Genotype-dependent Molecular Evolution of Sheep Bovine Spongiform Encephalopathy (BSE) Prions in Vitro Affects Their Zoonotic Potential*

Background: The results of serial passage of BSE in sheep are unknown. Results: In vitro modeling shows a sheep genotype-dependent switch in prion protein type and loss of the ability to convert human prion protein. Conclusion: Molecular evolution of sheep BSE prions in vitro is genotype-dependent and affects zoonotic potential. Significance: Zoonotic risk might be predicted from cell-free modeling. Prion diseases are rare fatal neurological conditions of humans and animals, one of which (variant Creutzfeldt-Jakob disease) is known to be a zoonotic form of the cattle disease bovine spongiform encephalopathy (BSE). What makes one animal prion disease zoonotic and others not is poorly understood, but it appears to involve compatibility between the prion strain and the host prion protein sequence. Concerns have been raised that the United Kingdom sheep flock may have been exposed to BSE early in the cattle BSE epidemic and that serial BSE transmission in sheep might have resulted in adaptation of the agent, which may have come to phenotypically resemble scrapie while maintaining its pathogenicity for humans. We have modeled this scenario in vitro. Extrapolation from our results suggests that if BSE were to infect sheep in the field it may, with time and in some sheep genotypes, become scrapie-like at the molecular level. However, the results also suggest that if BSE in sheep were to come to resemble scrapie it would lose its ability to affect humans.

Prion diseases are rare fatal neurological conditions of humans and animals, one of which (variant Creutzfeldt-Jakob disease) is known to be a zoonotic form of the cattle disease bovine spongiform encephalopathy (BSE). What makes one animal prion disease zoonotic and others not is poorly understood, but it appears to involve compatibility between the prion strain and the host prion protein sequence. Concerns have been raised that the United Kingdom sheep flock may have been exposed to BSE early in the cattle BSE epidemic and that serial BSE transmission in sheep might have resulted in adaptation of the agent, which may have come to phenotypically resemble scrapie while maintaining its pathogenicity for humans. We have modeled this scenario in vitro. Extrapolation from our results suggests that if BSE were to infect sheep in the field it may, with time and in some sheep genotypes, become scrapie-like at the molecular level. However, the results also suggest that if BSE in sheep were to come to resemble scrapie it would lose its ability to affect humans.
Prion diseases are transmissible fatal neurodegenerative disorders of the central nervous system that occur in animals and humans and are caused by unconventional agents termed prions. A hallmark of acquired prion diseases is a prolonged asymptomatic phase followed by the appearance of neurological signs, neuropathological changes, and the accumulation of a misfolded pathogenic isoform (PrP Sc ) 2 of the cellular prion pro-tein (PrP C ) in the central nervous system (1). Prion diseases can be transmitted within and between species, but crossing a socalled "species barrier" often results in prolonged incubation periods (2). It is only after adaptation of the agent to the new host that the incubation period shortens and the disease phenotype stabilizes. This species barrier effect is thought to be determined at least in part by prion protein sequence homology between the species involved (3).
Bovine spongiform encephalopathy (BSE) is the prototypic acquired prion disease of cattle, first recognized in 1986 in the United Kingdom (4) and reaching epidemic proportions before being brought under control (5). The appearance of a new form of Creutzfeldt-Jakob disease (variant Creutzfeldt-Jakob disease (vCJD)) in the United Kingdom during the following decade was strongly suspected to result from dietary exposure of the human population to the BSE agent (6), and subsequent investigations of its transmission properties are consistent with this explanation (7)(8)(9). The number of cases of vCJD occurring in the United Kingdom peaked in the year 2000 and has declined since. All cases of definite clinical vCJD are of one prion protein genotypic group (PRNP codon 129 MM). BSE is the only known zoonotic animal prion disease (5).
Scrapie is the most intensively researched animal prion disease. It is endemic in sheep in many countries, including the United Kingdom. Scrapie prion strain diversity can be inferred from differences in incubation period, in PrP Sc profile and distribution, and in vacuolar lesions in the brain (11). Scrapie susceptibility and incubation period are largely determined by polymorphic variation in the sheep prion protein gene (Prnp) with those at 136, 154, and 171 and the VRQ, ARQ, and ARR alleles having the greatest effect (12). Depending on the infecting source or strain (11), VRQ/VRQ and ARQ/ARQ sheep are the most susceptible to classical scrapie, whereas ARR/ARR is considered to be effectively resistant (13).
It was recognized early on that if BSE were to have infected sheep then its presence could be masked by endemic scrapie with potentially serious and long term effects for animal and human health (14,15). Sheep were confirmed to be experimentally susceptible to BSE (16 -19). The clinical signs of scrapie and ovine BSE are indistinguishable (20); however, the diseases can be distinguished by careful immunohistochemical analysis of PrP Sc accumulation patterns in the brain and lymphoid tissues (21)(22)(23)(24)(25) and by the biochemical characteristics of the protease-resistant prion protein (PrP res ) in the brain (26,27). Susceptibility to experimental BSE is associated with the ARQ/ ARQ genotype, but BSE in VRQ/VRQ sheep can likewise be obtained (28), and infection of ARR/ARR sheep has also been reported (19, 29 -31). Discriminatory tests for sheep BSE based on differences in molecular profile were shown to remain effective after three serial experimental passages of the BSE agent in ARQ/ARQ sheep, but there was an indication of a shift in the biochemical properties of PrP res during these passages away from that associated with BSE (27). It is therefore possible that adaptation of the BSE agent to an ovine host involves a change in the molecular properties of the BSE prions and that this change may have occurred prior to the introduction of statutory testing. In this scenario, sheep BSE in the field may have become indistinguishable from scrapie, but it may have retained its pathogenicity for humans.
PrP misfolding and prion replication can be mimicked in cell-free PrP conversion assays such as protein misfolding cyclic amplification (PMCA) (32,33). PMCA can be used to model the transmission and strain adaptation phenomena associated with prion replication in vivo but at an accelerated rate (34 -40). The results from these studies suggest that species, strain, and genotypic barriers to prion disease can be modeled in vitro. We have modeled serial passage of the BSE agent in sheep by conducting serial PMCA (sPMCA) using sheep BSE brain homogenates to seed ovine brain substrate of three Prnp genotypes and then testing whether this in vitro adaptation process results in changes in the potential of sheep BSE to convert human PrP C in a further PMCA reaction.

EXPERIMENTAL PROCEDURES
Ethics Statement-Human tissues were obtained from the CJD Brain and Tissue Bank, which is part of the Medical Research Council Edinburgh Brain Banks. Tissues were collected with consent for research use. Ethical approval for the use of the human tissues in this study was covered by LREC 2000/4/157 (Professor James Ironside). All studies, including experimental inoculations, care of animals, and euthanasia, were carried out in accordance with the United Kingdom Animal (Scientific Procedures) Act 1986. Sheep were obtained from one of two facilities. Experiments performed at the Moredun Research Institute were carried out under licenses from the United Kingdom Government Home Office number 60/2656 (renewed in 2005 with number 60/3646). The remaining sheep were obtained from experiments carried out at the Agricultural Development and Advisory Service facilities at High Mowthorpe under project license number 70/5155. Animals were monitored daily for the presence of neurological signs compatible with TSE and were euthanized once those signs reached a predetermined end point, when showing signs of intercurrent disease unresponsive to treatment, or for welfare reasons. In all cases, euthanasia was performed by intravenous injection of barbiturate overdose followed by exsanguination.
Uninfected Animal Brain Tissues-Nine samples of ovine brain tissue of the three major scrapie-susceptible or -resistant variants, differing in their Prnp polymorphism at codons 136, 154, and 171 (both PBS-perfused or non-perfused; two VRQ/ VRQ, three ARQ/ARQ, and four ARR/ARR) were obtained from a scrapie-free flock (ARSU flock) at the Animal Health and Veterinary Laboratories Agency (Weybridge, UK). The bovine (BSE-negative) sample came from cow with limited or no exposure to BSE reared under controlled conditions, and the tissues were provided by the Animal Health and Veterinary Laboratories Agency TSE Archive (Weybridge, UK). All brain tissues were stored at Ϫ80°C immediately after animals were sacrificed. The disease status of these animals was confirmed at source by prion protein immunohistochemistry and Western blot.
Experimental Sheep BSE, Cattle BSE, and Sheep Scrapie Tissues-Brain stem samples from five sheep experimentally infected with BSE (homozygous VRQ/VRQ, ARQ/ARQ, and ARR/ARR BSE-infected sheep), the scrapie-infected sheep, and the BSE-infected cattle brain tissues were produced or collected by Animal Health and Veterinary Laboratories Agency (Lasswade and Weybridge, UK). The BSE-positive cow was a field suspect that had been identified through passive surveillance, and the tissues were provided by the Animal Health and Veterinary Laboratories Agency TSE Archive. The disease status of the animals was confirmed at source by prion protein immunohistochemistry and Western blot.
Prnp Sequencing-Prnp genotyping of the sheep involved in this study was performed on blood samples by PCR amplification and sequencing of the whole open reading frame of the Prnp gene on a 3130 Genetic Analyzer with the BigDye terminator v3.1 cycle sequencing kit according to the manufacturer's protocol (Applied Biosystems).
Human Brain Tissues-All tissues were handled exclusively in the category 3* biosafety containment facility according to stringent health and safety protocols. Human brain tissues (frontal cortex) were sampled from a frozen half-brain collected at autopsy with the appropriate consent for tissue retention and research use. The vCJD specimen was from a patient (PRNP codon 129 MM) with definite variant CJD diagnosis as defined by established criteria. The sCJD sample was from a patient with a diagnosis of sporadic CJD (VV2 subtype). The non-CJD human brain specimens used for PMCA substrate preparation were from frontal cortex from patients with Guillain-Barré syndrome (PRNP codon 129 MM) and dementia with Lewy bodies (PRNP codon 129 VV), and their use as PMCA substrates has been described previously (39).
Preparation of Brain Homogenates-Bovine, ovine, and human tissue homogenates (10%, w/v) were prepared by manual homogenization of brain tissues using glass grinders (Fisher Scientific Ltd.) in PMCA conversion buffer (PBS, 1% Triton X-100, 150 mM NaCl) containing protease inhibitors (Complete Mini EDTA-free, Roche Applied Science). A non-ionic detergentinsoluble nucleocytoskeletal fraction was cleared by centrifugation at 424 ϫ g for 40 s. The supernatants were divided into aliquots and stored at Ϫ80°C until use as PMCA substrate.
Protein Misfolding Cyclic Amplification-EDTA (Fluka, Sigma-Aldrich) was added to all brain homogenate substrates at a final concentration of 5 mM prior to conducting the PMCA reaction. When human tissue was used as substrate, heparin (LKT Laboratories) (10) was added to the samples at a final concentration 1 mM. Substrates were seeded at 4°C with a dilution of brain homogenate (previously titrated by Western blotting) such that the seed material contained sufficient PrP res to be detectable at the lower end of the linear range of a standard Western blot. A negative control aliquot not subjected to PMCA (20 l) was immediately frozen at Ϫ80°C, and the remainder (sample of 100 l) was subjected to PMCA using a Misonix model 4000 sonicator. A single round of PMCA comprised 96 cycles of 20-s sonication at 80% total power output and a 29-min 40-s incubation at 37°C. When serial PMCA reactions were performed, each new round was seeded at a 1:10 dilution with the product of the previous PMCA round (i.e. a 10-fold dilution of the PMCA product in fresh brain homogenate substrate). An aliquot of each sPMCA reaction product was retained for a use as a seed for human brain tissue substrate in phase III of the study.
Sample Preparation for Western Blot Analysis-Equal volumes of PMCA products and their corresponding negative controls were digested using proteinase K (PK; Novagen, EMD Millipore) at a final concentration of 50 g/ml at 37°C for 1 h. Protease digestion was terminated by the addition of Pefabloc SC (Roche Applied Science) to a final concentration of 1 mM, and the samples were then collected by centrifugation at 20,817 ϫ g for 1 h at 4°C. Pellets were then resuspended in PMCA conversion buffer, and an equal volume of 4ϫ NuPAGE lithium dodecyl sulfate sample buffer (Invitrogen) was added to each sample aliquot to a final concentration 1ϫ.
Western Blot Analysis-Samples were boiled at 100°C for 10 min and loaded onto a NuPAGE 10% Bis-Tris gel (Invitrogen) and subjected to electrophoresis for 50 min at 200 V using preset gel cassettes (Invitrogen) and NuPAGE MES-SDS 1ϫ running buffer (Invitrogen). A BenchMark prestained protein ladder (Invitrogen) and a MagicMark TM XP Western protein standard (Invitrogen) were run alongside the samples. The gels were then electroblotted onto polyvinylidene difluoride (PVDF) membrane (Hybond-P, GE Healthcare) for 1 h at 30 V using 1ϫ transfer buffer consisting of 4% 20ϫ NuPAGE transfer buffer (Invitrogen), 16% MeOH, 80% distilled H 2 O. For immunodetection, the PVDF membrane was blocked with a solution of 5% (w/v) nonfat milk powder (Sigma-Aldrich) dissolved in TBS-T (200 mM Tris/HCl, 150 mM NaCl, pH 7.6) containing 0.1% Tween 20 (polyoxyethelenesorbitan monolaurate, Fisher Scientific Ltd.) for 60 min. Then the PVDF membrane was incubated with anti-PrP monoclonal antibody 6H4, 12B2, or 3F4 in TBS-T for 60 min and subsequently with horseradish peroxidase-conjugated sheep anti-mouse IgG (GE Healthcare) antibody for 60 min. The membranes were developed using ECL Plus (GE Healthcare) and imaged using the ChemiDoc TM XRSϩ System (Bio-Rad) following the manufacturer's instructions.
Quantitative Densitometry and Statistical Analysis-The difference in sheep BSE PrP res amplification efficiency in ovine substrates of different genotype was analyzed from three independent identical experiments. Serial PMCA products were analyzed by Western blot, and densitometry of the appropriate bands was performed using a volume tool of the XRSϩ System Image Lab TM 2.0 software. The background signal values of individual blots were subtracted before data were analyzed and portrayed using GraphPad Prism version 6.01 software.
Comparison of PrP C Levels in Substrate Brain Homogenates-The total protein concentration of the brain homogenates was obtained using DC TM Protein Assay kit (Bio-Rad). Equal amounts of protein were analyzed by Western blot using anti-PrP primary antibody 6H4. The PrP C concentration was then obtained by densitometry using a known amount of recombinant PrP C run on the same blot. In experiments designed to normalize PrP C levels in sheep brain homogenates to be used as PMCA substrates, equivalent volumes of 10% brain homogenates were analyzed by Western blotting using the anti-PrP primary antibody 6H4, the PrP C signal was quantitated by densitometry, and the dilution factors to reach equivalence were calculated.
Deglycosylation of Ovine PrP C -Ovine brain homogenates of equal PrP C amount (20 g) were denatured at 100°C for 10 min and then incubated with 500 units/l peptide-N-glycosidase F (PNGase F kit, New England Biolabs) for 2 h at 37°C according to the manufacturer's instructions. Deglycosylated proteins were precipitated using 4 volumes of ice-cold 80% MeOH and collected by centrifugation at 18,787 ϫ g for 30 min at 4°C. The supernatant was removed, and the pellets were resuspended in NuPAGE sample reducing agent (Novex, Invitrogen) and NuPAGE lithium dodecyl sulfate sampling buffer and analyzed by Western blot.

RESULTS
Overall Study Design- Fig. 1 gives an overview of the study. Phase I established the PMCA conditions that resulted in efficient amplification of cattle PrP res in normal cattle brain substrate and in normal ovine brain substrates of the VRQ/VRQ, ARQ/ARQ, and ARR/ARR genotypes. Phase II tested whether the BSE-like molecular signature of ovine BSE PrP res was stable when serially amplified in ovine substrates or whether there was a shift toward a scrapie-like molecular signature. Classical scrapie brain homogenates from VRQ/VRQ and ARQ/ARQ sheep were used to seed all three ovine polymorphic groups, providing a positive control for amplification and PrP res -type conservation. Phase III determined whether any molecular switch from a BSE-like to a scrapie-like molecular signature altered the ability of the sheep-adapted BSE PrP res to convert PrP C in human brain substrates. In this phase, the sPMCA products from rounds 1 and 8 produced in phase II were used to seed human brain substrates of the PRNP codon 129 MM and VV genotypes and subjected to a single round of PMCA. The PMCA products were then analyzed by Western blot after PK treatment using the anti-PrP mAb 3F4 (which detects human but not ovine or bovine PrP).

BSE Agent Replication Modeled by Amplification in Ovine
Substrates-Sheep BSE brain homogenates and positive control brain homogenates of cattle BSE and classical sheep scrapie gave the expected protein banding pattern corresponding to the three glycoforms of PrP res on Western blot analysis. The diglycosylated form of the protein (top band) exhibited the strongest signal followed by the mono-(middle band) and nonglycosylated (lower band) forms, which exhibited weaker signals ( Fig. 2A). The non-glycosylated band of PrP res of cattle BSE and sheep BSE had an apparent molecular mass of around 19 kDa (Fig. 2A, lane 1-6), whereas in classical scrapie, it was around 20 -21 kDa (Fig. 2A, lane 7). Next, we established the PMCA conditions needed to amplify BSE PrP res efficiently in normal bovine brain (Fig. 2B) and PrP res from classical scrapie in ovine brain (Fig. 2C). The results confirmed that the PrP res type of PMCA products resembled that of the seed (Fig. 2, B and C). In subsequent PMCA reactions, the initial seed dilution was sometimes adjusted, but the duration and power output of sonication, duration of incubation period, and number of sonication/incubation cycles per round were kept identical for all experiments of this study. We then used these standardized conditions to amplify BSE PrP res in ovine substrates of the three main ovine Prnp polymorphic genotypes (Fig. 2, D and E). We compared the efficiency of amplification in ovine brain substrates from animals that had been perfused with PBS ( Fig. 2D) with those that had not (Fig. 2E). Amplification was observed after a single round of PMCA in ovine substrate of the VRQ/ VRQ (Fig. 2, D and E, lane 2) and the ARQ/ARQ (Fig. 2, D and E, lane 4) genotypes but not in ARR/ARR (Fig. 2, D and E, lane 6) genotype substrates. Amplification was found to be slightly more efficient using substrates from perfused animals (Fig. 2D,  lanes 2 and 4) when compared with the non-perfused brains  ( Fig. 2E, lanes 2 and 4), and all further experiments were carried out using the former. Given the failure of the ARR/ARR substrate to support PrP res amplification, we sought to determine whether the PrP C levels present in all nine ovine substrates used were the same. The ARR/ARR brain substrates were found to have a consistently lower level of PrP C than the VRQ/VRQ and ARQ/ARQ brain substrates (Fig. 3, A and B).
BSE Agent Adaptation to a VRQ/VRQ Ovine Host Modeled by Serial PMCA-Next, we modeled sheep BSE agent adaptation by conducting sPMCA of experimental sheep BSE brain homogenate seeded in ovine brain substrates of different Prnp genotypes ( Fig. 1, phase II). Each round was composed of 96 cycles of sonication and incubation (as described under "Experimental Procedures"). Samples from each round were analyzed for PrP res by Western blotting using the anti-PrP mAb 6H4 (which recognizes both BSE and scrapie PrP res ) and 12B2 (which, after protease digestion, detects scrapie but not BSE PrP res ) (41). When the VRQ/VRQ ovine brain substrate was seeded with BSE brain homogenate from either VRQ/VRQ, ARQ/ARQ, or ARR/ARR sheep (Fig. 4A), amplification of sheep BSE PrP res was maintained, and a clear molecular switch of PrP res type from a BSE-like to scrapie-like molecular signature occurred at rounds 6 -8. The sheep-adapted BSE PrP res PMCA product exhibited a non-glycosylated band of higher molecular mass than that of the cattle BSE and was comparable with that of the natural scrapie positive controls (Fig. 4A, marked Sc). This switch from BSE-like to scrapie-like PrP res was confirmed by analyses using the scrapie PrP res -specific mAb 12B2. The sheep BSE PMCA products were only poorly detected with this anti-body during rounds 1 to 5 or 6, but a strong PrP res signal appeared at rounds 6 -8 (Fig. 4A, right). The experiment was conducted using identical conditions on three independent occasions (the last time in triplicate), and the switch from BSElike to scrapie-like PrP res type was found on each occasion. Semiquantitative densitometric assessment of the PrP res signal in the nine sPMCA reactions of ovine VRQ/VRQ substrate seeded with sheep BSE isolates (n ϭ 9 per substrate genotype) confirmed the visual observations (Fig. 4B). PrP res was efficiently amplified throughout the sPMCA (Fig. 4B, 6H4 analysis, blue bars), and a clear rise of scrapie-like PrP res signal detected by 12B2 was evident at rounds 6 -8 (Fig. 4B, red bars).
BSE Agent Adaptation to an ARQ/ARQ Ovine Host Modeled by Serial PMCA-The ARQ/ARQ ovine brain substrate seeded with sheep BSE of the VRQ/VRQ, ARQ/ARQ, or ARR/ARR genotype (the same as those used above) and subjected to eight rounds of sPMCA also supported PrP res amplification throughout sPMCA as seen in Western blotting using the 6H4 mAb (Fig. 4C, left). The sPMCA sheep BSE PrP res product exhibited the non-glycosylated band of the same molecular mass as the cattle BSE seeds when the 6H4 mAb was used, indicating retention of the BSE-like signature. Analysis of the same samples using the 12B2 mAb (Fig. 4C, right) showed low levels of PrP res with a molecular mass similar to that of the classical scrapie positive control. This suggests that a minor subpopulation of scrapie-type PrP res co-amplifies in the samples. This experiment was carried out using the same conditions on three separate occasions (the last time in triplicate), and the results were reproducible, and a semiquantitative densitometric assessment of the Western blots (n ϭ 9 per substrate genotype) is shown (Fig. 4D).
BSE Agent Adaptation to an ARR/ARR Ovine Host Modeled by Serial PMCA-In contrast, despite some evidence of amplification in the firsts PMCA rounds, the ovine ARR/ARR substrate failed to support amplification in subsequent rounds with any of the three sheep BSE seeds used as above (Fig. 4E). The PrP res signal was gradually lost whether the detection mAb was 6H4 or 12B2. The experiment was conducted three times (last time in triplicate), and the loss of PrP res signal during sPMCA using the ARR/ARR was found on each occasion. Densitometric assessment of the Western blots (n ϭ 9 per substrate genotype) is shown (Fig. 4F).
Confirmatory and Control PMCA Experiments Using Additional Seeds and Non-perfused Brain Substrates-The same set of experiments was repeated using non-perfused ovine brain substrates (VRQ/VRQ, ARQ/ARQ, and ARR/ARR), but the PrP res was lost after the third round in all experiments (data not shown). Amplification of scrapie PrP res in the perfused ovine substrates was also carried out as a control for sheep PrP res -type conservation during sPMCA. The VRQ/VRQ ovine substrate seeded with VRQ/VRQ and ARQ/ARQ sheep scrapie amplified scrapie-like PrP res throughout sPMCA, whereas the ARQ/ARQ ovine substrate amplified the ARQ/ARQ but not the VRQ/ VRQ sheep scrapie. The ARR/ARR substrate failed to support scrapie PrP res amplification. An example of ovine VRQ/VRQ substrates seeded with classical ARQ/ARQ scrapie, subjected to eight rounds of sPMCA, and analyzed by Western blot using mAbs 6H4 and 12B2 is shown (Fig. 5, A and B), demonstrating FIGURE 3. Ovine brain substrates of the VRQ/VRQ and ARQ/ARQ genotypes exhibit higher levels of PrP C than the ARR/ARR genotype. A, representative Western blot of the nine ovine brain substrates used in this study. Samples were loaded at 5 g of total protein/lane, and the Western blot was probed with mAb 6H4. B, graph portraying the levels of PrP C in the ovine brain substrates. Data were obtained by densitometry of three independent Western blot analyses (background-subtracted) of identical sample loading, an example of which is shown in A. Data are means Ϯ S.D. (error bars).
that the scrapie PrP res type was conserved during the sPMCA experiment. Next, we used the cattle BSE brain sample (rather than experimental sheep BSE) to seed the same (perfused) ovine brain substrates as used above and subjected them to seven rounds of sPMCA (Fig. 6). The VRQ/VRQ ovine substrate seeded with cattle BSE showed efficient amplification throughout the sPMCA experiment, and a switch of PrP res type from a BSE-like to a scrapie-like signature was again observed, this time at round 2-3 (Fig. 6, A and B). The ARQ/ARQ ovine substrate amplified both BSE-like and scrapie-like PrP res at lower but detectable levels (Fig. 6, C and D), and the ARR/ARR substrate again failed to amplify BSE PrP res to any readily detectable extent (Fig. 6, E and F).
Reintroduction of the VRQ/VRQ Sheep-adapted BSE sPMCA Product to a Bovine PMCA Substrate-The sheep substrate genotype-dependent switch of PrP res type can be seen most clearly when round 1 and round 8 sPMCA products using VRQ/VRQ and ARQ/ARQ substrates are compared on the same Western blot along with classical sheep scrapie and cattle BSE PrP res as reference standards (Fig. 7A). To test whether this switch in PrP res type was a reversible phenomenon, we conducted an experiment using the sheep-adapted BSE sPMCA product that had acquired a scrapie-like PrP res molecular phenotype to seed bovine brain substrate in sPMCA (Fig. 7, B-D). We seeded the bovine substrate with sPMCA-generated material (Fig. 4A, r8, final sPMCA product) and subjected it to four rounds of sPMCA. Although the amplification level was poor, Western blot analysis using the 6H4 mAb showed that the nonglycosylated band of PrP res in the amplified PMCA product was of molecular mass typical for BSE (Fig. 7B, lanes 2-5) rather than of classical scrapie run alongside the sPMCA samples (Fig.  7, B and C, lane 6, marked Sc) or of the scrapie-like seed (Figs. 4A, r8, and 7A, lane 3) used to seed this reaction. Analysis using the 12B2 mAb confirmed this observation as the PrP res signal detected by 12B2 antibody was lost in the subsequent passages of the sPMCA (Fig. 7C, lanes 2-5). Graphical representation of densitometry data confirmed this visual observation (Fig. 7D).
Controlling Serial Protein Misfolding Cyclic Amplification for de Novo PrP res Formation-In each of the above experiments, unseeded control PMCA reactions were run in parallel with the seeded PMCA reactions to test for de novo PrP res formation. At early stages of the project, PrP res was found in some of these unseeded controls. This only occurred when scrapie-seeded samples were run in the same sPMCA experiment. Results from these experiments were disregarded. After the introduction of stringent liquid handling precautions to avoid cross-contamination, unseeded controls remained PrP res -negative (Fig. 8). We concluded that the appearance of PrP res in the unseeded control reactions at the early stages of the project represented cross-contamination rather than de novo formation.
Competence of the ARR/ARR Substrate to Support PrP res Replication in PMCA-The failure of sheep ARR/ARR substrates to support amplification in any of the above sPMCA experiments might be attributed directly to the presence of arginine at position 171, but other aspects of PrP C expression may also be relevant. The ARR/ARR substrate contained lower levels of PrP C compared with VRQ/VRQ or ARQ/ARQ (Fig. 3). Densitometric analysis of a comparative Western blot indicated that the VRQ/VRQ substrate contained 2.2 times the amount of PrP C found in the ARR/ARR substrate. Hence, we repeated the sPMCA experiment using the ARR/ARR brain homogenate at 20% (w/v) instead of the standard 10%, effectively doubling the PrP C concentration inter alia. However, the results were again negative (Fig. 9, A and B). The reverse strategy was also attempted in which the 10% (w/v) VRQ/VRQ substrate was diluted to match the PrP C concentration of the 10% (w/v) ARR/ ARR substrate. Dilution of the VRQ/VRQ substrate reduced the amplification efficiency of VRQ sheep BSE seed, resulting in amplification in sPMCA rounds 1 and 2 that failed to overcome the effect of serial dilution in subsequent rounds (Fig. 9, C and D).
Nevertheless, amplification of the VRQ sheep BSE seed in the ϳ5% (w/v) VRQ/VRQ sheep substrate was greater than amplification of the same seed in the 10% (w/v) ARR/ARR substrate FIGURE 4. VRQ/VRQ and ARQ/ARQ ovine brain substrates support amplification of sheep BSE PrP res in serial PMCA. A, ovine VRQ/VRQ substrate seeded with sheep BSE, subjected to eight rounds (r1-r8) of sPMCA and analyzed by Western blotting using anti-PrP mAb 6H4 (left) and 12B2 (right). C, the ARQ/ARQ substrate seeded and analyzed as above. E, the ARR/ARR substrate seeded and analyzed as above. Ovine brain substrates were seeded with experimental sheep BSE VRQ/VRQ, ARQ/ARQ, and ARR/ARR as indicated in the figure and subjected to eight rounds of sPMCA. Ϫ, corresponding controls not subjected to PMCA. Classical scrapie (Sc) and BSE standards were used as controls of PrP res migration, antibody specificity, and blotting procedure. Samples were analyzed after PK treatment using anti-PrP mAb 6H4 (left) and 12B2 (right). Note that the signal from the classical scrapie control (Sc) that was included in each blot was consistently greater with 12B2 than with 6H4, indicative of a higher detection sensitivity of scrapie-like PrP res using the 12B2 antibody compared with 6H4.   A and B, representative Western blot analysis of ovine VRQ/VRQ substrate seeded with classical scrapie (ARQ/ARQ genotype) and subjected to eight rounds (r1-r8) of sPMCA. Classical scrapie (Sc) and BSE standards were used as a control for PrP res migration and the blotting procedure. Samples were analyzed after PK treatment using anti-PrP mAb 6H4 (A) and 12B2 (B). (Fig. 9, E and F) as shown by densitometric analysis (Fig. 9, D  and F). The 10% VRQ/VRQ substrate efficiently supported the sheep BSE PrP res amplification in all four rounds of sPMCA in agreement with our earlier results (Fig. 9, G and  H). Additionally, PrP C deglycosylation and Western blot analysis showed that all VRQ/VRQ, ARQ/ARQ, and ARR/ ARR (Fig. 10, A, lanes 2, 4, and 6, respectively, and B) homogenates contained full-length PrP C and the C1 fragment at approximately equal relative amounts in all three polymorphic variants, suggesting that differential proteolytic processing is not responsible for the failure of the ARR/ARR samples to support amplification.  A and B, the VRQ/VRQ ovine substrate seeded with BSE PrP res , subjected to seven rounds (r1-r7) of sPMCA, and analyzed by Western blot using mAb 6H4 (A) and 12B2 (B). C and D, the ARQ/ARQ substrate seeded with BSE PrP res , subjected to seven rounds of sPMCA, and analyzed by Western blot using mAb 6H4 (C) and 12B2 (D). E and F, the ARR/ARR substrate seeded with BSE PrP res , subjected to seven rounds of sPMCA, and analyzed by Western blot using mAb 6H4 (E) and 12B2 (F). Conditions of sPMCA were identical to those used throughout this study. Samples were analyzed by Western blotting after PK treatment. Corresponding controls for each sample that were not subjected to PMCA are shown on the left of leach blot (Ϫ). Classical scrapie (Sc) and BSE samples were used as reference standards, for antibody specificity, and for the blotting procedure. The molecular mass (in kDa) is marked to the left of each blot.  1-3) and ARQ/ARQ substrate (lanes 6-8) seeded with sheep BSE ARQ/ARQ isolate. B and C, four rounds (r1-r4) of sPMCA of bovine brain substrate seeded with the sheep-adapted BSE sPMCA product (PrP res of scrapie-like molecular phenotype; Fig. 4A, lane 3). D, graphical representation of densitometric analysis of PrP res amplification level in the reversion experiment shown in B and C. The data were background-subtracted and normalized by the scrapie standard run alongside the samples (B and C, lane 6). sPMCA conditions were identical to those used in previous sPMCA experiments. Ϫ, corresponding controls not subjected to PMCA. Classical scrapie (Sc) and BSE were used as reference standards and as a control for antibody specificity and the blotting procedure. Samples were analyzed after PK treatment using anti-PrP mAb 6H4 (A and B) and 12B2 (C). The molecular mass (in kDa) is marked to the left of each blot.

Zoonotic Transmission of in Vitro Sheep-adapted BSE Agent
Modeled by PMCA Using Human Substrates-In the third phase of the project (Fig. 1, phase III), we first seeded the human brain substrate of PRNP codon 129 MM and VV genotypes with experimental sheep BSE, cattle BSE, and classical scrapie samples and subjected these to a single round of PMCA (Fig. 11). Seeding with human vCJD (MM brain substrate) and sCJD (VV brain substrate) brain samples served as positive controls for the PMCA reactions. Samples were PK-treated and analyzed by Western blotting using anti-PrP mAb 3F4. The human brain substrate of the codon 129 MM genotype supported amplification of the VRQ/VRQ, ARQ/ARQ, and ARR/ARR experimental sheep BSE, cattle BSE, and vCJD PrP res but did not support amplification of scrapie PrP res (Fig. 11A). The human brain substrate of the codon 129 VV genotype only poorly amplified PrP res from sheep or cattle BSE, and no PrP res amplification was observed when it was seeded with scrapie. However, efficient amplification of sCJD PrP res with a matching codon 129 VV genotype was obtained (Fig. 11B), confirming the general competence of this substrate to support amplification. This confirms previous observations and indicates that species, strain, and genotypic barriers to prion diseases were respected in our in vitro system. Next, the sheep-adapted BSE sPMCA products from rounds 1 and 8 generated in phase II of this study were used to seed the human brain codon 129 MM and VV substrates. The sheep-adapted BSE PrP res that had retained its BSE-like molecular signature was found to have also retained its ability to convert human PRNP codon 129 MM PrP C (Figs. 12A,  lanes 2, 6, and 10, and 3, B, lanes 2, 4, 6, 8, 10, and 12, and C,  lanes 6 and 10). In marked contrast, the in vitro sheep-adapted BSE agent that had acquired a scrapie-like molecular signature was found to have lost its ability to convert human PrP C of the PRNP codon 129 MM genotype (Fig. 12A, lanes 4, 8, and 12). The result was confirmed in a second independent experiment conducted under identical conditions. The human brain sub-strate of PRNP codon 129 VV genotype failed to amplify PrP res when seeded with any of the sheep-adapted BSE sPMCA-generated products (data not shown).

DISCUSSION
BSE is a prion disease of cattle that has been shown to be transmissible to other animals and is zoonotic, causing vCJD in humans. If BSE entered the United Kingdom sheep flock and has been maintained since the 1980s and 1990s, it would currently be at its ϳ20th passage. It is possible that this process could have resulted in an adaptation of the agent to the ovine host that rendered BSE indistinguishable from scrapie at the molecular level. Because of the protracted nature (and the attendant financial and ethical considerations) of modeling such a process in vivo, we have chosen to model this scenario by conducting serial PMCA of ovine brain substrates seeded with experimental sheep BSE brain homogenates. We estimate that conducting eight serial passages in vivo using large animals (sheep) or small animals (appropriate ovinized transgenic mouse lines) followed by transmission to non-human primates or humanized transgenic mice would take over 8 years (in mice) and over 20 years (in sheep) to conduct. In contrast, eight serial rounds of ovine PMCA followed by a single round of PMCA in human substrate can be completed in 1 month.
Using this in vitro approach, we found that the major factor influencing the propagation of the sheep BSE agent was the genotype of the host sheep brain substrate ( Table 1). The VRQ/ VRQ substrate efficiently amplified sheep BSE PrP res during serial rounds of PMCA. Inclusion of appropriate comparable controls (scrapie and BSE) on each Western blot was important for distinguishing between the molecular types of PrP res propagated in the ovine substrates. In addition, a discriminatory antibody comparison method was used (mAbs 6H4 and 12B2) that, in conjunction with Western blotting, allowed us to unambiguously classify the predominant PrP res types present before and after PMCA. The Western blotting results demonstrated a clear and reproducible switch of mobility from that of BSE in sheep to that found in natural scrapie after five to six rounds of sPMCA that was confirmed by the relative 6H4/12B2 antibody binding. Aspects of the agent strain are often preserved for several passages following cross-species transmission (27,42,43). It has been proposed previously that a single round of PMCA reaction approximates to one in vivo passage, and a process of adaptation of certain prion strains to a new host was reported to occur following three to six rounds of serial PMCA (36,37). Our data appear consistent with these observations. The ARQ/ARQ substrate also supported efficient PrP res amplification throughout the serial rounds of PMCA, but in this substrate, a switch to a scrapie-like PrP res was not evident; instead a BSE-like PrP res predominated accompanied by a less readily detectable scrapie-like PrP res component. This suggests a role for valine (as opposed to alanine) at position 136 in favoring a scrapielike PrP Sc conformation ( Table 1). The relative genotypic specificities of sheep BSE seeds and the substrate genotype-dependent switch in predominant PrP res type were replicated when cattle BSE was used as a seed in sPMCA instead of the experimental sheep BSE samples. In an effort to determine whether the VRQ/VRQ substratedependent switch from a BSE-like to a scrapie-like PrP res type was reversible, we attempted to reamplify the sheep VRQ/VRQ sPMCA product in a bovine brain substrate. Sustained amplification of PrP res was not achieved for either PrP res type in the bovine substrate. This outcome is consistent with a previous report that sheep scrapie fails to amplify in a bovine PMCA substrate (44) and may reflect the known difficulty of transmitting scrapie to cattle (45).
sPMCA of the classical scrapie agent in the three ovine substrates served as a positive control in our experiments, and we saw conservation of the expected molecular banding pattern of typical scrapie PrP res . These results are in agreement with previous reports of efficient amplification when the donor seed and the host substrate are matched for the amino acid at position 136 (46) and more importantly with observations of scrapie agent propagation in vivo (13).
Our modeling of the sheep BSE agent amplification in ovine substrate is broadly consistent with experimental in vivo observations in which sheep of the ARQ/ARQ genotype propagated the BSE agent for several passages (27). Our findings are also in agreement with recently published observations assessing species and ovine polymorphic barriers to BSE, including the efficiency of amplification of the BSE agent in ovine ARQ/ARQ and VRQ/VRQ substrates by a single round PMCA (40). However, we do note a discrepancy between our and others' PMCA data (40) and the experimental in vivo data concerning BSE and the ARR/ARR genotype that show that sheep of the ARR/ARR genotype are susceptible to experimental BSE infection (19,31). The explanation for this difference between the in vitro and in FIGURE 9. Correction for low PrP C level in the ARR/ARR ovine brain PMCA substrate. A, Western blot analysis of 20% ovine ARR/ARR substrate (doubled substrate concentration to match the PrP C level of the 10% VRQ/VRQ substrate) seeded with sheep BSE VRQ/VRQ seed and subjected to four rounds (r1-r4) of sPMCA. C, the 5% VRQ/VRQ ovine substrate (substrate concentration diluted to match the PrP C level of the 10% ARR/ARR substrate) seeded with sheep BSE VRQ/VRQ seed and subjected to four rounds of sPMCA. E, the 10% ARR/ARR ovine substrate seeded with sheep BSE VRQ/VRQ seed and subjected to four rounds of sPMCA (r1-r4). G, the 10% VRQ/VRQ ovine substrate seeded with sheep BSE VRQ/VRQ seed and subjected to four rounds of sPMCA. A, C, E, and G, samples were analyzed after PK treatment using anti-PrP mAb 6H4 (left) and 12B2 (right). Ϫ, corresponding controls not subjected to PMCA. The molecular mass (in kDa) is marked to the left of each blot. B, D, F, and H, graphical representation of semiquantitative densitometric analysis of PrP res amplification efficiency in sPMCA shown in A, C, E, and G, respectively. Western blots were probed with anti-PrP mAbs 6H4 (blue bars) and 12B2 (red bars) and analyzed by densitometry. The data were background-subtracted.
vivo results is not clear at present, but we have investigated some of the potential causes. We discount proteolytic processing of ovine PrP C as an explanation of genotype-dependent amplification potential (47,48). We found that the ARR/ARR substrates had half the amount of PrP C found in the VRQ/VRQ and ARQ/ARQ substrates, but doubling the concentration of the ARR/ARR substrate did not result in amplification. It therefore seems likely that the failure of the ARR/ARR substrate to support sustained amplification of sheep BSE PrP res is due to a genuine relative molecular barrier to conversion associated with the substitution of arginine for glutamine at position 171 ( Table 1) but one that can be overcome in the more complex process of infection and disease progression in vivo.
Another major objective of our study was to use PMCA to address molecular conversion barriers between potentially zoonotic animal diseases and humans. We have shown previously that brain samples from vCJD (in humans), BSE in cattle, and experimental sheep BSE (ARQ/ARQ genotype) can convert codon 129 MM human PrP C in PMCA reactions, whereas ARQ/ARQ scrapie does not do so to a measurable extent (38). This suggests that the known zoonotic potential of the BSE agent is recapitulated in this cell-free prion protein conversion assay irrespective of the host species (cow, sheep, or human). We have also shown that classical BSE cattle brain homogenates have a greater potential to convert human PrP C in PMCA than any other tested animal prion disease (39). Here we show that samples of experimental sheep BSE isolates of the VRQ/ VRQ, ARQ/ARQ, and ARR/ARR genotypes are all able to convert human PrP C during a single round of PMCA and that the preference is for the PRNP codon 129 M over the V allele (Table  1). We also show that there is an association between the PrP res type present and the ability to convert human PrP C : when sheep BSE of any genotype is amplified in an ARQ/ARQ substrate it retains its molecular signature and its ability to convert human PrP C , whereas if the same starting seeds are amplified in a VRQ/ VRQ substrate the PrP res adapts to a scrapie-like signature and loses its ability to convert human PrP C . This provides a more comprehensive data set than previous transgenic animal-based FIGURE 10. Relative levels of full-length and C1 fragment of PrP C present in brain substrates of the ovine genotypes. A, representative Western blot of VRQ/VRQ (lanes 1 and 2), ARQ/ARQ (lanes 3 and 4), and ARR/ARR (lanes 5 and 6) ovine brain substrates. Samples in lanes 1, 3, and 5 show the level of PrP C in the crude brain homogenates. Corresponding samples in lanes 2, 4, and 6 were treated with peptide-N-glycosidase F (PNGase F); the top band corresponds to full-length PrP C , and the lower band corresponds to the C1 fragment. The Western blot was probed using anti-PrP mAb 6H4. The molecular mass (in kDa) is marked to the left of the blot. B, graphical representation of densitometric analysis of ovine substrates treated with peptide-N-glycosidase F shown in A (lanes 2, 4, and 6). The positions of full-length PrP C and the C1 fragment are shown. FIGURE 11. Western blot analysis of human brain tissue seeded with experimental sheep BSE, cattle BSE, and natural scrapie and subjected to a single round of PMCA. A, human brain tissue (PRNP codon 129 MM) homogenate seeded with PrP res from experimental sheep BSE (VRQ/VRQ, ARQ/ARQ, and ARR/ARR), cattle BSE, and classical scrapie. B, human brain tissue (PRNP codon 129 VV) homogenate seeded as above. Ϫ, corresponding control of each sample not subjected to PMCA; ϩ, samples that were subjected to a single round PMCA for 48 h. Unseeded (Ϫ ve) control and positive control (human MM seeded with vCJD and human VV seeded with sCJD) are shown. Samples were analyzed by Western blotting after PK treatment using the anti-PrP mAb 3F4. Dotted lines indicate individual montaged Western blot results. studies (49,50) and points to a decisive role for the interaction of conformation-enciphered agent strain and host genotype in determining prion replication.
Extrapolation from our results using the serial PMCA method would suggest that if BSE were to infect sheep in the field the molecular phenotype of sheep BSE agent might, through time and in certain sheep genotypes, come to resemble scrapie; however, if this were the case, then there may be an associated increase in the molecular barrier of disease transmission to humans. The concepts of Darwinian evolution are increasingly being applied to the epigenetic molecular evolution of prions and amyloids, and the subject is becoming experimentally tractable (51,52). The accelerated in vitro molecular evolution of BSE prions in sheep shown here strongly indicates that both replicative efficiency and zoonotic potential may be determined in a direct manner by a combination of the PrP Sc conformation associated with the agent and aspects of host prion protein genetics that determine key polymorphic residues in the prion protein sequence.

TABLE 1
The relevant polymorphic residues in the human and ovine prion protein, and the amino acids present at equivalent position in the human, ovine, and bovine proteins The MARX genotype of cattle, humans, and sheep that is permissive to in vitro replication and associated with BSE PrP res type fidelity is shown in blue. The Val-136 substitution in sheep that is permissive to in vitro replication but is associated with a switch in PrP res type is shown in red. The Arg-171 and Val-129 residues in sheep and humans, respectively, that are associated with poor conversion in vitro are shown in green.