Ultra-efficient Replication of Infectious Prions by Automated Protein Misfolding Cyclic Amplification*

Prions are the unconventional infectious agents responsible for transmissible spongiform encephalopathies, which appear to be composed mainly or exclusively of the misfolded prion protein (PrPSc). Prion replication involves the conversion of the normal prion protein (PrPC) into the misfolded isoform, catalyzed by tiny quantities of PrPSc present in the infectious material. We have recently developed the protein misfolding cyclic amplification (PMCA) technology to sustain the autocatalytic replication of infectious prions in vitro. Here we show that PMCA enables the specific and reproducible amplification of exceptionally minute quantities of PrPSc. Indeed, after seven rounds of PMCA, we were able to generate large amounts of PrPSc starting from a 1 × 10-12 dilution of scrapie hamster brain, which contains the equivalent of ∼26 molecules of protein monomers. According to recent data, this quantity is similar to the minimum number of molecules present in a single particle of infectious PrPSc, indicating that PMCA may enable detection of as little as one oligomeric PrPSc infectious particle. Interestingly, the in vitro generated PrPSc was infectious when injected in wild-type hamsters, producing a disease identical to the one generated by inoculation of the brain infectious material. The unprecedented amplification efficiency of PMCA leads to a several billion-fold increase of sensitivity for PrPSc detection as compared with standard tests used to screen prion-infected cattle and at least 4000 times more sensitivity than the animal bioassay. Therefore, PMCA offers great promise for the development of highly sensitive, specific, and early diagnosis of transmissible spongiform encephalopathy and to further understand the molecular basis of prion propagation.

Prion diseases or transmissible spongiform encephalopathies (TSEs) 2 are neurodegenerative disorders of humans and animals. Historically, scrapie has been the most common TSE in animals, affecting sheep for over 200 years (1). The most recent and worrisome outbreak of an animal TSE disease is bovine spongiform encephalopathy (BSE) in cattle (2). BSE has important implications for human health, because the infectious agent can be transmitted to humans producing a new disease, termed variant Creutzfeldt-Jakob disease (3,4). TSEs are characterized by an extremely long incubation period, followed by a brief and invariably fatal clinical disease (5). To date no therapy or early diagnosis is available.
The pathogen responsible for TSEs, called "prion," is comprised mainly or exclusively of a misfolded protein named PrP Sc , which is a post-translationally modified version of the normal prion protein (PrP C ) (6,7). During the course of the disease, prions replicate by the autocatalytic conversion of PrP C into PrP Sc , triggered by the misfolded protein present in the infectious inoculum. The conversion seems to involve a conformational change whereby the ␣-helical content of the normal protein diminishes and the amount of ␤-sheet increases (8,9). The structural changes are accompanied by alterations in the biochemical properties as follows: PrP C is soluble in nondenaturing detergents; PrP Sc is insoluble; PrP C is readily digested by proteases; and PrP Sc is partially resistant (7,10).
To understand the mechanism of prion conversion, Caughey and co-workers (11) developed a cell-free conversion reaction that mimics in many aspects the prion replication process. One of the limitations of the cell-free conversion method is the relatively low efficiency of PrP Sc formation that diminishes its application to study the nature of the infectious agent and to attempt sensitive detection of the protein. More recently, we have developed a novel technique referred to as protein misfolding cyclic amplification (PMCA), in which it is possible to simulate prion replication in the test tube in an accelerated and efficient way (12). In a cyclic manner, conceptually analogous to PCR cycling, PrP Sc is incubated with excess PrP C to enlarge the PrP Sc aggregates, which are then sonicated to generate multiple smaller units for the continued formation of new PrP Sc (13). We have reported previously proof-of-concept experiments in which the technology was applied to replicate the misfolded protein from diverse species (12,14). The newly generated protein exhibits the same biochemical, biological, and structural properties as brain-derived PrP Sc , and strikingly, it is infectious to wild-type animals, producing a disease with characteristics that are identical to the illness produced by brain-isolated prions (15). The technology has been automated, leading to a dramatic increase in efficiency of amplification and its application to detect PrP Sc in blood of hamsters experimentally infected with scrapie (16).
In this study we describe in detail the methodological aspects for efficient PMCA amplification and the characterization of the technology for reproducibility, specificity, and sensitivity. We also study the minimum number of molecules needed to trigger amplification and the application of the PMCA technique to generate infectious material in vitro. Our results demonstrate that PMCA is capable of detecting as little as ϳ26 monomers of PrP, which, according to recent data on the minimal size of the infectious particle (17), would correspond to a single molecule of oligomeric infectious PrP Sc . The technique is highly reproducible and specific to amplify PrP Sc , because no signal is detected when no PrP Sc inoculum is present. Finally, our results show that PMCA enables the increase of infectivity by around 20 million-fold, converting a sample that is not infectious into a highly infectious one. These data demonstrate that PMCA has a similar power of amplification as PCR techniques used to amplify DNA and have great promise for the development of highly sensitive detection of PrP Sc and for understanding the molecular basis of prion replication.

EXPERIMENTAL PROCEDURES
Preparation of Tissue Homogenates-Healthy and sick animals were perfused with phosphate-buffered saline (PBS) plus 5 mM EDTA prior to harvesting the tissue. Ten percent brain homogenates (w/v) were prepared in conversion buffer (PBS containing 150 mM NaCl, 1.0% Triton X-100, 4 mM EDTA, and the Complete Protease Inhibitor Mixture from Roche Applied Science). The samples were clarified by a brief, low speed centrifugation (1500 rpm for 30 s) using an Eppendorf centrifuge, model 5414 (Hamburg, Germany). Dilutions of this brain homogenate were done in conversion buffer, and they are expressed in relation to the brain; for example, a 100-fold dilution is equivalent to a 1% brain homogenate.
In Vivo Infectivity Studies-Syrian Golden hamsters were used as an experimental model of scrapie. Animals were 4 -6weeks old at the time of inoculation. Anesthetized animals were injected intracerebrally stereotaxically in the right hippocampus with 1 l of the sample or intraperitoneally with 200 l of sample as described previously (15). The onset of clinical disease was measured by scoring the animals twice a week using the following scale: 1, normal animal; 2, mild behavioral abnormalities, including hyperactivity and hypersensitivity to noise; 3, moderate behavioral problems, including tremor of the head, ataxia, wobbling gait, head bobbing, irritability, and aggressiveness; 4, severe behavioral abnormalities, including all of the above plus jerks of the head and body and spontaneous backrolls; 5, terminal stage of the disease in which the animal lies in the cage and is no longer able to stand up. Animals scoring level 4 during 2 consecutive weeks were considered sick and were sacrificed to avoid excessive pain using exposition to carbonic dioxide. The scrapie infectious material used in these studies was titrated and 1 LD 50 was obtained in a brain dilution of ϳ1 ϫ 10 Ϫ9 .
Brains were extracted and analyzed biochemically and histologically. The right cerebral hemisphere was frozen and stored at Ϫ70°C for biochemical examination of PrP res using Western blot analysis as described below. The left hemisphere was fixed in 10% formaldehyde solution, cut into sections, and embedded in paraffin. Serial sections (6 m thick) from each block were stained with hematoxylin-eosin, using standard protocols, or incubated with monoclonal antibodies recognizing PrP or the glial fibrillary acidic protein. Immunoreactions were developed using the peroxidase-antiperoxidase method, following the manufacturer's specifications. Antibody specificity was verified by absorption. To compare the pattern of histopathological damage among animals, we calculated the lesion profile, using a variation of the method employed in mice (18). Briefly, the severity of vacuolation was scored in a scale from 0 to 5 in seven different brain areas, including medulla, cerebellum, superior colliculus, hippocampus, and cerebral cortex occipital, frontal, and lateral.
PMCA Procedure-Although the principle of PMCA remains the same as in our original publication (12), the system has been optimized and automated, thus enabling the routine processing of many more samples in the same amount of time while reaching a higher conversion efficiency. Aliquots of normal and scrapie brain homogenate prepared in conversion buffer were mixed and loaded onto 0.2-ml PCR tubes. Tubes were positioned on an adaptor placed on the plate holder of a microsonicator (Misonix model 3000, Farmingdale, NY) and programmed to perform cycles of 30 min of incubation at 37°C followed by a 40-s pulse of sonication set at 60% potency. Samples were incubated without shaking and immersed in the water of the sonicator bath. and the entire microplate horn was kept inside an incubator at 37°C. The detailed protocol, including troubleshooting, has been recently published elsewhere (19 -21).
Proteinase K Digestion-Samples were incubated with 50 g/ml PK for 60 min at 45°C. The digestion was stopped by adding electrophoresis sample buffer. In each experiment it is important to have a negative control consisting of normal brain homogenate and to check that PK digestion of PrP C is complete to avoid confusion between undigested PrP C and the signal from the PK-resistant core of PrP Sc . For this purpose special care has to be taken to observe the switch in the molecular weight after PK digestion, characteristic of bona fide PrP Sc .
Western Blot-Proteins were fractionated by SDS-PAGE under reducing conditions, electroblotted into nitrocellulose membrane, and probed with 3F4 antibody (22) (Signet, Dedham, MA) diluted 1:5000 in PBS. The immunoreactive bands were visualized by enhanced chemiluminescence assay (Amersham Biosciences). Densitometric analysis was done by using a UVP Bioimaging system EC3 apparatus (Upland, CA). Statistical significance of the values was evaluated by Student's t test.
ELISA-Twenty l of serial dilutions of an infected hamster brain into 10% normal brain homogenate were treated with 50 g/ml PK for 1 h at 45°C and 450 rpm. The digestion was stopped with 50 mM phenylmethylsulfonyl fluoride. The digested samples were mixed with PBS up to 100 l, loaded onto a 96-well ELISA plate (Maxisorp surface treatment), and incubated for 1 h at 37°C. The plates were blocked with 5% BSA in PBS for 1 h at 37°C, followed by incubation for 1 h with the monoclonal antibody 3F4 (Signet, Dedham, MA) diluted 1:3000 in 1% BSA in PBS. The antibody was washed four times with 200 l of PBS, 0.05% Tween 20, and samples were incubated with the polyclonal antibody anti-mouse IgG, conjugated with horseradish peroxidase (Amersham Biosciences), and diluted 1:1000 in 1% BSA in PBS. After four washes the samples were developed with ImmunoPure ABTS (Pierce) following the manufacturer's directions.
PTA Precipitation-Twenty l of serial dilutions of an infected hamster brain into 10% normal brain homogenate were first treated with benzonase (2.5 units/l in PBS, 1 mM MgCl 2 ) for 30 min at 37°C with constant agitation. Thereafter, we added 4% PTA (prepared in 70 mM MgCl 2 , titrated with NaOH to pH 7.4) and incubated at 37°C for 30 min with constant agitation. Samples were centrifuged at 15,800 ϫ g for 30 min at 37°C in an Eppendorf microcentrifuge. Pellet was resuspended in 20 -30 l of PBS containing 0.1% Sarkosyl. After treatment with PK, samples were analyzed by Western blot as described above.
PrP Sc Quantitation-To estimate the quantity and the number of molecules of PrP Sc in our samples, we analyzed several dilutions of scrapie brain homogenate by Western blot in the same gel as aliquots of known amounts of recombinant hamster PrP (supplemental Fig. 1A). The signal intensity was evaluated by densitometry, and the quantity of PrP in the sample was estimated by extrapolation of the calibration curve prepared with recombinant PrP (supplemental Fig. 1B). To minimize artifacts because of saturated or weak signal, several different dilutions were measured, and each dilution was analyzed in triplicate. To standardize the signal among the different blots, the densitometric data were expressed relative to the value of the signal of the same quantity of normal brain homogenate (without PK treatment). PrP Sc quantitation was also confirmed by ELISA using recombinant PrP as standard. In this way we estimated the average concentration of PrP Sc in the scrapie brain used in our studies to be 67 ng/l. The number of molecules of PrP Sc detected was estimated by mathematical calculation of the dilution used and the known concentration of PrP Sc in the brain homogenate.

RESULTS
The PMCA Technology and Its Reproducibility-PMCA consists of cycles of accelerated prion replication. The basis for PMCA is the observation that prion replication follows a seeding-nucleation model in which oligomeric PrP Sc in the infectious material converts PrP C by integrating monomeric proteins into the ends of the aggregate, inducing and stabilizing its misfolding (13,23). PMCA is a cyclic process consisting of two phases. During the first phase the samples containing minute amounts of PrP Sc and a large excess of PrP C were incubated to induce growing of PrP Sc polymers. In the second phase the sample is sonicated to break down the polymers, multiplying the number of nuclei. In this way, after each cycle the number of "seeds" is increased in an exponential fashion. The number of cycles can be repeated as many times as needed to reach the amplification rate desired. For practical operation, the system has been automated by using a programmable plate sonicator, as described under "Experimental Procedures." This improvement not only decreases processing time and allows for the routine processing of many more cycles than a single probe sonicator but also prevents loss of material. Cross-contamination is eliminated because there is no direct probe intrusion into the sample.
Reproducibility of amplification was measured by monitoring the PrP Sc signal obtained before and after PMCA cycling under different experimental conditions. Equivalent samples containing a 10,000-fold dilution of scrapie brain into 10% healthy hamster brain homogenate were placed in distinct positions of the microplate sonicator and subjected to 48 PMCA cycles. The levels of amplification were studied by Western blot after PK digestion (Fig. 1A), and data were quantitated by densitometric analysis of the Western blot signal (Fig. 1B). Although some small variability was observed on the signal obtained in distinct wells, the differences were not significant and could not be attributed to a position effect, but rather they probably reflect some small experimental variability.
To analyze further the reproducibility of the procedure, equivalent samples containing a 10,000-fold dilution of scrapie brain homogenate into 10% healthy hamster brain homogenate were subjected to 48 PMCA cycles in experiments done on different days. Fig. 1C shows that at 7 distinct days the amplification efficiency was virtually the same. Again, densitometric analysis showed that the signal was not statistically different in the distinct experiments (Fig. 1D). For this experiment, the normal brain sample used on day 1 was freshly prepared, whereas for all the other days, frozen material was used. Therefore, no differences between fresh and frozen material were observed. However, it is important to note that brain substrate has to be frozen in aliquots to avoid repetitive freezing-thawing that decreases PMCA efficiency.
The influence of different, but equivalent, inocula on the conversion efficiency was studied by amplifying preparations of 10,000-fold diluted scrapie brain homogenate obtained from five distinct hamsters into the same substrate (Fig. 1E). After 48 PMCA cycles, a large and similar conversion of PrP C into PrP Sc was observed. A similar result was obtained when normal brain homogenate from five different hamsters was used as a substrate for the amplification of a unique PrP Sc inoculum (Fig.  1G). However, densitometric analysis of the experiments shown in Fig. 1, E and G, indicate that in both cases one sample gave a statistically significant different level of amplification than the other four samples (Fig. 1, F and H, respectively). These differences are not because of changes in the levels of PrP Sc or PrP C in the samples, which were not significantly different. These results suggest that perhaps individual variability on the expression of PrP or conversion factors may lead to changes on the extent of prion conversion in vitro. In none of the experiments shown in Fig. 1 was PrP Sc detectable in samples containing the same material but kept frozen without amplification.
Specificity of PMCA Amplification-Specificity of cyclic amplification was evaluated in a blind study in which 10 brain samples of scrapie-affected hamsters and 11 samples of healthy animals were subjected to 48 PMCA cycles, and PrP Sc was detected by Western blot analysis after PK digestion (Fig. 2, A  and B). The results show that although 100% of the samples derived from sick animals (Fig. 2, A and B, lanes 1, 2, 3, 5, 7, 11, NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 13, 17, 19, and 21) were positive after PMCA, none of the samples coming from normal animals (lanes 4, 6, 8, 9, 10, 12, 14, 15, 16, 18, and 20) show any significant PrP Sc signal. Of the 10 positive control samples, 7 corresponded to a 10,000-fold dilution of brain, 2 corresponded to a 50,000-fold dilution ( Fig. 2A,  lanes 13 and 17 in), and 1 corresponded to a 100,000-fold dilution ( Fig. 2A, lane 19). None of these 10 samples showed any PrP Sc signal in Western blot without PMCA amplification (data not shown). The interpretation of the data is that, under the conditions used, PMCA leads to 100% specificity for PrP Sc detection.

In Vitro Replication of Infectious Prions
As demonstrated before, the amplification rate using PMCA depends upon the number of incubation/sonication cycles carried out (12). Thus, we decided to evaluate whether a PrP Sc -like signal might appear on negative samples after many PMCA cycles. For this purpose, a 10% healthy hamster brain homogenate in the absence (negative control) or in the presence (positive control) of an aliquot of a 50,000-fold diluted scrapie brain was subjected to 24, 48, 96, or 144 PMCA cycles, and the PrP Sc signal was detected by Western blot analysis (Fig. 2C). The results clearly indicate that PrP Sc reactivity was detected only after PMCA in the positive control samples with an intensity that depended upon the number of cycles performed. In comparison, in the negative control samples, no PrP Sc was ever detected, regardless of the number of PMCA cycles carried out (Fig. 2C).
Specificity was further studied in an even more challenging situation in which several rounds of PMCA were done after FIGURE 1. Reproducibility of automated PMCA. A, samples containing a 10,000-fold dilution of 263K scrapie hamster brain prepared in 10% normal hamster brain were either immediately frozen (F) or placed in 10 different positions of the plate holder and subjected to 48 PMCA cycles. Thereafter, samples were treated with PK and PrP Sc reactivity analyzed by Western blotting. B, densitometric analysis of three different blots obtained using the samples described in A. C, equivalent samples prepared as described in A were subjected to 48 PMCA cycles performed at seven different times, and the extent of PrP Sc formation was evaluated by Western blotting. D, densitometric analysis of three different blots obtained using the samples described in C. E, five different scrapie hamster brains (I1 through I5) were diluted 10,000-fold into the same 10% normal hamster brain homogenate and subjected to 48 PMCA cycles. PrP Sc signal was detected by Western blotting after PK treatment. F, densitometric analysis of three different blots obtained using the samples described in E. G, a single scrapie brain homogenate was diluted 10,000-fold into 10% solutions prepared from five different normal hamster brains (S1 through S5). Again 48 PMCA cycles were performed and PrP Sc detected by Western blotting. H, densitometric analysis of three different blots obtained using the samples described in G. F, frozen samples; A, amplified samples. NBH, normal brain homogenate. All samples were treated with PK before electrophoresis, except those in which ϪPK is indicated. Data were statistically analyzed by one-way analysis of variance and Student's t test. * p Ͻ 0.05.
diluting the material to refresh the substrate. We have shown that PrP Sc can be kept replicating indefinitely in vitro by several rounds of successive PMCA (15) and that serial PMCA enables the ultrasensitive detection of PrP Sc in brain and blood samples (16). Brains from healthy hamsters and from animals infected with the 263K scrapie strain were diluted 10 4 -fold into a 10% normal hamster brain homogenate. Samples were subjected to 48 PMCA cycles. After this first round of PMCA, a small aliquot of the amplified samples was taken and diluted 10-fold into more normal brain homogenate. These samples were again amplified by 48 PMCA cycles. This procedure was repeated several times, and PrP Sc generation was determined by Western blot analysis after PK digestion. As shown in Fig. 2D, 10 rounds of PMCA to reach a final dilution of the original brain equivalent to 10 Ϫ13 led to continuous formation of PrP Sc only when the initial inoculum was derived from scrapie-infected animals. No PrP Sc was ever detected in the absence of PrP Sc inoculum, indicating that the system retains high specificity, even after 480 PMCA cycles. These same samples were used for further amplification, and after a dilution of more than 10 Ϫ63 of initial inoculum a robust and continuous amplification was observed in samples containing PrP Sc . When many additional rounds of saPMCA were performed, we observed a scatter appearance of a protease-resistant band identical to PrP Sc even in samples without PrP Sc inoculum (data not shown). Spontaneous gener-ation of PrP Sc was seen mainly when more than 10 rounds of PMCA were done. At present we do not know whether this newly generated PrP Sc is the result of cross-contamination or the de novo generation of PrP Sc . More experiments need to be performed to analyze the reproducibility of this observation and to distinguish between these two possibilities. In case we ruled out the possibility of contamination, these findings would indicate that PMCA can amplify a low frequency event in which PrP C spontaneously converts into PrP Sc . This could be the basis for the origin of sporadic prion diseases.
Sensitivity of PMCA and Minimum Number of Molecules Detected after Amplification-We have recently reported that sensitivity of detection after 140 cycles of PMCA was increased by around 6600-fold (16). We also reported previously that to further increase replication efficiency, we needed to refresh the substrate periodically using a methodology we termed serial automated PMCA (saPMCA) (16). By performing two rounds of saPMCA, we were able to detect PrP Sc up to a 5 ϫ 10 Ϫ10 dilution of scrapie hamster brain, to reach an increase of sensitivity of around 10 million-fold (16). To estimate the minimum number of molecules of PrP Sc that our technology can detect in a given sample, we diluted a scrapie brain homogenate 1 ϫ 10 Ϫ12 -fold into conversion buffer and subjected this material to saPMCA. According to our estimations, the PrP Sc concentration in the scrapie-infected brain used for these studies was ϳ67 ng/l (see "Experimental Procedures" and supplemental Fig. 1). This result indicates that a 1 ϫ 10 Ϫ12 -fold dilution should contain ϳ6.7 ϫ 10 Ϫ20 g/l or 1.3 molecules of PrP Sc monomer per l. Because in our experiments we use a volume of 20 l, the sample tested contains ϳ26 molecules of monomeric PrP Sc . Strikingly, after five rounds of saPMCA, we were able to detect a signal in one of the four replicates used, and after seven rounds of amplification, we detected a signal in three of the four replicates (Fig. 3). Importantly, no amplified product was detected when a 10 Ϫ14fold dilution of brain was used (a sample that should contain no molecules of PrP Sc ) or in any of the control samples in which no PrP Sc was present (Fig. 3). No signal was detected either in a 10 Ϫ13 -fold dilution (data not shown).
The serious consequences of the BSE epidemics and the increasing concern regarding the iatrogenic transmission of variant Creutzfeldt-Jakob disease have motivated the development of several biochemical methods to detect PrP Sc . Several tests have been approved by the European community and are A, blind study in which aliquots from 11 different normal brain samples and 10 distinct diluted scrapie-infected brain samples were subjected to 48 PMCA cycles to attempt detection of PrP Sc . For all scrapie-infected samples, except for the ones in lanes 13, 17, and 19, the brain was diluted 10,000-fold into 10% normal brain homogenate. For lanes 13 and 17, the inoculum was diluted 50,000 and for lane 19, the scrapie brain was diluted 100,000-fold. After amplification, the samples were treated with PK and analyzed by Western blotting. B, densitometric analysis of three different blots from the samples shown in A. C, samples from healthy (denoted ϪPrP Sc inoculum) or scrapie-infected (denoted ϩPrP Sc inoculum) brain were diluted 50,000 into 10% normal brain homogenate and subjected to 0, 24, 48, 96, or 144 PMCA cycles, and PrP Sc signal was analyzed by Western blotting. D, brains from healthy hamsters or from animals infected with 263K scrapie were diluted 10 4 -fold into a 10% normal hamster brain homogenate. Samples were subjected to 48 PMCA cycles. The amplified material was diluted 10-fold into normal brain homogenate and amplified again. This procedure was repeated several times to reach a 10 Ϫ13 dilution of initial material. The figure shows the PrP Sc signal obtained in Western blots of only some of the dilutions, representative of all the results obtained. NBH, normal brain homogenate. All samples were treated with PK before electrophoresis, except those in which ϪPK is indicated. NOVEMBER 17, 2006 • VOLUME 281 • NUMBER 46 widely used in BSE surveillance in several countries (24,25). All these tests consist of the immunological detection of PrP Sc either by Western blot (26), ELISA (27), or the conformational dependent immunoassay (28). To compare the sensitivity of these standard tests with PMCA detection, we performed studies in parallel with different methods using the same samples, all prepared in 10% normal brain homogenate to facilitate the comparison (Table 1). Western blot is the most standard but least sensitive of these tests, allowing for the detection of a minimum of 4.0 ng of PrP Sc in 20 l of sample, which is equivalent to 8 ϫ 10 10 molecules of misfolded protein. Simple ELISA was 8-fold more sensitive than Western blotting (Table 1). One strategy that has been used to enhance PrP Sc detection is the specific precipitation and concentration of the protein using PTA (29,30). Under our experimental conditions, PTA precipitation of PrP Sc from scrapie brain led to a 50-fold increase in detection (Table 1). According to literature reports, conformational dependent immunoassay detects PrP Sc up to a maximum dilution of hamster scrapie brain homogenate equivalent to 5 ϫ 10 Ϫ5 (29), indicating that it is 60-fold more sensitive than Western blotting (Table 1). By comparison, one round of 100 PMCA cycles resulted in an average of 2500-fold more sensitive detection of PrP Sc as compared with Western blotting. This sensitivity threshold indicates that one round of PMCA can detect as little as 3.2 ϫ 10 7 molecules of PrP Sc in a 20-l sample volume (Table 1). On the other hand, two rounds of 100 PMCA cycles were able to systematically detect PrP Sc up to a maximum dilution of the scrapie brain equivalent to 5 ϫ 10 Ϫ10 , indicating a sensitivity 6 million times higher than standard Western blotting (Table 1). In other words, two rounds of 100 PMCA cycles can detect as little as 13,000 molecules of PrP Sc . As shown before, seven successive rounds of PMCA cycles produce a signal after amplification even when the starting material was a 1 ϫ 10 Ϫ12 -fold dilution of sick brain homogenate. This amplification leads to an increase of sensitivity of 3 billion times with respect to standard Western blot (Table 1). Until now, the animal bioassay of infectivity was by far the most sensitive assay available for detection of prions (31,32). Among the animal bioassays, hamsters infected with the 263K scrapie strain are the most rapid and sensitive, because animals can be infected with the lowest quantity of infectious agent, and disease symptoms are observed at the shortest time after inoculation (33). Indeed, a 1 ϫ 10 Ϫ9 dilution of sick brain is the minimum amount that can still produce disease in 50% of the animals (mean lethal dose or LD 50 ). In our experiments the minimum dilution that produced disease in all animals was 4 ϫ 10 Ϫ9 , indicating that the bioassay can detect as little as 107,000 molecules of misfolded protein, which represent a 725,000-fold higher sensitivity than Western blotting (Table 1). Remarkably, our findings with saPMCA using the same samples as for the FIGURE 3. Minimum quantity of PrP Sc detected by saPMCA. Aliquots of scrapie hamster brain homogenate were serially diluted into conversion buffer to reach 1 ϫ 10 Ϫ12 and 1 ϫ 10 Ϫ14 dilutions. Four aliquots of 20 l of each dilution were mixed in four separated tubes with 80 l of normal brain homogenate and subjected to 144 PMCA cycles. Thereafter, a volume of 20 l was used for Western blotting after PK digestion, and 10 l were diluted into 90 l of normal brain homogenate, and the samples were subjected to a second round of 144 PMCA cycles. The procedure was repeated several times to reach seven successive rounds of PMCA. S1, S2, S3, and S4 correspond to the four replicated tubes in each dilution. As a negative control, we used normal brain homogenate diluted 10 Ϫ12 into conversion buffer and subjected to the same scheme of saPMCA. This experiment was also done in four replicated tubes, and C1, C2, C3, and C4 represent each result. infectivity studies demonstrate that two and seven rounds of saPMCA are Ͼ8 and Ͼ4000 times more sensitive than the most efficient animal bioassay, respectively ( Table 1). Generation of Infectivity in Vitro from Sub-infectious Quantities of PrP Sc -As described in Fig. 3, we were able to generate large quantities of PrP Sc from a very high dilution (1 ϫ 10 Ϫ12 ) of hamster scrapie brain. Mathematical calculation estimates that such dilution contains the equivalent of ϳ26 molecules of PrP monomers, which are at least 3-4 orders of magnitude below the minimum quantity of PrP Sc needed to produce infectivity in some animals at very long times after inoculation. To study whether the in vitro generation of PrP Sc was associated with an increase of infectivity, we inoculated intraperitoneally the sample obtained after seven rounds of amplification starting with 1 ϫ 10 Ϫ12 dilution (Fig. 3, sample S 1 ). All six wild-type hamsters inoculated showed typical signs of 263K scrapie disease and were sacrificed at an average of 299.6 Ϯ 20.5 days post-inoculation. In our experience with this animal model, this incubation time is similar to that obtained with a 5 ϫ 10 Ϫ4 dilution of scrapie infected brain inoculated by the intraperitoneal route. Therefore, seven rounds of PMCA produced a 2 ϫ 10 7 increase of infectivity. As expected none of the control animals inoculated with the same dilution without amplification developed the disease even after 500 days post-inoculation.

In Vitro Replication of Infectious Prions
The clinical signs observed in animals inoculated with the in vitro amplified samples were identical to those of the animals treated with infectious brain material and included hyperactivity, motor impairments, head wobbling, muscle weakness, and weight loss. Brain samples from these animals contain a large quantity of protease-resistant PrP Sc , which has an identical glycosylation profile to the protein observed in brain-inoculated animals (Fig. 4A). Conversely, no protease-resistant protein was detected in the brain of negative control animals. Histological analysis showed typical brain spongiform degeneration, PrP accumulation, and astrogliosis (Fig. 4B). No differences in the lesion pattern profile were observed compared with animals inoculated by brain infectious material. These findings suggest that PrP Sc generated in vitro corresponds to the same strain of 263K used to trigger amplification.

DISCUSSION
Our results indicate that PMCA is able to amplify exceptionally small quantities of PrP Sc in a very specific and reproducible manner. Indeed, by using several successive rounds of PMCA, our data demonstrate that we are able to induce the conversion of PrP C with as little as the equivalent to ϳ26 molecules of PrP Sc monomers. Recent data have shown that the minimum size of the particle able to sustain infectivity and to induce the cell-free conversion of PrP C into PrP Sc contains between 14 and 28 molecules of PrP monomers (17). Taken together, this suggests that saPMCA can amplify a single particle of oligomeric infectious PrP Sc . This unprecedented amplification efficiency is comparable only to the effectiveness of PCR amplification of DNA (34). Moreover, although at these levels of amplification PCR often results in artifactual amplification products, we rarely see a false-positive using PMCA.
The applications of such a powerful amplification technology are many and impact various fields. A particularly important application is on the development of a highly sensitive biochemical detection of PrP Sc , which constitutes the best surrogate marker for TSE diagnosis (24). A comparison of saPMCA with the sensitivity of some of the current methodologies used for PrP Sc detection (Table 1) showed that our technology is several millions or even billions of times more sensitive than the standard Western blot or ELISA-based assays currently used to detect prions in cattle. At present, it is not possible to diagnose the disease in the pre-clinical phase or in live individuals using body biological fluids (24). The successful implementation of saPMCA should lead to the identification of individuals and tissue samples infected with even the tiniest quantities of PrP Sc , enabling us to decrease the risk of further propagation of TSE to the minimum. Indeed, we have recently reported for the first time the biochemical detection of PrP Sc in blood samples using saPMCA both at the symptomatic (16) and pre-symptomatic stages of the disease (35). Although technically more challenging, the PMCA technology has been adapted to amplify prions from a variety of origins, including human (14). One of the major limitations in the case of human samples is the availability of normal brain tissue to use as a substrate of PMCA, but this can be overcome by using transgenic mice brain or cell lysates.
Another important application of PMCA is to understand the underlying biology of prions and the nature of the infectious agent. In this study we show that it is possible to generate infectivity by amplification of a scrapie brain sample diluted 10 Ϫ12fold, containing an estimated single particle of oligomeric PrP Sc . By successive in vitro replication of this single particle of PrP Sc , we have successfully generated many millions of PrP Sc A, samples of brain homogenate from animals inoculated with PrP Sc amplified from a 1 ϫ 10 Ϫ12 dilution of scrapie brain (line 4), with brain infectious agent (line 3), or with control normal brain homogenate (line 2) were treated with 50 g/ml PK for 60 min at 45°C. PrP signal was detected by Western blot using the 3F4 antibody. Line 1 corresponds to the control normal brain homogenate without PK treatment. B, histological evaluation of brain sections from an animal inoculated with in vitro amplified PrP Sc showed the typical spongiform brain degeneration after staining with hematoxylin-eosin (left panel), PrP accumulation as evaluated by staining with the 3F4 anti-PrP monoclonal antibody (center panel), and reactive astrogliosis as detected by staining with glial fibrillary acidic protein antibodies (right panel). Similar results were obtained with several other animals inoculated with the same material. The profile of the brain lesion pattern was indistinguishable from those obtained by inoculation of brain infectious material. molecules and raised infectivity by more than 7 orders of magnitude. These data support and extend our previous findings showing the in vitro generation of prion-infectious material by cell-free replication of PrP Sc using PMCA (15). These findings provide strong support for the prion hypothesis. However, we cannot rule out that other molecules present in the brain homogenate may also contribute to infectivity.
In addition, PMCA may be useful for understanding the molecular mechanism of prion replication and the identification of endogenous factors modulating PrP Sc formation. Indeed, Supattapone and co-workers (36 -39) have used PMCA to show that metal cations, such as copper and zinc, and polyanions, including diverse types of RNA molecules, can modulate PrP conversion in vitro. PMCA may also be used to examine and quantify the species barrier phenomenon and to understand the mechanism encoding prion strains. Finally, PMCA may be used as an efficient high throughput screening assay for identification of molecules able to inhibit or reverse prion replication and thus to discover novel potential drugs for TSE treatment. An efficient treatment to stop further prion conversion coupled with an early pre-symptomatic diagnosis to identify patients before irreversible brain damage has occurred seems the most promising approach to combat these devastating diseases.