A New Method for the Characterization of Strain-Specific Conformational Stability of Protease-Sensitive and Protease-Resistant PrPSc

Although proteinacious in nature, prions exist as strains with specific self-perpetuating biological properties. Prion strains are thought to be associated with different conformers of PrPSc, a disease-associated isoform of the host-encoded cellular protein (PrPC). Molecular strain typing approaches have been developed which rely on the characterization of protease-resistant PrPSc. However, PrPSc is composed not only of protease-resistant but also of protease-sensitive isoforms. The aim of this work was to develop a protocol for the molecular characterization of both, protease-resistant and protease-sensitive PrPSc aggregates. We first set up experimental conditions which allowed the most advantageous separation of PrPC and PrPSc by means of differential centrifugation. The conformational solubility and stability assay (CSSA) was then developed by measuring PrPSc solubility as a function of increased exposure to GdnHCl. Brain homogenates from voles infected with human and sheep prion isolates were analysed by CSSA and showed strain-specific conformational stabilities, with mean [GdnHCl]1/2 values ranging from 1.6 M for MM2 sCJD to 2.1 for scrapie and to 2.8 M for MM1/MV1 sCJD and E200K gCJD. Interestingly, the rank order of [GdnHCl]1/2 values observed in the human and sheep isolates used as inocula closely matched those found following transmission in voles, being MM1 sCJD the most resistant (3.3 M), followed by sheep scrapie (2.2 M) and by MM2 sCJD (1.6 M). In order to test the ability of CSSA to characterise protease-sensitive PrPSc, we analysed sheep isolates of Nor98 and compared them to classical scrapie isolates. In Nor98, insoluble PrPSc aggregates were mainly protease-sensitive and showed a conformational stability much lower than in classical scrapie. Our results show that CSSA is able to reveal strain-specified PrPSc conformational stabilities of protease-resistant and protease-sensitive PrPSc and that it is a valuable tool for strain typing in natural hosts, such as humans and sheep.


Introduction
Transmissible spongiform encephalopathies (TSEs), or prion diseases, are neurodegenerative disorders that afflict humans and others mammals. TSEs may have genetic, infectious, or sporadic origins. Creutzfeldt-Jakob disease (CJD) is the most common TSE in humans and may be sporadic (sCJD), genetic (gCJD), or acquired (iatrogenic CJD). A novel human acquired prion disease, variant CJD (vCJD), appeared from 1995 onwards and was postulated to be caused by consumption of beef from cows infected with bovine spongiform encephalopathy (BSE). The most common forms of TSE in animals, scrapie in small ruminants, BSE in cattle and chronic wasting disease (CWD) in deer, are all acquired. New atypical forms of BSE in cattle, namely BSE-H [1] and BSE-L or BASE [2] and atypical scrapie in small ruminants, namely Nor98 [3], are supposed to be sporadic.
All TSEs are characterised by the accumulation of PrP Sc , a misfolded form of the cellular protein PrP C . Besides differing in conformation, the two forms of the prion protein have divergent physical properties: while PrP C is soluble in nondenaturating detergents, rapidly digested by proteases, and rich in a-helical structure, PrP Sc is insoluble in detergents, partially resistant to proteolysis and contains mostly b-sheet [4,5,6].
A major problem for the protein-only hypothesis, which postulates that prions are composed mainly or exclusively of PrP Sc [7], was to explain the existence of multiple strains of prions. Prion strains are infectious isolates that, when propagated in the same host, exhibit distinct prion disease phenotypes that persist upon serial transmission [8,9,10]. Thus, prion strains cannot be encoded by differences in PrP primary structure but anyway carry information that are independent from the host [11]. The prion hypothesis equates strains to different self-propagating conformational variants of PrP Sc [12].
Several studies demonstrated that prion strains can be distinguished based on different biochemical properties of PrP Sc , thus allowing a molecular strain typing approach to TSEs. These studies are based on the electrophoretic features of the protease resistant core of PrP Sc [13,14,15], the relative proteinase K resistance of PrP Sc [16,17,18], or the physico-chemical behaviour of PrP Sc during denaturation [19,20] and strongly support the view that distinct strains show different PrP Sc conformations.
A conformational stability assay (CSA), combining guanidine hydrochloride (GdnHCl) denaturation with limited proteolysis using proteinase K, showed that different prion strains may exhibit distinct denaturation profiles [20]. A conformation-dependent immunoassay (CDI) showed the existence of multiple strainspecified PrP Sc conformers by quantifying the immunoreactivity of native and denatured PrP Sc of eight hamster prion isolates [19]. This technique also showed that prion-infected brains contain both protease-sensitive (sPrP Sc ) and protease-resistant PrP Sc (rPrP Sc ). More recently, it was shown by CDI that sPrP Sc represents as much as 90% of total PrP Sc in the brain of patients with sporadic CJD [21].
These findings along with the recent discovery in humans [22] and animals [3] of previously undetected TSEs characterised by relatively protease sensitive PrP Sc , highlights the need of molecular strain typing methods able to recognize PrP Sc populations based on their physical properties rather than only based on protease digestion.
In this study we aimed at developing a new conformational stability assay based on the differential solubility of PrP C and PrP Sc [23,24], that we called CSSA (conformational stability and solubility assay). We have previously transmitted CJD and scrapie isolates to bank voles [25,26], which showed to be a valuable tool for the biological characterization of the most common forms of sCJD, gCJD and scrapie. We took advantage of these vole-adapted strains in order to evaluate the potential of CSSA for strain discrimination.
We first set up experimental conditions allowing the most advantageous separation of PrP C and PrP Sc , and thus performed the conformational stability assay by measuring PrP Sc solubility in homogenates treated with increasing concentrations of GdnHCl in the absence of proteinase K. Indeed, insoluble PrP was inversely correlated to GdnHCl concentration, and dose-response curves allowed estimation of the concentration of GdnHCl able to solubilise 50% of PrP Sc .
Additionally, we extended the study to natural isolates of sheep scrapie and human sCJD cases to investigate the potential of CSSA for strain discrimination in natural hosts. Our results show that this method is valuable for the biochemical typing of strains in voles and it is also a promising tool for molecular analysis of natural prion isolates.

PrP species in healthy and diseased brain
It has been previously reported that normal PrP is composed of full-length PrP (FL-PrP) as well as of 2 C-terminal fragments derived from physiological cleavage at the a and b sites: C1, which is the most represented, derives from a PrP cleavage at position 111/112, while C2 is usually barely detectable and is cleaved around the octarepeat region [27]. Interestingly, a cleavage disrupts the conserved neurotoxic and amyloidogenic region comprising residues 106-126 of PrP, preventing the generation of PrP Sc , while b cleavage occurs upstream this conserved region. Accordingly, C2 was reported to be enriched in diseased brains and insoluble in nondenaturing detergents, similarly to PrP res [28].
In order to determine if C1 and C2 were detectable in voles, we analysed healthy and diseased brain homogenates either before or after deglycosylation, with SAF84, whose epitope is present in both fragments, and 12B2 which recognises an epitope that is present only in C2 (Fig. 1A). In normal brain homogenate (NBH), FL-PrP was accompanied by substantial amounts of C1 and lower amounts of C2. In contrast, in voles infected with the Italian scrapie isolate SS7 (SBH) [26] C2 was strongly enhanced while C1 was not easily detectable ( Fig. 1B and C). Furthermore, the PrP res fragment generated after PK digestion in SBH was similar to C2 ( Fig. 1B and C). Finally, in SBH PrP dimers were present which were completely absent in NBH.

Separation of PrP C and PrP Sc
In order to develop a conformational stability assay based on the differential solubility of PrP C and PrP Sc we first set up the experimental conditions which enabled the most advantageous separation of PrP C from PrP Sc . This was obtained through a conventional procedure based on centrifugation in the presence of detergents.
By varying concentrations of different detergents, times of centrifugation and centrifugal force (see Material and Methods), we found that treatment with 2% sarcosyl followed by centrifugation at 20.000 g for 1 h enabled an optimal separation of PrP C from PrP Sc .
In these conditions, .95% of the total PrP C in NBH was soluble and was found in the supernatant fraction ( Fig. 2A, left panel) while in SBH most of PrP was sedimented ( Fig. 2A, right panel). Unfractionated SBH contained high amounts of PrP res which exclusively sedimented to the pellet fraction ( Fig.2A). Insoluble PrP from SBH was indeed mostly PK-resistant (,90% of insoluble PrP), while soluble PrP was PK-sensitive. All fractions from NBH were devoid of PrP res ( Fig. 2A).
Furthermore, soluble and insoluble PrP fractions from SBH displayed slightly different banding patterns which suggested that C2 was mainly insoluble, while soluble PrP contained FL-PrP and the C1 fragment ( Fig. 2A). Moreover, PrP Sc dimers were clearly segregated in the insoluble fraction ( Fig. 2A). The differential solubility of C1 and C2 in SBH was confirmed by the analysis of deglycosylated PrP species (Fig. 2B), which showed that C1 was completely soluble while C2 was mainly sedimented in the pellet fraction (compare ''S'' and ''P'' lanes in Fig. 2B, left panel). With the aim to mimic a situation comparable to that expected in animals with pre-clinical disease, we also studied the differential solubility of C1 and C2 after mixing equal amounts of NBH and SBH (Fig. 2B). Indeed, under these conditions the PrP C content is increased compared to SBH, as can be seen by the higher proportion of C1 in mixed NBH/SBH compared to SBH alone (compare ''Tot'' lanes in the two panels of Fig. 2B). Also in this condition, C1 was completely soluble and the pellet fraction was enriched in C2 (compare ''S'' and ''P'' lanes in Fig. 2B, right panel).
Finally, we investigated the efficacy of our solubility assay for separating PrP C and PrP Sc in voles infected with other, non scrapie-derived prion strains (Fig. 2C). For these experiments we used voles infected with MM1 and MM2 sCJD [25]. In both strains a substantial amount of PrP was found in the insoluble fraction after detergent treatment and centrifugation. As already observed in SBH, the banding patterns of insoluble and soluble PrP were distinct, suggesting a specific precipitation of PrP species associated to disease, namely C2 and PrP dimers (compare ''S'' and ''P'' lanes in Fig. 2C). Furthermore, after PK digestion, PrP res was strongly enriched in the pellets and virtually absent in the supernatants.
Collectively, these findings strongly suggest that under the experimental conditions described above we were able to specifically precipitate PrP Sc in brain homogenates from voles infected with different prion strains.

Conformational stability and solubility assay (CSSA)
The near complete separation of PrP C from PrP Sc allowed us to develop a procedure for biochemical strain typing based on the conformational stability of PrP Sc after exposure to GdnHCl. The conformational solubility assay was set up by measuring PrP Sc solubility in positive brain homogenates treated for 1 hour with increasing concentrations of GdnHCl.
As expected, in SBH PrP Sc was solubilized by increasing GdnHCl concentrations ( Fig. 3A and B). Indeed, with concentrations of GdnHCl equal or greater than 1.5 M, PrP Sc was partially solubilised and was progressively found in the supernatant instead of the pellet (compare Fig. 3A and 3B). With 3.5 M GdnHCl, virtually all PrP from SBH was found in the soluble fraction ( Fig. 3A and 3B). The solubilization of PrP Sc was guanidine-dependent and dose-response The location of SAF84, 12B2 and SAF32 epitopes used for the FL/C1/C2 discrimination are shown. B and C: Normal brain homogenates (NBH) and scrapie brain homogenates (SBH) of voles were analyzed by western blot using SAF84 (B) and 12B2 (C). The samples, were analyzed either before or after deglycosylation (N-Gly F) as indicated. The brackets on the left indicate the position of glycosylated and unglycosylated bands of FL, C2 and C1: from 35 kDa to 27 kDa for FL, from 26 kDa to 18 kDa for C2 and from 24 kDa to 16 kDa for C1. These PrP species are reduced to single unglycosylated PrP bands after deglycosylation, which are indicated by dashes on the right of the blots. In SBH, PrPSc dimers (Dim) are also indicated. In NBH both C1 and C2 were present, although C2 was poorly represented; in contrast in SBH C2 fragment was the most abundant PrP species while C1 was barely detectable. Tissue equivalents (TE) loaded per lane were 0,15 mg and 0.5 mg for samples respectively before and after PK digestion. curves enabled estimation of the concentration of GdnHCl able to solubilise 50% of PrP Sc ([GdnHCl] 1/2 ). The [GdnHCl] 1/2 value was similar when estimated either in the pellet or in the supernatant fractions (Fig. 3C). In contrast, PrP from NBH remained soluble within the range of GdnHCl concentrations tested (Fig. 3D). Denaturation of PrP Sc was complete after incubation with GdnHCl for 1 hour, as very similar denaturation curves were obtained when the treatment was extended up to 4 hours (Fig. S1). When the denaturation curves where measured in replica blots with mAbs recognising different PrP species, namely SAF32 and SAF84 (see scheme in Fig. 1A), we obtained similar [GdnHCl] 1/2 values, which suggest that C2 and FL PrP Sc share the same conformational stability (Fig. S2).
We then studied the relationship between insolubility and PKresistance of PrP Sc during denaturation. In order to investigate if denaturation equally affects insolubility and PK-resistance, we compared denaturation curves derived from the same SBH, either obtained by insoluble PrP Sc (CSSA) or by PK-resistant PrP Sc (CSA). As shown in Fig. 4, the curves of insoluble PrP Sc (Fig. 4A) and PrP res (Fig. 4B) didn't show differences (Fig. 4C), suggesting that insolubility and PK-resistance were equally susceptible to denaturation by GdnHCl. This finding was further confirmed in experiments aimed at investigating whether, after denaturation, solubilized PrP Sc could partially preserve its resistance to proteinase K. Indeed, after denaturation with 3 M GdnHCl, soluble PrP was fully susceptible to protease digestion (Fig. 4D).

Conformational stability of scrapie, sCJD and gCJD PrP Sc from bank voles
Under the experimental conditions described above, we investigated the potential of CSSA for differentiating prion strains.
As reported in previous studies [25,26] sCJD, gCJD and some scrapie isolates present distinct and specific patterns of transmission in voles, based on survival times, lesion profiles, PrP Sc deposition and PrP res biochemical properties. We have shown that voles infected with MM1/MV1 sCJD and E200K gCJD isolates were characterised by a PrP res fragment of ,19 kDa, MM2 sCJD showed a PrP res fragment of ,17 kDa, while natural scrapie isolates and murine scrapie ME7 were characterised by a PrP res fragment of ,18 kDa, intermediate between types 1 and 2 CJD.

Conformational stability of human and sheep isolates
Since we were interested in exploiting our conformational solubility assay (CSSA) for strain discrimination in natural prion diseases, we analysed 3 of the isolates used as inocula for transmission to voles, namely the human MM1 and MM2 sCJD isolates and the sheep scrapie SS7.
At first we tested the efficiency of separation of PrP Sc from PrP C in human and sheep brain homogenates, under the same experimental conditions used for vole brain homogenates. In all samples, PrP Sc was enriched in the pellet, as showed by the segregation of PrP res in the insoluble fraction and by the distinct banding patterns shown by soluble and insoluble PrP (Fig. S3). In negative sheep brain homogenates the fraction of PrP segregating in the pellet was somewhat higher than in vole brains (.10% in some experiment). However, this sedimented PrP C was insensitive to GdnHCl and thus did not interfere with our assay (Fig. S4).

Conformational stability of classical and Nor98 scrapie isolates from sheep
In order to test the ability of CSSA for studying the conformational stability of protease-sensitive PrP Sc , we took advantage of the recently described atypical scrapie strain, Nor98 [3], which has been shown to induce the accumulation of relatively protease-sensitive PrP Sc [29]. As previously observed for all other strains investigated, insoluble and soluble Nor98 PrP Sc showed different segregation of PrP species (Fig. 6). Interestingly the 12 kDa PrP res fragment, which is a characteristic feature of Nor98 PrP res [3,30], was observed before PK-digestion and segregated with insoluble PrP, in a manner similar to what was observed for C2 in all other strains. However, in Nor98 only a minimal fraction of insoluble PrP was protease resistant (Fig. 6), while PrP res represented .90% of insoluble PrP in classical scrapie (Fig. 6). These findings confirm that in Nor98 samples disease- associated PrP is mostly PK-sensitive, although accompanied by low amounts of genuine PK-resistant PrP Sc .
We then studied the conformational stability of Nor98 (n = 5) and classical (n = 4) Italian field scrapie cases of different PrP genotypes (Table 2). Classical scrapie samples gave denaturation curves very similar to that of SS7, with [GdnHCl] 1/2 values ranging from 1.96 to 2.31 ( Fig. 7 and Table 2). Nor98 samples gave remarkably similar denaturation profiles, independently of PrP genotype, and showed high sensitivity to GdnHCl denaturation (Fig. 7)

Discussion
The main pathogenetic event in TSEs involves the transconformation of PrP C to PrP Sc , which leads to the accumulation of detergent-insoluble and partially protease-resistant aggregates. Thus, resistance to PK digestion and insolubility are the hallmarks of PrP Sc . However, previous studies on the molecular character-ization of PrP Sc were mainly focused on the PK-resistant core of PrP Sc , due to the difficulties in differentiating PrP C from PrP Sc in infected brain homogenates. Notwithstanding, it is becoming increasingly clear that insoluble but protease-sensitive isoforms of PrP are involved in different animal and human prion diseases [19,22,31,32,33,34,35]. These PrP isoforms have been detected by several methods, including conformation-dependent immunoassay [19,21,33], immunological capture [36,37], differential centrifugation [31,32], thermolysin digestion [35] and cold PK digestion [34].
Among the above-mentioned techniques able to detect sPrP Sc , the CDI was also reported to distinguish eight hamster prion strains based on PrP Sc conformation, by plotting the ratio of antibody binding to denatured/native PrP as a function of the concentration of PrP Sc [19]. However, CDI depends on the availability of monoclonal antibodies able to recognise buried epitopes in PrP Sc , such as the 3F4. We developed a protocol for the molecular characterization of PrP Sc aggregates which does not rely on their protease resistance but is based on a conventional procedure of differential centrifugation in the presence of detergents and on the solubilization of PrP Sc aggregates upon denaturation with increasing concentrations of GdnHCl. Our findings show that CSSA is reliable and straightforward, and that it is able to discriminate PrP Sc conformers associated with different  TSE strains. Compared to CDI, we believe that CSSA offer some interesting advantages: i) it does not depend on the antibody used and can be thus exploited to compare PrP Sc conformational stability in different species, ii) the proof that CDI is able to discriminate prion strains in natural hosts is still lacking, while here we show the potential of CSSA to be used in human and sheep natural isolates.
In previous studies [20,38,39] a conformational stability assay (CSA) able to discriminate PrP Sc conformers was set up by measuring the extent of loss in protease resistance as a function of increased exposure to GdnHCl. This method proved to be very helpful for molecular strain typing in different species [40,41,42] and was also recently exploited for investigating some basic mechanisms of prion replication [39]. The protocol that we developed is conceptually similar to CSA, as it derives information on the conformational stability of PrP Sc aggregates. Indeed, we showed that CSA and CSSA gave very similar denaturation profiles in vole-adapted scrapie brain homogenates (see Fig. 4). However, we believe that CSSA represents a step forward in prion molecular typing and offers several advantages compared with previously used protocols. With CSA, in fact, susceptibility to PK is exploited to distinguish denaturated from native PrP Sc . However, different prion strains may have distinct susceptibilities to PK, while CSA uses the same PK concentration to derive the level of PrP Sc denaturation in different strains. Furthermore, when performing CSA it is necessary to dilute the GdnHCl to allow the activity of proteolytic enzyme. However, it is known that PrP Sc unfolding can be a partially reversible phenomenon and, as reported, the dilution of the denaturant could restore the original protease-resistance of PrP Sc [43]. With CSSA this problem was circumvented by avoiding any change in denaturant concentration during the assay. Indeed, the denaturation step is followed by the centrifugal separation of soluble and insoluble fractions, which is performed under conditions identical to the denaturation step.
Most importantly, CSSA allowed the characterization of protease-sensitive PrP Sc , and thus the direct comparison of the conformational stabilities of TSE strains associated with proteasesensitive and protease-resistant PrP Sc . We investigated the potential of CSSA for strain typing of ovine field isolates of Nor98, which are characterised by high amounts of insoluble sPrP Sc [29, present paper]. Our findings seem very encouraging, since Nor98 samples displayed a distinct denaturation profile, although the isolates derived from sheep of various PrP genotypes and presumably at different stage of the disease. Furthermore, Nor98 samples were easily discriminated from classical scrapie, which represents a different strain. Thus CSSA enabled characterization of the conformational stability of a protease sensitive strain, for which methods based on proteolysis would have characterised only a minimal amount of PrP Sc present in the brain homogenate. This is promising in view of recent studies [21,35] showing that in sCJD and vCJD isolates, as much as 90% of PrP Sc in the brains was estimated to be sPrP Sc . Furthermore the recent discovery of protease sensitive prionopathy (PSPr), a previously unrecognised human prion disease [22], might suggest that prion diseases characterised by protease-sensitive PrP isoforms are more frequent than previously thought. Brain homogenates from PSPr cases were reported to contain mainly a protease sensitive form of insoluble PrP, accompanied by low amounts of typical protease-resistant PrP [22], similarly to our findings in Nor98 samples. It would be interesting to investigate the potential of CSSA for characterising these human protease-sensitive diseaserelated isoforms of PrP.
Another interesting feature of CSSA is that, by avoiding PK treatment, it allowed the characterization of FL-PrP Sc , including the N-terminus which is cleaved upon PK digestion. Besides FL-PrP, in vivo trimmed PrP Sc fragments were selectively precipitated and thus CSSA also conveyed information on their conformational stability. These fragments comprised C2 C-terminal fragments in scrapie and sCJD isolates, as well as the 12 kDa internal fragment  characteristic of Nor98, and may be easily distinguished from FL-PrP, also by means of differential antibody binding (data not shown; see for example Fig. S2). This is interesting because it enables a comparison of the conformational properties of PrP Sc aggregates made-up of FL-PrP or in vivo trimmed PrP Sc fragments.
In situ epitope-mapping of PrP Sc , indeed, showed that FL and C2 PrP Sc aggregates have different cellular localizations [44]. Furthermore, by analysing thermolysin-resistant PrP Sc in sheep scrapie and BSE isolates, Owen and colleagues have recently shown the potential of C2 fragments for strain typing [45].
The main potential drawback of CSSA might be the incomplete separation or PrP C and PrP Sc . Herein, we developed a protocol able to minimise this problem, notwithstanding 1-4% of PrP C in brain homogenates from healthy voles was found in the pellet under our working conditions. There are however several lines of evidence suggesting that this potential drawback did not interfere with CSSA results. We have shown that brain homogenates from clinically affected voles contain low levels of soluble PrP (bona fide PrP C ) (see Fig. 2). Based on the results obtained with NBH, less than 5% of PrP C is expected to be found in the pellet; on the other hand, in diseased brains bona fide PrP C was 5-30% of total PrP. From these considerations, it can be argued that the PrP Sc /PrP C ratio in the pellet should be higher than 2 orders of magnitude, which represents the working range of CSSA. Furthermore, we obtained clear-cut differential solubility of PrP C and PrP Sc even after increasing the PrP C /PrP Sc ratio in the homogenate, by mixing SBH with NBH (Fig. 2). Most importantly, we found that the sedimented PrP C fraction was insensitive to denaturation (Fig.  S4) and thus it does not interfere with CSSA.
We investigated the potential of CSSA for strain discrimination by analysing vole-adapted strains deriving from human sCJD, gCJD and sheep scrapie which we have previously shown to give distinct strains in voles [25,26]. In previous studies we have shown that voles infected with MM1/MV1 sCJD and E200K gCJD isolates showed identical transmission patterns and were characterised by an unglycosylated PrP res fragment of ,19 kDa, while MM2 sCJD showed an unglycosylated PrP res fragment of ,17 kDa. In contrast, all Italian scrapie isolates studied so far [26,46], as well as the murine scrapie strain ME7 [25], were characterised by the accumulation of an unglycosylated PrP res with a molecular weight (,18 kDa) intermediate between CJD types 1 and 2, upon transmission to voles. These different PrP res types derive from different PK-cleavage sites of PrP Sc , which in turn are believed to reflect distinct conformations of PrP Sc aggregates. With CSSA we showed that these three PrP res types are actually characterised by PrP Sc aggregates with distinct susceptibilities to denaturation by GdnHCl. Indeed, PrP Sc from voles infected with MM1 sCJD, MV1 sCJD and E200K gCJD, characterised by 19 kDa PrP res , showed the highest resistance to denaturation, while sCJD MM2, characterised by 17 kDa PrP res , was the most susceptible and scrapie SS7, that was characterised by 18 kDa PrP res , displayed intermediate susceptibility. Interestingly, the within group variability of [GdnHCl] 1/2 values was very low (see table 1) and allowed statistical comparisons among the different groups, which strengthened the view that [GdnHCl] 1/2 values reflect strain-specific rather than individual PrP Sc properties.
It has been recently suggested that the conformational stability of PrP Sc is directly proportional to the length of the incubation time in mice [39]. On this point, it is worth noting that our results in voles, although based on only 5 isolates, seem to contradict this conclusion. Indeed, the lowest conformational stability was associated with prions with the longest survival times, i.e. MM2 sCJD with a survival time of ,330 days post-inoculation (dpi), while scrapie SS7 (survival time of ,90 dpi) and MM1/MV1 sCJD (survival time of ,130 dpi) showed higher conformational stabilities and shorter survival times compared to MM2 sCJD. Further studies with an extended panel of isolates are needed to investigate if the direct relationship between conformational stability and incubation time observed in mice holds true also for vole prions.
We also explored the potential of CSSA for strain typing of natural prion diseases. To this aim we analysed three of the human and sheep isolates which were used for bioassay in vole. Interestingly, we found that the two isolates from MM1 sCJD and MM2 sCJD patients could be easily discriminated by their conformational stabilities, with MM1 sCJD displaying lower susceptibility to denaturation compared to MM2 sCJD ([GdnHCl] 1/2 values of 3,31 M for MM1 sCJD and 1,63 M for MM2 sCJD). These findings compare well with previous studies by CSA, which showed that PrP Sc associated with MM1 sCJD was , 2-fold more stable than that of MM2 sCJD, with [GdnHCl] 1/2 values of 2,76 M and 1,63 M for MM1 sCJD and MM2 sCJD [40]. Of course, this needs confirmation in a larger set of isolates.
Furthermore, this approach allowed us to compare the conformational stability of prions in their natural host and after transmission to voles. It has been previously shown that prion strains can either maintain their biological properties or mutate upon propagation in a new host species [10]. More recently it was reported that a change in conformation was accompanied by the emergence of a new prion strain during interspecies transmission, while the conformational stability of PrP Sc was preserved when a strain ''bred true'' in the new host [38]. Our findings show that the rank order of conformational stability of sCJD and scrapie isolates was generally preserved upon adaptation in voles (see Fig. 5). In particular, the conformational stability of MM2 sCJD was identical before and after transmission in voles. We have also previously reported that MM2 sCJD, and to a lower extent MM1 sCJD, encountered a very low transmission barrier to voles when compared to several other prion strains [25]. These observations may suggest that MM2 and possibly MM1 sCJD faithfully propagated their strain properties upon transmission in voles. The ''conformational selection'' model, postulates that host PrP C primary structure influences which of the portfolio of possible PrP Sc types are thermodynamically preferred during propagation [47]. In this model, the transmission barrier is determined by the degree of overlap between the subset of PrP Sc types allowed or preferred by PrP C in the host and donor species. It can be thus speculated that the vole PrP sequence is prone to adopt some human PrP Sc conformations and we have previously shown that this property may reside in the presence of some peculiar amino acid at relevant positions of vole PrP [46,48]. Finally, we exploited CSSA for discriminating sheep field prion isolates, showing that this method is valuable for strain discrimination in natural hosts. Indeed, the conformational stability of prions was strongly associated with the strain, either classical scrapie or Nor98, and did not depend on individual properties such as age, clinical stage or PrP genotype (see table 2). Although preliminary, these findings suggest that CSSA might be exploited as complementary approach for increasing the discriminatory power of strain typing in small ruminant TSEs. Studies with natural and experimental TSEs, including sheep BSE and CH1641-like isolates, are underway.

Materials and Methods
Terminology Throughout the manuscript we used ''prion'' to refer to the infectious agent of TSEs, ''PrP Sc '' to refer to the abnormal diseaseassociated PrP isoform, ''rPrP Sc '' to refer to the protease-resistant fraction of PrP Sc , ''sPrP Sc '' to refer to the protease-sensitive fraction of PrP Sc and ''PrP res '' to refer to protease-resistant PrP Sc fragments deriving from rPrP Sc trimming by proteinase K.

Natural TSE isolates
Human brain tissues were from cerebral cortices of two sCJD cases (MM1 and MM2 types) previously characterised by vole bioassay [25].
The medulla oblongata of SS7 scrapie was obtained from an ARQ/ARQ Sarda sheep with clinical scrapie reported in Italy in 1997 and previously characterised by vole bioassay [26]. All others sheep brain tissues were from Italian field cases detected by active and passive surveillance and characterised by discriminatory western blotting for molecular strain typing and by complete sequencing of the PRNP open reading frame, according to published protocols [49,50]. The details of these TSE cases are reported in Table 2.

Vole samples
Brain tissues were obtained from voles infected with vole-adapted human and sheep TSEs as previously reported [25,26]. The passage number and mean survival times 6 standard deviation of the vole groups from which the samples used in the present study derived were as follows: sCJD MM1, second passage, 12968 dpi; sCJD MV1, second passage, 128615 dpi; sCJD MM2, second passage, 339627 dpi; sCJD MM2, third passage, 323641 dpi; gCJD E200K, second passage, 143612 dpi; scrapie SS7, third passage, 9365 dpi.

Separation of PrP Sc and PrP C by differential centrifugation
Brain homogenates (20% w/v) were prepared in 100 mM Tris-HCl with Complete protease inhibitor cocktail [Roche] at pH 7.4. The homogenates were either used directly or stored at 220uC.  Table 2. doi:10.1371/journal.pone.0012723.g007 The experimental conditions for PrP C /PrP Sc separation were set up in vole brain homogenates by studying the effect of different detergents, centrifugal force and time of centrifugation.
Brain homogenates (3% to 12% w/v) were added with equal volumes of different buffers (100 mM Tris-HCl at pH 7.4 containing Sarcosyl 4% or 2%; 100 mM Tris-HCl at pH 7.4 containing NaDoc 1%, NP40 1%; 100 mM Tris-HCl at pH 7.4 containing Triton X-100 2%) and incubated for 1 hour at 37uC with gentle shaking. Then samples were centrifuged at 10000 to 20000 g for 1 h or 2 h. The obtained pellets were re-suspended with 100 mM Tris-HCl (pH 7.4) containing the relevant detergent. The experimental conditions then used throughout the paper in vole, sheep or human brain tissues included solubilisation in 100 mM Tris-HCl at pH 7.4 containing sarcosyl 2% and centrifugation at 20000 g for 1 h. For each of the different experimental conditions tested, equivalent aliquots of brain homogenate before centrifugation, along with supernatant and pellet fractions, were analysed by western blot either with or without PK digestion.

Conformational stability and solubility assay
Aliquots of brain homogenates (3% to 6% w/v) were added with an equal volume of 100 mM Tris-HCl (pH 7.4) containing sarcosyl 4% and incubated for 1 h at 37uC with gentle shaking. Aliquots of 100 ml were treated with 100 ml of guanidine hydrochloride (GdnHCl) solutions with a final concentration ranging from 0 to 4.0 M. GdnHCl stock solutions were prepared from an 8 M solution (Pierce) diluted in water. After treatment with GdnHCl for 1 h at 37uC with gentle shaking, samples were centrifuged at 20000 g for 1 h at 22uC. Pellets were re-suspended in 90 ml NuPage LDS Sample Buffer (Invitrogen) and 10 ml NuPage Sample Reducing Agent (Invitrogen). Aliquots of supernatants were precipitated with 4-fold volume excess of prechilled methanol 30 min at 220uC, centrifuged at 15000 g for 30 min at 4uC and then were re-suspended in 90 ml NuPage LDS Sample Buffer (Invitrogen) and 10 ml NuPage Sample Reducing Agent (Invitrogen). Supernatant and pellet fractions were analysed by Western blotting.
Individual denaturation curves were analyzed and best-fitted by plotting the fraction of PrP Sc remaining in the pellet as a function of GdnHCl concentration, and using a four parameter logistic equation (GraphPad Prism). In order to fit denaturation curves for each prion strain, the mean fraction of PrP Sc remaining in the pellet 6 SD were plotted. Statistical comparison of [GdnHCl] 1/2 values were made by comparing the best-fit value for each data set with GraphPad Prism. This was performed by either fitting each data set independently or doing a global fit with a shared [GdnHCl] 1/2 value, and then the results were compared with an F test. The simpler model was selected unless the extra sum-ofsquares F test had a P value,0.05.

Digestion with proteinase K after denaturation with GdnHCl
Aliquots of the same brain homogenate were treated in parallel according to CSSA and conformational stability assay (CSA) protocols. The CSA was performed as described [20], with minor modifications. Aliquots of brain homogenates (6% w/v) were added with an equal volume of 100 mM Tris-HCl (pH 7.4) containing sarcosyl 4% and incubated for 1 h at 37uC with gentle shaking. Aliquots of 50 ml were added with 50 ml of GdnHCl to give a final concentration ranging from 0 to 4.0 M. After 1 h of incubation at 37uC all samples were diluted to a final concentration of 0.4 M GdnHCl. Proteinase K (50 mg/ml) was added and the samples were incubated for 1 h at 37uC with gentle shaking. The reaction was stopped with 3 mM PMSF (Sigma). Aliquots of samples were added with an equal volume of isopropanol/butanol (1:1 v/v) and centrifuged at 20000 g for 5 min. Pellets were re-suspended in NuPage LDS Sample Buffer (Invitrogen) and were analysed by Western Blotting.

Western blot analysis
Electrophoresis and Western blotting were performed as previously described [25]. Samples were denatured by adding NuPage LDS Sample Buffer (Invitrogen, Carlsbad, California, United States) and NuPage Sample Reducing Agent (Invitrogen), and heating for 10 min at 90uC. After centrifugation at 10000 g for 5 min each sample was loaded onto 12% bis-Tris polyacrylamide gels (Invitrogen). After electrophoresis and Western blotting on PVDF membranes (Immobilon-P; Millipore, Bedford, MA, USA), the blots were processed by SNAP i.d. TM Protein Detection System (Millipore) as described by the manufacturer instructions.
The membranes were developed with an enhanced chemiluminescence method (SuperSignal Femto, Pierce). Chemiluminescence signal was detected with the VersaDoc imaging system (Bio-Rad) and was quantified by QuantityOne software (Bio-Rad).
Deglycosylation was performed by adding 18 ml of 0.2 M sodium phosphate buffer (pH 7.4) containing 0.8% Nonidet P40 (Roche) and 2 ml (80 U/ml) di N-Glycosidase F (Roche) to 5 ml of denaturated samples and by incubating overnight at 37uC with gentle shaking. Samples were then analysed by Western blotting as described above.