Lyotropic phase behaviour of dilute, aqueous hen lysozyme amyloid fibril dispersions

Colloidal dispersions of amyloid fibrils were produced by incubating solutions of the native protein in aqueous HCl for 2 weeks at 65 °C in a closed hot oven to guarantee a homogeneous temperature environment. Solutions were kept in closed polypropylene vials (2 mL volume) that were tightly sealed with PTFE tape to prevent evaporation. [Note that, for unknown reasons, amyloid formation appeared to proceed at a slower rate if solutions were incubated in glass vials.]

Aligned thin films were prepared by doctor-blading concentrated amyloid dispersions on glass microscopy slides, which was achieved by shearing the dispersion with a razor blade that was held in close proximity to the glass surface. Amyloid films were stained with 1 wt% Congo Red (or PTAA) aqueous solution after drying; excess dye was removed by rinsing with distilled water.

The pH of dispersions was determined with a Biotrode pH-meter (Mettler Toledo).

TEM was performed with a Philips CM200 microscope, operated at 200 kV. Samples were prepared by depositing diluted amyloid dispersions (~0.5 mM with respect to initial protein concentration) on 200 mesh carbon-coated copper grids (Agar Scientific) and, after drying, stained with 2 wt% aqueous uranyl acetate solutions (three times for 10 s followed by rinsing with distilled water; Sigma–Aldrich).

Optical microscopy was carried out with an Olympus BH2 polarising microscope. Wet samples were sandwiched between glass microscopy and cover slides.

Polarised UV–Vis absorbance spectra were recorded using a Perkin Elmer Lambda 950 UV–Vis spectrophotometer.

Amyloid fibril dispersions from hen egg white lysozyme in aqueous hydrochloric acid (HCl) were prepared at previously established growth conditions, i.e. at a pH between 1 and 3 and a temperature of 65 °C [21, 22, 26, 27]. These conditions are ideal for amyloid formation because this range of pH is situated far below the isoelectric point of lysozyme (pI ~ 11).¹ The high positive net charge of the protein results in long-range repulsion, which permits the slow growth of regular β-sheet structures. Instead, incubation close to the pI would lead to rapid aggregation and thus the formation of irregular particulates [28].

Irrespective of the acid and protein concentration, nanofibrillar structures with a diameter of approximately 5 nm and a length in excess of one micrometre could be observed after subjecting protein solutions to elevated temperature, as evidenced by a series of transmission electron micrographs in Fig. 1. Nevertheless, the nucleation and growth rate of amyloid fibrils is likely to vary with composition and pH [26, 27], which may affect in particular their length distribution but also internal structure. For instance, based on a theoretical study, it has been argued that the length distribution of hen lysozyme amyloid fibrils depends on the initial protein concentration [29]. Hydrolysis of lysozyme can also occur, especially at pH < 3 and elevated temperature, but both the full length protein as well as its fragments are incorporated into amyloid fibrils [27]. In addition, it should be noted that despite the presence of numerous amyloid fibrils, it cannot be ruled out that a fraction of the initial protein remains present even after 2 weeks of incubation.

In order to elucidate the appearance of anisotropic phases, we examined amyloid dispersions with polarised optical microscopy (Fig. 2a). At dilute protein concentration, e.g. less than 0.2–0.7 mM initial lysozyme content, no birefringence was observed, indicating homogeneous isotropic dispersions, in agreement with a previous study [21]. In stark contrast, at intermediate concentrations around 0.7–1.4 mM, birefringent spherulitic structures suspended in an isotropic liquid had developed, which previously have been associated with β-sheet fibrils radially grown from a more disordered nucleus [20]. In addition, for sufficiently acidic dispersions, i.e. for acid concentrations of ~30 mM HCl and above (corresponding to −log[HCl] ≤ 1.5; c.f. Fig. 2), a heterogeneous distribution of fibrillar material in the form of spherulitic domains was found to extend to at least 4 mM initial lysozyme concentration. Here, it is critical to note that with increasing protein content spherulites ceased to be surrounded by an isotropic medium but, instead, were dispersed in a more birefringent liquid, which due to limitations of our imaging equipment is not apparent from the optical micrographs displayed in Fig. 2a. Less acidic dispersions, in contrast, featured a homogeneous nematic texture above a lysozyme concentration of 1.5 mM. However, spherulitic domains were not found to develop, when diluting nematic liquids, which instead separated into isotropic and anisotropic fractions, confirming that the spherulitic nature is particular to phase-separation during amyloid formation [20]. This is also consistent with our observation that spherulites did not form in protein solutions that were subjected to vortexing during incubation, which appeared to disrupt the formation of regular structures and gave rise to disordered birefringent aggregates instead, as also reported for amyloid fibrils from bovine insulin [30].

Interestingly, the distribution of homogeneous isotropic, homogeneous anisotropic and heterogeneous regions (i.e. the coexistence of either an isotropic and lyotropic phase or two lyotropic phases) is in good qualitative agreement with phase equilibria predicted for solutions of rod-like particles (Fig. 2b) [31], which is astonishing considering the fact that samples were prepared at vastly different incubation conditions. The narrow isotropic, biphasic and extended nematic region at low acid concentration, i.e. −log[HCl] > 1.5, result from appropriate volume fractions of rigid rod-like amyloid fibrils, as predicted by Onsager theory [32], which is thoroughly discussed in Ref. [23]. In contrast, the transition from a homogeneous to a heterogeneous distribution of material around −log[HCl] = 1.5 can be understood in terms of the change in electrostatic interaction of amyloid fibrils with the aqueous dispersion medium, which may—at least in part—give rise to the observed phase behaviour. Previous studies of amyloid fibrils from β-lactoglobulin or bovine insulin have revealed that the propensity for aggregation is related to the electrostatic repulsion between fibrils, which could be altered through partial screening. This was accomplished by varying the ionic strength of the dispersion medium [23, 30], as well as by adjusting the pH, resulting in aggregation around the isoelectric point, at which fibrils carry no net electric charge [24]. Similarly, for the here investigated system it appears that at high acid concentration and thus presence of a high amount of chloride ions, similar screening effects or, potentially, salt bridges led to formation of heterogeneous structures; interactions, which in another report were found to become significant for peptide-based β-sheet fibrils at pH ≤ 2 [16]. Hence, the tendency to develop heterogeneous textures only dominates at sufficiently low pH, i.e. at pH < 1.7–1.9, depending on the protein concentration (Fig. 3). It should also be noted that in the here investigated pH range, i.e. pH 1–3, aspartic and glutamic acid residues are likely to be protonated and, therefore, the net electrical charge of lysozyme will be largely invariant.

Heterogeneous dispersions, i.e. those that contain spherulites, don’t easily permit uniaxial alignment of the nanofibrillar β-sheet rich material. In contrast, homogeneous isotropic dispersions have been shown to order when subjected to flow fields [21, 33], which can be retained through suitable surface interactions [11, 12]. Homogeneous nematic dispersions instead, which readily feature local alignment of amyloid fibrils, usually displayed a high density of disclinations (c.f. Fig. 2a). However, largely defect-free thin solid films could be prepared through doctor-blading such liquids. Ordering was most visible when staining dried amyloid films with the conjugated electrolyte Congo Red, which resulted in pronounced apple-green birefringence when observed between crossed polarisers and gave rise to anisotropic absorbance as evidenced by polarised UV–Vis absorbance spectroscopy (Fig. 4). The absorbance maximum of the dye red-shifted from 502 to 507 nm, confirming binding to β-sheets [26]. Previous studies have shown that Congo Red and therefore its dipole preferentially orient with respect to the long axis of amyloid fibrils [34‐36], suggesting that here these structures tend to align with the shear direction during film deposition. Likewise, staining with the conjugated polyelectrolyte poly(thiophene acetic acid) (PTAA) gave rise to maximum light absorbance when illuminated with light polarised parallel to the shear direction and displayed a shift in peak absorbance from 430 to 450 nm (not shown), in accordance with another report [11]. In contrast to studies on single polyelectrolyte-decorated amyloid fibrils [11, 12], we could only achieve marginal orientation of the conjugated moiety as evidenced by the relatively low dichroic ratios of our oriented films. Besides insufficient ordering of the template, this may have resulted from the commonly observed helical nature of lysozyme amyloid fibrils [21, 26, 27], which will give rise to misalignment of the bound dye and, hence, its absorbing dipole with respect to the shear direction. In principle, however, straight fibrils, e.g. grown from shorter peptides [17], may facilitate efficient ordering of conjugated electrolytes in thin films, provided homogeneous lyotropic dispersions can be prepared as outlined in this study.