The protonation state of histidine 111 regulates the aggregation of the evolutionary most conserved region of the human prion protein

Abstract

In a group of neurodegenerative diseases, collectively termed transmissible spongiform encephalopathies, the prion protein aggregates into β‐sheet rich amyloid‐like deposits. Because amyloid structure has been connected to different prion strains and cellular toxicity, it is important to obtain insight into the structural properties of prion fibrils. Using a combination of solution NMR spectroscopy, thioflavin‐T fluorescence and electron microscopy we here show that within amyloid fibrils of a peptide containing residues 108–143 of the human prion protein [humPrP (108–143)]—the evolutionary most conserved part of the prion protein ‐ residue H111 and S135 are in close spatial proximity and their interaction is critical for fibrillization. We further show that residues H111 and H140 share the same microenvironment in the unfolded, monomeric state of the peptide, but not in the fibrillar form. While protonation of H140 has little influence on fibrillization of humPrP (108–143), a positive charge at position 111 blocks the conformational change, which is necessary for amyloid formation of humPrP (108–143). Our study thus highlights the importance of protonation of histidine residues for protein aggregation and suggests point mutations to probe the structure of infectious prion particles.

Abbreviations

EM

electron microscopy

humPrP

human prion protein

NMR

nuclear magnetic resonance

ThT

thioflavin T

TOCSY

total correlation spectroscopy.

Introduction

Misfolding and aggregation of different proteins is regarded as the hallmark of many neurodegenerative diseases.1, 2, 3 Increasing evidence suggests that the β‐sheet rich, fibrillar stage of the aggregation process seems not to be the main culprit of cellular toxicity, but rather transient intermediate states, often called oligomeric species.4 While this notion has gathered wide‐spread support in recent years, it has also become apparent that differences in fibril morphology can lead to different phenotypes and different degrees of pathology and toxicity.5, 6 It is, therefore, of importance to gather insights into the structure of protein fibrils as found in neurodegenerative diseases. In particular, the profound change in secondary structure—from α‐helical to β‐sheet—and the concomitant aggregation of the prion protein (PrP) has been closely associated with transmissible spongiform encephalopathies.7, 8, 9 In humans, Creutzfeldt–Jakob disease and Gertsmann–Sträussler–Scheinker syndrome represent two of the most common manifestations of prion‐related pathologies.10, 11 An M/V polymorphism in position 129 of PrP is responsible for two different variants of CJD,12 whereas stop‐inducing mutations at positions Y145 and Q160 are related to Gertsmann–Sträussler–Scheinker‐like phenotypes.13 It has also been proposed that residues 138–141 play a central role in conformational change, regulating the early stages of aggregation.14

We have previously shown that the integration of chemical shifts, measured by solid‐state NMR spectroscopy15 with the structure calculation program CS‐Rosetta16, 17 allows modeling of the amyloid core of a fibrillar form of a small prion peptide, humPrP (108–143).18 Furthermore, we showed that within this model, the polymorphic residue 129 is buried inside the core of the fibril.18, 19 Using a combination of solution NMR spectroscopy, thioflavin‐T fluorescence and electron microscopy, we here probe selected features of the fibril model of humPrP (108–143) and show that the protonation state of H111 strongly influences the ability of humPrP (108–143) to aggregate into amyloid fibrils.

Results and Discussion

humPrP (108–143) contains two histidine residues at positions 111 and 140 [Fig. 1(A)]. According to NMR‐based hydrogen/deuterium exchange measurements and hydroxyl radical probing detected by mass spectrometry, histidine 111 is located inside the solvent‐protected fibril core, in contrast to histidine 140.19 In addition, the previously generated model of amyloid fibrils of humPrP(108‐143), which was calculated using CS‐Rosetta16 on the basis of solid‐state NMR chemical shifts15 and distance restraints reflecting the cross‐β structure of amyloid fibrils,18 predicted histidine 111, and serine 135 to be in close proximity.

Figure 1.

H111 and H140 are exposed to the same microenvironment in solution. A: Schematic representation of the main regions and characteristics of the prion protein. The mutation sites in red correspond to inherited forms of prion diseases. Residues 1–22 and 231–240 are removed during maturation of the protein. Orange boxes mark the location of α‐helices in the native structure of PrP.24 Residues 111–138 (green) are solvent‐protected in amyloid fibrils of humPrP (108–143). B: Superposition of a selected region from two‐dimensional TOCSY spectra of humPrP (108–143) showing cross‐peaks of the histidine side‐chain protons at increasing pH (in different colors). pH values are indicated. C: Chemical shifts of H111 and H140 as a function of pH.

We first asked whether histidine 111 and 140 are in the same microenvironment in the free monomeric state. To this end, we performed a pH titration of disordered monomeric humPrP (108–143), which probes the response of the two histidine residues to protonation. We changed the pH from 8 to ∼2 and followed the chemical shifts of the two histidine residues by two‐dimensional ¹H‐¹H TOCSY. Figure 1(B,C) show that both histidine residues respond in a similar manner as the pH decreases, resulting in a common pKa of 5.8 ± 0.5. In addition, substitution of histidine 111 by a lysine residue did not affect the pKa of histidine 140. The analysis showed that the imidiazole ring of both H111 and H140 in the monomeric, unfolded state of humPrP (108–143) has a pKa value close to that observed in an isolated histidine amino acid.20, 21, 22

Next, we examined the role of histidine 111 and 140 for amyloid formation of humPrP (108–143). To this end, we inserted point mutations into the humPrP (108–143) sequence and performed aggregation assays with both the wild‐type and the mutated peptides. Amyloid formation of the peptides was evaluated by binding assays using the amyloid‐specific dye thioflavin T (ThT) as well as electron microscopy (Fig. 2). The measurements showed that replacement of H111 by a positively charged side‐chain (lysine) inhibited the ability of humPrP (108–143) to form amyloid fibrils [Fig. 2(A)]. Moreover, introduction of a negatively charged aspartic acid at position 135, which is in direct spatial proximity to the imidazole ring of H111 in the amyloid fibril model of humPrP (108–143) [S135D, Fig. 2(D)], restored its aggregation capabilities [Fig. 2(A)]. The lag phase of amyloid formation was only slightly longer in the H111K/S135D variant when compared to the wild‐type protein, while its elongation rate was even faster. Consistent with a charge compensation mechanism, a peptide, in which the positive and negative charge at positions 111 and 135 were interchanged, was also able to form amyloid fibrils [Fig. 2(C)]. Notably, the aggregation lag phase of the H111D/S135K variant was roughly half of that of the wild‐type peptide, suggesting that additional factors, such as residues in direct proximity of the introduced mutation, might influence the aggregation kinetics of humPrP (108–143). In contrast to these charge effects, substitution of H111 by phenylalanine and S135 by alanine had less effect on the aggregation kinetics of humPrP (108–143) [Fig. 2(C), red and green data points, respectively].

Figure 2.

The protonation state of H111 regulates the aggregation of humPrP (108–143). A–C: Time‐dependent ThT‐fluorescence intensity during incubation of different variants of humPrP (108–143). Panel (A) shows that insertion of a positive charge at position 111 (H111K) abolishes fibril formation, while neutralizing it with a counteracting negative charge (H11K/S135D) restored fibrillization. B: Influence of pH on aggregation of humPrP (108–143). Protonation of H111 by lowering the pH from 7.4 to 4.5 prevents fibril formation, while protonation of H140 only slightly extended the lag phase of aggregation (H111K/S135D pH 4.5). C: Swapping of residues at positions 111 and 135 (H111D/S135K) accelerates aggregation, while insertion of noncharged amino acids (H111F, S135A) only slightly changes aggregation kinetics. D: Model of the structure of humPrP (108–143) in amyloid fibrils.18 E: Estimation of aggregate formation by a protein sedimentation assay. F: Electron micrograph of amyloid fibrils formed by the H111K/S135D variant of humPrP (108–143) at pH 4.5. The bar represents 200 nm.

The finding that replacement of H111 by a lysine blocked aggregation of humPrP (108–143) suggested that the protonation state of the imidazole ring of H111 could be an important regulator of the fibrillization of humPrP (108–143). In order to test this hypothesis, we repeated the aggregation of both wild‐type humPrP (108–143) and the H111K/S135D variant at pH 4.5. Because the NMR‐based pH titration had shown that the pKa value of H111 and H140 is 5.8, the imidazole ring of both histidine residues carries a positive charge at pH 4.5. All other ionizable residues, on the other hand, should not be strongly affected. Figure 2(B) shows how the decrease to pH 4.5 influenced the aggregation of humPrP (108–143). While wild‐type humPrP (108–143) rapidly formed amyloid fibrils at pH 7.4, no fibrillization was detected for it at pH 4.5 by both ThT‐fluroescence intensity [Fig. 2(B)] and protein sedimentation [Fig. 2(E)]. In contrast, H111K/S135D humPrP (108–143) efficiently fibrillized at both pH 7.4 and pH 4.5 [Fig. 2(B,E,F)]. The duration of the lag phase of H111K/S135D humPrP (108–143) increased by approximately 1 h at pH 4.5, while the elongation rate was similar at both pH values.

Taken together our study demonstrates that a positive charge in position 111 is able to block the conformational change that is necessary for fibril formation of humPrP (108–143), consistent with its location in the solvent‐protected core of the amyloid fibril. In contrast, H140 is located outside of the fibril core and does not play a critical role in amyloid formation of humPrP (108–143). The point mutations identified in the current work, such as H111K and S135D, will hopefully provide a useful tool to probe structural features in infectious prion aggregates.

Experimental Procedures

Variants of humPrP (108–143) with purity exceeding 95% were obtained from EZBiolab. Variants of lyophilized humPrP (108–143) were dissolved at a concentration of 0.35 mM with icecold 25 mM Tris buffer, pH 7.5, 0.02% sodium azide, and dialyzed overnight against 500 mL of 25 mM Tris, pH 7.5, 0.02% sodium azide at 4°C in a 500–1000 Da molecular mass cutoff dialysis membrane. The peptide was filtered through a 0.22 μm Millipore filter. Concentration was adjusted to 0.15 mM in a 500 μL final volume, and the solution was incubated at 12°C in low binding protein 1.5 mL Eppendorf tubes. After different incubation times, a 20 μL aliquot of the sample was mixed with 1 mL of the thioflavin T assay solution (0.1 mM thioflavin T in 100 mM NaH₂PO₄, 140 mM NaCl, pH 8.5), incubated 5 min at room temperature and transferred to a 10‐mm cuvette. The fluorescence emission was measured between 460 and 600 nm on a Varian Cary Eclipse fluorescence spectrophotometer (Agilent Technologies) with excitation at 442 nm at 20°C. To have a second measure for aggregate formation of different variants of humPrP (108–143), aggregates were collected after 2.7 and 24 h of incubation by centrifugation at 16,000g for 30 min at 4°C. Subsequently, the pellet was resuspended in 100 μL of 25 mM Tris–HCl, pH 7.5, and diluted in a 1:5 ratio in 7.5M guanidine hydrochloride, 25 mM NaH₂PO₄, pH 6.5. Peptide concentrations were determined by UV absorption at 280 nm.

NMR experiments were recorded at 10°C on a Bruker Avance 700 MHz spectrometer equipped with cryogenic probe. Two‐dimensional TOCSY spectra were recorded with 400 complex points in the indirect dimension, 16 scans per increment and spectral widths of 8389 and 1844 Hz, respectively. NMR spectra were processed with Topspin (Bruker) and analyzed with CCPN Analysis.23 The titration results were fitted with a sigmoidal equation and plotted as a function of the pH using the software SciDAVis.