Methionine 129 variant of human prion protein oligomerizes more rapidly than the valine 129 variant: implications for disease susceptibility to CJD

The human PrP gene ( PRNP ) has two common alleles that encode either methionine or valine at codon 129. This polymorphism modulates disease susceptibility and phenotype of human transmissible spongiform encyphalopathies (TSEs) but the molecular mechanism by which these effects are mediated remains unclear. Here, we compared the misfolding pathway that leads to the formation of β -sheet-rich oligomeric isoforms of the methionine 129 variant of PrP to that of the valine 129 variant. We provide evidence for differences in the folding behaviour between the two variants at the early stages of oligomer formation. We show that Met129 has a higher propensity to form β - 10 sheet-rich oligomers whereas Val129 has a higher tendency to fold into α -helical-rich monomers. An equimolar mixture of both variants displayed an intermidate folding behaviour. We show that the oligomers of both variants are initially a mixture of α -rich and β -rich conformers that evolve with time to an increasingly homogeneous β -rich form. This maturation process, which involves no further change in proteinase-K resistance, occurs more rapidly in the Met129 form than the Val129 form. Although the involvement of such β -rich oligomers in prion pathogenesis is speculative, the misfolding behaviour could, in part, explain the higher susceptibility of individuals that are methionine homozygote to both sporadic and variant CJD.


INTRODUCTION
The transmissible spongiform encephalopathies (TSEs) are a group of fatal, neurodegenerative disorders that affect humans and animals and are believed to be caused by a novel class of infectious pathogen, the prion (1,2). These diseases have attracted considerable interest not only because of their unique biology but also because of the appearance of new variant of Creutzfeldt-Jakob disease (vCJD) (3), which appears to be caused by dietary exposure to the causative agent of bovine spongiform encephalopathy (BSE) (4,5). At the heart of disease pathogenesis lies a poorly understood structural rearrangement of PrP, a host-encoded glycoprotein of the nervous and lymphoid systems. The normal cellular form of the prion protein (PrP C ) 10 undergoes a conversion that leads to the accumulation of an abnormal, conformationally-altered isoform (PrP Sc ). According to the "protein-only" hypothesis of prion propagation, PrP Sc is the principal or sole component of transmissible prions (6). PrP Sc differs from PrP C by increased β-sheet content, increased resistance to proteinase K and an oligomeric rather than a monomeric state (7).
The human PrP gene (PRNP) exists in two major allelic forms that encode either methionine or valine at codon 129, with allele frequencies of 0.63 and 0.37 in western European and American populations, respectively (8,9). This polymorphism is a key determinant of susceptibility to sporadic (10) and acquired (8,11) prion diseases and 20 may affect age at onset (12)(13)(14). Based on the analysis of 300 sporadic CJD subjects, Parchi et al. identified six distinct clinicopathological variants of sCJD, which appeared to be specified largely by the genotype at codon 129 and the physiochemical properties of PrP Sc (14). More recently, an extensive analysis, which included the metal ion-dependent conformation of PrP Sc , of a large number of sCJD cases showed that PrP Sc types are associated with the residue encoded at codon 129, the duration of illness and with neuropathological phenotype (15). PrP Sc types 1 and 4 have so far been detected only in Met129 homozygotes, type 4 being uniquely associated with vCJD, type 3 in cases containing a Val129 allele and type 2 in any PRNP codon 129 genotype (16). The codon 129 polymorphism seemed to modulate the pattern of 30 neuropathology and the extent of lesions in sCJD, as shown in a study that involved 70 patients who died in France between 1994 and 1998 (17). Codon 129 polymorphism also has epistatic effects on the phenotypic effects of mutations elsewhere in the prion gene. For example, the Asp178Gln mutation combined with a methionine at position 129 results in fatal familial insomnia (FFI) (18). In contrast, the same mutation with a valine encoded at position 129 results in familial CJD (19).
Similarly, Val129 homozygotes in association with the Phe198Ser mutation predispose patients to the Indiana kindred variant of Gerstmann-Straussler-Scheinker (GSS) disease (20). Furthermore, an increased prevalence of genotype Val/Val at the polymorphic site 129 has been described in patient with early onset Alzheimer's disease (21) and a shift in cognitive decline towards early age in carriers of PRNP129 Val/Val has recently been reported (22). 10 Despite the clear importance of the polymorphism at codon 129 in the PRNP gene and its link to disease susceptibility and pathogenesis, the molecular mechanisms by which these effects are mediated remain unclear. The in vitro, thermodynamic stability of recombinant PrP is not affected by the Met129Val mutation or by other substitutions related to inherited human prion diseases (23). Structural studies, however, have shown evidence for hydrogen bonding between residues 128 and 178, which might provide a structural basis for the highly specific influence of polymorphism in position 129 on disease phenotype that segregates with Asp178Gln (24). Molecular dynamic simulations of low pH-induced conformational conversion of PrP has provided further clues as to the role of residue 129 (25). In these 20 simulations, Met129 seems to interact with Val122 leading to the recruitment of more N-proximal residues into an expanding β-sheet. As a complementary approach, we have chosen to study the folding properties of the two PrP allelomorphs experimentally, by examining the misfolding pathway that leads to the formation of β-sheet-rich oligomeric isoforms (26). We show that PrP-Met129 has a higher propensity to form β-sheet-rich oligomers whereas PrP-Val129 has a higher tendency to fold into α-helical-rich monomers. We also provide evidence that the dynamics of maturation of the oligomers differ between the two variants. The maturation process occurs over time at the expense of the α-helical-like monomers at a faster rate for PrP-Met129 than for PrP-Val129. Once the oligomers from either allelomorph have 30 been formed they show similar proteinase K resistance that does not change throughout the oligomer maturation process. The observed differences in the misfolding of PrP-Met129 and PrP-Val129 could explain the high susceptibility of individuals that are methionine homozygote to sporadic as well as variant CJD.

MATERIALS AND METHODS
Cloning of PrP genes and protein purification. Genomic DNA encoding methionine/valine at codon 129 of PRNP gene (18) (26). This rapid oligomer formation on SEC-HPLC corresponded to time zero as opposed to time course analysis of oligomer formation during dialysis. This time course analysis of oligomer formation was carried out as follows: 1 ml of 6M guanidine-hydrochloridedenatured proteins (10 mg/ml or 30 mg/ml), was dialysed at room temperature against 2 litres of 20 mM sodium acetate, 0.2 M NaCl, pH 3.7, 2 M urea by use of a Slide-A-Lyser dialysis cassette (Perbio Science UK Ltd, Tattenhall, UK) of 10 kDa cut off.
Aliquots were withdrawn, after carefully shaking the cassette to ensure homogeneity, 10 after 30 min, 2, 4 and 24 hours and analyzed by SEC as described above. Protein peaks were manually collected for subsequent analysis. Circular dichroism was carried as previously described (28) to assess the secondary structure of the protein. On line capillary HPLC nanospray mass spectrometry. Mass spectrometry was carried out on a Quattro II tandem quadrupole mass spectrometer (Micromass UK Ltd, Altrincham, UK) equipped with on-line capillary HPLC as detailed in (29).
Briefly, the capillary HPLC was 180 µm i.d. and was packed with 3.5 µm Jupiter C18 resin (Phenomenex, Macclesfield, UK). A flow rate of 1 µl/min was used and proteins were eluted with a gradient from 0-70% solvent A to B, where solvent A was 95:5 water:acetonitrile with 0.05% trifluoroacetic acid and solvent B was 95:5 water:acetonitrile with 0.05% trifluoroacetic acid. The eluent was passed directly to the mass spectrometer which was operated in continuous flow nanospray mode. Full scan mass spectra (m/z 300-2100) were acquired every 5 seconds. 10

Characterization of recombinant human PrP-Met129 and PrP-Val129
Recombinant human prion protein variants, HuPrP 90-231 Met129 and HuPrP  Val129, were purified to homogeneity by immobilized-metal affinity chromatography (see supplementary figure 1) followed by RP-HPLC. Purified proteins were analysed by mass spectrometry to confirm purity and identity. Omitting DTT from the buffer used to solubilize the inclusion bodies and allowing disulfide bond to form in the oxidizing environment of the IMAC was found to yield a fully oxidized protein (see supplementary figure 1) that has characteristics of PrP C , such as a monomeric state, 20 high α-helical content and proteinase K sensitivity. All the studies presented in this paper were performed on proteins that included a C-terminal His-tag.
PrP-Met129 has an intrinsically higher propensity to oligomerize than PrP-

Val129
To assess the effect of the codon 129 polymorphic residue on the formation of nonnative isoforms, both protein variants were denatured in 6 M guanidine hydrochloride, 50 mM Tris-HCl pH 7.2, and allowed to fold under conditions favouring the formation of β-oligomer species (26). To dissect the in vitro folding pathway of prion protein misfolding we used size exclusion chromatography (SEC), dialysis and 30 circular dichroism (CD). The denatured proteins (30 mg/ml) were injected onto a SEC column that had previously been equilibrated in 20 mM sodium acetate, 0.2 M NaCl, pH 3.7, 1 M urea and 0.02% sodium azide and eluted with the same buffer. The elution profile of the Met129 variant shows two major peaks and one minor peak (Fig.   1A). Peak I (terminology as given by Baskakov et al (26)) eluted at 5.68 min and corresponded to high molecular weight aggregates (Fig. 1A). Peak II eluted at 6.54 min and corresponded to the oligomeric isoform while peak III, which eluted at 9.08 min, had a similar retention time to monomeric protein (Fig. 1A). The elution profile of Val129 variant shows the same three peaks but in different proportions, with peak II eluting as a shoulder to peak I (Fig. 1B). In order to mimic a heterozygote situation of an individual that carries both alleles we have analyzed the folding behaviour in a 1:1 mixture of Met129 and Val129 ( Fig 1C). The elution profile of the equimolar mixture showed three peaks as seen before ( Fig 1C). Integration of the peaks in figure  10 1A and 1B revealed that less than 2% of the Val129 variant had formed oligomers under these rapid refolding conditions (within the first 10 minutes of the SEC-elution) as compared to more than 70% in the case of Met129 (Fig. 1D). In addition, the percentage of monomeric population that was formed within 10 minutes was calculated to be about 66% for the Val129 variant but only 24% for the Met129 variant ( Fig. 1D). Integration of the peaks in the equimolar mixture showed that the percentage of oligomeric form was reduced to about 44%, corresponding to approximately 30% reduction in the amount of oligomers as compared to the situation of the Met129 alone (Fig. 1D). Also the amount of the monomeric population in the mixture 1:1 increased by about 17% when compared to that of the Met129 alone ( Fig.  20 1D). The monomeric forms of Met129 and Val129 variants yielded CD spectra with two minima at 208 nm and 222 nm, indicative of a predominantly α-helical conformation (data not shown). The CD spectra of protein from peak II of PrP-Met129 and peaks I and II of PrP-Val129 did not give clear information about the protein conformations, but suggested that these fractions were composed of a mixture of α-helical and β-sheet forms.
To investigate the formation and evolution of the oligomers that eluted in peak II in more detail, we used a slower dialysis method of refolding that allowed analysis at different times after initiation of the refolding. Two protein concentrations, 30 and 10 30 mg/ml, were analyzed in parallel. The composition of the dialysis buffer was the same as that used in SEC elution buffer, except that, to maximize the formation of oligomers, the urea concentration was increased to 2 M (28). After 30 minutes refolding, and at 30 mg/ml, both allelomorphs adopted predominately oligomeric forms (peak II) with virtually no monomeric protein present (peak III); this elution profile did not change over the course of the experiment ( Fig. 2A). Similar results where obtained with a mixture (1:1) of Met129 and Val129 (data not shown). The CD spectra of oligomers collected at 30 min and 1 h showed that the population was not dominated by β-sheet rich structures but contained some α-helical rich protein ( Fig.   2, C-D). However, by four hours the CD spectra of the protein contained in peak II from both allelomorphs were typical of proteins possessing high amounts of β-sheet and the spectra remained constant thereafter.  From these data we conclude that, under these conditions, the Met129 variant has an intrinsically higher propensity to form oligomer isoforms than the Val129. The high oligomerization propensity of the methionine variant becomes even more apparent under the highest protein concentration and rapid refolding conditions. Since this is true for denatured, aggregated or natively-folded monomeric protein, the higher propensity of oligomerisation is clearly independent of the starting state of the protein.

The oligomers of both Met129 and Val129 variants are conformationally heterogeneous and show different kinetics of maturation
It has been shown that the oligomeric isoforms of mouse and Syrian hamster PrP are formed of populations of structurally heterogeneous proteins (26). RP-HPLC analysis showed that reduced forms of recombinant mouse PrP existed in multiple β-sheet rich 10 isoforms with distinct retention times (30,31). Accordingly, we used RP-HPLC to investigate the effect of the residue encoded at codon 129 on the composition of the oligomer population that eluted from SEC in peak II. Aliquots of this fraction were incubated at room temperature in SEC elution buffer (20 mM sodium acetate, 0.2 M NaCl, pH 3.7, 1 M urea and 0.02% sodium azide) and analyzed at various times (Fig.   3). After two days incubation, the RP chromatogram of PrP-Met129 showed two peaks with distinct retention times (Fig. 3A), however for PrP-Val129 only one major peak was observed (Fig. 3B). A second peak could be seen only after 12 days incubation. The two peaks were designated (IIa) and (IIb), with shorter and longer retention times, respectively. In order to rule out the possibility that the two peaks 20 represented unexpected modification to the proteins, we analysed them by mass spectrometry (Fig. 4). The deconvoluted mass spectra of PrP-Met129 IIa and PrP-Met129 IIb demonstrate that they represent proteins of the same molecular mass (Fig.   4, A-B). Similarly, the deconvoluted mass spectra of PrP-Val129 IIa and PrP-Val129 IIb showed that they also represented proteins with similar molecular masses (Fig. 4, . From these analyses we infer that no covalent differences are evident and that the two peaks observed by RP-HPLC represent different conformations, with different amounts of hydrophobic residues exposed at the protein surface.
Two important observations can be clearly made from RP-C4 chromatograms. Firstly, 30 the proportions of the two forms of PrP-Met129 and PrP-Val129 were different; the proportion of protein eluting in peak (IIa) was higher for PrP-Met129 after 12 days incubation (Fig. 3, A vs. B), while for PrP-Val129 there was more protein that eluted in peak (IIb). Secondly, the proportions of the two peaks changed over time at different rates for Met129 and Val129 variants. In particular, the proportion of protein eluting in peak (IIb) decreased at a higher rate in the oligomer of PrP-Met129 (Fig.   3A) than in the oligomer of PrP-Val129 (Fig. 3B). The peak (IIb) of the oligomer from both allelomorphs seemed to have a retention time similar to that of the corresponding α-helical monomers. Because RP-HPLC uses differences in the hydrophobic properties to achieve separation between bio-molecules, it can be concluded that the protein population present in the highly retained peak (IIb) has more surface-exposed hydrophobic residues than the protein population present in peak IIa. Furthermore, the protein population present in peak (IIb) appears to undergo conversion to become 10 recruited into the population that eluted in peak (IIa). Over time, this conversion seemed to occur much quicker with PrP-Met129 than with PrP-Val129 (Fig. 3, A vs.

B).
Because of the long storage time of both allelomorphs in SEC elution buffer at room temperature, we have used mass spectrometry to assess the integrity of the samples (see supplementary figure 3). After 2 and 12 days of oligomerization, both allelomorphs had molecular masses that matched those predicted from the amino acid sequences. However, after 30 days of oligomerization, the deconvoluted mass spectra of both variants showed an additional species with a molecular mass that 20 corresponded to the cleavage of the N-terminal two residues from the proteins (see supplementary figure 3). The two variants showed the same level of cleavage that increased over time to yield, 300 days later, proteins that were completely lacking two residues from the N-terminus (see supplementary figure 3). The cleavage occurred in the peptide bond linking an aspartyl residue to a prolyl residue (Fig. 5F). The lability of Asp-Pro bonds under acidic conditions has previously been reported (32) and provides a simple explanation for our observation. The cleavage occurred at the same rate for both Met129 and Val129 variants and at a region of the N-terminal that was PK sensitive (see below) and, therefore, was unlikely to impact the differences that were observed during oligomerization of the two allelomorphs. 30

The oligomers of Met129 and Val129 show similar proteinase K resistance
Limited PK digestion has been used widely to distinguish the disease isoform, PrP Sc , from PrP C as well as to probe differences between PrP Sc strains (33,34). We therefore used limited PK digestion in combination with on-line capillary HPLC mass spectrometry to probe differences between Met129 and Val129 variants during oligomerization. After treatment for up to 60 min with a 1:24 ratio of PK to protein, capillary HPLC_MS demonstrated that digestion of the oligomers from both allelomorphs yielded a range of short peptides that eluted before 20 min (insets in Fig.   5, A-B) that were liberated from the N-terminal and C-terminal His-tag regions of the protein. The central and C-terminal domains remained intact and eluted after 20 min as two peaks from the reversed phase column (insets in Fig. 5, A-B). These 10 polypeptides separated into two peaks that started either from residues 117 or 135 respectively, as a result of differential cleavage within the hydrophobic region (

DISCUSSION
In this work we show that the polymorphism at codon 129 affects the kinetics of oligomerization that leads to non-native isoforms of human prion protein. Starting from an unfolded state and when forced rapidly to adopt the oligomeric form under 30 low pH and at high protein concentration, the Met129 variant clearly exhibited a preference for the oligomeric state, whereas Val129 variant favoured the α-helicalrich, monomeric state. The apparent molecular mass of the oligomers from both PrP variants was estimated to be around 400 kDa, close to that previously reported (26).We have ruled out the possibility that the denatured starting protein contained any oligomeric forms of PrP by SEC analysis under denaturing conditions using 6M guanidine hydrochloride, 50 mM Tris-HCl pH 7.2 as elution buffer. Indeed the chromatograms showed only monomeric population of both PrP variants (see supplementary figure 4). How then does residue129 influence this folding behaviour?
The first clue can be derived from the solution structure of human PrP determined by NMR (35). Residue 129 lies within the first β-strand (residues 128-131) in human PrP. Riek et al (36) proposed that this short β-sheet might be a 'nucleation site' for a conformational transition from PrP C to PrP Sc that could include the loops connecting were potential sites for conversion to β-structures (25). In particular, the first β-strand served as a structural nucleus for the formation of new, and propagation of existing, β-like structure (25). The interactions between Met129 and Val122 and between Tyr128 and Tyr162 illustrated the role of tertiary interactions between side chains nucleating secondary structure formation, as has been observed for helix formation in barnase (40).
Our experimental data show that, under conditions that favour rapid formation of oligomers, SEC can be used to partition protein folded into aggregated, monomeric and oligomeric states so that differences in folding of Met129 and Val129 variants 30 become clear. It has been shown that folding of recombinant, murine prion protein into α-helix rich conformation is extremely rapid (41) and is under kinetic control (28). We show in this work that the propensity of recombinant human prion protein to fold into monomeric α-helical-rich population is increased when residue 129 was occupied by a valine. However, when the same position is occupied by methionine a tendency for oligomerization can be observed, with comparatively little monomeric α-helical-rich population. These differences in folding behaviour can be explained by a fundamental event in protein folding, which is driven by a small number of specific residues involved in the formation of a well-defined, specific nucleus that then allows the rest of the structure to coalesce efficiently around it (42,43). The high β-sheet propensity of valine residues as compared to methionine residues (44,45) makes PrP in which valine is present at residue 129 more favourable to adopt the native local βsheet that would temporarily stabilize the Val129 allelomorph in the monomeric α-10 helical PrP.
Differences between the oligomers formed from Met129 and Val129 could be seen when we explored the hydrophobicity of the oligomers as they evolved over time.
These differences were consistent with early events that were observed during the rapid misfolding process. That is, the proportion of oligomer subunits with hydrophobic properties resembling those of the α-monomer was higher in PrP-Val129 than in the PrP-Met129. Over the course of several weeks of incubation, the oligomerization process in both variants occurred at the expense of the α-helical-like population that became a substrate for oligomer recruitment (28,46). The presence of 20 the α-helical-like species in the oligomer reflected the heterogeneous nature of this non-native form of PrP. This finding was consistent with the previously published supra molecular electrospray ionisation mass spectrometry results (26). On the basis of these observations we propose a model (Fig. 6) that depicts the formation of nonnative β-oligomers in vitro. Under appropriate conditions (low pH and high protein concentration) the acquisition of this non-native form can happen from either denatured, aggregated or native forms and involves two steps: a rapid oligomerization phase followed by a slow maturation phase characterized by a structural rearrangement of the oligomers. The presence of a methionine at codon 129 affects both the kinetics of oligomerization and those of the maturation phase. 30 A hallmark of the disease isoform, PrP Sc , is its partial resistance to proteinase K digestion compared to normal cellular PrP. This property has also been used to probe differences between PrP Sc conformations. We found that once oligomers from both allelomorphs are formed they show similar proteinase K resistance that does not change throughout the oligomer maturation process. This indicates that if one relies solely on PK digestion to investigate PrP conformations, one could miss the detection of dynamic conformational changes influenced by the polymorphism at codon 129 in human PrP during disease pathogenesis. Similar PK-cleavage site (starting at residue 117) in the protease-resistant core of the oligomers from both human PrP variants has been reported for recombinant mouse PrP   (47). It is important to emphasis that this PK-cleavage site is not the same as the termini around codon 90 found for authentic PrP Sc (48). 10 It has been reported that two mutations in α-synuclein that are linked to early-onset Parkinson's disease can accelerate the oligomerization but not fibrilization suggesting that formation of nonfibrillar oligomer is likely to be critical in disease pathogenesis (49). There is no evidence so far that these β-oligomers exist in vivo or can directly Phylogenetically, methionine is the ancestral amino acid at codon 129 (52), and Val129 is a mutation found in humans. Worldwide PRNP haplotype diversity and coding allele frequencies suggest that strong balancing selection at this locus occurred during the evolution of modern human and made heterozygosity at PRNP a significant 30 selective advantage (53). Our data could provide a mechanism based on prion protein folding properties as an additional selective pressure against, particularly, Met129 homozygosity.         sodium acetate, 0.2 M NaCl, pH 3.7, 1 M urea and 0.02% sodium azide) at room temperature, we have used mass spectrometry to assess the integrity of the samples.

Figure legends
After 2 days of oligomerization, both allelomorphs had molecular masses that matched the theoretical ones. However, after 30 days of oligomerization the deconvoluted mass spectra of both variants showed an additional species with a molecular mass that corresponded to the cleavage of two residues from the N-terminal part of the proteins. The two allelomorphs showed the same level of cleavage that seemed to continue over time to yield, 300 days later, proteins that were completely lacking two residues from the N-terminus. The amino acid sequence of cloned human 10 PrP showed that the cleavage occurred in the peptide bond linking aspartyl residue to prolyl residue. The lability of Asp-Pro bonds has been reported before and therefore provides a simple explanation for our observation in the context of acidic conditions used during the oligomerization. The observed cleavage occurred at the same rate and equally between Met129 and Val129 and at a region of the N-terminal that was PK sensitive, therefore it was unlikely to impact the observed oligomerization differences