Back to the oligomeric state: pH-induced dissolution of concanavalin A amyloid-like fibrils into non-native oligomers

Abstract

The subtle interplay between long range electrostatic forces, hydrophobic interactions and short range protein–protein interactions regulates the onset/evolution of protein aggregation processes as well as the stability of protein supramolecular structures. Using a combination of FTIR spectroscopy, light scattering and advanced imaging, we present evidence on the main role of electrostatic forces in the formation and stability of amyloid-like fibrils formed from concanavalin A (ConA), a protein showing structural homology with the human serum amyloid protein. At high protein concentration, where protein–protein interactions cannot be neglected, we highlight a thermal-induced aggregation pathway in which amyloid-like aggregates are readily formed. When dissolved in solutions at different pHs, these aggregates show either a reduced β-sheet content keeping the same morphology (3 < pH < 10) or they are promptly dissolved leaving in solution non-native oligomers (pH > 10). The latter result can be ascribed to the change of the charge state for the ConA amino acid side chain with high pK_a values. Our results support the idea of fibrils as a reservoir of oligomeric species that can be released if changes or discontinuities in the aggregate microenvironment occur.

---

Introduction

The phenomenon of protein aggregation and in particular the formation of amyloid fibrils is nowadays the focus of many research activities in basic and applied sciences. The formation of these ordered elongated aggregates is implicated in a large number of debilitating pathologies such as Alzheimer's and Parkinson's disease.^1,2 Moreover, amyloid fibrils may be biologically functional and have important roles in vivo.³ Recently, the great potential of these β-sheet-rich nanostructures as self-assembled biomaterials has also been highlighted.^4–6 Indeed, they are characterized by remarkable physico-chemical properties^7–10 and they may form superstructures with tunable size and morphology in aqueous solution.^9,10 Such superstructures may be suitable for a wide range of applications in tissue engineering, regenerative medicine and drug delivery.^9–15 As a consequence, the detailed understanding of fundamental interactions and mechanisms regulating amyloid self-assembly and stability is essential both for the development of new therapeutic strategies, for amyloid-related pathologies and for the design and the manipulation of novel biomaterials.

In specific conditions, proteins associate and form amyloid fibrils regardless of the sequences or the structures of the native state; this has led to the idea that the ability to form amyloids is a generic feature of polypeptide chains^2,3,16 and that the amyloid state is thermodynamically more stable than all other states accessible to a protein chain.^17,18

In cells, heterogeneous environmental conditions, e.g. different local concentration of proteins or local presence of different ions, changes in pH, carbohydrates compositions, may regulate multiple mechanisms. They can either minimize aggregation and maintain functional state favourable or, if necessary, favor protein association. In different cell compartments local environmental properties may tune inter- and intra-molecular forces involved both in protein folding and aggregation. These two phenomena involve inter-residue interactions of essentially identical character and it is possible to infer that they are controlled by general physical principles.^12,19,20 Interplay between long range electrostatic forces, hydrophobic interactions and short range protein–protein interactions regulates the onset and the evolution of aggregation processes and the final morphology of the aggregates but it is also responsible for their stability.^12,21–23 Physico-chemical interactions between protein aggregates and the environment may drive their stabilization or dissolution, and these events are potentially connected to the biological effects. It was recently shown that by varying the solvent properties it is possible to induce fibril dissolution. As an example, dimethylsulfoxide (DMSO) and 2,2,2-trifluoroethanol (TFE) were found to induce disaggregation by interfering with H-bonds stabilized β-structure.^24–26 Exposing fibrils to ethanol or TFE–water mixtures resulted in fibril disassembly and reassembly, thus suggesting that changes in dielectric constant of the surrounding medium is fundamental for the amyloid stability.^26,27 Salt concentration and pH changes were shown to both promote aggregation and trigger fibril association and dissolution.^28,29 Moreover, UV-light irradiation is able to induce modification of single hydrophobic residues causing the generation of new charges in the region of the intra- and inter-sheet space.³⁰ This causes unfavorable electrostatic interactions disrupting the β-sheet stacking and induces fibril dissociation.³⁰

All these observations underline how the subtle balance between electrostatic and hydrophobic interactions affects the formation and stability of the organized hydrogen bond network constituting fibril spine.^31–33 It has been also shown that electrostatic interactions between single proteins control the supramolecular assembly kinetics and the resulting aggregate morphology.^19,34 These interactions can be easily modified by changing the pH value, which regulates protein net charge. In particular, both experimental and theoretical studies support the idea that highly charged states, e.g. in solution far from iso-electric point, favour the formation of aggregates with fibril-like structures.^12,32,35 Both specific interactions (salt bridges stabilizing H-bond spine) and unspecific interactions may play a fundamental role in fibril stability and may regulate the physico-chemical properties of the aggregates.

Here, we explored the effects of pH on concanavalin A (ConA) fibril formation and on the stability of ConA fibrils. ConA represents a suitable model for its biological and structural peculiarities: (1) it induces apoptosis on tumoral cells with a mechanism related to its aggregation,^32,36 (2) it shows a large structural homology to the human serum amyloid protein, generally present in all the in vivo fibril deposits³⁷ and (3) its enhanced flexibility and its “all β” structures allow isolating specific features from the amyloid fibril formation route.

Previous studies report that protein–solvent interactions regulate ConA aggregation kinetics and that the pH of the solution determines the structure and the morphology of the final aggregates.^32,38–40 At basic pH away from the iso-electric point (pI = 5.1), when the net charge of the protein is mainly negative, well defined long and thin amyloid-like fibrils are formed via a non-nucleated assembly mechanism. This involves the formation of intermolecular β-sheets.^38,39 At pH values close to the pI the process leads to the formation of amorphous aggregates.³⁸

At low pH (pH 3) in the regime of low protein concentration (<30 μM) ConA was found to form large fibril-like aggregates. In these conditions away from pI, ConA adopts a more flexible structure called “unlocked” and undergoes a thermal aggregation process characterized by the formation of an on-pathway long-lived intermediate. The subsequent coagulation of such “crinkled” precursors generates amyloid-like fibrils.⁴⁰

In the present study we focus on ConA aggregation process at low pH in high protein concentration regime where the effect of short range protein–protein interactions on the aggregation kinetics cannot be neglected. In these conditions we analyse ConA association and pH induced dissociation mechanisms.

Using FTIR spectroscopy, light scattering and fluorescence microscopy techniques, we analyse the effect of altering the pH of the aggregate environment, mainly focusing on disaggregation conditions. Our results highlight the role of polypeptide chain net charge in aggregate stability. Data suggest the possible role of specific residues whose changes in the charge state drive the sudden fibril dissolution and formation of small oligomeric species.

Fast dissociation of large fibril-like aggregates in small species further confirms the idea of amyloid fibrils as a reservoir of toxic oligomers in cells.⁴¹ In specific conditions, local physiological or pathological (e.g. due to inflammation states) pH discontinuities may allow the fast production of highly-concentrated diffusible oligomers, which may interact or disrupt cell membranes. These mechanisms may act in parallel with molecular recycling mechanisms⁴² producing association and dissociation of oligomers from mature fibrils which may have effects at longer time scales. The response to environmental stimuli for fibril-like aggregates may also have intriguing applications in biomaterials. Protein aggregate gel may be indeed used as scaffold to support the growth of other materials and be readily destroyed by washing procedure with aqueous solution at suitable pHs.

Experimental

Sample preparation

Concanavalin A (ConA, type IV, L7647) and thioflavin T (ThT) were purchased from Sigma Aldrich, ConA-Alexa647 was purchased from ThermoFisher (C21421), and all are used without further purification. All the measurements were performed in deuterated solution with 0.1 M NaCl. pD was adjusted with DCl to a final value of about 2.6. Each solution was freshly prepared and filtered through 0.20 μm filters (MS 17761, Sartorius). Concentration measurements were performed by means of a Jasco V-770 UV-Vis spectrophotometer equipped with One-Drop Unit SAH-769. Protein concentration was obtained using absorption at λ = 280 nm and ε = 33 280 cm⁻¹ M⁻¹. To follow the aggregation kinetics by FTIR and light scattering, we prepared ConA samples at 8.0 ± 0.1 mg ml⁻¹ and pD ∼ 2.6. pD values were corrected by 0.4 (i.e. pD = pH meter reading + 0.4).

Fourier transformer infrared spectroscopy

Fourier transformer infrared (FTIR) measurements were carried out with a Bruker Vertex 70 spectrometer equipped with a DTGS (doped triglycine sulfate) detector, in a sample compartment under continuum purging in N₂ dry atmosphere. To follow the kinetics of aggregation, a 50 μl-aliquot of freshly prepared ConA was placed between two CaF₂ windows separated by a 50 μm Teflon spacer. Each final spectrum is an average of 128 scans in the 400–7000 cm⁻¹ range with a spectral resolution of 2 cm⁻¹. The measurements were performed at 60 °C and a VWR temperature controller has been used. FTIR spectrum of empty FTIR cell, taken under identical conditions, was used as a reference to calculate the absorption spectrum. To eliminate the contribution of the background from amide I′ bands, the buffer spectrum measured in the same conditions was subtracted. The subtraction procedure was optimized in the region at about 3860 cm⁻¹. Amide I′ bands were thus normalized. We performed fittings of the amide I′ bands in terms of seven Gaussian components. The following expression was used:
where ν_i, σ_i and are the peak frequency, width, and the area of the i^th component. The components at 1604 cm⁻¹ and 1707 cm⁻¹ in the present analysis are considered as extrapolations and their peak frequency, width, and area are maintained constant in the fittings. Moreover, the peak frequency of the other five components was kept fixed at 1620, 1635, 1657, 1672 and 1693 cm⁻¹. We defined “fractional area” . Since each component will be assigned to a specific backbone conformation, the fractional area indicates how the protein secondary structure changes during the thermal treatment. This standard type of analysis assumes identical extinction coefficients for the different structural elements, and has been shown to be subject to absolute errors of 2.5–4% in structural content.^43,44 The choice of fitting FTIR spectra profile using Gaussian components instead of Voigt or Laurentian profile is euristic, and based on the assumption that the intrinsic structural heterogeneity of protein samples gives rise to Gaussian broadening of spectral peaks. The deconvolved components were found empirically to have Gaussian envelopes and the excellent agreement obtained between observed and calculated spectra demonstrates a posterior the validity of the selected model.

Light scattering

Scattering intensity was measured by employing a Zetasizer Nano ZS (Malvern Instruments) with a 633 nm light beam and operating in the back scattering configuration (173°). Experiments were performed at 60 °C, inducing protein aggregation on 1 ml of freshly prepared ConA sample (8 ± 0.1 mg ml⁻¹; pD ∼ 2.6) in a PMMA UV-Grade cuvette (Kartell). Measurements were recorded every 3 min during the thermal incubation.

Confocal fluorescence microscopy

After incubation of ConA (8.0 ± 0.1 mg ml⁻¹ and pD ∼ 2.6) at 60 °C for 300 minutes (hereafter called F-ConA), the sample was stained using ThT. 300 μl aliquots of stained sample were placed on microscope slides and imaged at 1024 × 1024 pixel resolution using a Leica RCS SP5 confocal laser scanning microscope with a 63× oil objective NA = 1.4 (Leica Microsystems, Germany) and a scanning frequency of 400 Hz. The excitation was set at 458 nm and emission detected in the range 480–600 nm. Pinhole was 95 mm.

Fibril dissolution experiments

To study the effect of electrostatic forces, we changed the pD of mature fibrils sample (F-ConA sample). Identical 200 μl aliquots of F-ConA sample were titrated using NaOH at different concentration in D₂O. The FTIR spectrum of the equilibrated samples was measured. In all the experiments reported here, structural changes readily occur and the final structure of the sample remains stable for different days indicating that a new equilibrium is reached in solution.

Raster image correlation spectroscopy (RICS)

To perform RICS experiments, freshly prepared ConA (8.0 ± 0.1 mg ml⁻¹; pD ∼ 2.6) has been mixed with ConA-Alexa647 (final concentration 80 nM) and then incubated at 60 °C for 300 min. Images were acquired in one channel with an Olympus FluoView1200 confocal laser scanning microscope (Olympus, Tokyo, Japan) using an UPLSAPO 60× 1.2 NA objective. We used 633 nm laser for Alexa647. The bandwidth of the emission filter used for the red emission channel was 655–755 nm. For RICS analysis, 70 frames image stack were acquired (256 × 256 pixels) using a scan speed of 12.5 μs per pixel, (line time was 4.325 ms and frame time was 1.15 s). The electronic zoom was set at ×16.3 (pixel size of 0.05 μm). Analysis was performed using the RICS algorithm of SimFCS program (for details on RICS analysis see ref. 45 and 46). RICS autocorrelation function has been fitted using the diffusion model to calculate the characteristic parameters of the process: the amplitude of the correlation function G₀ and the diffusion coefficient D. The excitation volume was calibrated using a solution of Alexa647 in water containing 0.01% Tween 20 w/w. The diffusion coefficient D of the dye was set to 300 μm² s⁻¹ to determine the waist (ω₀) of the laser beam which resulted ω₀ = 0.35 μm. For this measurement the scan speed was reduced to 4 μs per pix. Importantly, control measurements were performed to assess that pH variations in the range 2–12 did not change Alexa647 fluorescence signal in such a way to hinder RICS analysis. Measurements of the same sample (ConA containing ConA-Alexa647) before thermal incubation and as a function of pH were also used for “internal calibration” of the system and are reported in the ESI.†

Results and discussion

Fig. 1a shows the temporal evolution of the amide I′ band during the thermal treatment of a ConA sample (8.0 ± 0.1 mg ml⁻¹; pD ∼ 2.6) at 60 °C for ∼300 minutes.

Fig. 1 (a) Time evolution of ConA (8.0 ± 0.1 mg ml⁻¹, pD ∼ 2.6) FTIR spectrum in the amide I′ region during thermal incubation at 60 °C. The arrows indicate the direction of changes as a function of time. (b) Deconvolution of spectrum acquired at 6 minutes in terms of Gaussian components (begin of aggregation kinetic, after thermal equilibration of the sample): black circles are the experimental points; black dashed lines are the Gaussian components, the continuous red line is the overall fitted spectral profile. (c) Deconvolution of the spectrum acquired at 300 minutes (end of the kinetic) using same spectral components as in (b) (see the text for more details). The insets show fit residuals on an expanded scale.

FTIR spectra in this region gives quantitative information on protein secondary structures⁴⁷ and their changes take into account for detailed modifications occurring during aggregation process.^48,49 At the beginning of the kinetics the absorption spectrum shows a dominant peak at 1635 cm⁻¹ (typically assigned to native β-sheets) together with a small shoulder centred at about 1693 cm⁻¹ (attributed to β-sheet antiparallel conformation).^50,51 As a function of time a significant increase of the peak centred at 1620 cm⁻¹ is observed together with the growth of a shoulder centred at ∼1672 cm⁻¹. In parallel, the decrease of the absorbance signal at 1635 cm⁻¹ and at 1693 cm⁻¹ occurs. The peak at 1620 cm⁻¹ is typical of aggregate β-sheets with very strong intermolecular hydrogen bonds and it is considered one of amyloid structure peculiar features.⁵¹ The peak at higher wavenumbers (1672 cm⁻¹) coupled to the feature at 1620 cm⁻¹ suggests the presence of intermolecular antiparallel β-sheet.^50,51

In order to have a deeper information on ConA secondary structure changes during thermal treatment, we consider deconvolution of amide I′ peak in terms of single Gaussian components. In Fig. 1b and c we show, as an example, the deconvolution of the spectrum acquired at 6 minutes (beginning of aggregation process after thermal equilibration) and after 300 minutes (end of aggregation kinetics) of incubation at 60 °C, respectively. Same analysis with a similar good match between data and fitting was performed for different data points during the kinetics (not shown). Following a well-established approach to interpret the amide I′ band of proteins,^47,50 we modelled the data using five main absorption components (1620, 1635, 1657, 1672 and 1693 cm⁻¹, plus two extrapolation components), each assigned to a specific backbone conformation, neglecting any possible contribution from residue side chains. Peak assignments are in line with other results obtained for ConA where both native, oligomeric and fibrillar structure were analyzed by means FTIR.^32,51,52

The main structural changes occurring during the aggregation process in the present conditions (high protein concentration and acid pH) are related to an increase of the band at 1620 cm⁻¹ and at 1672 cm⁻¹. These are related to the formation of aggregated β-sheets with very strong multiple-stranded structures and of turn structures. Moreover, we observe a simultaneous progressive decrease of the band at 1635 cm⁻¹, at 1657 cm⁻¹ and at 1693 cm⁻¹, assigned to native intramolecular β-sheet, to random coil conformation, and to native β-sheet antiparallel conformation, respectively.

The presence of isosbestic points at ∼1630 and ∼1690 cm⁻¹ (usually indicating interconverting chemical or structural species) implies the conversion between two secondary structures, i.e. from a native to an aggregated one. In particular, intramolecular β-sheet conformation (1635 cm⁻¹ band) and β-sheet antiparallel conformation (1693 cm⁻¹ band) characterizing the native structure convert into aggregated β-sheet structures (1620 cm⁻¹ band and 1672 cm⁻¹ band).

As can be seen, all the selected spectral components except the one at 1693 cm⁻¹ are maintained at the end of aggregation kinetics. In particular the peak at ∼1635 cm⁻¹ is still present. Similarly, native turn and loops seems to be partially conserved as suggested by the persistence of peak at ∼1657 cm⁻¹. This possibly indicates that part of protein native structure is retained upon aggregation.

Here, the main aim of Gaussian deconvolution procedure is to better visualize temporal features of the process with a focus on the fate of significant secondary structure components. The time evolution of the fractional area (i.e. the intensity of each peak contribution normalized to the total area of amide I′ peak) of each Gaussian component is reported in Fig. 2 together with the time evolution of the elastic scattering intensity measured for the same sample at the same temperature (red circles, right axis in the top panel).

Fig. 2 Time evolution of the elastic scattering intensity for the same sample as in Fig. 1 during thermal incubation at 60 °C (red circles, right axis) together with fractional area of the different Gaussian components resulting from the analysis of data in Fig. 1 (black circles, left axis) (see the text for details).

Variations of the fractional area account for relative absorption variations in the spectrum and therefore can be interpreted as a change in the protein secondary structure during thermal treatment. During the growth of the fractional areas of the components assigned to aggregated β-sheet structures (1620 and 1672 cm⁻¹) there is a simultaneous decrease of the fractional area of the components typical of the native structures.

The formation of large objects is taken into account by the fast and monotonic growth in the elastic scattering intensity. Data clearly indicate that an aggregation process occurs already at the very early stages of the process without any significant lag phase. Interestingly, macroscopic supramolecular assembly monitored by scattering intensity appears to be completed after only 15 minutes, while secondary structure changes monitored by FTIR are observed until ∼300 minutes. In previous studies, the aggregation process of ConA has been monitored in several different conditions. It was shown that conformational and structural changes, which depend on environmental conditions, occur along the reaction and the specificity of these changes leads to different aggregate morphologies. In particular, a common feature is that in the low concentration regime (<1.5 mg ml⁻¹) and at pH values below or above protein isoelectric point, fibrils are formed via intermolecular β-structures stabilization from native β-sheets. Data in Fig. 1 and 2 reveal that ConA in the present experimental conditions (high protein concentration and acidic pH) undergoes fast massive supramolecular association while changes in molecular structures leading to intermolecular β-sheets formation are slower and possibly occur in the already formed aggregates.

Global observation of data in Fig. 1 and 3 reveals a rather simple monotonic aggregation process, where the formation of intermolecular structures appear to parallel the decrease in native structure content, with no evident intermediate steps. This behaviour together with the isosbestic points presence suggests that part of native β-structures converts in intermolecular structures. In particular, the growth of the 1620 cm⁻¹ peak suggests the amyloid-like structure of the aggregates.

Fig. 3 1024 × 1024 representative laser scanning fluorescence confocal microscopy images of ConA aggregates stained with thioflavin T. Preparation at the end of the aggregation kinetics after incubation for 300 minutes at 60 °C. Images were detected from different areas of the sample.

The morphology of the aggregated structures at the end of thermal treatment at 60 °C was investigated by means of confocal microscopy which provides spatial information on the microscopic scale directly in solution. Representative confocal microscopy images of the sample stained with the amyloid-sensitive dye thioflavin T⁵³ are reported in Fig. 3. Micrometric ThT-positive aggregates with homogeneous morphology are observed. The supramolecular assembly of the observed structures appears to be constituted by a meshed net of elongated curly aggregates. This resembles the “crinkled” aggregates formed at low concentration (c = 0.5 mg ml⁻¹) and under acidic condition.⁴⁰

In the low concentration regime and at low pH, aggregation process of ConA at high temperature was found to be regulated by long range electrostatic forces and hydrophobic interactions, leading to a two-step temperature-dependent pathway leading to coagulation of early on-pathway intermediates into amyloid-like elongated fibrils. Structural and conformational changes were found to occur at the same time as the aggregates growth. Moreover, at 60 °C molecular reorganization of the aggregates was also highlighted during coagulation of non-conventional fibrils (crinkled intermediates) into more compact mature fibrillar structures. In the present conditions, at the same temperature used in our previous study⁴⁰ but at significantly higher protein concentration, short range protein–protein interactions become important. These interactions may have a dominant role favouring fast association leading to the formation of disordered aggregates. These species, due to the high temperature and charge of the molecules, progressively modify their structures toward more ordered systems.

The transition toward ordered and more stable large fibril-like structure requires the rearrangement or even the dissociation of the molecules. This is favoured by electrostatic repulsive interactions and drives the system toward the optimization of the intermolecular interactions of the backbone as well as of the side-chains. Specifically, the positively and negatively charged side chains may favour the fibrillar packing into similar supramolecular structures both in the low and high concentration regime.²² The nature of the association process in these solution conditions (independently on concentration regime) possibly implies that amino acid interactions and in particular hydrophobic, salt bridges and H-bonds must be energetically coupled. If electrostatics determines the ordering of the final structures, changes in such interactions should either stabilize or destabilize the aggregate structure.

Variation of the pH of the solution affects titratable groups charges and results in the modification of both global protein charge and detailed interactions between single residues. This results in protein conformation changes whose extent depends on the peculiarities of the polypeptide chain. Protonation or de-protonation of single residues may then occur, affecting protein global hydrophilicity, aggregation kinetics and aggregate stability.^9,12,19,29

In the following we focus on the analysis of the effects of pH changes on mature fibril-like. We explore pH-induced changes in ConA fibril-like samples obtained after incubation at 60 °C for 300 minutes.

Identical aliquots of fibril-like aggregates were titrated to reach desired pD (pD values are used instead of pH as fibrils are obtained in deuterated solution) and FTIR measurements were performed after sample equilibration. Fig. 4 shows the FTIR spectra at different pDs (red lines) in comparison with ConA fibril-like aggregates at pD ∼ 2.6 (black lines). The difference between the spectrum after pD variation and original sample is reported in green. In the hypothesis that extinction coefficient of secondary structure species does not significantly vary with pD in the sample,^43,44 our results show that fibril structure undergoes significant changes when pD is modified. In the samples where the pD is raised to values up to pD 10 (Fig. 4a–e) control measurements by confocal microscopy show that the fibril size and morphology is retained on the micrometric scale (see ESI Fig. S1†). Changes in the FTIR spectra shape are observed, and in particular they are characterised by a reduction of the intensity of the peak at 1620 cm⁻¹. Moreover an increased absorption in the spectral range between 1630 cm⁻¹ and 1690 cm⁻¹ is observed. The extent of the intensity variations is different at different pDs. It appears that if pD is increased in the range between 3 and about 7, i.e. at pD values closer to isoelectric point, protein molecules change their structure toward a more disordered structure and the highly ordered β-sheet is partially destroyed as a consequence of the reduction of protein net charge. When the sample is brought to final pD values in the range 7–10, the intermolecular β-sheet content reduction is less evident and ordered β-structures are maintained, although they are slightly reduced in the sample. This is not surprising as ConA protein is known to form amyloid fibril at basic pH so that fibrillar structure is also favored in these conditions.³⁸ In all these measurements it is possible that native-like elements are recovered as suggested by the growth of absorption intensity in the native β-sheet (1635 cm⁻¹) or β-turns and random coil structure (around 1650 cm⁻¹) region. However, the complete recovery of the ConA native structure is never found as clearly evident from the residual presence of dominant peak at 1620 cm⁻¹.

Fig. 4 Fibril-like aggregate dissolution experiment: multiple aliquots of the same sample containing mature fibril-like aggregates of ConA obtained after 300 minutes of thermal incubation at 60 °C (pD ∼ 2.6) were titrated with NaOH to change sample pD. FTIR spectrum of sample at pD ∼ 2.6 (black line); FTIR spectrum of sample at different pDs, (red line); difference FTIR spectra subtracting the spectrum at pD ∼ 2.6 from those at higher pD (green line).

Our observations indicate that when pD is changed in the range between about 2.6 and 10 the fibril-like aggregate reorganize their secondary structure with a reduced β-sheet content and keep the same morphology.

Interestingly, when the pD is changed from about 2.6 to pD ∼ 12, FTIR spectrum reported in Fig. 4f shows that the peak at 1620 cm⁻¹ relative to intermolecular β-sheets is significantly lowered. Spectral components at higher wavenumbers increase, indicating the similar modification described above but with different extent. In particular we note that a peak at 1685 cm⁻¹ is distinguishable, possibly indicating that antiparallel molecular β structures are retained or stabilized in these solution conditions. Moreover the pD change from about 2.6 to 12 results in optical modifications of sample state: the solution turbidity due to the high scattering is dramatically reduced (see ESI Fig. S2†) suggesting at least partial dissolution of aggregates.

This latter hypothesis can be deeply investigated using fluorescence imaging methods and RICS. RICS provides a powerful tool to quantitatively measure molecule dynamics in multiple environments, mapping also aggregation processes. This technique allows to couple morphological information with quantitative measurements of fluorescent molecule diffusion coefficients.^45,46 RICS allows direct measurements in solution and allows in somehow to overcome “diffraction limited” resolution issues of confocal microscopy and to monitor dynamical events.

We applied RICS analysis on a ConA aggregated sample prepared using exactly the same protocol as described above but containing 80 nM labeled ConA-Alexa647 (molar ratio ConA-Alexa/ConA ∼ 1/3000). Fig. 5a shows a 1024 × 1024 representative confocal image of the aggregate sample. As can be seen the shape and distribution of fluorescent molecules is identical to the one reported in Fig. 3. As for images obtained for ThT-stained samples in Fig. 3, we observed a homogeneous distribution of clusters of “crinkled” fibril-like aggregates with size in the μm range. This measurement indicates that labelled ConA is widely distributed in the sample and that the final morphology of the aggregates is not changed by the presence of labeled molecules. For this sample the correlation function shape stems for not diffusing species (even after removing immobile fraction; see ESI Fig. S3†). A representative image of the same sample when the pD value was titrated to pD ∼ 12 is reported in Fig. 5b. The image is characterized by uniform fluorescence where no microscopic objects are distinguishable above instrumental resolution (about 200 nm). In line with previous observations, it is possible to infer that oligomers or free protein molecules whose size is smaller than the instrumental spatial resolution are present in the sample. To verify this hypothesis RICS autocorrelation function was calculated for 256 × 256 region of interest (ROI) in the same sample.

Fig. 5 (a) Representative 1024 × 1024 confocal image of fibril-like aggregates obtained by thermal incubation at 60 °C of 8.0 mg ml⁻¹ ConA sample (pD ∼ 2.6) containing ConA-Alexa647 (molar ratio ConA-Alexa/ConA ∼ 1/3000). (b) 1024 × 1024 representative confocal image of the same sample where the pD has been increased to 12. (c) Spatial autocorrelation (RICS) function measured for the sample in the panel (b). (d) Fit of the correlation function shown in (c). The values of the amplitude results G₀ = 0.017 ± 0.005 and the diffusion coefficient D = 10.8 ± 0.5 μm² s⁻¹.

The resulting spatial correlation function and the fit are reported in Fig. 5c and d, respectively. The shape of the autocorrelation function indicates that fast molecules are freely diffusing in the analysed ROI. RICS autocorrelation function has been fitted using the diffusion model.^45,46

The value of the diffusion coefficient obtained from the analysis is D = 10.8 ± 1 μm² s⁻¹. Several measurements were repeated in different ROIs and analogous results were obtained with diffusion coefficients ranging from 9 to 12 μm² s⁻¹. These values indicate that sample is mainly composed by small oligomeric diffusing species. To qualitatively give an estimation of the size of diffusing molecules we note that using the Stokes–Einsten diffusion law, we obtain an oligomer mean radius about 4 times higher than ConA “dimer” radius.

Moreover as further control and to have an internal reference, which takes into account non-ideal measurements conditions, we also measured diffusion coefficients of labelled ConA in 8.0 mg ml⁻¹ ConA solution (i.e. in similar crowding conditions with respect to the sample in Fig. 5b) at different pH (see ESI Fig. S4†). At pD ∼ 2.6 where ConA is expected to be in dimeric state,⁵⁴ results give a D = 46 ± 0.5 μm² s⁻¹; a significantly higher value with respect to the one measured for sample in Fig. 5b.

Summarising all the presented results indicate that pD changes from about 2.6 to 12 causes fibril-like aggregates dissociation in small oligomers which are stable for several hours. Altogether, presented experimental results indicate that fibril dissociation process produces oligomeric structures still containing β-aggregates as indicated by the presence of the peak at 1620 and 1680 cm⁻¹ in Fig. 4f.

This last result is also confirmed from far-UV CD measurements reported in ESI (see ESI Fig. S5†).

ConA fibril-like structures are produced in experimental conditions well below the isoelectric point (i.e. positive net charge of the protein). These structures can be readily dissolved when the solution pD is brought to extremely basic pD values, well above the isolectric point of the protein. Changes in solution pH, i.e. change of global charge of amino acids residues, may cause the destabilization of the aggregates and possibly their disruption. The complete loss of the stability of the aggregates observed away from the isoelectric point may either be due to the abrupt “inversion” of the global net charge of the polypeptide chain or to changes in the charge state of single residues. In particular, it is possible to ascribe this behaviour to the amino acid side chain characterized by pK_a values close to or above 12. In ConA structure, 6 arginine, 7 tyrosine and 12 lysine residues⁵⁵ are present, whose pK_a has been reported to be in this range.⁵⁶ These residues are likely to participate in salt bridges capable of kinetically stabilizing the fibrils. When pH rises above the pK_a value the charge of these residues is modified from positively to negatively charged²⁹ causing fibrils disruption. In solution at high pH values these residues modify their net charge affecting the complex balance of interactions which stabilizes the fibrils. The variation of electrostatic forces may produce the reorganization of hydrogen bonds, protein–protein, protein–solvent and solvent–solvent hydrogen bonds and this may induce the rupture of some intermolecular hydrogen bonds constituting fibril spine.

All the above evidences points towards highly anisotropic interactions between protein molecules as responsible for the aggregate stability. These are strongly dependent on the spatial distribution of the chargeable residues on the single protein surface.^57,58 This, together with the initial not fully folded state of ConA, makes the quantitative prediction of the microscopic interactions not trivial.⁵⁹ However, in the energy landscape perspective it is possible to infer that an abrupt change of pH results in modifications of energy landscape shape in such a way that favourable conditions for fibril-like structures may not be readily accessible and native or oligomeric intermediate minima are more favourable. These mechanisms are similar to point mutations which dramatically alter the protein energy landscape and trigger formation of aggregate structures with different structural peculiarities and stability through different aggregation pathways.⁶⁰ The effect of varying the environmental conditions in samples containing amyloid fibrils formed from other proteins^29,61,62 were recently reported highlighting the common nature of the effects reported here.

Conclusions

Here we investigated aggregation and disaggregation process of ConA in particular highlighting the role of protein charge and of electrostatic interactions on the association process and on aggregate stability. Experiments were performed at high temperature in a concentration regime in which molecules collisions are favored and protein–protein solvent-mediated interactions cannot be neglected. In these conditions, notwithstanding the high charge on the protein, a fast massive supramolecular association triggered by temperature occurs followed by a structural reorganization leading to intermolecular β-sheets formation. Amyloid-like elongated curly aggregates are observed in the micrometric scale and they are positively stained by ThT. The analysis of experimental data together with the information already reported in the literature, suggests an association process regulated by hydrophobic and electrostatic interactions leading to the formation of intermolecular β-structures. In particular, structural peculiarities of ConA allow the observation of monotonic structural conversion from native β structures to intermolecular β-sheets. The characteristics peaks of native parallel and antiparallel β-sheets monotonically decrease in parallel with the growth of peaks of parallel and antiparallel intermolecular structures. These structural changes proceed for a longer temporal scale compared to the formation of large supramolecular assemblies.

Association process is observed in conditions where protein net charge is high, so that final aggregate structure is the result of modulation of long range repulsive electrostatic interaction, hydrophobic and short range electrostatic attractive interactions. If solution pH is changed in the range between about 2.6 and 10, molecular structure of the aggregates is modified and the highly ordered intermolecular β-sheet are reduced in the proximity of the isoelectric point, but supramolecular structure of the aggregates is maintained. When the pD is brought to a value that critically inverts the charge state of the protein and in particular is brought to values above the pK_a of aminoacid residues possibly involved in stabilizing salt bridges, the fibrils are promptly dissolved, leaving in solution small oligomers that retain some intermolecular β-structures.

Our results have a two-fold impact. On one side our findings contribute to a more general understanding of protein aggregation processes as such. We indeed highlight the possibility to access new minima (i.e. oligomers) in the generalized energy landscape of a protein starting from a fibril-like structure. This could represent an alternative route for in vivo oligomer accumulation in the context of neurodegenerative disease.^26,63,64 On the other side, advances in protein-based biomaterials are very much related to the development of procedures to handle the protein material. This is in turn related to the intimate physico-chemical interactions between proteins and between protein and surroundings.⁶⁵ Our data bring about the possibility to tune protein self-assembled structure stability by changing the physical interactions rather than covalently modifying the protein.⁶⁵ If further elaborated, this will provide a new platform for the rational design of protein aggregate biomaterials.

Acknowledgements

We wish to acknowledge Molecular Biophysics and Nanotechnology group at the Department of Physics and Chemistry, University of Palermo and in particular Maurizio Leone, Matteo Levantino and Antonio Cupane for useful discussions. We thank Federica Piccirilli for helpful advices during the early stages of this work. Fluorescence microscopy measurement were performed at Microscopy and Bio-imaging Lab (Advanced Technologies Network Center, University of Palermo), http://www.chab.center/home-en/.

References

1. C. A. Ross and M. A. Poirier, Nat. Med., 2004, 10, S10.
2. T. P. J. Knowles, M. Vendruscolo and C. M. Dobson, Nat. Rev. Mol. Cell Biol., 2014, 15, 384.
3. D. M. Fowler, A. V. Koulov, W. E. Balch and L. W. Kelly, Trends Biochem. Sci., 2007, 32, 217.
4. S. Zhang, Nat. Biotechnol., 2003, 21, 1171.
5. I. Cherny and E. Gazit, Angew. Chem., Int. Ed., 2008, 47, 4062.
6. S. L. Gras, Adv. Chem. Eng., 2009, 35, 161.
7. T. P. J. Knowles and M. J. Buehler, Nat. Nanotechnol., 2011, 6, 469.
8. I. Usov and R. Mezzenga, ACS Nano, 2014, 8, 11035.
9. V. Foderà, V. Vetri, T. S. Wind, W. Noppe, C. Cornett, A. M. Donald, L. Morozova-Roche and B. Vestergaard, J. Phys. Chem. Lett., 2014, 5, 3254.
10. G. Schirò, V. Vetri, C. B. Andersen, F. Natali, M. M. Koza, M. Leone and A. Cupane, J. Phys. Chem. B, 2014, 118, 2913.
11. M. L. Smith, J. S. Sharp and C. J. Roberts, Soft Matter, 2012, 8, 3751.
12. V. Vetri and V. Foderà, FEBS Lett., 2015, 589, 2448.
13. U. Shimanovich, I. Efimov, T. O. Mason, P. Flagmeier, A. K. Buell, A. Gedanken, S. Linse, K. S. Akerfeldt, C. M. Dobson, D. A. Weitz and T. P. J. Knowles, ACS Nano, 2015, 9, 43.
14. S. T. Kumar, J. Meinhardt, A. K. Fuchs, T. Aumüller, J. Leppert, B. Büchele, U. Knüpfer, R. Ramachandran, J. K. Yadav, E. Prell, I. Morgado, O. Ohlenschläger, U. Horn, T. Simmet, M. Görlach and M. Fändrich, ACS Nano, 2014, 8, 11042.
15. V. N. Uversky, A. V. Kabanov and Y. L. Lyubchenko, J. Proteome Res., 2006, 5, 2505.
16. F. Chiti and C. M. Dobson, Annu. Rev. Biochem., 2006, 75, 333.
17. D. Thirumalai and G. Reddy, Nat. Chem., 2011, 3, 910.
18. A. J. Baldwin, T. P. J. Knowles, G. G. Tartaglia, A. W. Fitzpatrick, G. L. Devlin, S. L. Shammas, C. A. Waudby, M. F. Mossuto, S. Meehan, S. L. Gras, J. Christodoulou, S. J. Anthony-Cahill, P. D. Barker, M. Vendruscolo and C. M. Dobson, J. Am. Chem. Soc., 2011, 133, 14160.
19. V. Foderà, A. Zaccone, M. Lattuada and A. M. Donald, Phys. Rev. Lett., 2013, 111, 108105.
20. C. M. Dobson, Nature, 2003, 426, 884.
21. J. Adamcik and R. Mezzenga, Soft Matter, 2011, 7, 5437.
22. B. Raman, E. Chatani, M. Kihara, T. Ban, M. Sakai, K. Hasegawa, H. Naiki, Ch. M. Rao and Y. Goto, Biochemistry, 2005, 44, 1288.
23. O. Griffith Jones and R. Mezzenga, Soft Matter, 2012, 8, 876.
24. G. Zhang, V. Babenko, W. Dzwolak and T. A. Keiderling, Biochemistry, 2015, 54, 7193.
25. N. Hirota-Nakaoka, K. Hasegawa, H. Naiki and Y. Goto, J. Biochem., 2003, 134, 159.
26. M. G. Di Carlo, V. Vetri, G. Buscarino, M. Leone, B. Vestergaard and V. Foderà, Biophys. Chem., 2016, 216, 23 .
27. S. Jordens, J. Adamcik, I. Amar-Yuli and R. Mezzenga, Biomacromolecules, 2011, 12, 187.
28. N. Nespovitaya, J. Gath, K. Barylyuk, C. Seuring, B. H. Meier and R. Riek, J. Am. Chem. Soc., 2016, 138, 846.
29. S. L. Shammas, T. P. J. Knowles, A. J. Baldwin, C. E. MacPhee, M. E. Welland, C. M. Dobson and G. L. Devlin, Biophys. J., 2011, 100, 2783.
30. T. J. Measey and F. Gai, Langmuir, 2012, 28, 12588.
31. K. E. Marshall, K. Morris, D. Charloton, N. O'Reilly, L. Lewis, H. Walden and L. C. Serpell, Biochemistry, 2011, 50, 2061.
32. V. Vetri, R. Carrotta, P. Picone, M. Di Carlo and V. Militello, Biochim. Biophys. Acta, 2010, 1804, 173.
33. G. Sancataldo, V. Vetri, V. Foderà, G. Di Cara, V. Militello and M. Leone, PLoS One, 2014, 9, e84552.
34. J. Jung, G. Savin, M. Pouzot, C. Schmitt and R. Mezzenga, Biomacromolecules, 2008, 9, 2477.
35. M. R. Krebs, K. R. Domike and A. M. Donald, Biochem. Soc. Trans., 2009, 37, 682.
36. V. Vetri, G. Ossato, V. Militello, M. A. Digman, M. Leone and E. Gratton, Biophys. J., 2011, 100, 774.
37. J. Emsley, H. E. White, B. P. O'Hara, G. Oliva, N. Srinivasan, I. J. Tickle, T. L. Blundell, M. B. Pepys and S. P. Wood, Nature, 1994, 367, 338.
38. V. Vetri, C. Canale, A. Relini, F. Librizzi, V. Militello, A. Gliozzi and M. Leone, Biophys. Chem., 2007, 125, 184.
39. R. Carrotta, V. Vetri, F. Librizzi, V. Martorana, V. Militello and M. Leone, J. Phys. Chem. B, 2011, 115, 2691.
40. V. Vetri, M. Leone, L. A. Morozova-Roche, B. Vestergaard and V. Foderà, PLoS One, 2013, 8, e68912.
41. M. Stefani, Prog. Neurobiol., 2012, 99, 226.
42. N. Carulla, G. L. Caddy, D. R. Hall, J. Zurdo, M. Gairí, M. Feliz, E. Giralt, C. V. Robinson and C. M. Dobson, Nature, 2005, 436, 554.
43. D. M. Byler and H. Susi, Biopolymers, 1986, 25, 469.
44. G. Vedantham, H. G. Sparks, S. U. Sane, S. Tzannis and T. M. Przybycien, Anal. Biochem., 2000, 285, 33.
45. M. A. Digman and E. Gratton, Microsc. Res. Tech., 2009, 72, 323.
46. M. A. Digman, C. M. Brown, P. Sengupta, P. W. Wiseman, A. R. Horwitz and E. Gratton, Biophys. J., 2005, 89, 1317.
47. A. Barth, Biochim. Biophys. Acta, Bioenerg., 2007, 1767, 1073.
48. M. G. Santangelo, R. Noto, M. Levantino, A. Cupane, S. Ricagno, M. Pezzullo, M. Bolognesi, M. R. Mangione, V. Martorana and M. Manno, Proteins, 2011, 80, 8.
49. W. Dzwolak, V. Smirnovas, R. Jansen and R. Winter, Protein Sci., 2004, 13, 1927.
50. A. Barth and C. Zscherp, Q. Rev. Biophys., 2002, 35, 369.
51. F. Piccirilli, G. Schirò, V. Vetri, S. Lupi, A. Perucchi and V. Militello, Biophys. Chem., 2015, 199, 17.
52. G. Zandomeneghi, M. R. H. Krebs, M. G. McCammon and M. Fändrich, Protein Sci., 2004, 13, 3314.
53. H. LeVine, Protein Sci., 1993, 2, 404.
54. D. F. Senear and D. C. Teller, J. Biochem., 1981, 20, 3076.
55. J. H. Naismith, C. Emmerich, J. Habash, S. H. Harrop, J. R. Helliwell, W. N. Hunter, J. Raftery, A. J. Kalb and J. Yariv, Acta Crystallogr., 1994, 50, 847.
56. C. N. Pace, G. R. Grimsley and J. M. Scholtz, J. Biol. Chem., 2009, 284, 13285.
57. R. Mezzenga and P. Fischer, Rep. Prog. Phys., 2013, 76, 046601.
58. M. G. Noroamd and D. Frenkel, J. Chem. Phys., 2000, 113, 2941.
59. G. Foffi and F. Sciortino, J. Phys. Chem. B, 2007, 111(33), 9702.
60. S. Ricagno, M. Pezzullo, A. Barbiroli, M. Manno, M. Levantino, M. G. Santangelo, F. Bonomi and M. Bolognesi, Biophys. J., 2010, 99, 3402.
61. K. Yamaguchi, Y. O. Kamatari, M. Fukuoka, R. Miyaji and K. Kuwata, J. Biochem., 2013, 52, 6797.
62. D. Kurouski, R. K. Dukor, X. Lu, L. A. Nafie and I. K. Lednev, Chem. Commun., 2012, 48, 2837.
63. Z. Ye, D. Bayron, J. J. Agiulera, S. Srinivasan, Y. Wang, L. C. Serpell and W. Colón, J. Biochem., 2011, 50, 9184.
64. C. Haass and D. J. Selkoe, Nat. Rev. Mol. Cell Biol., 2007, 8, 101.
65. M. J. Webber, E. A. Appel, E. W. Meijer and R. Langer, Nat. Mater., 2016, 15, 13.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra16690c

---