Sensitive protein misfolding cyclic amplification of sporadic Creutzfeldt–Jakob disease prions is strongly seed and substrate dependent

Unlike variant Creutzfeldt–Jakob disease prions, sporadic Creutzfeldt–Jakob disease prions have been shown to be difficult to amplify in vitro by protein misfolding cyclic amplification (PMCA). We assessed PMCA of pathological prion protein (PrPTSE) from 14 human sCJD brain samples in 3 substrates: 2 from transgenic mice expressing human prion protein (PrP) with either methionine (M) or valine (V) at position 129, and 1 from bank voles. Brain extracts representing the 5 major clinicopathological sCJD subtypes (MM1/MV1, MM2, MV2, VV1, and VV2) all triggered seeded PrPTSE amplification during serial PMCA with strong seed- and substrate-dependence. Remarkably, bank vole PrP substrate allowed the propagation of all sCJD subtypes with preservation of the initial molecular PrPTSE type. In contrast, PMCA in human PrP substrates was accompanied by a PrPTSE molecular shift during heterologous (M/V129) PMCA reactions, with increased permissiveness of V129 PrP substrate to in vitro sCJD prion amplification compared to M129 PrP substrate. Combining PMCA amplification sensitivities with PrPTSE electrophoretic profiles obtained in the different substrates confirmed the classification of 4 distinct major sCJD prion strains (M1, M2, V1, and V2). Finally, the level of sensitivity required to detect VV2 sCJD prions in cerebrospinal fluid was achieved.


Results
Efficient amplification of sCJD is substrate dependent. Infected brain homogenates (IBH) from 14 patients with confirmed sCJD of MM1 (#1-3), MM2 (#4,5), MV1 (#6,7), MV2 (#8, 9), VV1 (#10,11), or VV2 (#12-14) subtype were used as PrP TSE seeds for the PMCA reaction and compared against the vCJD WHO reference case from the NIBSC (#15) ( Table 1). In Fig. 1, western blot analysis of the non-amplified materials (10 −3 dilution w/v) from each IBH is represented as an illustration of the different molecular profiles type 1/type 2 PrP res of each subtype as well as their initial PrP res level. Serial PMCA amplification of each IBH were performed using three different substrates referred to as TgMet, TgVal and BV hereafter in presence of heparin as a cofactor. Dilutions from 10 −4 to 10 −9 were submitted to serial PMCA and PrP res generated after 4 rounds was analyzed by western-blot. The number of 4 rounds was fixed based on vCJD previous data for which dilution limit of brain homogenate was obtained after 3 PMCA rounds therefore a plateau in the amplification could be expected after 4 rounds. PMCA overall results are summarized in Table 1 and representative PrP res signals are illustrated in Fig. 2. Seeding efficiencies varied in a subtype and substrate dependent manner. No PrP res signal was observed after PMCA in all the unseeded substrates (N; n = 45). For each discordant results, experiments were repeated three times.
After 4 rounds of PMCA, vCJD amplification was achieved in all 3 substrates with detection of the 10 −9 dilution (last dilution tested). Although sCJD MM1 prions were unable to convert TgMet, 2 out of 3 cases could seed the PMCA reaction in TgVal with moderate efficiency, down to a 10 −6 dilution. Amplification using BV NBH was efficient and consistent down to the 10 −6 /10 −7 dilutions. Optimal amplification of sCJD MM2 was achieved in BV (10 −9 dilution) while no or moderate amplification was obtained in TgMet and TgVal respectively. Prions from sCJD MV1 and MV2 patients were preferentially amplified in TgVal, allowing the detection of the 10 −6 /10 −7 and 10 −9 dilutions, respectively. As with sCJD MM1 and MM2 prions, TgMet did not amplify MV1 prions. Moderate amplification of MV1 was obtained in BV, allowing the detection of the 10 −5 /10 −6 dilutions. Regarding sCJD MV2 prions, PMCA in the TgMet substrate allowed the detection of the 10 −6 /10 −7 dilutions and discordant results were obtained in BV. No amplification for one case (#8) and amplification allowing detection down to the 10 −8 dilution with the second case (#9) were obtained with this substrate. These results were confirmed in two additional series of PMCA experiments. The sCJD VV1 prions were the most difficult to amplify in our conditions: no amplification in TgMet, very limited amplification (10 −4 ) in TgVal for only one case (#11) and unexpectedly higher amplification in BV (10 −7 dilution for case #10) but with again discordant results between VV1 cases. In Fig. 1, the two VV1 cases showed slight differences in the PrP res size which could support the marked differences in PMCA amplification. Signals shown in Fig. 2 in TgVal and BV are from the same sCJD VV1 case (#10).
Unlike the other sCJD subtypes, VV2 prions were easier to amplify and like vCJD, could be amplified in all 3 substrates, except case #14 in BV but PrP res level in this brain tissue was lower compared to the others (Fig. 1). www.nature.com/scientificreports/ The best substrate for the sCJD VV2 amplification was TgVal allowing the detection of the 10 −9 dilution and the least effective was BV allowing detection only up to 10 −4 /10 −5 dilutions. The efficient amplification of sCJD MM1 prions in BV prompted us to challenge overexpressing bank vole PrP using TgBV substrate. Unfortunately, the seeding potential of sCJD MM1 was not increased in TgBV compared to the wild type BV, (detection of the 10 −6 dilution) (see Supplementary Fig. S1).
Amplified PrP res could shift according to the PMCA substrate. Independently of amplification efficiency of IBH/substrate couples, we exploited the 9A2 antibody, which recognizes both human and bank vole PrP, to examine by western blot (WB) the specific electrophoretic profiles of the different PMCA amplicons according to the substrate used. When type 2 subtypes IBH (MV2 and VV2) seeded TgMet substrate or when type 1 subtypes IBH (MM1, MV1 and VV1) seeded TgVal substrate , a molecular type1/type2 shift of the unglycosylated isoform of PrP res was observed (Fig. 3). Amplicons generated from BV substrate conserved their electrophoretic typing profiles compared to the initial IBH used to seed the reaction (Fig. 3a).
To assess whether the observed shift was due to a modification of the proteinase K cleavage site, we analyzed TgMet or TgVal shifting PMCA amplicons using 3F4 Ab (epitope 109-112 of human PrP) and 12B2 Ab (epitope Table 1. Summary of PMCA results after 4 rounds according to seed/substrate combination (dilutions tested from 10 −4 to 10 −9 ). Bold indicates that the last 10-9 dilution tested was positive.  Figure 1. Western blot analysis of non-amplified brain samples. A panel of 14 brain samples (#1-#14) from patients with sCJD was obtained from the French CJD National Surveillance Network. #15 corresponds to vCJD reference brain sample provided by the NIBSC. The PrP TSE signal was assessed by means of western blot analysis using 3F4 antibody after proteinase K digestion. For each sample, the equivalent of 20 µL of 0.1% (w/v) brain homogenate was loaded onto the gel. www.nature.com/scientificreports/ 89-93 of human PrP, specific for type 1 PrP res ) (Fig. 3b). After 4 rounds of PMCA amplification using TgMet substrate, type 2 subtypes shifted to a type 1 profile, which was confirmed by the presence of a WB signal using 12B2 Ab on generated amplicons. In contrast, type 1 subtypes shifted to a type 2 profile after PMCA in TgVal, as confirmed by the absence of WB signal using 12B2 Ab.

IBH subtype
To explore the dynamic property of shifting amplicons, we performed a back-seeding PMCA using the shifted PMCA amplicons as seeds. We first used MV2/VV2-TgMet amplicons as seeds back into TgVal substrate (Fig. 3c). Serial dilutions of the amplicons from 10 −2 to 10 −5 were submitted to 2 PMCA rounds. Type 2 profiles were recovered in TgVal substrate with efficient seeding potential (10 −5 dilution detected after 2 rounds). Thus, while amplification of the VV2 subtype in TgMet substrate led to a type 1 amplicon, its seeding activity back in TgVal was completely different to the pure Type 1 VV1 subtype which failed to seed or poorly seeded the TgVal substrate (#10 and #11 in Table 1). These findings thus highlight a reversible molecular PrP res signature of MV2/ VV2 sCJD prions upon PMCA conversion in human PrP M129 substrate without apparent alteration of their seeding properties. We next complemented this data by performing a reverse experiment, in which MV1/VV1-TgVal amplicons (shifted to type 2) were used to seed the TgMet substrate (Fig. 3c). The MV1-TgVal amplicons harbored a low seeding potential in the TgMet substrate (detection of only the 10 −2 dilution after 2 rounds), accompanied with a back shift into type-1 PrP res . This effect was not observed with direct seeding of PMCA reactions with the sCJD MV1 subtypes in TgMet (Table 1), highlighting that PMCA in TgVal allowed the MV1 prions to acquire a moderate seeding potential back into the TgMet substrate; however this seeding effect was clearly lower from that of the pure MV2 sCJD subtype in the TgMet substrate (#8 and #9 in Table 1). By contrast, the VV1-TgVal could not seed back the TgMet substrate (Fig. 3c). Taken together, these findings are in agreement with the notion that the observed PrP res shifts during the heterologous PMCA in the human PrP substrates are PrP res electrophoretic profiles of shifting PMCA amplicons (+) and their corresponding initial seeds (−) were compared using 3F4 Ab (recognizing both type1 and type2 PrP res ) and 12B2 Ab (specific for type1 PrP res ). NBH refers to normal brain homogenate from TgMet mice without any proteinase K digestion. For (a) and (b), loaded materials are from different seeded dilutions and have been adjusted in order to better visualize unglycosylated bands: www.nature.com/scientificreports/ not associated with major changes in the seeding properties of sCJD prions, although in the case of MV1 prions some slight modifications could be observed during back seeding experiments.
Electrophoretic signatures of CJD PMCA amplicons concord with the existence of four strains of sCJD. Seeding behaviors of IBH/substrate couples were consistent with previously described observations from in vivo models (Table 1). In addition to presenting a similar amplification potential (10 −6 -10 −7 dilutions detected after 4 rounds of PMCA), VV2 and MV2 subtypes harbored a type 1 electrophoretic profile in TgMet substrate (shift to type1), while maintaining a type 2 profile when seeded in TgVal substrate with a maximal amplification efficiency (i.e. detection of the 10 −9 dilution). These two subtypes could be considered as a same strain of prion, i.e. the V2 strain 42 . Regarding MM1 (#2 and #3) and MV1 subtypes, similar behaviors were shown and are reminiscent of the M1 prion strain. Indeed, with a positive amplification down to the 10 −6 dilution, the PMCA amplicons obtained in TgVal substrate adopted a type 2 PrP res profile ( Fig. 3a; Table 1) and no amplification was observed in TgMet substrate. Both MM1 and MV1 subtypes could be amplified in BV substrate with, however, a better seeding activity of MM1 (detection of the 10 −6 /10 −7 dilutions) compared to MV1 (detection of the 10 −5 /10 −6 dilutions).
When analyzing the amplification of different sCJD subtypes in BV substrate, a specific feature appeared for MM2 subtype with a maximum level of amplification with this subtype (detection of the 10 −9 dilution).
VV1 subtype also displayed a unique behavior as no, poor, or variable amplification was obtained with all of the substrates tested in our PMCA settings.
Amplification results obtained by PMCA in this study fit well with the previously proposed classification of M1 (MM1 and MV1), V2 (VV2 and MV2), M2 (MM2) and V1 strain (VV1), on the basis of the seeding efficiency with the different substrates and shifting properties of PrP res from the PMCA amplicons.
PMCA amplification allowed the detection of sCJD prions in CSF samples. CSF samples (n = 9) collected from patients with probable or definite CJD were submitted to four to six serial PMCA rounds using the most adapted substrate. All patients were homozygous at codon 129: 5 MM genotypes including 2 definite MM1 cases and one MM2 confirmed case, three VV genotypes including one definite VV2, and one definite vCJD case. According to our abovementioned results obtained using IBH, BV was used to amplify MM genotype, TgVal for VV genotype and TgMet, TgVal and BV for vCJD samples. Results are summarized in Fig. 4. Generated PMCA amplicons were analyzed by WB using 9A2 Ab. As previously shown 20 , vCJD amplification from CSF samples using TgMet is efficient with a positive signal observed from the second round. While TgVal and BV also supported vCJD amplification from CSF samples, 3 and 5 PMCA rounds, respectively, were required to detect a PrP res signal (Fig. 4b). Among the 5 sCJD-MM cases analyzed, only two ( §1 and §5) gave a positive result after prolonged PMCA rounds (i.e. 5 rounds for the sCJD-MM2 case ( §5) and 6 rounds for the MM probable sCJD case §1). WB analysis of PMCA amplicons showed a type 2 PrP res preservation for sample §5 from MM2 subtype (Fig. 4c). The three CSF samples from patients with probable or definite VV sCJD ( §6 to §8) could seed the PMCA reaction in TgVal substrate and gave a positive PrP res signal after three or four rounds. Considering the very low PMCA efficiency to amplify the VV1 subtype with TgVal substrate, we could assume that the two probable VV sCJD cases were likely of VV2 subtype.

Discussion
A major issue in TSE therapeutics research is the antemortem characterization of the prion agent/strain involved that would allow CJD patient stratification in clinical trials. This is of special interest since the efficacy of antiprion compounds may vary according to strain 43 , including CJD strains 44 . Recently, new in vitro approaches, based on the seeding properties of pathological prion aggregates were developed for the diagnosis of human TSEs. For example, RT-QuIC is a very sensitive and specific assay for the diagnosis of sCJD and is about to be implemented worldwide in neurology-specialized health care facilities 45,46 . In the present study, we provide a comprehensive analysis of the PMCA seeding activity across the spectrum of sCJD subtypes, using three different brain substrates derived from humanized PrP transgenic mice (methionine or valine homozygous at codon 129) and bank voles. Contrary to the vCJD prions, obtaining an efficient PMCA of sCJD prions was not as straightforward. That is why we became interested in cofactors, specifically heparin that had been used as a powerful enhancer of prion amplification by PMCA 47 . Based on its common usage, we have added heparin in all of the PMCA reactions described in this study. Overall, we found major differences in the seeding properties of sCJD subtypes. Depending on the seed/ substrate PMCA pairing, a shift in the electrophoretic mobility of the generated PrP res amplicons was observed in some cases. This heterogeneity highlights some important in vitro differences in the capacity of sCJD subtypes to seed PMCA reactions in a substrate-dependent manner.
According to the sensitivity of the PMCA and the PrP res electrophoretic profiles obtained using 3 different substrates, we were able to classify the different sCJD subtypes into 4 distinct groups, i.e. MM1/MV1, MM2, VV1, and MV2/VV2. This is in agreement with the previously proposed existence of 4 distinct sCJD prion strains based on bioassay transmission studies in human transgenic mice, bank voles and non-human primates 35,[48][49][50] , as inferred from the analysis of molecular characteristics of PrP TSE in vitro 12,51 or by combining both approaches 51 and referred to as M1, M2, V1 and V2 respectively.
Our data obtained with human transgenic mice also support the notion that the PrP C polymorphism at codon 129 from the substrate can directly influence the PrP res electrophoretic profile obtained after PMCA, sometimes leading to a shift (type-1 or type-2) from the initial PrP res signature. In TgMet brain substrate, both MV2 and VV2 sCJD acquired a type-1 PrP res , whereas in TgVal substrate, MM1, MV1 and VV1 sCJD sources, when amplified, shifted towards a type-2 PrP res . These observations are in partial accordance with the results of previous in vivo  48 . A similar in vivo observation was also reported after transmission to the same TgMet mice used in our study 49   www.nature.com/scientificreports/ amplified with substrate containing Valine129 PrP, gave type 2 amplicons 29 . Interestingly, at least for the V2 sCJD strain (MV2 and VV2), these phenotypic modifications were not accompanied by major changes in the initial seeding characteristics, as illustrated by the strong ability to replicate back in the TgVal substrate in the form of type-2 PrP res . The V2 strain amplified in TgMet substrate, while acquiring a type 1 profile, differed clearly from the V1 sCJD strain, which amplified very poorly in the TgVal substrate. This is reminiscent of the traceback phenomenon observed in vivo by passaging sCJD V2 prions to 129 M then 129 V human PrP-expressing mice 52 .
In contrast to the shift observed with PMCA in humanized transgenic mice substrates, our study revealed that amplification of different sCJD types in the bank vole (Met 109 ) substrate, in addition to sustaining human prion amplification in almost all cases tested, allowed a faithful conservation of the initial PrP res type present in the brain of the sCJD patients. This contrasts with a change from type-1 to type-2 PrP res reported by Redaelli et al. after PMCA amplification of MM1 sCJD in bank vole substrate 33 . Our results are however in good agreement with previous bioassay transmission studies in either wild-type bank voles 35 or transgenic mice overexpressing bank vole PrP (both Met 109 ) 36 , from which MM1 and MV1 sCJD propagate as type-1 PrP res and MM2, VV2 or vCJD propagate as type-2 PrP res . However, regarding glycosylation ratio, another hallmark of PrP TSE characterization, we noticed a systematic augmentation of the diglycosylated isoforms in all amplicons generated by PMCA, whatever the initial seed used (from IBH or CSF) or the substrate used. Using human prion strains, this modification in the glycosylated ratio from mono-to diglycosylated isoforms was previously mentioned from in vivo 35,36,38,53 , in vitro 33 or ex vivo 54 results.
The level of amplification obtained in the TgMet substrate in our study was somewhat disappointing, especially in the homologous PMCA context for MM1 and MM2 prions. Indeed the compatibility of the genotype at codon 129 between the seed and substrate has been proposed to be one of the most important factors for an efficient amplification 27 . Also, MM1 sCJD prions were transmitted very efficiently in TgMet mice 14,55 , which contrasts with our in vitro observations. Our failure to amplify MM1 and MM2 sCJD prions with TgMet substrate could be due to prion strain characteristics that were not fully compatible with the PMCA settings (sonication power and/or duration, time of incubation steps), and were thereby not efficiently sustaining the seeding of PMCA reactions, presumably in the early steps (initiation/elongation kinetics). It could be due also to an intrinsic inability of the Met129 PrP (TgMet) to support in vitro MM1 prion amplification, however, the same substrate allowed the amplification of vCJD very efficiently and repeatedly 56 , and to a lesser extent also the V2 sCJD strain. It is worth noting that, considering the PMCA outcome using substrates containing human PrP (TgMet and TgVal), we obtained the best amplification efficiency for V2 strain (VV2 and MV2) and vCJD which show the highest stability with regards to PK digestion 57 .
Given the encouraging PMCA amplification potential obtained for certain sCJD types in specific substrates and in order to demonstrate the feasibility of using PMCA on easily-accessible peripheral tissues or body fluids, we tested a small panel of CSF samples collected at clinical stage from patients with probable or definite sCJD. These included 5 MM cases that were tested in the bank vole substrate, 3 VV cases tested in TgVal substrate, as well as 1 vCJD case tested in all 3 substrates, which served as a positive control and was previously detected blindly using TgMet substrate 20 . For vCJD prion detection in CSF, the TgMet substrate appeared to be the most efficient, although positive results were obtained using TgVal and bank vole substrates after prolonged PMCA rounds.
Most importantly, we could detect a PMCA seeding activity in some of the sCJD CSF samples analyzed. Detection of sCJD PrP TSE from CSF samples by PMCA had been previously described by Rubenstein et al. 31 , however their specific PMCA setting associated to Surround Optical Fibre ImmunoAssay (SOFIA) as readout precluded sCJD subtypes differentiation. Regarding the seeding activity from the five CSF from MM patients, efficiency was variable as only two out of five cases were detected positive after 5/6 PMCA rounds. Our PMCA failed to identify CSF samples from the two definite MM1 sCJD patients. Considering RT-QuIC studies, the analytical sensitivity required to detect a seeding activity in the CSF of MM1 patients might be around the 10 −8 dilution of the brain samples. This level is not achieved in this study even by using bank vole PrP as substrate. Nevertheless, the CSF sample from a definite MM2 patient seeded the PMCA reaction in BV substrate and gave a positive type 2 signal after 5 PMCA rounds. Like PrP TSE of brain origin, PrP TSE from CSF samples displayed similar seeding behavior in BV substrate with regards to type 1/2 preservation. sCJD in CSF samples from VV patients was consistently detected using TgVal substrate after 3-4 PMCA rounds. These results indicate that all three VV sCJD patients were probably of VV2 subtypes, the second most common sCJD subtype after MM1.
In conclusion, this study demonstrates the potential of PMCA for the sensitive amplification of sCJD prions across the spectrum of human sCJD subtypes, which showed marked seed/substrate amplification heterogeneities. Noteworthy, PMCA could accurately discriminate between 4 sCJD prion strains thus recapitulating the current in vivo classification, establishing PMCA as an efficient in vitro model of sCJD prions propagation. For some prion strains-notably the V2 strain-the level of sensitivity achieved suggests that PMCA might be amenable to sCJD prion detection in peripherally accessible samples.

Methods
Brain tissues. Brain tissue obtained at autopsy from sporadic or variant CJD cases came from either the UK National Institute for Biologicals Standards and Control (NIBSC) CJD Resource Centre in the form of 10% (w/v) homogenate in 0.25 M sucrose (WHO reference NHBYO/0003 for vCJD) or from the French CJD National Surveillance Network as a block of frontal cortex tissue. The diagnosis was confirmed neuropathologically. All six currently defined subtypes of sCJD (MM1-MM2-MV1-MV2-VV1-VV2) were represented (2-3 different patients for each subtype-MM2 are of MM2 cortical type) (see Supplementary Table S1). Infected brain homogenates (IBH) at 20% (weight/volume) in 5% glucose were prepared using a high-speed homogenizer (MiniBeadbeater). Kingdom and by the university hospital (CHU) of Montpellier, France. The CSF had been collected in polypropylene tubes under standardized conditions. The samples were transferred to one laboratory within 4 h of being collected and centrifuged at a rate of 1000g for 10 min at 4 °C. It was then aliquoted into 0.5-mL polypropylene tubes and stored at − 80 °C for further analysis. CJD cases were classified as sporadic or variant CJD (definite or probable) using internationally recognized criteria 58 (see Supplementary Table S2).
PMCA. Normal PrP used as substrate was obtained from two humanized transgenic mouse lines: TgMet overexpressing sixfold the level of human PrP with a methionine at codon 129 (Tg650 line) 55 and TgVal overexpressing four-eight-fold the level of human PrP with a valine at codon 129 (Tg152 line) 59 ; wild type bank vole (BV) carrying methionine at codon 109 35 ; and bank vole transgenic mice (TgBV) overexpressing 4.9 fold the level of bank vole PrP with a methionine at codon 109 (Tg22019±) 60 . After collection, brains were rinsed in cold PBS and immediately frozen on dry ice before long-term storage at − 80 °C. Normal brain homogenates (NBH) were prepared at 10% (weight: volume) in conversion buffer (phosphate-buffered saline containing 150 mM sodium chloride, 1% Triton, protease inhibitor cocktail (Roche), EDTA 1 mM) and clarified at 2000×g for 20 s before freezing at − 80 °C in single-experiment aliquots. For amplification by PMCA, 10 µL of the different infected brain homogenates (IBH) serially tenfold diluted (10 −4 -10 −9 w/v) or 20 µL of CSF were mixed with 90 µL of PMCA substrate supplemented with heparin at 10 µg/ mL (Sigma) in PCR-tubes containing three Teflon beads (diameter 2.388 mm; Marteau & Lemarié). Unseeded substrates were also included in each PMCA experiments as Negative controls. Each PMCA cycle is composed of an incubation step (14 min 40 s at 37 °C) and a sonication step (20 s at 240 W). Successive rounds of 160 cycles were performed by diluting the amplified material in fresh heparin-supplemented PMCA substrate (1:10 for serial dilutions and 1:5 for CSF). For IBH dilutions, 4 amplification rounds were applied. CSF samples were amplified in up to 6 rounds of PMCA. To avoid any cross-contamination, experiments were carried out under strict quality control PCR conditions. Proteinase K (PK) digestion and SDS-PAGE/immunoblotting. After amplification, protease-resistant prion protein was detected by western blot as described previously 61 . After proteinase K digestion (200 µg/ mL) for 60 min at 45 °C and denaturation at 100 °C in SDS-PAGE denaturing buffer, samples were run on 12% polyacrylamide gel electrophoresis, before being electro-transferred onto PVDF membrane. Western blot (using the SNAP ID system, Millipore) was performed using 3F4 (mAb 3F4, epitope 109-112 of human PrP-Ozyme, France), 12B2 (mAb 12B2, epitope amino acid residues 89-93 of human PrP-Wageningen Bioveterinary, Netherlands), 9A2 (mAb 9A2, epitope amino acid residues 99-101-Wageningen Bioveterinary, Netherlands) or 6D11 (mAb 6D11, epitope 93-109 of human PrP sequence-Ozyme, France), and anti-mouse IgG peroxidaselinked secondary antibody (GE Healthcare, UK) linked to a chemiluminescent reaction (ECL blotting detection reagent, GE Healthcare, France), and imaged using films except for Fig. 3c using the imaging system Fusion FX7 (Vilber, France). The detection limit was determined visually after a maximum time exposure of 30 min and as result the dilution scored positive when the three characteristic PrP res bands were observed.
Ethics statement. The human CSF and brain samples used in this study were provided by the Laboratory of Clinical Proteomics (Montpellier, France) and the French National Center of Reference for Prions (Paris, France). A written informed consent for autopsy and research use was provided by patients' relatives, according to French legislation (L.1232-1 to L.1232-3, Code de la Santé Publique). Collection, preservation and preparation of human samples for research purpose have been declared to the French Ministry of Research (number DC-2008-417 and DC-2009-957) according to French regulation (L.1243-3 and L. 1243-49, Code de la Santé Publique). A few CSF samples were also provided by the NCJDRSU (UK) with informed consent obtained from the next of kin for research use (05/MRE00/67). All experiments on human samples were carried out in accordance with French regulation (L.1243-3 and L. 1243-49, Code de la Santé Publique) and all protocols were approved by the EFS research committee and supervised by the Scientific Direction of ANSM (Project No. P69).
All the animal experiments made to collect the brains from mice or bank voles at euthanasia (using carbon dioxide) were carried out in strict accordance with EU directive 2010/63. Mouse experiments were carried out in strict accordance with the recommendations from the Guide for the Care and Use of Laboratory Animals, as provided by the French Ministry of Agriculture and of the European Union (project authorization number 02298.03 provided by the French Ministry of Research after ethical evaluation). Bank voles were obtained from the breeding colony of Istituto Superiore di Sanità (ISS). The experimental protocol was approved and supervised by the Service for Biotechnology and Animal Welfare of the Istituto Superiore di Sanità and authorised by the Italian Ministry of Health (Decree No. 1119/2015-PR). All procedures were carried out in accordance with European Council directives 86/609 and 2010/63, as well as in compliance with the Italian Legislative Decree 26/2014.