Fig 1.
(A) The native state of human transglutaminase (PDB-ID: 2XZZ) (B) a model protein in the native state, (C) disease-related amyloid β-sheet of the Aβ (1-42) peptide (PDB-ID: 2NAO) [28] and (D) the seed structure of an amyloid fibril represented in the lattice model used for this work. Side chains of hydrophobic residues are coloured in yellow, of polar residues in grey, of positively charged residues in blue and of negatively charged residues in red. For β-stranded structures the backbone is coloured in green. In the folded protein, a hydrophobic core can be observed, where the hydrophobic residues are shielded from the water by the hydrophilic and charged residues. Similarly, the core sequence regions of amyloid fibril structures tend to be strongly hydrophobic.
Fig 2.
Main techniques used in this work.
This work contains four main components: (I) Theory developed from previously published results, (II) A lattice model of fibril elongation, (III) Experimental measurements of the free energy and enthalpy of amyloid fibril elongation and the enthalpy of short peptide assembly, (IV) Meta-data analysis of the enthalpy of fibril elongation.
Fig 3.
Temperature dependence of the free energy components.
(A) Free energy, (B) Entropy and (C) Enthalpy of fibril elongation. The differences between the fully formed fibril (Cext = 21) and the monomeric state (Cext = 0) are calculated from the simulation of several different fibrils. The interaction strength of the Leucine-based fibrils is weaker, leading to a lower enthalpic contribution, effectively weakening the fibrils. Dotted lines indicate estimates for the hydrophobic contributions showing from left to right and
; these estimates are generated using Eqs 7 and 8 with corresponding α, ΔCh = −6 and with an offset, Eint = ΔH based on simulations with the equivalent peptide for α = 0; stars indicate the change of an exothermic to an endothermic process, based on the
estimate. It is clear that in our model the slope of the enthalpy of fibril elongation as a function of temperature, is dominated by the temperature dependence of the hydrophobic effect.
Fig 4.
Top: free energy landscapes, and bottom: representative snapshots are shown for simulations at three different temperatures modelled with a strong hydrophobic temperature dependence, α = 60. The snapshots of the peptides are representative for the low free energy conformations; the circle represents the exact order parameters for the snapshot in the corresponding free energy landscape. Within the snapshots the seed fibril is indicated in a lighter shade (see caption Fig 1 for colour coding). At high temperatures only transient (external) contacts are formed and the protein molecules that are not part of the seed remain in their monomeric form. At intermediate temperatures the free peptides adopt a regular, fibrillar structure at the end of the seed. At low temperatures the heatmap shows that there is not a single distinct conformation with a low free energy: the simulated free peptides are attached to the seed fibril, but no hydrophobic core is formed between the peptides.
Fig 5.
Enthalpy of amyloid fibril elongation and peptide and amino acid crystallisation at different temperatures.
The enthalpy of fibril elongation of α-lactalbumin, α-synuclein, glucagon and β-2-microglobulin [33] (solid lines), as well as the enthalpies of crystallisation of L-phenylalanine [60], diphenylalanine [61] and of GNNQQNY in water (dashed lines) are shown as a function of temperature. For α-synuclein, ITC experiments were conducted in two set-ups: fibrils titrated into monomer solution (red symbols) and monomers titrated into fibril solution (grey symbols). See the Materials and methods and S2 Methods for experimental details. (A) Eq 11 was fitted through each of the curves to estimate the strength of the hydrophobic effect (γ) in . (B) Relationship between the fitted values of γ and the total hydrophobic surface area of the proteins (Ah). 110 mM HCl, 1 mM Na2SO4, data from [34]. 210 mM HCl, 1 mM Na2SO4. 310 mM HCl, 30 mM NaCl.
Fig 6.
The effect of changes in ionic strength and temperature on the stability of α-synuclein fibrils.
We show the fraction of depolymerised fibrils as a function of urea concentration, determined by intrinsic fluorescence using the F94W mutant of α-synuclein. Fibril destabilisation was measured at four different conditions: at high (dark, 10 mM PB + 150 mM NaCl) and low (light, 10 mM PB) ionic strength, and at 28°C (red) and 4°C. The depolymerisation curves were fitted with the isodesmic linear polymerisation model, Eq 12. The data shows that fibrils become less stable at low temperatures, and that an increase in ionic strength stabilises the fibrils by screening repulsive charges in the fibril.