Multiple Substitutions of Methionine 129 in Human Prion Protein Reveal Its Importance in the Amyloid Fibrillation Pathway*

Background: A polymorphism in position 129 in the human prion protein modulates susceptibility to prion infection and disease phenotype. Results: Mutations to various amino acids highlights the importance of position 129 during amyloid fibrillation. Conclusion: Position 129 is a key site for early intermolecular interactions during fibrillation. Significance: Insight into early mechanisms of aggregation implicates a means to prevent fibrillation. The role of the polymorphism Met or Val in position 129 in the human prion protein is well documented regarding disease susceptibility and clinical manifestations. However, little is known about the molecular background to this phenomenon. We investigated herein the conformational stability, amyloid fibrillation kinetics, and seeding propensity of different 129 mutants, located in β-strand 1 of PrP (Met129 (WT), M129A, M129V, M129L, M129W, M129P, M129E, M129K, and M129C) in HuPrP(90–231). The mutations M129V, M129L, M129K, and M129C did not affect stability (midpoints of thermal denaturation, Tm = 65–66 °C), whereas the mutants M129A and M129E and the largest side chain M129W were destabilized by 3–4 °C. The most destabilizing substitution was M129P, which lowered the Tm by 7.2 °C. All mutants, except for M129C, formed amyloid-like fibrils within hours during fibril formation under near physiological conditions. Fibril-forming mutants showed a sigmoidal kinetic profile and showed shorter lag times during seeding with preformed amyloid fibrils implicating a nucleated polymerization reaction. In the spontaneous reactions, the lag time of fibril formation was rather uniform for the mutants M129A, M129V, and M129L resembling the wild type. When the substituted amino acid had a distinct feature discriminating it from the wild type, such as size (M129W), charge (M129E, M129K), or rotational constraint (M129P), the fibrillation was impeded. M129C did not form ThT/Congo red-positive fibrils, and non-reducing SDS-PAGE of M129C during fibrillation conditions at different time points revealed covalent dimer formation already 15 min after fibrillation reaction initiation. Position 129 appears to be a key site for dictating PrP receptiveness toward recruitment into the amyloid state.

The role of the polymorphism Met or Val in position 129 in the human prion protein is well documented regarding disease susceptibility and clinical manifestations. However, little is known about the molecular background to this phenomenon. We investigated herein the conformational stability, amyloid fibrillation kinetics, and seeding propensity of different 129 mutants, located in ␤-strand 1 of PrP (Met 129 (WT), M129A, M129V, M129L, M129W, M129P, M129E, M129K, and M129C) in HuPrP(90 -231). The mutations M129V, M129L, M129K, and M129C did not affect stability (midpoints of thermal denaturation, T m ‫؍‬ 65-66°C), whereas the mutants M129A and M129E and the largest side chain M129W were destabilized by 3-4°C. The most destabilizing substitution was M129P, which lowered the T m by 7.2°C. All mutants, except for M129C, formed amyloid-like fibrils within hours during fibril formation under near physiological conditions. Fibril-forming mutants showed a sigmoidal kinetic profile and showed shorter lag times during seeding with preformed amyloid fibrils implicating a nucleated polymerization reaction. In the spontaneous reactions, the lag time of fibril formation was rather uniform for the mutants M129A, M129V, and M129L resembling the wild type. When the substituted amino acid had a distinct feature discriminating it from the wild type, such as size (M129W), charge (M129E, M129K), or rotational constraint (M129P), the fibrillation was impeded. M129C did not form ThT/Congo red-positive fibrils, and non-reducing SDS-PAGE of M129C during fibrillation conditions at different time points revealed covalent dimer formation already 15 min after fibrillation reaction initiation.
Position 129 appears to be a key site for dictating PrP receptiveness toward recruitment into the amyloid state.
The prion protein (PrP) 2 is most abundant in mammalian neurons but is ubiquitously expressed throughout various cells and tissues. The functional role of native PrP is not fully understood. PrP is associated with a number of different prionoses, both sporadic, inherited, and acquired, all of which are invariably fatal. The common molecular pathognomonic marker for prionoses is the presence of severe vacuolation within the CNS rendering a sponge-like tissue. Concomitant with the presence of spongiosis is the presence of a conformational isoform of PrP, which has converted from a largely helical globular protein PrP C that misfolds into an aggregation prone ␤-sheet conformation, PrP Sc , which often assemble into protein deposits with conspicuous similarities to amyloid (1).
Several inherited prion disease-causing point mutations are found in the human PrP gene (PRNP) (2,3). There are also two nonpathogenic polymorphisms, methionine or valine in position 129 (M129V) and glutamic acid or lysine at codon 219 (E219K) (4,5). These polymorphisms are not directly pathogenic but on the contrary hetrozygosity at either of these positions is associated with resistance to sporadic Creutzfeldt-Jakob disease (6,7). The polymorphism in position 129 is also associated with a sporadic Creutzfeldt-Jakob disease phenotype and influences the molecular type of prion strains (8). Furthermore, differences in codon 129 determine the phenotype of patients suffering pathogenic mutations elsewhere in the PRNP gene (9 -13). Homozygous individuals are more susceptible to acquired prion disease such as Kuru, variant Creutzfeldt-Jakob disease, and iatrogenic Creutzfeldt-Jakob disease (14 -19).
Taken together, there is ample evidence that position 129 in PrP is a key site for prion disease susceptibility and hence most likely for PrP misfolding. The effect of the natural polymorphism at position 129 has been studied in vitro in several studies regarding both amyloid fibrillation kinetics and oligomerization. Reorganization of HuPrP(90 -231) into mature amyloid fibrils under denaturing conditions is enhanced if position 129 encodes a valine, whereas methionine in this position favors formation of soluble oligomers (20 -22).
Amyloid fibril formation is a multiordered, nucleationdependent process (23). Because nucleation is a multiorder reaction, its rate especially at low protein concentrations should be stochastic. This feature is reflected by a time variation of the lag phase of fibril formation (24). During amyloidogenesis, the fibril elongates through monomer addition and conversion, and in addition, the fibril is fragmented leading to multiplication of fibril elongation sites. It is thus evident that fragmentation of amyloid fibrils is of great importance for fibril replication. The rate of the growth phase is a measure of fibril fragility rather than of cooperativity in monomer addition (25)(26)(27).
The amyloid formation pathways of many proteins involve formation of oligomeric species. These oligomers comprise varying numbers of non-native protein molecules, but most often, they do not display the features of amyloid regarding ThT fluorescence. It has been shown that oligomers of many proteins involved in neurodegenerative diseases are the most toxic species in cell culture (28,29), including the prion protein (30,31). PrP forms oligomers via different pathways, giving rise to different distinctly different oligomers from the same primary structure (32). The formation of different oligomers can also be modulated by the polymorphism in position 129. The propensity to form oligomers and the properties of the oligomers are not affected by the polymorphism, but the stacking of oligomers was suggested to be aggravated by the valine residue (33).
In our current study, we elucidated the conformational stability and amyloid propensity of recombinant HuPrP(90 -231) with variations in position 129. We made use of the range of properties offered by amino acid substitution. To investigate the role of increasing number of methyl groups, alanine, valine, and leucine were used. To elucidate the role of the sulfur atom in the wild type methionine, cysteine was included in the study. Lysine and glutamic acid should give information on the involvement of electrostatic interactions. Tryptophan was employed to examine influence of bulkiness. Proline was used to investigate how rotational constraint influences amyloid formation.

Expression and Purification of HuPrP(90 -231) and
HuPrP(121-231)-Plasmids (pRSET-A) containing HuPrP WT (Met 129 ) genes of desired lengths were obtained from K. Wüthrich (34). Site-directed mutagenesis was performed using the QuikChange mutagenesis kit (Stratagene, La Jolla, CA) according to manufacturer's instructions, and mutants were verified by DNA sequencing (GATC Biotechnology, Konstanz, Germany). The plasmids were transformed into BL21/DE3 cells and were grown on plates overnight. The cells were transferred to 1.5 liters of Terrific Broth (MP Biochemicals) supplemented with 100 g/ml ampicillin at 37°C with shaking. At A 600 ϳ 2.5, isopropyl ␤-D-thiogalactopyranoside was added to a final concentration of 1.5 mM to induce protein production. Agitation at 37°C was maintained during protein expression, and cells were harvested by centrifugation after overnight induction, and the pellet was resuspended in buffer G (6 M guanidine hydrochloride, 10 mM Tris-HCl, pH 8.0, 100 mM Na 2 PO 4 , 10 mM reduced glutathione) and frozen at Ϫ80°C until needed.
After thawing, sonication, and centrifugation, the soluble fraction was added to 3 ml nickel-nitrilotriacetic acid-agarose (Qiagen) in a disposable column. The agarose was washed with 10 ml of buffer G, 7 ml of 50% buffer G and 50% buffer B (10 mM Tris-HCl, 100 mM Na 2 HPO 4 , pH 8.0), 5 ml of buffer B, and finally, 5 ml of buffer B supplemented with 50 mM imidazole. The protein was eluted using 10 ml of buffer E (10 mM Tris-HCl, 100 mM Na 2 PO 4 , 500 mM imidazole, pH 5.8). The washing procedure was repeated until the elution fractions no longer contained protein detectable by the Bradford assay (Bio-Rad). The elution fractions were pooled, concentrated, and applied to a Hiload 16/60 Superdex 75 prep grade column (GE Healthcare) using a BioLogic LP system (Bio-Rad) with a continuous flow rate of 1 ml/min. Buffer F (50 mM Na 2 HPO 4 (pH 7.4), 100 mM NaCl, 50 mM KCl) was used for running buffer. Purity of obtained proteins was confirmed by SDS-PAGE (35). All described experiments with PrP were conducted in a biosafety P3** facility.
CD Spectroscopy-A Chirascan CD spectrometer (Applied Photophysics) and 1-mm cuvette was used for all measurements. Far UV spectra between 195 and 250 nm were measured at 4°C to assess the native protein structure. All measurements were performed with PrP dissolved in Buffer F. Thermal denaturation was monitored by recording the molar ellipticity at 222 nm, whereas the sample was heated from 4 to 90°C at 1°C/min. The sample was cooled down, and the far UV spectrum was again measured at 4°C for the refolded protein. The respective T m value was calculated according to John et al. (36).
Amyloidogenesis of HuPrP under Near Native Conditions-The protein was diluted in buffer F to the desired concentration (6 M), ThT was added to a concentration of 2 M, and the protein solution was aliquoted (100 l) into the wells of microtiter plates (CoStar NT 96-well, black with clear bottom, Corning). The plate was sealed with sealing tape. Typically, six identical samples were prepared for each fibrillation experiment. The plate was shaken in a Tecan SafireII plate reader at high speed in linear mode. Every 15 min, the shaking was halted, and fluorescence intensity was measured form the bottom of the plate by excitation at 440 nm and emission at 480 nm. Kinetic traces were fitted according to Almstedt et al. (35) to obtain the lag phase and the growth rate for each trajectory. Statistical analyses comparing experiments were performed using unpaired two-tailed t test using GraphPad Prism software (version 5). A p value Ͻ 0.05 was considered significant.
Polarization Microscopy-Congo red for microscopy analysis was added to mature fibrils in solution at a molar ratio of fibril:dye 3:1. Stained fibrils were left to self-sediment overnight at 4°C. The supernatant was removed, and the pelleted fibrils were transferred to Superfrost plus glass slides (Thermo Fisher, Waldorf, Germany) and covered with Mounting medium (Dako, Glosrup, Denmark), and coverglass was sealed with transparent nail polish. Congo red-stained samples were analyzed using a Nikon light microscope equipped with polarizers for both incoming light and in front of the detector.

Conformational Stability of Native HuPrP(90 -231) 129-Mutants-
The conformation of the native HuPrP(90 -231) protein variants was monitored by far-UV CD spectroscopy. All variants displayed a spectrum with two negative peaks at 222 and 208 nm followed by a positive peak at wavelengths below 200 nm (Fig. 1a). These features are consistent with native folded PrP dominated by ␣-helical structure and is in agree-ment with spectra in the literature (37). The thermodynamic stability of the protein variants was thereafter monitored by far-UV CD spectroscopy. Spectra were recorded at 4°C, followed by sequential steps of thermal denaturation monitored at 222 nm. After complete denaturation at 90°C, the sample was cooled to 4°C, and far-UV CD spectra were recorded again to assess any irreversible change of secondary structure upon one unfolding-refolding cycle.
All mutants except for M129P showed similar far UV CD spectra before denaturation (Fig. 1a). M129P displayed decreased positive amplitude between 195 and 200 nm compared with WT PrP, with a concomitant shift of the peak toward 197 nm indicating loss of helical structure (38). The thermal stability curves of all mutants showed a sloping pretransitional baseline followed by a cooperative thermal unfolding transition. Comparing the thermal stability of the mutants showed that the WT, M129V, M129L, M129K, and M129C were the most stable, and all within error showed T m values of 65-66°C ( Fig. 1b and Table 1). The M129A, M129E, and M129W mutants were destabilized by 3-4°C, and M129P was the most destabilized mutant with a T m lowered by 7.2°C. After thermal denaturation, refolding was allowed at 4°C, and the reversibility of the thermal unfolding was compared. Interestingly, the spectral feature of M129P described above at 197 nm also appeared in the spectra for M129W and M129E and, to some extent, M129A (Fig. 1c) after refolding. The M129P, M129W, M129E, and M129A mutants were also the most destabilized mutants and the amplitude of the irreversibility at 195-200 nm correlated well with the respective T m . Traditionally, the amount of ␣-helix and reversibility of helical structure is easily assessed at 222 nm. The reversibility of the CD signal at 222 nm for M129C and M129W showed the largest difference between the predenaturation and postdenaturation spectra (Fig. 1c), corroborating the spectral findings discussed above in the 195-200 nm region. The difference in molar ellipticity at 222 nm in the spectrum before and after thermal denaturation   (59), which calculated water-accessible non-polar surface areas (Å 2 ) of the side chains in the fully extended ␤-form of the peptide Gly-X-Gly by using the Richmond algorithm (60) and van der Waals radii of Richards (61).
was between 9 -12% for stable variants and 15-22% for mutants with lower T m values. One of the most stable variants, M129C, also displays the largest loss of helical structure upon refolding. This mutant lost 30% of its helical structure upon thermal denaturation-refolding (Table 1). As a reference for the CD measurements, the shorter variant HuPrP(121-231) (Met 129 ) comprising only the folded globular domain was employed. Comparing the shorter 121-231 construct and longer 90 -231 construct (HuPrP(90 -231)), the native spectrum, T m , and reversibility was indistinguishable within error for these parameters (Table 1) but did reveal a more cooperative thermal transition compared with the longer construct (Fig. 1b). Nevertheless, the identical T m rules out strong interactions between the globular domain 121-231 with the unstructured partial N-terminal 90 -120 domain on the folding properties of HuPrP(90 -231) in this assay.
Amyloid Fibril Formation of HuPrP 129 Mutants-Previously, we have described in detail that vigorous shaking of natively folded PrP in 2-ml cryotubes induced protein aggregation and amyloidogenesis in physiological buffer at 37°C (35). Aggregates produced according to this protocol were collected after 7 h of agitation, stained with Congo red, and placed on glass slides. Imaging with a polarization microscope revealed that all mutants except for M129C formed Congo red-stained birefringent aggregates, indicating the presence of symmetric Congo red binding sites, the tinctorial hallmark of amyloid fibrils (Fig. 2a). The aggregates of all mutants except for M129C were also positive for ThT fluorescence following excitation at 440 nm with an emission peak at 485 nm, which was absent for native protein (Fig. 2b). The ThT fluorescence intensity of the fibrils varied depending on mutant (Table 2).
Nonreducing SDS-PAGE of M129C-The lack of correlation between thermal stability and refolding capability together with the inability to form Congo red and ThT-positive fibrils of M129C led us to investigate whether this was due to intermolecular disulfide formation. Aliquots from fibrillation reactions of WT and M129C were collected at 0 min, 15 min, 2 h, and 4 h. Samples were boiled in non-reducing SDS containing sample loading buffer and loaded on SDS-PAGE gels followed by Coomassie staining. The WT displayed distinct monomeric band during the monitored time course. In the case of M129C, a dimeric protein was present already at the initial time point. As fibrillation proceeded, the monomeric band faded, and the dimeric band was enhanced (Fig. 3). At the end of the time course, a faint tetrameric band could also be observed. Dimer formation should be caused by intermolecular disulfide formation between cysteine residues at position 129.
Amyloid Fibrillation Kinetics-Formation of amyloid fibrils of HuPrP under near physiological conditions was monitored by measuring ThT fluorescence intensity during the fibrillation process in a 96-well plate format. Six separate samples of identical composition were used for each PrP variant. Every individual trace showed a characteristic lag, growth, and equilibrium trajectory (Fig. 4a, red lines). This is consistent with the notion that the amyloid fibrillation process of PrP is a nucleationdependent polymerization reaction. The ensembles of individual mutant trajectories of identical samples revealed that the protein showed a rather variable lag phase, indicating that PrP fibrillation kinetics is a process with a stochastic component, which appeared to be mutation-dependent. This notion is consistent with previous data showing an initial aggregation reaction preceding fibril formation, which renders a kinetic trap from which fibrils protrude (35). This is the background for the variations noted in the fibrillation trajectories. Lag times for all mutants are listed in Table 3. Comparing lag times during the spontaneous fibril formation reaction the wild type-like mutants (M129A, M129V, and M129L) were within the same  interval as the WT (ϳ5 h) (Fig. 4, a and b, and Table 3). Substitutions to conformationally constrained proline, bulky tryptophan, or charged residues (glutamic acid and lysine) substantially elongated the lag time of amyloid formation ( Fig. 4b and Table 3). The M129C mutant did not form ThT-positive fibrils within the time course of the experiment (Fig. 4a, note the different y axis, and Table 3). The growth phase of the fibril formation trajectories was defined as the slope of the ThT fluorescence following the end of the lag phase to half the normalized maximum ThT intensity for each trajectory (35). In the amyloid fibril formation reactions, the highest mean growth rates were obtained for WT and M129A. The clinically relevant mutant M129V, known to prolong incubation time of prion disease in heterozygote carriers and even protect against prion infection, had a slower growth rate than the wild type protein (Fig. 4, a and b, and Table 3), (p Ͻ 0.0001) on par with M129W, M129K, and M129E.
Effect of Seeding-To investigate the seeded fibrillation reactions, 1% seed from previous fibrillation reactions were added to fresh protein, and the fibrillation kinetics was monitored using the same setup and analyzing the same parameters as for the spontaneous reaction.
Seeding of the mutants with preformed fibrils of the same sequence efficiently shortened the lag time for the M129(M/A/ L/V/W/P/E) variants (Fig. 4a, black traces). The lag time of the M129K mutant was unaffected by seeding (Fig. 4, a and b, and Table 3). Addition of seed did however induce all samples (6/6) to convert into ThT-positive fibrils within the monitored time frame, whereas only 3/6 samples converted spontaneously (Fig.  4, a and b, and Table 3). The M129C mutant, which did not form ThT fluorescent fibrils, also was unaffected by seeding with 1% M129C protein subjected to a previous round of fibril formation conditions (Fig. 4a, note the different y axis, and Table 3). Interestingly, addition of 1% seed from preformed wild type fibrils did induce formation of ThT-positive fibrils of M129C on the same time scale as the wild type protein itself (Fig. 4a, inset). The strongest seeding effect was observed for the M129W and M129P mutants and the least affected mutant in terms of lag time (M129K excluded) was M129V (Fig. 4b).
Seeding of M129W also showed a prominent effect on the growth rate of amyloid formation, whereas the growth rates for the rest of the studied mutants were only affected modestly ( Fig.  4b and Table 3).

DISCUSSION
Misfolding of PrP C into PrP Sc in the prionoses frequently entails the accumulation of PrP aggregates with compelling similarities to amyloid deposits formed in local and systemic amyloidoses. High resolution structural models of PrP Sc have been presented (39). In two models, the PrP C molecule is partially refolded starting from the unstructured segment (from 90 or 114) up to position 175. The PrP molecules are herein arranged in intermolecular ␤-sheet configurations (␤-helical or spiral), comprising trimeric building blocks (40,41). The C-terminal helical domain is to a large extent intact in these models. On the contrary, there are other more dramatic structural rearrangements of misfolded PrP proposed by Surewicz et al. Herein, the PrP sequence, including 90 -231, has been shown to be essentially completely misfolded in vivo in prion-infected mice as deduced by hydrogen deuterium exchange kinetics (42) into an in-register parallel cross-␤-sheet amyloid state. The structural differences between these models and the evidence that deposited PrP is conformationally diverse within distinct prion strains (43), implicates that aggregated PrP is polymorphic. Therefore, several positions within the PrP sequence should influence PrP Sc structure and PrP C conversion. In the human prion protein, the 129 position has been shown to constitute the most prevalent single residue linked to prion pathogenesis and is known to be polymorphic for either Met or Val. These substitutions in position 129 also modulate the HuPrP(90 -231) amyloidogenicity under near physiological conditions in vitro. Surewicz and co-workers (44) have reported that M129V and WT have the same fibril formation kinetics when subjected to mildly denaturing conditions. In that study, the WT and M129V variants did not fibrillate under native conditions when the agitation was slow, in accordance with our previous report on stagnant incubation of WT HuPrP(90 -231) (35). Nevertheless, studies of the D178N mutation in context with either WT or M129V showed faster fibrillation rates both regarding a shorter lag phase and faster growth rate when in context with Met in position 129 (44). This observation is consistent with our data on M129V in this work (Table 3), where the growth rate for M129V was slower compared with WT both for unseeded (0.11 Ϯ 0.06 h Ϫ1 versus 0.37 Ϯ 0.11 h Ϫ1 ) and for self-seeded reactions (0.11 Ϯ 0.02 h Ϫ1 versus 0.44 Ϯ 0.09 h Ϫ1 ) (p Ͻ 0.0001). In addition the lag time for self-seeded WT (1.5 Ϯ 0.2 h) compared with self-seeded M129V (2.9 Ϯ 1.0 h) was significantly faster (p Ͻ 0.005). Here, we have compared the amyloidogenicity and conformational stability of nine different recombinant HuPrP(90 -231) sequences with substitutions in position 129 (including Met and Val) to investigate its possible effect on misfolding of the protein into the amyloid-like state under near native conditions.
The conformational stability of the M129V, M129L, M129K, and M129C mutations were all within error indistinguishable

Mutations in Position 129 of Human Prion Protein
from the WT protein. The M129A, M129E, M129P, and M129W mutants were less stable than WT. A possible correlation for the destabilization as a consequence of side chain were found when comparing the influence on stability with the apolar size of the side chain compared with the WT Met 129 ( Table  1). Influence of the unstructured 90 -120 sequence, which has  (90 -231) mutants. a, normalized kinetic traces (ThT) of amyloid fibril reactions for all investigated mutants were monitored for 24 h in 96-well plates. Spontaneous reactions were displayed in red, and the reaction was seeded with 1% fibrils displayed in black. The kinetic traces of M129C have been related to the intensity of the wild type (129M) reaction (note the different y axis). The inset in the M129C graph shows the kinetics of M129C seeded with 1% wild type fibrils (n ϭ 6 for each reaction). b, amyloid fibril formation kinetic parameters of HuPrP(90 -231) variants in the absence and presence of 1% seed from Fig. 4a (data for 129C not shown). The top panels show lag times and growth rates of the spontaneous reactions, and the bottom panels show the same parameters for fibrillation reactions seeded with 1% fibrils from the same sequence. The average lag times and growth rates were calculated for samples that converted within 24 h. The color code for mutant identification is shown at the bottom of the figure. The insets in the bottom panels show the differences (⌬) of lag time and growth rates compared with spontaneous reactions. been reported to become structured at neutral pH, was found to be very low because of indistinguishable conformational stability of the HuPrP(121-231) and the HuPrP(90 -231) variants. The most severely destabilizing mutation was M129P, which decreased the stability by 7.2°C. This observation is consistent with ␤-sheet breaker propensity of proline showing that ␤-strand 1 is rather important for PrP stability. To obtain a reference to exemplify the magnitude of destabilization of substitutions in position 129, we compared the stability results with two previously reported highly destabilized disease mutants associated with inherited prionoses: H187R (45) and F198S (46). When analyzed under the exact same conditions in the context of HuPrP(90 -231), the H187R mutant had a T m of 53.2 Ϯ 0.6°C and F198S had a T m of 48.6 Ϯ 0.3°C, providing ⌬T m values as compared with WT by 12.4 and 17.0°C, respectively (Table 1). Thus, the M129P mutation, although substantial, is less destabilizing than these highly destabilized naturally occurring pathogenic mutations.
Congo red birefringence and ThT fluorescence showed that eight highly diverse mutants in the 129 position formed amyloid fibrils under the conditions used herein (intense agitation of folded PrP in physiological buffer, at 37°C). In the kinetic trajectories (ThT assay), the fibril conversion competent mutants showed a lag phase followed by a rather steep growth phase which ended in an equilibrium phase at the end of the reaction. This behavior indicated that the protein regardless of substitution in the 129 position displayed nucleated polymeri-zation kinetics. This was especially evident when the amyloid fibril formation reactions were run in the presence of 1% preformed PrP fibrils as seeds, which rather consistently shortened the lag times. A schematic model of the fibril formation reactions is presented in Fig. 5. Spontaneous fibril formation (Fig.  5i) is initiated by dimerization followed by aggregation into oligomeric aggregates that slowly convert to amyloid fibrils (35). Fibril seeds added to a new fibrillation reaction (Fig. 5ii) render direct addition and rapid conversion of monomeric PrP.
In stark contrast to the other eight variants in position 129, M129C completely resisted spontaneous conversion into amyloid fibrils. Investigation of this mutant by non-reducing SDS-PAGE suggests that this is due to an intermolecular disulfide bond formation very early in the nucleation process (Fig. 3). This, in turn, abrogates further nucleation and fibrillation of this mutant (Fig. 5iv). Addition of preformed seeds from the WT sequence, however, expediently enabled formation of ThT-positive fibrils from the M129C sequence. This shows that the wild type seeds are able to recruit M129C monomers and convert them into the amyloid conformation, thus diverting M129C out of the kinetic trap introduced by covalent dimerization and places M129C on the seeded conversion track (Fig.  5ii). Importantly, dimerization has been shown to be the ratedetermining step in amyloid fibril conversion of mPrP albeit under different conditions compared with those used in our study (47). Interestingly, a dimer model within a crystallization interface where the ␤-stand 1 was paired in an antiparallel con-  forms a native dimer (step 1), which is further converted into a fibrillationcompetent misfolded oligomeric aggregate (step 2). This conformational rearrangement is stabilized by intermolecular interactions and exposes different surface amino acids than the native protein. The protein further converts into fibrillation nuclei (step 3), and fibril elongation is initiated (step 4), followed by fibril fragmentation (step 5) in the exponential growth phase.
ii, seeded reaction. The fibrillation nuclei can directly recruit and convert monomers into the fibrillar state (step 6). iii, in the case of M129K, marked in orange, the recruitment of native monomer to the nuclei is abrogated by positive charge repulsion caused by charged residues exposed on the surface of the fibrillation-competent conformation. The seeding mechanism is distorted. The spontaneous reaction also is, to some extent, delayed by positive charge repulsion (step 7). iv, for M129C, the formation of a native dimer enables intermolecular covalent disulfide formation (step 8), and conversion into fibrillation competent conformation is blocked.
figuration between two PrP monomers in a co-crystal complex with an antibody has been reported ( Fig. 6) (48). This observation shows that ␤-stand 1 has a propensity for inducing dimerization of PrP in the native state. Our data suggest that this state, if locked in this conformation, would be a conversionincompetent state. We base that conclusion on all of the other conversion competent 129 mutations in our study. This dramatic conformational reorganization of the C-terminal domain follows initial dimerization has been shown by structural studies of recombinant PrP amyloid fibrils (49) and studies of PrP mutants with intramolecular disulfide cross-links within the C-terminal domain that efficiently inhibited fibril formation (50). Although these studies observed that PrP fibrils formed under denaturing conditions, the latter study proposes the building block of mPrP fibrils to be composed of domain swapped dimers.
In the spontaneous reaction, the lag time of fibril formation was rather consistent for all wild type-like mutants (M129A, M129V, M129L). Residues with special features like bulkiness (M129W), rotational constraint (M129P), and charge (M129E and M129K) displayed longer lag times. We speculate that the initial aggregation/oligomerization step of the conversion reaction is affected by the M129W mutation by heightening the barrier of the conformational conversion from the oligomeric aggregate to amyloid fibril (step 3 in Fig. 5i). The pronounced seeding efficiency for the M129W mutation strengthens this hypothesis because this trap is circumvented. The M129P mutant showed a highly variable lag phase in the spontaneous reactions, which could reflect different conformational populations of proline cis-trans isomers during the initial dimerization and aggregation steps (steps 1-2 in Fig. 5a). The M129K mutant displayed an indifferent lag time when seeded but increased the conversion ratio (Table 3). This observation, in combination with the notion that glutamic acid in this position drastically shortens the lag time during seeding, leads us to conclude that positive charge (charge repulsions in the intermolecular interface of initial aggregation) is contributing to this alteration and hence follows an alternate route of association and conversion (step 7 in Fig. 5iii). Thus, the recruitment of native monomer to the nuclei appears abrogated by positive charge repulsion caused by charged residues exposed on the surface of the fibrillation-competent conformation. The seeding mechanism is distorted. The spontaneous reaction also is retarded by positive charge repulsion. This positive charge repulsion effect is supported indirectly by previous reports on polyanion (DNA, RNA, glycosaminoglycan, and luminescent conjugated polythiophene) modulation of PrP fibril formation and PrP C to PrP Sc conversion (51)(52)(53)(54)(55)(56).
Formation of intermolecular disulfides in the M129C variant was an early event in aggregation and fibrillation (Fig. 5iv). Hence, arresting the native PrP conformation by antiparallel locking of two monomers at ␤-stand 1 is putatively a strategy for intervention of PrP misfolding. The extent of the dimerization interface is unlikely to be limited to ␤-strand 1 but could involve an extended portion of PrP. This currently is being investigated in our laboratories. Nevertheless, given that a high resolution target for an antiparallel arrangement of ␤-stand 1 in a native state dimer interface was suggested by recent x-ray crystallography data (Fig. 6) (48), this could be an exploitable avenue for classical drug discovery endeavors. Similar small molecule kinetic stabilization of the native state has been exploited fruitfully to avoid misfolding of the tetrameric protein transthyretin (57,58). FIGURE 6. A putative dimerization interface of ␤-strand 1. The structure of HuPrP(121-231), highlighting ␤-strand 1 sequence 127-131 (in red) with an intermolecular distance of 3.9 Å between neighboring Met 129 residues was recently shown by x-ray crystallography The interface was revealed as a crystal packing interaction (48). Locking the antiparallell configuration of ␤-strands 1 and 1Ј by a disulfide bond in the M129C dimer results in a nonconvertable state. The structure was drawn using the program PyMOL using the Protein Data Bank code 2W9E.