10830
J. Phys. Chem. B 2009, 113, 10830–10837
Self-Organization Pathways and Spatial Heterogeneity in Insulin Amyloid Fibril Formation Vito Fodera`,*,†,‡ Sebastiano Cataldo,§ Fabio Librizzi,† Bruno Pignataro,§ Paola Spiccia,† and Maurizio Leone†,‡ Dipartimento di Scienze Fisiche ed Astronomiche, UniVersita` degli Studi di Palermo, Via Archirafi 36, 90123 Palermo, Italy, CNRsIstituto di Biofisica, U.O. Via U. La Malfa 153, 90146 Palermo, Italy, and Dipartimento di Chimica Fisica “F. Accascina”, UniVersita` degli Studi di Palermo, Parco d’Orleans II Viale delle Scienze, 90100 Palermo, Italy ReceiVed: December 12, 2008; ReVised Manuscript ReceiVed: June 12, 2009
At high temperature and low pH, the protein hormone insulin is highly prone to form amyloid fibrils, and for this reason it is widely used as a model system to study fibril formation mechanisms. In this work, we focused on insulin aggregation mechanisms occurring in HCl solutions (pH 1.6) at 60 °C. By means of in situ Thioflavin T (ThT) staining, the kinetics profiles were characterized as a function of the protein concentration, and two concurrent aggregation pathways were pointed out, being concentration dependent. In correspondence to these pathways, different morphologies of self-assembled protein molecules were detected by atomic force microscopy images also evidencing the presence of secondary nucleation processes as a peculiar mechanism for insulin fibrillation. Moreover, combining ThT fluorescence and light scattering, the early stages of the process were analyzed in the low concentration regime, pointing out a pronounced spatial heterogeneity in the formation of the first stable fibrils in solution and the onset of the secondary nucleation pathways. 1. Introduction Elucidation of the molecular mechanisms leading to the formation of protein aggregates is a crucial point for a proper development of protein delivery systems.1 Further, clarifying such aspects may also have a great relevance in different medical fields.2,3 In fact, several neurodegenerative pathologies have been shown to be strictly related to the presence of ordered protein aggregates in the injured tissues, with their role in the diseases still being a matter of debate.4–8 These ordered protein deposits, referred to in literature as amyloid fibrils, display a quite regular tridimensional arrangement rich in β-sheets structures. Noticeably, the formation of such species does not take place only in ViVo and, under specific conditions, fibril occurrence can be observed also in Vitro on a number of different proteins. As a consequence, the capability of proteins to lose their native structure and self-assembly into amyloid fibrils seems to be a general property of any polypeptide chain.9,10 For this reason, studying in Vitro amyloid formation of model proteins allows us to shed new light on the mechanisms taking place in ViVo. Amyloid aggregation processes are generally indicated to proceed through different steps with different temporal and length scales. Conformational changes are commonly accepted to be the first stage in the aggregation pathway,8,11–14 in which partially destabilized and aggregation prone structures appear. Afterward, the process is generally described in terms of the so-called “nucleation and elongation” model.15,16 Destabilized structures are able to interact with each other and form a new high energy species called a nucleus; subsequently, protein molecules interact with such new species, starting the elongation * Towhomcorrespondenceshouldbeaddressed.E-mail:
[email protected]. Phone: +390916234223. Fax: +390916234223. † Dipartimento di Scienze Fisiche ed Astronomiche, Universita` degli Studi di Palermo. ‡ CNRsIstituto di Biofisica. § Dipartimento di Chimica Fisica “F. Accascina”, Universita` degli Studi di Palermo.
phase and leading to the formation of mature amyloid fibrils. The elongation process may proceed Via different mechanisms, leading to specific temporal features for the fibril growth. In fact, as shown for a number of proteins,17–19 not only an endto-end attachment of protein molecules to the nucleus (homogeneous nucleation) but also highly cooperative pathways may take place (secondary nucleation), involving the already formed fibrils as activators for new fiber filaments (branching, fragmentation, heterogeneous nucleation).16 As pointed out by Ferrone (1999), the kinetic features of both these growth processes can be mathematically rationalized, in terms of a quadratic or an exponential growth time dependence for the homogeneous and secondary nucleation, respectively.16 Insulin is a protein hormone largely used as a model system for the study of amyloid formation.20–22 In fact, under specific conditions, i.e. high temperature and low pH, it is very prone to form amyloid fibrils. Moreover, the fibrillogenesis pathway for insulin results in a highly diversified scenario for both the involved mechanisms and the occurring species.23–29 Insulin fibrillation kinetics are characterized by a long lag phase and a fast fibril growth, and it is now generally accepted that secondary nucleation mechanisms determine such a temporal profile.23,24,28,30 In particular, a crucial role is likely played by the surfaces of already formed fibrils able to act as nucleation points. Such a suggestion was first reported by Waugh31,32 and was recently confirmed by means of imaging and spectroscopic tools,23,30 so that insulin fibrillation can be considered a highly autocatalytic process. Moreover, it is worthy of note that also the kinetics profile can be affected by different experimental conditions, diverging from the classical three-steps curves (lag phase, rapid growth, and saturation) and showing a range of different species in solution.29 Regarding the species in the fibrillation path, several studies have shown that different amyloid morphologies can appear,23,25–27 being dependent on the solvent composition as well as on hydrostatic pressure.29 Interestingly, it must be noted that even the nucleus structure for insulin occurring in
10.1021/jp810972y CCC: $40.75 2009 American Chemical Society Published on Web 07/09/2009
Spatial Heterogeneity in Insulin Fibril Formation acetic acid solutions has been recently elucidated by means of small angle X-ray scattering (SAXS), and the authors proposed a novel mechanism for fiber elongation, involving a structural nucleus which is also the elongating unit of fibrils.33 Further, great interest has also been addressed to the effects of the experimental conditions on in Vitro insulin fibrillation and to the intrinsic uncertainty of the kinetic parameters, underlying a high degree of stochasticity for this process.25,28 We present an experimental study on thermally induced fibril formation in bovine insulin samples under acidic conditions. With the aim of studying the effect of protein concentration on aggregation mechanisms, progress curves for fibril formation at different insulin concentrations were detected using the fluorescent amyloid-selective dye thioflavin T (ThT).34–40 This probe has been recently reported to be chemically and optically stable under the experimental conditions here considered (high temperature and low pH),41 being a useful and fast tool for fibril monitoring. Further, atomic force microscopy has been used to identify different morphologies along the fibrillation pathway. Finally, dynamic light scattering together with ThT assay have also been used for the study of the early stages of the process in the low protein concentration regime. For this study, evolution of the whole ThT spectrum has been revealed using a CCD camera with a high sensitivity and a fast acquisition time, leading to the recording of one spectrum per second. This temporal peculiarity allowed a quantitative detection of the signals in the early stages of the process, pointing out a pronounced spatial heterogeneity in the formation of the early stable aggregates. 2. Materials and Methods 2.1. Sample Preparation. Bovine insulin and thioflavin T (ThT) were purchased from Sigma Aldrich and used without further purification. To avoid formation of salt crystals, the solvent (25 mM HCl, 0.1 M NaCl, pH 1.6) was freshly prepared before each experiment and filtered through 0.22 µm filters (MS 16534, Sartorius), before bovine insulin was added. Samples at different insulin concentrations were freshly prepared, centrifuged (3000 rpm for 10 min), and filtered through 0.22 µm filters (MS 16534 Sartorius) prior to each experiment. Protein concentration was determined by UV absorbance at 276 nm using an extinction coefficient of 1.0 for 1.0 mg/mL.42 Experiments were performed at 60 °C, inducing fibrillation on a sample volume of 1.1 mL in a PMMA UV-Grade cuvette (Kartell) with a 1 cm path. Cuvettes were carefully cleaned by milli-Q water and dried with a nitrogen flux for 5 min before each experiment. 2.2. Thioflavin T Fluorescence. Thioflavin T (ThT) was prepared in a stock solution (1 mM in Milli-Q water), filtered through 0.22 µm filters (MS 16534, Sartorius), and stored at 4 °C protected from light to avoid photobleaching. The dye concentration in water was estimated by absorbance spectroscopy using a molar extinction coefficient of 36 000 M-1 cm-1 at 412 nm.38 Before each fluorescence experiment, aliquots of this stock solution were added to the insulin solutions to a final ThT concentration of 20 µM (in situ procedure). A Varian Eclipse spectrophotometer was employed to study the effect of insulin concentration (0.5-10 mg/mL, data in Figures 1 and 2a) on the kinetics profiles. The evolution of the emission spectra during fibrillation at 60 °C was detected using emission and excitation bandwidths of 5 nm, a scan-speed of 120 nm/min, and an integration time of 0.5 s, and spectra were recorded at 1 nm intervals using an excitation wavelength of 450 nm. To study the early stages of fibril formation in the low concentration regime (0.5, 0.75, 1, 1.5, and 2.5 mg/mL, data in Figures 3a, 4, and 5), a fluorescence system (ACTON Instru-
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10831
Figure 1. Kinetics of bovine insulin fibrillation in 25 mM HCl, 0.1 M NaCl, 60 °C (λexc ) 450 nm λem ) 480 nm). ThT emission normalized to the plateau value is shown as a function of incubation time at different insulin concentrations.
ments) equipped with a PIXIS charge coupled device as a detector (acquisition time up to µs) was used. A Xenon lamp (XS-433) was used as a source. The intensity of the incoming beam was also recorded during the experiments, and its temporal evolution was used to take into account the effects of the intrinsic (very small) intensity fluctuations of the lamp on the fluorescence data. As a consequence, the presented emission spectra were already corrected for the incidental instability of the xenon lamp. Moreover, no correlation between the intrinsic drift of the source and the raw intensity fluctuation of ThT and scattering was obtained, making reliable the experimental data. Further, a removable optical lens system allows us to tune the size of the beam impinging on the cuvette (from mm to cm), changing the portion of the illuminated sample volume. Using a lens on the excitation light, a highly focalized beam is produced, leading to an illuminated volume of ∼0.4% of the total sample volume, whereas taking off the lens system, a portion of ∼60% of the total volume turns out to be illuminated. For each experiment performed, the entrance slits (source and excitation) and the exit slit were set at 300 and 500 µm, respectively. A homemade sample holder was connected to a temperature controller, having a confidence of (0.2 °C. The evolution of the ThT emission spectra (440-650 nm) during fibrillation at 60 °C was detected using an excitation wavelength of 450 nm. For the data analysis of the normalized kinetics profiles (Figure 5), TableCurve 2D v5.01 (Copyright 2002, SYSTAT Software Inc.) was employed. Several best fit functions were sampled (sigmoidal curves category) and, with the aim of estimating intensity fluctuations, the ones able to properly describe the general three-steps feature of each kinetics were chosen. As a consequence, no hypothesis on the molecular mechanism was made in using the fitting functions. Visualization of intensity fluctuations was provided by the percentual residuals (∆%) obtained from the fitting procedure. Possible overestimation of residuals ascribable to the fitting procedure was carefully minimized for a reliable detection of the fluctuations. For a quantitative description of fluctuations, the parameter F as defined in eq 1 was used
F)
∑ (∆%)2
(1)
extending the summation over the entire length of the intensity jumps after the end of the lag phase. Other measurements to check for the fluctuations were also performed using a NMR-
10832
J. Phys. Chem. B, Vol. 113, No. 31, 2009
Fodera` et al.
Figure 2. Fibrillation kinetics of 4 mg/mL bovine insulin in 25 mM HCl, 0.1 M NaCl, 60 °C. (a) ThT fluorescence curve during the process. (b) SFM images at different instants of the kinetics. Roman numerals indicate samples at different incubation times as shown in part a: (I) 0 min; (II) 40 min; (III) 60 min; (IV) 90 min; (V) 140 min; (VI) 180 min. In the blue dotted boxes, zooms on the fibril morphologies in the final part of the kinetics are reported.
like sample holder with a volume of 100 µL of protein solution or using a not focalized beam.
2.3. SFM Measurements. Dynamic scanning force microscopy (SFM)43 was carried out in air using a Multimode
Spatial Heterogeneity in Insulin Fibril Formation
J. Phys. Chem. B, Vol. 113, No. 31, 2009 10833 Such a procedure was used for the acquisition of six images during the kinetics. To make as reliable as possible SFM data, images were acquired from several sample areas. 2.4. Dynamic Light Scattering. Protein sizes and scattering intensity in the early stages of the fibrillation were measured by dynamic light scattering employing a Zetasizer Nano ZS (Malvern Instruments) with a 633 nm light beam and operating in the back scattering configuration (178°). Experiments were performed at 60 °C, inducing fibrillation on a sample volume of 1.1 mL in a PMMA UV-Grade cuvette (Kartell). For the correlation data, an average of 12 scans with 10 s of integration time was used and measurements were recorded every 3 min during the incubation.
Figure 3. (a) ThT fluorescence and (b) DLS intensity as a function of the incubation time during the early stages of fibrillation of 0.5 mg/ mL bovine insulin in 25 mM HCl, 0.1 M NaCl, 60 °C. Solid lines are quadratic fitting curves of data during the lag phase (primary nucleation). Dashed lines are exponential growth fitting curves of the whole data set (secondary nucleation).
Figure 4. 300 emission spectra (λexc) 450 nm) as detected for 300 s at the end of the lag phase (16800-17100 s of incubation) of 0.5 mg/ mL of bovine insulin in 25 mM HCl, 0.1 M NaCl at 60 °C (20 µM ThT). Red lines are indicated to better visualize the evolution of elastic scattering (λem) 450 nm) and ThT intensity (λem ) 480 nm).
Nanoscope V workstation (Veeco). Etched-silicon probes with a pyramidal-shape tip having a nominal curvature of 8 nm and a nominal internal angle of 25° were used. During scanning, the 125-µm-long cantilever, with a nominal spring constant in the range of 20-80 N m, oscillated at its resonance frequency (286 kHz). Height and phase images were collected by capturing 512 × 512 points in each scan, and the scan rate was maintained below 1 line per second. During imaging, temperature and humidity were about 20 °C and 40%, respectively. Dynamic scanning probe microscopy measurements were carried out during fibrillation kinetics of 4 mg/mL of bovine insulin in 25 mM HCl, 0.1 M NaCl, pH 1.6 at 60 °C. During the kinetics, aliquots have been taken out and dilute 200 times in water solution of 25 mM HCl without salt. Afterward, 8 µL of such a solution was positioned on a freshly cleaved mica surface. A nitrogen flux was used for 5 min to dry the protein solution on the mica before performing SFM measurements.
3. Results and Discussion 3.1. Different Processes and Morphologies Occurring during the Fibrillation Kinetics. Figure 1 shows the progress curves of ThT fluorescence as a function of the incubation time at different insulin concentrations, ranging from 0.5 to 10 mg/ mL, incubated at 60 °C with 20 µM of ThT. At low concentrations (4 mg/mL), besides the absence of the lag time, the biphasic behavior does not take place or, probably, the occurrences of the two processes could be too fast to be clearly detected. For the other concentrations investigated (
