Ab(1–40) used in the CD measurements was produced by

Neosystem Laboratoire (Strasbourg, France) The peptides

studied were unmodified at the N- and C-termini It is

known that, in physiological preparations, both

non-modi-fied and C-terminally amidated forms of the peptide are

found For NMR HSQC and TROSY measurements,

uniformly 15N-labeled Ab(1–40) from Alexo-Tech AB

(Umea˚, Sweden) was used In the diffusion NMR study,

unlabeled Ab(1–40) was obtained from rPeptide (Athens,

GA, USA) The peptides were used without further

purifi-cation SDS was purchased from ICN Biomedicals Inc

(Irvine, CA, USA), LiDS was from Sigma (Stockholm,

Sweden) and deuterated SDS-d25was obtained from Cam-bridge Isotope Laboratories (Andover, MA, USA)

Preparation of the peptide CD

To remove aggregation seeds in the sample, the peptide was dissolved in HFiP for 1 h, followed by freezing and lyophi-lizing The lyophilized peptide was dissolved in 10 mm NaOH, 4 mgÆmL)1, and sonicated in water bath for 1 min

NMR

The peptide was stored at –18C and thawed before use In the titration experiments, the concentration of the peptide was 75 lm, which was determined by weight When prepar-ing the sample, a previously described protocol was used [6] NaOH (10 mm) was added to the peptide yielding a concen-tration of 2 mgÆmL)1 and the sample was sonicated in ice bath for 1 min Cold distilled water and D2O (10% of D2O was added for signal locking) were added to half the final sample volume and, again, the sample was sonicated for

1 min Sodium phosphate buffer (20 mm) was added to reach the final sample volume The peptide concentration in the assignment experiment was 300 lm and, for that reason, the peptide was dissolved directly in LiDS in distilled water to avoid aggregation After sonication in an ice bath, D2O was added and, after another sonication, 20 mm sodium phos-phate buffer was added For all experiments, the pH was adjusted to 7.2 by adding small amounts of NaH2PO4and

Na2HPO4using the pH meter Orion PerpHecT LogR meter (San Diego, CA, USA) All sample preparations were performed on ice For diffusion NMR measurements, the peptide was dissolved as described, but at pH 7.4

FTIR

The sample was prepared in the same way as for the CD experiments but using deuterated reagents

Preparation of detergent solution

The 200 mm SDS solution was prepared in 10 mm sodium phosphate buffer (pH 7.3) or 10 mm Tris–HCl buffer (pH 7.3) LiDS was dissolved in 20 mm of sodium phos-phate buffer and two stock solutions (50 and 500 mm) were made to minimize the dilution effects SDS-d25 was dis-solved in D2O and two stock solutions were used (10 mm and 100 mm)

CD spectroscopy

CD spectra were recorded at 3C and 20 C in LiDS and SDS, respectively, and for different titration steps in

Trang 10

deter-gent at concentrations in the range 0.05–20 mm The

spec-tral region was recorded from 190–250 nm, with a 0.2 nm

step resolution, on a Jasco J-720 CD spectropolarimeter

(Jasco Inc., Easton, MD, USA) equipped with a PTC-343

temperature controller using quartz cells of 1.0 mm optical

path length At 20C, the scanning speed was 100 nmÆ

min)1 and the spectra were collected and averaged over 20

scans At 3C, a scanning speed of 50 nmÆmin)1 was used

and ten scans were employed The background signals

were subtracted from the CD spectra of the peptides The

peptide concentration was 75 lm in all experiments The

same peptide sample was used in one titration series

NMR spectroscopy

The NMR measurements were used to follow the structural

changes in the Ab(1–40) peptide caused by LiDS titration

Experiments were performed on a Varian Inova 600 MHz

spectrometer at 3C (Varian NMR, Inc., Palo Alto, CA,

USA) 1H-15N HSQC experiments were acquired in 1H

dimension in a 6 kHz window centered at 4.98 p.p.m using

a 0.12 s acquisition time and eight scans In the15N

dimen-sion, 256 increments were acquired in a 2.5 kHz window

centered at 118.5 p.p.m These parameters were used also

in the assignment experiment and, after every temperature

change of 5C, the sample equilibrated for at least 20 min

before the next run The TROSY experiment was

per-formed to study the state characterized by NMR signal loss

in the HSQC spectra The same parameters as for HSQC

were used and 96 scans were averaged Solvent suppression

was performed with excitation sculpting NMR data

pro-cessing and integration of peak volume were performed in

Varian vnmr software, whereas the spectra were presented

using sparky [28] The diffusion experiments were

per-formed with the pulsed field gradient spin-echo experiment

(pulsed field gradient longitudinal eddy–current delay) with

the 600 MHz Varian Inova spectrometer, which is equipped

with a z-axis gradient coil Thirty different linearly spaced

gradient strengths were used with a delay between the

gradient pulses of 150 ms and a gradient length of 2 ms

Calibration of the pulsed field gradients was performed by

diffusion coefficient of HDO in D2O at 25C is 1.90 · 10)9

m2Æs)1[29]

FTIR spectroscopy

FTIR spectra were collected in a Bruker Tensor 37

spec-trometer (Bruker, Ettlingen, Germany) at 20C with a

4 cm)1 spectral resolution Three series of 100 scans each

were acquired and averaged, and the second derivative was

performed with nine smoothing points using opus software

(Bruker) For clarity of the results, the negative second

derivative of the spectra normalized for the trifluoroacetic

acid band (approximately 1675 cm)1) is shown To improve signal-to-noise ratios, the peptide concentration was

100 lm

ThT and native–PAGE experiments

For ThT measurements, the final ThT concentration was kept at 15 lm for all samples (10 mm sodium phosphate buffer, pH 7.3, 20C) SDS titrations were performed both in the absence and presence of peptide (75 lm) Samples were excited at 450 nm (1 nm slit width) and

(1 nm slit width) were performed in a Jobin-Yvon Flu-oroMax spectrofluorometer (HORIBA Jobin-Yvon Inc., Edison, NJ, USA) Titrations were carried out in the same way as for the CD experiments During the SDS titration in the presence of peptide, aliquots were sampled

at 0, 1.1, 2.2, 4.6 and 25.5 mm of SDS These aliquots

(gel ran for 3 h) and subsequent silver staining was per-formed using a silver staining kit (Bio-Rad, Hercules,

CA, USA)

Acknowledgements

We thank Andreas Barth for generous access to the FTIR spectrometer, and L E Go¨ran Eriksson for helpful discussions We thank Torbjo¨rn Astlind for technical help with the NMR experiments This study was supported by the Swedish Research Council and

by the Catalan Government postdoctoral fellowship

‘Beatriu de Pino´s’ (2005 BP-A 10085 to A.P.-M.) Further support was obtained from the European Commission (contract LSHG-CT-2004-512052), the Carl Trygger Foundation, the Marianne and Marcus Wallenberg Foundation and the Swedish Foundation for Strategic Research (Bio-X)

References

1 Hardy J & Selkoe DJ (2002) The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road

to therapeutics Science 297, 353–356

2 Wilquet V & De Strooper B (2004) Amyloid-beta precursor protein processing in neurodegeneration Curr Opin Neurobiol 14, 582–588

3 Haass C & Selkoe DJ (2007) Soluble protein oligomers in neurodegeneration: lessons from the Alzheimer’s amyloid b-peptide Nat Rev Mol Cell Biol

8, 101–112

4 Danielsson J, Jarvet J, Damberg P & Gra¨slund A (2002) Translational diffusion measured by PFG-NMR

on full length and fragments of the Alzheimer Ab(1-40) peptide Determination of hydrodynamic radii of

Ngày đăng: 30/03/2014, 02:20

TỪ KHÓA LIÊN QUAN

TÀI LIỆU CÙNG NGƯỜI DÙNG

TÀI LIỆU LIÊN QUAN

🧩 Sản phẩm bạn có thể quan tâm