|Home | About | Journals | Submit | Contact Us | Français|
Misfolding of the natively α-helical prion protein into a β-sheet rich isoform is related to various human diseases such as Creutzfeldt-Jakob disease and Gerstmann-Sträussler-Scheinker syndrome. In humans, the disease phenotype is modified by a methionine/valine polymorphism at codon 129 of the prion protein gene. Using a combination of hydrogen/deuterium exchange coupled to NMR spectroscopy, hydroxyl radical probing detected by mass spectrometry, and site-directed mutagenesis, we demonstrate that stop mutants of the human prion protein have a conserved amyloid core. The 129 residue is deeply buried in the amyloid core structure, and its mutation strongly impacts aggregation. Taken together the data support a critical role of the polymorphic residue 129 of the human prion protein in aggregation and disease.
Misfolding of the natively α-helical prion protein (PrPC)3 into a β-sheet-rich isoform is related to various human diseases such as Creutzfeldt-Jakob disease and Gerstmann-Sträussler-Scheinker syndrome collectively known as transmissible spongiform encephalopathies (1). The form of disease is determined by two prion types (2), which are a cross species phenomenon (3). Prion types are characterized by differences in stability against denaturing agents, different proteinase K-cleavage sites, and different forms of prion aggregate deposits (3). These differences were attributed to distinct prion conformations (4). In humans, a methionine/valine polymorphism at codon 129 of the prion protein gene contributes in determining disease phenotypes (2). Although both prion types can be formed with each polymorphism at codon 129, methionine shows a strong association with prion type 1 and valine with prion type 2 (5). Homozygosity at this position predominates in sporadic Creutzfeldt-Jakob disease, where up to 89% of patients of large epidemiological studies had either Met/Met or Val/Val at codon 129 (6). It has also been demonstrated that to achieve a more efficient prion transmission in human PrP-transgenic mice models, both the inoculum and the mouse should bear the same amino acid at position 129 (7).
Although the high-resolution structures of PrPC from many species are well described (8), little is known about the structure of PrPSc. X-ray fiber diffraction showed that infectious prions have cross-β structure, confirming that prions can form amyloid (9). Antibody mapping studies suggested that when PrPC is converted into PrPSc, a conformational rearrangement occurs in the region comprising residues 90–176 (10). Mass spectrometric analysis of hydrogen-deuterium exchange of brain-derived PrPSc suggested that prion protein conversion involves refolding of the entire region from residues ~80–90 to the C terminus (11). Electron microscopy of two-dimensional crystals of the 27–30-kDa infectious fragment, PrP27–30, suggested that PrP amyloid fibrils might consist of stacked trimers of left-handed β-helices with the core formed by residues 89–140 (12–14).
Characterization of PrPSc at high resolution is currently not possible. Therefore, a variety of studies have investigated the structure of amyloid fibrils produced in vitro from recombinant prion protein. Mass spectrometry-coupled exchange experiments of amyloid fibrils of human prion protein (humPrP) comprising residues 90–231 indicated strong solvent protection in the region encompassing residues ~160–230 (15). In contrast, for mouse PrP(23–231), residues 24–98 and 182–212 were protected (16). Hydrogen exchange measurements by mass spectrometry and NMR spectroscopy of mouse PrP(89–143) bearing a P101L mutation identified a high level of protection from solvent exchange for residues 102–109 and 117–136 (17). In addition, molecular dynamics simulations provided models for the conformation of PrP in prions (18).
Hereditary prion diseases include C-terminally truncated variants of the prion protein, Y145X, Q160X, Y226X, and Q227X, where X indicates the truncation site (see Fig. 1A). Although a single experiment to transmit human 145stop mutant prions to mice failed (19), the stop mutants provide a particularly valuable model for studying misfolding of prion protein (20, 21) as they: (i) lack the normally present glycosylphosphatidylinositol anchor and are secreted into the intercellular medium, which appears to favor fibrillization and aggregation (21, 22); (ii) show in vivo prion deposits in the form of amyloidogenic plaques or plaque-like aggregates, like other prion diseases (23); and (iii) aggregate under nondenaturing conditions in vitro (20, 22). Previously, we showed that the β-sheet content is highly similar in amyloid fibrils of the Y145X and Q160X prion stop mutants (22), and solid-state NMR spectroscopy revealed extended β-sheet conformation in the 112–140 region (24–26). Here we investigated the importance of the polymorphic residue 129 for the formation and structure of amyloid fibrils of human prion stop mutants. Using a combination of hydrogen/deuterium (H/D) exchange coupled to NMR spectroscopy, hydroxyl radical probing detected by mass spectrometry, and site-directed mutagenesis, we demonstrate that the valine/methionine residue 129 is deeply buried in the amyloid core of the stop mutants, supporting its critical role in aggregation and disease.
Plasmids encoding Q160X and N174X were expressed in Escherichia coli BL21(DE3) by using M9 minimal medium with [15N]NH4Cl (1 g/liter) as the only nitrogen source and as required [13C6]glucose (4 g/liter), to obtain uniformly 15N- or 13C,15N-labeled protein, and purified according to Ref. 27. Variants of humPrP(108–143) with purity exceeding 95% were obtained from EZBiolab.
Aggregation of Q160X and N174X was started by transferring protein solutions into a phosphate buffer (50 mm potassium phosphate, pH 6.5) (22). The reaction was carried at 25 °C without agitation, and fibril formation was monitored using a thioflavin T fluorescence assay.
Variants of lyophilized humPrP(108–143) were dissolved at a concentration of 0.35 mm with ice-cold 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 cut-off 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 16,000 × g 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 7.5 m guanidine hydrochloride, 25 mm NaH2PO4, pH 6.5. Peptide concentrations were determined by UV absorption at 280 nm.
The sample was bound to carbon-coated grids and stained with 1% uranyl acetate. Pictures were taken at 72,000× magnification at a CM 120 with a 2048 × 2048 pixel TemCam (Tietz) in spotscan mode.
Fibrils were collected by centrifugation (10,000 × g for 5 min) and washed 3–4 times with 50 mm phosphate buffer (pH 6.5) to remove residual monomeric protein or low molecular weight oligomers. To initiate exchange, the pellet was resuspended in D2O buffer (0.1% formic acid in D2O, pD 2.5) and incubated at 4 °C for 3 and 7 days. The samples were then immediately frozen in liquid nitrogen and lyophilized. For the NMR experiments, lyophilized proteins were resolubilized in ice-cold solution of 2 m guanidinium thiocyanate in 50% H2O, 50% D2O, pD 2.5, and a series of 1H,15N heteronuclear single quantum coherence (HSQC) spectra was recorded over a period of 24 h to follow the back-exchange process (28). NMR spectra were acquired at 278 K on a Bruker Avance 900 MHz spectrometer equipped with a 5-mm triple resonance cryogenic probe head. Experiment time was ~23 min, and the dead time for sample preparation and acquisition setup was 15 min. NMR data were processed and analyzed using NMRPipe (29). For nonresolvable residues within the octarepeat region, protection levels were calculated using averaged intensities. Signals originating from residues Gly-30, Trp-31, Gly-35, Arg-37, Tyr-38, Tyr-49, Gly-53, Gly-55, Trp-57, Met-109, Lys-110, Met-112, Leu-125, Ile-138, and Tyr-145 were overlapping in the denaturing condition. As signal overlap might distort the analysis and lead to false protection factors, these residues were not included into the analysis. Exchange curves were fitted to Equation 1 using IgorPro 5.01
where I0 is the signal intensity observed in the first HSQC spectrum after dissolution, and A1–A3 are fitting variables.
Sequential backbone assignment of humPrP mutants in fibril dissolving buffer was obtained from three-dimensional HACANNH, HNCACB, and HNN experiments (30, 31) recorded at 278 K on a Bruker Avance spectrometer operating at 900 MHz and equipped with a 5-mm triple resonance cryo probe. The samples used for resonance assignment contained typically 0.4 mm 13C,15N-labeled variant humPrP in 2 m guanidinium thiocyanate, 90% H2O, 10% D2O, pH 2.5. Due to severe signal overlap in the octarepeat region, only 89 resonances for 123 non-proline residues of human humPrP(23–159) were observed in two-dimensional 1H,15N HSQC spectra of the denatured monomer. 64 of these resonances could be assigned unambiguously, whereas the rest could only be located to the octarepeat region.
The molecular weight of the amyloid fibrils was estimated using NMR pulsed-field gradient methods experiments (32). 10 mg of unlabeled fibrils were placed in a rotor, and diffusion coefficients were measured using the LED pulse sequence with bipolar gradients (33). Spectra were recorded on a Bruker Avance 900-MHz spectrometer equipped with a 4-mm high-resolution magic angle spinning probe. The sample was spun at 7.2 kHz, total diffusion time was D = 100 ms, and gradient pulse length d = 3 ms. Sine-shaped gradients were used, and their strength was incremented in 16 steps from g = 0.07–0.48 G/mm. The diffusion coefficients were calculated by fitting the intensity decay curves (see Fig. 1C) with the equation
where γ is the gyromagnetic ratio of the 1H nucleus, and t = 200 μs is the delay for gradient switching. For molecular weight estimation, 1,4-dioxane (88 Da) and bovine serum albumin (66 kDa) were used as internal and external references.
humPrP(108–143) was probed in the monomeric and fibrillar state by chemically generated hydroxyl radicals, as described in Ref. 34. 70 μl of 50 μm sample were added to an Eppendorf tube containing 10 μl of 13 mm Fe-EDTA (Sigma-Aldrich and Merck, respectively) solution and 10 μl of 0.2 m sodium ascorbate (Sigma-Aldrich) solution, both in 50 mm Na2HPO4, pH 6.5 buffer. For control purposes, a 17-μl aliquot was transferred to 40 μl of quenching buffer (2 m Tris, pH 7.2, Prolabo). Finally, 10 μl of 3% v/v H202 (Sigma-Aldrich) were added to the Fenton mixture to start the reaction. At 10, 60, 150, and 300 s, 17-μl aliquots were removed and transferred to new Eppendorf tubes containing 40 μl of quenching buffer.
To disassemble the probed peptides prior to further analysis, samples were dissolved in 200 μl of acetonitrile. The total volume was reduced to ~20 μl. Subsequently, 150 μl of 100 mm Tris, 10 mm CaCl2 buffer and 0.5 μg of chymotrypsin (Roche Applied Science) were added, and digestion was allowed to proceed overnight at 25 °C. The digested samples were desalted with C8 (3M) STAGE tips (35).
For LC-MS/MS analyses, samples were dissolved in 30 μl of sample solvent (5% v/v acetonitrile, 1% v/v formic acid). 5 μl were injected onto a nano-liquid chromatography system (Agilent 1100 series, Agilent Technologies) including an ~2-cm-long, 150-μm inner diameter C18 trapping column in-line with an ~15-cm-long, 75-μm inner diameter C18 analytical column (both packed in-house, C18 AQ 120 Å 5 μm, Dr. Maisch GmbH, Ammerbuch, Germany). Peptides were loaded on the trapping column at a flow rate of 10 μl/min in buffer A (0.1% formic acid in H2O, v/v) and subsequently eluted and separated on the analytical column with a gradient of 7.5–37.5% buffer B (95% acetonitrile, 0.1% formic acid in H2O, v/v) with an elution time of 37 min and a flow rate of 300 nl/min.
Online electrospray mass ionization-MS was performed with an LTQ-Orbitrap Velos instrument (Thermo Scientific), operated in data-dependent mode using a TOP10 method. MS scans were recorded in the m/z range of 350–1600. The 10 most intense ions were selected for subsequent MS/MS. Both precursor ions as well as fragment ions were scanned in the Orbitrap. Fragment ions were generated by higher energy collision dissociation activation (normalized collision energy = 40) and recorded from m/z = 100. As precursor ions as well as fragment ions were scanned in the Orbitrap, the resulting spectra were measured with high accuracy (<5 ppm) both in the MS and in the MS/MS level.
Data analysis was performed using MaxQuant 184.108.40.206 (36). Database search was performed against humPrP(108–143) with Andromeda (37) considering a total of 14 variable modifications: hydroxylation (+16 Da) in His, Val, Met, Leu, Tyr, Arg, Pro, Ile, Phe; dioxidation (+32 Da) of Met, Tyr, Phe; deguanidination (−43 Da) in Arg; and ring opening (−22) in His. The oxidation levels at any given time point were calculated on a per-residue basis. By making use of extracted ion chromatograms, obtained with Xcalibur (10 ppm tolerance), the ratios of oxidized to nonoxidized peptides in which a specific residue is involved were determined according to Ref. 38 and as illustrated below for Met-112
where O indicates the oxidized residue. Because ionization efficiencies of modified and unmodified species are similar, but not identical, the results need be interpreted in a relative context (i.e. fibrillar versus monomeric states). For quantification of peptides that contain several modifications in different combinations that lead to the same m/z value (termed isobaric oxidized peptides), we made use of the fact that these elute at different retention times from the LC into the mass spectrometer (ranging from 1 to 9 min with respect to the nonoxidized peptide). To the same purpose, only the most abundant charge state of each peptide was considered.
Two independently prepared fibril samples and two monomeric samples were measured and used for error calculation, and the error informed corresponds to the S.E. It should be noted that the sum of the intensities corresponding to different combinations of oxidized residues (the case for most methionines) lead to higher uncertainties, which stresses the need for the results to be interpreted in a relative context.
To characterize the molecular weight of humPrP(23–159) fibrils (Fig. 1B), we have used pulsed-field gradient methods (32) combined with high-resolution magic angle spinning (Fig. 1C). High-resolution magic angle spinning averages out NMR signal broadening caused by inhomogeneity in magnetic field anisotropy and thus allows detection of flexible regions in aggregates, whereas NMR-based pulsed field gradient methods provide an estimate of the diffusion coefficient of aggregates. To take into account the nonspherical shape of fibrils and contribution from rotational diffusion, the equations described by Perrin (39) were applied to calculate the corresponding diffusion constant of a spherical molecule of the same molecular weight. For this calculation, the average diameter and length of the fibrils were estimated from EM micrographs as 15 and 500 nm, respectively. Using the Stokes-Einstein relation, the molecular weight of humPrP(23–159) fibrils was estimated to be in the order of 4 MDa.
The solvent accessibilities of single residues in amyloid fibrils (Fig. 2) of the hereditary stop mutant Q160X were probed using NMR-detected H/D exchange (40). To allow detection by liquid-state NMR and at the same time preserve the H/D exchange pattern, fibrils that had been exposed to H/D exchange were rapidly converted to monomers in a 2 m solution of guanidinium thiocyanate in 50% H2O, 50% D2O (28). Back-exchange was then monitored in the denatured state by a series of two-dimensional 1H,15N HSQC spectra (Fig. 2). Neglecting back-exchange before and during the first HSQC after dissolution, the intensity ratio is 1.0 for an amide proton, for which the protection level in the fibril is 50%, and reaches a minimum of 0.5 for an amide proton with a 100% protection level in the fibrillar state.
To allow a residue-specific analysis of the H/D exchange, an assignment of the denatured protein in the dissolving buffer was required. Due to severe signal overlap in the octarepeat region, 89 distinct resonances for 123 non-proline residues of human humPrP(23–159) were observed in two-dimensional 1H,15N HSQC spectra of the denatured monomer. Sequence-specific backbone assignment was obtained for 64 of these resonances from a set of standard three-dimensional HACANNH, HNCACB, and HNN experiments (30, 31). The rest could only be located to the octarepeat region. Based on the sequence-specific resonance assignment, a solvent protection map was constructed that revealed back-exchange ratios of more than 1.2 for the segments 23–108 and 143–173 (Fig. 3A). In the central domain comprising residues 109–142, two regions could be distinguished: residues 109–120 and 140–142 with intermediate amide proton back-exchange ratios (exchange ratios of 1.0–1.2) and residues 121–139 with high protection values (>70%). A very similar protection map was obtained after 3 instead of 7 days of H/D exchange (Fig. 3A).
Comparison of H/D exchange profiles for individual residues (Figs. 2B and and33A) provides further insight. Firstly, within the region 109–140, Gly-119 and Ala-120 showed the most pronounced increase in signal intensity during the back-exchange process, suggesting that a turn may be formed by these residues. Secondly, the striking change in the exchange profiles from Ile-139 to His-140 points to a well defined boundary of the fibrillar core. Thirdly, residues 144–156 that form helix1 in the native structure of the human prion protein (8) showed high back-exchange ratios, comparable with the flexible N-terminal domain and suggesting that the helix1 region is highly solvent-accessible in amyloid fibrils of humPrP(23–159). Fourth, the polymorphic residue Met-129 is strongly protected from solvent exchange.
To probe the structural consequences of residues 160QVYY163, which form β-strand 2 in the native PrP structure, we performed H/D exchange on amyloid fibrils of the designed stop mutant N174X. The solvent protection map of N174X was very similar to the exchange profile of the Q160X stop mutant (Fig. 3B). The high solvent accessibility of residues 160–170 suggested that this region is not part of the fibrillar core in the N174X stop mutant. In longer peptides, however, more extensive structure could be present in this sequence stretch. The combined H/D exchange data indicate that the hydrogen-bonded core of amyloid fibrils of Q160X and N174X stop mutants of the human prion protein comprises residues 121–139, flanked by residues 109–120 and 140–142.
Next, we investigated the organization of single side chains in the amyloid structure of prion stop mutants using hydroxyl radical probing (Fig. 4). Hydroxyl radicals covalently modify side chains of solvent-accessible amino acids. The modifications induce a mass shift that can be detected by mass spectrometry (41). Measurements were performed for the peptide humPrP(108–143) that covers the fibrillar core of the stop mutants (Fig. 3) (24–26) and for which amyloid fibrils are morphologically similar to those of the stop mutants (Fig. 5). When compared with the stop mutants, the humPrP(108–143) peptide has two important advantages: (i) peptides obtained by solid-phase synthesis and stored as lyophilized powder are less prone to oxidation of methionine residues, a problem previously reported for recombinant prion protein (42); and (ii) the disordered N-terminal tail comprising residues 23–107 does not modulate the rate of aggregation.
Hydroxyl radical probing was applied to humPrP(108–143) in both the disordered monomer and the fibrillar state (Fig. 4 and supplemental Fig. S1). Oxidative modifications were observed for residues Met-109, Met-112, Tyr-128, Met-129, Leu-130, Met-134, Arg-136, Pro-137, Ile-138, Ile-139, His-140, and Phe-141 (Fig. 4 and supplemental Fig. S2). Other amino acid residues were not modified because of their lower reactivity or solvent accessibility. Strikingly, a pronounced reduction in oxidation was observed for the side chains of Tyr-128, Met-129, and Leu-130 in the fibrillar state (Fig. 4B). In addition, small differences in oxidation levels were present for Ile-138 and Phe-141. No changes in oxidation were observed for the side chains of Met-109, Met-112, Met-134, Arg-136, Pro-137, Ile-139, and His-140 despite the fact that the backbone amide protons of several of these residues are part of a fibrillar β-strand (24).
The burial of the side chain of Met-129 in the fibrillar core suggests an important role of this residue for aggregation of the prion protein. To test this hypothesis, we introduced point mutations into humPrP(108–143). Removal of the hydrophobic side chain by the substitutions M129N and M129D inhibited the aggregation of humPrP(108–143) (Fig. 5). In contrast, substitution of Met-109 by lysine did not have any effect, consistent with the location of Met-109 outside of the solvent-protected region (Fig. 3, Table 1). Importantly, substitution of Met-129 with valine reduced the lag time of fibril formation. Moreover, when mixing humPrP(108–143) with the Val-129 variant at a ratio of 1:3, the lag phase of the aggregation was in between that of the two pure peptides, whereas at an equal ratio of the two variants, aggregation was retarded when compared with the Met-129 variant alone (Fig. 5A). Thus, a 1:1 mixture of Met-129 and Val-129 humPrP(108–143) aggregates slower than the pure peptides at the same total concentration, suggesting that the two variants are not as active in the formation of aggregation nuclei as the pure sequences.
The importance of the 129 position for prion aggregation was further supported by seeding experiments. After 6 h of aggregation, Met-129 humPrP(108–143) amyloid fibrils were fragmented and homogenized with three 5-min sonication cycles. Three different volume/volume percentages (0.1, 1, and 5%) were used as seeds on fresh preparations of monomeric Met-129, Asn-129, and Asp-129 peptides. Regardless of the seed concentration, there was no seeding effect on either Asn-129 or Asp-129 peptides, whereas the Met-129 seed on Met-129 fresh monomer completely removed the nucleation phase (Fig. 5D).
Biochemical analyses of aggregated prion protein in patients with a stop codon mutation suffering from a prion disease have shown that the protein resulting from the mutated allele forms the aggregates (23, 43, 44). However, a single experiment to transmit human 145stop mutant prions to New Zealand White mice failed (19).
Using H/D exchange coupled to NMR spectroscopy, we showed that the Q160X and N174X stop mutants of the human prion protein have a common solvent-protected core (Fig. 3) that starts at residue 109 and ends at residue 142. Residues 160QVYY163, which form β-strand 2 in the native PrP structure, are not part of the fibrillar core of these stop mutants. In addition, residues 144–156 that form helix1 in the native structure (8) remain highly solvent-accessible in amyloid fibrils of humPrP(23–159). In vitro aggregated amyloid fibrils of Q160X and Y145X are morphologically similar according to electron microscopy (20, 22) and have similar proteinase K cleavage sites (22). Moreover, solid-state NMR measurements of Y145X located the β-sheet-rich amyloid core to residues 112–140 (24). Taken together the data suggest that their amyloid fibrils share a common structure starting at residue 109–112 and ending at residue 140–142. The identified fibril core is consistent with a major PrP fragment of ~7 kDa that was detected in amyloid fibrils purified from the 145stop mutant (43), from the 227stop mutant (44), and from Gerstmann-Sträussler-Scheinker brains and spans residues 81–82 to 144–153 (45). Moreover, the fibrillar core represents the most conserved sequence element in mammalian and nonmammalian prion proteins (46) and encompasses the peptide fragment 112–119 that is essential for the replication of the infectious agent in tissue culture (47).
Recently, some of us determined the backbone fold of amyloid fibrils of the Y145X stop mutant by application of the CS-Rosetta structure calculation program to experimental solid-state NMR chemical shifts (24, 48). The calculations revealed a left-handed β-helix formed by three β-strands. The Met-129 side chain is located in the beginning of the third β-strand and points inward to the hydrophobic core of the left-handed β-helix. The structural model is well supported by the solvent accessibility of individual side chains obtained by mass spectrometry. Met-129 shows a pronounced reduction of solvent accessibility in the amyloid state (Fig. 4), in line with the structural model proposed for the Y145X stop mutant (48).
Besides Met-129, the side chains of Tyr-128 and Leu-130 are solvent-protected in the amyloid state according to mass spectrometry (Fig. 4). In the β-helix model of the Y145X stop mutant, the two residues belong, together with Met-129, to a single β-strand formed by 128YMLGSAMSR136 (48). The experimental solvent accessibilities of the 128YMLGSAMSR136 β-strand suggest that both sides of this β-strand are protected. The side harboring the Met-129 side chain forms the hydrophobic core of the β-helix. In contrast, the side harboring Tyr-128 and Leu-130 is not protected in a single filament. A possible explanation for the solvent protection of the Tyr-128 and Leu-130 side chains in light of the β-helix model could be the pairing of two or more filaments, a hypothesis that would be supported by the dimensions of amyloid fibrils of the prion stop mutants observed by electron microscopy (22).
In humans, the disease phenotype is modified by a methionine/valine polymorphism at codon 129 of the prion protein gene in addition to the prion type. Increasing evidence suggests that changes between methionine and valine do not affect the folding or stability of the native structure, implying that its influence takes place at downstream stages in the disease. Based on the broadening of the signals of Met-129 in solid-state NMR measurements of the Y145X stop mutant, it was suggested that Met-129 is located in a flexible loop (24). In contrast, using a combination of H/D exchange coupled to NMR spectroscopy and hydroxyl radical probing detected by mass spectrometry, we demonstrate that the amide proton of Met-129 has slow solvent exchange in the fibrillar state and that its side chain becomes buried upon aggregation into amyloid. We therefore attribute the line broadening observed in solid-state NMR experiments for residues 121–141 (25) to dynamics between different filaments. At the same time, a high rigidity of a single PrP filament is supported by the finding that submicrosecond time scale dynamics are uniform across the amyloid core (25).
In agreement with the burial of Met-129 in the amyloid core, the presence of a valine residue at position 129 strongly enhanced aggregation of humPrP(108–143) (Fig. 5A). The strong effect of the Val-129 substitution might at first sight appear surprising as the Val-129 variant of humPrP(23–144) converted to amyloid fibrils only slightly faster than Met-129 in previous in vitro studies (49). We attribute the differences to the entropic barrier provided by the unstructured residues 23–108 thereby masking differences between different PrP variants. In support of an entropic barrier, aggregation of humPrP(23–144) and humPrP(23–159) requires severalfold higher concentrations than humPrP(108–143) (Fig. 5A) (22). Importantly, both the Met-129 and the Val-129 humPrP(108–143) peptide displayed a shorter nucleation phase than their 1:1 combination modeling the heterozygous situation. Indeed, humans that are heterozygote at codon 129 of the prion protein gene are relatively protected from prion diseases, and homozygotes, either methionine or valine, are 2–4-fold more susceptible (50).
The finding that the Val-129 variant aggregates faster than Met-129 of humPrP(23–144) is in contrast to previous aggregation experiments using recombinant PrP(90–231), where methionine is more aggregation-promoting than valine (51). Besides differences in the experimental settings, an explanation may be that the C-terminally truncated mutants aggregate to a different prion type than the N-terminally truncated one. From large epidemiological studies, we know that 87% of the methionine-homozygous sporadic Creutzfeldt-Jakob disease subjects accumulate prion type 1 and 97.5% of the valine homozygous Creutzfeldt-Jakob disease subjects accumulate prion type 2 (2). Taking this into account, the observed structure, starting at 109–112 and ending at 140–142, may represent the structural motif of prion type 2.
In summary, we demonstrated that the polymorphic residue 129 is deeply buried in the amyloid core of stop mutants of the human prion protein. In line with its burial in the amyloid core structure, the identity of the 129 side chain is important for aggregation into amyloid fibrils. Our data support a critical role of the polymorphic residue 129 in the packing of protein chains into prion particles.
We thank Dr. Michael Kramer for discussions.
*This work was supported by the VolkswagenStiftung (Grant ZN1294 to W. J. S.-S.), the Deutsche Forschungsgemeinschaft (Heisenberg scholarship to M. Z., ZW 71/2-2 and 3-2; Grant SFB 860 to M. Z., R. V. H., and H. U., INST 186/859-1), and the European Community's Seventh Framework Programme FP7/2009 under Grant Agreement 238316.
This article contains supplemental Figs. S1 and S2.
3The abbreviations used are: