Effect of Freezing on Amyloid Peptide Aggregation and ... - Springer Link

ISSN 1061-933X, Colloid Journal, 2008, Vol. 70, No. 4, pp. 501–506. © Pleiades Publishing, Ltd., 2008. Original Russian Text © A.V. Filippov, A.V. Suleimanova, G. Grobner, O.N. Antsutkin, 2008, published in Kolloidnyi Zhurnal, 2008, Vol. 70, No. 4, pp. 544–549.

Effect of Freezing on Amyloid Peptide Aggregation and Self-Diffusion in an Aqueous Solution A. V. Filippova, A. V. Suleimanovaa, G. Grobnerb, and O. N. Antsutkinc a

Kazan State University, ul. Kremlevskaya 18, Kazan, 420008 Tatarstan, Russia b Umea University, Department of Chemistry SE-90187 Umea, Sweden c Lulea University of Technology, Division of Chemistry SE-97187 Lulea, Sweden Received August 21, 2007

Abstract—Pulsed-field gradient 1H NMR is employed to investigate the self-diffusion of amyloid Aβ-peptide in an aqueous buffer solution (pH 7.44) with a protein concentration of 50 µmol at 20°C. The self-diffusion coefficient of the peptide in a freshly prepared solution corresponds to its monomeric form. The storage of the solution at 24°C causes part of the peptide molecules to form amyloid aggregates as soon as over 48 h. However, the 1H NMR echo signal typical of aggregated molecules is not observed because of their dense packing in the aggregates and a large mass of the latter. A freezing–fusion of the solution after the aggregation does not cause changes in the self-diffusion coefficients of the peptide. After a peptide solution free of amyloid aggregates is subjected to a freezing–fusion cycle, part of the peptide molecules also remains in the monomeric form in the solution, while another part forms amyloid aggregates, with a portion of the aggregated peptide molecules retaining a high rotational mobility with virtually absolute absence of a translational mobility. The results obtained are interpreted in terms of the formation of “porous aggregates” of amyloid fibrils, with “pores” having sizes comparable with those of peptide molecules, though, being larger than water molecules. Peptide molecules, which do not form fibrils, are captured in the pores. Temperature regime is shown to be of importance for the aggregation of amyloid peptides. In particular, freezing, which is traditionally considered to be a method for the prevention from or temporary interruption of aggregation, may itself lead to the formation of amorphous amyloid aggregates, which remain preserved in solutions after their unfreezing. DOI: 10.1134/S1061933X08040157

INTRODUCTION Alzheimerís disease (AD) is a chronic illness of the brain that affects a considerable portion of the population. Along with some other diseases, such as spongy encephalopathy, AD belongs to amyloid diseases, which are caused by the deposition of Äβ-peptide, which is soluble under ordinary conditions, in the form of neurotoxic aggregates [1]. Aggregates of amyloid Äβ-peptides composed of 39–43 aminoacid residues form plaques or fibrils in patientís brains, thus leading to the apoptosis of nerve cells. The causes and mechanisms of the changes in the character of deposition of peptides, which, under ordinary conditions, have globular or native structures, with the formation of toxic aggregates still remain to be investigated despite of a considerable progress [1–4]. Structural and biological studies devoted to the aggregation of Äβ-peptides into fibrillar structures suggested that the aggregation processes strongly depend on physical conditions. One of the approaches to studying the aggregation of amyloid Äβ-peptides consists in the investigation of their aqueous solutions [1, 5–7]. In this work, we studied the influence of freezing of an aqueous solution of amyloid Äβ-peptide on its aggregation. We used a method that enabled us to directly observe aggregates in a solution, that is, the

pulsed-field gradient NMR technique [5–13]. This method made it possible to identify aggregates owing to a simple relation between the mass of a particle (a molecule or an aggregate) and its self-diffusion coefficient. The self-diffusion coefficient of a particle in a solution may be calculated through the Stokes–Einstein equation describing the motion of a rigid sphere in a viscous medium as follows: kT kT D = ------ = --------------, 6ηπR f

(1)

where k is the Boltzmann constant, í is temperature, f is the coefficient of friction, η is the viscosity of a solvent, and R is the hydrodynamic radius of the sphere. The equation for the self-diffusion coefficient of a particle with mass å has the following form: 4πN 0 ρ Tk D = ⎛ ---------------⎞ ⎛ 3 ------------------------------------⎞ , ⎝ 6πηî⎠ ⎝ 3M ( V 2 + δ 1 V 1 )⎠

(2)

where ρ is the particle density, N0 is Avogadro number, î is the form factor, V1 is the specific volume of the particle, V2 is the specific volume of the solvent, and δ1 is the solvent fraction bonded to the particle. In order to calculate particle mass å from a known self-diffusion

501

502

FILIPPOV et al.

coefficient D, we need to know the values of η, V2, and δ1. In addition to the study of the self-diffusion, ordering, which is typical of amyloid aggregates, was monitored by the methods of the circular dichroism and fluorescence of thioflavin T. The amyloid is characterized by the presence of fibrils in which β-structures are located normal to the fibril axis [14], so that the formation of β-structures may be established from the circular dichroism spectra [3]. Thioflavin T binds to amyloid fibrils, thus increasing the intensity of the fluorescence in the region of 485 nm caused by an excitation radiation in the region of 455 nm [15]. OBJECTS AND METHODS OF THE STUDY Amyloid peptide Äβ(1–40)E22G arctic mutant [16] was investigated, the primary structure of which is described by the following aminoacid sequence: DAEFRHDSGYEVHHQKLVFFAGDVGSNKGA-IIGLMVGGVV. Peptide molecular mass was 4260 Da. Amyloid peptide (0.5 g) was dissolved in deuterated trifluoand the solution roethanol (CF3–CD2–OD) concentration was brought to 50 µmol. This solvent is known to stabilize the native α-helical secondary structure of the peptide. (The circular dichroism spectra indicate the presence of an α-helical conformation and the self-diffusion coefficient of peptide molecules (1.35 × 10–10 m2/s at 20°ë) corresponds to the monomeric form [8].) Then, the sample was dried at room temperature under nitrogen and evacuated for 12 h. A buffer solution (TRIS (10 mmol/l), KCl (10 mmol/l), and EDTA (0.5 mmol/l); pH 7.44) was added to the dried powdered peptide to prepare its solution with a concentration of 50 µmol/l. The examined solution was transparent. It was divided into two portions; each portion was placed into a thin-walled glass ampule and sealed. Identical samples thus prepared were denoted as N1 and N2. The time period from the onset of the dissolution and the first measurement of the self-diffusion was 12 h. Visual examinations and self-diffusion measurements were performed for 18 days. In the course of the experiments, samples were frozen for a short time (4 h), whereas, during most part of the time, including the time of measurements, both of the samples were kept at 20°ë. Samples were completely frozen over 2–3 min in a freezing chamber. Self-diffusion measurements were carried out using a Varian/Chemagnetic CMX NMR spectrometer operating at a frequency of 100 MHz for 1H employing the stimulated echo sequence [17]. Diffusive decay (DD), A(k), was registered. Here, Ä(k) refers to the integral values of the spectra and k = γ2δ2g2td, where γ is the proton gyromagnetic ratio, δ is the duration of a pulsedfield gradient, g is the amplitude of the pulsed-field gradient, td = (∆ – δ/3) is the diffusion time, and ∆ is the distance between the second and the third radiofrequency

pulses. In the case of a single self-diffusion coefficient, Ä(k) depends on the experiment parameters as follows: A ( k ) = A ( 0 ) exp ( – kD ).

(3)

In all of the experiments, constant magnetic field and pulsed-field gradient had the same directions. The gradients had a constant amplitude of g = 1.15 T/m, while the δ value was varied over the range 0.3–3 ms. Diffusion time was equal to 100 ms. To provide the necessary signal-to-noise ratio, the number of acquisitions was as large as 640; the relative error in the determination of the self-diffusion coefficients was below 10%. The set of spectra was obtained via the Fourier transformation of experimental echoes. Integral spectrum value– pulsed-field gradient dependences were presented as DDs. Amyloid structures resulting from peptide aggregation in the solution were determined by the circular dichroism method (a Jasco J-810 spectropolarimeter equipped with quartz cells with a volume of 150 µl and an optical path of 0.2 mm). The measurements were carried out in a wavelength range of 190–250 nm. Each sample was measured eight times, and the spectra were averaged. Thioflavin T fluorescence spectra were registered with a FluoroMax-2 spectrofluorimeter (Jobin Yvon/Psex Instruments, the United States) in a wavelength range of 450–550 nm, excitation wavelength of 440 nm, and slit width of 5 nm. RESULTS 1. Self-Diffusion of Monomeric Aβ-Peptide in a Solution Figure 1 illustrates a DD typical of a transparent freshly prepared solution at the initial stage of an experiment. At low pulsed-field gradients, a signal attributed to water is observed in the DD (the arrow denotes the region in which water and peptide protons contribute to the echo signal). However, as the gradient is enhanced, the contribution of water decreases, so that, beginning with the sixth point of DD, the signal becomes to be due to peptide protons alone. This assignment is also confirmed by the patterns of the NMR spectra measured at different pulsed-field gradients, where the signal characteristic of water is observed as a line with a chemical shift of approximately 4.6 ppm only at low values of the pulsed-field gradient. The DD region corresponding to the peptide has the pattern described by Eq. (3), thus making it possible to determine the self-diffusion coefficient of the peptide, which, at 20°ë, is equal to 1.59 × 10–10 m2/s. Several procedures are employed to determine the state of protein molecules from measured values of their self-diffusion coefficients. Price et al. [11] calculated the self-diffusion coefficient of monomeric lysozyme from the data on a three-dimensional protein structure in terms of the Han–Herzfield [18] scaling COLLOID JOURNAL

Vol. 70

No. 4

2008

EFFECT OF FREEZING ON AMYLOID PEPTIDE AGGREGATION Dense precipitate

Amplitude 10000

N1 N2

Turbid solution –20°C for N1 Transparent solution 1000

2 × 1010

1 × 1010

3 × 1010 γ2δ2g2 (∆–δ/3)

Fig. 1. Diffusive decay of a stimulated spin echo measured for a transparent amyloid peptide solution (sample N1) at the beginning of the study at 21°ë. The dashed line refers to the slope corresponding to the calculated self-diffusion coefficient of monomeric peptide under these conditions (D ≈ 1.59 × 10–10 m2/s). The deviation from the linear dependence at low magnetic field gradients is caused by the presence of water protons.

theory with allowance for the effect of molecular collisions. Wilkins et al. [13] related the effective hydrodynamic radius of protein molecules (native and denaturated ones) to the number of aminoacids in a polypeptide chain. In this work, the state ofÄβ-peptide molecules was determined by comparing experimentally measured self-diffusion coefficients with the coefficient calculated for monomeric peptide. When calculating the self-diffusion coefficient of the monomer, we accepted the following simplifications: (1) the monomer has a globular conformation and (2) particles are spherical (î = 1) and the effects of interparticle collisions in dilute solutions may be ignored. The self-diffusion coefficient of monomeric Äβ-peptide at 20°ë (1.51–1.59) × 10–10 m2/s) was determined using Eq. (2) and the following characteristics reported in the literature: density ρ = 1300 kg/m3 [5], viscosity η = 1.002 × 10–3 Pa s, and peptide-bonded water fraction δ1 = 0−0.328 [9]. The calculated coefficient value coincides with that experimentally determined. Thus, in a freshly prepared buffer solution, Äβ-peptide exists in the monomeric form. 2. The Effect of Temperature on the Self-Diffusion of Aβ-Peptide in a Solution This experiment is schematically represented in Fig. 2. In the scheme, the ranges and temperatures of freezing, as well as the moments of self-diffusion measurements performed for samples N1 and N2, are shown in the time scale. Sample N2 was stored at 24°ë almost to the end of the experiment; it was subjected to freezing–fusion over such a long period as 17 days. Initial sample was COLLOID JOURNAL

Vol. 70

No. 4

2008

–20°C for N2

–55°C for N1 0

0

503

2 4 6 8 10 12 14 16 18 20 Days elapsed after peptide dissolution

Fig. 2. Schematic representation of the effect of temperature on peptide solutions and their state. Light rectangles refer to the periods of sample freezing at temperatures denoted in the figure. Triangles and circles correspond to self-diffusion measurements (one measurement a day).

transparent; however, over 48 h, the solution became turbid, thus indicating aggregation. This turbidity remained preserved to the end of the experiment. It is surprising that the visually observed state of the sample and the self-diffusion do not completely correspond to one another. The DD of sample N2 was similar to that demonstrated in Fig. 1, irrespective of whether the solution was transparent (during the first day after the peptide was dissolved) or turbid (during all subsequent days), as well as after a freezing–fusion cycle carried out on the 18th day after the sample was prepared. Sample N1 was placed into the measuring unit of the NMR spectrometer at 20°ë, and the measurements were performed for 3 days. During the measurements, visible signs of peptide aggregation and variations in the self-diffusion coefficients were not observed. After the first freezing–fusion cycle, the sample became turbid and its DD changed (Fig. 3). The comparison of the obtained DD with the DD measured before the sample was frozen (Fig. 1) testifies that the pattern of the former is more complex. The analysis indicates that this DD may be represented as the sum of two components; the signal of one of them is independent of the imposed pulsed-field gradient and characterizes the protons of part of peptide molecules that have virtually no translational mobility (D ≈ 0, solid line), and the other component coincides with the DD of a monomeric peptide solution (triangles). As follows from the above considerations, in the course of a freezing–fusion cycle, the peptide is partly aggregated and part of it (probably occurring in the aggregated state) has an extremely low translational mobility, whereas another part of the peptide remains in the solution as a monomer. The observed self-diffusion coefficient of the aggregated peptide cannot be attributed to its dimeric or another aggregated form containing a small number of molecules in an aggregate. Simple estimation through hydrodynamic relation (2) suggests that an aggregate must contain at least 104 peptide molecules. At the same time, the turbidity of the solution indicates that the sizes

504

FILIPPOV et al.

Amplitude 10000

Amplitude 10000

3 1000

1000

2 1

100

100 0

1 × 10

10

2×

1010

3× γ2δ2g2 (∆–δ/3) 1010

Fig. 3. Diffusive decay of stimulated spin echo in sample N1 after the first freezing–fusion cycle (triangles). One of the components of this DD corresponds to part of peptide that is characterized by a self-diffusion coefficient ≈0 (solid line). The second component (circles), which was determined by subtracting the first component from the experimental DD, virtually corresponds to the self-diffusion of monomeric peptide (see Fig. 1) in the solution. The dashed line refers to the slope corresponding to the calculated selfdiffusion coefficient of monomeric peptide.

of the aggregates are rather large and exceed the wavelength of visible light (approximately 600 nm). This turbidity remains preserved, and the DD depicted in Fig. 3 remains reproducible for several days until the next freezing of the sample. After sample N1 was subjected to the second freezing–fusion cycle, the turbidity disappeared, and a precipitate with a particle size of about 0.1 mm was formed at the bottom of the ampule. However, this change had no effect on the DD pattern, which is illustrated in Fig. 3. After the third freezing, which was more profound (to –55°ë) as compared with the two previous ones (to –20°ë), the contribution of the “immobile” component of the DD rises 2.5-fold. The completely changed DD of sample N1 is presented in Fig. 4. Circular dichroism spectra were registered for freshly prepared solutions and those stored for 19 days. The spectra of the freshly prepared solutions were typical of native peptides occurring in the α-helical conformation. After the experiment was finished, the spectra demonstrated the prevalence of β-structures (Fig. 5, curves 1, 2), with their content being higher for sample N2. At the same time, the thioflavin fluorescence spectra were measured, which demonstrated an increase in the fluorescence by 7.7 and 11 times for samples N1 and N2, respectively. Thus, both of the methods attested to the formation of amyloid aggregates, which was more intense for sample N2. Finally, after the experiment was completed, both of the samples were dried, and the precipitates were dissolved in trifluoroethanol. In both cases, the circular dichroism spectra demonstrated the predominance of the α-helical conformation

2 × 1010

1 × 1010

0

3 × 1010 γ2δ2g2 (∆–δ/3)

Fig. 4. Diffusive decay of stimulated spin echo in (1) initial sample N1 and after (2) its first and second freezing at –20°ë and (3) the third freezing at –55°ë. The dashed line refers to the slope corresponding to the calculated self-diffusion coefficient of monomeric peptide. Amplitudes of the initial DD are reduced 1.6-fold for convenience of consideration.

(curve 3 in Fig. 5 was obtained for the precipitate dissolved in trifluoroethanol). DISCUSSION As follows from the obtained self-diffusion coefficients, the examined Äβ-peptide is present in an aqueous solution as a monomer, which is in concordance with the published data [5–7]. In the course of peptide Ellipticity, millidegree α-helix β -fold

2 1

Random coil

2

200 220 240

0 1 –1

3

–2 190 200 210 220 230 240 250 λ, nm Fig. 5. The circular dichroism spectra of samples (1) N1 and (2) N2 18 days after the onset of the experiment and (3) precipitate of sample N1 dissolved in trifluoroethanol. Measurement temperature is 20°ë. The circular dichroism spectra characteristic of three different peptide conformations [20] are illustrated in the insert. COLLOID JOURNAL

Vol. 70

No. 4

2008

EFFECT OF FREEZING ON AMYLOID PEPTIDE AGGREGATION

aggregation, no signal attributed to dimers arises, while the aggregation itself is easy to observe visually. The absence of the signal due to Äβ-peptide dimers was previously discussed in the literature and explained by an enhanced spin–spin NMR relaxation of protons in oligomeric Äβ-peptides [6], as well as by the fact that oligomeric Äβ-peptides with low molecular masses are not stable intermediate products of the aggregation of Äβ-peptides [7–8]. We observed an unusual phenomenon, namely, the appearance of a signal attributed to protons of the peptide occurring in a state characterized by very low translational mobility; this state appears after a freezing–fusion cycle. Upon aggregation at room temperature, this component does not arise in the spectrum of self-diffusion coefficients, as it is seen by the example of sample N2. Obviously, when aggregation proceeds under normal conditions, the spin echo signal cannot be observed because of too short spin–spin relaxation times (≤100 µs) of peptide protons, which, therefore, are not registered by the employed procedure of NMR in a liquid. It should be taken into account that the feasibility of the signal observation is also associated with the averaging of the dipole–dipole interactions of protons in this state, when peptide molecules rotate isotropically as a whole with a correlation time of smaller than 10–6 s. Let us consider some features of this state. (1) As two components of self-diffusion are observed (one of them corresponding to the monomer and the other having a self-diffusion coefficient close to zero), the time conditions of the experiment correspond to the “slow exchange” between the two aforementioned states of peptide molecules. This statement means that peptide molecules remain in “phases” with different diffusion coefficients for noticeably longer time periods than the diffusion time in our experiments (100 ms). The translational mobility of peptide molecules in one of the phases is hindered; at the same time, their rotation is rather free and isotropic. (2) The translational mobility of peptide molecules rather than water molecules and buffer components is limited in the formed aggregates. Hence, we may assume that these aggregates contain “pores”, which are larger than water molecules but smaller than peptide molecules. Thus, aggregates formed without sample freezing are obvious to contain denser packed and more ordered peptide molecules, because they are formed under conditions, which are closer to the thermodynamic equilibrium (a lower local peptide concentration and room temperature). In these aggregates, the rotational mobility of all peptide molecules is limited (probably by the rotation of an aggregate as a whole). The more ordered β-structure of aggregates formed without solution freezing is also confirmed by the more intense fluorescence of thioflavin. Aggregates formed with freezing are more amorphous, because they are formed under nonequilibrium conditions and at a high local peptide COLLOID JOURNAL

Vol. 70

No. 4

2008

505

concentration (because the water is partly frozen). Under these conditions, the aggregate matrix is formed from peptide molecules (β-structures) whose rotational mobility is hindered; however, this matrix contains cavities with captured peptide molecules (that probably occur in a monomeric form) whose high rotational mobility remains preserved. These aggregates are permeable to water and other small molecules of the buffer. It is of interest that this situation is similar to polymeric carcerand molecules obtained and described by Sherman and Cram [19]. In these compounds, hydrocarbon backbones with additional benzene and acidic phenol groups form molecular cavities filled with freely rotating guest molecules (CH3)2NCHO, (CH3)2NCOCH3, and CH3SO), which cannot leave the cavities or “wards”. Irrespective of their prehistory, amyloid aggregates appear to be unstable with respect to the action of trifluoroethanol. CONCLUSIONS Amyloid peptide solutions have been shown to be sensitive to temperature and freezing even at low concentrations. In samples (both subjected and not subjected to freezing), peptide self-diffusion coefficients correspond to its monomeric form rater than aggregates of few particles. Probably, such aggregates are unstable or the spin–spin relaxation times of protons in these systems are about 1 ms and below. Under the aggregation conditions close to the thermodynamic equilibrium, more ordered aggregates are formed with high degree of molecular packing and larger fraction of peptide molecules transformed into the amyloid form. A freezing–fusion cycle gives rise to the formation of a system of bonded amorphous aggregates containing captured peptide molecules, which freely rotate in cavities inaccessible to other peptide molecules present in a solution. These aggregates are permeable to small molecules, such as water molecules. It has been found that freezing, which is commonly considered to be a procedure for prevention systems from aggregation or temporary interruption of this process, can itself lead to the formation of complex networks of amyloid aggregates, which remain preserved in solutions after fusion. ACKNOWLEDGMENTS We are grateful to V.V. Zamotin for his help in the analysis of thioflavin fluorescence. This work was supported by the Ministry of Education and Science of Russian Federation, the program “Development of Scientific Potential of Higher School”, grant no. 2.1.1.3222 and CRDF REC-007-3.

506

FILIPPOV et al.

REFERENCES 1. Rochet, J.-C. and Lansbury, P.T., Curr. Opin. Struct. Biol., 2000, vol. 71, p. 60. 2. Antzutkin, O.N., Magn. Reson. Chem., 2004, vol. 42, p. 231. 3. Bokvist, M., Lindstrom, F., Watts, A., and Gröbner, G., J. Mol. Biol., 2004, vol. 335, p. 1039. 4. Gnanakaran, S., Nussinov, R., and Garcia, A.E., J. Am. Chem. Soc., 2006, vol. 128, p. 2158. 5. Danielsson, J., Jarvet, J., Damberg, P., and Gräslund, A., Magn. Reson. Chem., 2002, vol. 40, p. 89. 6. Jarvet, J., Damberg, P., Bodell, K., et al., J. Am. Chem. Soc., 2000, vol. 122, p. 4261. 7. Narayanan, S. and Reif, B., Biochemistry, 2005, vol. 44, p. 1444. 8. Filippov, A., Sulejmanova, A., Antzutkin, O., and Gröbner, G., Appl. Magn. Reson., 2005, vol. 29, p. 439. 9. Krishnan, V.V., J. Magn. Reson., 1997, vol. 124, p. 468. 10. Nesmelova, I.V. and Fedotov, V.D., Biochim. Biophys. Acta, 1998, vol. 1383, p. 311.

11. Price, W.S., Tsuchiya, F., and Arata, Y., J. Am. Chem. Soc., 1999, vol. 121, p. 11503. 12. Tillett, M.L., Lian, L.-Y., and Norwood, T.J., J. Magn. Reson., 1997, vol. 133, p. 379. 13. Wilkins, D.K., Grimshaw, S.B., Receveur, V., et al., Biochemistry, 1999, vol. 38, p. 16424. 14. Sunde, M. and Blake, C., Adv. Protein Chem., 1997, vol. 50, p. 123. 15. Naiki, H., Higuchi, K., Hosokawa, M., and Takeda, T., Anal. Biochem., 1989, vol. 177, p. 244. 16. Nilsberth, C., Westlind-Danielsson, A., Eckman, C.B., et al., Nature Neurosci., 2001, vol. 4, p. 887. 17. Tanner, J.E., J. Chem. Phys., 1970, vol. 52, p. 2523. 18. Han, J. and Herzfield, J., Biophys. J., 1993, vol. 65, p. 1155. 19. Sherman, J.C. and Cram, D.J., J. Am. Chem. Soc., 1989, vol. 111, p. 4527. 20. Greenfield, N. and Fasman, G.D., Biochemistry, 1969, vol. 8, p. 4108.

COLLOID JOURNAL

Vol. 70

No. 4

2008